Abstract
Historically, since the metabolism of administered peptide/protein drugs (“biotherapeutics”) has been expected to undergo predictable pathways similar to endogenous proteins, comprehensive biotherapeutic metabolism studies have not been widely reported in the literature. However, since biotherapeutics have rapidly evolved into an impressive array of eclectic modalities, there has been a shift toward understanding the impact of metabolism on biotherapeutic development. For biotherapeutics containing non-native chemical linkers and other moieties besides natural amino acids, metabolism studies are critical as these moieties may impart undesired toxicology. For biotherapeutics that are composed solely of natural amino acids, where end-stage peptide and amino acid catabolites do not generally pose toxicity concerns, the understanding of biotherapeutic biotransformation, defined as in vivo modifications such as peripherally generated intermediate circulating catabolites prior to end-stage degradation or elimination, may impact in vivo stability and potency/clearance. As of yet, there are no harmonized methodologies for understanding biotherapeutic biotransformation and its impact on drug development, nor is there clear guidance from regulatory agencies on how and when these studies should be conducted. This review provides an update on biotherapeutic biotransformation studies and an overview of lessons learned, tools that have been developed, and suggestions of approaches to address issues. Biotherapeutic biotransformation studies, especially for certain modalities, should be implemented at an early stage of development to 1) understand the impact on potency/clearance, 2) select the most stable candidates or direct protein re-engineering efforts, and 3) select the best bioanalytical technique(s) for proper drug quantification and subsequent pharmacokinetic profiling and exposure/response assessment.
Introduction
Peptide and protein drugs (“biotherapeutics”) have become increasingly important modalities for the treatment of grievous diseases. In 2010, there were more than 200 biotherapeutics approved for use (Walsh, 2010). The types, or modalities, of these biotherapeutics are quite varied. They include hormones, growth factors, and other replacement peptides/proteins; monoclonal antibodies (mAbs); subunit vaccines; fusion proteins; and therapeutic enzymes (Walsh, 2010). Moreover, a recent report showed that annual revenue for biotherapeutics had grown steadily during the previous 10 years, and in 2011, biotherapeutics accounted for 15.6% of the total global pharmaceutical market, which was valued at $138 billion globally. This valuation is expected to increase to over $320 billion by the year 2020 (http://giiresearch.com/press/6805.shtml).
Understanding the biological absorption, distribution, metabolism, and excretion (ADME) properties of small molecules (SMs) is a critical element in SM drug discovery and development, spanning early discovery to late development (Baillie, 2008). In contrast, the understanding and mechanistic details of the ADME properties of biotherapeutics are not as well developed, although there has been recent interest in closing this gap (Prueksaritanont and Tang, 2012). One reason for the disparity has been attributed to the relatively young age of the era of biotherapeutics compared with that of SM drugs (Waldmann, 2003). At any rate, as new ADME characterizations of biotherapeutics become available in the literature, it behooves the scientific community to assess the state of the science in toto. This review focuses on the current understanding of metabolism properties of biotherapeutics versus SM drugs and the concomitant impact on biotherapeutic discovery and development.
To avoid potential confusion resulting from the common association of the term “metabolism” with SM drugs or with end-stage protein lysosomal degradation, the term “biotransformation” will be used to describe the physical alteration of a biotherapeutic due to peripheral intermediate catabolism and truncation, not including inherent chemical stability. Other biotransformation events, such as deamidation, oxidation, or other amino acid modifications, will be mentioned briefly but are generally beyond the scope of this review. The main objectives of this article are to review work that has been done and tools that have been developed to examine protein (molecular mass >5 kDa) biotherapeutic biotransformation, clarify the impact on candidate selection and bioanalysis, and suggest approaches for assessing biotransformation based upon protein therapeutic modality.
Metabolism: Small-Molecule Drugs Versus Biotherapeutics
Unlike traditional small-molecule drugs, biotherapeutics generally exhibit poor oral bioavailability; therefore, they are dosed in a parenteral manner, usually by i.v. or s.c. injection. During the absorption and distribution phases, biotherapeutics undergo metabolism and clearance through mechanisms that differ substantially from SM drugs. Metabolism for SM drugs is defined as essential biochemical modifications, which occur predominantly in the liver, that often render the drugs more susceptible to elimination (for example, by increasing hydrophilicity). Although detoxification and enhanced clearance are the predominant outcomes of SM drug metabolism, there is the potential for increased toxification after metabolism and undesired accumulation of toxic metabolites in tissues or organs. As a result of this potential for untoward effects, the understanding of metabolism is an absolute requirement for moving a small-molecule candidate forward through the development pipeline. The analytical tools for understanding small-molecule drug metabolism have been well established and validated by the pharmaceutical industry. Regulatory agencies also have very specific guidelines with respect to requirements for characterization of small-molecule drug metabolism that have to be addressed in any New Drug Application filing.
In contrast to metabolism of small molecules, metabolism of biotherapeutics is predominantly defined as the enzymatic hydrolysis (catabolism) of polypeptides to produce smaller peptides and amino acids for deactivation and increased clearance. Unlike small-molecule drugs, the final small peptide and amino acid breakdown products of all biotherapeutics are produced predominantly by lysosomal degradation after active or passive cellular uptake, and thus are expected to be similar. Since these end products are not likely to pose clinical toxicity concerns, in-depth studies of biotherapeutic metabolism have not been abundant, and analytical tools for probing biotherapeutic metabolism have not been systematically adopted. In addition, guidance from regulatory agencies with respect to biotherapeutic metabolism is minimal and reflects a certain indifference. As noted by Hill, “The ICH [International Conference on Harmonization] S6 (R1) guidance [from 1998] suggests that the consequential metabolism of proteins and peptides is via expected routes and implies only a little more investigation may be needed…” (Hill, 2010). Newer guidance from the European Medicines Agency (2007; http://www.ema.europa.eu/docs/en_GB/document_library/Scientific_guideline/2009/09/WC500003029.pdf) states that “metabolites that have pharmacodynamic activity should preferably be measured, for example, through chromatographic separation, collection and further in vivo bioassay quantification. However, in cases where measurement of separate active metabolites or peptide fragments is not technically feasible, pharmacokinetics of the active moiety could be determined.” This newer guidance implies selective metabolism studies may be in order, especially in cases where amino acid sequences have been altered. Safety concerns can arise with these types of biotherapeutics—for example, hybrid biotherapeutics where protein scaffolds contain unnatural or D-amino acids, polyethylene glycol (PEG), or small-molecule moieties such as those found in antibody-drug conjugates (ADCs) (Hamuro and Kishnani, 2012).
Even with biotherapeutics that contain only natural amino acid sequences, the need for metabolism studies has to be presently assessed in light of the rapid production of widely varied protein therapeutic modalities in recent years, ranging from small peptides to large multispecific chimeric proteins. The catabolic breakdown of biotherapeutics may involve a series of kinetically disparate proteolytic events, as illustrated in Fig. 1. In this example, an antibody (Ab)/peptide fusion construct is administered, and during the absorption and distribution phases, a large intermediate catabolite forms rapidly (kcatabolism) whereby the fused peptides are proteolytically truncated. This catabolite may be sustained in circulation for a significant amount of time after dosing before ultimate lysosomal degradation and/or renal excretion (kelim1 or kelim2) (Hall et al., 2010; Hager et al., 2013). This peripheral, nonlysosomal catabolism of biotherapeutics is referred to as “biotransformation” and is systemic, but the major sites of enzymatic breakdown are blood, liver, and kidney (Werle and Bernkop-Schnürch, 2006; Lin, 2009). Proteases with exo- or endopeptidase activity have been identified and characterized in these tissues/organs (Werle and Bernkop-Schnürch, 2006). Although clinical safety may not be generally affected, the potential impact of circulating catabolites on potency and clearance may require consideration or assessment. Knowledge of specific sites of proteolytic cleavage is critical for engineering more stable biotherapeutic candidates by mutation of susceptible residues or application of other types of stabilization. In addition, the traditional bioanalytical tools for quantification of biotherapeutics and determining pharmacokinetic (PK) exposure such as ligand-binding assays (LBAs) may not take catabolism into account, and the resulting concentration measures may therefore either underestimate or overestimate true exposure. The impact on bioanalysis is fully addressed in the next section.
Impact of Protein Biotherapeutic Biotransformation on Bioanalysis
LBA is the standard method by which biotherapeutics, especially larger protein biotherapeutics, are quantified in preclinical and clinical samples for PK and pharmacodynamic (PD) profiling. Noncompetitive “sandwich” enzyme-linked immunosorbent assay is the standard LBA for bioanalysis. In this assay format, capture and detection reagents track distinct epitopes of the analyte of interest. Often, the capture and detection reagents are chosen based upon assay optimization using only the purified, intact biotherapeutic analyte spiked into serum or plasma. For this reason, biotransformation has generally not been considered during LBA development. For the purposes of correct PK/PD profiling, the optimal LBA will track only those entities (parent and bioactive catabolites) that contribute to true exposure. For instance, glucagon-like peptide-1 (GLP-1) is a 37–amino acid peptide that has been investigated for treatment of type II diabetes. To increase its half-life and bioavailability, it has been conjugated to the N terminus of a carrier mAb (Murphy et al., 2010). After i.v. administration of this construct to mice, quantification of the therapeutic was achieved by two different LBA formats that both used anti-idiotypic capture of the Ab portion of the construct; however, two different detection reagents were used: 1) antihuman IgG, for detection of the mAb only (“total” assay), or 2) anti–GLP-1 (N-terminal specificity), for detection of intact GLP-1 (“N-terminal” assay). The N-terminal assay produced drug concentration results that were significantly lower than the total assay throughout the PK time course. These results indicated that, whereas the concentration of the mAb portion of the construct was sustained, the N-terminal region of the GLP-1 portion was quickly degraded. More importantly, the pharmacological action of GLP-1 is contingent upon having an intact N terminus (Deacon, 2004). Therefore, if the total assay were to be used for PK profiling, it would grossly overestimate the true exposure, as the reported concentration of the biotherapeutic is predominantly derived by inactive catabolites that contain degraded GLP-1. Thus, the N-terminal assay would be the appropriate LBA for PK/PD profiling. Additionally, since the discrepancy between the total and N-terminal LBAs was significant, subsequent mass spectrometry (MS) studies were undertaken to determine the exact vulnerable proteolytic loci of the GLP-1. This information was used to stabilize the GLP-1 N-terminal region with the goal that the results of the total and N-terminal assays would ultimately approach convergence. The indispensable utility of MS for analysis of biotherapeutic biotransformation will be addressed in the next section of this review.
This GLP-1/mAb fusion example demonstrates the possibility for LBA to significantly overestimate exposure if biotransformation is not considered. Alternatively, the following example demonstrates how exposure can be underestimated. Romiplostim is a novel protein biotherapeutic used for the treatment of idiopathic thrombocytopenia, consisting of tandem repeats of thrombopoietin mimetic peptides (TMPs) fused to the C terminus of the fragment crystallizable region (Fc) of human IgG1 (Molineux and Newland, 2010). In the intact construct, there are a total of four TMPs. Two LBA formats were developed to quantify romiplostim in plasma samples obtained from preclinical in vivo experiments (Hall et al., 2010). For capturing romiplostim, both assays used a rabbit polyclonal reagent raised against TMP. In the first assay (bridging assay), detection of the captured romiplostim was achieved by using the same polyclonal reagent. In the second assay (TMP/Fc assay), the detection reagent was an antihuman Fc mAb. In an in vivo rat PK study, there was a large discrepancy between the two assays with the bridging assay producing markedly lower drug concentration values than the TMP/Fc assay. The TMP/Fc assay values were similar to radioassay results where 125I was incorporated into the Fc region of romiplostim; thus, this LBA appeared to track all Fc-containing catabolites and could be considered a total assay with the caveat that at least one TMP was present in the construct. Exploration of what the bridging assay measured revealed that the entire construct had to be essentially intact to produce any measurable assay signal; a romiplostim analog containing only one TMP on each monomer of Fc failed to produce a signal in this assay. Further mass spectrometric studies showed extensive catabolism of the terminal TMPs with the internal TMPs remaining largely intact after 24 hours postdose. Although relative bioactivities were not available, it could be reasonably hypothesized that all catabolites of romiplostim that contained at least one viable TMP would be bioactive and should be monitored for accurate PK profiling. Thus, in this case, the TMP/Fc assay would be most appropriate for reporting romiplostim concentrations, as it reflected both the intact construct plus its bioactive metabolites. Using the bridging assay was too specific, providing concentration information for intact romiplostim alone, thereby leading to underestimation of true exposures.
As was the case with the GLP-1/mAb construct, MS studies were crucial to assess 1) the exact molecular details of romiplostim biotransformation, 2) which catabolites any given LBA was detecting, and 3) the most appropriate LBA for PK assessment. In addition, the extensive biotransformation of romiplostim led to the design of a newly stabilized construct where the TMPs were inserted between the CH2 and CH3 domains of Fc, thus rendering the TMPs less susceptible to catabolism (Hall et al., 2010).
To circumvent these issues of the effects of biotransformation on LBA development and specificity, one natural inclination is to bypass LBA entirely and move toward use of quantitative MS assays where specificity concerns are ameliorated. Biotherapeutics or associated surrogate peptides as well as any important catabolites could be specifically quantified by MS. In fact, several researchers have published studies with this ultimate goal in mind (Dubois et al., 2007, 2008; Heudi et al., 2008; Ezan et al., 2009; Liu et al., 2011b; Bronsema et al., 2012; Li et al., 2012). Historically, there have been several drawbacks associated with quantitative MS methods compared with LBA, including the following: 1) MS has been generally less sensitive, 2) large protein analytes have suffered from low throughput, and 3) quantitative MS methods for macromolecules are much more nascent compared with established antibody-based methodologies. Presently, however, detection sensitivity has improved to nearly match that of LBA due to advances in MS instrumentation and work flows such as the utilization of nanoflow liquid chromatography (LC) coupled to nanospray MS (Duan et al., 2012) and use of pre-enrichment techniques such as affinity MS, and throughput has increased substantially by the implementation of automation. We are at an exciting juncture where PK profiling achieved by quantitative MS is becoming more commonplace and may be eventually used with similar frequency as LBA. Until then, LBA remains the gold standard for biotherapeutic bioanalysis, and as such, critical information about biotransformation, whether it is obtained from MS or other techniques, can be used to develop the correct LBA for the most accurate PK profiling.
Workflows for Studying Protein Biotherapeutic Biotransformation
Mass Spectrometry.
MS is becoming the analytical tool of choice for assessing biotransformation of protein biotherapeutics. Although differential LBA results, as demonstrated in the examples in the previous section, offer an indication that gross metabolism is occurring by utilizing multiple formats that assess different regions of the analyte (e.g., total versus intact), LBA results can only serve as an important yet inexact screen for probing biotransformation. Moreover, these results cannot easily provide exact molecular details about the specific locations of proteolytically sensitive residues, or “hot spots.” Due to the high mass resolution of current mass spectrometers, however, this molecular understanding is readily obtainable; the observance of the catabolic loss of just one amino acid is easily achieved. In addition, other biotransformation events, such as oxidation of vulnerable amino acids such as methionine and tryptophan, can be easily tracked by high-resolution MS; these events would be nearly impossible to probe by LBA.
Sample Generation and Collection.
Biotransformation studies require the generation of appropriate samples for analysis. Since in vitro test systems have been successfully implemented for the analysis of the metabolism of SM drugs, researchers have also attempted to use in vitro test systems to understand biotherapeutic biotransformation. For example, studies have been conducted in whole blood/serum/plasma, and liver and kidney homogenates (Boulanger et al., 1992; Powell et al., 1992; Fredholt et al., 2000; Sofianos et al., 2008). A couple of cautionary issues for using in vitro test systems need to be addressed. First, the level of catabolism, especially for tissue homogenates, is generally much greater than that observed in in vivo models (Werle and Bernkop-Schnürch, 2006). Second, the in vitro catabolic profile may not adequately describe that which actually occurs in vivo. An example of this is given by the study of catabolism of stromal-derived factor-1, also known as C-X-C motif chemokine 12 (CXCL12), an 8-kDa chemokine (Antonsson et al., 2010). CXCL12 catabolism was studied by both in vitro and in vivo mouse models. In vitro, truncation of the N terminus of CXCL12 up to five amino acids was observed; however, the protein was further truncated in vivo by two additional amino acids (Antonsson et al., 2010). Additionally, when a methionine residue was added to the N terminus, it completely protected the N terminus from degradation in vitro but not in vivo (Antonsson et al., 2010). Therefore, although in vitro test systems are attractive due to ready availability and low cost, at present, in vivo studies produce the most reliably accurate descriptions of physiologically relevant biotherapeutic biotransformation. Furthermore, since in vivo studies are preferable, analysis of biotherapeutic biotransformation and in vivo stability has been nearly exclusively monitored in collected blood/plasma/serum.
To begin an in vivo biotransformation study, an animal species has to be selected. Preclinical studies of catabolism are dominant in the literature, demonstrating that issues of biotransformation should be addressed early in a research program so that potential candidates can be stabilized by re-engineering if liabilities are found. One could theorize that biotransformation studies in nonhuman primates are most likely to be extrapolatable to humans, although there is little to no literature to support this. One study reports that biotherapeutic catabolism profiles between rodent and monkey studies were identical except for the temporal appearance of any given catabolite (Hager et al., 2013). This is an important linkage given the higher expense of conducting monkey studies. Additional studies are needed to examine cross-species differences in biotherapeutic biotransformation, and more importantly, the correlation of preclinical to clinical data.
Once the animal species has been chosen, the biotherapeutic is administered by the desired mode of introduction, usually i.v., s.c., or i.p. There are no strict guidelines as to the dosing level, as there have been no studies to date showing any effect of dose on catabolic profiles; however, the dose should be high enough to track catabolism for the desired period of time postdosing. Furthermore, due to the generally lower sensitivity of MS methods compared with LBA, a dosing level higher than that used for LBA is usually implemented. The route of administration may potentially impact the resultant catabolic profile. The appearance of catabolites may be delayed following s.c. versus i.v. dosing, but the identities of the catabolites generated may be independent of the route of administration. It is generally believed that, after s.c. administration, the lymphatic system is the primary route of absorption for protein biotherapeutics with molecular masses >16 kDa (McLennan et al., 2005). At the injection site and during lymphatic transport, it can be hypothesized that a biotherapeutic may be exposed to unique proteases that could result in biotransformation prior to systemic exposure. Although there are reports that biotherapeutics are stable after in vitro incubation in lymph (Charman et al., 2000; Wang et al., 2012), catabolic degradation of biotherapeutics has been observed in subcutaneous skin tissue homogenate and lymph node suspensions (Wang et al., 2012). It is not known, however, if this catabolic activity results in complete destruction of the administered biotherapeutic or can produce larger catabolic fragments that appear in systemic circulation. More studies are clearly needed to assess the impact of route of administration upon biotherapeutic biotransformation.
Blood collection from dosed animals needs to be carefully considered to maintain biotherapeutic integrity. The guidelines for sample collection for maintaining integrity of protein biomarkers and plasma proteomes can serve as a useful template for biotherapeutic biotransformation studies (Omenn et al., 2005; Rai et al., 2005; Rai and Vitzthum, 2006; Tammen, 2008). Any catabolites discovered during a study need to be attributed to biotransformation and not to artifactual proteolysis occurring during sample collection. Collection of plasma is generally preferential to that of serum to avoid any degradation due to activated proteolysis that could occur during blood clotting. In addition, ex vivo stability should be monitored by adding the analyte of interest to control plasma/serum and carrying this sample through the same ensuing processing and analysis steps as the actual in vivo samples. Furthermore, collection of very early time points (<5 minutes) can serve as a positive control where in vivo catabolism is expected to be minimal; the presence of proteolyzed analyte in this case would suggest artifactual ex vivo degradation. Although not routine, the addition of protease inhibitors during collection could also inhibit any ex vivo proteolysis.
Sample Preparation.
Once plasma/serum samples have been collected properly, the next step is preparing the samples for MS analysis. Plasma/serum is a very complex proteinaceous mixture with endogenous proteins that span an enormous dynamic range of concentrations—at least 9 orders of magnitude (Adkins et al., 2002). Therefore, direct analysis of biotherapeutic catabolites from plasma/serum directly is nearly impossible. Extraction methods are thus needed to remove endogenous proteins and concentrate the biotherapeutic and cognate catabolites of interest. Methods include protein precipitation, solid-phase extraction, and affinity purification (Ji et al., 2003; Dai et al., 2005; Ackermann and Berna, 2007; Heudi et al., 2008; Ezan et al., 2009; Lu et al., 2009; Liu et al., 2011b; Li et al., 2012). The first two methods are more generally applied to peptides and low-molecular-weight proteins and will not be addressed in detail here. Affinity purification, however, provides a highly selective way to enrich a protein biotherapeutic and its associated catabolites. Selection of an appropriate affinity matrix depends upon the biotherapeutic of interest. Polyclonal Abs against the therapeutic can serve as an appropriate matrix with multiepitope capture. In addition, polyclonal Abs are much easier and less expensive to generate than monoclonal Ab reagents. Alternatively, if human Ab or Ab fragments (e.g., Fc) are part of the biotherapeutic, these can serve as useful catabolically stable “handles” for enrichment. Commercially available protein A or Abs specific for human Ab or Ab fragments can be used in these cases.
Besides enrichment, other sample preparation steps may be appropriate. MS analysis can be greatly helped by reducing the molecular complexity and size of the analyte of interest. For dimeric or higher-order structure biotherapeutics that are held together by disulfide bonds, reduction and alkylation can be beneficial either before or after enrichment. If N-linked glycosylation is present, treatment with glycosidases such as PNGase F can greatly help to reduce the complexity. In other cases, enzymatic digestion with site-specific proteases such as trypsin, chymotrypsin, Lys-C, and Asp-N can release smaller polypeptide fragments where MS analysis is greatly facilitated. For example, the N-terminal catabolism of glucose-dependent insulinotropic polypeptide was monitored by tracking the smaller N-terminal tryptic peptide as opposed to the intact molecule (a difference of 26 amino acids), resulting in a sensitivity improvement of 250-fold (Siskos et al., 2009).
In many cases, enrichment is sufficient for direct analysis by MS. There are cases, however, where orthogonal separations such as LC may be necessary for additional purification. Generally, reverse-phase LC is the method of choice for peptide/protein separation. The chromatography can be done offline, or in line with the mass spectrometer.
Choice of Mass Spectrometer.
Protein biotherapeutic catabolites can be investigated by a number of different types of ion sources [e.g., electrospray ionization, matrix-assisted laser desorption ionization, surface-enhanced laser desorption ionization (SELDI)] and mass analyzers [e.g., quadrupole, time of flight (TOF), ion trap/orbitrap]. Since the parameters for mass spectrometer choice are essentially the same as for general proteomic studies and protein analysis, the details will not be described here, as there are already excellent review articles covering this topic (Mann et al., 2001; Ens and Standing, 2005; Domon and Aebersold, 2006; Ahmed 2008; Tipton et al., 2011). Catabolites and other biotransformed entities can be confirmed by mass differences in the resultant mass spectra and corroborated by gas-phase fragmentation data if necessary.
Criticality of Biotransformation Studies Based on Biotherapeutic Modality
Cytokines, Growth Factors, and Replacement Proteins.
This broad therapeutic class is characterized by biotherapeutics that have a large range of molecular masses and have endogenous counterparts. Marketed examples include epoetin alfa for renal failure, filgrastim for neutropenia, and interferon-β for multiple sclerosis. The circulatory half-lives can vary widely, which can be the result of a combination of both catabolic deactivation and/or inherent clearance via other pathways. For example, if a biotherapeutic has a very short half-life mainly due to rapid clearance of the intact molecule by the kidney or by tissue-mediated drug disposition, then extensive studies regarding possible circulating catabolites may not be necessary. However, it may be difficult to know this a priori, and thus it is challenging to suggest general guidance for the need of biotransformation studies. Several enlightening examples of biotransformation studies in this biotherapeutic class do, however, merit discussion.
The chemokine RANTES (regulated on activation, normal T cell expressed and secreted) plays a role in leukocyte trafficking and homing (Schall et al., 1990). A 68–amino acid mutant of RANTES, [44AANA47]-RANTES, has been shown to inhibit the recruitment of native RANTES and reduce the severity of a murine model of multiple sclerosis (Johnson et al., 2004). Chemokines have been shown to be susceptible to deactivation by proteases in vivo; therefore, studies were performed to analyze the catabolism of [44AANA47]-RANTES (Favre-Kontula et al., 2006). Using immobilized anti-RANTES polyclonal Abs for enrichment followed by interrogation by SELDI TOF-MS, it was shown that [44AANA47]-RANTES quickly formed two major catabolites, the 3-68 and 4-68 forms where the first two or three N-terminal amino acids were lost, respectively. These catabolites are important in that loss of the initial N-terminal residues can significantly alter the biology of this chemokine (Proost et al., 1998). Using the SELDI TOF-MS results, a quantitative MS approach that would be able to easily track the parent and catabolites was developed. An alternative approach would have been to use LBA with reagents that have specificity for the N terminus of the molecule.
Native human parathyroid, hPTH (1-84), has a very complex endogenous variant and catabolic pool with substantial amounts of both N-terminal and C-terminal truncated species (D’amour and Brossard, 2005; Lopez et al., 2010). Both hPTH (1-84) and a truncated variant hPTH (1-34) have been shown to induce bone formation, and hPTH (1-34) has been approved for the treatment of osteoporosis (Neer et al., 2001; Quattrocchi and Kourlas, 2004). One reason for the prevalent use of the shorter analog may be the avoidance of the catabolic production of C-terminal fragments of PTH, which have been shown to antagonize the bone growth effects of hPTH (1-34) (D’amour and Brossard, 2005). Furthermore, C-terminal fragments of hPTH (1-84) can accumulate in renal failure patients, causing PTH resistance, and may potentially cause other bone disease (D’amour and Brossard, 2005). With respect to hPTH (1-34), it has been shown that this short PTH analog is catabolized readily in vitro using rat kidney, liver, and lung homogenates, but these results have not been confirmed in vivo (Liao et al., 2010). Interestingly, no formal in vivo catabolism studies of hPTH (1-34) have been reported. In fact, a LC–tandem MS method for quantification of this analyte has recently been developed, even though the authors assert that there has been no indication of catabolites of hPTH (1-34) that could interfere with the already developed LBA methods (MacNeill et al., 2013).
Monoclonal Antibodies.
More than 30 mAbs, including denosumab for the treatment of osteoporosis and bevacizumab for the treatment of various cancers, have been approved as drugs by the U.S. Food and Drug Administration (Reichert, 2012). Of all of the biotherapeutic modalities, mAbs arguably have the richest wealth of knowledge concerning their PK and in vivo disposition properties (Lobo et al., 2004; Tabrizi et al., 2006; Kuang et al., 2010; Deng et al., 2012). This class of biotherapeutic is composed of human or humanized IgG molecules. An IgG molecule consists of two identical light chains (lc; ∼25 kDa) and two identical heavy chains (hc; ∼50 kDa). Each lc is linked to a hc by a disulfide bond, and the two hc are covalently linked to each other by two or more disulfide bonds. Human IgG hc has four subclasses (IgG1–1gG4), although most biotherapeutic mAbs are of the IgG1 or IgG2 subclass. The human IgG lc has two subclasses (κ and λ).
mAbs are attractive biotherapeutics due to their intrinsically long circulatory half-lives. The long half-life is predominantly dictated by neonatal Fc receptor recycling, which also protects the molecule from lysosomal catabolism (Roopenian and Akilesh, 2007; Suzuki et al., 2010). Due to this protection as well as the inherent stability of the molecule, intermediate catabolism and the presence of mAb fragments is not expected. Therefore, for mAbs, extensive biotransformation studies are not generally warranted. Indeed, there are several studies in the literature that have revealed other types of mAb biotransformation events, such as oxidation, deamidation, glutamate/pyroglutamate conversion, and C-terminal lysine processing (Liu et al., 2009, 2011a; Cai et al., 2011), but very little has been published about mAb biotherapeutic catabolic fragments. One study has reported the presence of mAb fragments by incubation of a fluorescently labeled mAb in whole blood followed by analysis by capillary electrophoresis (Correia, 2010). The exact molecular nature of the fragments was not, however, identified (i.e., if the fragments were actually proteolytic fragments), and the relevance to potential in vivo catabolism is not clear.
Peptide/Protein Fusions with Half-Life Extenders.
A relatively new class of biotherapeutic involves attaching a peptide or small protein with intrinsically high clearance (short half-life) to a large scaffold with a long half-life. The goal is to produce a biotherapeutic with desired pharmacological activity that has sustained circulatory concentration that is dictated by the half-life extender. Examples of half-life extenders are mAbs, fragments of mAbs such as Fc, albumin, transferrin, and PEG (Kontermann, 2011). The half-life extenders are chosen due to their longevity and in vivo stability with respect to biotransformation; however, this stability is not necessarily conferred to the fused pharmacoactive peptide or protein. Examples of this disparity and the impact on protein engineering and bioanalysis have already been discussed for GLP-1/mAb and Fc/TMP (romiplostim) constructs in a previous section of this review. Another recent example of the criticality of understanding biotransformation for this type of biotherapeutic is that of fibroblast growth factor 21 (FGF21) fusions to human Fc (Hager et al., 2013). FGF21 is a promising biotherapeutic for the treatment of type II diabetes. Native FGF21 has a very short half-life and is cleared predominantly by renal excretion after administration (Hager et al., 2013). To attempt to create a longer-acting therapeutic, FGF21 was initially recombinantly fused to the C terminus of human Fc (Hecht et al., 2012). However, by using differential enzyme-linked immunosorbent assay coupled with ligand-binding MS, it was found that the C terminus underwent very fast peripheral catabolism at a proline residue 10 amino acids upstream of the terminus. Since the C terminus of the construct had to remain intact to retain potency, protein engineering efforts were undertaken to stabilize this catabolic liability. After numerous constructs were generated, the construct with this proline mutated to a glycine residue showed retained potency with complete stability against biotransformation in this region (Hecht et al., 2012; Hager et al., 2013). These efforts demonstrated the necessity of understanding biotransformation, using this information to stabilize the molecule for retained potency and decreased clearance, and selecting the proper LBA to track only the bioactive entities during PK profiling.
In contrast, there have been reported studies where proteolytically labile peptides have become protected by fusion to half-life extenders. For example, a 31–amino acid peptide coined DAPD (dual-acting peptide for diabetes) is a hybrid peptide that has both GLP-1 agonist and glucagon antagonist activity (Pan et al., 2006). DAPD is vulnerable to deactivation by proteases including dipeptidyl protease IV. Due to the fast clearance of DAPD, the half-life was extended by conjugation to a high mass branched PEG (43 kDa) via maleimide conjugation through the C-terminal cysteine residue (Claus et al., 2007). By introducing the branched PEG, the in vivo half-life was significantly increased by protecting the DAPD from both catabolism and renal filtration.
In short, since the use of half-life extenders such as Fc, mAb, albumin, or PEG is invoked to increase the in vivo persistence of quickly clearing pharmacoactive peptides and small proteins, it is paramount to confirm that biotransformation does not inadvertently derail this extended half-life strategy.
Antibody-Drug Conjugates.
ADCs represent a novel modality where the understanding of in vivo stability and biotransformation is especially crucial. ADCs are composed of mAbs that have been conjugated with small-molecule toxins or chemotherapeutic agents. The attachment of the small molecule moieties to the Ab is generally through different types of linkers and can be directed nonspecifically (e.g., through lysine ε-amino or endogenous cysteine groups) or specifically through engineered sites (e.g., free thiols or other reactive groups) (Nolting, 2013; Perez et al., 2013; Behrens and Liu, 2014; Tian et al., 2014). The stoichiometry of small-molecule drugs conjugated to each carrier antibody is referred to as the drug-to-antibody ratio, and this ratio is desirably preserved until the ADC reaches its target. Overall, the intended pharmacology of ADCs is to use the Ab portion of the conjugates to deliver the conjugated toxins directly and specifically to drug targets differentially expressed on tumors and reduce toxicity issues of systemic administration of high doses of the chemotherapeutic agent alone. For example, calicheamicins are highly potent cytotoxic agents that have been conjugated to mAbs that target surface targets, such as CD22, CD33, and LeY, that are highly expressed on various types of tumors (Bross et al., 2001; Boghaert et al., 2004). Clearly, it is undesirable if the chemotherapeutic agent deconjugates from the Ab carrier in circulation before it engages the intended antigen target, as this will impact overall potency and more importantly increase potential toxicity due to free toxin.
Numerous groups have studied the in vivo stability of ADCs with respect to toxin attachment. Differential LBA has been suggested as a way to assess ADC stability. In this case, assays are generated with differential specificity to measure conjugated and unconjugated forms of carrier mAbs. For example, in one report, two assays were developed to track total calicheamicin concentration (free or conjugated to an ADC) versus concentration of the mAb carrier only (Hussain et al., 2014). Using these assays, the concentration ratio of total calicheamicin to that of the mAb carrier did not change for the first 6 hours after dosing but declined in a log-linear manner such that ∼50% of the conjugated calicheamicin was deconjugated over the 336-hour PK time course.
In addition, affinity MS methods have been used to understand the loss of toxins from ADCs in vitro and in vivo. This is likely due to the aforementioned argument that MS methods provide molecular-level details more readily than differential LBA methods do, and the MS methods have the added benefit of not requiring procurement or generation of LBA reagents. More explicitly, changes in the drug-to-antibody ratio of any ADC due to in vivo or in vitro loss of the conjugated toxins from the carrier Ab can be easily tracked by MS, whereas this information is difficult to obtain by alternative methods. For example, Shen et al. (2012) described the impact of conjugation site on in vivo stability and potency of an ADC composed of trastuzumab conjugated to maleimide-monomethyl auristatin E (MMAE) through engineered cysteine sites in various places in the Ab. They used affinity LC-MS to monitor the loss of MMAE after incubation of the various ADCs in human plasma, and found that certain sites of attachment led to more stable ADCs. They ultimately concluded that stability was conferred to those sites with less solvent accessibility whereby the mechanism of MMAE deconjugation was abated. Significantly, these in vitro results were also observed in vivo using a mouse model.
Conclusion and Future of Field
Unlike small molecules, the need for studies of biotherapeutic metabolism (i.e., catabolism/biotransformation) has only recently been addressed. Although there has been minimal guidance from regulatory bodies with respect to biotherapeutic biotransformation, some guidelines can be established based upon the literature and recent experiences, some of which have been compiled in this review. For biotherapeutics that contain moieties that are not natural amino acids, studies of catabolites containing these entities may require intense investigation due to potential clinical safety issues. For biotherapeutics composed solely of natural amino acid sequences, biotransformation investigations may be less urgent due to the general lack of toxicity concerns of catabolites. However, for certain modalities, circulating catabolites may have a significant impact on drug potency and clearance. Understanding the degree of biotransformation is crucial to the success of a drug development campaign, especially at the early stages, where proteolytically labile sites can be stabilized through biotherapeutic re-engineering. In addition, bioanalytical assays that are used for PK profiling must be able to detect bioactive catabolites and exclude detection of pharmacologically inactive catabolites to define the most accurate PK exposure.
The current method of choice for analyzing biotherapeutic biotransformation is MS due to its exquisite molecular resolution. Differential LBA can be used to screen for gross catabolic liabilities, although pinpointing of specific vulnerable loci is nearly impossible. In addition, other biotransformation events, such as amino acid modifications, cannot be efficiently probed by LBA.
Of the biotherapeutics that are composed only of natural amino acids, the modality that is most prone to intermediate catabolite formation is pharmacologically active peptides or small proteins fused with stable half-life extenders such as mAb, fragments of mAb (e.g., Fc), transferrin, and albumin. The stability of the half-life extender may not be conferred to the fused peptide/protein. For replacement proteins and cytokines, biotransformation studies may be required if enhanced circulatory stability is required that is not engendered by the native endogenous protein itself or if the nature of circulating catabolites must be known to choose the best quantitative bioanalytical method. The modality with the least need for examination of catabolites is mAb alone; the literature suggests that mAbs do not generally form stable, circulating catabolites.
In vivo assessment of biotransformation is presently the most informative, whereas in vitro assessments, although offering some information, generally do not provide a completely accurate correlation to that which occurs in vivo. The development of in vitro assays that are more predictive of in vivo biotransformation is an unmet need that would help this field immensely. Furthermore, the translation of biotransformation results across animal species and particularly to humans has not been systematically explored. This information is critical for this field, as preclinical studies must be relevant to clinical translation. Future studies should address the extent of translation. In the worst case scenario, if the translation is less than adequate, this would necessitate the need for further refinement of in vivo preclinical models, such as identification and utilization of animal species with comparable proteolytic enzyme profiles to humans, or the development of truly correlative in vitro human models.
Acknowledgments
The author thanks Marc Retter for critical review of this manuscript.
Authorship Contributions
Wrote or contributed to the writing of the manuscript: Hall.
Footnotes
- Received March 31, 2014.
- Accepted June 19, 2014.
Abbreviations
- Ab
- antibody
- ADC
- antibody-drug conjugate
- ADME
- absorption, distribution, metabolism, and excretion
- CXCL12
- C-X-C motif chemokine 12
- DAPD
- dual-acting peptide for diabetes
- Fc
- fragment crystallizable region
- FGF21
- fibroblast growth factor 21
- GLP-1
- glucagon-like peptide-1
- hc
- heavy chain
- hPTH
- human parathyroid hormone
- LBA
- ligand-binding assay
- lc
- light chain
- LC
- liquid chromatography
- mAb
- monoclonal antibody
- MMAE
- maleimide-monomethyl auristatin E
- MS
- mass spectrometry
- PD
- pharmacodynamic
- PEG
- polyethylene glycol
- PK
- pharmacokinetic
- PTH
- parathyroid hormone
- RANTES
- regulated on activation, normal T cell expressed and secreted
- SELDI
- surface-enhanced laser desorption ionization
- SM
- small molecule
- TMP
- thrombopoietin mimetic peptide
- TOF
- time of flight
- Copyright © 2014 by The American Society for Pharmacology and Experimental Therapeutics