Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

The p185^HER2 oncogenic protein is overexpressed in human carcinomas and can be associated with a poor prognosis (Slamon et al., 1987; Giovanella et al., 1991). Murine monoclonal antibodies such as the 4D5 antibody to human p185^HER2 have been generated and inhibit the proliferation of cultured human breast cancer cells that express this oncogenic protein (Shepard et al., 1991). The 4D5 antibody subsequently was "humanized," wherein the murine antibody framework sequences were replaced with human sequences; the humanized 4D5 antibody is a potential therapeutic and imaging agent for breast carcinoma (Carter et al., 1992). Although the p185^HER2 protein is expressed on the plasma membrane of breast cancer cells (Press et al., 1993; Scott et al., 1993), immunocytochemistry of prostate tumors shows that this protein is largely located intracellularly (Giri et al., 1993). Therefore, if an antibody that is directed against the p185^HER2 protein is to form an effective complex with an intracellular target, the antibody must be enabled to undergo endocytosis into the target cell. An additional problem in targeting antibodies for cancer treatment or diagnosis is that the antibody must be enabled to undergo trans-endothelial migration (Jain, 1996). This is because antibodies are confined largely to the intravascular compartment, undergo exodus from the circulating plasma only slowly and have long plasma residence times.

An antibody drug delivery strategy that is intended to facilitate both trans-endothelial migration and target cell endocytosis is antibody cationization (Pardridge et al., 1995). In this strategy, the pI of the antibody is increased by converting surface carboxyl groups of the protein to extended primary amino groups. Cationized homologous proteins have no measurable tissue toxicity and have minimal immunogenicity (Pardridge et al., 1996). In addition, monoclonal antibodies may be cationized with retention of affinity for the target protein (Bickel et al., 1994). Therefore, the present studies examine the feasibility of using this drug delivery strategy for the humanized 4D5 antibody. The effects of cationization on uptake and biological activity of the humanized 4D5 antibody in human SK-BR3 breast carcinoma cells in tissue culture is examined, as is the effect of cationization on the in vivo biodistribution of the antibody in rats.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Materials. The humanized 4D5 antibody, also designated rhu MAb HER2 for recombinant humanized anti-p185^HER2 monoclonal antibody, and the p185^HER2 ECD were provided by Genentech, Inc. (South San Francisco, CA) under the Material Transfer Agreement. The SK-BR3 human breast carcinoma cell line that overexpresses p185^HER2 (Shepard et al., 1991) was obtained from the American Type Culture Collection (Rockville, MD), [¹²⁵I]iodine was obtained from Amersham Corp. (Arlington Heights, IL.), NHS-fluorescein was obtained from Pierce Chemical Co. (Rockford, IL). Male Sprague Dawley rats weighing 320 to 350 g were obtained from Harland Sprague Dawley (Indianapolis, IN). Human IgG and all other chemicals were obtained from Sigma Chemical Co. (St. Louis, MO).

Antibody cationization of rhu MAb HER2. To a 1-ml solution (5 mg/ml) of humanized 4D5 antibody, also designated as rhu MAb HER2, in acetate/saline buffer, was added 2.0 ml of 2 M hexamethylenediamine (pH = 6.8) and 108 µl of a 250 mg/ml solution of fresh N-ethyl-N'-3-(diamethylamino) propyl carbodiimide. The pH was readjusted to 6.8 and the solution was stirred gently for 3 hr at room temperature followed by quenching with the addition of 1 ml of 1 M glycine. The solution was stirred an additional 30 min at room temperature and then dialyzed overnight at 4°C against 4 l of 0.005 M Na₂HPO₄/0.15 M NaCl/pH = 7.4. Tween-20 was added to a final concentration of 0.05%, and the solution was aliquoted and stored at 20°C. SDS-PAGE under reducing and nonreducing conditions and polyacrylamide slab gel IEF were performed as described previously (Pardridge et al., 1995). Human IgG was cationized as described previously (Pardridge et al., 1996), and used as a control.

Antibody iodination. The native, humanized 4D5 antibody (50 µg) was iodinated with lactoperoxidase (0.03 U), [¹²⁵I]iodine (1 mCi) and H₂O₂ (2.4 nmol). The iodination was quenched by the addition of 200 µl stop solution (2 mg/ml tyrosine, 10% glycerol, 0.15 M NaCl, 0.02 M Na₂HPO₄, pH = 7.4) and unbound radioactive iodine was removed on a 0.7 × 28 cm column of Sephadex G-25 in PHSH buffer (0.001 M Na₂HPO₄, 0.5 M NaCl and 0.1% HSA). The final specific activity of the native, humanized 4D5 antibody was 7.1 µCi/µg with a TCA precipitability of 98.5%. The cationized, humanized 4D5 antibody (50 µg) was iodinated with 0.12 U of lactoperoxidase, 2.5 mCi [¹²⁵I]iodine and 44 nmol of H₂O₂. The final specific activity was 5.9 µCi/µg with a TCA precipitability of 96.9%.

IRMA. The affinity of the native or cationized, humanized 4D5 antibody for the p185^HER2 ECD was determined with an IRMA described previously (Bickel et al., 1994). In this assay, 1 µg of HER2 ECD was applied to enzyme-linked immunosorbent assay plate wells in 0.1 M NaHCO₃ (pH = 8.3) at room temperature for 90 min. The wells were aspirated, washed with PBS (0.01 M Na₂HPO₄, 0.15 M NaCl, pH = 7.4) and blocked with 100 µl/well of PBS containing 0.05% Tween-20 (PBST buffer) and 0.1% HSA. To each well was then added 100 µl of PBST buffer containing 0.4 µCi/ml of ¹²⁵I-labeled native 4D5 antibody, and 0.5 to 50 µg/ml concentrations of native 4D5 antibody or cationized human IgG, or 0.5 to 100 µg/ml cationized 4D5 antibody. After incubation for 2.5 hr at room temperature, the solutions were aspirated, the wells were washed with cold PBST buffer (plus 0.1% HSA) and counted for bound radioactivity, which was expressed as a percent of the total radioactivity added per well.

SKBR3 assays. SK-BR-3 human breast adenocarcinoma cells (ATCC HTB 30) were grown in 6-well-cluster dishes in McCoy's 5A medium containing 10% fetal calf serum, 30 mM Hepes buffer (pH = 7.2) in a humidified atmosphere of 95% air and 5% CO₂. The medium also contained penicillin (100 µg/ml), streptomycin sulfate (100 µg/ml) and fungizone (1.25 µg/ml). Cells were plated in 6-well-cluster dishes at a density of 230,000 cells/well or 96-well-cluster dishes at a density of 40,000 cells/well. For the dye uptake proliferation assay (Kern et al., 1993), the cells were incubated in 96-well-cluster dishes (200 µl/well) containing 1, 3, 10 or 30 µg/ml concentrations of native 4D5 antibody, cationized 4D5 antibody or cationized human IgG. Three days later, the medium was aspirated, the cells were washed in PBS, fixed with 180 µl/well of 80% ethanol at 4°C for 30 min and stained with 0.5% crystal violet in 20% methanol for 5 min at room temperature. The excess crystal violet was removed by aspiration, and the wells were washed three times with 180 µl/well of 20% methanol. Dye was eluted from the cells by addition 200 µl/well of 0.1 M sodium citrate (pH = 4.2) in 50% ethanol at room temperature for 60 min, and the absorbance at 540 nm of this solution was measured in a spectrophotometer.

Immunocytochemistry. SK-BR3 cells were grown to near confluency in 35-mm Petri dishes. After removal of the media, washing in PBS, the cells were fixed in 4% paraformaldehyde/0.1 M Na₂HPO₄, pH = 7.4 for 20 min at 4°C. Endogenous peroxidase activity was inactivated with 0.1% H₂O₂ for 5 min at room temperature, the cells were blocked with either 3% horse or goat serum and 10 µg/ml solutions native 4D5 antibody, mouse IgG1 or human IgG was added for 2 hr at room temperature. A biotinylated goat anti-human IgG or a biotinylated horse anti-mouse IgG secondary antibody was added and the immunocytochemical signal was detected with avidin biotin peroxidase immunocytochemistry (Hsu et al., 1981). The cells were not counterstained before light microscopy.

Fluorescein labeling of antibody and confocal microscopy. The native or cationized humanized 4D5 antibody was fluoresceinated with NHS- fluorescein. A 1.0-ml solution (1 mg/ml) of either native or cationized 4D5 antibody was adjusted to pH = 8.3 with 0.05 M NaHCO₃ and 15 µl of 2.36 mg/ml of NHS-fluorescein in 100% dimethyl sulfoxide was added per tube followed by mixing in the dark at room temperature for 90 min. PBS was added to a final volume of 2.0 ml and the volume was reduced to 200 µl with a Centricon 30 concentrator (Amicon Corp., Beverly, MA); this process was repeated two additional times. The final material was brought to a volume of 1.0 ml with PBS and 0.1% HSA and stored in the dark at 20°C until used for confocal microscopy. Approximately five fluorescein molecules were added per native or cationized 4D5 antibody; this was quantitated with NHS-fluorescein as a standard in a Farrand ratio fluorometer with a 7-59 primary filter (300-480 nm) and a 3-69 secondary filter (>510 nm). The affinity of the fluoresceinated/native 4D5 antibody for the HER2 ECD was determined with the IRMA described above.

Pharmacokinetics and in vivo biodistribution. Rats were anesthetized with 100 mg/kg of ketamine and 4 mg/kg of xylazine intraperitoneally, and a femoral vein and femoral artery were cannulated with PE50 cannulas. A 0.2-ml solution of buffered Ringers-Hepes solution (pH = 7.4) containing 7.5 µCi of ¹²⁵I-labeled cationized 4D5 antibody or ¹²⁵I-labeled native 4D5 antibody and 0.1% native rat serum albumin was injected into the femoral vein. An aliquot of arterial plasma was removed at 0.25, 1, 2, 5, 15, 30, 60, 90 and 120 min after injection. Animals were sacrificed at 120 min and brain, liver, kidney, lung and heart were removed, weighed and counted for ¹²⁵I-radioactivity. Aliquots of plasma were also counted for total radioactivity and for radioactivity that was precipitable by 10% TCA.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

The native, humanized 4D5 antibody immunostained the plasma membrane of SK-BR3 human breast carcinoma cells in tissue culture as reported previously (Scott et al., 1993). In addition, many SK-BR3 cells in culture expressed immunoreactive p185^HER2 on a tubular intracellular network that resembled the ER as shown in figure 1. These immunocytochemical reactions were performed with a biotinylated goat anti-human secondary antibody. No immune reaction was detected if a biotinylated horse anti-mouse secondary antibody was used or if the primary antibody was replaced with a mouse IgG1 or human IgG.

View larger version (123K): [in this window] [in a new window]   Fig. 1.   Avidin-biotin immunocytochemistry demonstrating localization of the native humanized 4D5 antibody to the plasma membrane and to the tubular network comprising the ER of SK-BR3 human breast cancer cells in tissue culture. Magnification = 1670×.

The molecular weight of the humanized 4D5 antibody was not altered by cationization as revealed by SDS-PAGE under reducing conditions (fig. 2A). When SDS-PAGE was performed under nonreducing conditions, both the native or cationized 4D5 antibody migrated at 150 kdaltons. The pI of the native 4D5 antibody was approximately 8.6 (fig. 2B). Although the cationized 4D5 antibody comigrated with a pI = 9.3 standard (fig. 2B), this uppermost part of the gel was highly alkaline and direct pH recording of the gel showed the pH to be 12.0. The pH of the gel surrounding the point of migration of the native 4D5 antibody was 8.8. 

View larger version (55K): [in this window] [in a new window]   Fig. 2.   (A) Reducing SDS-PAGE of cationized (cat) humanized 4D5 antibody (lane 1), native (nat) humanized 4D5 antibody (lane 2) and molecular weight standards (lanes 3 and 4). The molecular weight of the standards is shown on the right-hand border. The migration of the IgG heavy (H) and light (L) chains is shown in the left-hand border. (B) Polyacrylamide gel IEF of native humanized 4D5 antibody (lane 1), cationized humanized 4D5 antibody (lane 2) and pI standards (lane 3). The pI of the standards is shown in the right-hand border. The top half centimeter of the gel, which co-migrates with both the pI = 9.3 standard and the cationized 4D5 antibody, has a pH of 12.0, as determined by direct pH measurement of the gel. The direct pH measurement of the region of the gel migrating with the pI = 8.6 standard and the native 4D5 antibody is 8.8.

The ED₅₀ of the binding of the native humanized 4D5 antibody to the p185^HER2 ECD was 1.2 µg/ml as determined with the IRMA (fig. 3). The binding isotherm underwent a right shift with an ED₅₀ of 35 µg/ml for the cationized 4D5 antibody; cationized human IgG caused no reaction in the IRMA (fig. 3), which indicates cationization does not cause binding to HER2 ECD.

View larger version (25K): [in this window] [in a new window]   Fig. 3.   Immunoradiometric assay showing binding of 125I-labeled native humanized 4D5 antibody to p185HER2 ECD in the presence of varying concentrations of unlabeled native 4D5 antibody, cationized 4D5 antibody and cationized human IgG (hIgG). The concentration of antibody that gives 50% inhibition of the binding (ED50) is shown in the figure.

The native 4D5 antibody resulted in 50 to 60% inhibition in SK-BR3 cell division based on results with the dye uptake proliferation assay (fig. 4), whereas the cationized 4D5 antibody inhibited proliferation approximately 40 to 50%. Cationized human IgG showed no inhibition of SK-BR3 cell proliferation (fig. 4).

View larger version (19K): [in this window] [in a new window]   Fig. 4.   The absorbence at 540 nm is plotted versus the concentration of native humanized 4D5 antibody, cationized humanized 4D5 antibody or cationized hIgG added to the medium of cultured SK-BR3 cells for a 72-hr period. After incubation, the cells were fixed, and the uptake of crystal violet by the cells, which is a measure of the total number of cells present in the wells at a given antibody concentration, was measured by spectrometry. Data are mean ± S.E. (n = 3 wells/point).

The ¹²⁵I-labeled native humanized 4D5 antibody was bound avidly by SK-BR3 cells, and virtually all this binding was resistant to a mild acid wash (fig. 5, left panel). Conversely, approximately 50% of the ¹²⁵I-labeled cationized 4D5 antibody that was bound to the cells was resistant to mild acid wash (fig. 5, right panel).

View larger version (21K): [in this window] [in a new window]   Fig. 5.   Binding and uptake (endocytosis) of125I-labeled native humanized 4D5 antibody (left) or125I-labeled cationized humanized 4D5 antibody (right) for a 3-hr incubation of the radiolabeled antibody with SK-BR3 cells at 37°C. Both total binding and acid-resistant binding is shown. Data are mean ± S.E. (n = 3 wells/point). The medium TCA precipitability was 94 ± 1% and 97 ± 1% for the 125I-labeled cationized 4D5 antibody and the 125I-labeled native 4D5 antibody, respectively, at all time points.

Confocal microscopy showed the fluoresceinated, native 4D5 antibody bound to the surface of SK-BR3 cells in tissue culture, and by 90 min of incubation at 37°C, the native antibody was distributed to a compartment of the cell immediately contiguous with the plasma membrane (fig. 6C). In contrast, the fluoresceinated/cationized humanized 4D5 antibody underwent rapid endocytosis into the SK-BR3 cells. At 3 min of incubation at 37°C most of the fluoresceinated/cationized 4D5 antibody was associated with the cell membrane (fig. 6D), but by 30 and 90 min of incubation, virtually all of the fluoresceinated/cationized 4D5 antibody had migrated from the plasma membrane to deep within the cell in the endosomal system. The failure of the fluoresceinated/native 4D5 antibody to undergo significant endocytosis into the cell was not caused by a loss of affinity of the antibody for the p185^HER2 target protein after fluoresceination. As shown with the IRMA (fig. 7), the affinity of the fluoreseinated, native 4D5 antibody for the p185^HER2 ECD was still high with a ED₅₀ of 7 µg/ml.

View larger version (67K): [in this window] [in a new window]   Fig. 6.   Confocal microscopy of SK-BR3 cells incubated with either fluoresceinated/native humanized 4D5 antibody (A, B, C) or fluoresceinated/cationized humanized 4D5 antibody (D, E, F). Cells were preincubated for 60 min at 4°C with fluoresceinated antibody and then incubated at 37°C for 3 min (A, D), 30 min (B, E) or 90 min (C, F). The magnification bar in panel F is 8 µm.

View larger version (18K): [in this window] [in a new window]   Fig. 7.   Immunoradiometric assay showing % binding of125I-labeled native humanized 4D5 antibody to p185HER2 ECD in the presence of varying concentrations of unlabeled native humanized 4D5 antibody (4D5) or fluoresceinated humanized 4D5 antibody (fluoro-4D5).

Cationization markedly altered the pharmacokinetics and in vivo biodistribution of the ¹²⁵I-labeled humanized 4D5 antibody (fig. 8). The native antibody was removed from plasma slowly (fig. 8, left panel), whereas the plasma clearance of the cationized 4D5 antibody, 1.87 ± 0.45 ml/min/kg (table 1), was 23-fold greater than the plasma clearance of the native antibody. The systemic volume of distribution of the cationized 4D5 antibody, 225 ± 36 ml/kg, was 11-fold greater than the V_ss of the native antibody (table 1), which was equal to the plasma volume (V_C) of the animal. The¹²⁵I-labeled cationized 4D5 antibody was taken up by brain, heart, lung, liver and kidney, and the organ V_D of the cationized 4D5 antibody was 2-, 5-, 13-, 18- and 49-fold greater in brain, heart, kidney, lung and liver, respectively, at 2 hr after intravenous injection (table 2). The organ V_D of the native 4D5 antibody in brain, heart, lung or kidney was not significantly different from the plasma volume in these organs, which indicates no trans-endothelial migration of the native 4D5 antibody occurred in any organ except for liver. Conversely, the cationized 4D5 antibody underwent trans-endothelial migration in all organs as indicated by the severalfold enrichment in the organ V_D value relative to the plasma volume for the respective organ (table 2).

View larger version (22K): [in this window] [in a new window]   Fig. 8.   (Left) Plasma concentration, expressed as % of injected dose (ID)/ml plasma, of 125I-labeled native humanized 4D5 antibody or 125I-labeled cationized humanized 4D5 antibody at various times after intravenous injection. Data are mean ± S.E. (n = 3 rats/point). (Right) The % of plasma radioactivity that is precipitable by TCA is shown for the native or cationized 4D5 antibody at each of the respective time points. Mean ± S.E. (n = 3).

View this table: [in this window] [in a new window]   TABLE 2 Organ clearance of 125I-labeled cationized and native humanized 4D5 after intravenous injection in the rat Data are expressed as mean ± S.E. of three rats. Measurements were 120 min after injection. The Vo values used for calculation of PS products are: brain (10 µl/g), heart (150 µl/g), lung (225 µl/g), liver (119 µl/g) and kidney (131 µl/g), as reported previously (Pardridge et al, 1994a).

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

The results of the present studies are consistent with the following conclusions. First, the p185^HER2 oncogenic protein can be localized to the intracellular compartment in many SK-BR3 cells (fig. 1), in addition to the plasma membrane reported previously (Press et al., 1993; Scott et al., 1993). Second, the humanized 4D5 antibody may be cationized to high pI with minimal change in mobility on SDS-PAGE (fig. 2). Third, the affinity of the cationized 4D5 antibody for the target p185^HER2 ECD is reduced, but not abolished, after cationization (fig. 3), and the cationized 4D5 antibody retains biological activity in the cell proliferation assay (fig. 4). Fourth, the native 4D5 antibody is bound by SK-BR3 cells in culture (fig. 5), but confocal microscopy shows that this uptake is restricted to the periplasma membrane compartment of the cell (fig. 6); conversely, the cationized 4D5 antibody undergoes rapid internalization into the cell (fig. 6). Fifth, the pharmacokinetics and in vivo biodistribution of the humanized 4D5 antibody are altered markedly after cationization with a 11-fold increase in systemic volume of distribution, a 23-fold increase in plasma clearance, and a generalized increase in organ uptake (fig. 8, tables 1 and 2).

The intracelluar localization of p185^HER2 shown in figure 1 for cultured SK-BR3 cells is an exception because this protein generally is expressed on the plasma membrane in human breast cancer cells (Press et al., 1993). Conversely, in human prostate gland tumors, immunoreactive p185^HER2 is predominantly an intracellular protein (Giri et al., 1993). A tumor antigen expressed on the plasma membrane is still not accessed easily by a circulating antibody owing to the tumor microvascular barrier. Moreover, the antibody must circumvent two barriers in series (the endothelial barrier and the tumor cell plasma membrane) to target an intracellular antigen. In either an intracellular target or a plasma membrane antigen, antibody cationization enables the antibody to traverse the limiting biological transport barrier, i.e., the capillary endothelium and the tumor cell plasma membrane.

The cationization of the antibody must be performed in a way that largely retains the affinity of the antibody for the target antigen. The ED₅₀ of the binding of the native humanized 4D5 antibody for the p185^HER2 ECD ranges from 1.2 to 1.6 µg/ml (figs. 3 and 7), which is equivalent to 8 to 11 nM. Athough the K_D of the binding of the humanized 4D5 antibody to the HER2 ECD is reported to be as low as 0.1 nM (Carter et al., 1992), the ED₅₀ of binding described here approximates the K_D of 4D5 antibody binding to HER2, 6 nM, originally reported by Shepard et al. (1991). The affinity of the humanized 4D5 antibody for the p185^HER2 ECD is retained after cationization as shown by both the IRMA (fig. 3) and the dye uptake proliferation assay (fig. 4). The affinity of the cationized humanized 4D5 antibody for the p185^HER2 protein may be augmented in future studies by performing cationization with site protection. In this approach, the antibody is cationized in the presence of a molar excess of target antigen, and this previously has increased the affinity of a cationized antibody for the target protein (Bickel et al., 1994; Pardridge et al., 1994b). The retention of antibody binding of target antigen after the cationization modification is consistent with the absence of any structural changes in the antibody that alter mobility on SDS-PAGE (fig. 2A). The size of the cationized light chain, ~30 kdaltons (fig. 2A), is slightly higher than the size of the antibody light chains. However, the mass of the native light chain is of the same size (fig. 2A), which may be related to the "humanization" of the antibody, whereby previous variable sequences are grafted on to a human antibody framework (Carter et al., 1992). The native humanized 4D5 antibody is characterized by an alkaline pI of 8.6 (fig. 2B). However, an alkaline pI is not sufficient to cause absorptive-mediated endocytosis, as shown both in the confocal studies with the native antibody (fig. 6), and in previous studies with a murine monoclonal antibody, which also had an alkaline pI in the native state (Pardridge et al., 1995).

The cationized antibody drug delivery strategy is intended to facilitate antibody transport through two biological barriers: the capillary endothelial barrier and the target cell plasma membrane. The poor penetration of monoclonal antibodies into solid tumors has been attributed generally to the large size of the tumor. However, recent studies show that there is poor antibody penetration and a "binding-site barrier" phenomenon even in small micrometastases (Saga et al., 1995). In micrometasases, the tumor/blood ratio of the antibody at a dose of 2.3 mg/kg is only 0.9 at 72 hr after administration (Saga et al., 1995). The binding site barrier in micrometastases is caused by poor antibody transport across the microvascular endothelial barrier (Jain, 1996). In contrast, cationization facilitates the absorptive-mediated transcytosis of antibodies across microvascular barriers (Triguero et al., 1989, 1990). Similarly, the iodinated native murine 4D5 antibody also distributes to experimental tumors poorly in vivo, because the tumor/blood ratio of the antibody is only 1.6 at 72 hr after administration to beige/nude mice transplanted with p185^HER2 positive experimental tumors (DeSantes et al., 1992). A tumor/blood ratio of iodinated 4D5 antibody of only 1.6 is attributed to both a low tumor "signal" and a high blood "noise" component of the ratio. The high noise component is caused by the high concentration of the iodinated, native 4D5 antibody in blood. This high concentration of labeled antibody in blood arises from the relatively slow egress of native antibodies from the plasma compartment, as exemplified by the slow exodus from blood of the ¹²⁵I-labeled native humanized 4D5 antibody (fig. 8). Conversely, the cationized humanized 4D5 antibody undergoes rapid exodus from the plasma compartment (fig. 8) and achieves significant levels of biodistribution into multiple organs in the rat (table 2). The low "signal" of the iodinated native 4D5 antibody is caused by poor antibody penetration across the endothelial microvascular barrier of the tumor. Although previous studies provide evidence for endocytosis of the native murine 4D5 antibody (DeSantes et al., 1992), the present experiments with confocal microscopy demonstrate that the endocytosis of the native humanized 4D5 antibody into SK-BR3 cells is minimal compared with the rapid endocytosis of the cationized antibody (fig. 6). The minimal endocytosis of the humanized 4D5 antibody (fig. 6) is consistent with recent studies showing the ErbB receptors, other than the epidermal growth factor receptor, are endocytosis-impaired receptors (Baulida et al., 1996). The confocal microscopy results also illustrate the limitations in the use of acid-wash assays as a measure of endocytosis (Tagliabue et al., 1991). Virtually all of the ¹²⁵I-labeled native 4D5 antibody bound to the cell is resistant to mild acid wash (fig. 5), yet the confocal microscopy results show this antibody is confined to the periplasma membrane space (fig. 6). Conversely, the portion of the cationized antibody that is not endocytosed, and only bound to the surface of the cell, is removed by mild acid wash (fig. 5).

In summary, these studies provide evidence that 1) the humanized 4D5 antibody may be cationized with retention of affinity for the target antigen, 2) the cationized antibody undergoes rapid removal from plasma and rapid uptake by organs in vivo and 3) the cationized antibody undergoes rapid endocytosis into target cells as demonstrated by confocal microscopy. Therefore, it is hypothesized that a radio-iodinated, cationized, humanized 4D5 antibody may be a preferred imaging modality for the early and sensitive radiodetection of human carcinoma overexpressing p185^HER2. The signal/noise ratio of the detection modality may be enhanced with the use of cationized antibodies owing to the dual effects of antibody cationization, i.e., 1) increased trans-endothelial migration and target cell endocytosis (i.e., increased signal), and 2) rapid removal from the plasma compartment (i.e., reduced noise). The use of cationized monoclonal antibodies as imaging or therapeutic agents in humans may be limited by the immunogencity of the cationized protein, because cationization increases the immunogenicity of heterologous proteins (Muckerheide et al., 1987). However, when cationized homologous proteins are administered daily in relatively large doses (7.5 mg/kg/day), there is no measurable immune response or tissue toxicity in experimental animals (Pardridge et al., 1995). Therefore, humanized monoclonal antibodies may be preferred agents for cationization.