DNA‐Mediated Assembly of Multispecific Antibodies for T Cell Engaging and Tumor Killing

Abstract Targeting T‐cells against cancer cells is a direct means of treating cancer, and has already shown great responses in clinical treatment of B‐cell malignancies. A simple way to redirect T‐cells to cancer cells is by using multispecific antibody (MsAb) that contains different arms for specifically “grabbing” the T‐cells and cancer cells; as such, the T‐cells are activated upon target engagement and the killing begins. Here, a nucleic acid mediated protein–protein assembly (NAPPA) approach is implemented to construct a MsAb for T‐cell engaging and tumor killing. Anti ‐CD19 and ‐CD3 single‐chain variable fragments (scFvs) are conjugated to different l‐DNAs with sequences that form the Holliday junction, thus allowing spontaneous assembly of homogeneous protein–DNA oligomers containing two anti‐CD19 and one anti‐CD3 scFvs. The new MsAb shows strong efficacy in inducing Raji tumor cell cytotoxicity in the presence of T‐cells with EC50 ≈ 0.2 × 10−9 m; it also suppresses tumor growth in a Raji xenograft mouse model. The data indicates that MsAbs assembled from protein–DNA conjugates are effective macromolecules for directing T‐cells for tumor killing. The modular nature of the NAPPA platform allows rapid generation of complex MsAbs from simple antibody fragments, while offering a general solution for preparing antibodies with high‐order specificity.


scFv expression and purification
Gene fragment encoding H6-MBP-(TEV site)-Strep-scFv were subcloned into the pET21(+) vector (Novagen) to generate scFv expression plasmids. Expression constructs were transformed into SHuffle T7 competent E.coli cells (New England Biolabs). The cells were grown at 37 ℃ in LB media until the OD600 reached 0.6~0.8. Then the cell culture was cooled to 16 ℃ before adding 0.1 mM β-D-thiogalatopyranoside (IPTG), and protein expression lasted overnight at 16 ℃. The cells were harvested and resuspended in lysis buffer (20 mM Tris-HCl, 200 mM NaCl，2 mM EDTA, 10 μM TCEP, pH 7.4), followed by sonication with a 6 mm probe (20% output) for 5 min. The cell lysis was centrifuged at 40,000 g for 25 min at 4 ℃ and the supernatant was loaded to an amylose column (New England Biolabs) pre-equilibrated with the lysis buffer. The fusion protein was eluted with 4 column volumn (CV) of lysis buffer supplemented with 10 mM D-maltose (Sigma). The purified fusion protein was cleaved with tobacco etch virus (TEV) protease (at a concentration of 10 μg/ml) for 2 h at 37 ℃ to remove the MBP, followed by further purification by size exclusion chromatography (Superdex 200 Increase, GE lifescience). The elution fractions containing the Strep-tagged scFv were collected and stored at -80 ℃.

SPR measurement of scFv affinity to target antigens
SPR measurements were performed using the Biacore 8K instrument (GE Life Sciences) at 25 C. Human CD19 (CD19 20-291 -10xHis, Acro Biosystems, Cat# CD9-H52H2) or CD3 (CD3 23-126 -6xHis, Abcam, Cat# ab220577) antigen proteins were coupled to the CM5 chip surface by amino coupling. To measure binding, purified MBP-scFvs in the SPR buffer was injected at increasing concentrations (0.0078125, 0.015625, 0.03125, 0.0625, 0.125, 0.25 and 0.5 M). The response in signal (RU) at each of the above protein concentration was recorded ( Figure S1). SPR data processing and analysis were performed using the program BIAevaluation (GE Life Sciences). Equilibrium signal (R eq ) at different MBP-scFv concentration (C) were plotted and fit to the equation R eq = R max /(1 + K D /C) to determine the equilibrium binding constant (K D ). R max is the binding signal at saturation.

scFv-DNA ligation
The L-DNA with a 5'-NH 2 group was dissolved in PBS (10 mM Na 2 HPO 4 , 1.8 mM KH 2 PO 4 , 2.7 mM KCl, 137 mM NaCl, pH7.4) to a concentration of 0.5 mM. In order to ligate scFv with L-DNA, a mixture of L-DNA and SM(PEG) 2 (ThermoFisher) at a molar ratio of 1 : 50 was first incubated at 25 ℃ for 2 h to ligate SM(PEG) 2 with L-DNA. Precipitate formed immediately after adding 2 volumns of anhydrous ethanol. Following incubation at -20 ℃ for 20 min, the precipitate was collected by centrifuging at 12000 rpm for 10 min and the supernatant was removed. The pellet was washed with 75% ethanol for 4 times. The residual ethanol was vaporized and the pellet was solubilized with double distilled water to a concentration of ~ 5 mg/ml. The SM(PEG) 2 -DNA was stored at -80 ℃.
ScFv was mixed with SM(PEG) 2 -DNA at a molar ratio of 1:1.2 -1:2 at 4 ℃ for overnight to ligate scFv with L-DNA. The conjugation yield was ~90%. Unligated SM(PEG) 2 -DNA was removed by passing through a strepTactin (Qiagen) column ( Figure S6a). The elution was further loaded on a HiTrap Q column (GE lifescience) and scFv-DNA was eluted by a 20 CV linear gradient from buffer A (20 mM Tris-HCl, 15 mM NaCl, pH 8.5) to buffer B (20 mM Tris-HCl, 1 M NaCl, pH 8.5). The elution fractions containing scFv-DNA was collected and dialysis against PBS for 1.5 h for twice ( Figure S6b). The sample was concentrated to a concentration of 10 μM.
Human peripheral blood mononuclear cells (PBMC). For in vitro efficacy study, PBMCs were separated from the whole blood of healthy donors by a density gradient centrifugation method using Ficoll Histopaque. For in vivo studies, PBMCs were purchased from appropriate vendors (ASTARTE BIOLOGICS).

In vitro cytotoxicity of NAPPA001 and NAPPA002
In each of the 96-well plates, Raji cells (10 4 ) were mixed with fresh human PBMCs (10 5 ) at E:T ratio of 10:1, followed by the addition of various concentrations of (scFv CD19 ) 2 -scFv CD3 (NAPPA001). Raji cell lysis was assessed after 48 hours of MsAb treatment using a LDH detection kit (Beyotime, China). LDH stands for lactate dehydrogenase; it is released into the extracellular media by dead cells due to impaired membrane integrity. Cytotoxicity (%) was calculated as [sample LDH release -background LDH release]/[maximum LDH release -background LDH release] × 100, where background LDH release is from Raji cells incubated with PBMC without MsAbs and maximum LDH release is from Raji cells treated with 1% Triton X-100. All measurements were performed in triplicates. The same protocol was used for other positive and negative control experiments. Exactly the same protocol was also used for measuring in vitro cytotoxicity of (scFv CEA ) 2 -scFv CD3 (NAPPA002) on the LS174T colon carcinoma cells.

Animal study
Mice. Female (non-pregnant and nulliparous) NOD-scid IL2Rgamma null (The Jackson Laboratory, Lot# 005557) mice (age 6-8 weeks) were used for human carcinoma xenograft mouse model. Mice were maintained under pathogen-free and standardized environmental conditions (20±1 C, 50±10% relative humidity, 12 hours light-dark daily cycles). All experiments were done according to the American Animal Protection Law with permission from the responsible local authorities.
In vivo efficacy of (scFv CD19 ) 2 -scFv CD3 (NAPPA001). Raji cells (5×10 5 cells) (50 L) were pre-mixed with human PBMC (2×10 6 cells) (50 L) at an E:T ratio of 4:1, and further mixed with Matrigel (100 L) in a 1:1 ratio yielding 200 L suspension for subcutaneous implantation into the right flanks of NOD/SCID mice. The (scFv CD19 ) 2 -scFv CD3 (0.5 mg per 1 kg of body mass) was injected intravenously twice a week for four weeks starting 3 days after Raji/PBMC co-grafting. Caliper tumor volume measurements were made twice-weekly on all surviving animals. Tumor volumes were estimated by the following formula: Tumor Volume = 0.5 × (length × width 2 ), where length is the longest diameter measurement and width is the shortest diameter measurement.
In vivo efficacy of (scFv CEA ) 2 -scFv CD3 (NAPPA002). LS174T cells (10 6 ) were pre-mixed with PBMCs (5×10 6 ) at an E:T ratio of 5:1, and further mixed with Matrigel in a 1:1 ratio yielding 200 L suspension for subcutaneous implantation into the right flanks of NOD/SCID mice. The (scFv CEA ) 2 -scFv CD3 (1 mg per 1 kg of body mass) was injected intravenously twice a week for 1.5 weeks starting 3 days after LS174T/PBMC co-grafting. Caliper tumor volume measurements were made twice-weekly on all surviving animals. Tumor volumes were estimated by the following formula: Tumor Volume = 0.5 × (length × width 2 ), where length is the longest diameter measurement and width is the shortest diameter measurement.
In vivo imaging of MsAbs targeting solid tumor. LS174T cells (10 6 ) were pre-mixed with PBMCs (5×10 6 ) at an E:T ratio of 5:1, and further mixed with Matrigel in a 1:1 ratio yielding 200 L suspension for subcutaneous implantation into the right flanks of NOD/SCID mice. When tumor volume reached ~500 mm 3 , (scFv CEA ) 2 -scFv CD3 -Cy5.5 (220 L, 0.45 mg/ml) or PBS (vehicle) was injected intravenously. Fluorescence intensity in mice was recorded and quantified by a small animal imager (PerkinElmer IVIS Kinetic III) at different time points (1, 2, 4, 8, 24, 48, 96 h). Fluorescence intensity vs. Time curve was generated to determine halflife of the fluorescent MsAb in mice.   To examine whether cancer cell killing was due to T-cell redirecting, Raji cells were treated with 500 pM NAPPA001 in the presence (1) or absence (2) of hPBMC (E:T=10:1) for 2 days. To examine whether the cancer cell killing on the left was due to T-cell  Raji engaging mediated by the bispecific NAPPA001, Raji cells were treated with a mixture of scFv CD19 and scFv CD3 (500 pM) in the presence of hPBMC (E:T=10:1) (3). To test whether scFv CD19 -scFv CD3 assembled with a simple L-DNA duplex can function as a normal BsAb, scFv CD19 -L-DNA1 and scFv CD3 -L-DNA4 were mixed at 1:1 ratio. Raji cells were treated with 500 pM of BsAb in the presence of hPBMC (E:T=10:1) for 2 days (4). To ensure that antibody assembly was mediated by L-DNA oligomerization, the scFv CD19 and scFv CD3 were mixed with L-DNA Holliday junction at 500 pM each and used to treat Raji cells in the presence of hPBMC (E:T=10:1) (5).
Tumor lysis (cytotoxicity) was measured using the LDH (lactate dehydrogenase) detection kit, which quantifies LDH release from dead cells with impaired membrane integrity. Figure S4. Body weights of the five mice during in vivo efficacy study of (scFv CD19 ) 2 -scFv CD3 (NAPPA001). Figure S5. In vivo efficacy of NAPPA002 in human colorectal adenocarcinoma (LS174T cell line) xenograft mouse model. LS174T tumor cells were implanted or co-implanted with human PBMC (E:T ratio is 5:1) subcutaneously into NSG mice (n = 6) and treated with either PBS (vehicle) in the absence (gray) or presence (black) of PBMC, or NAPPA002 (20 g/mouse) (red) via intravenous injection twice a week starting 3 days after LS174T/PBMC co-grafting for 1.5 weeks.