Imaging Brain Tumors by Targeting Peptide Radiopharmaceuticals through the Blood-Brain Barrier¹

Human brain tumors are the leading cause of cancer deaths of persons 15–34 years of age (1). Fatality rates could be diminished with early diagnosis of human brain tumors. However, present-day imaging modalities of brain tumors require a disrupted BBB^,3 and the BBB is generally intact in human brain tumors in the early stages (2, 3). Moreover, the brain adjacent tumor, which is the site of tumor growth and extension into normal brain, has an intact BBB (4). This makes it difficult to detect residual glioma in the perioperative or postoperative state with present-day imaging modalities. The biological properties of human gliomas include overexpression of peptide receptors relative to normal brain. Human gliomas overexpress a functional receptor for EGF (5, 6). On this basis, radiolabeled EGF has been proposed as a peptide radiopharmaceutical for imaging brain tumors (7). However, EGF does not cross the BBB (8) and cannot image brain tumors in the early stage when treatment is most needed, and the BBB is intact. Similarly, MAbs to the EGFR have been proposed as antibody radiopharmaceuticals for imaging human gliomas (9), but these also do not cross the BBB (10). In brain tumor sections in vitro, EGF labels human gliomas to a greater extent than that achieved with the EGF receptor MAbs (11).

EGF can be transported across the BBB in brain tumors if this peptide radiopharmaceutical is conjugated to a BBB drug delivery system (12). The latter is comprised of a peptide or a peptidomimetic MAb that undergoes receptor-mediated transcytosis through the BBB via one of several endogenous peptide transport systems localized within the brain capillary endothelial plasma membrane, which forms the BBB in vivo. The OX26 murine MAb to the TfR undergoes receptor-mediated transport through the BBB via the endogenous transferrin transport system (13, 14). Peptides that are not normally transported through the BBB, such as EGF, may be conjugated to brain drug delivery vectors, such as the OX26 MAb, using avidin-biotin technology (12). In this approach, the nontransportable peptide is monobiotinylated in parallel with the preparation of a conjugate of the OX26 MAb and SA, and this conjugate is designated OX26/SA. Previous studies (15) showed that EGF could be monobiotinylated with retention of high affinity for the human EGF receptor, when the biotin was attached to the EGF via a 14-atom bis (aminohexanoyl) linker, designated-XX-. However, when the EGF-XX-biotin was bound to OX26/SA, there was no significant binding to the EGF receptor because of steric hindrance caused by binding of the EGF to the brain delivery vector (15). The steric hindrance could be removed by replacement of the 14-atom-XX-linker with a >200-atom linker comprised of PEG of M_r 3400 molecular weight, designated PEG³⁴⁰⁰. When the EGF-PEG³⁴⁰⁰-biotin was bound to OX26/SA, there was retention of high-affinity binding of the conjugate to the EGFR. The PEG³⁴⁰⁰ linker restored the bifunctionality of the conjugate, and the conjugate bound both the EGF receptor, for imaging brain tumors, and the BBB transferrin receptor, for mediating BBB transport (15). Therefore, the present studies were designed to test the hypothesis that experimental brain tumors expressing the human EGFR could be imaged in vivo with an EGF peptide radiopharmaceutical that is enabled to undergo transport through the BBB because of conjugation to a peptidomimetic MAb that transcytoses through the BBB. Human U87 glioma cells were used to form experimental brain tumors in nude rats. The EGF was conjugated with DTPA to enable radiolabeling with the ¹¹¹In radionuclide. The EGF that is dual conjugated with DTPA and PEG³⁴⁰⁰-biotin is designated [¹¹¹In]-labeled DTPA-EGF-PEG³⁴⁰⁰-biotin.

The technique used to prepare [¹¹¹In]-labeled DTPA-EGF-PEG³⁴⁰⁰-biotin conjugated with OX26/SA has been described (12, 15). NHS-PEG³⁴⁰⁰-biotin was reacted with EGF at room temperature for 60 min in a molar ratio of 5:1, where NHS is N-hydroxysuccinimide and PEG³⁴⁰⁰ is polyethyleneglycol of M_r 3400. This mixture was reacted with DTPA dianhydride for 60 min in a molar ratio of 50:1. EGF, containing a single DTPA group and a single PEG³⁴⁰⁰-biotin, and designated DTPA-EGF-PEG³⁴⁰⁰-biotin, was purified by two Superose 12HR fast protein liquid chromatography columns in series, followed by ¹¹¹In chelation to the DTPA group. The conjugate of the OX26 MAb and recombinant SA, designated OX26/SA, was prepared with a thiol-ether linkage (16). Thiolated OX26 was reacted with m-maleimidobenzoyl N-hydroxysuccinimide ester-activated SA. The number of biotin binding sites per OX26/SA conjugate was 3.3 ± 0.3, as determined with a [³H]biotin binding assay (16). The conjugate of [¹¹¹In]-labeled DTPA-EGF-PEG³⁴⁰⁰-biotin and OX26/SA was prepared by mixing in a molar ratio of 1:1.6.

The human glioblastoma multiforme cell line U87 MG was obtained from the American Type Culture Collection (Rockford, MD) and was grown in monolayer culture (37°C, 5% CO₂) in MEM with 1 mm sodium pyruvate and 10% fetal bovine serum in 24-well cluster dishes. Cells were washed with 0.01m HEPES, 0.15 m NaCl (pH 7.4), and 0.1% BSA (HBSB buffer), followed by 15–120 min incubation at 37°C with 200 μl of HBSB buffer containing 1.0 μCi/ml (0.23 nm) of [¹¹¹In]-labeled DTPA-EGF-PEG³⁴⁰⁰-biotin with and without BBB transport vector OX26/SA in the presence and absence of 1 μm unlabeled EGF. After incubation, supernatants were aspirated, and the cells were washed two times with cold 0.01 m HEPES, 0.15 m NaCl (pH 7.4) and solubilized by the addition of 0.5 ml of 1 n NaOH and incubation at 37°C for 4 h. ¹¹¹In radioactivity was counted by gamma counter, and protein content of the cells was measured with the bicinchoninic acid protein assay (Pierce Chemical Co., Rockford, IL). Cell-associated binding (surface-binding and intracellular accumulation) was expressed as a percentage of medium radioactivity bound per mg of cell protein (12).

Male athymic nude rats (Hsd:RH-rnu) weighing 180–210 g were purchased from Harlan Sprague Dawley (Indianapolis, IN), and were implanted with U87 MG cells. A total of 15 animals was used for this study; 12 animals were examined with QAR, and 10 animals were investigated with immunocytochemistry. A burr hole was drilled 3 mm to the right of midline and 1 mm posterior to bregma. U87 MG cells were suspended in serum-free MEM containing 1.2% methylcellulose. Five μl of cell suspension (3 × 10⁵ cells) were injected into the right caudate nucleus at a depth of 4 mm over 10 s, using a 10-μl Hamilton syringe with fixed needle. The syringe was withdrawn after 5 min. At 16 days after U87 MG tumor cell implantation, nude rats were i.v. injected with either [¹¹¹In]-labeled DTPA-EGF-PEG³⁴⁰⁰-biotin (100 μCi) with OX26/SA and 16 nmol of unlabeled EGF or [¹¹¹In]-labeled DTPA-EGF-PEG³⁴⁰⁰-biotin (100 μCi) alone. The coinjection of unlabeled human EGF saturates peripheral uptake of the peptide radiopharmaceutical via the hepatic EGFR, making more radiopharmaceutical available for uptake by the brain (12). Two h after isotope injection, nude rats were sacrificed, and the brain was rapidly removed from the cranium and sectioned into 3-mm slabs. The slabs were plunged into powdered dry ice for rapid freezing over 30 min. Frozen sections of 15 μm were cut with a Bright cryostat and thaw-mounted onto glass coverslips. After drying, these coverslips were placed in apposition to Kodak Biomax MS X-ray film and exposed for 6 days at −70°C with an intensifying screen. The film was scanned with a Hewlett-Packard ScanJet IIcx/T flatbed scanner and transferred to Adobe Photoshop on a Power Macintosh 7100/66 microcomputer, followed by colorization with NIH image software, and prints were generated with a Kodak printer. The grayscale image of the brain was quantified with NIH Image, and means ± SE was computed from three separate measurements.

The expression of the EGF receptor in the U87 MG glioblastoma multiforme was investigated with immunocytochemistry both in cell culture and in vivo as brain tumors in athymic nude rats. The U87 MG cells cultured in a 35-mm dish were fixed in 100% methanol (−20°C for 20 min). The 15-μm frozen sections of experimental tumor in athymic nude rat brain were fixed in 100% acetone (−20°C for 20 min). Endogenous peroxidase activity was inactivated with 0.5% H₂O₂ for 5 min at room temperature, and the cells were blocked with 3% horse serum for 30 min at room temperature. Two different primary antibodies were used with comparable results. The antihuman EGFR mouse monoclonal antibody (Upstate Biotechnology, Lake Placid, NY) was used at 10 μg/ml. A second primary antibody used in this study was the 528 mouse monoclonal antibody to the human EGFR that had been generated from conditioned medium produced from 528 hybridoma cells, which were obtained from the American Type Culture Collection. The 528 antibody has been demonstrated in previous studies to illuminate the human EGFR of U87 MG cells in nude mice with immunocytochemistry (17). As isotype controls, mouse IgG1 and mouse IgG2a were used in parallel. After reaction with a biotinylated horse antimouse antibody that had been preabsorbed with rat immunoglobulin, the specimens were exposed to avidin and biotinylated peroxidase using the Vector ABC method (Vector Labs, Burlingame, CA).

The structure of the bifunctional conjugate, designated EGF-PEG³⁴⁰⁰-biotin/SA-OX26, is illustrated in Fig. 1 A, which depicts the bifunctionality of the conjugate. The two domains of the conjugate, the tumor EGFR binding domain and the BBB TfR binding domain, are separated by a >200-atom PEG³⁴⁰⁰ linker. The linker domain is further comprised of a biotin attached to the tip of the PEG moiety and SA, which is conjugated to the OX26 anti-rat TfR MAb via a stable thiol-ether bond. A radionuclide domain is attached to human EGF and is comprised of ¹¹¹In chelated to DTPA, which is conjugated to an EGF lysine.

The conjugate retains high-affinity binding to the human EGF receptor on U87 glioma cells, and this was initially investigated in cell culture as shown in Fig. 1, B and C. There was time-dependent binding of [¹¹¹In]-labeled DTPA-EGF-PEG³⁴⁰⁰-biotin, which is [¹¹¹In]-labeled EGF in Fig. 1,B. This time-dependent binding was nearly completely saturated by the inclusion of unlabeled human EGF (Fig. 1,B). When the radiolabeled EGF was conjugated to OX26/SA to form the overall construct shown in Fig. 1,A, there was similarly a time-dependent increase in binding of the conjugate to the U87 cells, and this was inhibited by unlabeled EGF but was not inhibited by unlabeled and unconjugated OX26 MAb (Fig. 1,C). The abundant expression of the human EGFR on the U87 cells in cell culture is shown in Fig. 1 D, which is an immunocytochemical study using a murine monoclonal antibody to the human EGFR. The basolateral pattern of the immunostaining is consistent with the expression of the EGFR on the plasma membrane of the U87 glioma cells.

The sustained expression of the immunoreactive human EGFR in the U87 cells in vivo in the form of experimental brain tumors in nude rats was demonstrated with immunocytochemistry as shown in Fig. 2,A for large tumors and in Fig. 2,C for small tumors. When [¹¹¹In]-labeled EGF was conjugated to OX26/SA and injected i.v. into the nude rats with U87 gliomas, the tumors were imaged in the case of both large tumors (Fig. 2,B) and small brain tumors (Fig. 2,D). Conversely, the brain tumor uptake of the [¹¹¹In]-labeled EGF, which was not conjugated to the BBB delivery system, was negligible (Fig. 2, E and F). Quantitation of the integrated density over same square area of tumor or normal brain was performed with NIH Image (see “Materials and Methods”). The integrated densities over the tumor and normal brain were 1079 ± 15 and 55 ± 9, respectively, for studies with the OX26-EGF conjugate (Fig. 2,B). Conversely, the integrated density over the tumor or normal brain was not different from the background integrated density, 0.49 ± 1.53, for studies with the unconjugated EGF (Fig. 2 F).

These studies describe an EGF peptide radiopharmaceutical that is conjugated to a BBB drug targeting system to enable transport of the EGF through the BBB in an experimental human U87 glioma. The EGF peptide radiopharmaceutical that is conjugated to the BBB targeting system successfully images small and large brain tumors (Fig. 2, B and D), but the unconjugated EGF radiopharmaceutical does not image the tumors (Fig. 2,F), because of lack of transport of the unconjugated peptide EGF through the BBB in vivo. The conjugate is a bifunctional molecule (Fig. 1 A). The EGF part of the conjugate binds the human EGFR and targets the human glioma cell, and the OX26 MAb part of the conjugate binds the rat TfR and targets the rat brain capillary endothelium, perfusing the tumor and forming the tumor BBB. The OX26 MAb part of the conjugate enables transport through the BBB, and the EGF part of the conjugate enables specific binding to the glioma. The EGF does not bind normal brain because of the paucity of EGFRs in normal brain, and the OX26 MAb does not bind the tumor because the OX26 MAb is specific for the rat TfR and does not bind the human TfR expressed on the human glioma cells. The bifunctionality of the conjugate is retained by the use of the extended PEG³⁴⁰⁰ linker, which physically separates the EGF and OX26 halves of the conjugate and enables binding of the conjugate to both the EGFR on the tumor cell and the TfR on the BBB of the tumor (15).

The EGF peptide radiopharmaceutical conjugated to the anti-TfR MAb (Fig. 1,A) was used in previous studies to image experimental brain tumors in Fischer rats in vivo (12). These animals had been implanted in the caudate putamen with C6 rat glioma cells that had been transfected previously with a gene encoding the human EGFR (18). This gene construct was under the influence of a glucocorticoid inducible promoter. The C6 glioma cells expressed the EGFR in tissue culture in the presence of 1 μm dexamethasone but did not express the human EGFR when grown in vivo in the form of experimental brain tumors (12). These previous studies demonstrated that EGF does not cross the BBB, and that EGF can be made transportable through the BBB by conjugation of the peptide to the BBB delivery system. However, successful in vivo imaging of brain tumors was not possible in previous studies because the human EGFR was not expressed on the C6 glioma cells in vivo (12). These previous studies showed that the expression of the EGFR on the tumor cells in vivo is a necessary condition for tumor imaging in vivo, and that tumor imaging does not arise from the uptake of the OX26 MAb conjugate nonspecifically by Fc receptors present on the tumor cell. Additional evidence for the specificity of the conjugate binding to the tumor cell EGFR is the radioreceptor assay, showing inhibition of conjugate binding to the U87 glioma in the presence of excess EGF but not excess OX26 MAb (Fig. 1 C). The OX26 MAb is specific for the rat TfR and does not bind the human TfR on the U87 glioma cells. Rather, the OX26 MAb binds the rat TfR expressed on the capillary endothelium, perfusing the tumor and comprising the BBB of the tumor.

The present studies use human U87 glioma cells, which express the human EGFR in vivo in experimental tumors in nude mice (17). Because the OX26 BBB drug delivery system is specific for rats and not mice,⁴ U87 experimental tumors were generated in the present studies in nude rats. Prior investigations have shown that nude rats may be used for experimental U87 gliomas (19).

In vivo imaging of brain tumors requires that the radiopharmaceutical be labeled with a stable radionuclide. In the present formulation, the radionuclide, ¹¹¹In, was chelated to human EGF via a DTPA linker, which was attached to one of the lysine moieties on the human EGF (12). A greater metabolic stability of the EGF radiopharmaceutical in vivo is achieved using the ¹¹¹In radionuclide, as opposed to a ¹²⁵I radionuclide (12). The SA was attached to the OX26 MAb via a stable thiol-ether linkage. The SA bound the biotin moiety at the tip of the PEG³⁴⁰⁰ strand, which in turn was conjugated to another internal lysine residue on the human EGF. The structure of the human EGF after the dual conjugation of DTPA and PEG³⁴⁰⁰-biotin was confirmed in previous studies (12) with matrix-assisted laser desorption ionization mass spectrometry (12). The [¹¹¹In]-labeled DTPA-EGF-PEG³⁴⁰⁰-biotin/SA-OX26 is shown in Fig. 1 A. Previous studies have characterized: (a) the structure of the conjugate; (b) the dual binding of the conjugate to the EGF and transferrin receptors; (c) the in vivo plasma pharmacokinetics of the conjugate; and (d) the metabolic stability of the conjugate in vivo (12).

The lack of significant transport of the unconjugated [¹¹¹In]-labeled EGF into the U87 human gliomas shown in Fig. 2,F parallels the absence of [¹¹¹In]-labeled EGF entry into C6 gliomas in Fischer rats in vivo reported previously (12). Although the BBB is disrupted in brain tumors including U87 experimental brain tumors (20), the disruption of the BBB is not sufficient to enable imaging of the brain tumor with unconjugated peptide radiopharmaceuticals such as EGF. Successful tumor imaging requires the conjugation of the peptide radiopharmaceutical to a BBB drug delivery system (Fig. 1 A). The tumor imaging is not derived from the coadministration of unlabeled EGF, which is given to inhibit peripheral degradation of the EGF conjugate (12). EGF does not cross the BBB (12), and the coadministration of unlabeled EGF with labeled EGF does not increase brain uptake of EGF.⁴

Once the EGF peptide radiopharmaceutical is delivered through the BBB via the endothelial TfR, the EGF is bound to the tumor EGFR. The sustained binding of the peptide to the EGFR may be attributable to the biological characteristics of EGF binding to its receptor. The human EGFR binds both EGF and transforming growth factor α. However, once internalized, transforming growth factor α is rapidly dissociated from the EGFR, whereas EGF remains bound to the EGFR after internalization (21). This property of EGF binding to its cognate receptor may underlie the high signal of the experimental brain tumor seen in Fig. 2 B. Another aspect of EGFR biology that may enhance the tumor image in patients with brain tumors is the up-regulation of the EGFR on human glioma cells exposed to radiation therapy (22). However, EGF peptide radiopharmaceuticals cannot access the EGFR on brain tumor cells, which are localized behind the BBB, if a BBB drug delivery system is not used in humans.

The present studies demonstrate that peptide radiopharmaceuticals such as EGF can be used to image brain tumors when the molecules are conjugated to a BBB drug delivery system. Brain tumors could be imaged with either a peptide radiopharmaceutical, e.g., EGF, or an antibody radiopharmaceutical, e.g., an anti-EGFR MAb. Previous investigations have attempted to use a radiolabeled anti-EGFR MAb to image human brain grade 3–4 gliomas (9, 23), but tumor uptake of nonspecific antibodies was also observed in these patients, indicating that the tumor was advanced with a fully disrupted BBB (9). However, specific imaging of the tumor in the early phase of the tumor growth when the BBB is intact is probably not possible unless the peptide or antibody radiopharmaceutical is conjugated to a specific BBB drug targeting system such as that used in the present studies. Although the OX26 MAb is specific for rats, similar studies can also be performed in humans using BBB transport vectors that bind to human BBB receptors. A MAb to the human insulin receptor is active in humans and Old World primates (24), such as the Rhesus monkey, and has a BBB transport coefficient 9-fold greater than that found with anti-TfR MAbs (25). In addition to neuroimaging brain tumors, the use of a BBB drug delivery system and a peptide pharmaceutical could also be directed toward the therapy of human brain tumors. Potato carboxypeptidase inhibitor, a T-knot protein, is an EGFR antagonist that blocks cell growth when the peptide is bound to the tumor EGFR (26). Virtually any neurodiagnostic or neurotherapeutic agent can be conjugated to the BBB drug delivery system for noninvasive brain drug delivery in vivo (27).

Daniel Jeong skillfully prepared the manuscript, and Margarita Tayag provided excellent technical assistance. Dr. Harry Vinters aided in the low magnification photography of the brain immunocytochemistry specimens.