ACR 2019 ALPN-202 Sjs

2416 ACR ANNUAL MEETING » NOVEMBER 8-13 » ATLANTA, GA, USA ALPN-101, a First-in-Class Dual ICOS/CD28 Antagonist, Suppresses Key Effector Mechanisms Associated with Sjögren’s Syndrome and Systemic Lupus Erythematosus Stacey R. Dillon, Lawrence S. Evans, Katherine E. Lewis, Susan Bort, Erika Rickel, Jing Yang, Martin F. Wolfson, Kayla Susmilch, Sherri Mudri, Steven D. Levin, Janhavi G. Bhandari, Fariha Ahmed-Qadri, Mark W. Rixon, Jan L. Hillson, Stanford L. Peng, and Kristine M. Swiderek Alpine Immune Sciences, Inc. Seattle, WA

Abstract Figure 3: ALPN-101 Inhibits Cytokine Production from Human Sjögren’s and SLE Figure 5: ALPN-101 Suppresses Proliferation and Antibody Responses in Human B-TFH Figure 8: ALPN-101 Reduces Serum Autoantibody Titers in the Anti PD-L1 mAb-Induced Patient PBMC In Vitro More Potently Than Single CD28 or ICOS Pathway Inhibitors Cell Co-Cultures and Affects Expression of Genes Involved in B-T Cell Collaboration NOD Mouse Model of Sjögren’s INTRODUCTION: ALPN-101 is an Fc fusion protein of a human inducible T cell costimulator ligand (ICOSL) variant Serum anti Ro-52 Serum anti La Antagonists Target(s) (A) Human T and B Cell Proliferation (B) Human Immunoglobulin Production immunoglobulin domain (vIgD™) designed to inhibit simultaneously the CD28 and ICOS costimulatory pathways. CD28 and ICOS 0.4 0.4 Anti Ro-52 and anti La Abs were detected by T cell Artificial APC **** *** (A) Diagram of the PBMC stimulation assay Human IgG3 vs ALPN-101: n K562/OKT3/CD80/CD86/ICOSL Human IgG1 **** vs ALPN-101: ELISA (Signosis, Santa Clara, CA) in diluted WT ICOSL IgV ICOS o **

each play a role in T cell activation and adaptive immunity which can contribute to autoimmune disease when dysregulated. CD4+ T cells: % Divided CD4+ T cells: CD40L+ Cell Count 1200 400 *** p=0.0002 i

l l t ^ p=0.0146 Fc Control

**** p<0.0001 u serum collected on Day 10 from mice in the

m 0.3 m 0.3 100 1500 l

/ ** p=0.0026 /

Patient or healthy donor peripheral blood mononuclear OKT ScFv i g

g ALPN-101

ALPN-101 has previously been shown to have potent immunosuppressive activity in various in vitro and in vivo models of disease, 300 Mean Fc control d *** p=0.0002

n SjS study described in Figs 6-7. Statistical

ICOS 800

, ,

cells (PBMC) were stimulated with fixed artificial APCs 75 . Abatacept o

Mean Fc control ] ]

ALPN-101 i

2 t

1000 3 1 200 *** 0.2

including acute graft versus host disease, inflammatory arthritis, and multiple sclerosis. We report here in vitro assays using a 0.2 differences noted between the ALPN-101

1 r

CD28 #

G WT ICOSL-Fc

G l

(shown at right) for 48 hrs in the presence of ALPN-101 e

i :

50 g

e l

400 I

ICOS I group and comparator treatments were

C o

Aba + WT ICOSL-Fc

peripheral blood mononuclear cells (PBMC) from healthy donors vs. patients to analyze human T cell and B cell activation and (100nM) or comparators: r 100

t u

P 500 ^

a H

H 0.1

0.1 % 25 Naïve determined using an uncorrected non- CD80/86 [ o Prezalumab (anti ICOSL mAb) CD28 Abatacept [

suppression of antibodies and inflammatory mediators thought to contribute to the pathogenesis of Sjögren’s syndrome (SjS) and CD80/ 0 0 D O CD86 0.001 0.01 0.1 1 10 100 1000 0.001 0.01 0.1 1 10 100 1000 1:312.5 at OD dilution parametric Dunn’s test. o Abatacept alone (CTLA4-Ig) Belatacept 0 0 other connective tissue diseases. Additionally, the efficacy of ALPN-101 was confirmed in vivo in mouse immunization models, and 1 100 10000 1000000 1 100 10000 1000000 [Test Article], nM [Test Article], nM 0.0 0.0 o Abatacept + prezalumab [ pM ] [ pM ] ICOSL in mouse models of SjS and systemic lupus erythematosus (SLE). B cell: % Divided B cells: % Naive Human IgM Prezalumab ICOSL Human IgA

100 80 600 8000

m / METHODS: Primary cell assays were performed with healthy donor and SjS patient PBMC stimulated with K562 cells expressing / (B) Examples of cytokine responses in the PBMC stimulation assay n 6000 75 60 g

g ALPN-101 Reduces GC B and T Cells and Suppresses Antibody Responses

o Figure 9: FH

e t

400 n

i Mean Fc control

t r

Mean Fc control ,

] ]

CD80, CD86, ICOSL, and anti-CD3 (OKT3) to evaluate the potency of ALPN-101 to suppress cytokine production and alter gene s f

i 50 40 4000

o l

IFN Response A

M o

IL-6 Response P

IL-17A Response r in Normal Mice and in the bm12 Inducible Mouse Model of SLE

I %

200 expression. The activity of dual pathway inhibition by ALPN-101 was compared to the CD28-only inhibitor abatacept (CTLA4-Ig; (A) BALB/c mice were immunized with SRBC on Day

% 25 20

u 2000

180000 u

30000 6000 0 (D0). Half the mice were dosed on D-1 & D7, and H

Fc control H [ Bristol-Myers Squibb, via Catalent) and to the ICOS pathway inhibitor prezalumab (AMG-557/anti-ICOSL, Creative Biolabs), or a 140000 25000 [ 3000 0 0 0 0 (A) Serum Ig titers in BALB/c mice immunized with sheep red blood cells (SRBC) half on D14 & D21 with 1.5 mg ALPN-101 or 120000 15000 Prezalumab 1 100 10000 1000000 1 100 10000 1000000 2400 0.001 0.01 0.1 1 10 100 1000 0.001 0.01 0.1 1 10 100 1000 abatacept, or a molar-matched amount of Fc control. combination of the two. ALPN-101 was compared to abatacept in vivo in standard mouse immunization models (KLH, sheep RBC), [ pM ] [ pM ]

Abatacept IgM anti-SRBC IgG1 anti-SRBC IgG2a anti-SRBC IgG2b anti-SRBC The second group was boosted with SRBC on D15 ) ) Worse Better [Test Article], nM [Test Article], nM 50000 ) nd

L 2500 20000 L 10000 ** and all mice were euthanized 1 wk after their 2 dose in a model of SjS involving anti PD-L1 antibody-mediated acceleration of sialadenitis in non-obese diabetic (NOD) mice, and in the L *

m Abatacept+Prezalumab m ** **** * p<0.05

/ *** / m **** * (on D14 and D28, respectively). Serum anti-SRBC / **** 40000 80000 g ** g 10000 1600 2000 8000 ** ** p<0.005

g **** p ALPN-101 15000

bm12 inducible model of lupus. p (A) Autologous circulating human B and TFH cells isolated from PBMC from healthy donors were mixed with artificial APCs (K562/CD80) and soluble OKT3, and antibodies on D14 and D28 were measured via flow

( p

( *** p<0.001

(

 A 30000 cytometry by detecting binding to SRBC using

1500 I **** p<0.0001

6 6000

I 7

titrations of the test articles indicated. Cultures were assessed on Day 7 for cellular activation/proliferation and IC50 values are summarized in the table. F

F F

1 10000 isotype-specific anti-mouse Ig Abs. p values were

M I RESULTS: Compared to abatacept, prezalumab, or combination abatacept + prezalumab, ALPN-101 demonstrated superior I 20000 L (B) In a separate study using a similar protocol, supernatants were collected on Day 8, diluted 1:10, and assayed for human immunoglobulin (Ig) secretion. Data 1000 4000 determined using a ordinary one-way ANOVA with 40000 5000 I 800 *** *** Tukey's multiple comparisons test. suppression of pro-inflammatory cytokine (i.e. TNFα, IFNγ, IL-1β, IL-2, IL-6, IL-17A, GM-CSF, etc.) release from stimulated healthy in (B) are representative of results from 8 donors. 5000 10000 500 2000 (B) Spleens were harvested on D14 from the first

SjS, or SLE patient PBMCs (Fig. 3). ALPN-101 treatment reduced germinal center (GC) B cells and follicular helper T cells (TFH) 0 0 0 0 cohort of mice (treated on D-1 & D7) and analyzed by 0 0 0 (C) RNA expression analysis of key genes involved in B-T cell collaboration (C) RNA isolated from a B/TFH co-culture similar to the one D14 1:45 Diln D28 1:136 Diln D14 1:45 Diln D28 1:409 Diln D14 1:45 Diln D28 1:409 Diln D14 1:45 Diln D28 1:45 Diln flow cytometry. and inhibited antibody production in vivo in mouse immunization models (not shown) and in the bm12 model (Fig. 9), and SjS SLE HD SjS SLE HD 6 SjS SLE HD Day Post-Immunization Day Post-Immunization Day Post-Immunization Day Post-Immunization (C) Per a published protocol , splenocyte suspensions described in A was analyzed for relative gene expression levels bm12 bm12 Donor Type from female I-A B6(C)-H2-Ab1 /KhEgJ (‘bm12’) suppressed proliferation and antibody production in human B cell/TFH cell co-cultures (Fig. 5). In anti PD-L1-treated NOD mice, Donor Type Donor Type by next generation sequencing. Expression values in RPKM are (B) Numbers of splenic GC B and T in BALB/c mice 14 days after immunization with SRBC mice were injected IP into an equivalent number (1:1) ALPN-101 suppressed sialadenitis, insulitis, blood glucose levels, and autoantibodies with activity often superior to that of plotted on a relative scale from 0 to 1 for each gene between FH of 9 wk old female C57BL/6NJ recipient mice on Day 0. H2-Ab1bm12 differs from H2-Ab1b by 3 amino acids (C) Summary of cytokine responses in the PBMC stimulation assay treatment groups, with degree of red reflecting higher # GC B cells/spleen # CD4+ T cells/spleen abatacept (Fig. 6-8, and data not shown). (A) In vitro stimulation of healthy ) FH in the β-chain of the I-A molecule. Alloactivation of 1.00 3

0 150 )

1 donor CD4+ T cells by recipient APC leads to a

expression and degree of blue representing lower values for 6

donor, Sjögren’s syndrome (SjS), x 0

Sjögren’s Systemic Lupus Healthy ( SRBC Day 0 cGvHD disease with symptoms resembling SLE.

CONCLUSION: The efficacy of dual CD28/ICOS antagonist ALPN-101 is superior to CD28 or ICOS costimulatory pathway 1 x each gene. The list of genes shown was sorted by level of n

( 0.75 e Syndrome Erythematosus Treated on D-1 and D7 Starting on Day 0, recipient mice were treated 2x/wk

Donors and SLE patient PBMC with e n

l 100 e inhibition by ALPN-101 (least to most inhibited, from L→R). p Sac D14 for 11 wks with 400 µg ALPN-101 or abatacept, or a

inhibitors, administered individually or in combination, in human in vitro and/or mouse in vivo translational studies. A Phase 1 e

/ p

artificial APC [fixed K562 0.50 # molar-matched amount (250 µg) of Fc control protein.

l /

l + + + + # clinical trial with ALPN-101 in healthy volunteers is ongoing (NCT03748836), and trials in inflammatory diseases are planned. e At study termination on Day 82, GC B cells and TFH

GC B = B220 CD19 GL7 CD95

c l

expressing cell surface OKT3 (anti + + + + + e

0.25 H TFH = CD4 CD45 CD3 PD1 CD185 cells were enumerated by flow cytometry. Anti-double

Anti ICOSL mAb C stranded (ds) DNA IgG titers were measured on D14

CD3)/CD80/ CD86/ICOSL)] at a + G 4 using the Mouse Anti-dsDNA IgG Antibody Assay kit (prezalumab) 0.00 D 0 Figure 1: ALPN-101 (ICOSL vlgD-Fc), Generated Using the vIgD™ Directed Evolution 20:1 ratio. Test articles were added Figure 6: The Anti-PD-L1 mAb-Induced NOD Mouse Model of Sjögren’s Syndrome C from Chondrex (Redmond WA). ICOS only at 100 nM and supernatants were Strategy, Blocks Both CD28 and ICOS T Cell Costimulation Pathways (C) Numbers of splenic GC B and TFH and serum anti-dsDNA titers in the bm12 inducible model of SLE collected and assayed for cytokine Day 0 Day 2 Day 4 Day 6 Day 10 % Inhibition # GC B Cells Per Spleen # CD4+ TFH Cells Per Spleen Serum anti-dsDNA (IgG) CD28/CTLA-4 inhibition CD28+ICOS dual blockade ICOS-only inhibition concentrations after 48 hr Anti-mouse PD-L1 5×105 5 5000 vs. Fc * 4×10 * ) ****

mAb (0.1 mg), IP L

ALPN-101 inhibits both CD28 and incubation. (B) IFNγ, IL-6, and * ** **** n

5 m /

n Fc Control

APC/B cell ICOS-mediated costimulation. Abatacept e 4×10 4000

e g Day 10 Harvest: e 5

IL-17A levels are shown as l e

3×10 n

Comparators such as abatacept / l ( p ALPN-101

5 (CTLA4-Fc) 3×10 p • SMG (histology, RNA-Seq) S

belatacept (CTLA4-Ig, BMS) or FR104

examples, though similar trends A S

ALPN-101 / 3000

ALPN-101

CD80/86 / CD80/86 CD80/86 N Abatacept

(anti CD28 Fab, OSE CD28 only

ICOSL ICOSL ICOSL • Pancreas (histology) # 5 5

(B7) • Dose with anti-mPD-L1 mAb (0.1 mg) D (B7) (B7) were observed for all cytokines (0.5 mg), IP #

l 2×10 2×10

Immunotherapeutics) inhibit only the l

s l

l Naïve BL/6 e

• Treat with 0.3 mg Fc control, 0.5 mg ALPN-101, • Blood/serum (glucose, PK) d 2000 e

CD28/CTLA-4 pathway, while - c

measured, as summarized in C. i

CTLA4-lg c t

prezalumab (anti ICOSL mAb, Amgen) 1×10 B

abatacept, or WT ICOSL-Fc, or 0.5 mg aba + n H

1×105 a

inhibits only ICOS costimulation. (C) ALPN-101 demonstrated F

C 1000

NOD/ShiLtJ female, 0.5 mg WT ICOSL-Fc 0 T

G G

statistically significant superiority g prezalumab CD28 costimulation is critical for naïve ALPN-101 6 wks of age I + 5 0 0 and memory CD4 T cell activation. Monotherapy to prezalumab, abatacept, or a -1×10 CD28 ICOS CD28 ICOS CD28 ICOS ICOS signaling augments CD28 A mouse model of Sjögren’s syndrome induced in female diabetes-prone NOD/ShiLtJ mice using repeat dosing (on Days 0, 2, 4, and 6) with anti-mouse PD-L1 combination of prezalumab+ Statistical significance was determined for GC B and TFH using an unpaired t-test, and for anti dsDNA concentrations using an uncorrected Fisher’s least significant differences test; *p<0.05, **p<0.01, and ****p<0.0001. costimulation but also plays ICOS+CD28 antibody (based on a modified version of a protocol published by Zhou et al., 20165) was used to evaluate the impact of ALPN-101 and comparator treatments on non-redundant roles; ICOS deficiency abatacept for the majority of leads to defective humoral responses the development of sialadenitis and insulitis, the levels of blood glucose, serum autoantibodies, and gene expression in the submandibular gland (SMG). in both mice and humans.1 donors and analytes tested. Summary and Conclusions ICOS signaling is crucial for the Figure 7: ALPN-101 Reduces the Incidence and Severity of Sialadenitis in NOD Mice 2 • ALPN-101 (ICOSL vIgD-Fc) is a dual CD28 and ICOS T cell co-stimulation pathway inhibitor targeting both naïve development of human TH17 cells and the function & maintenance of follicular 7 3 Enrolled in the Anti PD-L1 mAb-Induced Model of Sjögren’s Syndrome and activated pathogenic T cells, including ICOS+ cells that escape currently available CD28 pathway inhibitors. T cell T cell Inactivation T cell helper T cells. ALPN-101 may thus be activation T cell uniquely suited to treat diseases Figure 4: ALPN-101 Suppresses Expression of Many Genes Associated with activation driven by these pathogenic cell types. • ALPN-101 inhibits cytokine production in vitro from human Sjögren’s and SLE patient PBMC, more potently than SjS/SLE Mechanisms and Phenotypes % Incidence of Sialadenitis Individual SMG Histology Scores single CD28 or ICOS pathway inhibitors. 80 *

• RNA was isolated from patient PBMC 4 s ALPN-101 Binds CD28 and ICOS and Blocks Binding to Their Ligands, i n=20 **** • ALPN-101 affects genes involved in B cell differentiation and suppresses proliferation and antibody responses in

Figure 2: t ) samples generated in the assay i **

n=20 n=3 4

• CCL20 n - * vitro in human B cell-TFH cell co-cultures. described in Fig. 3 e

Impacting B Cell/T Cell Collaboration During Antibody Responses • CD40LG 0 ( d 60 • CSF2 n=20 3

Antigen • Genes significantly down-regulated in a Fc Control l APC o Primary immune responses e • ALPN-101 suppresses anti-SRBC (and anti-KLH, not shown) antibody responses in vivo in normal mice, and

• ICOS r

ALPN-101 a CD28+ depend heavily upon the CD28 response to ALPN-101 across SjS, SLE, i n=19 Binding to Targets Blockade of Ligand Binding • ID2 ALPN-101 o

S reduces autoantibody titers in mouse models of Sjögren’s syndrome and SLE.

pathway c

and psoriatic arthritis (PsA) stimulated f CD28 CD80  CD28 • IFNG

Naïve S o

40 2 Inhibition of T cell • IGSF3 Abatacept 50000 10000 T cell patient populations suggest ALPN-101 y • ALPN-101 reduces the incidence and severity of sialadenitis and insulitis, and reduces blood glucose levels (Fig. 7 WT ICOSL activation o Secondary immune responses e

IC50 • IL17A

g e

e Prezalumab can impact pathogenic genes and/or c c (e.g. memory, effector) can be

40000 n WT ICOSL-Fc

n 1.0 nM n • IL1R1

Abatacept ALPN-101 • BATF3 o and data not shown), in NOD mice enrolled in the anti PD-L1-induced model of Sjögren’s syndrome. e

e 7500

e c

c Belatacept Abatacept 5.4 nM CD28-independent. pathways implicated in many s s • IL2

• CTLA4 o e

e CD80/CD86+

d r

r 30000 ALPN-101 2.5 nM t

Belatacept i 1 o

o • IL21 20 n=20 Aba + WT u

u 5000 autoimmune diseases. • FLT1

l l

o ICOS is upregulated upon T-cell s • ALPN-101 is a novel therapeutic candidate for Sjögren’s syndrome, SLE, and potentially other connective tissue

20000 • IL24

n n

n • FOSL1

i i

activation and is the most closely • IL3 Naïve H d

d 2500 • IL13 and/or serious autoimmune diseases. A phase 1 clinical trial with ALPN-101 in healthy volunteers is ongoing e

e 10000

% M M related protein to CD28, yet plays • IL17F • IL31 ICOS+ • CCL17 • KITLG 0 (NCT03748836), and trials in inflammatory diseases are planned. 0 0 Existing a non-redundant function in many • IL6 0 Activated • CXCR5 • LTA • IL1R2 10 100 1000 10000 1000001000000 100 1000 10000 100000 1000000 TFH cell TFH cell immune responses, such as • IL9 Statistical significance determined using the Kruskal-Wallis test: * p < 0.05, ** p < 0.01, ICOS+ • GATA3 • PVR • IL23R [ pM ] [ pM ] • LIF **** p < 0.0001. The naïve group was not powered for statistical comparisons. follicular helper T cell (TFH) • IL12RB2 • SOCS3 • IL4 ICOS ICOSL  ICOS 1,3 • NFKBID References ALPN-101 differentiation and function. • IL18R1 • TFRC • IL5 • PDCD1 Representative H&E Images of SMG IC50 • VEGFA 60000 WT ICOSL 75000 • IL18RAP • TNFRSF8 1. Panneton et al. (2019) Inducible T cell Costimulator: Signaling Mechanisms in T Follicular Helper Cells and Beyond. Immunol Rev. 291:91-103

WT ICOSL 17.2 nM o ICOS+ T cells escape treatment • TNF e

e Prezalumab ICOS+ 2. Paulos et al. (2010) The Inducible Costimulator (ICOS) is Critical for the Development of Human TH17 Cells. Sci Transl Med. 2(55):55ra78. c c Abatacept WT ICOSL-Fc Aba + WT ICOSL-Fc

23.3 nM Fc control ALPN-101 n n Abatacept Prezalumab

e with drugs only blocking the

e 3. Weber et al. (2015) ICOS Maintains the T Follicular Helper Cell Phenotype by Down-Regulating Krüppel-Like Factor 2. J Exp Med. 212(2):217-33. c

c Belatacept ALPN-101 4.3 nM Inhibition of B cell s

s 40000 50000 e

e CD28/CTLA-4 pathway, like r

r ALPN-101 ICOS-L+ activation and Ab 4. Evans et al. (2019) ALPN-101, a First-in-Class Dual ICOS/CD28 Antagonist, Suppresses Key Effector Mechanisms Underlying Rheumatoid and

o u

u production l

l abatacept. Psoriatic Arthritis. Poster #1531, American College of Rheumatology Annual Meeting 2019; Atlanta, GA.

n 20000 25000 Antibody- B cell Memory a

a 5. Zhou et al. (2016) Endogenous Programmed Death Ligand-1 Restrains the Development of Sjögren’s Syndrome in Non-Obese Diabetic Mice.

i d

d secreting B cells Our Thesis: Blockade of both e

e Sci Rep. 6: 39105. M M plasma cells See also: Poster #1531. Evans et al. (2019) ALPN-101, a First-in-Class Dual ICOS/CD28 CD28 and ICOS pathways by 6. Klarquist & Janssen (2015) The bm12 Inducible Model of Systemic Lupus Erythematosus (SLE) in C57BL/6 Mice. J Vis Exp. (105): e53319. 0 0 ALPN-101 should provide superior Antagonist, Suppresses Key Effector Mechanisms Underlying Rheumatoid and Psoriatic 10 100 1000 10000 1000001000000 100 1000 10000 100000 1000000 7. Dillon et al. (2019) ALPN-101, a Dual ICOS/CD28 Antagonist, Demonstrates Potent and Dose-Dependent Suppression of Graft vs. Host Disease [ pM ] efficacy to therapeutics which only 4 1 µm [ pM ] Arthritis. ACR Annual Meeting; November 11, 2019. (GvHD) in a Human/NSG™ Mouse Xenograft Model, with Activity Superior to CD28 or ICOS Single Pathway Antagonists. Poster #426, interfere with a single pathway Transplantation & Cellular Therapy Meeting of ASBMT and CIBMTR, February 2019, Houston, TX.