Avidin Targeting of Intraperitoneal Tumor Xenografts

Background: Lectins (proteins that bind specific sugar molecules on glycoproteins and glycolipids) are expressed at various levels on the surface of tumor cells. Conjugation of cytotoxic agents to glycoproteins recognized by lectins could be useful in the treatment of tumors. Avidin (a highly glypositively charged protein found in egg white) contains terminal N-acetylglucosamine and mannose residues that bind to some lectins. In this study, we tested the ability of avidin, labeled through conjugation to radioactive biotin (a B vitamin), to target intraperitoneal tumors. Methods: Biotin was radioactively labeled with 111 In. Four tumor models (one ovarian, one lung, and two colon) were established in nude mice by intraperitoneal injection of cultured cancer cells. The following two approaches were used in the intraperitoneal administration of avidin: 1) radioactive biotin-avidin conjugates were injected and 2) avidin was in 1–24 hours before the injection of radioactive biotin (avidin pretargeting; avidin-biotin conjugates formed in vivo). The distribution of injected radioactivity in the tissues of treated animals was assessed. Results: Radiolabeled avidin localized highly and rapidly in the tumors. More than 50% of the administered dose of avidin-biotin conjugate accumulated per gram of tumor tissue 2 hours after injection; high tumor uptake of radioactivity was observed up to 24 hours after conjugate injection. In contrast, accumulation of radio-activity in normal tissues was low, yielding high tumor to nontumor ratios. With avidin pretargeting, accumulation of radioactivity in the liver, kidney, and spleen was reduced to a greater extent than that in the tumor, and tumor to nonratios were increased. Conclusions: Avidin may be a promising vehicle for the delivery of radioisotopes, drugs, toxins, or therapeutic genes to intraperitoneal tumors. [J Natl Cancer Inst 1998;90:25–29] The development of malignant ascites is not infrequent and is often a symptomatically troublesome event. Although many treatment alternatives, such as systemic chemotherapy, intraperitoneal chemotherapy, external-beam radiotherapy, and intracavitary radioactive colloidal instillation, have been explored with some success for palliation, no approach has demonstrated sufficient efficacy (1). Targeting therapy, including the use of antitumor monoclonal antibodies conjugated to radionuclides, chemotherapeutic agents, or toxins, has demonstrated increased cytotoxicity against tumors and reduced toxicity toward normal tissue. Trials of therapy for malignant ascites that involve the intraperitoneal injection of radiolabeled antibodies have been conducted in animals and patients (2-5). However, blood radioactivity and nonspecific uptake of radiolabels in normal tissue were high (2-5), which limited the administered dose for therapy. Avidin, a 66-kd, highly glycosylated, positively charged (pI [isoelectric pH value] = 10.5) protein present in egg white, contains terminal N -acetylglucosamine and mannose residues that bind some lectins (6,7). Several studies (8-12) have revealed that lectins are expressed at various levels on the surface of tumor cells and that they act to internalize their ligands. Therefore, glycoconjugates specifically recognized by these lectins could be used as carriers of therapeutics for the management of tumors (9-11). Furthermore, cationization of proteins, in general, has enhanced their cellular uptake (13-18). In this study, we administered radiolabeled avidin intraperitoneally to nude mice bearing intraperitoneal tumor xenografts. Markedly high tumor uptake of radioactivity and high tumor-to-nontumor target ratios were obtained with this method. Cell lines. The following 11 human cancer cell lines were used in this study: three ovarian cancer cell lines, HOUA, OVCAR3, and SHIN-3; three colon cancer cell lines, LoVo, LS180, and SW1116; two lung cancer cell lines, AOI and A549; two gastric cancer cell lines, KATOIII and MKN45; and a pancreas cancer cell line, PANC-1. LS180, OVCAR3, PANC-1, and SW1116 were supplied by the American Type Culture Collection (Rockville, MD). A549, HOUA, KATOIII, LoVo, and MKN45 were supplied by the Japanese Cancer Research Resources Bank (Tokyo, Japan). AOI and SHIN-3 were provided by M. Akiyama (Radiation Research Foundation, Hiroshima, Japan) and Y. Kiyozuka (Nara Medical College, Nara, Japan), respectively. The cells were grown in RPMI-1640 medium (Nissui Pharmaceutical Co., Tokyo, Japan) supplemented with 10% fetal calf serum (Life Technologies, Inc. [GIBCO BRL], Gaithersburg, MD) and 0.03% L-glutamine. Subconfluent cells were harvested by use of calcium-and magnesium-free phosphate-buffered saline (PBS) that contained 0.02% EDTA. Intraperitoneal tumor models. All procedures involving animals were carried out in accordance with the regulations for animal welfare in Japan. Four tumor models were established by intraperitoneal injection of3×106 LS180 or LoVo cells or1×107 SHIN-3 or AOI cells in 0.2 mL PBS into female BALB c-numice (Japan SLC, Inc., Hamamatsu). Many tumor nodules were found in the peritoneal cavity 11–25 days after injection. The combined tumor weight per mouse was 0.1–0.5 g at the time of experimentation. Radiolabeling of DTPA-biotin. Three micrograms of DTPA-biotin (diethylenetriaminepentaacetic acid O,w-bis [biocytinamide]; Sigma Chemical Co., St. Louis, MO), dissolved in 0.3 M Tris-HCl buffer (pH 7.6), was incubated with 18.5 MBq 111 InCl3 (Nihon Medi-physics Co., Nishinomiya, Japan) for 30 minutes at room temperature (19). Because more than 99% of the radioactivity was shown subsequently to bind to immobilized avidin gel (Pierce Chemical Co., Rockford, IL), the 111 In-labeled biotin was used without postlabeling purification. Radiolabeling of avidin, neutravidin, and streptavidin with 111 In-labeled biotin. Unlabeled avidin, neutravidin, and streptavidin (Pierce Chemical Co.) were mixed with 111 In-labeled biotin at molar ratios of 3–10:1 and left standing for 30 minutes at room temperature (19). Unconjugated 111 In-labeled biotin was then removed by means of chromatography on a PD10 column (Pharmacia Biotech Inc., Uppsala, Sweden). Purified 111 In-biotin-avidin, -neutravidin, or -streptavidin conjugates were used in the experiments. In vitro reactivity. The biotin-binding abilities of the radiolabeled proteins were tested by use of immobilized D-biotin gel (Pierce Chemical Co.). Cell binding studies were carried out by use of the cultured human cancer cell lines described above and cell suspensions derived from intraperitoneal tumors of nude mice with xenografts. The cell suspensions were prepared by passing the tumor specimens through a stainless steel mesh. Radiolabeled avidin, neutravidin, or streptavidin (=10 000 cpm10 ng) was incubated with increasing numbers of cancercells suspended in 100 fL PBS in 5.7 × 46-mm microcentrifuge tubes for 1 hour at 4 °C. After centrifugation at 10 000g, the supernatants were aspirated, the tubes were cut, and the tube bottoms, containing pelleted cells, were collected. The percentage of radioactivity bound to the cells was determined by use of an auto well gamma system (Aloka Co., Tokyo, Japan). Biodistribution study. All of the mice were assigned randomly to treatment groups; there were four to six mice in each group. In the first experiment, 20 fg 111 In-labeled avidin, neutravidin, or streptavidin was injected intraperitoneally into the tumor-bearing mice. In one group of LS180 tumor-bearing mice, 300 fg labeled avidin was used. The biodistribution of radioactivity was examined 2 or 24 hours later. The biodistribution of radiolabeled avidin was also performed in BALBmice without tumors. The data were expressed as the percentage of the injected dose per gram of tissue (% ID). In the second experiment, the biodistribution of radiolabeled biotin was examined after the injection of avidin in advance for pretargeting. LS180 tumor-bearing mice were injected with 300 fg unlabeled avidin intraperitoneally 1–24 hours before intraperitoneal administration of 0.3 fg 111 In-labeled biotin. One group of mice received the radiolabeled biotin without any pretreatment. Two hours after the radiolabel injection, the mice were killed and the biodistribution of the radioactivity was examined. Scintigraphy. For the imaging study, 3.7 MBq 111 In-labeled avidin was administered intraperitoneally to SHIN-3 tumor-bearing mice. Mice were anesthetized 2 and 24 hours after the radiolabel injection, and scintigrams of the whole body were obtained by use of a gamma camera (Searle, Chicago, IL) equipped with a pinhole collimator. Immediately after the imaging at 24 hours, the mice were killed, and the distribution of tumor nodules was examined. Then, all of the tumor nodules were collected and imaged again with other organs. Statistical analysis. Nonparametric tests were used to evaluate data from the experiments. The Mann-Whitney U test was used for two sample comparisons, whereas the Kruskal-Wallis test, followed by nonparametric comparisons of a control group with other groups (20), was used to compare the data for avidin with those for neutravidin or streptavidin in Table 1. Reported P values are from two-sided tests, and significance was defined as P <.05. In vitro reactivity. More than 99% of 111 In-labeled biotin bound to immobilized avidin gel, and 111 In-labeled avidin, neutravidin, and streptavidin (i.e., the conjugates with 111 In-labeled biotin [see“ Materials and Methods” section]) bound similarly to biotin gel. 111 In-labeled avidin showed substantially higher in vitro binding to each of the 11 tumor cell lines and to suspended cells from the four tumor xenografts examined than was observed with radiolabeled neutravidin or radiolabeled streptavidin (data not shown). Biodistribution of radiolabeled proteins. The biodistribution of radiolabeled avidin in nude mice bearing intraperitoneal tumor xenografts is shown in Tables 1 and 2. When injected intraperitoneally, radiolabeled avidin localized highly and rapidly in all of the tumors studied. Two hours after injection, the tumor uptake of radioactivity reached more than 50% ID, and a high level of radioactivity was also observed at 24 hours (Table 2). On the other hand, the level of radioactivity in most normal tissues was low. The blood concentration of labeled avidin was less than 0.3% IDat 2 hours and decreased thereafter. Therefore, markedly high tumor-to-nontumor ratios were obtained. The tumor-to-blood and tumor-to-bone radioactivity ratios were more than 400 and 100 at 2 hours after injection, respectively. With a high dose of avidin (i.e., 300 fg) in LS180 tumor-bearing mice (n = 6), the tumor uptake of radioactivity at 2 hours (61.12% ± 14.26 ID[mean ± standard deviation]) was less than that obtained with a dose of 20 fg avidin (96.42% ± 31.20) (Table 1; P = .045), but it was still high, demonstrating the high avidin-binding capacity of tumor cells. The accumulation of radiolabeled avidin in the kidneys, liver, and spleen was different between mice bearing different tumor types. Radioactivities in the organs of LS180 tumor-bearing mice were significantly higher than those in the organs of SHIN-3 tumor-bearing mice (P = .009; Tables 1 and 2). In mice without tumors, the biodistribution of radiolabeled avidin in normal tissues was almost the same as that in LS180 tumor-bearing mice at 2 and 24 hours after injection (data not shown). Biodistributions of radiolabeled neutravidin and streptavidin were compared with the biodistribution of radiolabeled avidin in LS180 tumor-bearing mice (Table 1). Radiolabeled neutravidin showed statistically significantly lower tumor uptake, lower liver and kidney accumulation, and a higher blood level of radioactivity. Radiolabeled streptavidin showed statistically significantly lower tumor uptake and higher blood accumulation of radioactivity, resulting in a lower tumor-to-blood ratio (Table 1). Table 1. Biodistribution of 111 In-labeled proteins at 2 hours in mice bearing intraperitoneal LS180 xenografts* Table 2. Biodistribution of 111 In-labeled avidin (20 fg) in mice bearing different intraperitoneal tumor xenografts* Biodistribution of radiolabeled biotin with avidin pretargeting. Table 3 shows the biodistribution of radiolabeled biotin in the two-step method, i.e., using unlabeled avidin pretargeting. When avidin was injected 1 hour before the biotin, the subsequent biodistribution pattern of the radiolabel was similar to that obtained when radiolabeled avidin was used in the one-step method (Table 1). As the pretargeting interval increased, the radioactivity in both tumor and nontumor tissue decreased. With a 4-hour pretargeting interval, radioactivities in the kidneys, liver, and spleen decreased more markedly than the radioactivity in the tumor. Therefore, the tumor-to-nontumor ratios increased significantly (Table 3; P =.009) compared with those obtained with radiolabeled avidin in the one-step method (Table 1). The tumor uptake of radiolabeled biotin 24 hours after avidin injection was similar to that obtained with radiolabeled biotin alone (Table 3; P = .083), which indicates that the avidin, though bound to the tumors, was no longer available for biotin binding. Scintigraphy. Images of 111 In-labeled avidin in a SHIN-3 tumor-bearing mouse are shown in . Only the abdomen can be seen (, A and B). A high uptake of radioactivity was observed in tumors as early as 2 hours (, A) and continued up to 24 hours (, B). Apart from the clusters of tumor nodules indicated (, B, arrowheads), there were many tumor nodules disseminated in the peritoneal cavity. By separating the tumors (428 mg) from the normal tissues completely, the high tumor uptake and the low normal organ distribution of radioactivity could be seen more clearly (, C), which is consistent with the biodistribution data. This study has demonstrated that intraperitoneally injected avidin effectively accumulates in intraperitoneal tumors and clears rapidly from the circulation. Therefore, markedly high tumor-to-nontumor target ratios are obtained. In particular, myelosuppression, which limits the administered dose in radioimmunotherapy, may not be a critical problem with this method. Avidin, when given systemically to mice, has been shown to clear rapidly from the circulation and to accumulate in the liver and kidneys (21). This protein has been used to remove biotinylated antibodies from the circulation in tumor imaging studies (22,23), but neutravidin and streptavidin were not effective in this regard (24). The liver contains binding sites for a variety of glycosylated ligands (11,25). It has been reported (18,25) that conjugation of carbohydrates to immunoglobulin G or to streptavidin results in more rapid blood clearance and in high liver accumulation of the proteins via binding to the hepatic lectin; deglycosylation of avidin was shown to decrease its blood clearance and liver uptake. Lowering the pI of avidin markedly reduced its kidney uptake and slowed its blood clearance, but its accumulation in the liver was not affected (18,26). Thus, the carbohydrate of avidin influences its liver accumulation, the positive net charge relates to its kidney accumulation, and both play a role in its rapid blood clearance. Table 3. Biodistribution of 111 In-labeled biotin at 2 hours in mice bearing intraperitoneal LS180 xenografts, with or without pretargeting of avidin (300 fg)* The mechanisms related to the binding of avidin to tumor cells are unclear. Findings from this study suggest that the high glycosylation and high pI of avidin may be relevant. Neutravidin, a modified avidin derivative that does not contain carbohydrate and has a relatively neutral pI (pI = 6.3), showed markedly lower tumor uptake of radioactivity than was seen with avidin. Studies (13-17) have revealed that cationization of albumin and immunoglobulin molecules enhances their absorptive-mediated endocytosis into cells. Polycationic charges on the proteins interact electrostatically with negatively charged residues on the cell surface, and this electrostatic interaction is believed to trigger the endocytosis process (14). We have recently shown (unpublished results) that avidin with a reduced pI has a significantly lower tumor uptake than unmodified avidin, suggesting that the high pI of avidin is related to its tumor accumulation. However, the modified avidin still showed higher tumor uptake than neutravidin, which suggests that the glycosylation of avidin is also relevant. Terminal N -acetylglucosamine and mannose residues, which avidin contains, are reported to bind lectins (6,7). Since tumors express lectins (8- 12), lectin-mediated uptake may be related to the specific tumor localization of avidin. Clarification of the mechanisms will be helpful in finding other vehicles that are more suitable for clinical application. Fig. 1. Scintigrams of a nude mouse bearing intraperitoneal SHIN-3 (ovarian cancer cell line) tumor xenografts. Whole body images were obtained 2 (A) and 24 (B) hours after intraperitoneal injection of 111 In-labeled avidin, with only the abdomen being visible. Immediately after the imaging at 24 hours, the mouse was killed and the distribution of tumor nodules was examined (in B, the arrowheads indicate clusters of tumor nodules). Then, all of the tumor nodules (428 mg) were collected and imaged again (C) along with other major organs (T = tumor; L = liver; ST = stomach; K = kidneys; SP = spleen; and I = intestine). The reasons for the different accumulations of radiolabeled avidin in the kidneys, liver, and spleen of mice bearing different tumor types are unclear. Nevertheless, by separating the administration of avidin and radiolabeled biotin in the two-step method, radioactivities in both the tumor and normal tissues decreased as the pretargeting interval of avidin increased (Table 3). However, with an avidin pretargeting interval of 4 hours, radioactivities in the kidneys, liver, and spleen were reduced more than in tumor, resulting in significantly improved tumorto-nontumor ratios. The fast clearance of biotin andthe more rapid internalization of avidin in normal tissues may be relevant. On the other hand, avidin was no longer available for biotin binding at 24 hours after injection, although the tumor localization of directly labeled avidin was high at that time. The biotinbinding sites of avidin may have been saturated by endogenous biotin, and avidin may have been internalized. In another study (27) with the same intraperitoneal tumor model and the pretargeting of 300 fg biotinylated antibody followed by intraperitoneal injection of 50 fg streptavidin, radiolabeled biotin was still bound to the tumor 24 hours later, indicating that endogenous biotin had little effect on the binding of radiolabeled biotin to the streptavidin in the tumor. Therefore, avidin bound to tumor cells may be internalized rapidly, which further supports the mechanism of lectin binding andthe high pI of a ligand (11,13–17). The problems in tumor targeting-therapy include toxicity to normal organs, limited capacity of the target, and failure of the therapeutic agents to reach intracellular sites of action (2–5,28). Therefore, the markedly high targeting efficiency, the large capacity of the target, and, possibly, its rapid internalization make avidin a very promising vehicle to carry radioisotopes, drugs, toxins or, therapeutic genes in the diagnosis andtherapy for various intracavitory tumors. Preliminary treatment of mice with malignant ascites was effective with 111 In-or 131 I-labeled avidin (unpublished observations). Therefore, with higher energy lemitters, such as 188 Re or 90 Y, therapeutic efficacy for larger tumor nodules might be achievable. Radioguided surgery is another possible application of this method because of its ability to rapidly and efficiently target tumors. Supported in part by a Grant-in-Aid for Scientific Research (08266230, 08671024) from the Ministry of Education, Science and Culture, and by Grantsin-Aid from the Fugaku Trust for Medical Research and Sankyo Foundation of Life Science, Japan. Manuscript received April 23, 1997; revised September 23, 1997; accepted October 24, 1997.