Role of Biotin-Binding Affinity in Streptavidin-Based Pretargeted

Lymphoma-bearing nude mice pretargeted with 1F5 Antibody-SA-Wild Type (WT) bioconjugates produced 125I-bis-biotin tumor concentrations of 2.2%ID/g and...
1 downloads 0 Views 458KB Size
Download PDF
Bioconjugate Chem. 2005, 16, 131−138

131

Role of Biotin-Binding Affinity in Streptavidin-Based Pretargeted Radioimmunotherapy of LymphomaX Kevin J. Hamblett,†,⊥ Oliver W. Press,§,| Damon L. Meyer,§,⊥ Don K. Hamlin,‡ Don Axworthy,§ D. Scott Wilbur,‡ and Patrick S. Stayton*,† Departments of Bioengineering, Medicine, and Radiation Oncology, University of Washington, Seattle, Washington 98195, and The Fred Hutchinson Cancer Research Center, Seattle, Washington 98109-1024. Received March 30, 2003; Revised Manuscript Received June 23, 2004 One pretargeting approach to cancer radioimmunotherapy utilizes an antibody-streptavidin conjugate that is first localized to the tumor. A “clearing agent” is then administered to remove the excess bioconjugate from blood, followed by injection of the radiolabeled biotin therapeutic. In this study, the role of streptavidin-biotin affinity in this pretargeting system was investigated for the first time in vivo, with a reduced affinity, site-directed streptavidin mutant and with radiolabeled bis-biotin reagents. The S45A streptavidin mutant (SA-S45A), which displays a faster off-rate for biotin, was utilized with a bivalent biotin carrier that retains high avidity for the streptavidin mutant. Mice were fed either a normal or biotin-deficient diet, yielding serum endogenous biotin concentrations of 31 nM and 5 nM, respectively. Lymphoma-bearing nude mice pretargeted with 1F5 Antibody-SAWild Type (WT) bioconjugates produced 125I-bis-biotin tumor concentrations of 2.2%ID/g and 7.0%ID/g in mice fed normal diets vs biotin-deficient diets. 125I-bis-biotin tumor concentrations of mice pretargeted with 1F5-SA-S45A were 12%ID/g and 10%ID/g for mice fed normal and biotin-deficient diets, respectively. However, poor clearance of the 1F5-SA-S45A with the biotinylated clearing agent led to high normal organ concentrations of 125I-bis-biotin. A galactosylated human serum albumin (HSA) modified with bis-biotin was then tested, and normal organ 125I-bis-biotin concentrations were significantly reduced. Tumor-to-organ ratios achieved for 1F5-SA-S45A with the HSA-bis-biotin clearing agent in mice with high serum biotin were similar to those achieved with 1F5-SA-WT in mice with low serum biotin. These results demonstrate that exchange of bound endogenous biotin with lower affinity streptavidin mutants is possible, and that corresponding use of bis-biotin carriers can nearly eliminate the differences in therapeutic radioactivity at the tumor site in animals on normal vs biotindeficient diets. The results also interestingly demonstrate, however, that improved clearance agents capable of removing the lower affinity streptavidin-antibody conjugate are needed to achieve comparable specificity in tumor to blood or normal organ ratios.

INTRODUCTION

Anti-CD20 antibody therapies have yielded high response rates in patients with B cell lymphomas (1). However, only 6% to 20% of patients achieve complete remissions, and few are cured using the antibody as a X Part of the Special Issue collection for Imaging Chemistry that began in issue 6, 2004. A preliminary description of this work was presented at the Symposium on Chemistry and Biological Applications of Imaging Agents and Molecular Beacons, at the spring 2004 National Meeting of the American Chemical Society. * To whom correspondence should be addressed. Phone: 206685-8148. Fax: 206-685-8256. E-mail: [email protected]. † Department of Bioengineering. | Department of Medicine. ‡ Department of Radiation Oncology. § The Fred Hutchinson Cancer Research Center. ⊥ Present address: Seattle Genetics, Inc., Bothell, WA 98026. 1 Abbreviations: Ab-SA, monoclonal antibody-streptavidin conjugate: BSA, bovine serum albumin; CA, clearing agent; HSA, human serum albumin; NAGB, N-acetylgalactosaminebiotin; RIT, radioimmunotherapy; rt, room temperature; SA, streptavidin; SA-S45A, streptavidin mutant having the serine at residue 45 replaced with an alanine, SA-WT, wild type streptavidin; 1F5-SA-WT, monoclonal antibody 1F5 conjugated to wild-type streptavidin; 1F5-S45A, monoclonal antibody 1F5 conjugated to the streptavidin mutant S45A; cpm, counts per minute; %ID/g, percent of injected dose per gram.

single agent. The therapeutic effectiveness of anti-CD20 antibodies can be enhanced by conjugating to radionuclides. This has been demonstrated in a recent randomized clinical trial that compared 90yttrium-labeled ibritumomab tiuxetan (Zevalin), an anti-CD20 antibody, to rituximab (2). The radiolabeled anti-CD20 antibody achieved an overall response rate of 80% and complete remissions in 30% of lymphoma patients, compared with an overall response rate of 56% (p ) 0.002) and a complete response rate of 16% (p ) 0.04) with the nonradioactive anti-CD20 antibody (2). To further improve the response rates, myeloablative doses of radioactivity can be administered to further improve the response rates; however, this strategy subsequently requires stem cell transplantation (3). A potential strategy for delivering higher levels of radioactivity to tumors without requiring myeloablative doses is multistep radioimmunotherapy (RIT) with pretargeting. One pretargeting approach consists of an antibody-streptavidin (Ab-SA) conjugate, a “clearing agent” to remove Ab-SA conjugate from the blood, and the radiolabeled biotin therapeutic. Clearing the excess Ab-SA conjugate from the blood has two significant advantages. First, the Ab-SA conjugate in the blood is quickly eliminated through the liver, limiting the nonspecific radiation exposure due to the presence of excess Ab-SA in the blood. Second, higher concentrations of AbSA can be injected because this nonspecific exposure is

10.1021/bc034049g CCC: $30.25 © 2005 American Chemical Society Published on Web 12/16/2004

132 Bioconjugate Chem., Vol. 16, No. 1, 2005 Chart 1. Structures for Bis-biotin-NCS and

Hamblett et al. 125I-Bis-biotin

reduced, leading to higher concentration gradients and better tumor penetration of the antibody-streptavidin complex. Pretargeting has been shown to be an effective strategy for directing radioactivity to tumor xenografts in mice (4-6). Although SA-biotin pretargeting has demonstrated promise, the immunogenicity of SA and the presence of competing endogenous biotin represent potential limitations (7, 8). Endogenous biotin binds and effectively blocks biotin-binding sites of SA due to its slow dissociation rate (T1/2 ) 6.5 h) (9). Mice are generally fed a biotin-deficient diet for several days prior to pretargeted radioimmunotherapy to reduce endogenous biotin concentration in serum (7, 10). The extent of the endogenous biotin problem in humans is less clear, but patient dietary restrictions are difficult at best and even low levels of endogenous biotin can block available biotinbinding sites of streptavidin (4, 8). Serum biotin levels also do not reveal the entire extent of endogenous biotin, as the liver has been suggested as a storage/release depot for biotin (11), and biotin levels in colon adenocarcinoma were found to be 3-fold higher than the surrounding tissue (12). In this study we have investigated the role of the high affinity streptavidin binding in the pretargeting system. The high affinity of streptavidin for biotin provided the initial rationale for using this binding pair in pretargeting, but it conversely leads to the limitations associated with blocking by endogenous biotin. We have recently shown that SA-mutants can be used in combination with a bivalent biotin derivative to allow exchange of biotin, while retaining the high affinity with the bivalent biotin ligand in an in vitro binding system (13). Here we have characterized a site-directed mutant of SA in an antiCD20 pretargeting system that exhibits lower affinity and a faster biotin off-rate than wild-type SA (SA-WT) (9, 14-16). We report here an in vivo investigation using a mouse lymphoma xenograft system to determine whether this modified streptavidin-biotin system could circumvent blocking of biotin-binding sites by endogenous biotin while maintaining blood clearance capabilities and tumor targeting specificity. The study provided interesting insight into the tradeoff of high affinity in relation

to endogenous ligand competition versus clearing agent effectiveness. EXPERIMENTAL PROCEDURES

Reagent Preparation. SA-WT and SA-S45A were expressed, refolded, and purified as previously described (9, 15, 16). Conjugation of intact 1F5, an anti-CD20 antibody, to SA was performed as previously described (5, 17). Briefly, Ab was reduced using DTT and SA was conjugated with SMCC in two separate reaction vials. Reduced Ab and maleimide-modified SA were reacted and then purified in two steps using iminobiotin and cation exchange columns. The Ab-SA conjugates were analyzed by HPLC and found to be >90% of the preparations had antibody:SA ratios of 1:1. 125I-bis-biotin (Chart 1) was prepared and radioiodinated as previously described (Chart 1) (13). To prepare the HSA-bis-biotin clearing agent, 0.3 mL of a 20 mg/mL in 50 mM sodium bicarbonate, pH 8.5, solution of HSA (6 mg, 0.09 µmol) was added to 0.15 mL of a 9.5 mg/mL solution of bis-biotin-NCS (1.4 mg, 1.5 µmol), shown in Chart 1, in the same buffer. The reaction mixture was slowly tumbled at room-temperature overnight. The entire mixture was then passed over a Sephadex G-25 column (PD-10, Phamacia, Piscataway, NJ) and collected in 0.9% saline. The protein containing fractions were combined, concentrated in a Centricon30 (Millipore, Bedford, MA), and washed three times with 0.9% saline, concentrating between each wash. To the final volume of 0.25 mL was added 0.75 mL of 0.5 M sodium borate, pH 8.0, followed by 3 mg (9.6 0.09 µmol) of solid β-galactopyranosylphenyl isothiocyanate (Sigma, St. Louis, MO). The mixture was then allowed to tumble at room-temperature overnight before passing over a Sephadex G-25 column. The protein containing fractions were combined, concentrated, and washed three times with 0.9% saline, concentrating between washes. Effect of Diet on Serum Biotin Levels. To measure the levels of biotin in serum, a competition assay was performed as previously described (18). Bovine serum albumin (BSA) mixed with biotinylated-BSA (100:1) at

Bioconjugate Chem., Vol. 16, No. 1, 2005 133

Streptavidin−Biotin Affinity in Pretargeting Therapy

a total protein concentration of 10 mg/mL was added to a 96-well plate at a total protein concentration of 10 mg/ mL and allowed to incubate at room temperature for 1 h. The wells were washed three times with 10 mM Tris, 150 mM NaCl, 0.05% Tween 20 at pH ) 7.5, referred to as TNTw buffer. Biotin standards or samples were added to the wells, followed by the addition of SA-HRP (1.25 mg/mL) (Zymed Laboratories, San Francisco, CA) at a 1:5000 dilution mixed with BSA at 1 mg/mL, and the mixtures were incubated at room temperature for 30 min. Wells were washed three times with TNTw buffer, followed by the addition of 0.1 M citric acid, 0.012% H2O2, and o-phenylenediamine dihydrochloride at a concentration of 1 mg/mL, pH ) 4.5, which was incubated for 30 min at room temperature. The absorbance at 450 nm was measured for each well using a plate reader. The absorbance of biotin standards were plotted versus the logarithm of concentration. A standard curve was fit to a onesite competition model using Graphpad Prism version 3.02 for Windows, GraphPad Software, San Diego, CA (www.graphpad.com). BALB/c nu/nu mice were separated into two groups of four mice each. Both groups were initially fed a normal diet (Irradiated Lab Diet #5053: Rodent Diet #20, [Purina, St. Louis, MO]). Mice were starved for 2 h at the same time every day, followed by retro-orbital phlebotomy. After the first blood draw, defined as time ) 0, one group of mice was placed on a biotin-deficient diet (TD 81079 [Harlan Teklad, Madison, WI]), and the other group remained on the normal diet. Blood was removed for the next 4 days at the same time each day after a 2 h starvation. Blood was immediately collected into 1.5 mL tubes and allowed to clot. Blood samples were centrifuged at 4000 rpm for 3 min, the supernatant was removed, and the process was repeated. Supernatant was removed, and biotin concentrations were determined at various serum dilutions using the biotin assay described above. Serum samples for the mice at each time point were tested at three dilutions in order to obtain the most accurate reading range. Absorbance readings were used to calculate the corresponding biotin concentration in nmoles from the standard curve. Effect of Streptavidin-Biotin System on BisBiotin Tumor Concentration on Normal or BiotinDeficient Diets. BALB/c nu/nu mice (B & K Universal, Fremont, CA) were separated into two groups and fed either a normal or biotin-deficient diet. 10 × 106 Ramos lymphoma cells were injected subcutaneously into the flank of each mouse. One group of mice fed a biotindeficient diet was injected with 125I-bis-biotin ip, mice were sacrificed 24 h later, and tissues were harvested as described below. Upon the emergence of palpable tumors, the remaining mice were injected ip with 300 µg (6 nmol biotin binding sites) of either 131I-1F5-SA-WT conjugate or 131I-1F5-SA-S45A conjugate, labeled by the chloramine-T method. The mice were separated into four groups (n ) 5), a normal diet and biotin-deficient diet for each 1F5-SA conjugate. Twenty-four hours after conjugate injection, 50 µg (6 nmol biotin) of a N-acetylgalactosamine-biotin (NAGB) clearing agent, graciously supplied by the NeoRx Corporation (19), was administered ip. 125I-Bis-biotin (2.8 nmol biotin) was injected ip into the mice 3 h later, and the mice were sacrificed 24 h later. Organs were harvested and weighed, and the radioactivity was measured by a Cobra II gamma counter (Packard Bioscience). Tissue concentrations were calculated as the percent of injected dose per gram of tissue (%ID/g) and reported ( standard deviation. Statistical

Figure 1. The effect of diet on serum biotin concentration. Mice were fed either a normal diet (×) or a biotin-deficient diet (9), with four mice per group. Blood samples were removed by a retro-orbital blood draw, and serum biotin concentration was measured using a biotin assay.

analysis was performed using the nonparametric MannWhitney test to compare groups. Pharmacokinetics of 1F5-SA Using Different Clearing Agents. Alternative clearing agents were investigated to improve the clearance of the mutant conjugate, 1F5-SA-S45A. Human serum albumin (HSA) was modified by conjugation of galactose residues and 2-2.5 equiv of bis-biotin molecules, to produce HSA-bisbiotin. Mice were fed a normal diet and separated into cohorts of four mice. Mice were injected ip with 75 µg (5.6 nmol biotin binding sites) 125I-SA-S45A. At 4 h after the initial reagent, mice were injected ip with either 200 µg (11 nmol biotin) of HSA-bis-biotin, 100 µg of NAGB (12 nmol biotin), or 200 µg of HSA (control group). Blood samples were obtained at various time points throughout the experiment. The percent of injected dose per gram of blood (%ID/g) was calculated and plotted as a function of time. Biodistributions in Mice Pretargeted with 1F5SA, Cleared with a HSA-Bis-Biotin Clearing Agent. BALB/c nu/nu mice were injected with 10 × 106 Ramos cells subcutaneously in the right flank. Mice were fed either a normal or biotin-deficient diet. With the formation of palpable tumors, mice were injected with 300 µg (6 nmol biotin binding sites) of either 131I-1F5-SA-WT or 131I-1F5-SA-S45A. HSA-bis-biotin at a dose of 200 µg (11 nmol biotin) was administered 24 h later. Three hours later 125I-bis-biotin (2.8 nmol biotin) was injected into the mice, and mice were sacrificed 24 h later. Tissue samples were collected, organs were weighed, and radioactivity was measured. RESULTS

Assessment of Serum Biotin Concentrations for Normal vs Biotin-Deficient Diets. The serum levels of biotin were characterized for comparison between the normal mouse diets and the biotin-deficient diets. Prior to the determination of serum biotin levels, a standard curve was established using 11 different biotin concentrations ranging from 0.1 nM to 400 nM. The standards yielded a curve that was well fit to a one-site competition model after logarithmic transformation of the concentration. Mice that remained on the normal diet had serum biotin concentrations of 31 ( 5 nM. After beginning a biotin-deficient diet, serum biotin concentrations dropped to 4.7 ( 1.5 nM in 1 day as shown in Figure 1. Effect of Streptavidin Affinity on Bis-Biotin Tumor Concentration with Normal vs Biotin-Defi-

134 Bioconjugate Chem., Vol. 16, No. 1, 2005

cient Diets. A biodistribution study was conducted to test whether a bis-biotin therapeutic could be used in combination with a 1F5-SA-mutant conjugate to reduce the blocking of biotin-binding sites by endogenous biotin. SA-S45A has an affinity for biotin (2.1 × 10-10 M) (14) that is approximately 3 orders of magnitude lower than SA-WT (2.3 × 10-13 M) (20). Decreasing the affinity for biotin was expected to promote exchange of endogenous biotin, diminishing the blocking of biotin-binding sites by endogenous biotin on mice fed normal diets. The biodistribution of 125I-bis-biotin was first measured in mice fed a biotin-deficient diet 24 h postinjection (Supporting Information). The concentration in all tissues was below 0.1%ID/g, indicating that the molecule does not localize nonspecifically to any of the tissues tested. Following this initial evaluation, a dual-label biodistribution experiment was conducted in mice on normal vs biotin-deficient diets. An outline of the injection times and groups used in both biodistribution experiments is shown in Scheme 1. Tumor-bearing mice were split into two groups and fed either a biotin-deficient diet or a normal diet. These groups were further sub-divided and administered with either 1F5-SA-WT or 1F5-SA-S45A, with all groups injected with the NAGB clearing agent. The concentrations of the 131I-1F5-SA conjugates in different tissues are shown in Figure 2A, 24 h after 125I-bis-biotin administration. The blood concentration of 131I-1F5-SA-WT was 17.7 ( 8.6%ID/g with the normal diet and 7.9 ( 6.0%ID/g for the biotin-deficient diet, indicating that endogenous biotin blocked the action of the biotin-based clearing agent. The blood concentration of the 131I-1F5-SA-S45A conjugate was 21.4 ( 4.9%ID/g with the normal diet and remained relatively high at 14.2 ( 6.8%ID/g with the biotin-deficient diet. The high blood concentrations of 131I1F5-SA-S45A indicated suboptimal clearance of the low affinity mutant conjugate by the NAGB clearing agent. The concentration of 125I-bis-biotin in different tissues at 24 h after injection is displayed in Figure 2B for the four groups of mice. The 125I-bis-biotin concentrations at the tumors with 1F5-SA-S45A were 12 ( 4.6%ID/g and 10 ( 4.4%ID/g for normal vs biotin-deficient diets, respec-

Hamblett et al.

tively, whereas the concentrations with 1F5-SA-WT were 2.2 ( 1.1%ID/g and 7.0 ( 1.6%ID/g for normal vs biotindeficient diets, respectively. These results demonstrate that the blocking of radiolabeled biotin by endogenous biotin observed in the 1F5-SA-WT system can be overcome with the lower affinity SA-S45A mutant. However, the concentrations of 125I-bis-biotin in all normal organs was also significantly higher (p < 0.05) compared to mice receiving 1F5-SA-WT, due to the high blood concentrations of the 1F5-SA-S45A conjugate. The lower affinity of SA-S45A for biotin thus reduced the effectiveness of the NAGB clearing agent, necessitating the use of a new clearing agent with higher affinity to the mutant conjugate. Effect of Different Clearing Agents on the Pharmacokinetics of SA-S45A and 1F5-SA-S45A. In initial pilot experiments, we assessed the efficiencies of modified clearing agents at removing unconjugated 125I-SA-S45A from the circulation of mice fed a normal diet. The use of a divalent biotin-based agent was investigated based on the “avidity” effect previously demonstrated with the lower affinity streptavidin mutants (13). The NAGB clearing agent was tested at twice its usual dose (100 µg). Additionally, a galactosylated HSA-based clearing agent was constructed by conjugation with bis-biotin-NCS (see methods). Unmodified HSA was used as a control. Blood concentrations of 125I-SA-S45A as a function of time are shown in Figure 3 for groups of mice injected with each clearing agent. HSA-bis-biotin reduced the blood concentration of 125I-SA-S45A from 15.9 ( 2.8%ID/g to 1.0 ( 0.01%ID/g at 30 min after injection. The clearing results with the NAGB clearing agent showed a bimodal distribution where two mice showed no reduction in blood %ID/ g, while two mice showed a reduction equivalent to that with the HSA-bis-biotin (leading to the average reduction from 16.8 ( 2.4%ID/g to 7.5 (( 4.5-7.1) %ID/g, with the large standard deviation bars in Figure 3). Separate experiments showed that the NAGB clearing agent efficiency with the SA-S45A mutant streptavidin was sensitive to mouse diet, and the bimodal results most likely reflected that two of the mice had recently con-

Scheme 1. Schematic Depicting Time Course of Reagent Injections and Tissue Distribution in the Aminal Experimentsa

a 131I-1F5-SA was injected at time 0, followed 24 h later by clearing agent. Three hours later 125I-bis-biotin was administered, and mice were sacrificed after an additional 24 h. Two biodistribution experiments were performed, and the reagents used in each group are listed in the table.

Streptavidin−Biotin Affinity in Pretargeting Therapy

Bioconjugate Chem., Vol. 16, No. 1, 2005 135

Figure 2. Tissue distributions of (A) 131I-1F5-SA and (B) 125I-bis-biotin in a pretargeting protocol using the NAGB clearing agent. 131I-1F5-SA-WT and 131I-1F5-SA-S45A were injected into mice on either normal or biotin-deficient diets. Twenty-four hours after the initial injection, the NAGB clearing agent was injected, followed 3 h later by administration of 125I-bis-biotin. Mice were sacrificed 24 h after 125I-bis-biotin injection, the tissues were harvested and weighed, and radioactivity was measured. The %ID/g of each tissue is shown for each group. From left to right in each column: 1F5-SA-S45A, normal diet; 1F5-SA-S45A, biotin-deficient diet; 1F5-SA-WT, normal diet; and 1F5-SA-WT, biotin-deficient diet. Average values are shown ( standard deviation.

sumed normal diet pellets. The HSA-bis-biotin was chosen for further studies to match best to the reduced affinity of the SA-S45A reagents under the broadest set of diet conditions. Biodistributions in Mice Pretargeted with 1F5SA and Cleared with HSA-Bis-Biotin. In a second biodistribution experiment, Ramos-bearing nude mice were split into two groups and fed either a biotin-deficient diet or a normal diet. These groups were further subdivided and administered with either 1F5-SA-WT or 1F5SA-S45A. The HSA-bis-biotin clearing agent was used to determine if 1F5-SA-S45A in mice fed a normal diet could yield high 125I-bis-biotin tumor concentrations and low 125I-bis-biotin concentrations in the blood and normal organs. Figure 4A depicts the Ab-SA localization in each organ. The blood concentration of 131I-1F5-SA-S45A 24 h after the 125I-bis-biotin injection was 5.4 ( 3.0%ID/g with the normal diet and 4.4 ( 0.8%ID/g in the biotindeficient group. 1F5-SA-WT exhibited a blood concentration of 9.7 ( 3.7%ID/g in mice fed a normal diet and 4.2

( 2.7%ID/g in mice fed a biotin-deficient diet. 125I-bisbiotin biodistributions are displayed in Figure 4B. Tumor concentrations of 125I-bis-biotin 24 h after injection were 0.2 ( 0.03%ID/g for 1F5-SA-WT with the normal diet, and 0.1 ( 0.05%ID/g on the biotin-deficient diet. The tumor concentrations of 125I-bis-biotin were 4.9 ( 3.2%ID/g and 3.4 ( 0.6%ID/g for 1F5-SA-S45A on normal vs biotindeficient diets, respectively. Blood concentrations of 125Ibis-biotin for 1F5-SA-S45A with the normal vs biotindeficient diets were 2.5 ( 0.6%ID/g and 3.9 ( 0.7%ID/g, respectively. A comparison of tumor/normal organ ratios of the nonoptimized vs optimized 1F5-SA-S45A conjugate with the normal diet is compared to the optimized 1F5SA-WT conjugate on the biotin-deficient diet in Figure 5. It can be observed that the bis-biotin clearing agent improves the overall performance of the 1F5-SA-S45A conjugate similar to the best 1F5-SA-WT conjugate system, with the advantage of attaining this performance on a normal biotin diet.

136 Bioconjugate Chem., Vol. 16, No. 1, 2005

Figure 3. The effect of clearing agents on blood pharmacokinetics of SA-S45A. Three groups of mice (n ) 4) were fed a normal diet and injected with 125I-SA-S45A, and 4 h later the clearing agent was injected. Blood samples were removed at various time points, the radioactivity was measured, and %ID/g was calculated for each group: HSA control (9) NAGB (O), and HSA-bis-biotin (2). Average values are shown ( standard deviation. DISCUSSION

In this study, we have investigated the role of streptavidin affinity, modified therapeutic ligands, and clearing agents in a pretargeting system designed for treating B-cell lymphoma. In particular, the tradeoffs between the advantage of faster biotin off-rates for the rapid exchange of bound endogenous biotin with the biotinylated therapeutic, and the disadvantages of poorer therapeutic retention and blood clearance, have been clarified and initially optimized. The major conclusion is that the low affinity streptavidin mutant can be used with bis-biotin therapeutics of higher effective affinity (avidity) to achieve exchange of blocking endogenous biotin and excellent retention at the tumor. To achieve high tumor/normal organ ratios, it was also found that the clearing agent must be modified to the bis-biotin derivative in order to effectively clear the radiolabeled biotin therapeutic from the blood. When the optimized therapeutic and clearing agents were utilized with the lower affinity mutant, the performance of the pretargeting system with mice on normal diets reached that of the WT streptavidin system on biotin-deficient diets. In addition to providing useful insight into the role of biotin off-rates in pretargeting systems, this finding suggests that the engineered pretargeting system could reduce the impact of endogenous biotin and reduce the quantities of antibody-streptavidin conjugates needed to achieve effective radiation doses at the tumor site. The bis-biotin strategy was based on previous in vitro work using SA affinity mutants and bivalent biotin ligands (13). The bivalent binding of optimized bis-biotin ligands raised their affinity with a site-directed SA-Y43A mutant to that comparable to biotin and wild-type SA. This property allows displacement of monovalent endogenous biotin due to the faster off-rate of the SA mutant, while maintaining the high affinity retention of the bisbiotin once it is exchanged. Initial evaluation of the Y43A streptavidin mutant used in the in vitro studies suggested that a faster off-rate was necessary in the in vivo pretargeting system, and thus the SA- S45A mutant was utilized in this study. The half-life of biotin dissociation from SA-WT is over 6.5 h at 37 °C, while for SA-S45A it is only 14 s. To test the impact of high and low biotin serum concentrations on wild-type and mutant SA biotin and bis-biotin systems, we conducted a series of experi-

Hamblett et al.

ments with mice possessing variable serum concentrations of endogenous biotin as a result of dietary manipulations. The biotin-deficient diets rapidly lowered the serum biotin concentration by 6-fold, from 31 nM to 4.7 nM. While a biotin-deficient diet was sufficient to reduce endogenous biotin serum concentrations 6-fold in mice within 1 day, in rats a 6-fold reduction of plasma biotin concentration took six weeks (7). Endogenous biotin levels in human plasma were reported between 1.3 and 2.2 nM (10, 21), which is lower than reported above in nude mice (31 nM) and in rats (5.3 nM) (7). Tumor biopsies of patients in the clinic treated with NR-LU-10-SA (WT) revealed a reduction in the biotin-binding capacity over time, which was hypothesized to be due to blockage by endogenous biotin (22). As highlighted by Sharkey et al. (23), the presence of biotin storage sites may play a role in blocking biotin-binding sites. Development of cutaneous and neurological disorders in the aforementioned rats by 3-5 weeks suggests that reducing endogenous biotin levels in humans with such a diet would require close monitoring. The tumor concentration of 125I-bis-biotin in mice pretargeted with 1F5-SA-WT and low concentrations of endogenous biotin (biotin-deficient diet) was 7.0%ID/g compared to 2.2%ID/g in mice with high concentrations of serum endogenous biotin (normal diet). Tumor concentration of 125I-bis-biotin was 12%ID/g in mice with high serum biotin concentrations treated with 1F5-SAS45A, indicating endogenous biotin was unable to block the biotin-binding sites of 1F5-SA-S45A as it did 1F5SA-WT. There was a concomitant increase in the concentrations of 125I-bis-biotin in normal tissues, especially the blood. This finding highlighted an important tradeoff with the SA mutant: the normal biotin-based clearing agent also had an increased off-rate that prevented effective blood clearance. Biotinylated HSA modified with galactose residues was one of the first clearing agents used to remove Ab-SA conjugates from the blood (4). The same strategy was used to construct the multivalent clearing agent that was designed to exhibit the same increase in effective affinity through the avidity effect observed with the bis-biotin therapeutic. The HSA-bisbiotin clearing agent, used with the 1F5-SA-S45A conjugate and 125I-bis-biotin, reduced normal organ concentrations in mice fed a normal diet to similar levels as those observed with the combination of the NAGB clearing agent, the 1F5-SA-WT conjugate, and 125I-biotin in mice fed biotin-deficient diets (Figure 2). We have thus shown that the blocking of biotin-binding sites by endogenous biotin can be mitigated by using a SA-mutant with a lower affinity for biotin coupled with a bivalent bisbiotin therapeutic and clearing agent. Further optimization of the clearing agent and the streptavidin affinity will be necessary to fine-tune the composite activity and specificity (tumor-tissue ratios) further, but these results demonstrate the potential for engineering protein and ligands to optimize the overall pretargeting system. ACKNOWLEDGMENT

We thank Dr. John Pagel and Dr. David Hyre for insightful discussions. This work was supported by the NIH (NCI Grant CA76287 and DK49655), the Whitaker Foundation (Graduate Fellowship for K.J.H.), and the Department of Energy, Medical Applications and Biophysical Research Division, Office of Biological and Environmental Research under grant number DE-FG0398ER62572.

Streptavidin−Biotin Affinity in Pretargeting Therapy

Bioconjugate Chem., Vol. 16, No. 1, 2005 137

Figure 4. Tissue distributions of (A) 131I-1F5-SA and (B) 125I-bis-biotin in a pretargeting protocol using the HSA-bis-biotin clearing agent. 131I-1F5-SA-WT and 131I-1F5-SA-S45A were injected into mice on either normal or biotin-deficient diets. Twenty-four hours after injection HSA-bis-biotin clearing agent was injected, followed 3 h later by administration of 125I-bis-biotin. Mice were sacrificed 24 h after 125I-bis-biotin injection, the tissues were harvested and weighed, and radioactivity was measured. The %ID/g of each tissue is shown for each group. From left to right in each column: 1F5-SA-S45A, normal diet; 1F5-SA-S45A, biotin-deficient diet; 1F5-SA-WT, normal diet; and 1F5-SA-WT, biotin-deficient diet. Average values are shown ( standard deviation.

Figure 5. Graph displaying tumor/normal tissue ratios. Tumor/normal tissue ratios of 125I-bis-biotin were calculated from the biodistribution experiments. The following groups are displayed from left to right: 1F5-SA-S45A, NAGB clearing agent, normal diet; 1F5-SA-S45A, HSA-bis-biotin clearing agent, normal diet; and 1F5-SA-WT, NAGB clearing agent, biotin-deficient diet.

138 Bioconjugate Chem., Vol. 16, No. 1, 2005 LITERATURE CITED (1) Mclaughlin, P. (2002) Progress and Promise in the Treatment of Indolent Lymphomas. Oncologist 7, 217-225. (2) Witzig, T. E., Gordon, L. I., Cabanillas, F., Czuczman, M. S., Emmanouilides, C., Joyce, R., Pohlman, B. L., Bartlett, N. L., Wiseman, G. A., Padre, N., Grillo-Lopez, A. J., Multani, P., and White, C. (2002) Randomized Controlled Trial of Yttrium-90-Labeled Ibritumomab Tiuxetan Radioimmmunotherapy Versus Rituximab Immunotherapy for Patients With Relapsed or Refractory Low-Grade, Follicular, or Transformed B-Cell Non-Hodgkin’s Lymphoma. J. Clin. Oncol. 20, 24532463. (3) Press, O. W., Eary, J. F., Appelbaum, F. R., Martin, P. J., Nelp, W. B., Glenn, S., Fisher, D. R., Porter, B., Matthews, D. C., Gooley, T., and Bernstein, I. D. (1995) Phase II Trial of I131-B1 (Anti-CD20) Antibody Therapy with Autologous Stem Cell Transplantation for Relapsed B Cell Lymphomas. Lancet 346, 336-340. (4) Axworthy, D. B., Reno, J. M., Hylarides, M. D., Mallett, R. W., Theodore, L. J., Gustavson, L. M., Su, F. M., Hobson, L. J., Beaumier, P. L., and Fritzberg, A. R. (2000) Cure of Human Carcinoma Xenografts by a Single Dose of Pretargeted Yttrium-90 with Negligible Toxicity. Proc. Natl. Acad. Sci. 97, 1802-1807. (5) Press, O. W., Corcoran, M., Subbiah, K., Hamlin, D. K., Wilbur, D. S., Johnson, T., Theodore, L., Yau, E., Mallett, R., Meyer, D. L., and Axworthy, D. (2001) A Comparative Evaluation of Conventional and Pretargeted Radioimmunotherapy of CD20-Expressing Lymphoma Xenografts. Blood 98, 2535-2543. (6) Schultz, J., Lin, Y., Sanderson, J., Zuo, Y., Stone, D., Mallet, R., Wilbert, S., and Axworthy, D. (2000) A tetravalent singlechain antibody-streptavidin fusion protein for pretargeted lymphoma therapy. Cancer Res. 60, 6663-6669. (7) Mock, D. M., and DuBois, D. B. (1986) A sequential, solidphase assay for biotin in physiologic fluids that correlates with expected biotin status. Anal. Biochem. 153, 272-278. (8) Rusckowski, M., Fogarasi, M., Fritz, B., and Hnatowich, D. J. (1997) Effect of endogenous biotin on the applications of streptavidin and biotin in mice. Nucl. Med. Biol. 24, 263-268. (9) Klumb, L., Chu, V., and Stayton, P. S. (1998) Energetic Roles of Hydrogen Bonds at the Uriedo Oxygen Binding Pocket Streptavidin-Biotin Complex. Biochemistry 37, 7657-7663. (10) Baker, H. (1985), Assessment of Biotin Status: Clinical Implications. Ann. N. Y. Acad. Sci. 447, 129-132. (11) Shriver, B., Roman-Shriver, C., and Allred, J. B. (1993), Depletion and Repletion of Biotinyl Enzymes in Liver of Biotin-Deficiet Rats: Evidence of a Biotin Storage System. J. Nutr. 123, 1140-1149. (12) Baker, H., Frank, O., Chen, T., Feingold, S., DeAngelis, B., Baker, E. R. (1981) Elevated vitamin levels in colon

Hamblett et al. adenocarcinoma as compared with metastatic liver adenocarcinoma from colon primary and normal adjacent tissue. Cancer 47, 2883-2886. (13) Hamblett, K. J., Kegley, B. B., Hamlin, D. K., Chyan, M., Hyre, D. E., Press, O. W., Wilbur, D. S., and Stayton, P. S. (2002), A Streptavidin-Biotin Binding System That Minimizes Blocking By Endogenous Biotin. Bioconjugate Chem. 13, 588-598. (14) Hyre, D. E., Le Trong, I., Freitag, S., Stenkamp, R. E., and Stayton, P. S. (2000) Ser45 plays an important role in managing both the equilibrium and transition state energetics of the streptavidin-biotin system. Protein Sci. 9, 878885. (15) Chu, V., Freitag, S., Le Trong, I., Stenkamp, R. E., and Stayton P. S. (1998) Thermodynamic and structural consequences of flexible loop deletion by circular permutation in the streptavidin-biotin system. Protein Sci. 7, 848-859. (16) Chilkoti, A., Tan, P. H., and Stayton, P. S. (1995) Sitedirected mutagenesis studies of the high-affinity streptavidin-biotin complex: contributions of tryptophan residues 79, 108, and 120. Proc. Natl. Acad. Sci. 92, 1219-1224. (17) Hylardies, M., Mallett, R. W., and Meyer, D. L. (2001) A Robust Method for the Preparation and Purification of Antibody/Streptavidin Conjugates. Bioconjugate Chem. 12, 421-427. (18) Huang, E. Z., and Rogers, Y. (1997) Competitive enzymatic assay of biotin. Methods Enzymol. 279, 304-308. (19) Weiden, P. L., Breitz, H. B., Press, O., Appelbaum, J. W., Bryan, J. K., Gaffigan, S., Stone, D., Axworthy, D., Fisher, D., and Reno, J. (2000) Pretargeted radioimmunotherapy (PRIT) for treatment of non-Hodgkin’s lymphoma (NHL): initial phase I/II study results. Cancer Biother. Radiopharm. 15, 15-29. (20) Green, N. (1975) Avidin. Adv. Protein Chem. 29, 85-133. (21) Wood, H. G. (1998) Biotin, in Vitamins: Fundamental Aspects in Nutrition & Health (Combs, G. F., Ed.) pp 349363, Academic Press, San Diego, (22) Breitz, H. B., Weiden, P. L., Beaumier, P. L., Axworthy, D. B., Seiler, C., Su, F., Graves, S., Bryan, K., and Reno, J. M. (2000) Clinical optimization of pretargeted radioimmunotherapy with antibody-streptavidin conjugate and 90Y-DOTAbiotin. J. Nucl. Med. 41, 131-140. (23) Sharkey, R. M., Karacay, H., Griffiths, G. L., Behr, T. M., Blumenthal, R. D., Mattes, M. J., Hansen, H. J., and Goldenberg, D. M. (1997) Development of a Streptavidin-AntiCarcinoembryonic Antigen Antibody, Radiolabeled Biotin Pretargeting Method for Radioimmunotherapy of Colorectal Cancer. Studies in a Human Colon Cancer Xenograft Model. Bioconjugate Chem. 8, 595-604.

BC034049G