Development of a high affinity Anticalin® directed against human CD98hc for theranostic applications

Abstract

Enhanced amino acid supply and dysregulated integrin signaling constitute two hallmarks of cancer and are pivotal for metastatic transformation of cells. In line with its function at the crossroads of both processes, overexpression of CD98hc is clinically observed in various cancer malignancies, thus rendering it a promising tumor target. Methods: We describe the development of Anticalin proteins based on the lipocalin 2 (Lcn2) scaffold against the human CD98hc ectodomain (hCD98hcED) using directed evolution and protein design. X-ray structural analysis was performed to identify the epitope recognized by the lead Anticalin candidate. The Anticalin - with a tuned plasma half-life using PASylation^® technology - was labeled with ⁸⁹Zr and investigated by positron emission tomography (PET) of CD98-positive tumor xenograft mice. Results: The Anticalin P3D11 binds CD98hc with picomolar affinity and recognizes a protruding loop structure surrounded by several glycosylation sites within the solvent exposed membrane-distal part of the hCD98hcED. In vitro studies revealed specific binding activity of the Anticalin towards various CD98hc-expressing human tumor cell lines, suggesting broader applicability in cancer research. PET/CT imaging of mice bearing human prostate carcinoma xenografts using the optimized and ⁸⁹Zr-labeled Anticalin demonstrated strong and specific tracer accumulation (8.6 ± 1.1 %ID/g) as well as a favorable tumor-to-blood ratio of 11.8. Conclusion: Our findings provide a first proof of concept to exploit CD98hc for non-invasive biomedical imaging. The novel Anticalin-based αhCD98hc radiopharmaceutical constitutes a promising tool for preclinical and, potentially, clinical applications in oncology.

Keywords: CD98hc; PASylation; cancer theranostics; lipocalin; protein engineering; tumor targeting.

Figure 1

Phage display selection of lipocalin variants with affinity towards the ectodomain of hCD98hc. (A) Graphical depiction of CD98hc linked to a CD98 light chain via a disulfide bridge, illustrating the two main biochemical functions: amino acid transport and integrin signaling. The structure of the CD98hc ectodomain, which was employed as a recombinant protein for phage display selection of Anticalins is highlighted in pink. (B) Analytical SEC of the CD98hc-specific lipocalin variants P1E4, P3A12 and P3D11 that were selected via phage display, in comparison with wtLcn2 (all purified as soluble proteins expressed in E. coli), revealing monomeric protein preparations (minor aggregate contaminations elute at the void volume, V₀). (C) SPR real-time binding analysis (single cycle) of the high-affinity variant P3D11 versus hCD98hcED, demonstrating a picomolar dissociation constant (see Table 1).

Figure 2

X-ray structure of an Anticalin in complex with the human CD98hc ectodomain. (A) Cartoon representation of the P3D11•hCD98hcED complex (P3D11 pink, hCD98hcED gray). The epitope loops L1 and L2 of hCD98hc are colored green and cyan, respectively. (B) Illustration of the interface after separating the two complex partners. Molecular contact surfaces are colored according to the opposite structure (cf. panel A) with hydrogen bond donors and acceptors highlighted blue and red, respectively, and hydrogen-bonded water molecules depicted as red spheres. (C) Surface representation of the P3D11•hCD98hcED complex including a model of the four complex N-glycans (light blue spheres) on the target protein.

Figure 3

Affinity and stability engineering of the lipocalin variant P3D11 via bacterial surface display. (A) FACS analysis of E. coli cells presenting wtLcn2 (as negative control, left), the initial lipocalin variant P3D11 (as reference, middle) and the error-prone library (right), each incubated with 100 nM (cycle 1, upper panel) or 1 nM (cycle 6, lower panel) biotinylated hCD98hcED. (B) Single clone FACS analysis, using 1 nM biotinylated hCD98hcED, of P3D11 versus the second generation variant, D11.1, obtained after six selection cycles, in comparison with wtLcn2. (C) Thermal denaturation of P3D11 and its improved version D11vs measured by CD spectroscopy.

Figure 4

Cytofluorometric and immunofluorescence microscopy of CD98hc expressing human cancer cell lines using a PASylated Anticalin. (A) Flow cytometry analysis of cell lines from colorectal adenocarcinoma (Caco-2), prostate carcinoma (DU-145 and PC-3) and B-cell lymphoma (SU-DHL-4, Raji and Ramos) with Sulfo-Cy5.5-labeled PASylated D11vs (blue histograms). In a control experiment, a 10-fold molar concentration of either unlabeled D11vs-PAS200 or soluble hCD98hcEDg was co-applied for competition (red and yellow histograms, respectively). Untreated cell lines are shown in gray. (B, C, D) Immunofluorescent detection (red) of hCD98hc using Sulfo-Cy5.5-labeled PASylated D11vs shown for (B) Ramos, (C) PC-3 and (D) Caco-2 cells (middle). In control experiments, the cells were treated with Sulfo-Cy5.5-labeled wtLcn2 (left), or binding of Sulfo-Cy5.5-labeled PASylated D11vs was competed with a 10-fold molar concentration of unlabeled D11vs-PAS200 (right). Cell nuclei were counterstained with DAPI (blue).

Figure 5

In vivo PET/CT imaging of a prostate carcinoma xenograft model. (A) Mice (♀) bearing PC-3 xenografts were injected i.v. with 2.85 ± 0.15 MBq D11vs-PAS200-Dfo•⁸⁹Zr and subjected to PET/CT imaging 24, 48 and 72 h post injection. Prominent signals were apparent in the xenograft tumor (arrowhead) as well as liver (l), kidneys (k), bladder (bl) and joints (*). (B) PET images were analyzed by threshold-based image segmentation (lower threshold at 50% of the hottest voxel), and activity trajectories were plotted. Data for mice with (green) and without (red) blocked target are shown. Epitope blocking was accomplished by i.v. injection of a 100-fold molar amount of the unlabeled PASylated Anticalin 2 h prior to application of the protein tracer. (C) Consecutive 10 µm cryosections were prepared from the tumor tissue followed by autoradiography and H&E staining. (D) The second half of each tumor was embedded in paraffin and sections were stained for CD98hc and CD31 (blood vessels), using appropriate antibodies, as well as with H&E.

Figure 6

In vivo and ex vivo PET/CT imaging and biodistribution analysis. (A) Mice (♂) bearing PC-3 tumors were injected with 3.96 ± 0.12 MBq D11vs-PAS200-Dfo•⁸⁹Zr whereas mice in the blocked group (B) additionally received a 250-fold molar amount of unlabeled D11vs-PAS200 2 h prior to the protein tracer injection. For each mouse the in vivo PET/CT, ex vivo PET/CT and a photograph of the explanted tumor are depicted. (C) For biodistribution analysis, the mouse organs were dissected and their radioactivity and weight were quantified to allow calculation of the %ID/g values.