Neuroblastoma is the most frequent cancer in children under 5 years of age. Gross peak annual sales for the EMA and FDA-approved anti-GD2 antibody dinutuximab (Unituxin®) for the treatment of neuroblastoma could reach €117.4 million in 2022, with about €38 million in the EU, €51 million in the US, and €28.4 million in the rest of the world. The global market for neuroblastoma therapeutics is estimated to about €1 billion per year. Importantly, patients treated with Unituxin® develop dose-limiting acute toxicity, raising the need for improving tolerance to this treatment. One possibility is to target an O-acetylated form of the GSL GD2 (OAcGD2) that is present on neuroblastomas, but absent from peripheral nerves, a situation that could be leveraged by our lectibody technology.
According to the proposed program, we have advanced according to the following steps.
Step 1. Bead-based screening
In a first set of experiments, we screened our lectibody library on bead-immobilized OAcGD2 that was provided to us by our biotech partner OGD2 Pharma. The quality of these functionalized beads was confirmed in binding experiments using an GD2/OAcGD2-specific antibody (Fig. 1). The depletion step in the screening protocol was performed on GD2 beads, and the selection cycle was repeated 3 times on OAcGD2-containing beads. Phage DNA was extracted from bacteria obtained after the last rounds of selection and used to transform competent TG1 cells from which 96 clones were picked. These were sequenced, and phages were isolated from each of them with the use of helper phages for characterization by FACS on Lan-1 neuroblastoma cells that express OAcGD2. Unfortunately, none of the identified clones was enriched during the screening procedure, and none bound to Lan-1 cells (not shown). We reasoned that the bead protocol was imposing constraints as to the way in which OAcGD2 is presented that may be incompatible with lectibody-based phage display screening. We therefore reverted to the cell-based screening protocol that we had used in our previous proof-of-concept studies.
Step 2. Cell-based screening on Lan-1 versus Lan-6 cells
Lan-1 neuroblastoma cells express OAcGD2, while Lan-6 cells do not (Fig. 2). We therefore applied the screening protocol with our lectibody library to 3 cycles of depletion on Lan-6 cells and selection on Lan-1 cells. As described for the bead-based screening, 96 clones were picked and characterized. In this case, some clones bound to OAcGD2-positive Lan-1 cells, and much less to OAcGD2-negative Lan-6 cells (Fig. 3). Their sequences are shown in Fig. 4. Based on this difference, we reasoned that these are promising candidates for further characterization. Surprisingly, these lectibodies bound to Lan-1 cells from which GSLs had been depleted using Genz-123346 (Fig. 5) as efficiently as to non-GSL-depleted Lan-1 cells (Fig. 6). This unexpected result called into question whether the target of these lectibodies was indeed OAcGD2, or as for that any other GSL.
Step 3. Cell-based screening on GSL-depleted Lan-1 versus control Lan-1 cells
The initial proof-of-concept lectibody screening study had been done on GSL expression inhibited mouse embryonic fibroblasts for the depletion step. We therefore set up another screen in which for depletion, GSL expression inhibited Lan-1 cells were used, followed by selection on Lan-1 cells that expressed normal GSL levels. We again were able to identify clones that specifically bound to the cells on which the screen had been done, i.e. GSL-expressing Lan-1 cells (not shown). However, once again the secondary validation experiments failed to confirm a GSL dependency for binding using Genz-123346-treated Lan-1 cells (not shown).
Step 4. In-depth characterization of a representative clone from Step 2
A representative clone from Step 2 was obtained by chemical synthesis and in vitro folding, as described in WO2020245321. This clone, termed B8, reproduced the preferential binding to OAcGD2 expressing Lan-1 cells over OAcGD2-negative Lan-6 cells (Fig. 7). Furthermore, also the unexpectedly GSL-independent binding to Lan-1 cells was reproduced (Fig. 8).
We then used another cell system to further analyze the GSL dependency question. HeLa cells express the natural STxB receptor, the GSL Gb3, and other GSLs. Both, wild-type STxB and B8 bound to HeLa cells, and GSL depletion led to the expected loss of wild-type STxB binding (Fig. 9). In contrast, the binding of B8 was not affected by GSL depletion (Fig. 9).
Our partner OGD2 Pharma used an overlay assay to directly probe for an interaction between B8 and OAcGD2. While positive controls with wild-type STxB binding to its natural receptor Gb3 showed the expected interaction, no binding could be detected between B8 and OAcGD2.
Very clearly, all these experiments confirmed that for yet unexplained reasons, the use of our lectibodies library to screen for OAcGD2 binders on Lan-1 cells leads to the isolation of clones that show GSL-independent binding.
In the Lectibodies program, we have invested the full work effort, as initially planned. Some unexpected results have led us to deviate from the initially planned path. All decisions have been discussed and taken in agreement with our industrial partner, OGD2 Pharma. The relationship with this partner has been fully honored throughout, and we currently continue to work with OGD2 Pharma.
