Section 1

Bio-Analystical Applications

SIMULTANEOUS QUANTITATION OF SPECIFIC PROTEINS USING IMMUNO-ASSAY ICP-MS

Zoë A. Quinn, Vladimir I. Baranov and Scott D. Tanner

MDS SCIEX, 71 Four Valley Drive, Concord, Ontario, Canada L4K 4V8

1 INTRODUCTION

ICP-MS has been used for many biological applications in the past, including: the quantitation of toxic elements in blood and other body fluids; studying the transport and destiny of metals in vivo; monitoring the absorption and metabolism of both nutrient metals and metallo-drugs; characterizing metallo-proteins and metal homologues of amino acids and stable isotope tracing (uptake, metabolism, diagnostics). Typically in all these methods, the bio-molecules analyzed with the ICP-MS are already associated with a metal component. However with the lowering of levels of detection for P and S, ICP-MS has been recently used to determine the state of protein phosphorylation (see Bandura et al.,). In addition, it is also possible to analyze bio-molecules that do not contain metal, through the use of conjugated metal tags. In this paper, we will describe the potential role for ICP-MS in the field of proteomics as well as describe the development of several novel element-tagged ICP-MS immunoassays (IA ICP-MS).

Proteomics, stated simply, is the characterization of proteins expressed by the genome (or genetic information stored in the DNA of an organism). It can be said that Proteomics is the next step after Genomics, and faces considerably greater challenges. In the pharmaceutical setting, the primary motive of Proteomics is often to determine a means to control the pathway to disease (e.g., using small molecule drugs). Therefore, it is very desirable to map the Human Proteome, not only to better understand disease and identify new drug targets, but also to determine the intricate mechanisms of human development. From the ~30,000 genes discovered in the Human Genome project, there are more than 500,000 protein variants (drug targets) predicted due to residue wobbling, splicing and other post-translational modifications. In comparison, genes are built by stringing together a series of 4 different nucleotides (of similar chemical properties) into a simple alpha helix. Proteins, however, are constructed from a pool of 20 different amino acids, each one distinctive in charge, hydrophobicity and size. Therefore the resulting 3-D structure of a protein is a unique globular mass packed with different secondary structures such as beta sheets, loops, and alpha helices. However, establishing the amino acid sequence, 3-D structure and function of each and every protein is complicated further by the over 20 forms of post-translational modifications (eg. phosphorylation, glycosylation), varying protein-protein interactions, tissue-specific expression patterns and responsiveness to environmental factors. For example, it is estimated that over 5 million tissue-specific protein complexes need to be identified and characterized in terms of function for the Human Proteome Project to be complete.

On a technical level Proteomics is hindered by the limited abilities of bio-molecule quantitation. We have found that the ICP-MS can be a valuable tool and provides sufficient sensitivity to measure specific proteins at endogenous levels, with the advantage of remaining largely oblivious to the bulk protein and salt matrix. In our methods, the quantitation of proteins using the ICP-MS is facilitated through the use of affinity separation, which allows for the isolation of a target species. Immuno-separation can be used prior to ICP-MS analysis, using specifically tagged antibodies to isolate target proteins (antigens) from cell cultures or serum, which allows absolute quantitation of a specific protein. In its most commonly used form, immunoassays employ antibodies with fluorescent, enzymatic or radiological tags for detection (in which the outcome of an enzymatic or fluorometric reaction is proportional to the amount of antigen present). We have found that the development of metal-tagged antibodies (for purposes other than element detection) makes possible the involvement of atomic spectroscopy in these biotechniques.

ICP-MS has several unique attributes that are complementary to the conventional protocols, including: sensitivity, large dynamic range, independence of the sample matrix, and the ability to detect a large number of elements and isotopes simultaneously. Combining these qualities with the specificity of an immuno-reaction offers a new approach to the proteomic challenge. The premise of the method is that antigens of interest are reacted with complementary, metal-tagged antibodies and physically separated from non-reacting proteins. The atomic composition of the tag (conjugated to the antibody) is then measured to determine the antigen concentration of the sample. In this reaction, the sensitivity of the method is a linear function of the number of atoms of a given isotope in the tag. This allows for absolute protein quantitation (assuming the number of atoms in the tag has a narrow distribution). The advantage of multiplexing also exists, where multiple antibodies can be labeled with distinguishable element tags (as elements, isotopes or in unique combinations - preferably those that occur at naturally low levels), and used for simultaneous determination of multiple antigens.

We have made use of immuno-reagents that already contain element tags for different purposes, such as gold-tagged antibodies which are used in the localization of cellular proteins through electron microscopy (and have also been successfully analyzed by ICP-MS). These nano-particle tags contain small gold or lanthanide clusters (less than 2 nm in diameter). This is an advantage when using ICP-MS for detection, as the elemental nano-particles used are of uniform size and contain a significant number of atoms per conjugate. In addition, there is also the option of increasing the Au signal response even further by using silver enhancement. In addition, four lanthanide (Eu, Tb, Dy, and Sm) tags have been conjugated to various antibodies and are marketed for use in an automated fluoroimmunometric Enzyme Linked Immunosorbent Assay (ELISA)-based system (Wallac AutoDELFIA). This method requires the addition of chelators to release the lanthanide ions forming fluorescent chelates. It is apparent that this approach allows greater possibilities for multi-analyte analysis, as the Wallac AutoDELFIA system is currently capable of detecting up to four analytes, provided that the fluorescence responses can be spectroscopically resolved.

We report the development of six novel ICP-MS linked immunoassays, in which some commonly-used immunoaffinity separation techniques (centrifugal filtration, gel filtration, protein A sepharose affinity, and direct, sandwich, and competitive ELISAs) were successfully coupled to the ICP-MS. In this manner, the elemental component of the reacted tagged antibodies was used to detect and accurately quantify the specific concentrations of target proteins in complex biological samples. It is demonstrated that the detection levels of these methods are as low as 0.05 ng/mL of target protein and yield a linear response to protein concentration over 4 orders of magnitude. Furthermore, we demonstrate a useful and important immunosorbent method using maleic anhydride plates linked to ICP-MS. In this 96 well format, two protein targets were analyzed simultaneously in each well. This brings to light future possibilities of quantitatively determining protein-protein interactions and the stoichiometry of biological complexes. In addition, several benefits of elemental detection over fluorescence have been observed: i) elimination of chelator step; ii) multi-analyte detection (e.g., elemental analysis provides the possibility of detecting multiple antigens per sample); iii) relative insensitivity to concomitant organic or metallic species; iv) inherent ability to provide absolute quantitation; v) linearity of response; and vi) long-term sample integrity.

2 METHOD AND RESULTS

2.1 Novel ICP-MS Linked Immunoassays.

The ICP-MS linked immunoassays we have developed require both element-tagged antibodies (or competing peptides) and the separation of antigen-bound antibodies from free antibodies. We have developed several ICP-MS immunoassays using centrifugal filters, Protein Sepharose A, Sephacryl S200 gel filtration, and various ELISA plates. In all of these assays, we make use of two sets of element-tagged antibodies: Fab'-nanoAu fragments and Eu-labeled whole molecule antibodies. One potential disadvantage of using gold in this manner is that gold has a high affinity for surfaces of a typical ICP-MS sample introduction system. However, it was observed that this effect is significantly reduced in the presence of proteins in a sample, probably due to complexation and/or passivation of surfaces. The concentrations of commercially available Fab'-nanoAu is not specified by the manufacturer as conventional applications do not require absolute quantitation. However, the manufacturer does specify that each gold cluster is approximately 70 atoms. The elemental composition of the Fab'-nanoAu reagents has been previously characterized by ICP-MS. As discussed previously, we have found that although chromatographically purified by the manufacturer, approximately 50–60% of the gold signal is not associated with the Fab'-nanoAu (see Baranov et al.,). However this result is obviously lot-dependent. We expect that these nanoAu impurities have a minimal impact on the results since they are easily removed during separation and washing procedures. We have used iridium as an internal standard for gold because gold and iridium have close atomic mass numbers and their ionization potentials are similar (9.225 and 9.1 eV, respectively). Therefore, the instrument response is expected to be comparable for both analytes using an ICP ion source. In addition, both Au and Ir are stable in HCl, which is a useful medium for dissolving protein samples. In this work, the ratio of the ion signals, 197Au/(191Ir and 193Ir) in a standard solution containing 1% BSA acidified (1:1) with 10% (v/v) HCl, 0.1% HF, 1 ppb Au and 1 ppb Ir was used to quantify the gold content in experimental samples. Similarly, we have chosen holmium as an internal standard for europium measurements since europium and holmium have close atomic mass numbers and their ionization potentials are 5.67 and 6.02 eV, respectively. The Eu-labeled lgG molecules contain approximately 6–10 atoms of Eu per IgG molecule as specified by the manufacturer. Eu has two isotopes of similar abundance, and we therefore expect that 3–5 atoms of each Eu isotope (151Eu, 47.8% abundant and 153Eu, 52.2% abundant) exist per tag.

2.2 Centrifugal Filtration IA ICP-MS.

In the first method, centrifugal filters were used to quantitate Human IgG by producing a calibration curve in the range 10-100ng/ml (Figure 1A). An excess of prefiltered anti-human Fab'-labeled with Nanogold in 1% BSA/PBS was used, and we assumed that all IgG/Fab'-nanoAu complexes, plus nonspecifically bound Fab'-nanoAu, would be retained on the top of the filter. The remainder of unbound Fab'-nanoAu would pass through the filter. This method was not found to be very practical for two reasons: i) significant retention of the unbound antibody supposedly due to nonspecific absorption on the filter despite blocking with 1% BSA; and ii) the molar concentration of the IgG/Fab'-nanoAu complex is significantly lower than the concentration of IgG in the original solution (perhaps due to poor extraction from the filter). It was found that the gold cluster response (and noise) will limit the DL to ~3 ng/mL (or ~20 fmol/mL) of IgG. As well, ICP-MS coupled size exclusion filter immunoassays can provide an accurate means of detecting specific antigens over a wide range of concentrations (10-5000 ng/mL). However, high background in the blank solutions might degrade the lower range limit.

2.3 ICP-MS Immunoprecipitation Assay.

The second ICP-MS linked immunoassay method developed was found to have less background and more sensitivity. This method utilizes protein A sepharose (PAS) to immunoprecipitate the target antigen that is then quantitated by binding with an element-tagged detection antibody. In this experiment, we analyzed the concentration range of 0.1-100 ng/mL (0.667-667 fmol/mL) Human IgG (Figure 2A). We found that despite passivation, non-specific binding limits the dynamic range of this method at low concentration of antigen. Deviation from linearity in the low concentration region on the level of 50-10 ppt of gold is largely dependent on the Fab'-nanoAu concentration in the reaction mixture. According to our previous investigation of Kd for this reaction, we expect that the amount of gold detected is directly proportional to the amount of PAS/IgG/Fab'-nanoAu complexes formed in the reaction. In this method, we found that in order to detect very low amounts of antigen, it is necessary to maintain a low background of gold. For this set of experiments (see Figure 2A), the background and sensitivity translate into DL = 0.5 amol/µL for IgG. Therefore, this method should significantly gain from reducing non-specific binding and by improvements in efficiency of the ICP-MS sample introduction system (reduction of sample volume, longer acquisition time).

In a comparison study, we quantified the range of 0.5-500 ng of human IgG using Western blotting and densitometry analysis. Western blotting is widely used to detect specific proteins in complex biological samples and also to estimate protein concentrations. In Western blotting, proteins are separated by mass in a polyacrylamide gel using electrophoresis and then transferred to nitrocellulose using electroblotting. Proteins of interest can be detected in the nitrocellulose blot by probing first with a specific primary antibody (raised against the target protein) and then with an anti-primary, peroxidase-tagged antibody to detect the existing protein-antibody complexes. After exposure to a chemiluminescent substrate and sensitive film paper, proteins of interest can be identified by the appearance of bands on the film. In this study, serial dilutions of native human IgG were prepared and divided in two, for both ICP-MS linked immunoprecipiation and Western blotting (Figure 2A,B). In Western blotting it is apparent that 1 ng is the lowest amount of human IgG that could be detected. In both Westem blots, 2.5 ng could be reliably detected at both 5- and 20-s exposures. It is also apparent that the relative densities of the blots do not accurately reflect the concentration of human IgG over the range of concentrations. The expected linear relationship between signal and protein concentration was observed only in the interval 2.5-25 ng of IgG, which might be attributed in part to the limited capability of the imager (256 levels of gray) in data quantification. In comparison, the ICP-MS linked immunoprecipitation assay exacted a linear relationship over the entire concentration range of Human IgG (0.5-500ng) and could be easily performed in triplicate to yield statistically significant results. However, it should be noted that the Western blot method has one advantage over the IA ICP-MS methods: the ability to analyze small sample volumes of 5-30ul. In the ICP-MS linked immunoprecipitation assays, we kept the sample volume in the range 0.5-1.5 mL, which was convenient for the standard sample introduction system, with an aspiration rate of approximately 100 µL/min. Miniaturization of the sample introduction system and reduction of the sample uptake rate would be desirable for real biological samples as well as improve the DL of the ICP-MS immunoassays.

To further investigate the accuracy and precision of antigen detection through the ICP-MS immunoprecipitation assay, we were able to quantify the concentration of human IgG in a commercially available immunology control serum. In this assay, a calibration curve was generated using serial dilutions of human IgG in the concentration range of 0.1-100 ng. From four serial dilutions of both control sera (Levels 1 and 2), the amount of IgG present in Levels 1 and 2 were determined to be 875 [+ or -] 54 and 2303 [+ or -] 181 mg/dL, respectively. It should be noted that these determinations rely on the manufacturer's specified concentration of human IgG in the standard solution. The uncertainty in this concentration is unknown and was not included in the uncertainties of quantitation of the control serum given above. The manufacturer of the control serum used four different analyzers to estimate the mean concentration of IgG in Level 1 to be 738 mg/dL (concn range, 590-886 mg/dL) Beckman Array (nephelometric/IFCC RPPHS), 714 mg/dL (concn range, 571-857 mg/dL) BMC Hitachi (IFCC RPPHS), 664 mg/dL (concn range, 531-797 mg/dL) Dade ACA/Star (IFCC RPPHS), and 747 mg/dL (concn range, 598-896 mg/dL) Dade Behring (nephelometric/ IFCC RPPHS). Similarly, the mean concentration of IgG in Level 2 has been estimated by the manufacturer to be 2098 mg/dL (concn range, 1678-25 18 mg/dL) Beckman Array (nephelometric/IFCC RPPHS), 2 109 mg/dL (concn range, 2687-2531 mg/dL) BMC Hitachi (IFCC RPPHS), 1901 mg/dL (connrange, 1521-2281 mg/dL) Dade ACA/Star (IFCC RPPHS), and 2150 mg/dL (concn range, 1720-2580 mg/dL) Dade Behring (nephelometric/IFCC RPPHS). It is apparent that the concentration of specific antigen in a complex biological matrix can be accurately quantitated by IA ICP-MS.