Manufacture of an Allogeneic CAR-T Stem Cell Memory Product Candidate for Multiple Myeloma, P-BCMA-ALLO1, Is Robust, Reproducible and Highly ScalableStacey A. Cranert, Maximilian Richter, Min Tong, Leslie Weiss, Yening Tan, Eric M. Ostertag, Julia Coronella, Devon J. Shedlock

CONCLUSIONS

Poseida Therapeutics, Inc., 9390 Towne Centre Drive, Suite 200, San Diego, CA, 92121

INTRODUCTION

ABSTRACT

METHODS & RESULTS

Chimeric Antigen Receptor (CAR) T cell therapy has generated unprecedented efficacy in the treatment ofmultiple hematologic malignancies. For relapsed/refractory Multiple Myeloma (MM), autologous CAR-T productsdirected against the B cell maturation antigen (BCMA), such as Poseida’s P-BCMA-101, have demonstratedsignificant efficacy. P-BCMA-101 is comprised of a high-percentage of stem cell memory T cells (TSCM), resultingin a product that is much safer and potentially more durable than other anti-BCMA autologous productcandidates. However, as individualized products, all autologous CAR-T products are expensive to manufactureand dependent upon patient T-cells of variable quality. We are developing P-BCMA-ALLO1, an off-the-shelfallogeneic (allo) BCMA-specific CAR-T product candidate derived from healthy donor material, which providesnumerous advantages over autologous products, increasing patient access by being immediately available andgreatly reducing manufacturing cost and variability.

P-BCMA-ALLO1 is produced using two key platform technologies: the nonviral piggyBac® (PB) DNA ModificationSystem and the high-fidelity Cas-CLOVER™ (CC) Site-Specific Gene Editing System. The mRNA coding forhyperactive, or “Super PB” transposase (SPB), and CC enzymes are codelivered with the P-BCMA-ALLO1 PB-basedDNA transgene via electroporation to healthy donor T cells to stably integrate the transgene, as well as toknockout (KO) several mediators of allo graft-versus-host and host-versus-graft responses to maximize patientsafety and durability of response. The P-BCMA-ALLO1 transgene encodes three genes, a BCMA-specific single-domain variable heavy chain (VH)-CAR (VCAR) gene, a drug selection gene to generate a ~100% CAR+ product, aswell as a caspase-based safety switch gene to reduce or eliminate the product in vivo, if desired. The CC Systemis used to KO the endogenous T Cell Receptor (TCR) and beta-2 microglobulin, thereby decreasing MajorHistocompatibility Complex (MHC) class I expression. KO of these key targets is aimed to prevent graft-versus-host disease, as well as reduce host-versus-graft rejection of the product. The CC System can efficiently editresting T cells, thereby maintaining a high-percentage of TSCM cells, and does not create unwanted off-targetmutations, another important consideration when creating an allo product candidate. To maximize the numberof doses produced from a single manufacturing run, we have developed a proprietary “booster molecule” thatallows for significant expansion of TCR-KO CAR-TSCM cells to potentially produce hundreds of doses.

To date, large-scale manufacturing of significant doses of potent allo CAR-T products has been challenging for thefield. P-BCMA-ALLO1 manufacturing uses a potentially unlimited number of individual serial donors. We havecurrently produced P-BCMA-ALLO1 at both research and near-commercial scale from >35 donors with >97%manufacturing success. While a range of TCR-KO efficiencies was observed (~50-90%), the final product wasalways >99% homozygous TCR-KO after a purification step. Overall expansion of TCR-KO cells ranged from ~2-20fold, and after removal of unedited TCR+ cells ~0.42-7.04x10e9 TCR-KO cells were recovered from 0.75x10e9starting cells. However, working at clinical production scale (starting with ~3x10e9 cells), up to 250 doses of P-BCMA-ALLO1 could be manufactured per run, at a dose of 150x10e6 cells/patient. Importantly, with this level ofdonor and manufacturing robustness, no significant prior screening of donor material, other than to meetstandard FDA requirements, would be needed.P-BCMA-ALLO1 made from multiple donors were comprised of an exceptionally high-percentage of the desirableTSCM cells (CD45RA+CD62L+CD45RO-) and had minimal to no expression of exhaustion markers, such as PD-1 orLag3. Furthermore, P-BCMA-ALLO1 demonstrated potent efficacy in the RPMI-8226 xenograft model in NSG miceacross multiple products generated from separate individual healthy donors. Altogether, these data demonstratea robust, reproducible and highly scalable manufacturing process. Moreover, this manufacturing process caneasily be expanded for use with additional CAR targets for treatment of other hematologic or solid tumormalignancies.

Patient

Patients

HealthyDonor

Anti-BCMACAR

TCR knock-outMHC I knock-out

AUTOLOGOUS ALLOGENEIC

CHALLENGES• Long vein-to-vein time• Variable potency & manufacturability of patient

material• High cost of an individualized product

ADVANTAGES• Healthy donor material for consistent potency &

manufacturability• Universal, off-the shelf product for increased patient

access & decreased cost• Large number of doses for numerous patients from

single manufacturing run

ApheresisProduct

IsolatedT-Cells

Nucleoporation, Gene Editing, and

Gene Transposition

Selection Expansion

PB-CAR Transposon DNA

Transposase mRNA+

Cas-CLOVERTM mRNA + gRNAs

+

Cryostored CAR-T Cell Product

TCR KOPurification

Overview of P-BCMA-ALLO1 Manufacturing Process

Adapted from Gattinoni et al. Nat. Med. 2017 CD62L

CD

45R

A

TSCM

89.9%

TCM

0.6%TEM

0.3%

TEFF

9.4%

Phenotype of P-BCMA-ALLO1 Product

Fig. 1 Cas-CLOVERTM Gene Editing System combines the strengths of CRISPR/Cas9 and TALEN/ZFN editing approaches, and efficiently edits in restingT cells. A) The site-specific Cas-CLOVERTM Gene Editing System consists of a dimeric gRNA-guided nuclease. Editing activity is guided by a pair of gRNAsallowing for ease of design as well as multiplexing ability found in CRISPR/Cas9-based systems. At the same time editing activity is contingent ondimerization of two half domains, similar to TALENs or ZFNs, leading to exquisite editing specificity. B) Cas-CLOVERTM efficiently edits resting T cells.Upon editing, efficient disruption of surface expression of TCRα/β and CD3 was detected via flow cytometry and graphed.

Cas-CLOVER™ Gene Editing System

CRISPR/Cas9

TALEN/ZFN

• Ease of use/design

• Low cost

• Multiplexing ability

• Specificity

A) B)

1 2 3 4 5 6 7 8 9 10 11

0

50

100

Before purification

% T

CR

-KO

Fig. 3 A Booster Molecule increases the yield of P-BCMA-ALLO1 cells while maintaining proliferative potential and the desirable early memoryphenotype. P-BCMA-ALLO1 CAR-T cells were produced in presence or absence of the Booster Molecule. Addition of the Booster Molecule led tosignificant increases in cell expansion during production A) and increased the percentage of TCR KO cells at the end of production B) leading toincreased product yield of P-BCMA-ALLO1. C) At the same time, P-BCMA-ALLO1 grown in the presence of the Booster Molecule maintained a highfraction of desirable TSCM cells and D) in an in vitro serial restimulation assay, P-BCMA-ALLO1 had a similar proliferative potential to TCR-replete(unedited) CAR-T.

- B

oo

ste

r M

ole

cu

le

+ B

oo

ste

r M

ole

cu

le

0

5

1 0

1 5

2 0

2 5

Fo

ld e

xp

an

sio

n

A L L O C A R -T c e ll

e x p a n s io n

Without With

Fold

Exp

ansi

on

Booster Molecule

~5x Expans

ion Without Affecting Profile

A) B) C)

Booster Molecule Increases Yield of P-BCMA-ALLO1

1 2 3 4 5 6 7 8 9

020406080

100120140200

300

400

128

104

104

29

29

88

88

251

82

82

11

11

42

42

33

33

Donor

Do

ses a

t 150x10

6

1 2 3 4 5 6 7 8 90

20

40

60

80

100

Donor

%o

f S

ub

se

t

TSCM

*TCM

TEM

TEFF

*Includes both TCM

and TSCM cells

in transistion

Unpurified

β2M

CD

3 CD3-84.2%

Purified

CD3-99.9%

β2M

CD

3

Anti-BCMA CAR

95.2%

No CAR Control

P-BCMA-ALLO1

A) B)

Fig. 4 The final P-BCMA-ALLO1 product is >99% TCR KO and >95% CAR-positive. A) At the end of P-BCMA-ALLO1 production, a clinically-scalablepurification step is added to deplete remaining TCR-positive CAR-T cells and yield a highly pure allogeneic CAR-T cell product comprised of more than>99.9% TCR KO cells. B) The presence of a selectable marker in the P-BCMA-ALLO1 transposon allows for the generation of a P-BCMA-ALLO1 cellproduct that is >95% CAR-positive.

P-BCMA-ALLO1 - a Highly Pure Allogeneic CAR-T Product

Fig. 5 P-BCMA-ALLO1 exhibits potent anti-tumor effects in an in vivo mouse model of MM. P-BCMA-ALLO1 (ALLO) was tested in alongside withhealthy donor non-edited cells (HDNE, CAR-positive/TCR-positive) in an in vivo model of MM tumor control. A) NGS mice were implanted sub-cutaneously with 1x107 RPMI-8226 BCMA+ tumor cells and tumors were established for 7 days before injection with 1x107 CAR-T cells. P-BCMA-ALLO1 exhibited potent and sustained anti-tumor efficacy that was comparable to non-edited control CAR-T cells. B-C) P-BCMA-ALLO1 alsodemonstrated robust in vivo proliferation that could be detected in the blood of treated animals by day 14 after T cell administration. C) The peak ofexpansion (Day 14) correlated with the timing of tumor control observed in A) and was similar to expansion levels observed for TCR-replete(unedited) CAR-T.

P-BCMA-ALLO1 Exhibits Potent Anti-tumor Effects In Vivo

0 7 14 21 28 35 42 49

0

500

1,000

1,500

2,000

Days post CAR-T infusion

Tu

mo

r V

olu

me [

mm

3] No CAR T

Unedited anti-BCMA CAR-T

Donor 1

Donor 2

Donor 3

Donor 4

Donor 5

P-BCMA-ALLO1

Pre 7 14 21 28 35 42 49

1

10

100

1,000

10,000

100,000

Days post CAR-T infusion

CA

R+

cells/

l b

loo

d

Uneditedanti-BCMA CAR-T

P-BCMA-ALLO1

CellTraceTM Violet

UneditedControl

P-BCMA-ALLO1

40%

39%

2%

0%

0%

0%

Donor #1 (PBMC)

Donor #2 (PBMC)

Autologous PBMC

P-BCMA-ALLO1 Unedited Control

0.53% 0.01%

0.65%

0.29% 0.02%

0.39%

2.43% 83.0%

12.6%

Unstimulated +Anti-CD3/CD28 +Anti-CD3/CD28CD69

CD

25

A) B)

Fig. 6 P-BCMA-ALLO1 is not capable of mediating allogeneic responses. A) To show TCR KO in the P-BCMA-ALLO1 final product on a functional level,cells were exposed to an activation regiment comprised of anti-CD3 and anti-CD28 antibodies for 24 h. P-BCMA-ALLO1 did not show anyupregulation of the activation markers CD25 and CD69 by flow cytometry. In contrast, unedited control CAR-T cells showed a strong induction ofCD25 and CD69 in response to activation. B) P-BCMA-ALLO1 cells or donor-matched, unedited control CAR-T cells were exposed to PBMCs frommismatched donors as well as autologous PBMCs. While the unedited control CAR-T cells showed robust induction of CAR-T cell proliferation inresponse to the allogeneic PBMCs. In contrast, P-BCMA-ALLO1 cells did not proliferate, showing that P-BCMA-ALLO1 cannot mediate an allogeneicresponse.

P-BCMA-ALLO1 Does not Mediate an Allogeneic Response

• Highly desirable TSCM-rich product phenotype and potent proliferative potential• Allogeneic CAR-T equivalent or better than unedited healthy donor CAR-T in vitro & in vivo• Robust non-viral manufacturing process compatible with majority of healthy donors &

ability to generate hundreds of doses per manufacturing run• Superior safety due to TSCM phenotype, no/low off-target gene editing, and safety switch• Results support rapid advancement of P-BCMA-ALLO1 into the clinic for treatment of MM

Fig. 2 Cas-CLOVERTM editing in the TRAC and B2M loci shows no off-target editing activity by NGS. A) Cas-CLOVERTM was used to edit resting T cellsusing gRNA pairs targeting TRAC or B2M and potential off-target sites for different combinations of gRNAs were surveyed via NGS. B) NGS revealedefficient on-target editing for both TRAC and B2M loci while no above-background off-target editing events were detectable. Non-edited resting T cellsfrom 3 donors were used to assess the level of non-editing related background mutations.

Donor:

A) B)

Cas-CLOVER™ Shows no Off-Target Activity

Cas-CLOVER™ Gene Editing System - The Best of Both Worlds

No

CA

R T

Au

to

Do

no

r 1

Do

no

r 2

Do

no

r 3

Do

no

r 4

Do

no

r 5

1

10

100

1,000

10,000

100,000

CA

R+

cells/

l b

loo

d

Allo

C)

D)

Pre 7 14 21 28 35 42 49

1

10

100

1,000

10,000

100,000

Days post CAR-T infusion

CA

R+

cells/

l b

loo

d

Uneditedanti-BCMA CAR-T

P-BCMA-ALLO1

TCR-α β2M