CD74 CAR-T vector construction and expression

A second-generation CAR was designed, where single chain variable fragments (scFV) of the mouse anti-human CD74 was joined with 4-1BB and CD3ζ chain via the CD8 hinge region and transmembrane domain. A CMV promoter was added in front of the entire sequence for constitutive expression. The sequence was then codon-optimized, synthesized (Twist Bioscience, OR), and subcloned into a lentiviral vector pCDH (SBI Bioscience, CA). Lentiviral particles were made by transfecting Lenti-X 293T cells (Clontech, Takara Bio USA Inc, CA) with the CAR construct and packaging plasmids including PsPax2, (Addgene #12260) and pMD2.G (Addgene #12259), via Lipofectamine 2000 (Invitrogen, MA). The viral supernatant was harvested 48 h post-transfection, passed through a 0.22 um filter, and then snap-frozen at − 80 °C until it was used.

In silico scFV optimization and mutagenesis

Amino acid paratope in the regions of complementarity-determining region (CDR)1, CDR2, and CDR3 was first identified by ProABC-2. The scFV was then in silico reconstructed as PDB by ABodyBuilder2 [24] and prepared for antigen docking (CD74, PDB# 1IIE). The scFV was then docked to the antigen using HADDOCK 2.4 [25]. To refine the docking model, alanine scanning on the amino acid residues on each CDR was performed. Among all model cluster candidates, the one with the lowest HADDOCK score was picked for computational mutagenesis study (Additional file 1: Figure S1). The binding residue candidates were substituted and docked against CD74 antigen with all possible 20 amino acids. By using Eris molecular suite [26], mutations were introduced in the scFV regions on those binding residues. The estimated free energies of mutant conformations were compared with the wild-type and lead mutants were selected for site-directed mutagenesis and downstream functional assays. Mutagenesis on the CDR regions were performed by GeneMorph II EZClone domain mutagenesis kit (Agilent, CA) according to the manufacturer’s instructions. Three CAR constructs with anti-CD74 scFV (named 543, 553, and 563) from public domains were also developed and used for comparison. The mutant plasmid was transformed, and lentiviral production was performed for each CAR mutant. To facilitate the screening, each mutant was displayed on Jurkat cells (T-ALL cell line), and the expression level of the CAR was uniquified by sorting the GFP+ cells from each mutant by flow cytometry and CD3ζ by immunoblot (Additional file 2: Figure S2A and B). The mutant performance was assessed and compared with the parent 74bbz CAR using 4 parameters: (1) functional binding affinity to a chimeric CD74 extracellular domain (ECD)-Fc fusion protein (Additional file 2: Figure S2C); (2) CD69 activation marker expression upon engaging CD74+ target cells; (3) repeated antigen proliferation assay; and (4) in vitro cytotoxicity assay against CD74+ target MCL cell line Mino cells.

Cell culture and isolation

The MCL cell lines JeKo-1, Mino, Sp53, UPN-1, Granta-519, and Z-138 were obtained from ATCC (Manassas, VA) and cultured under the manufacturer’s instruction. All the cell lines used were routinely tested for mycoplasma with MycoAlert (Lonza, MA) and passaged for no more than 2 months. All the cell lines were regularly STR authenticated. Human peripheral blood mononuclear cells (PBMCs) were isolated from healthy blood donors by Ficoll-Paque Plus (GE Healthcare Life Science, PA) gradient density under an Ohio State University Institutional Review Board approved protocol. Human T cells from peripheral blood were isolated using CD4 and CD8 microbeads in a ratio of 1:1 following the manufacturer’s instructions (Miltenyi Biotech, CA). To determine if activation of T cell and B cells induces CD74 upregulation, T cells were treated overnight with CD3 and CD28 soluble antibodies (10 ng/mL, BioLegend, CA) and 250 U/mL IL-2. B cells were isolated from PBMCs using Easysep human B cell isolation kit (StemCell Technologies, MA) and then treated with LPS (10 ng/mL) and anti-IgM antibody (10 µg/mL). The activation status of the cells was confirmed by flow cytometry. After being isolated from peripheral blood, primary MCL patient cells were cryopreserved and later thawed for use. These cells were cultured in RPMI 1640 with 10% FBS and 5% CO2 and used immediately after thawing.

Antibodies and flow cytometry

Antibodies used in this study included anti-CD3 (clone SK7 and 145-2C11), CD56 (clone N901), CD14 (clone MφP9), human CD45 (clone 2D1), mouse CD45 (clone 30-F11), human LIN (CD3, CD19, CD20, CD56), CD4 (clone RPA-T4), CD69 (clone FN50), CD25 (clone BC96), CD33 (clone WM53), CD11b (QA20A58), and HLA-DR (clone L243). All antibodies are from BioLegend, CA, except CD74 (clone MB741) was from BD Biosciences, CA. Cells were washed once with PBS, blocked with Trustain human Fc blocker (BioLegend), stained with antibodies for 20 min at room temperature and analyzed with a LSRII flow cytometer (BD Biosciences, CA, USA). For parent or mutant CAR-T detection on primary T cells, a truncated EGFR (tEGFR) tag was used in the CAR construct and detected by anti-EGFR antibody (AY13, BioLegend).

For CD74 antigen density determination, molecules of soluble fluorochrome (MESF) FITC-5 premix beads (Bangs Laboratory Inc, IN) were used. These sets of calibrated beads contain a specific number of fluorophore molecules bound per bead. They are used to standardize and convert the mean fluorescence intensity (MFI) in flow cytometry into a count of fluorophores. These allow for the calculation of the number of antigens per cell when using antibodies under saturating conditions, considering the Fluorophore to Protein Ratio (F:P) of each antibody. A calibration curve correlating instrument detection channel values and standardized fluorescence intensity units on constructed with R2 of 0.9995. With the F:P ratio of 1 for the CD74 FITC antibody we used, the correlation equation was used to calculate the antigen density of CD74 on MCL cells from the MFI obtained on the same day with settings according to the manufacturer’s instruction.

A flow cytometry based functional binding affinity assay protocol was modified to measure the CD74 scFV on Jurkat cells to CD74 extracellular domain (ECD)-Fc [27]. Briefly, 1 × 106 Jurkat cells (viability > 90%) carrying parent or mutant scFV were washed twice with cold PBS and blocked with Fc blocker (Trustain human Fc blocker, BioLegend) for 10 min at room temperature. The cells were stained with 5 µg/mL CD74-ECD (Sino Biological US Inc., PA) in excess for 45 min on ice, then washed twice with PBS before staining with anti-Fc flow cytometric grade antibody and SYTOX Blue dead cell stain (Invitrogen) for 15 min at room temperature. The cells were washed twice with cold PBS and analyzed immediately by flow cytometry. To confirm the activation of Jurkat cells or primary T cells, the expressions of CD69 and CD25 were determined by flow cytometry.

Repeated antigen stimulation assay

The stimulation and proliferation methods were performed as previously reported [28]. Jurkat cell or primary T cell clones were co-cultured with irradiated Mino cells (stimulator) at effector-to-target ratio (E:T ratio) of 1:2 at a total cell density of 3 × 105/mL with RPMI 1640 medium containing reduced FBS at 1%. The culture was refreshed with medium every three days and restimulated with freshly irradiated Mino cells for a total of three times before the absolute cell counts were determined by trypan blue exclusion assay. Parent CD74bbz CAR and untransduced-Jurkat or T cells with or without stimulators were used as negative control.

Cytotoxicity assay

The cytotoxicity of the CAR-T cells was performed with ToxiLight™ non-destructive cytotoxicity bioassay kit (Lonza) and as manufacturer’s instruction described [29]. Briefly, the MCL cell lines or primary MCL patient samples (lymphoma % ranged from 73.7% to 98.7%) were co-cultured with 74bbz CAR-T cells or untransduced T cell control (UTT) cells at E:T ratio of 5:1 for 24 h. At 24 h, the wells for maximum lysis were added with 100% Lysis Buffer (Lonza) for 10 min at room temperature. The volume in other wells were adjusted with the provided Tris AC buffer. The cell supernatant from each well was harvested and reacted with the provided substrate for 5 min before the plates were read for bioluminescence with a Synergy HT microplate reader (Biotek, VT).

Enzyme-linked immunosorbent assay (ELISA)

Measurement of human IFN-γ in the culture supernatant was performed with ELISA MAX™ Deluxe Set human IFN-γ kit (BioLegend) according to manufacturer’s instruction. Cell-free supernatant was harvested from cell culture 24 h after coculture with effectors (E:T ratio 1:2). The plate was then washed, incubated with tetramethylbenzidine substrate (Agilent Technologies, CA), and read at 450 nm using a Synergy HT microplate reader (Biotek, VT).

In vivo experiments

All animal studies were approved by The Ohio State University Institutional Animal Care and Use Committee. 4–6-weeks old female NSG mice were used to establish a subcutaneous (s.c.) MCL model using CD74+ Mino cells. On day 0, 1 × 106 Mino cells were mixed with Matrigel matrix (Corning, AZ) at 1:1 v/v and subcutaneously injected in the right flank of the mice. After 3 days, 5 × 106 74bbz CAR-T cells or UTT control cells were injected intratumorally. The mice were sacrificed when predetermined removal criteria were met (body condition score < 2, hind limb paralysis, and 20% body weight loss). To determine the activity of 74bbz CAR-T cells on human immune cell subsets in vivo, we established a humanized mouse model using CD34+ stem cells purified from human umbilical cord blood as previously described [30]. Briefly, 4-week-old female NSG mice were irradiated at 125 cGy (RS-2000, Rad Source Technologies, GA) and then intravenously injected with 5 × 105 human CD34+ cells, isolated by human CD34 microbead ultra pure kit (Miltenyi Biotech), per mouse. The purities of the human CD34+ cells were > 85% (n = 3). The mice were intraperitoneally injected with stem cell factor (SCF), granulocyte macrophage colony-stimulating factor (GM-CSF), and IL-3 at 4 ng/mL every other day for 3 weeks to support the expansion of the myeloid compartment. After 10-week post-transplant of the human CD34+ cells, human chimerism in circulating cells was determined by facial/submandibular venous blood sampling and flow cytometry. The baseline absolute numbers of human immune cell subsets were collected, and mice were randomized into groups of 5–7 to receive either 5 × 105 UTT or 5 × 105 42105-74bbz CAR-T cells. The absolute number of human B cells (human CD45+ CD33− CD19+), monocytes (human CD45+LIN−CD11b+CD33+CD14+cells), granulocytic myeloid-derived suppressor cells (G-MDSC, human CD45+LIN−CD11b+CD33+CD14−HLA-DR−), monocytic myeloid-derived suppressor cells (M-MDSC, human LIN−CD45+CD11b+CD33+CD14+HLA-DR−) and NK cells (human CD45+CD33−CD3−CD56+) were measured on Day 3, 11, 18, 23 post UTT/CAR-T injections. The UTT and CAR-T cells were traced and identified as human CD45+CD3+ and CD45+CD3+EGFR+, respectively.

Statistical analysis

For independent data, two-sample t-test or ANOVA were used for two group comparisons or for multiple group comparisons, respectively. For paired or correlated data, such as cells from the same subjects treated with different treatment conditions, paired t-tests or linear mixed effects models were used to adjust the correlation among observations from the same object. To test the dependence of antigen expression on cytotoxicity, linear regression model was used. For survival studies, log rank test was used for comparison and survival curves were displayed with Kaplan–Meier survival curves. P-values were adjusted for multiple comparisons by Holm’s procedure when needed. A p-value less than 0.05 for two group comparisons or after adjustment for multiple comparisons was considered statistically significant.