Bioanalysis of recombinant proteins by mass spectrometry

Biotechnological developments have led to an increased number of recombinant proteins or antibodies in drug development that offer high potential in various diseases, such as cancer, growth disturbances and diabetes. This class of compound represents more than 30% of drugs approved during the last few years and nearly 200 products have now gained approval.¹ Study of the pharmacokinetics of recombinant proteins is essential during their preclinical and clinical evaluation. In vivo transformation, binding to circulating targets and immunogenicity are some examples of the pharmacokinetic properties that drive the selection of bioanalytical assays for assessing the blood level of these proteins.

Current methods for pharmacokinetic evaluation of proteins are mainly based on immunoassays. However, these sensitive methods involve the time-consuming step of obtaining polyclonal or monoclonal antibodies (mAbs), and are susceptible to cross-reaction with metabolized fragments, circulating receptors or endogenous analogous proteins. Moreover, recombinant proteins are often immunogenic and lead to endogenous antibodies present in blood samples. This generates major interferences in both competitive and sandwich immunoassays, leading to false-positive or negative results depending on the immunoassay format.²

Increased regulatory requirements for recombinant proteins or therapeutic mAbs will necessitate a large panel of different analytical strategies. Among the various alternatives, liquid chromatography coupled with mass spectrometry (LCMS) is potentially useful as it may offer rapid development and improved assay specificity. LCMS methods are already the technique of choice for small molecular weight drugs and are beginning to be of interest for recombinant proteins too. A few LCMS methods have been used for preclinical toxicity evaluation and clinical pharmacokinetics, and recent literature demonstrates that these methods can satisfy the requirements of specificity, reproducibility and sensitivity (Table 1). The most sensitive techniques use an immunoaffinity extraction as sample pretreatment and allow a detection limit close to 10–100 pM, which is one-or two-log above immunoanalytical methods, but largely satisfactory for drug monitoring in biofluids.

Table 1 Recent analytical developments for the quantification of recombinant proteins or antibodies in biological matrix.

Analytical strategies

Two main alternatives are possible, although they offer different levels of sensitivity (Figure 1). The first is a direct strategy where the sample preparation step is reduced to protein precipitation or direct trypsination.³ For the second, the analyte is specifically extracted by immunoconcentration with specific antibodies,⁴ or antigens in the case of recombinant antibodies.⁵ An interesting feature of the second approach is the increase of sensitivity, up to 10–100-fold, owing to sample concentration and removal of matrix interferences in the mass spectrometer. It should be kept in mind that proteins with a molecular weight above 10–15 KDa generally have to be submitted to trypsin digestion.

Figure 1

The first approach is preferred when the required sensitivity is above 10–100 nM. The sample preparation may consist only in protein precipitation or removal of abundant proteins, however, one main drawback is related to the remaining endogenous compounds in the extracted plasma, leading to significant matrix effects that induce a lower sensitivity. When the sensitivity requirement is high, the second alternative may be chosen. Here, a specific step such as an immunocapture affords sample concentration and removal of matrix interferences. For ease of assay development, magnetic beads coated with G proteins that bind the monoclonal or polyclonal antibody can be used. After binding, the targeted protein is eluted from the beads using organic solvents, high-ionic strength solutions or low pH solutions. At this step, a 10–100-fold increase in sensitivity can be obtained thanks to sample concentration, as a plasma volume of 1 mL may be reduced to less than 0.1 mL after extraction. Furthermore, this sample pretreatment removes unrelated proteins, thus reducing the ionization efficiency in the mass spectrometer.

In the literature, the two strategies are illustrated as reported in Table 1. Analysis of small therapeutical proteins by a direct approach, such as solid phase extraction followed by mass spectrometry quantification, as described for rK5 by Ji et al.⁶ or for Sifuvirtide by Dai et al.⁷ resulted in poorer sensitivity (10000 and 1000 pM respectively) than analysis performed with immunoaffinity. The application of the second strategy to Synacthen by Thevis et al.⁸ or to rhEPO by Guan et al.⁹ led to better sensitivity (100 and 10 pM respectively). These two alternatives were compared for a small therapeutic protein comprising 56 amino acids (MW=6256 Da; Figure 1). Extraction was performed with protein precipitation (first alternative) or by immunocapture (second alternative), followed by liquid chromatographic separation of the intact protein and ions at m/z 1039–1225. The internal standard comprised a chemical analog (cystein derivatization with iodoacatamide) that was introduced at the beginning of the sample treatment to monitor the entire process. A 10-fold higher sensitivity was observed with the immunocapture approach (limit of quantification at 0.5 versus 5 ng/mL), giving a sensitive assay of the protein able to monitor plasma concentration up to 72 h after injection.³ Methods with both extraction processes were validated according to FDA recommendation and compared with the original enzyme-linked immunosorbent assay (ELISA) , where they showed very good agreement.

A second example is related to mAb quantification. One difficulty for therapeutic mAbs is their similarity to human endogenous immunoglobulins, one of the most abundant proteins in plasma with concentration ;10 mg/mL. This is 1000–10000-fold greater than that of circulating therapeutic antibodies and could represent a potential source of analytical interference. A second difficulty is that the assay should be able to specifically determine the therapeutic active form of the mAb in the biological matrix (i.e., the form able to bind its endogenous target). Using the first alternative, with either solid phase extraction or albumin depletion, Yang et al.¹⁰ , Hagman et al.¹¹ and Heudi et al.¹² recently published quantification of mAb with a sensitivity ranging 3000–33000 pM. Conversely, we used the second strategy for the quantification of Erbitux, which is a mAb used for the treatment of colorectal cancer. Magnetic beads covalently linked to the extra cellular region of the epidermal growth factor receptor, the target of Erbitux, were used to extract the mAb from human serum. After binding and elution, the antibody was submitted to trypsin digestion, and tryptic peptides representative of the variable regions of both the light and heavy chains were selected for quantification (Figure 1). These peptides (LT3 and HT4) were monitored by LCMS. The method allowed a detection limit of 20 ng/mL, which is better than most of ELISA for therapeutic antibodies.⁵ In this case, the internal standard comprised a mouse mAb specific to EGFR.

Specific problems

Although the examples provided in the literature demonstrated that LCMS approaches for the assessment of recombinant proteins in biological fluids can provide sensitive and specific assays, few companies utilize the technique for their pharmacokinetic studies. A few technical difficulties remain to be fully solved; among them, is the use of specific capture step amenable to assay automation using online extraction columns or easy-to-perform manual steps.

A second problem is the need of an internal standard for accurate and reproductive quantification, especially when sample preparation involves numerous steps. Although tryptic peptides labelled with stable isotopes are often used, they do not allow for checking of the sample preparation steps, such as immunocapture and enzymatic digestion. As demonstrated previously, analogs to the targeted protein may be used, but they do not represent the ideal situation. In the bioanalysis of small molecular weight drugs, the current strategy is to use a drug labelled with stable isotopes. In the same way, isotope-labelled forms of the entire proteins would constitute the ideal internal standard for quantification. Considering this concept, colleagues from the Atomic Energy Commission (CEA; EDyP Laboratory at CEA/Grenoble, France) have developed a new methodology called the Protein Standard Absolute Quantification (PSAQ).13 This utilizes isotope-labelled, full-length proteins as standards. These standards are produced and labelled in 'cell-free' systems, allowing efficient labelling and facilitating the purification step. As the standards perfectly match the biochemical properties of the target proteins, they can be directly added to samples at the very first step of the analysis. In head-to-head comparisons with previous quantification strategies (labelled tryptic peptides), PSAQ more accurately detected the quantities of two Staphylococus aureus toxins that were present in drinking water and urine. Applications of the PSAQ strategy for recombinant proteins — and also in very different research domains such as disease biomarker evaluation, quality control of vaccines, allergens and toxins detection — can be anticipated.

Conclusion

Mass spectrometry approaches and technologies derived from proteomic studies are readily implemented for absolute quantification of therapeutic proteins in biological fluids. Table 2 summarizes the advantages and disadvantages of mass spectrometry compared with ELISA. During the next few years, it is expected that LCMS methods, which are already the reference for determining the pharmacokinetic profiles of small molecular weight drugs, will also gain importance for biotech drugs.

Table 2 Advantages and disadvantages of mass spectrometry compared with ELISA.

References

1. G. Walsh, Nat. Biotechnol., 24, 769–776 (2006).

2. G. Shankar, C. Pendley and K.E. Stein, Nat. Biotechnol.,25, 555–561 (2007).

3. F. Becher et al., Anal. Chem., 78, 2306–2313 (2006).

4. M. Dubois et al.,Rapid Commun. Mass Spectrom., 21, 352–358 (2007).

5. M. Dubois, Anal. Chem.,80, 1737–1745 (2008).

6. Q.C. Ji et al., Anal. Chem., 75, 7008–7014 (2003).

7. S. Dai et al.,RCM, 19, 1273–1282 (2005).

8. M. Thevis et al., RCM, 20, 3551–3556 (2006).

9. Y. Guan et al., Anal. Chem., in press (2008)

10. Z. Yang et al., Anal. Chem., 79, 9294–9301 (2007).

11. K. Hagman et al., Anal. Chem., 80, 1290–1296 (2008).

12. O. Heudi et al., Anal. Chem., 80, 4200–4207 (2008).

13. V. Brun et al., Mol. Cell Proteomics, 6, 2139–2149 (2007).

Eric Ezan is Group Leader.

Mathieu Dubois is Researcher.

François Becher is Researcher.

All at Pharmacology and Immunoanalysis Unit, Institute of Biology Technologies–Saclay, Atomic Energy Commission (Gif-Sur-Yvette, France).