Conclusions

FS222, a tetravalent CD137/PD-L1 mAb² was developed which binds PD-L1 and, in a PD-L1-dependent manner, stimulates CD137 agonism to elicit potent T cell activation in vitro withgreater potency than a combination approach. FS222 has comparable in vitro potency incynomolgus monkeys and so a preliminary toxicity study was undertaken in this species whichshowed PD responses related to dose and no evidence of liver toxicity (as seen clinically withCD137 agonists). A surrogate mouse CD137/PD-L1 mAb2 with mechanism matched to FS222significantly reduced tumour growth in two models with greater survival benefit than acombination approach in the MC38 model. In a CT26 model, a dose-dependent significantsurvival benefit was seen at doses between 0.1 and 0.3 mg/kg and above. This was coincidentwith intra-tumoural and peripheral T cell expansion. Within these two compartments FS222surrogate mAb2 showed rapid distribution to T cells via PD-L1 receptor occupancy and drugbinding assays.This data supports the further development of FS222, a best-in-class CD137/PD-L1 tetravalentbispecific mAb2 with IgG format for the treatment of patients with advanced malignancies.

Matthew A. Lakins, Alexander Koers, Jose Munoz-Olaya, Raffaella Giambalvo, Robert Hughes, Daniel Jones, Sarka Pechouckova, Emma Goodman, Sylwia Marshall, Mateusz Wydro, Cristian Gradinaru, Francisca Wollerton, Sarah Batey, Daniel Gliddon, Michael Davies, Michelle Morrow, Mihriban Tuna, Neil Brewis

F-star, Cambridge, UK

FS222 mAb², a Bispecific Conditional Agonist Antibody Targeting CD137 and PD-L1, Induces Potent Lymphocyte Activation and has a Favourable Safety Profile

5. FS222 surrogate mAb² causes dose-dependent survival benefit concomitant with peripheral and tumour expansion of T cells

B

Background

FS222 surrogatemAb2

Median survival (days)

p-values Log-rank (group wise

comparison with lower dosed group)

10mg/kg 39 0.2

1mg/kg 29 0.02

0.3mg/kg 24 0.007

0.1mg/kg 21 0.6

IgG ctrl 21

0 20 40 600

50

100

Days post inoculation

Pe

rce

nt

surv

ival

10 mg/kg

1 mg/kg

0.3 mg/kg

0.1 mg/kg

IgG ctrl

ns

***

ns

A

E

Compound IgG ctrl PD-L1 mAb CD137 mAbPD-L1 mAb + CD137 mAb

FS222 surrogate mAb2

Tumour-free animals

Number 0/12 2/12 2/12 4/12 12/12

Percentage 0 % 16 % 16 % 33 % 100 %

Figure 4. A FS222 surrogate mAb² activity in a CD8+ OT-1mouse T cell activation assay with cell-based crosslinkingprovided by B16-F10 tumour cells pulsed with ovalbuminpeptide and that express mouse PD-L1. Survival data formice treated with surrogate FS222 mAb² in B MC38, CCT26, and D individual plots for mice inoculated withMC38 tumour cell line and subsequently treated on day 7,9, and 11 with 1mg/kg FS222 surrogate mAb² - E Summarytable of tumour-free animals by end of study for MC38experiment described above.Additionally, FS222 surrogate mAb2 shows significanttumour growth inhibition in B16-F10 syngeneic mousetumour model (data not shown).

10 20 30 40 50

0

500

1000

1500

Days post inoculation

Tum

ou

r V

olu

me

mm

3

IgG ctrl

10 20 30 40 50

0

500

1000

1500

Days post inoculation

PD-L1 mAb

10 20 30 40 50

0

500

1000

1500

Days post inoculation

CD137 mAb

10 20 30 40 50

0

500

1000

1500

Days post inoculation

PD-L1 mAb + CD137 mAb

10 20 30 40 50

0

500

1000

1500

Days post inoculation

FS222 surrogate mAb²

B

A

C

D

0 20 40 600

50

100

Days

Per

cen

t su

rviv

al

FS222 surrogate mAb2

PD-L1 mAb

PD-L1 mAb +CD137 mAb

CD137 mAbIgG ctrl

MC38 (1mg/kg)

0 20 40 60 800

50

100

Days

Per

cen

t su

rviv

al

IgG ctrl

FS222

surrogate mAb2

CT26 (10mg/kg)

PD-L1 mAb

CD137 mAb

AACR Annual Meeting 2019 | 29 Mar-03 Apr | Atlanta | Poster Number: 1540 Strictly for personal use - DO NOT POST ONLINE

T cell

PD-L1+ cell

CD137

PD-L1

Fcab™Fc TARGET B TARGET B

TARGET ATARGET A

mAb²™

Directed amino acid substitutions to

create new binding sites in Fc

Rapid conversion into tetravalent IgG i.e. bivalent for both

targets

Plug-and-play into any mAb

1. FS222 simultaneously binds to both PD-L1 and CD137 with sub-nanomolar affinity

Figure 1. A Representation of FS222, a tetravalent bispecific CD137/PD-L1 antibody on a human IgG1 backbone with FcgR-bindingsignificantly reduced by the L234A and L235A (LALA) mutation highlighted in green. The anti-PD-L1 IgG1 with a distinct CD137binding capability in the Fc region consists of two homodimers, each comprised of a heavy and light chain. The complementarity-determining regions (CDRs) of the heavy and light chains are highlighted in orange. The CH3 domain AB and EF binding loops arehighlighted in cyan. B Surface plasmon resonance showing FS222 simultaneous binding to both human PD-L1 and human CD137. C– E Cell binding of FS222 to: C DO11.10 T cells expressing human CD137, D HEK cells expressing human PD-L1, or E activatedhuman primary CD4+ and CD8+ T cells.

C

0

20

40

60

80

100

Time (h) Post-dose

% K

i67

+

26 24 48 72 96120

1920

20

40

60

80

100

Time (h) Post-dose

% K

i67

+

26 24 48 72 96120

192

0

20

40

60

80

100

Time (h) Post-dose

% K

i67

+

26 24 48 72 96120

1920

20

40

60

80

100

Time (h) Post-dose

% K

i67

+

26 24 48 72 96120

192

IgG ctrl (10 mg/kg)

FS222 surrogate mAb² (10 mg/kg)

FS222 surrogate mAb² (1 mg/kg)

Figure 6. A PD-L1 receptor occupancy. Free PD-L1 was determined using a competing anti-mouse PD-L1 antibody. The datapresented shows the positive population as a percentage of PD-L1 receptor occupancy compared to a 100% mAb2 saturatedsample of CD8+ or CD4+ T cells from the tumour or blood. PD-L1 receptor occupancy is maintained longer in the tumourmicroenvironment than the blood. B FS222 binding data to T cells ex vivo. Anti-human Fc secondary antibody detected thehuman Fc region of FS222 mAb² bound to cells and the positive population is presented as a percentage of all CD8+ or all CD4+ Tcells from the tumour or blood. FS222 mAb² rapidly binds both CD8+ and CD4+ T cells in both tumour and blood and thepercentage positive T cell population is positively correlated with dose level.

6. FS222 surrogate mAb² shows dose-dependent PD-L1 receptor occupancy and T cell binding with rapid intratumoral and peripheral distribution

E

A

-600

-400

-200

0

200

400

600

800

1000

0 100 200 300 400 500 600 700 800 900 1000

Adjusted sensorgramRU

Re

sp

on

se

(0 =

bas

eli

ne

)

sTim e

4

5

FS222 binding to immobilised human PD-L1-Avi-His-Biot

Res

po

nse

(R

U)

B

Human CD137-mFc binding to captured FS222

CD137 binding sites

PD-L1 binding sites

LALA mutation

C D

0.01 1 1000

5000

10000

15000

Antibody concentration (nM)

MFI

AF6

47

FS222 mAb²CD137/mock mAb²CD137 mAbIgG ctrl

0.01 1 1000

50000

100000

150000

Antibody concentration (nM)

MFI

AF6

47

FS222 mAb²CD137/mock mAb²PD-L1 mAbIgG ctrl

Cell binding on DO11.10 overexpressing human CD137

Cell binding on HEK overexpressing human PD-L1

0.001 0.01 0.1 1 10 100

1000

2000

3000

4000

5000

Antibody concentration (nM)

MFI

AF6

47

FS222 mAb²CD137/mock mAb²PD-L1 mAbCD137 mAbIgG ctrl

0.001 0.01 0.1 1 10 1001000

2000

3000

4000

5000

6000

Antibody concentration (nM)

MFI

AF6

47

FS222 mAb²CD137/mock mAb²PD-L1 mAbCD137mAbIgG ctrl

CD4+ T cells CD8+ T cells

0

20

40

60

80

100

Time (h) Post-dose

% p

osi

tive

fo

r

FS2

22

su

rro

gate

mA

b²

or

IgG

ctr

l

26 24 48 72 96120

1920

20

40

60

80

100

Time (h) Post-dose

26 24 48 72 96120

192

0

20

40

60

80

100

Time (h) Post-dose

% p

osi

tive

fo

r

FS2

22

su

rro

gate

mA

b²

or

IgG

ctr

l

26 24 48 72 96120

192

0

20

40

60

80

100

Time (h) Post-dose

26 24 48 72 96120

192

FS222 surrogate mAb² (10mg/kg) FS222 surrogate mAb² (1mg/kg)

IgG ctrl (10mg/kg)

Blo

od

Tum

ou

r

0

20

40

60

80

100

Time (h) post-dose

PD

-L1

re

cep

tor

occ

up

ancy

%

26 24 48 72 96120

1920

20

40

60

80

100

Time (h) post-dose

26 24 48 72 96120

192

0

20

40

60

80

100

Time (h) post-dose

PD

-L1

re

cep

tor

occ

up

ancy

%

26 24 48 72 96120

1920

20

40

60

80

100

Time (h) post-dose

26 24 48 72 96120

192

FS222 surrogate mAb² (10mg/kg) FS222 surrogate mAb² (1mg/kg)

IgG ctrl (10mg/kg) + spikeIgG ctrl (10mg/kg)

0.0001 0.001 0.01 0.1 1 10 1000

10000

20000

30000

Antibody concentration (nM)

mIF

Ng

pg/

ml

PD-L1 mAb

CD137 mAb(crosslinked)

CD137 mAb +PD-L1 mAb

FS222surrogate mAb²

CD137/mock mAb²

4. FS222 surrogate mAb² in vitro activity replicates that of FS222 mAb² and leads to significant in vivo survival benefit, superior to combination/monotherapy

0.001 0.01 0.1 1 10 1000

2500

5000

7500

10000

Antibody concentration (nM)

hIL

2 (

pg/

ml)

FS222 (100% HEK.hPD-L1)

FS222 (50% HEK.hPD-L1)

FS222 (25% HEK.hPD-L1)

FS222 (12.5% HEK.hPD-L1)

FS222 (6.5% HEK.hPD-L1)

FS222 (0% HEK.hPD-L1)

CD137 mAb (0% HEK.hPD-L1)

IgG ctrl (0% HEK.hPD-L1)

0.001 0.01 0.1 1 10 100

5000

10000

15000

20000

Antibody concentration (nM)

hIF

Ng

(pg/

ml)

FS222 mAb²

CD137/mock mAb²

PD-L1 mAb

CD137/mock mAb² +PD-L1 mAb

IgG ctrl

Figure 2. A FS222 activity in a CD8+ T cell activation assay with varying mixed populations of HEKs that are positive or negative forPD-L1. B FS222 activity in MLR (EC50 0.07 nM) against monospecific component entities that make up the complete mAb² eitheralone or in combination.

A B

2. FS222 is superior to combinations in human T cell activation assays, with CD137-activity dependent on binding to PD-L1

Figure 3. A Pharmacokinetic prolife of FS222 incynomolgus monkey has a half-life of ~6 days. (SD= single dose) B Changes in clinical chemistryparameters relating to liver function ofcynomolgus monkey in FS222 repeat dose phase.No change outside of normal parameters (lowerand upper limit columns) was observed. C - FKinetic changes after 4x weekly repeated dose ofFS222 in cynomolgus monkey C frequency ofperipheral CD4+ central memory cells expressingKi67 and D NK cells expressing Ki67. E Frequencyof peripheral CD8+ central memory cellsexpressing Ki67 and F profile of serum solublePD-L1.

PD-1/L1 axis blockade shows durable responses and extended overall survival across cancer types in asubset of patients. Tumour Necrosis Factor Receptor (TNFR) superfamily activation is also being testedclinically to improve patient responses. Current interventions using therapeutic CD137 (4-1BB)agonists to activate T cells are restricted by severe dose limiting toxicities and poor efficacy asmonotherapies. The generation of a bispecific agonist of CD137 dependent on PD-L1 crosslinkingcreates a greater therapeutic window with improved safety and efficacy.

Here, we show how a novel tetravalent bispecific antibody (mAb²™) binds to both CD137 and PD-L1,induces potent in vitro T cell activity in a PD-L1-dependent manner and results in significant tumourcontrol across two syngeneic tumour models without toxicity. Intra-tumoural and peripheralpharmacodynamic changes implicate an increase in the proliferative CD8+ T cell response as amechanism of action.

CD8+ T cell activation assay Mixed Lymphocyte Reaction

A B

0 100 200 300 400

0.001

0.01

0.1

1

10

100

1000

10000

Time post-dose (h)

FS2

22

mA

b²

con

cen

trat

ion

(µ

g/m

l)

Male 30mg/kg Female 30mg/kg

Male 10mg/kg Female 10mg/kg

Male 3mg/kg (SD) Female 3mg/kg (SD)

Male 1mg/kg Female 1mg/kg

Male 0.1mg/kg Female 0.1mg/kg

A B

3. FS222 mAb² has a dose proportional PK profile and elicits immune activation with no liver toxicity in a preliminary toxicity study in cynomolgus monkey

0

20

40

60

80

Time (d)

Rel

ativ

e (%

)

-5 1 4 8 11 15 18 22 25 0 200 400 6000

1000

2000

3000

4000

Time (h)

Seru

m P

D-L

1 (p

g/m

L)

Male 0.1mg/kg

Female 0.1mg/kg

Male 1mg/kg

Female 1mg/kg

Male 10mg/kg

Female 10mg/kg

Male 30mg/kg

Female 30mg/kg

100% survival

Lower limit of normal range

FS222Upper limit of normal range

AST (U/L) 20 23-69 94

ALT (U/L) 21 19-111 112

ALP (U/L) 140 485-1310 1350

TBIL (mg/dL) 0.06 0.07-0.38 0.43

0

10

20

30

40

50

Time (d)

Re

lati

ve (

%)

-5 1 4 8 11 15 18 22 250

10

20

30

40

50

60

Time (d)

Re

lati

ve (

%)

-5 1 4 8 11 15 18 22 25

Ki67 expression on CD4+

central memory T cellsKi67 expression on

NK cells

No change in clinical chemistry parameters relating to liver function

C

Serum levels of soluble PD-L1

E F

Blood Tumour

Ki67 Expression on CD8+

central memory T cells

D

Figure 5. A Kaplan-Meier of dose-range findingstudy in CT26 of FS222 surrogate mAb² in therange 0.1 to 10 mg/kg. B Increased dose of FS222surrogate mAb² is correlated with increasedsurvival. C Pharmacodynamic changes observedin the tumour and the periphery for CT26tumour-bearing mice upon treatment with asingle dose (10mg/kg or 1mg/kg) of FS222surrogate mAb² .

CD

4+

T ce

llsC

D8

+T

cells

Cell binding on human primary T cells:

FS222

PD-L1 receptor occupancy on CD8+ T cells

PD-L1 receptor occupancy on CD4+ T cells

FS222 surrogate mAb² bound on CD8+ T cells

FS222 surrogate mAb² bound on CD4+ T cells