NK Cells Engineered to Express a Bispecific CD123/NKG2D Chimeric Antigen Receptor (CAR) and IL-15 As Off-the-Shelf Therapy for Acute Myeloid Leukemia Chao Guo, PhD 1,2, Luxuan Buren, PhD1,2, Alexander Aronov1, James Trager, PhD1* 1Nkarta Inc, South San Francisco CA 94080, USA 2Contributed equally *Corresponding author

Figure 5. Gain of function and loss of function studies demonstrate that bispecific constructs can fire through low Figure 1. Nkarta Natural Killer((NK) cell platform showing ex vivo natural killer (NK) cell NKG2D (A) Results of 4-hour cytotoxicity assays performed by against the CD123 HL60 cell line. (B) A significant difference (P < 0.05) between lysis in the presence and absence of NKG2D blockade is indicated by an asterisk. activation and expansion procedures Background A. H L 6 0 4 h r s 1 4 d p o s t t r a n s d u c t i o n

Natural killer (NK) cells have an emerging role in 8 0 THAW & G F P ISOLATE EXPAND ENGINEER EXPAND CRYOPRESERVE cellular immunotherapy for patients with acute NK cell ADMINISTER C D 1 2 3 - C A R y t myeloid leukemia (AML). NK cells are highly i 6 0 C D 1 2 3 - N K G 2 D - C A R x o t N K G 2 D - C A R o

active and suitable for allogeneic use as they are t y C not HLA-restricted and do not cause GVHD. We 4 0 % sought to demonstrate that engineering of NK Donor Patient γ-retrovirus Tumor targeting – NKG2D, other targets cells can improve their ability to recognize and kill Stimulatory cell line Homing – Tumor microenvironment evasion 2 0

1 2 Persistence – mbIL-15 ; : AML cells and control tumor burden. Natural killer 1 1 group 2D (NKG2D) is a transmembrane protein E : T belonging to the CD94/NKG2 family of C-type Figure 2. Targeting AML with CAR-NK cells and CD123-NKG2D-CAR design Low residual lysis of HL60 when NKG2D is blocked B. lectin-like receptors. Ligands of NKG2D (NKG2D- (A) Bone marrow aspirate showing AML (B) SFI levels of the various NKG2DL obtained with each single patient 80 + - + - + - + - anti-NKG2D blocking L) are highly expressed in AML tissues but weakly of the different leukemia entities. (C) Primary patient AML samples express CD123. (D) CD123-NKG2D-CAR GFP design CD123-CAR 60 y

on healthy tissues, making NKG2D an attractive t * Hilpert et al, Journal of immunology, 2012 Gill et al, Blood, 2014 i * CD123-NKG2D-CAR A. B. x o target for NK cell engineering. The interleukin-3 C. t NKG2D-CAR o 40 t * y

receptor alpha chain (CD123) has also been C

% * identified as a potential immunotherapeutic target 20 for high risk AML patients due to its 0 overexpression in AML blasts and leukemia stem P R R F R A A G A C C -C - - 3 D D cells compared with normal hematopoietic stem 2 2 2 1 G G D K K C -N N cells. We therefore developed a CD123-NKG2D 3 2 1 D tandem CAR construct that contains human NKG2D Ligands C D. Figure 6. CD123-NKG2D bispecific CAR confers best durable cytotoxicity of NK cells for CD123-specific single-chain variable fragment CD123 CAR: CD123 scFv CD3z mIL-15 at least 2 weeks after NK transduction. Cytotoxicity of NK-CARs and GFP-transduced NK cells against the AML (scFv) and NKG2D. We co-expressed this NKG2D CAR: NKG2D CD3z mIL-15 cell line 2 weeks post-transduction. (A) 4 hour cytotoxicity assay. (B) In vitro cytotoxicity assay was monitored for up chimera in NK cells with membrane-bound IL-15 to 4 days in culture with IncuCyte® live-cell imaging. E:T ratio is 1:8. CD123/NKG2D Bispecifics: CD123 scFv NKG2D CD3z mIL-15 to support prolonged cell survival and A. B. G F P Schematic of retroviral vector encoding CD123-NKG2D-CD3ζ and membrane bound IL-15 8 0 0 T H P - 1 4 h r s p o s t t r a n s d u c t i o n d 1 4 proliferation. C D 1 2 3 - C A R

N K G 2 D - C A R 5 0

C D 1 2 3 - N K G 2 D - C A R 6 0 0

Figure 3. NKG2D-CAR (NKX101) shows dose-dependent efficacy in an AML model T H P - 1 o n l y s

4 0 r

Results y e t i Antitumor activity of NKG2D transduced NK cells in THP-1 AML NSG xenograft model (A) Schematic timeline of b x m 4 0 0 o 5 u t

the AML NSG xenograft model. (B) Luciferase labeled THP-1 cells (2x10 ) were injected intravenously in NSG at n

o Retroviral transduction of NK cells with tandem 3 0 l l t e

6 y Day 0. Control mice (GFP; n=5) received 5x10 NK cells. The remaining 15 mice received a single intravenous c C CAR (n=6) resulted in 68.4% and 37.8% median 6 6 6 injection of 5x10 , 10x10 and 20x10 NK cells, transduced with NKG2D-CAR followed by IL-2 injections every % 2 0 0 2 0 CD123 and NKG2D CAR positivity respectively other day for 1 week. Photoluminescence signals were measured at weekly intervals by IVIS Spectrum In Vivo

Imaging system. 1 0 when measured 14 days post-transduction 0 1 2 : : 1 1 0 3 5 8 0 3 5 8 0 3 5 8 0 3 5 8 ...... A. Days: 0 0 0 0 1 1 1 1 2 2 2 2 3 3 3 3 (Figure 4). In a 4 hour cytotoxicity assay, the D0 D3 D6 D12 D18 D24 D30 IL-2 E:T E : T = 1 : 8 Imaging Imaging Imaging Imaging CD123-NKG2D tandem CAR transduced NK cells THP-1 (IV) Infuse Imaging (2 x 105) NKs (IV) were no more effective than NKG2D-CAR alone Figure 7. Selective cytotoxicity against CD123high THP-1 line demonstrates that bispecific constructs can low B. 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 C. fire through CD123. A significant difference (P < 0.05) between lysis in the presence and absence of CD123 peptide in killing the CD123 AML cell line HL60, but is indicated by an asterisk. Day 6 showed better cytotoxicity than Mock (GFP)- G F P Low residual lysis of THP-1 when CD123-CAR is blocked by CD123 peptides (1ug/ml) transduced NK cells, indicating that the tandem N K X 1 0 1 ( 5 M ) N K X 1 0 1 ( 1 0 M ) + - + - + - + - CD123 peptide Day 12 8 0 9 5 ´ 1 0 N K X 1 0 1 ( 2 0 M ) CAR can trigger robust cytotoxicity through G F P

) 9 4 ´ 1 0 c C D 1 2 3 - C A R NKG2D (figure 5). CD123-NKG2D tandem CAR Day 18 e s 6 0 / y 9 n 3 ´ 1 0 t i + o

t C D 1 2 3 - N K G 2 D - C A R x NK cells were effective at killing the CD123 o * * h

9 o 2 ´ 1 0 p t (

N K G 2 D - C A R I Day 24 o 4 0 L THP-1 AML cell line (figure 6a), exhibiting t 9

B 1 1 0 ´ y C significantly greater cytotoxicity than Mock (GFP)- 0 % 2 0 Day 30 6 2 8 4 0 6 y 1 1 2 3 3 a y y y y y transduced NK cells, or NK cells expressing either d a a a a a d d d d d CD123-CAR (Figure 6a); in a kinetically resolved T i m e ( d a y s ) Day 36 0

R R R cytotoxicity assay, the tandem CAR was more P A F A A C C C G - - - 3 D D 2 effective than the NKG2D CAR against THP-1 GFP NKX101(5M) NKX101(10M) NKX101(20M) 2 2 1 G G D K K C N N (figure 6b). In addition, the cytotoxicity of CD123- - 3 2 1 Figure 4. CD123/NKG2D bispecifics can be expressed at high levels in NK cells. D NKG2D tandem CAR transduced NK cells was C efficiently blocked by either CD123 peptides (A) CD123 expression and (B) NKG2D expression in expanded NK cells from 6 donors transduced with a retroviral vector containing GFP only or a vector encoding a CAR construct. Conclusion (Figure 7) or anti-NKG2D antibodies (Figure 5b), In summary, our data demonstrate the capacity of CD123-NKG2D tandem further demonstrating the ability of the tandem A. Summary of %CD123 expression (n=6 donors) B. Summary of %NKG2D expression (n=6 donors) 1 0 0 + + 1 0 0 CAR NK cells to kill both CD123 and NKG2D-L targets and support the receptor to be triggered by both component G F P n o

i 8 0

8 0 s C D 1 2 3 C A R clinical use of tandem CAR NK cells for patients with AML. s binding moieties. Studies to test the in vivo anti- g n e a l r o C D 1 2 3 / N K G 2 D B i s p e c i f i c s i F

p 6 0 6 0 - s x 3 tumor efficacy of CD123-NKG2D tandem CAR s References E 2

e N K G 2 D C A R

r 1 D p 4 0

D 4 0

transduced NK cells in an NSG mouse model of 2 x

C 1. Chang et al. Cancer Res; 73(6) March 15, 2013. E G

% K THP-1 are under way. 2 0 2 0

N 2. Hilpert et al, Journal of immunology, 2012, 189: 1360–1371. % 3. Gill et al, Blood 2014 123:2343-2354. 0 0

7

7 y 4. Fernandex-Messina et al, 2012 Front. Immunol., 25 September ;3:299. y a a D d 5. Carlsten and Childs, Front Immunol. 2015 Jun 10;6:266.

James Trager, PhD Contact [email protected] www.nkartatx.com