Effects of Charge on Antibody Tissue Distribution and

Nov 5, 2010 - Address correspondence to Leslie A. Khawli, Early Development Pharmacokinetics and Pharmacodynamics, Genentech, Inc., South San Francisc...
0 downloads 8 Views 3MB Size
Bioconjugate Chem. 2010, 21, 2153–2163

2153

Effects of Charge on Antibody Tissue Distribution and Pharmacokinetics C. Andrew Boswell,† Devin B. Tesar,‡ Kiran Mukhyala,§ Frank-Peter Theil,† Paul J. Fielder,† and Leslie A. Khawli*,† Department of Pharmacokinetic and Pharmacodynamic Sciences, Department of Antibody Engineering, and Department of Bioinformatics, Genentech Research and Early Development, South San Francisco, California 94080, United States. Received June 8, 2010; Revised Manuscript Received October 3, 2010

Antibody pharmacokinetics and pharmacodynamics are often governed by biological processes such as binding to antigens and other cognate receptors. Emphasis must also be placed, however, on fundamental physicochemical properties that define antibodies as complex macromolecules, including shape, size, hydrophobicity, and charge. Electrostatic interactions between anionic cell membranes and the predominantly positive surface charge of most antibodies can influence blood concentration and tissue disposition kinetics in a manner that is independent of antigen recognition. In this context, the deliberate modification of antibodies by chemical means has been exploited as a valuable preclinical research tool to investigate the relationship between net molecular charge and biological disposition. Findings from these exploratory investigations may be summarized as follows: (I) shifts in isoelectric point of approximately one pI unit or more can produce measurable changes in tissue distribution and kinetics, (II) increases in net positive charge generally result in increased tissue retention and increased blood clearance, and (III) decreases in net positive charge generally result in decreased tissue retention and increased whole body clearance. Understanding electrostatic interactions between antibodies and biological matrices holds relevance in biotechnology, especially with regard to the development of immunoconjugates. The guiding principles and knowledge gained from preclinical evaluation of chemically modified antibodies will be discussed and placed in the context of therapeutic antibodies that are currently marketed or under development, with a particular emphasis on pharmacokinetic and disposition properties.

INTRODUCTION Antibody therapy is an increasingly important therapeutic modality based on identification and pharmacological exploitation of molecular targets associated with inflammatory, oncologic, and autoimmune diseases. Advances in hybridoma technology and molecular biology have enabled the development of murine, chimeric, humanized, and fully human monoclonal antibodies (mAb). As a result, in recent decades several novel therapeutic antibodies have been approved for the treatment of a variety of diseases and have been mass-produced as highly specific and relatively homogeneous reagents (1, 2). One major goal in biotechnology is to improve the clinical utility of these reagents with respect to antigen targeting and potency to encourage effective disease therapy. Achieving this goal depends not only upon a thorough understanding of molecular properties underlying antibody behavior and function, but also upon the development of techniques to manipulate these properties in such a way that specifically enhances properties desired for a particular application without exacerbating undesirable ones. For instance, pharmacodynamic (PD) response is often directly proportional to tissue exposure and, thus, plasma half-life (3). As such, a typical goal in biotherapeutic development is to identify a candidate molecule having desirable pharmacokinetic (PK) properties or, alternatively, to manipulate a molecule’s properties to improve PK properties while preserving antigen recognition. * Address correspondence to Leslie A. Khawli, Early Development Pharmacokinetics and Pharmacodynamics, Genentech, Inc., South San Francisco, CA 94080. Tel. 650-225-6509; Fax 650-742-5234; E-mail: [email protected]. † Department of Pharmacokinetic and Pharmacodynamic Sciences. ‡ Department of Antibody Engineering. § Department of Bioinformatics.

Pharmacokinetic and pharmacodynamic (PKPD) properties of antibodies are governed by both molecule-dependent and species-dependent parameters (4). Biological processes are important determinants in antibody PK; examples include binding to antigens (5) and other cognate receptors (6). Furthermore, an antibody’s valency, shape, size, isoelectric point (pI), dosage, and injection method may also influence its disposition kinetics and clearance (3, 7). Molecular charge is only one of the many parameters that collectively influence antibody PK (4, 8), and its long-established importance (9) will be emphasized in this review. Following a brief overview of antibody structure and function, the results of preclinical studies involving charge-modified antibodies will be summarized. Finally, the physiological implications of electrostatic properties of antibodies will be considered at cellular and tissue levels.

ANTIBODY STRUCTURE AND FUNCTION Immunoglobulins (Igs or antibodies) are the fundamental protein players of both innate and acquired immunity. Immunoglobulins are tetramers consisting of two light chains (LCs, ∼25 kDa each) and two heavy chains (HCs, ∼55 kDa each) arranged in a shape that resembles the letter “Y” (Figure 1a,b). The LCs contain an N-terminal variable domain and a Cterminal constant domain, while the HCs consist of a variable N-terminal domain and three to four C-terminal constant domains. The Fab domain (antigen binding fragment) is composed of the LC and the first two N-terminal domains of the HC. The association of the LC and HC variable domains creates an N-terminal “hypervariable region” which confers the specificity of the antibody by virtue of the interaction of amino acids within the hypervariable region with molecular components (often amino acids) of an antigen. The Fc region (crystallizing fragment) is a dimer of the two or more C-terminal domains of the HC that are not part of the Fab. Whereas the

10.1021/bc100261d  2010 American Chemical Society Published on Web 11/05/2010

2154 Bioconjugate Chem., Vol. 21, No. 12, 2010

Boswell et al.

Figure 1. The four-chain structure of an immunoglobulin. The cartoon representation in (a) highlights the overall structural features of a human IgG, with CDRs depicted in yellow, and the Fab, Fv, and Fc regions indicated using brackets. The image of IgG1 (PDB code 1igy) in (b) was generated using the PyMOL molecular visualization system (106). In both (a) and (b), the two heavy chains are depicted in blue and orange and the two light chains in green and magenta.

Figure 3. Graphical illustration of antibody charge modification. Electrostatic surface images of native IgG1 (PDB code 1igy) (a) and two simulated charge-modified variants (b,c) are displayed using the PyMOL molecular visualization system (106) to illustrate regions of neutral (white), positive (blue), and negative (red) charge. Note that, despite the extreme degree of protein modification in these simulated charge variants, the estimated pI values are roughly within the range of values shown for experimental test molecules listed in Table 2. The variant structures were created from the native protein by manual point mutation, using DeepView v 4.0.1 Swiss PdbViewer (http://www.expasy.org/spdbv/) (107), of all 86 Lys and 38 Arg residues (b) or all 60 Asp and 64 Glu residues (c) to Ala. Electrostatic potential maps were computed with default settings using the PBEQ Solver within CHARMMGUI running from The University of Kansas Department of Molecular Biosciences Center for Bioinformatics Lab (http://www.charmmgui.org/) (108). Estimated pI values were calculated from the corresponding sequences using Scripps Protein Calculator v 3.3 (http:// www.scripps.edu/∼cdputnam/protcalc.html).

Figure 2. Natural/artificial groups contributing to antibody charge. Chemical structures of naturally occurring amino acids that are usually charged at physiological pH values (a) are shown. Also presented are structures of (b) cationic groups, (c) anionic groups, and (d) neutral groups for charge modification with wavy bonds indicating the potential site of protein conjugation. Abbreviations: HMD, hexamethylenediamine; DTPA, diethylene triamine pentaacetic acid; DOTA, 1,4,7,10tetraazacyclododecane-1,4,7,10-tetraacetic acid; PDP, pyridyl disulfide propionate.

Fab is responsible for antigen recognition and binding, the Fc regulates the biological functions of the antibody by binding to Fc receptors and eliciting immune effector functions. Binding to Fc receptors can also mediate transport of antibodies within cells or protect them from proteolytic degradation. Of particular importance for therapeutic Abs, the Fc portion of IgG binds pH-dependently to the neonatal Fc receptor (FcRn) (10) which protects antibodies from degradation and partially accounts for their long serum half-lives (6, 11). In mammals, IgG is the most abundant Ig in serum (∼80% of total serum Ig) and is the principle Ig that participates in the adaptive immune response. There are four human IgG subclasses (IgG1, IgG2, IgG3, and IgG4), which are distinguished from one another by differences in the constant region of the HC sequence. These sequence differences affect the length of the hinge region as well as the position and number of disulfide bonds between the two HCs of the IgG molecule. Additionally,

different IgG subclasses bind to different Fc receptors with varying affinities. Thus, IgGs of one subclass may be more or less efficient at eliciting certain effector functions than IgGs of another subclass. For example, IgG3 is the most effective subclass at activating the complement, followed by IgG1 and IgG2, whereas IgG4 cannot activate the complement. Circulating IgG participates in adaptive immunity by binding to antigens in the circulation. Antigens are cleared through subsequent activation of complement and/or other effector functions such as opsonization (phagocytosis of antigen/pathogen by macrophages and neutrophils) and antibody-dependent cell-mediated cytotoxicity (ADCC).

ANTIBODY CHARGE: NATURAL HETEROGENEITY AND METHODS FOR MODIFICATION Heterogeneity in the physicochemical properties of antibodies is a ubiquitous phenomenon resulting from the very complex, macromolecular nature of immunoglobulins. For instance, the influence of charged amino acids (Figure 2a) within a complex three-dimensional structure explains the presence of charged surfaces that often encourage protein-protein interactions under physiological conditions (12). Illustrated in Figure 3a is a typical variable charge distribution pattern that exists intramolecularly, across different surface regions of a single antibody. Note that, in some antibodies, a sizable degree of polarity may result from the presence of oppositely charged regions within different areas of the macromolecule. Furthermore, significant intermolecular heterogeneity (i.e., polydispersity) exists among different antibody molecules comprising any given preparation. A fundamental property of antibodies and other proteins is the isoelectric point (pI), which is defined as the pH at which the macromolecule carries no net electrical charge. While the approximate

Reviews

Bioconjugate Chem., Vol. 21, No. 12, 2010 2155

Table 1. List of Available Experimental (EXP) and Theoretical (CALC) Isoelectric Point (pI) Values for Selected FDA-Approved Therapeutic Antibodiesa antibody

trade name

type/source

pI(EXP)

muromonab abciximab rituximab basiliximab daclizumab trastuzumab

Orthoclone ReoPro Rituxan Simulect Zenapax Herceptin

rodent chimeric Fab chimeric chimeric humanized humanized

alemtuzumab ibritumomab adalimumab tositumomab omalizumab bevacizumab

Campath Zevalin Humira Bexxar Xolair Avastin

humanized murine human murine humanized humanized

cetuximab

Erbitux

chimeric

na ∼8.3 na na 8.2-9.0 8.6-9.1 ∼8.6 9.2 8.5-9.0 na 8.6-9.1 na na 8.3-8.6 8.2-8.3 na

ref. 99

100 101 28 95 102 103

104 105

pI(CALC) HC

pI(CALC) LC

pI(CALC) full

8.53 7.25 8.58 8.59 8.34 8.49

6.34 4.99 8.61 8.6 8.32 7.76

8.26 6.16 8.65 8.65 8.42 8.41

8.6 7.61b na 8.58 8.21 na

8.81 7.8b na 8.61 5.39 na

8.73 7.83b na 8.65 7.01 na

8.69

6.35

8.44

a

Theoretical pI values were computed by inputting the heavy chain (HC), light chain (LC), or full protein sequence of each antibody (http:// www.drugbank.ca) (97) into the ProtParam tool within the ExPASy Proteomics Server of the Swiss Institute of Bioinformatics (http://www.expasy.ch/ tools/protparam.html) (98). b Charges imparted by the conjugated chelator, tiuxetan, were not factored into the theoretical pI calculations.

surface charge distribution (Figure 3a) and overall pI (Table 1) of an antibody can be predicted from its sequence and structure, charge variation (often in the form of covalent modification of amino acid residues or glycan moieties) is known to spontaneously arise during production and manufacturing. Such alterations vary between different production methods and is of particular relevance in the biotechnology industry (reviewed in refs 13-15). Despite an inherent charge polydispersity that arises spontaneously in production, efforts to isolate charge variants and evaluate them on an individual basis suggest that differences of less than approximately one pI unit are inconsequential for PK (16). As such, methods have been developed for deliberately modifying the net charge of antibodies, often yielding more drastic changes in pI values (Figure 3b,c). The rationale for these efforts is twofold: (I) to permit exploratory preclinical study of the relationships between the net molecular charge and in vivo disposition kinetics and (II) to potentially improve the PK properties of antibodies for the diagnosis or treatment of disease. One promising approach to this endeavor is antibody engineering technology, which can produce changes in pI and biological clearance through substitutions involving charged amino acids (Figure 2a) (17). An alternative approach is the deliberate modification of antibody charge through direct chemical modification of amino acid side chains with positive (Figure 2b), negative (Figure 2c), or neutral (Figure 2d) chemical groups. Chemical modification of antibodies is commonly used to deliver payloads such as radionuclides (18) or toxins (19), but can also allow alteration of net molecular charge. As a point of reference, the amino acid modifications in Figure 3b,c would be equivalent to conjugation of 25 diethylene triamine pentaacetic acid (DTPA) molecules (Figure 2c) through Lys residues (assuming a maximum net -5 charge reduction (20)) or 62 hexamethylenediamine (HMD) molecules (Figure 2b) through Asp/Glu residues (assuming a maximum net +2 charge increase (21)), respectively. Such modifications can also produce significant changes in pI and, in turn, noticeable changes in plasma and tissue PK as demonstrated by numerous examples in the proceeding section.

PK AND TISSUE DISPOSITION OF CHARGE-MODIFIED ANTIBODIES Deliberate chemical modifications of antibodies with a variety of acidic, basic, or neutral chemical entities have been exploited to change the net molecular charge of antibodies (Figure 2b-d) (22). Apart from the chemical nature of the appended functional

group, the site of modification (e.g., tyrosines, lysines, cysteines, oligosaccharides) has also been shown to affect tissue distribution and elimination of modified antibodies (23). Cationization. Progress in molecular biology has yielded novel therapeutic strategies including gene therapy and cellular protein delivery, which rely on the entry of charged macromolecules into intracellular compartments. Cationization of antibodies is a drug delivery strategy wherein the pI is raised, often by conversion of surface carboxyl groups to primary amines or other positively charged groups. A number of cationic groups can be utilized, including HMD, ethylene diamine, and triethylene tetramine (Figure 2b). The resulting modified proteins have demonstrated altered biological properties, with a general trend toward increased plasma clearance and volume of distribution and increased disposition in both target (e.g., tumor) and normal tissues (Table 2) (24-29). Increased volume of distribution due to cationization can be demonstrated visually by whole body radiographic imaging, such as for an indium-111 (111In)-labeled murine antibody against the human epidermal growth factor receptor (EGFR) (Figure 4) (30). A common goal for therapeutic antibodies is to extend plasma half-life as a means to increase exposure, often expressed in terms of the area under the plasma concentration-time curve (6). In contrast, cationization tends to shorten plasma half-life due to an enhancement in both the rate and the magnitude of tissue distribution. However, cationization might prove to be a useful strategy in specific applications in which prolonged antibody exposure may be sacrificed for the sake of rapid, enhanced tissue uptake (e.g., targeting antibodies to efficiently cross the bloodbrain barrier (25)). Cationization of antibodies has also been explored as a means to encourage extravasation, antigen binding, and receptor-mediated endocytosis (31) of antibodies into target cells. In contrast to native Abs, which are generally excluded from cell membranes in the absence of receptor-mediated endocytosis, cationized Abs are better able to reach the intracellular space via absorptive-mediated endocytosis (32, 33). Electrostatic interactions between positively charged proteins and negatively charged cell membranes (34) could permit cell entry via nonspecific membrane flow and have been implicated in the mechanism by which cationized antibodies are rapidly endocytosed by cells in vitro (35). A similar phenomenon is suggested to induce absorptive-mediated transcytosis across microvascular endothelial barriers in vivo (36). Although cationization has been shown to improve endocytosis and target tissue retention, several undesirable effects have also been reported. In general, retention in nontarget tissues is also significantly increased for cationized antibodies relative to

2156 Bioconjugate Chem., Vol. 21, No. 12, 2010

Boswell et al.

Table 2. Summary of Reported Preclinical Studies Involving Charge-Modified Antibodies test molecules/target (species)

modifying group/∆ pI

effect on tissue disposition

effect on blood, plasma, or whole body clearance (CL)†

ref

Cationized Antibodies 125

I-goat colchicine-specific polyclonal IgG (rat) 125 I-AMY33/anti-β-amyloid (BALB/c mice) 111 In-AMY33/anti-β-amyloid (BALB/c mice)

HMD 5.85-9.0 to 8.65-10.3

v (74-fold) in VD

v (114-fold) in systemic CL

24

HMD 7 to 9

v (28-fold) in plasma CL

25

v (3.2-fold) in plasma CL

25

111

In-murine anti-EGFR (SCID mice) I-anti-HIV immune globulin (SCID mice) 125 I-murine mAb D146/ anti-Ras peptide (BALB/c mice) 125 I-humanized 4D5/anti-HER2 (rat) mouse monoclonal IgG, MOPC21 (SCID mice)

HMD (but also DTPA) 8.1-8.3 to 8.7

v in %ID/g (2 h) for liver, kidney, lung, heart, brain v in %ID/g (2 h) for liver, kidney; not in lung, heart, brain v in %ID/g (1 h) for liver, lung

HMD 6.3-7.3 to 8.1- >9.5

v in blood CL (i.e., decrease in AUCSS) v (1600-fold) in blood CL

30

125

v (58-fold) in plasma CL

27

v (23-fold) in plasma CLSS

28

v in plasma CL (i.e., decrease in k, time constant of concentration decay in the plasma: 500-1500 vs 3400-8200 s)

29

mouse monoclonal IgG, MOPC21 (SCID mice)

succinate 6.0 to 3.0-3.9

29

111

In-rabbit IgG/antihuman serum albumin (BALB/c mice) human, nonspecific, polyclonal IgG (rat)

DTPA 9.43-4.65 to 7.9-4.27

v in plasma CL based on decrease in k, time constant of concentration decay in the plasma n/a

111

succinate (as well as DTPA) pI values n/a

HMD (but also DTPA) 7 to (n/a)

v in VD (2 h) for liver, kidney, lung, brain v in VD (2 h) for liver, kidney, lung, heart, brain

HMD 8.5 to >9.5 HMD 8.6 to >9.3

v in VD (2 h) for brain, heart, lung, liver, kidney v in tumor microvascular permeability; Pv (10-7 cm s-1): 4.25-5.27 vs 1.47-4.07

HMD 6.0 to 8.6 - 9.3

26

Anionized Antibodies

In-bovine IgG (ddY mice)

V in tumor microvascular permeability; Pv (10-7 cm s-1): 1.11-2.53 vs 1.47-4.07 V in %ID/g (4 h) for intestine, colon v in %ID/g (4, 24 h) for liver V in %ID/g (up to 24 h) for most tissues

DTPA 6.6-9.3 to 6.0-8.3

v in %ID (1 h) for liver Suc50-IgG > Suc22-IgG > IgG

No ∆ in plasma CL based on terminal half-life (t21/2 (h) ) 20.5 ( 1.9 vs 20.7 ( 1.6) v in total CL: CLtotal (mL/h) at 20 mg/kg dose: 0.46 (Suc22-IgG) vs 2.86 (Suc50-IgG) vs 0.4 (IgG)

51 52 53

Neutralized Antibodies 125

I-chTNT-3/B/antihistone HI and DNA [BALB/c mice (PK); athymic nude mice (distribution)] 125/131 I-B72.3/murine anti-TAG72 IgG1 (BALB/c mice (PK); athymic nude mice (distribution)) 125/131 I-Lym-1/murine IgG2a anti-HLA-Dr (BALB/c mice (PK); athymic nude mice (distribution)) 125/131 I-TNT-1/antihistone HI and DNA (BALB/c mice (PK); athymic nude mice (distribution)) 125 I-TNT-1/antihistone HI and DNA (BALB/c mice (PK); athymic nude mice (distribution)) 125 I-TNT-2/antihistone HI and DNA (BALB/c mice (PK); athymic nude mice (distribution)) 125 I-TNT-3/antihistone HI and DNA (BALB/c mice (PK); athymic nude mice (distribution))

biotin >9.6 to 8.0-9.6

V in %ID/g (72 h) for most tissues

v in whole body CL based on whole body t1/2: 90.0 ( 8.1 vs 134.2 ( 4.0 h

59

SPDP 4.8-5.0 to 4.5-4.7

V in %ID/g (72 h) for most tissues (except tumor)

v in whole body CL based on whole body t1/2: 20 vs 120 h

60

SPDP 7.2-7.8 to 5.1-6.0

V in %ID/g (7 d) for most tissues (except tumor)

v in whole body CL based on whole body t1/2: 60 vs 144 h

60

SPDP 5.8-6.0 to 4.8-5.1

V in %ID/g (72 h) for most tissues

v in whole body CL based on whole body t1/2

60

biotin >9.6 to 7.2

V in %ID/g (72 h) for most tissues

V in whole body CL based on whole body t1/2: 17.4 ( 2.4 vs 30.4 ( 1.8 h

58

biotin >9.6 to 8.2

V in %ID/g (5 d) for most tissues

No ∆ in whole body CL based on whole body t1/2 175.1 ( 3.0 vs 178.7 ( 7.2 h

58

biotin >9.6 to 7.2

V in %ID/g (72 h) for most tissues

v in whole body CL based on whole body t1/2: 90.1 ( 3.8 vs 134.2 ( 4.0 h

58

control antibodies (Figure 4). In addition, cationized human IgGs have been shown to form soluble immune complexes that induce activation of human platelets at much lower concentrations than native, noncationized human IgGs (37). Anionization. Attempts have also been made to produce anionized Abs by chemically converting neutral or positively charged groups into negatively charged groups; this modification typically increases plasma clearance while decreasing tissue uptake and retention. While it may seem counterintuitive that both cationization and anionization increase plasma clearance, different underlying mechanisms are involved: tissue sequestration in the former case, and more rapid whole body clearance in the latter. Various anionic groups may be exploited for this

type of modification, including DTPA (38), 1,4,8,11-tetraazacyclododecane-N,N′,N′′,N′′′-tetraacetic acid (DOTA) (39, 40), and succinic acid (41) (Figure 2c). Similar induction of negative charge is imparted by protein conjugation to 1,4,8,11-tetraazacyclotetradecane-N,N′,N′′,N′′′-tetraacetic acid (TETA) (39, 42) and other polyaminopolycarboxylate chelates that are routinely used in radiopharmaceutical applications (18, 43). Considering that so few macrocyclic chelates are occupied by metal cations in high-specific-activity radiotracer preparations, alteration of antibody pI occurs irrespective of the presence or absence of cationic metal radionuclides. The majority of antibodies possess a more basic isoelectric point (pI > 6) than most serum proteins (pI < 5.5) (44); this is

Reviews

Bioconjugate Chem., Vol. 21, No. 12, 2010 2157

Figure 4. Effects of antibody cationization on tissue distribution. Figure reproduced with permission from Lee and Pardridge, 2003 (30). Whole body autoradiography (left) or gamma counter external detection (right) of female SCID mice implanted with U87 subcutaneous tumors in the right flank (arrows). Mice were injected with 50 µCi of either [111In]-native 528 mAb or [111In]-cationized 528 mAb. At 24 h after injection, the animals were anesthetized for external detection and then sacrificed for whole body autoradiography. The tumors were imaged 15 days after implantation.

particularly true for human IgGs (Table 1) and is a property that is in fact commonly exploited in their purification (45). This also makes them susceptible to ionic interactions with negatively charged glycan chains on cell surfaces (46) and heparin sulfate proteoglycans within the extracellular matrix (ECM) (47-50), collectively promoting higher tissue disposition and retention. Anionization partially abrogates these attractive forces, thereby reducing such non-antigen-dependent tissue uptake. Anodal shifts in antibody pI generally induce faster blood clearance and lower tissue accumulation, with the exception of liver (Table 2) (29, 51-53). It is important to note that the increased hepatic retention of radioactivity may be due in part to transchelation of 111In from the DTPA chelate to transferrin, with subsequent localization in the liver and other organs of the reticuloendothelial system (23). This behavior might be avoided by choosing a different radiometal/chelate pair such as 111In/DOTA, taking advantage of the higher kinetic stability of the metal complex formed by this macrocyclic chelate. It is evident that cautious interpretation is necessary when comparing uptake data obtained using both residualizing (e.g., radiometal chelate-based) and nonresidualizing (radiohalogenated tyrosine-based) probes. Specifically, it is advisible to only compare data obtained using a single labeling method when studying charge-dependent trends in tissue uptake. For instance, when antibodies conjugated to radiometal-chelate probes undergo endocytosis upon binding in antigen-expressing tissues, a greater cumulative tissue uptake of radioactivity is observed due to probe residualization following lysosomal catabolism (54). This effect is not related to molecular charge of the intact antibody; rather, it is driven by the inability of the resulting highly polar, charged radiometal-chelate-amino acid metabolites to cross the cell membrane, in contrast to the diffusible degradation products of antibodies that are directly radioiodinated through tyrosine residues. Radiometal-chelate probes are often preferred for tumor-targeting diagnostic and therapeutic applications for this reason. For instance, an 111In-DTPA-labeled IgG against the carcinoembryonic antigen (CEA) resulted in 41% of the injected dose per gram being localized within human colorectal xenografts at 24 h compared to 9% for a nonbinding 111 In-DTPA-labeled control IgG and 19% for the analogous radioiodinated anti-CEA antibody (55). Such trends induced by probe residualization are most profound in antigen-expressing tumors or in antigen-expressing normal tissues. In contrast, non-

antigen-expressing tissues have demonstrated little or no differences in levels of radioactivity between residualizing (111In) and nonresidualizing (125I) labeling methods (54). Neutralization. Improvements in the biodistribution of antibodies have been sought using another approach in which an uncharged group is appended to a charged amino acid residue (e.g., lysine). A variety of neutral, uncharged groups may be utilized, including biotin and the heterobifunctional linker, pyridyl disulfide propionate (PDP) (Figure 2d). Additionally, PEGylation (PEG ) poly(ethylene glycol)) is a broad modification strategy for therapeutics (reviewed in refs 56, 57). Such modifications will not be further considered here, however, as the large size (1-25 kDa) of PEG groups routinely used in antibody PEGylation often induces simultaneous changes in multiple molecular properties, including conformation, electrostatic binding, and hydrophobicity. Neutralization of positive charge (i.e., net reduction of charge) is much more common than neutralization of negative charge due to facile chemical methodologies which can be used to cap the nucleophilic amino groups of lysines. In some cases, antigen binding may be preserved in chargemodified antibodies while still reducing nonspecific organ uptake. For instance, favorable results were obtained for a series of biotin-modified TNT mAbs, which yielded preparations having 65-69% immunoreactivity in a cell binding assay, more rapid whole body clearance, and higher tumor-to-background ratios due to lower uptake in normal tissues (Figure 5) (58). The masking of positive charges in proteins often produces changes in biological behavior similar to those induced by anionization (Figure 5) (Table 2) (58-60). Once again, such generalizations must be approached with caution, as the nature of the capping group can heavily influence the impact of the modification on antibody disposition. pI/PK Relationships. By analyzing the directional changes indicated by arrows in Table 2, certain overall trends from preclinical evaluation of charge-modified antibodies should become apparent to the reader. In general, antibodies that were chemically modified to achieve higher, more basic pI values (Table 2, Cationized Antibodies) are associated with increased tissue uptake and increased blood clearance. In contrast, modified antibodies having more acidic pI values (Table 2, Anionized/Neutralized Antibodies) displayed decreased tissue uptake and increased whole-body (including blood) clearance. Nevertheless, cautious interpretation of data must be exercised

2158 Bioconjugate Chem., Vol. 21, No. 12, 2010

Figure 5. Effects of neutralization of positive charge in antibodies on tissue distribution. Data adapted with permission from Khawli et al. 2002 (58). Five-day tissue biodistribution and tumor uptake of 125Ilabeled parental (striped) and biotinylated (mAb/biotin ratio ) 1:5, white; 1:10, black) chTNT-2 mAbs in Madison 109 murine lung adenocarcinoma-bearing BALB/c mice.

due to additional consequences of charge modification beyond electrostatic interactions, such as conformational changes in overall antibody structure and steric effects (41). It is also plausible that chemical modification of antibodies may, in some cases, deleteriously impact an antibody’s binding affinity to FcRn (10) which, in turn, could trigger a reduction in serum half-life (61-64). Furthermore, the degree to which these putative changes would arise is highly dependent on the number, bulkiness, and physicochemical properties of small molecule charge modifiers attached to the protein. Another potential unwanted effect associated with antibody modification is increased immunogenicity due to creation of new antigenic determinants (65, 66). Such responses might compromise antibody efficacy by direct neutralization, acceleration of clearance, prevention of further dosing, and induction of serum sickness (67). However, accumulating experimental data shows that incidence rates of clinical immunogenicity for some chemically modified antibodies are quite acceptable. For instance, 4% for a yttrium-90-labeled anionic chelate-modified murine anti-CD20 IgG (ibritumomab tiuxetan, i.e., Zevalin) (68),