The Pharmacokinetic Characteristics of Glycolated Humanized Anti-Tac Fabs Are Determined by Their Isoelectric Points

A common problem associated with the i.v. use of low molecular weight biological reagents, such as antibody fragments, peptides, or cytokines, is their rapid clearance and high renal uptake (1, 2, 3, 4, 5, 6, 7). With anticancer agents such as toxins conjugated to small antibody fragments, nephrotoxicity may result (8). With radiolabeled antibody fragments, high renal accumulation can also result in reconstruction artifacts that may adversely affect visualization of lesions at the level of the kidneys during single-photon emission computed tomography. In addition, when small fragments are considered for radioimmunotherapy (9), high renal accumulation adversely affects dosimetry. This is expected to be a more difficult problem when radiometals are used for radiolabeling than when radioiodine is used because radiometals tend to be retained in the tissues once the antibodies are internalized and catabolized (10).

Wochner et al. (11) demonstrated that glomerular filtration is important in the clearance of immunoglobulin fragments by the kidneys. Several groups of researchers have shown that small polypeptides are filtered by the kidneys and absorbed by the proximal tubules (12, 13). Once reabsorbed, these molecules undergo catabolism in proximal tubule cells. The radioiodine from radiolabeled molecules is reabsorbed into the circulation, whereas many of the radiometals are mostly retained intracellularly (3, 10, 14, 15). It has been noted that administration of basic amino acids such as lysine or arginine blocks renal accumulation of small polypeptides by the kidney. This is thought to be due to a charge interaction at the luminal surface (16, 17, 18, 19, 20, 21). Several groups of researchers have, thus, proposed that infusing lysine or lysine-containing amino acid solutions could block renal accumulation of radiolabeled peptides or antibody fragments (3, 6, 15, 17, 22, 23). A small, recent clinical study using commercially available amino acid solutions demonstrated a significant but rather small degree of renal blockade of ^99mTc-Fab’ uptake (24). In an attempt to develop other strategies to block renal uptake, we previously demonstrated that glycolation of disulfide-stabilized Fv fragments decreases their renal accumulation and postulated that this was due to changes in the pIs² of these fragments (25). More recently, we demonstrated that, by glycolating Fab fragments of HuTac monoclonal antibody we significantly decreased its pI and renal accumulation (26). In these studies, the preparations used had a wide range of pIs that overlapped significantly. Other groups have demonstrated that by substituting onto the lysine side chain of IgG or Fab, decreases in pI were obtained that resulted in alteration in biodistribution (27, 28). In the case in which the pI of IgG was modified, this resulted in differences in clearance and tumor accumulation (28). When the pI of an ¹¹¹In-antimyosin Fab was decreased by conjugation with synthetically charged polymers, this resulted in lower renal accumulation in the kidney of dogs when compared to the nonmodified Fab (27).

In this study, we have better characterized our glycolated preparations into discrete pI fractions with little overlap and hypothesize that these fractions exhibit differences in their renal handling based on their pI. Using these different fractions, we evaluated the effect of charge on clearance from the circulation, renal accumulation, catabolism, urinary excretion, and tumor accumulation in nude mice, as compared to the original nonglycolated Fab fragment. Because lysine infusions have been proposed as a method of preventing renal uptake, we also studied whether the combination of anionic glyco-Fab and lysine would further reduce renal uptake of the fragments.

As our model system, we used antibody fragments that recognize IL-2Rα and sIL-2Rα. The Fab fragment was prepared from HuTac IgG (29) by papain digestion. HuTac antibody was constructed, as described previously, by combining the complementarity-determining region of the murine anti-Tac antibody with the human IgG1κ framework and constant regions (29). The HuTac IgG was produced in a continuous-perfusion bioreactor from SP2/0 cells that had been transfected with the genes encoding the heavy and light chains of the hyperchimeric antibody and purified on an IL-2 receptor affinity column. The eluted antibodies were further purified to contain >99% IgG.

The Fab fragment was produced with papain-agarose gel (Pierce Chemical Co., Rockford, IL). Briefly, 0.5 ml of HuTac IgG (∼20 mg/ml) was incubated with papain-agarose gel in PBS with 0.02 mm cysteine at pH 7.0 for 14 h at 37°C. The papain-agarose gel was then removed by filtration. The Fab fragments were purified with Protein A-Sepharose gel (Pierce) followed by dialysis (Slide-A-Lyzer; Pierce) to remove the undigested IgG, Fc fragment, and cysteine. The purity of HuTac Fab was > 99%, as determined by using size-exclusion HPLC with an UV elution profile on a TSK G2000SW column (TosoHaas, Philadelphia, PA; 0.067 m sodium PBS-0.1 m KCl, pH 6.8; 0.5 ml/min).

We synthesized TFP-glycolate as described previously (25): 100 mmol of TFP and 10 mmol of glycolic acid were reacted in 10 ml of acetonitrile using 12 mmol of dicyclohexylcarbodiimide at room temperature for 4 h. The ester derivatives were purified with normal-phase HPLC equipped with a μPorasil PrepPak cartridge (Waters, Milford, MA; hexane:ethyl acetate:acetic acid, 750:250:5; 1 ml/min).

To obtain HuTac Fab with different pIs, three formulations of glycolate were used at 7:1, 20:1, and 70:1 molar ratios of TFP-glycolate to HuTac Fab. The conjugation with the glycolated Fab was performed by reacting HuTac Fab (1.5 mg/ml) with TFP-glycolate dissolved in N,N-dimethylformamide in 0.1 m sodium carbonate buffer (pH 9.5) on ice for 30 min. The products were purified on a PD-10 column (Pharmacia Biotech AB, Uppsala, Sweden) with 0.05 m PBS.

To separate the Fabs by pI, the mixture of the above three preparations was separated using a cation-exchange column TSK-1000SW (TosoHaas, Philadelphia, PA; A, 0.02 m NaH₂PO₄, and B, 0.5 m NaH₂PO₄, pH 4.5; gradient from A to B for 30 min, 0.5 ml/min). We collected two fractions: Fraction 1 was the eluate from 2 to 7 min, whereas fraction 2 was the eluate from 15 to 30 min. Fraction 2 was further separated by an anion-exchange column DEDATSK-5SW (TosoHaas; A, 0.02 m PB, and B, 0.1 m PB with 0.5 m NaCl, pH 7.0; gradient from A to B for 30 min, 0.5 ml/min). Two well-separated peaks, consisting of fraction 2a, eluting from 2 to 7 min, and fraction 2b, eluting from 15 to 30 min, were collected from fraction 2. After separation, each of these three fractions (fractions 1, 2a, and 2b) was concentrated up to 1 mg/ml using a Centricon 30 filter (Amicon Inc., Beverly, MA) and used for this study. The three resultant preparations consisted of fraction 1, sa-glyco-Fab; fraction 2b, ma-glyco-Fab; and fraction 2a, mc-glyco-Fab. Unconjugated Fab was also used in this study.

All four preparations of Fab were labeled with either ¹²⁵I or ¹³¹I by a modified chloramine-T method (30, 31): Purified nonglycolated Fab or glyco-Fab (100 μg) in 0.05 m PB (pH 7.4) and 600 μCi of ¹²⁵I or ¹³¹I were mixed with 12 μg of chloramine-T dissolved in 0.05 m PB. After reacting for 2 min, the radiolabeled product was purified using a PD-10 column (Pharmacia). The specific activity of the preparations ranged from approximately 2 to 9 mCi/mg, and their purity wasg98%, as confirmed by instant TLC and size-exclusion HPLC.

Isoelectric focusing was used to determine the pIs of unconjugated Fab and the three fractions of glyco-Fab. Aliquots containing 1.0 μg of each of the four preparations and standard markers with pIs ranging from 3.5 to 9.3 were separated by isoelectric focusing (PhastGel IEF 3–9 system; Pharmacia). The gels were then stained with Coomassie Blue R 350 (Pharmacia; Fig. 1). In addition to the nonradiolabeled preparations, the four ¹³¹I-labeled Fab (glycolated and nonglycolated) preparations (∼80 kcpm), as well as known pI standards, were separated by isoelectric focusing using the same PhastGel system and autoradiographed with a bioimaging analyzer, BAS-150 (Fuji Medical System USA, Stamford, CT). The pIs of the different radioactivity bands were determined by comparing them with known standards (Fig. 1).

The immunoreactivities of the radiolabeled antibody fragments were determined by a modification of the cell-binding assay of Lindmo et al. (32) and an HPLC method, as follows. Aliquots of the radiolabeled preparations (3 ng/100 μl) were incubated for 2 h at 4°C with 1 × 10⁵ − 4 × 10⁷ SP2/Tac cells. The total cell-bound radioactivity was separated by centrifugation and counted in a gamma counter. Nonspecific binding to the cells was examined under conditions of antibody excess (25 μg of nonradiolabeled anti-Tac antibody). The nonspecific binding was subtracted from the total bound, and data were plotted (32). In addition, the proportion of each preparation that was in complexes between radiolabeled nonglycolated or glyco-Fab and sIL-2Rα (R&D Systems Inc., Minneapolis, MN) was determined. One hundred ng (2 pmol) of ¹²⁵I-labeled nonglycolated Fab or glyco-Fab were incubated with 3 μg (60 pmol) of sIL-2Rα for 15 min at room temperature and analyzed by size-exclusion HPLC, using a TSK G2000SW column equipped with an on-line NaI gamma detector (γ RAM; IN/US Systems, Fairfield, NJ). The HPLC elution profile clearly separated ¹²⁵I-labeled nonglycolated or glyco-Fab from the complexes (data not shown). The fraction of the total activity eluting at a higher molecular weight than that of the ¹²⁵I-labeled Fab was considered to be the immunoreactive fraction.

The binding assay was performed with the IL-2Rα-positive SP2/Tac cell line, a genetically engineered cell line developed by Protein Design Labs (33). This cell line was generated by transfecting SP2/0 cells, a non-immunoglobulin-secreting murine myeloma line (ATCC CRL 1581; American Type Culture Collection, Manassas, VA, with the gene that encodes IL-2Rα. Cells were provided by Dr. Thomas Waldmann of the Metabolism Branch of the National Cancer Institute (Bethesda, MD) and Protein Design Labs (Fremont, CA). Both cell lines were grown in DMEM (Life Technologies, Inc., Grand Island, NY) containing 10% FCS (Life Technologies, Inc.) and 0.03% l-glutamine at 37°C in 5% CO₂.

Tumor xenografts were generated with the ATAC4 cell line, which expresses the IL-2Rα receptor but does not secrete it into the circulation. This cell line was generated by genetically transfecting the plasmid encoding IL-2Rα and a neomycin-resistant gene into A431 cells (34). A431, which was obtained originally from G. Todaro (NIH, Bethesda, MD), is a human epidermoid carcinoma line that does not express IL-2Rα. Both cell lines were grown in RPMI 1640 (Life Technologies, Inc.) containing 10% FCS (Life Technologies, Inc.) and 0.03% l-glutamine at 37°C in 5% CO₂; the medium was occasionally supplemented with 750 mg/ml sulfate geneticin (Life Technologies, Inc.) to avoid growth of nontransformed ATAC4 cells.

Animal studies were performed under an approved Institutional Animal Care and Use Committee protocol. Female athymic nude mice (nu/nu), 5–7 weeks old and weighing 15–20 g, were used (Harlan Sprague Dawley, Frederick, MD). Tumor xenografts were established by s.c. inoculation of 3 × 10⁶ ATAC4 and 5 × 10⁶ A431 cells. Experiments on tumor-bearing mice were performed 10–14 days after implantation, when ATAC4 and A431 tumors both weighed a mean of 0.2 ± 0.1 g.

All mice were euthanized by CO₂ inhalation and exsanguinated by cardiac puncture prior to dissection. The organs were harvested, blot dried, and weighed on an analytical balance. Radioactivity was then counted in a gamma counter, decay corrected, and referred to a known standard of the injected dose.

A stock solution of l-lysine (270 mg/ml) was prepared in 0.1 m PB), pH 7.5, using l-lysine monohydrochloride (Pierce) and 1 n NaOH. This solution was coinjected i.v. and/or injected i.p. with ¹²⁵I-labeled glyco-Fab and ¹³¹I-labeled nonglycolated Fab at final concentrations of 250 and 150 mg/ml to block renal uptake (15).

Nine groups consisting of five normal nude mice each were injected i.v. with 1 μg/3 μCi of ¹²⁵I-labeled sa-glyco-Fab, 1 μg/7 μCi of ¹²⁵I-labeled ma-glyco-Fab, or 1 μg/2 μCi of ¹²⁵I-labeled mc-glyco-Fab. Groups were coinjected with 1 μg/4 μCi of ¹³¹I-labeled nonglycolated Fab. The mice were euthanized 15, 45, or 180 min after injection. Their blood was drawn, and their sera were separated and analyzed by size-exclusion HPLC. The organs were then harvested and radioactivity, together with aliquots of blood, was counted in a gamma counter (Packard Auto-Gamma, Meridien, CT). To determine whole-body retention, carcasses were also counted. The data were expressed both as a %ID/g of tissue and as receptor-positive tumor:normal tissue ratios. Moreover, when the mice were euthanized, their urine was also collected.

The serum and urine samples from the individual mice were analyzed by size-exclusion HPLC using a TSK G2000SW column (TosoHaas; 0.067 m sodium PBS-0.1 m KCl, H 6.8; 0.5 ml/min) equipped with an on-line NaI gamma detector (γ RAM). The fractional distribution of activity present as intact Fab or glyco-Fab (retention times: 18.6 min for intact Fab and 18.5 min for glyco-Fab), as high molecular weight complexes or as catabolites was quantified from the HPLC tracing. To determine the total amount of the %ID/g of blood present as Fab, complexes, or catabolites, we determined the fractional distribution in each compartment from the product of the fraction determined by HPLC and the total radioactivity retained in the blood.

Four groups consisting of five normal nude mice were injected i.v. with both 1 μg/3 μCi of ¹²⁵I-labeled sa-glyco-Fab and 1 μg/4 μCi of ¹³¹I-labeled nonglycolated Fab either in PBS or in PBS containing 50 mg of l-lysine and/or 1 μg of furosemide (Lasix, Biochem Lab, Springfield, IL). These studies were performed with sa-glyco-Fab only because the purpose of this experiment was to evaluate of the effect of increasing urinary flow for washing out the radioactivities from the renal tubules in case of either trapped or untrapped compound in the tubules. The mice were euthanized 15 min after injection, and biodistribution and HPLC analysis of serum and urine were performed as described above.

Five groups, each containing five mice bearing both A431 and ATAC4 tumor xenografts, were injected i.v. with both 1 μg/3 μCi of ¹²⁵I-labeled sa-glyco-Fab and 1 μg/4 μCi of ¹³¹I-labeled non-glycolated Fab. All groups were coinjected with 30 mg of l-lysine, and an additional 30 mg of l-lysine were administered i.p. 15 min later to all mice except the group that was sacrificed at 15 min. The mice were euthanized 15, 45, 120, 360, or 960 min after injection and biodistribution studies and HPLC of serum and urine were performed, as described above. Using the trapezoid method (Sigmaplot; Jandel Scientific, San Rafael, CA), the mean concentrations of radiolabeled preparations in each organ at each sacrifice time were determined and used to calculate the AUC from the time of injection to 960 min. The AUC for tumors was then compared with that of the other organs.

Scintigraphic imaging of nude mice injected with all four ¹³¹I-labeled Fab preparations was performed: four groups of mice were injected i.v. with 100 μCi/11–23 μg of ¹³¹I-labeled sa-glyco-Fab, ma-glyco-Fab, mc-glyco-Fab, or nonglycolated Fab. A gamma camera (Dynamo; Picker International Co., Cleveland, OH) equipped with a pinhole collimator was used to perform scintigraphy at 15, 45, and 180 min after injection. The mice were anesthetized with an i.p. injection of 0.6 mg of ketamine (Fort Dodge Lab. Inc., Fort Dodge, IA) and 0.1 mg of Rompun (Miles Inc., Shawnee Mission, KS) per mouse. Imaging data were recorded in a 64x 64 byte mode matrix using a 20% window centered over the 364-keV photopeak of ¹³¹I. Images were acquired for 45,000 counts.

Nude mice bearing ATAC4 and A431 were injected i.v. with 150 μCi/30 μg of ¹³¹I-labeled sa-glyco-Fab or 150 μCi/20 μg of ¹³¹I-labeled nonglycolated Fab. To block renal uptake, we coinfused with 30 mg of l-lysine followed by another with 30 mg of l-lysine i.p. 15 min after the ¹³¹I-labeled Fab was injected. Scintigraphy was performed 16 h after the antibody fragments were injected.

We performed the statistical analysis using the t test or 2the one-way ANOVA, with pairwise comparison using the Bonferroni method (Sigmastat; Jandel Scientific).

The pIs of nonradiolabeled or ¹²⁵I-labeled sa-glyco-Fab, ma-glyco-Fab, mc-glyco-Fab, and nonglycolated Fab were <4.5, 4.5–7, 7–9.3, and >9.3, respectively, as determined by both the Coomassie Blue staining method and autoradiography (Fig. 1). Peak retention times of ¹²⁵I-labeled sa-glyco-Fab, ma-glyco-Fab, mc-glyco-Fab, and nonglycolated Fab and ¹³¹I-labeled nonglycolated Fab on size-exclusion HPLC were similar and ranged from 18.5 to 18.9 min.

The immunoreactivities of all preparations were in the same range (Table 1).

¹²⁵I-labeled sa-glyco-Fab and ¹²⁵I-labeled ma-glyco-Fab showed similar distributions in the blood (P =0.7, 0.13, and 0.08 at 15, 45, and 180 min, respectively; Fig. 2) and all organs except the kidney (P < 0.01). In addition, ¹²⁵I-labeled mc-glyco-Fab and ¹³¹I-labeled nonglycolated Fab had similar distributions in the blood and all organs except the kidney (P < 0.01) at 45 and 180 min after injection. However, both ¹²⁵I-labeled sa-glyco-Fab and ¹²⁵I-labeled ma-glyco-Fab showed significantly different biodistributions when compared with both ¹²⁵I-labeled mc-glyco-Fab and ¹³¹I-labeled nonglycolated Fab in the blood, liver, lung, and spleen, especially 15 and 45 min after injection (P < 0.01). Early accumulation of radioiodine in the kidney increased with increasing pI, from 24 ± 2%ID/g for ¹²⁵I-labeled sa-glyco-Fab to 44 ± 6%ID/g for ¹²⁵I-labeled ma-glyco-Fab, 136 ± 11%ID/g for ¹²⁵I-labeled mc-glyco-Fab, and 191 ± 19%ID/g for ¹³¹I-labeled nonglycolated Fab 15 min after i.v. injection. These significant differences (P < 0.001) persisted until the 45-min point (Fig. 2).

Whole-body retention of radioactivity 15 min after injection was greatest for the group receiving ¹³¹I-labeled nonglycolated Fab (99 ± 4%ID/g) and was smallest in the group receiving ¹²⁵I-labeled sa-glyco-Fab (87 ± 3%ID/g; P < 0.01).

At all of the time points, HPLC demonstrated that most of the radioactivity in the serum was found in the intact Fab fraction for both ¹²⁵I-labeled anionic glyco-Fabs and that, at each time point, there were no significant differences in the amounts present as intact ¹²⁵I-labeled sa-glyco-Fab or ¹²⁵I-labeled ma-glyco-Fab (P > 0.13). The proportion present as catabolites of ¹²⁵I-labeled mc-glyco-Fab and ¹³¹I-labeled nonglycolated Fab was significantly larger than that of both ¹²⁵I-labeled sa-glyco-Fab and ¹²⁵I-labeled ma-glyco-Fab at all time points (P < 0.001). In cases of ¹²⁵I-labeled mc-glyco-Fab and ¹³¹I-labeled nonglycolated Fab, more than half of the radioactivity in the serum was found in the catabolite fraction at 45 and 180 min (Fig. 3). The amount of ¹²⁵I-labeled mc-glyco-Fab in the intact fraction was significantly larger than that of ¹³¹I-labeled nonglycolated Fab at 15 and 45 min (P < 0.001). Moreover, as the pI increased, the amount of catabolites increased significantly at 45 min (P < 0.001).

In mice receiving anionic glyco-Fabs, compared with mice receiving the cationic glyco-Fab, a greater proportion of the radioactivity was excreted as intact Fab at all time points (P < 0.001). The amount as catabolites in both ¹²⁵I-labeled sa-glyco-Fab and ¹²⁵I-labeled ma-glyco-Fab increased with time, but it increased more significantly in the ma-glyco-Fab group than in the sa-glyco-Fab group at 45 and 180 min (P < 0.001; Fig. 4).

Coinjecting 50 mg of l-lysine with the ¹²⁵I-labeled sa-glyco-Fab significantly reduced the renal accumulation at 15 min. Coinjection decreased the kidney accumulation from 25 ± 2%ID/g to 11 ± 1%ID/g at 15 min after injection (P < 0.001; Table 2). Addition of furosemide did not affect the biodistribution of glyco-Fab.

In the case of ¹³¹I-labeled nonglycolated Fab, lysine also blocked renal uptake, decreasing kidney accumulation at 15 min from 196 ± 19%ID/g to 25 ± 2%ID/g (Table 2). Furosemide added a small but significant effect in the lysine group (P < 0.001). In this group, coinjection of furosemide decreased the kidney accumulation at 15 min from 25 ± 2%ID/g to 18 ± 1%ID/g. Furosemide alone had a small effect on the biodistribution of nonglycolated Fab.

Whole-body retention of radioactivity at 15 min after injection was greatest for the group receiving ¹³¹I-labeled nonglycolated Fab with furosemide (101 ± 3%ID/g) and smallest in the group receiving ¹³¹I-labeled nonglycolated Fab with l-lysine and furosemide (51 ± 2%ID/g; Table 2).

Tumor-bearing mice coinjected with ¹²⁵I-labeled sa-glyco-Fab, ¹³¹I-labeled nonglycolated Fab, and lysine showed significantly lower renal accumulation of ¹²⁵I-labeled sa-glyco-Fab than of ¹³¹I-labeled nonglycolated Fab at 15, 45, and 120 min (P < 0.01) but significantly higher accumulation at 360 and 960 min (P < 0.01; Table 3). In the ATAC4 tumor, uptake of ¹²⁵I-labeled sa-glyco-Fab was significantly higher than that of ¹³¹I-labeled nonglycolated Fab at all time points (P < 0.01), the maximum ratio being 8.4 at 960 min after injection (Table 3). Although higher blood retention of the sa-glyco-Fab was also seen at all time points, at 960 min after injection, all of the ATAC4 tumor:normal tissue ratios or tumor:control tumor ratios of sa-glyco-Fab were significantly higher than those of the nonglycolated Fab (P < 0.01).

In the case of ¹²⁵I-labeled sa-glyco-Fab, the AUC of ATAC4 tumor (9664%ID·min/g) was >1.8-fold greater than that of blood (5227%ID·min/6) and 2.5-fold greater than that of kidney (3860%ID·min/g). The corresponding ratio for ¹³¹I-labeled nonglycolated Fab was 1.6 for blood (ATAC4 tumor: blood =2068: 1254%ID·min/g) and 0.4 for kidney (5237%ID·min/g; Table 3).

At all time points, HPLC analysis of serum from tumor-bearing mice coinjected with lysine demonstrated that, in the case of ¹²⁵I-labeled sa-glyco-Fab, most of the radioactivity was found in the intact Fab fraction. However, in the case of ¹³¹I-labeled nonglycolated Fab, more than half of the radioactivity in the serum was found in the catabolite fraction at 120 and 360 min (Fig. 5).

Urine showed a similar pattern to that of serum. In mice receiving sa-glyco-Fab rather than nonglycolated Fab, a greater proportion of the excreted radioactivity was in the form of intact Fab at all time points (P < 0.001). The fraction of the activity present as catabolites in both the sa-glyco-Fab and nonglycolated Fab groups increased with time but increased much more in the nonglycolated group (P < 0.001; Fig. 6).

In the presence of lysine, whole-body retention of ¹³¹I-labeled nonglycolated Fab was significantly less than that of ¹²⁵I-labeled sa-glyco-Fab at all time points (P < 0.001).

Serial imaging of normal nude mice was performed at 15, 45, and 180 min to evaluate the biodistribution of ¹³¹I-labeled Fabs. ¹³¹I-labeled sa-glyco-Fab and ¹³¹I-labeled ma-glyco-Fab showed higher blood pool at 15 and 45 min after injection than ¹³¹I-labeled mc-glyco-Fab and the ¹³¹I-labeled nonglycolated preparations. The sa-glyco-Fab mice did not show kidney imaging, whereas faint uptake was seen with the ma-glyco-Fab at all time points. By contrast, the ¹³¹I-labeled mc-glyco-Fab and ¹³¹I-labeled nonglycolated Fab showed high kidney uptake at all time points. In these mice, stomach uptake was also seen at 180 min, presumably because of the large number of catabolites (Fig. 7). The stomach was not visualized in mice receiving ¹³¹I-labeled sa-glyco-Fab or ¹³¹I-labeled ma-glyco-Fab.

Scintigraphy of the tumor-bearing mice injected with ¹³¹I-labeled sa-glyco-Fab clearly showed the ATAC4 tumor, but the scintigraphy of the tumor-bearing mice injected with ¹³¹I-labeled nonglycolated Fab showed low whole-body counts, high thyroid uptake, and no obvious ATAC4 tumor uptake (Fig. 8).

The handling of low molecular weight proteins (M_r <70,000) by the kidney is a complex process. Renal uptake of low molecular weight proteins depends on glomerular filtration, which, in turn, depends mostly on molecular size (35). In addition, charge is important in determining renal filtration (36, 37, 38). Once filtered, low molecular weight proteins generally undergo endocytosis in the proximal tubules. This process also depends on charge. Once internalized, these proteins are degraded by lysosomes (12). Therefore, modulation of renal handling of low molecular weight proteins such as Fab fragments may be affected by changing their size and/or charge, thereby altering their glomerular filtration and/or reabsorption by the proximal tubules.

In a previous study, we demonstrated that glycolation of Fab resulted in a decrease in renal accumulation related to the degree of glycolation (26). In that study, however, the pIs of the two glycolated preparations were broadly distributed and significantly overlapped. In this study, we confirm that glycolation of Fab results in decreased renal accumulation and demonstrate that this decrease is caused both by a decrease in renal filtration (evidenced by higher blood retention of the sa-glyco-Fab) as well as by a decrease in retention of the filtered fraction. This study shows that the decrease in renal filtration is related to changes in pI because anionically glyco-Fabs showed a slow clearance from the blood compared with cationically glyco-Fabs and nonglycolated Fabs, which showed fast clearance.

In this study, reabsorption into the proximal tubules decreased with both the sa- and ma-glyco-Fabs, as demonstrated by lower renal uptake and greater appearance of intact Fab in urine. By contrast, total renal accumulation showed charge-dependent differences, with sa-glyco-Fab having significantly less renal accumulation than ma-glyco-Fab (P < 0.01) and the fraction appearing as intact Fab in the urine being much larger for sa-glyco-Fab than for ma-glyco-Fab. In the case of cationic Fabs, renal accumulation of mc-glyco-Fab was significantly less than that of nonglycolated Fab (P < 0.01). Likewise, the fraction appearing as intact Fab in the urine was larger for the mc-glyco-Fab than for nonglycolated Fab. This shows that Fab reabsorption at the proximal tubules increases as Fab pI increases with glycolation (Fig. 9).

Because the size differences were small, we believe it is unlikely that a change in size caused by glycolation was responsible for the differences seen in blood retention and renal accumulation. There are 28 lysine molecules in HuTac Fab (29). If all 28 lysines are conjugated with glycolate [CH₂(OH)COOH; 58 D/1 glycolation], the maximum size increase is only 1624 (3.2% of Fab molecular weight). Using an isoelectric focusing gel, we showed that each of the three glycolated fractions contained five to seven different pI bands. Because each band is likely to represent a single additional glycolate substitution, the average size differences between neighboring fractions will be below 400.

This study again confirms that the renal accumulation of radiolabeled Fab in the kidneys can be considerably decreased by using lysine (6, 22, 23). Although lysine alone was very effective, the combined use of lysine and glyco-Fab further contributed to lower renal uptake as proposed in Fig. 9. Adding furosemide had a small effect on renal accumulation, with the exception of the group receiving nonglycolated Fab coinjected with lysine.

The higher tumor accumulation of the sa-glyco-Fab is most likely related to its more prolonged residence time (AUC) in the blood, which provides greater availability for the tumor. When the mean value of the activity retained in the blood was integrated from the time of injection to 960 min, the sa-glyco-Fab had 4.1 times higher activity in the blood than nonglycolated Fab. Similarly, when the activity in the tumor was integrated over time, the sa-glyco-Fab had 4.7 times more activity than the nonglycolated Fab. Although blood activity increased with the anionic Fabs, this higher background activity did not adversely affect the tumor:nontumor organ ratios. Therefore, as expected, high availability of sa-glyco-Fab resulted in more favorable visualization of tumor in our imaging studies.

Rather than chemical modification, a possible strategy for the future is to genetically engineer proteins by substituting or adding amino acids in such a way as to reduce their pI.

In conclusion, our data suggest that glomerular filtration of Fab depends grossly on its charge and that the proportion reabsorbed into the proximal tubules increases as pI increases. The use of anionic fragments alone or in combination with lysine can decrease their renal accumulation and increase tumor uptake. This should prove to be a promising approach that may be applicable to other small fragments with M_r 40,000 to 70,000.

We thank Dr. Ira Pastan for providing ATAC4 and A431 cell lines and Sue Kendall for her editorial assistance.