The in vivo characteristics of genetically engineered divalent and tetravalent single-chain antibody constructs

Nuclear Medicine and Biology 32 (2005) 157 – 164 www.elsevier.com/locate/nucmedbio

The in vivo characteristics of genetically engineered divalent and tetravalent single-chain antibody constructsB Uwe A. Wittela, Maneesh Jaina, Apollina Goela, Subhash C. Chauhana, David Colcherb, Surinder K. Batraa,* a

Department of Biochemistry and Molecular Biology, Eppley Institute for Research in Cancer and Allied Diseases, University of Nebraska Medical Center, Omaha, NE 68198, USA b Department of Radioimmunotherapy, Beckman Research Institute at City of Hope National Medical Center, Duarte, CA, USA Received 3 September 2004; received in revised form 16 November 2004; accepted 21 November 2004

Abstract Engineered multivalent single-chain Fv (scFv) constructs have been demonstrated to exhibit rapid blood clearance and better tumor penetration. To understand the short plasma half-life of multivalent single-chain antibody fragments, the pharmacokinetic properties of covalent dimeric scFv [sc(Fv)2], noncovalent tetrameric scFv {[sc(Fv)2]2} and IgG of MAb CC49 were examined. The scFvs displayed an ability to form higher molecular aggregates in vivo. A specific proteolytic cleavage of the linker sequence of the covalent dimeric or a deterioration of the noncovalent association of the dimeric scFv into tetravalent scFv constructs was not observed. In conclusion, sc(Fv)2 and [sc(Fv)2]2 are stable in vivo and have significant potential for diagnostic and therapeutic applications. D 2005 Elsevier Inc. All rights reserved. Keywords: Monoclonal antibodies; Immunotherapy; Pharmacokinetics; Cancer therapy

1. Introduction Antibody-based radio-immunotherapy has been more successful in the management of hematological malignancies than solid tumors [1–5] because of the easier access of the therapeutic antibody to its target antigen on tumor cells. In solid tumors, however, the antigens are well bprotectedQ due to poor vascular supply and high tumor interstitial pressures [6]. Additionally, the therapeutic antibody has to overcome large transport distances on its way to its target in solid tumors [7]. To overcome these obstacles, great effort has been undertaken to increase the diffusion of therapeutic antibodies into the tumor. A common approach for increasing tumor diffusion is the generation of antibodies with lower molecular weights [8]. Several strategies resulting in low molecular weight antibodies with unaltered binding of the antigen have been described [9]. We have developed several single-chain

B This work was supported by a grant from the United States Department of Energy (DE-FG0295ER62024). * Corresponding author. Tel.: +1 402 559 5455; fax: +1 402 559 6650. E-mail address: [email protected] (S.K. Batra).

0969-8051/$ – see front matter D 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.nucmedbio.2004.11.003

derivatives of CC49, a murine MAb directed against the cancer-associated glycoprotein TAG-72 [10]. These include monovalent (scFv) and divalent sc(Fv)2 forms as well [11,12]. One of the characteristics of these proteins was the formation of noncovalent dimers resulting in noncovalent dimeric (divalent) sc(Fv)2 and tetrameric (tetravalent) [sc(Fv)2]2 forms, respectively [11,12]. These noncovalent dimeric and tetrameric forms exhibited an increased biological half-life and higher tumor uptake than their monomeric counterparts [11,13,14]. With the decrease in molecular weight, however, these new molecules were subjected to a rapid in vivo clearance with a dramatic reduction of the biological half-life when compared to whole IgG [14,15]. The shortened in vivo halflife of the antibody fragments opened a whole new avenue for tumor imaging and cancer therapy with a lower delivery of the radioconjugate to healthy organs, while the delivery of the antibody to the tumor is characterized by an improved capability of diffusion [16,17]. The mechanisms underlying the rapid clearance of these antibody fragments, leading to their pharmacological superiority, are not fully understood. While the plasma half-life appears to correlate with the molecular weight from

158

U.A. Wittel et al. / Nuclear Medicine and Biology 32 (2005) 157–164

monovalent to tetravalent scFvs [18], very little is known about the possible proteolytic cleavage and/or aggregate formation of these antibody fragments in vivo. The multivalent constructs, like the sc(Fv)2 engineered with three linker sequences within the molecule, could be targets for rapid in vivo proteolytic cleavage followed by a fast renal elimination [9]. In the current study, we have examined the in vivo stability and pharmacological properties of the covalent dimeric scFv and the noncovalent tetrameric scFv derived from CC49 MAb as members of multivalent scFvs.

with 20 Ag Iodo-Gen (Pierce, Rockford, IL USA), 100 Ag protein was mixed in 0.1 M sodium phosphate buffer with 0.1 mCi Na125I or Na131I, and incubated for 3 min at room temperature. The samples were subjected to size exclusion chromatography over a Sephadex 25(Sigma) column for the removal of free iodine, and the labeled protein was collected in the void volume. Instant thin-layer chromatography (ITLC) was performed with the purified, labeled product and the reaction mixture with 25% methanol in water as a solvent to determine the percentage of free iodine. 2.4. HPLC analysis

2. Materials and methods 2.1. Antibody synthesis and purification The engineering and purification of divalent scFv construct VL-linker-VH-linker-VL-linker-VH-His6 of murine CC49 has previously been described [12]. Upon expression, the covalent dimeric scFv spontaneously formed the noncovalent tetrameric [sc(Fv)2]2. The scFvs were first purified by nickel affinity chromatography (Ni2+-nitrilo triacetic acid Superflow, Quiagen, Valencia, CA), and dimeric and tetrameric forms were further purified by size fractionation on a Superdex 200 column (1.660 cm, Pharmacia, Upsala, Sweden), concentrated (Cententricons Y-10, Millipore, Bedford, MA) and stored at 708C. Protein concentrations were assessed with the Bio-Rad DC protein assay (Bio-Rad Laboratories, Hercules, CA) using the BSA standard curve. The purity and integrity of the constructs were analyzed by SDS-PAGE according to the method described by Laemmli [19]. The electrophoresis was performed using a 10% acrylamide gel under reducing and nonreducing conditions. Gels were either stained with Coomassie blue R250 or exposed directly to film for autoradiography. 2.2. Determination of immunoreactivity The specific immunoreactivity of the proteins at various points in the purification process was assessed by solidphase competitive ELISA. As previously published [12], polystyrene plates were coated with 50 ng/well bovine submaxillary gland mucin (Sigma, St. Louis, MO) and blocked with 5% BSA at 378C for 1 h. Threefold serial dilutions of CC49 IgG and scFvs, prepared with 6 ng of biotinylated CC49 IgG, were added to the wells and incubated for 2 h at room temperature. After washing, the plates were incubated with alkaline phosphatase-conjugated streptavidin for 90 min. The reaction was developed using p-nitrophenyl phosphate as a substrate (KPL, Gaithersburg, MD) and the absorption was read at 410 nm using a Dynatech MR 5000 automatic microtiter plate reader. 2.3. I and 131I labeling and analysis of radiolabeled scFv forms The scFv constructs and CC49 IgG were labeled with Na125I as described by Colcher et al. [20]. In a tube coated

The degree of aggregation or degradation following the labeling procedure, as well as in mouse sera for the in vivo stability studies, was monitored by HPLC gel-filtration. Samples were injected onto a TSK G2000 and TSK G3000 (Toso Haas, Tokyo, Japan) size exclusion column connected in a series with 67 mM phosphate and 100 mM KCl buffer (pH 6.8) as the mobile phase at a flow rate of 0.5 ml/min. The absorption was monitored at 280 nm and 0.25-ml fractions were collected. The radioactivity of the fractions was determined with a Packard Minaxi AutoGamma 5000 gamma counter (Packard, Meriden, CT). 2.5. Assessment of the in vivo stability and blood clearance For the assessment of the in vivo stability, 5 ACi of radiolabeled scFvs or CC49 whole IgG were injected intravenously in the tail vein of 8-week-old female athymic mice (NIH). After 0.5, 1, 2 and 4 h, animals receiving the sc(Fv)2 and [sc(Fv)2]2 were euthanized and blood was collected from the left subclavian artery. Due to the longer t 1/2 of [sc(Fv)2]2, additional animals were euthanized after 6 and 12 h, while for CC49 IgG, the in vivo stability was determined after 1, 4, 6, 12, 18, 24, 48 and 72 h. The blood was collected for all animals, cells removed and 50 Al of the undiluted serum was applied to the HPLC gel-filtration columns, as described previously. The radioactivity of the fractions was plotted as percentages of the radioactivity loaded on the column. For the determination of the serum half-lives, 5 ACi of 125 I-labeled [sc(Fv)2]2 and 2.5 ACi 131I-labeled sc(Fv)2 were injected intravenously, and 5 Al of blood samples were collected at indicated time intervals. The time point of 1 min was defined as 100%. Regression curves and t 1/2 were calculated and plotted through the values obtained. 2.6. Biodistribution of the scFvs and CC49 MAb Female athymic mice (8 weeks old), bearing LS174T colon carcinoma xenografts, were used for biodistribution studies. Cells (3106) were implanted subcutaneously, and after 8 to 10 days, when most of the tumors reached a volume of 250–300 mm3, the animals were randomized in groups of five for the biodistribution experiment. Pair-label biodistribution was performed with simultaneous injection of 2.5 ACi of 131I-labeled sc(Fv)2 and 5 ACi of 125I-labeled

U.A. Wittel et al. / Nuclear Medicine and Biology 32 (2005) 157–164

[sc(Fv)2]2. For CC49 whole IgG, biodistribution was performed with 5 ACi of 125I-labeled protein. Animals were injected intravenously over the tail vein and sacrificed after 0.5, 4, 6, 16, 24 and 48 h. Blood, tumors and other major organs were harvested, weighed and counted in a gamma scintillation counter for the determination of radioactivity. The final values were calculated as a percentage of the total injected dose per gram wet weight, and the radio-localization indices were determined. To assess the influence of the applied dose on the tumor uptake, animals in groups of three received either 5 ACi of 125 I-labeled sc(Fv)2 or 5 ACi of 125I-labeled sc(Fv)2, in addition to 100 Ag of the unlabeled sc(Fv)2 construct. All animals were sacrificed after 4 h. The animals were assessed and processed as described previously. 3. Results The covalent dimeric scFv sc(Fv)2 and the noncovalent tetravalent scFv [sc(Fv)2]2 were secreted by the P. pastoris as soluble and functional proteins 96 h after the induction of the protein expression with methanol. The purification process resulted in more than 90% purity, as determined by SDS-PAGE under denaturing and nondenaturing conditions. The binding characteristics of the scFvs were evaluated by solid-phase competitive ELISA, where sc(Fv)2 and [sc(Fv)2]2 were able to inhibit 50% of the binding of biotinylated CC49 IgG at concentrations of 2.4410 7 and 5.6510 8 M, respectively. The CC49 whole IgG resulted in a 50% binding inhibition at 5.010 9 M (Fig. 1). These results are in accordance with our previous observation where we analyzed the binding affinity of these single chain constructs by a more direct approach of surface plasmon resonance [14]. Those analyses indicated that the K A of tetramer was slightly lower than (but comparable to) intact IgG). This lower K A of dimer and tetramer was attributable

159

Fig. 2. Pharmacokinetic analysis of IgG and scFv constructs of CC49 in athymic mice. Blood samples were collected from different groups at indicated time points and radioactivity was measured. The data displayed represent the averageFS.E.M. of three animals per group. The values for early time points representing the alpha phase were plotted separately (inset) to calculate t 1/2a.

to slow K on (association constant). In fact, the K off (dissociation constant) for tetramer is two to threefold slower than intact antibody, indicating its stronger association with the antigen [14]. Radio-iodination with 125I or 131I yielded proteins with specific activity of 0.8 mCi/mg, with less than 5% free iodine after size exclusion chromatography. SDS-PAGE of the radiolabeled products did not reveal breakdown products following the radiolabeling. HPLC analyses ruled out any aggregate formation in the radiolabeled proteins. The immunoreactivity of the radiolabeled constructs was analyzed by solid-phase radio-immunoassay (RIA) using BSMand BSA-coated beads to determine specific and nonspecific binding, respectively. The specific binding was found to be 85–95% while the nonspecific binding was 0.8–1.5%. 3.1. Blood clearance and biodistribution of CC49 sc(Fv)2 [sc(Fv)2]2 and IgG

Fig. 1. Solid-phase competitive ELISA between CC49 MAb and sc(Fv)2, [sc(Fv)2]2 or CC49 MAb to BSM. Varying amounts of IgG, dimer and tetramer were allowed to compete with 6 ng of biotinylated CC49 IgG for binding to BSM. Fifty percent inhibition of binding of biotinylated IgG was achieved by 2.4410 7 M of sc(Fv)2, 5.6510 8 of [sc(Fv)2]2 and 5.010 9 for IgG.

The plasma half-life of the scFvs, as well as the parent IgG, increased with the increase in the molecular weight of the protein (Fig. 2). By 8 h postadministration, N90% of the scFvs cleared from the circulation, while the corresponding blood levels of IgG were nearly 40% of that observed at the initial time point. The t 1/2h, the half-life representing the actual clearance of the molecule from the blood, was determined to be 70.2, 116.30 and 330.2 min for sc(Fv)2 [sc(Fv)2]2 and CC49 IgG, respectively. These differences were also reflected when the percentage of the injected dose in the blood was determined in biodistribution experiments (Fig. 3). The levels of sc(Fv)2 in the blood decreased rapidly from 22.8F0.9% ID/g at 30 min to 4.2F0.2% ID/g at 4 h postadministration. The % ID in the blood for the tetrameric [sc(Fv)2]2 was higher than that observed for sc(Fv)2 (27.6F0.8% ID after 30 min, 10.7F1.0% ID/g after 4 h and 6.3F0.3% ID/g after 6 h),

160

U.A. Wittel et al. / Nuclear Medicine and Biology 32 (2005) 157–164

Fig. 3. Comparative biodistribution of the scFv constructs with CC49 IgG in LS174T tumor xenograft-bearing athymic mice. Tumor-bearing mice were injected with radio-iodinated scFvs or IgG and euthanized at the indicated time points (x-axis). The organs were weighed and measured for radioactivity to calculate the percent injected dose uptake ( y-axis). The differences in plasma half-life are also represented in the % ID values obtained from the different constructs. The dose delivered to the kidneys correlates with the molecular weight of the construct and is largely reduced by the [sc(Fv)2]2 when compared to sc(Fv)2. The data displayed represents the averageFS.E.M. of five animals per group.

thereby confirming the slower pharmacological kinetics determined in the plasma clearance experiments (Fig. 2). Similarly, the slow blood clearance of CC49 IgG led to high values of the % ID/g in the blood, with almost 10% of the radioactivity present after 48 h. While there was an increased uptake of the dimer in the kidneys (20.0F0.7% ID/g after 30 min), the larger tetrameric [sc(Fv)2] 2 (9.7F0.3% ID/g) accumulated at levels comparable to

IgG (8.0F0.4% ID/g) (Fig. 4). The high radiation dose delivered to the kidney that was observed for the sc(Fv)2 construct did not occur in the 120-kDa [sc(Fv)2]2 (Fig. 3b). On the other hand, the radioactivity distributed to the liver was also not significantly increased in the animals receiving the noncovalent tetrameric [sc(Fv)2]2. In order to gain information on the pathway of excretion, we further investigated if the actual pathway of blood clearance can be saturated by an increase of the antibody dose by 20-fold. By the addition of 100 Ag of unlabeled sc(Fv)2 to the radiolabeled antibody fragment, the % ID/g in the blood after 4 h increased less than twofold from 1.38F0.2% ID/g in the low-dose animals to 2.11F0.2 in the high-dose animals (Fig. 5). This suggests that the mechanism involved in the removal of the radiolabeled sc(Fv)2 from the blood stream cannot be easily saturated. Also, this experiment demonstrated that the sc(Fv)2 is not subject to renal filtration and a saturable reabsorption mechanism that would lead to an increased loss of radiolabeled sc(Fv)2. 3.2. In vivo stability of sc(Fv)2 [sc(Fv)2]2 and CC49 IgG

Fig. 4. Biodistribution of radiolabeled sc(Fv)2 in the presence of excess cold protein. Tumor-bearing animals were injected with either 5 Ag of radiolabeled protein alone or in combination with 100 Ag of unlabeled (cold) sc(Fv)2. At a high dose, the degradation of the sc(Fv)2 could not be saturated, as is indicated by only slightly increased values for the % ID in the blood. The values shown represent the averageFS.E.M. of three animals per group, sacrificed 4 h after the treatment.

From the biodistribution and the blood clearance studies, only the quantitative presence of radioactivity in the blood can be determined. Information on the integrity of the proteins cannot be obtained by these experiments. In order to assess the in vivo stability of the single-chain antibodies, size fractionation by HPLC of serum obtained from nontumor-bearing mice at various time points after the injection

U.A. Wittel et al. / Nuclear Medicine and Biology 32 (2005) 157–164

Fig. 5. Stability analysis of injected radioiodinated scFvs and IgG in mouse serum. Mice were injected with 5 ACi of 125I-labeled scFvs or IgG via tail vein and sacrificed at the indicated time points. Serum was fractionated by HPLC and the radioactivity of the fractions was plotted as percentages of the total radioactivity loaded. Both dimeric and tetrameric scFvs exhibited in vivo aggregation. This aggregation was more predominant in the covalent dimer, but also detectable in the [sc(Fv)2]2. A cleavage into monomeric scFv with the resulting molecular weight of approximately 30 kDa could not be observed.

of the radiolabeled scFvs or whole IgG was performed. The purity and integrity of the radiolabeled proteins were determined by HPLC prior to injection in animals. Radiolabeled proteins were resolved as single peak corresponding to the molecular size of the construct with no indication of impurities, degradation or aggregation. In these studies, IgG proved to be extremely stable, not leading to any breakdown products or aggregates (Fig. 5). By these means, 72 h after the injection of 125I-labeled IgG, 87.7% of the radioactivity detected was found to be associated with the fractions

161

corresponding to the molecular weight of IgG, covering a size range of approximately 100 –200 kDa (Table 1). No significant increase in radioactivity in fractions representing a higher or lower molecular weight could be detected when compared to the initial time point (Fig. 5c). HPLC analysis of the sc(Fv)2 revealed 68.11% of the radioactivity in fractions corresponding to the molecular weight of sc(Fv)2 after 30 min, 57.2% after 1 h and 29.7% after 2 h (Fig. 5a). Interestingly, as demonstrated in Fig. 4, at 2 h, 25.19% of the radioactivity were detected as aggregates with the approximate molecular weight corresponding to the noncovalent tetrameric [sc(Fv)2]2 form. Furthermore, after 2 h, aggregates exceeding the molecular weight of the [sc(Fv)2]2 accounted for 14.8% of the total radioactivity. These aggregates, as well as the purified noncovalent [sc(Fv)2]2, were not stable under the mildest denaturation conditions and not detectable in SDS-PAGE (data not shown). Not only was the plasma half-life of the sc(Fv)2 short, but the breakdown was also rapid. After 4 h, low molecular weight forms accounted for 66.37% of the present radioactivity. At that time point, only 33.6% were still detected in the fractions corresponding to 60 kDa and above. Similar observations were made for the noncovalent tetrameric [sc(Fv)2]2 (Fig. 5b). As summarized in Table 1, up to 2 h after injection, less than 20% of the present radioactivity was detected with an approximate molecular weight range below the noncovalent tetrameric form, while up to 20.78% was detected as aggregates. No significant in vivo dissociation of the noncovalent bond between the two molecules of [sc(Fv)2]2 to dimeric sc(Fv)2 was observed. The in vivo stability experiments did not indicate any linker-specific breakdown of the scFv constructs, leading to variable light-chain or heavy-chain fragments or monovalent forms, explaining the different pathway of excretion revealed by comparative biodistribution between the divalent sc(Fv)2 and the noncovalent tetrameric [sc(Fv)2]2 (Fig. 6). 4. Discussion Engineered divalent and tetravalent constructs of MAb CC49 exhibited short plasma half-lives and rapid tumor accumulation. The reduction of the molecular weight of the antibody can enhance its ability to diffuse, leading to a more homogeneous tumor penetration [16,21]. On the other hand, the reduction of the molecular weight also leads to a significantly shortened biological half-life, especially when the molecules are smaller than the glomerular cutoff [8,22]. Thus, molecules like dimeric and tetrameric scFvs, and di-, tri- and tetrabodies, with their molecular weight between 60 and 120 kDa, seem to be the perfect compromise for radioimmunotherapy and tumor imaging, due to their low molecular weight and fast clearance [23]. At the same time, however, the benefit of a fast clearance in tumor imaging is a disadvantage in radioimmunotherapy, since the tumor

162

U.A. Wittel et al. / Nuclear Medicine and Biology 32 (2005) 157–164

Table 1 Distribution of the percentage of radioactivity in the fractions of HPLC-separated serum sc(Fv)2 (kDa)

0h

0.5 h

1h

2h

4h

N 200 100 –200 35 –100 b 35

0 0 100 0

6.49 15.38 68.11 10.05

10.52 17.12 57.2 15.19

14.8 25.19 29.66 30.31

10.21 8.04 15.39 66.37

[sc(Fv)2]2

0h

0.5 h

1h

2h

4h

6h

12 h

N 200 100 –200 35–100 b 35

0 100 0 0

18.65 52.69 17.99 10.64

20.78 60.27 10.90 8.04

19.75 61.84 8.79 9.63

20.23 54.53 10.40 15.20

16.54 54.94 10.42 18.35

20.00 38.75 12.3 29.44

CC49 IgG

0h

1h

4h

6h

12 h

18 h

24 h

48 h

72 h

N 200 100 –200 35–100 b 35

0 100 0 0

4.36 82.75 7.51 5.43

7.04 80.27 6.14 6.55

5.17 83.23 5.76 5.84

4.11 83.97 4.76 7.15

1.44 87.1 8.67 2.79

1.63 87.43 7.66 3.24

2.03 87.4 7.56 3.03

1.86 87.71 7.58 2.86

The given molecular weight was estimated from molecular weight standards. IgG and [sc(Fv)2]2 were found in fractions corresponding to the molecular weight between 100 and 200 kDa. The dimeric sc(Fv)2 is found in fractions representing a molecular weight of 35–100 kDa and low molecular weight molecules were found in fractions below 35 kDa.

uptake is closely linked to the pharmacological properties of the protein [24]. This was recently demonstrated in a study where nephrectomy resulted in an increased half-life of scFvs directed against the HER-2/neu [24]. In that study, constructs with an affinity within the range of 3.210 7 to 1.210 10 M reached tumor uptakes between 15% and 23% ID after 24 h. On the other hand, the same scFvs in nonnephrectomized mice only achieved tumor uptakes between 0.2% and 1.3% of the injected dose after the same time point [24]. Therefore, the increase in % ID seems to closely relate to the pharmacokinetic properties. The factors influencing the pharmacological properties of engineered proteins are not fully understood, but the design of the engineered antibody seems to play a crucial role [13,25]. It is believed that a shorter linker length makes the antibody less susceptible to proteolysis, increasing its half-

Fig. 6. Comparative analysis of the in vivo stability of sc(Fv)2 and [sc(Fv)2]2. At 2 h postinjection, a fraction of both constructs formed aggregates larger than tetrameric scFv. However, the major proportion of the aggregates formed by the sc(Fv)2 had a molecular size, comparable to the noncovalent [sc(Fv)2]2.

life [9]. This hypothesis is underlined by the observation of Adams et al. [26], where the plasma half-life of an iodinated HER2/neu diabody exceeded the half-lives of dimeric scFvs by several folds. Additionally, Wu et al. [27] found the plasma half-life of a diabody directed to CEA with t 1/2h to be 176 min, which is well above the plasma half-life of sc(Fv)2 with a t 1/2h of 70 min [12]. To investigate the proteolytic cleavage of the linker peptide of the scFvs, we performed size fractionations of the serum. Especially in the dimeric sc(Fv)2, after 2 h, a considerable fraction was found to be smaller then the original protein (Figs. 5 and 6). The approximate molecular weight of the small molecular weight fragments was found to be smaller than 3 kDa, clearly below the molecular weight of scFv or free V H/V L fragments, with an approximate molecular weight of 15 and 30 kDa, respectively [21]. Hence, a specific cleavage of the linker peptide, as predicted by Todorovska et al. [9] was not observed with the constructs described in this study. This is especially important with respect to the dose delivered to the kidneys, where the tetrameric [sc(Fv)2]2 is clearly superior to the divalent sc(Fv)2 (Fig. 3). The lack of the observed selective proteolytic cleavage of the [sc(Fv)2]2 does not produce dimeric and monomeric forms of the scFv that then would be subjected to glomerular filtration with a dramatic increase in the doses delivered to the kidneys [11]. Interestingly, we observed a tendency of the sc(Fv)2, as well as the [sc(Fv)2]2, to aggregate, so that after 2 h, 39.99% for the sc(Fv)2 and 19.95% for the [sc(Fv)2]2 were detected in fractions, representing a higher molecular weight than the injected constructs (Table 1). The nature of these aggregates, whether two sc(Fv)2 from a [sc(Fv)2]2 aggregate or whether the aggregates resemble unspecific aggregation to serum protein or immune complexes with circulating TAG72, is still subject to speculation. It is interesting to note that only

U.A. Wittel et al. / Nuclear Medicine and Biology 32 (2005) 157–164

53% of tetramer remains after 30 min but does not undergo significant degradation further at 4- and 6-h time points. On the other hand, the levels of intact dimer degrade rapidly from 68% at 30 min to 15% by 4-h postadministration. There is a possibility that tetramer is rapidly cleaved during initial period giving rise to dimer-like and other smaller fragments. Later, the degraded products may compete with the intact tetramer, thereby slowing down its degradation. Alternatively, tetramer degrades into dimer-like moieties, a fraction of which then aggregates to form complexes in the molecular weight range of tetramer. In this regard, it is important to mention that 40% of the circulating dimer aggregates into higher molecular weight complexes at 2 h postadministration. The in vivo appearance of aggregates is in contrast to previous in vitro observations with these scFv constructs. When analyzed in vitro at 378C in the presence of 1% BSA, the dimeric and tetrameric scFv constructs did not exhibit any aggregation or degradation even for up to 3 days [14,17]. The tendency of other genetically engineered scFvs (diabodies) to form aggregates in vitro has been previously reported by Wu et al. [27]. Apparently, due to fast clearance rates, the in vivo dimerization of monovalent scFvs was not observed previously [27]. Still, the observation of in vivo aggregation can become important, especially if the aggregates are dimerized sc(Fv)2. Then the dimerized products could be following the kinetics of the in vitro experiments and would preferably occur at high antibody concentrations, as used for therapeutic studies. These scFv aggregates then would not only loose their superior properties in tumor diffusion but would also become subject to rapid clearance by the spleen, with a decrease in availability in the blood. Hence, in addition to our findings, other factors must also contribute to the pharmacological properties of mono- or multivalent scFvs, since a significant deviation between experiments can generally be detected. One major factor to be considered is the choice of radioisotope used to label the proteins. Iodination seems to shorten the pharmacological half-life of scFvs in comparison to labeling with radiometals [14,17]. Wu et al. [27] observed a shorter serum half life of radiometal-labeled antibody fragments than the radioiodinated forms of the same construct [28]. Additional factors like the age of the animals, the size or type of the tumor or the occurrence of the antigen in the blood might be responsible for these observed differences. The last two factors are suggested by a clinical study, indicating that these deviations are not limited to genetically engineered antibodies or experimental situations alone. Patients with colorectal cancer and large liver metastasis showed especially high clearance rates of anti-CEA IgG (MN-14) [29], which is suggested to be related to increased hepatic enzyme activity or the occurrence of immune complexes with circulating CEA. With the onset of clinical trials of genetically engineered antibodies, we will be able to further understand and explain these differences in the pharmacological characteristics of

163

scFvs in different patient groups. After all, it is the rapid clearance of antibody fragments that makes them very attractive for radio-imaging. A better understanding of the mechanisms that underlie the pharmacological superiority of the engineered antibodies will help in designing improved targeting vehicles for optimal diagnosis and therapy. Acknowledgments The authors acknowledge Barbara J.M. Booth and Erik D. Moore for their technical support. We also acknowledge Ms. Kristi L.W. Berger (Eppley Institute) for editorial assistance. The CC49 scFv construct was a generous gift from Dr. Jeff Schlom of the Laboratory of Tumor Immunology at the National Cancer Institute and the Dow Chemical. References [1] Kaminski MS, Zasadny KR, Francis IR, Milik AW, Ross CW, Moon SD, et al. Radioimmunotherapy of B-cell lymphoma with [131I]antiB1 (anti-CD20) antibody. N Engl J Med 1993;329:459 – 65. [2] DeNardo SJ, DeNardo GL, Kukis DL, Shen S, Kroger LA, DeNardo DA, et al. 67Cu-2IT-BAT-Lym-1 pharmacokinetics, radiation dosimetry, toxicity and tumor regression in patients with lymphoma. J Nucl Med 1999;40:302 – 10. [3] Crippa F, Bolis G, Seregni E, Gavoni N, Scarfone G, Ferraris C, et al. Single-dose intraperitoneal radioimmunotherapy with the murine monoclonal antibody I-131 MOv18: clinical results in patients with minimal residual disease of ovarian cancer. Eur J Cancer 1995;31A: 686 – 90. [4] Juweid M, Sharkey RM, Behr TM, Swayne LC, Dunn R, Ying Z, et al. Clinical evaluation of tumor targeting with the anticarcinoembryonic antigen murine monoclonal antibody fragment. MN-14 F(ab)2. Cancer 1996;78:157 – 68. [5] Behr TM, Memtsoudis S, Vougioukas V, Liersch T, Gratz S, Schmidt F, et al. Radioimmunotherapy of colorectal cancer in small volume disease and in an adjuvant setting: preclinical evaluation in comparison to equitoxic chemotherapy and initial results of an ongoing phase-I/II clinical trial. Anticancer Res 1999;19:2427 – 32. [6] Jain RK. Transport of molecules in the tumor interstitium: a review. Cancer Res 1987;47:3039 – 51. [7] Jain RK, Baxter LT. Mechanisms of heterogeneous distribution of monoclonal antibodies and other macromolecules in tumors: significance of elevated interstitial pressure. Cancer Res 1988;48:7022 – 32. [8] Adams GP, Schier R. Generating improved single-chain Fv molecules for tumor targeting. J Immunol Methods 1999;231:249 – 60. [9] Todorovska A, Roovers RC, Dolezal O, Kortt AA, Hoogenboom PJ, Hudson PJ. Design and application of diabodies, triabodies and tetrabodies for cancer targeting. J Immunol Methods 2001;248: 47 – 66. [10] Muraro R, Kuroki M, Wunderlich D, Poole DJ, Colcher D, Thor A, et al. Generation and characterization of B72.3 second generation monoclonal antibodies reactive with the tumor-associated glycoprotein 72 antigen. Cancer Res 1988;48:4588 – 96. [11] Pavlinkova G, Beresford GW, Booth BJ, Batra SK, Colcher D. Pharmacokinetics and biodistribution of engineered single-chain antibody constructs of MAb CC49 in colon carcinoma xenografts. J Nucl Med 1999;40:1536 – 46. [12] Goel A, Beresford GW, Colcher D, Pavlinkova G, Booth BJ, Baranowska-Kortylewicz J, et al. Divalent forms of CC49 singlechain antibody constructs in Pichia pastoris: expression, purification, and characterization. J Biochem (Tokyo) 2000;127:829 – 36.

164

U.A. Wittel et al. / Nuclear Medicine and Biology 32 (2005) 157–164

[13] Batra SK, Jain M, Wittel UA, Chauhan SC, Colcher D. Pharmacokinetics and biodistribution of genetically engineered antibodies. Curr Opin Biotechnol 2002;13:603 – 8. [14] Goel A, Colcher D, Baranowska-Kortylewicz J, Augustine S, Booth BJ, Pavlinkova G, et al. Genetically engineered tetravalent single-chain Fv of the pancarcinoma monoclonal antibody CC49: improved biodistribution and potential for therapeutic application. Cancer Res 2000;60:6964 – 71. [15] Colcher D, Pavlinkova G, Beresford G, Booth BJ, Batra SK, Singlechain antibodies in pancreatic cancer. Ann N Y Acad Sci 1999:880: 263–80. [16] Yokota T, Milenic DE, Whitlow M, Schlom J. Rapid tumor penetration of a single-chain Fv and comparison with other immunoglobulin forms. Cancer Res 1992;52:3402 – 8. [17] Goel A, Baranowska-Kortylewicz J, Hinrichs SH, Wisecarver J, Pavlinkova G, Augustine S, et al. 99mTc-labeled divalent and tetravalent CC49 single-chain Fv’s: novel imaging agents for rapid in vivo localization of human colon carcinoma. J Nucl Med 2001; 42:1519 – 27. [18] Colcher D, Goel A, Pavlinkova G, Beresford G, Booth B, Batra SK. Effects of genetic engineering on the pharmacokinetics of antibodies. Q J Nucl Med 1999;43:132 – 9. [19] Laemmli UK. Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 1970;227:680 – 5. [20] Colcher D, Zalutsky M, Kaplan W, Kufe D, Austin F, Schlom J. Radiolocalization of human mammary tumors in athymic mice by a monoclonal antibody. Cancer Res 1983;43:736 – 42. [21] Colcher D, Bird R, Roselli M, Hardman KD, Johnson S, Pope S, et al. In vivo tumor targeting of a recombinant single-chain antigen-binding protein. J Natl Cancer Inst 1990;82:1191 – 7.

[22] Maack T, Johnson V, Kau ST, Figueiredo J, Sigulem D. Renal filtration, transport, and metabolism of low-molecular-weight proteins: a review. Kidney Int 1979;16:251 – 70. [23] Kortt AA, Dolezal O, Power BE, Hudson PJ. Dimeric and trimeric antibodies: high avidity scFvs for cancer targeting. Biomol Eng 2001; 18:95 – 108. [24] Adams GP, Schier R, McCall AM, Simmons HH, Horak EM, Alpaugh RK, et al. High affinity restricts the localization and tumor penetration of single-chain fv antibody molecules. Cancer Res 2001; 61:4750 – 5. [25] Colcher D, Pavlinkova G, Beresford G, Booth BJ, Choudhury A, Batra SK. Pharmacokinetics and biodistribution of geneticallyengineered antibodies. Q J Nucl Med 1998;42:225 – 41. [26] Adams GP, Schier R, McCall AM, Crawford RS, Wolf EJ, Weiner LM, et al. Prolonged in vivo tumour retention of a human diabody targeting the extracellular domain of human HER2/neu. Br J Cancer 1998;77:1405 – 12. [27] Wu AM, Chen W, Raubitschek A, Williams LE, Neumaier M, Fischer R, et al. Tumor localization of anti-CEA single-chain Fvs: improved targeting by non-covalent dimers. Immunotechnology 1996;2:21 – 36. [28] Yazaki PJ, Wu AM, Tsai SW, Williams LE, Ikler DN, Wong JY, et al. Tumor targeting of radiometal labeled anti-CEA recombinant T84.66 diabody and t84.66 minibody: comparison to radioiodinated fragments. Bioconjug Chem 2001;12:220 – 8. [29] Behr TM, Sharkey RM, Juweid MI, Dunn RM, Ying Z, Zhang CH, et al. Factors influencing the pharmacokinetics, dosimetry, and diagnostic accuracy of radioimmunodetection and radioimmunotherapy of carcinoembryonic antigen-expressing tumors. Cancer Res 1996;56: 1805 – 16.