Generation of nanobodies targeting the human, transcobalamin-mediated vitamin B12 uptake route

Cellular uptake of vitamin B12 in humans is mediated by the endocytosis of the B12 carrier protein transcobalamin (TC) via its cognate cell surface receptor TCblR (or CD320), encoded by the CD320 gene(1). Because CD320 expression is associated with the cell cycle and upregulated in highly proliferating cells such as cancer cells(2–4), this uptake route is a potential target for cancer therapy(5). We developed and characterized four camelid nanobodies that bind TC or the interface of the TC:TCblR complex with nanomolar affinities. We determined X-ray crystal structures of all four nanobodies in complex with TC:TCblR, which enabled us to map their binding sites. When conjugated to a toxin, three of these nanobodies are capable of inhibiting the growth of HEK293T cells and therefore have the potential to inhibit the growth of human cancer cells. We visualized the cellular binding and endocytic uptake of the most potent nanobody (TCNB4) using fluorescent light microscopy. The co-crystal structures of TC:TCblR with another nanobody (TCNB34) revealed novel features of the interface of TC and the LDLR-A1 domain of TCblR. Our findings rationalize the structural basis for a decrease in affinity of TC-B12 binding caused by the TCblR-Glu88 deletion mutant.


Introduction
Because vitamin B12 is a cofactor of enzymes involved in recycling of folates for DNA synthesis (5), the need for B12 is elevated in highly proliferating cells such as pluripotent cancer cells (4). This correlates with the overexpression of the CD320 protein, the cellular receptor (TCblR, sometimes called CD320) of transcobalamin-B12 complex (4,6). Targeting the cellular uptake route for vitamin B12 for potential cancer diagnosis or therapy has been described in previous studies (5), where strategies included the derivatization of vitamin B12 (7) or the coupling of a receptor binding antibody to saporin (8,9). In assays with cultured tissue, a saporin-conjugated antibody was able to inhibit a broad spectrum of cancer cell lines (9). Despite these efforts, no compound targeting the B12 uptake route is currently approved for clinical use (10). As an alternative way to exploit TCblR for potential cancer diagnosis or therapy, we have developed a set of nanobodies, which are heavy-chain-only antibody fragments that were originally derived from camelids and consist of a single IG domain. Nanobodies have found great popularity in basic science and biomedicine (11,12) because they are highly stable, can be loaded with drugs, are less immunogenic than therapeutic antibodies, able to penetrate tissues, and can conveniently be produced at large scale in bacterial expression hosts including E. coli (13). Many nanobodies have been raised against cancer cell markers (12), and several companies have been funded to develop nanobody-based drugs. In 2019 Ablynx (Ghent, BEL) obtained the first FDA approval for a therapeutic nanobody while ten nanobodies from various companies were in clinical trials (14).
We hypothesized that a nanobody-drug conjugate targeting the cellular vitamin B12 uptake route might be advantageous over previous approaches and, therefore, could have potential use in cancer therapy and diagnosis. We generated a set of nanobodies targeting the TCreceptor complex, characterized them biophysically and using X-ray crystallography, and probed their uptake into human embryonic kidney (HEK293) cells.

Expression and purification of TC:TCblR
TC:TCblR complex was produced by co-transfection of the respective gene constructs into SF9 cells and purified as described previously (15).

Nanobody selection, expression and purification
Nanobodies were generated, expressed and purified as described previously (16,17). with the following modifications: For the panning TC:TCblR complex was immobilized on a streptavidin matrix via biotinylation of TCblR as described previously (15). Purified nanobodies were desalted into 20 mM Tris pH 7.5, 0.5 mM CaCl2 and 150 mM NaCl.
To generate nanobody-saporin conjugates, a C-terminal AviTag TM (GLNDIFEAQKIEWHE), preceding the His6 Tag and flanked by flexible GGGS linkers, was fused to the nanobodies.
Biotinylation was performed as described previously (15 to avoid bias by color changes in the medium and inhibition of growth respectively total cellular viability was assayed using the MTS Assay Kit (Abcam). 100% viability was defined as the readout from non-treated cells (addition of PBS) and 0% viability was defined by the readout performed on wells where no cells where seeded. Measurements were conducted in triplicates and positive and negative controls were performed in quintuplicates.

Crystallization and data collection
After preincubating TC:TCblR at 10 mg/ml with a 1.  (23) and Refinement was performed in Phenix (24). To avoid bias from the comparably low-resolution map of TCNB26:TC2-TCblR, for this structure only rigid body and B-factor refinement was performed.

Figure preparation and data analysis
Protein sequence alignments were performed in CLC Genomics Workbench. Cell viability assay data were analyzed and plotted in GraphPad Prism. Protein structure images were generated in PyMol(25).

Cellular uptake of TCNB4
We selected the highest affinity binder, TCNB4, to test its potential for TC:TCblR-mediated cellular uptake. We incubated fluorescently labeled TCNB4 with HEK293 cells that constitutively overexpressed GFP-tagged TCblR (GFP-TCblR) (19). Using light microscopy, we observed co-localization of TCNB4 and TCblR (Fig 2). We further observed that the nanobody and the receptor protein were endocytosed. To test whether endocytosed nanobody was degraded, we repeated the experiment with cells overexpressing a fusion construct of TC and red fluorescent protein (DsRed)(27). Over the course of four hours, we observed the loss of DsRed fluorescence, indicating lysosomal degradation of DsRed-TC. For the labeled TCNB4 however, we did not observe any loss of the fluorescence signal, indicating that the nanobody remained intact upon endocytic internalization (SI Figure 1). Given that the fluorophore has an extended life span, some degradation of the nanobody cannot be excluded.

Inhibition of cell growth by saporin-conjugated nanobodies
To investigate whether nanobodies carrying toxic cargo could kill HEK293T cells we fused them to the cytotoxic, plant-derived enzyme saporin, which acts by irreversibly inactivating eukaryotic ribosomes(28). The fusion was accomplished by inserting a biotinylation site (AVI-tag) into the nanobodies and biotinylating them using the enzyme BirA(29). The biotinylated nanobodies were then coupled with biotin-reactive saporin (Streptavidin-ZAP, Advanced Targeting systems). The resulting nanobody-saporin conjugates were used for cytotoxicity assays.
We seeded highly proliferating HEK293T cells at low density to stimulate the expression of CD320 (8). When adding the saporin-conjugated nanobodies at a concentration of 50 nM, complete inhibition of cell growth was observed for TCNB4 (Fig. 3a). This effect was not observed for unconjugated TCNB4 or unconjugated saporin-ZAP (Fig. 3b). saporinconjugated TCNB11 and TCNB34 also inhibited cell growth, but to a lower extent than TCNB4. Interestingly, saporin-conjugated TCNB26 did not show significant cytotoxicity.
This might be due to the lower affinity of TCNB26 compared to the other nanobodies.

Structural characterization of nanobody binding
To delineate the interaction interfaces, we determined individual crystal structures of all four nanobodies in complex with TC:TCblR (Fig. 4, SI Table 1). The structures revealed three distinct binding epitopes. TCNB11, TCNB26, and TCNB34 bind to TC only, whereby TCNB11 and TCNB34 share overlapping binding epitopes. In contrast, TCNB4 binds at the interface of TC and TCblR and therefore specifically recognizes the TC:TCblR complex.
TCNB4 interacts with TCblR via complementarity-determining region 1 (CDR1) and a nonvariable loop (Fig. 4, SI Figure 2). CDR3 mediates most interactions with TC. A disulfide bond between CDR2 and CDR3 restricts the conformational freedom of CDR3 and presumably helps with its precise positioning. We observed density likely reflecting a Ca 2+ ion coordinated by backbone carbonyl groups of CDR3 and a glutamate side chain of CDR1. The coordinated ion appears to stabilize the conformation of the CDR3 loop further. TCNB11 binds its TC by means of contacts involving CDR2 and CDR3 (Fig. 4, SI Figure 3). CDR3 twists a flexible loop of TC near His173 by 180°. The flexibility of this loop has been observed previously: While His173 replaces the upper coaxial ligand of hydroxocobalamin(30), the cyano-group cyanocobalamin stays bound and displaces His173 from the B12 binding site (15). TCNB26 binds to TC with all three CDRs (Fig. 4, SI Figure 4). The relatively loosely packed binding interface might explain the comparably lower binding affinity of this nanobody. TCNB34 binds TC predominantly via CDR1 and CDR3 (Fig. 4d, SI Figure 5). CDR2 forms a disulfide bond to CDR3, similar to TCNB4.

Novel structural insight into the TC:TCblR complex
The crystal structure of TC:TCblR bound to TCNB34 improved the previously reported resolution of 2.1 Å(15) to 1.85 Å (SI Table 1). While the resolution was higher overall, two regions in particular were better resolved (Fig. 5) (15). We also observed a network of ordered water molecules that stabilize the complex of TC and TCblR by H-bonding. Second, our high-resolution map provided a more detailed view of the region around the disease-related residue Glu88 in CD320(32). Deletion of this residue has been associated with vitamin B12 deficiency(32). Our structure shows that the adjacent residue  Figure 3c).

Discussion
We have generated four high-affinity nanobodies targeting TC or the TC:TCblR complex.
Given that these nanobodies can enter human cells and inhibit cell growth when fused to a toxin, they may have a value in diagnostic or therapeutic approaches. As the nanobodies bind distinct epitopes, fusing two of them (for example, TCNB4 and TBNB11) might increase the affinity or avidity of TC:TCblR:nanobody complex formation in vivo. One might expect that such a bipartite nanobody could bind TC in the blood before binding to TCblR. To determine the applicability of our nanobody-drug conjugates, further studies in vivo in native environments will be required. Future investigations may also test the fusion of our nanobodies to other toxins or drugs for cancer therapy or radionuclides for tumor localization using imaging studies.
Our crystal structures revealed new features of the TC:CD320 complex, which helped further rationalize the lysosomal dissociation of TC-Cbl from TCblR at low pH (15)    of TC appears to bind to some extent to the cell surface, which may be due to direct (and low affinity) interaction with the extracellular domain of TCblR (panels A), but without internalization (the merge of panels A). Nanobody pre incubated with TC-Cbl appears to bind specifically to CD320-GFP (panels B) and is associated with CD320-GFP internalization (merge panels B). However, the CD320-GFP and the red nanobody do not appear to degrade at 4h as seen by intense fluorescence of TCblR-GFP and the nanobody.     TCNB4  78  TCNB11  78  TCNB26  81  TCNB34   100%   0%   Conservation   158  TCNB4  156  TCNB11  155  TCNB26 160 TCNB34 20 Figure 5