Dual targeting of CD19 and CD22 against B-ALL using a novel high-sensitivity aCD22 CAR

CAR T cells recognizing CD19 effectively treat relapsed and refractory B-ALL and DLBCL. However, CD19 loss is a frequent cause of relapse. Simultaneously targeting a second antigen, CD22, may decrease antigen escape, but is challenging: its density is approximately 10-fold less than CD19, and its large structure may hamper immune synapse formation. The characteristics of the optimal CD22 CAR are underexplored. We generated 12 distinct CD22 antibodies and tested CARs derived from them to identify a CAR based on the novel 9A8 antibody, which was sensitive to low CD22 density and lacked tonic signaling. We found no correlation between affinity or membrane proximity of recognition epitope within Ig domains 3–6 of CD22 with CART function. The optimal strategy for CD19/CD22 CART co-targeting is undetermined. Co-administration of CD19 and CD22 CARs is costly; single CARs targeting CD19 and CD22 are challenging to construct. The co-expression of two CARs has previously been achieved using bicistronic vectors. Here, we generated a dual CART product by co-transduction with 9A8-41BBζ and CAT-41BBζ (obe-cel), the previously described CD19 CAR. CAT/9A8 CART eliminated single- and double-positive target cells in vitro and eliminated CD19- tumors in vivo. CAT/9A8 CART is being tested in a phase I clinical study (NCT02443831).


Surface plasmon resonance
Recombinant anti-CD22 antibodies in IgG format, were captured on flow cells 2, 3 and 4 on a Series S Protein A sensor chip (Cytiva 29650263) to a density of 40-60 RU using a Biacore T200 instrument.HBS-P+ buffer was used as running buffer is all experimental conditions.Recombinant purified CD22 (Acro Biosystems SI2-H5228) at known concentrations was used as the 'analyte' and injected over the respective flow cells with 150 s contact time and 300s dissociation at 30 µl/minute of flow rate with a constant temperature of 25°C.In each experiment, flow cell 1 was unmodified and used for reference subtraction.A '0 concentration' sensogram of buffer alone was used as a double reference subtraction to factor for drift.Data were fit to a 1:1 Langmuir binding model.Since a capture system was used, a local Rmax parameter was used for the data fitting in each case.

Epitope binning
Recombinant anti-CD22 antibodies were immobilized on flow cell 2 of channels 1, 3, 5 and 7 of a Series S CM5 sensor chip (Cytiva 29104988) to a density of 820-1350 RU using a Biacore 8k instrument.HBS-P+ buffer was used as running buffer is all experimental conditions.Recombinant purified CD22 (Acro Biosystems SI2-H5228) at 200 nM (analyte 1) was injected over the flow channels for 150s at 30 µl/min, followed by 100 nM of the challenging anti-CD22 antibody (analyte 2) for 150s at 30 µl/min.A dissociation phase of 300s was included at the end.A condition without analyte 2 injection was used as baseline to assess analyte 2 contribution to RU.A '0 concentration' sensogram of buffer alone was used as reference subtraction to factor for drift.

Cell line engineering with STOP/SKIP
In order to obtain low CD22 expressing cells, we introduced a STOP/SKIP sequence upstream of the transgene.A STOPSKIP motif is a DNA sequence that bears a stop codon followed by a read-through sequence (e.g.CATG), with the aim of reducing the translational efficiency of any downstream transgene (Loughran et al. 2014).Different STOP codons led to different expression stringency and thus expression levels.

Antigen density measurement
Antigen density was determined by labelling the cell lines with aCD22 (BioLegend; 302506), aCD19 (BioLegend; 302208), or isototype (BioLegend; 400112) antibodies conjugated to PE. Concomitantly, we ran Quantibrite beads (BD Biosciences; 10626384), based on the beads' fluorescence and know density we created a standard curve.For each sample, the median fluorescence for PE was measured and their density calculated based on the Quantibrite standard curve.

Restimulation Assay
We set-up a co-culture of 5x10 4 effectors in the presence of 5x10 4 targets (SupT1 NT, SupT1 CD22 Mid , or CD22 Low ).Fresh targets were introduced every 4-6 days for stimulation 2 and 3. Stimulation 1 (Stim 1) was measured on day 3 after the assay set-up.The lysis has been normalised to that of SupT1 NT for each time-point.

Engineering Nalm6 CD19 KO
Nalm-6 cells (DSMZ clone ACC 128) were transduced with retrovirus encoding blasticidin S deaminase and Streptococcus pyogenes Cas9.Selection for transduced cells was carried out by culturing the cells in 20 mg/mL blasticidin (Invivogen) for 2 weeks.To generate a CD22 knockout, 1x106 Cas9-expressing Nalm-6 cells were nucleofected, using SF buffer and pulse code CV104 (Lonza), with 100 pmol of sgRNA (IDT) targeting the CD22 gene (5'-GAAACCCTCTACGCCTGGGA-3').Single cell clones were established and then transduced with retrovirus encoding the sort select marker RQR8 and chimeric CD22/CD19, consisting of the CD22 ectodomain fused to the CD19 transmembrane and truncated cytoplasmic domain (residues 333-556 deleted).These cells were nucleofected with sgRNA (IDT) targeting the CD19 gene (5'-TGGAATGTTTCGGACCTAGG-3') to generate CD19 knockout cells.A) The novel aCD22 CARs were screened into a Fab format, wherein the heavy and light chains were expressed separately.The light chain was fused to the IgK constant kappa domain, while the heavy chain was fused to the constant heavy chain of IgG followed by CD28 transmembrane, 4-1BB and CD3ζ.B) The kinetics of the binders were determined with Biacore technology.C) Epitope mapping was performed utilising 9A8.A) In order to obtain low expressing CD22 target cells lines we introduced the STOP/SKIP technology, which comprises a STOP codon followed by a SKIP sequence leading to the ribosome skipping the STOP codon.The STOP/SKIP upstream of the transgene reduces the expression of the transgene.B) The expression of CD22 on the engineered SupT1 was measured by flow cytometry.The cells were labelled with either isotype control or aCD22 antibody, as well as the marker gene upstream of the transgene was labelled to ascertain the cassette expression.C) The antigen density was calculated based on the flow cytometry while concomitantly using with Quantibrite beads to obtain a standard curve.The CD22 level on SupT1 CD22 Low was only subtly detectable by flow cytometry.D) Hence, we run western blot with different concentration of cell lysate loaded to investigate the presence of CD22 in SupT1 CD22 Low .E) Finally, in order to ascertain the presence of CD22 on the SupT1 by flow cytometry, we amplified the signal using QIFIKIT kit for one or two rounds of amplification.CD19 was downregulated in Nalm-6 WT cells by CRISPR technology.Flow-cytometry assessment demonstrated the lack of CD19 expression in the engineered Nalm-6 CD19 KO cells, while the expression of CD22 antigen is still present.

Figure S1 :
Figure S1: Novel aCD22 binders and their biophysical properties

FigureFigure S2 :
Figure S2: Engineered SupT1 cells to express high and low densities of CD22

Figure
Figure S3: Binder Screening