0 7 14 21 28 35 42

0

200

400

600

800

1000

Days on Study

Mea

n Tu

mor

Vol

ume±S

EM (m

m3 )

0.001 0.01 0.1 1 10 100 10000.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Concentration (nM)

OD

450n

m

XMT-2056, a well-tolerated, Immunosynthen-based STING-agonist antibody-drug conjugate which induces anti-tumor immune activity

Jeremy R. Duvall, Raghida A. Bukhalid, Naniye M. Cetinbas, Kalli C. Catcott, Kelly Slocum, Kenneth Avocetien, Keith W. Bentley, Stephen Bradley, Susan Clardy, Scott D. Collins, Elizabeth Ditty, Timothy Eitas, Brian D. Jones, Eugene W. Kelleher, Winnie Lee, Travis Monnell, Rebecca Mosher, Marina Protopopova, LiuLiang Qin, Pamela Shaw, Elena Ter-Ovanesyan, Joshua D. Thomas, Phonphimon Wongthida, Ling Xu, Liping Yang, Jeffrey Zurita, Dorin Toader, Marc Damelin, Timothy B. Lowinger

Mersana Therapeutics, Inc., Cambridge, MA

Abstract ID: 1738

Abstract - STING pathway agonism has emerged as a potential therapeuticmechanism to stimulate an innate anti-tumor immune response. However, thesystemic administration of a free STING agonist may be limited by toxicity, andbroad biodistribution may not be ideal. Antibody-drug conjugates (ADCs)constitute a proven therapeutic modality that enables tumor-targeted delivery andthus is ideally suited to systemic administration with reduced toxicity. To developan optimized STING agonist ADC platform, we designed a novel STING-agonistspecifically tailored for use in an ADC. Determination of the co-crystal structureconfirmed that the agonist binds to the closed, or ‘active’, conformation of theSTING homodimer.The resulting Immunosynthen platform, which was developed specifically for theselected STING agonist payload, was used to generate XMT-2056, a tumorantigen-targeted STING-agonist ADC with excellent drug-like properties and>100-fold increased potency as compared to the free STING-agonist payload. Inmice, XMT-2056 induced robust anti-tumor immune activity, with only minimalincreases in systemic cytokine levels, and exhibited significant benefit over thebenchmark free STING-agonist payload in both regards. Additionally, in vitro andin vivo studies demonstrate that XMT-2056 is able to activate the STINGpathway in both tumor-resident immune cells and tumor cells, offering a potentialadvantage over other innate immune activating pathways. XMT-2056 was well-tolerated in non-human primates at significantly higher exposure levels thanthose required for anti-tumor activity, and the ADC exhibited favorablepharmacokinetics after repeat doses. Together these data support the clinicaldevelopment of XMT-2056.

IRF3

STING PathwayCytosol

dsDNA

cGAMP

Type I IFN, ISGs

Endoplasmic Reticulum

Nucleus

cGAS

STINGTBK1

IRF3IRF3

Anti-Tumor Immune Responses

Depiction of the STING pathway (adaptedfrom Corrales et al1). cGAS recognizes andbinds to cytosolic dsDNA and synthesizescGAMP, a natural STING agonist that activatesSTING, which in return activates TBK1-mediatedphosphorylation of IRF3 transcription factor,leading to expression of type I interferon andother interferon stimulated genes and eventuallyto an anti-tumor immune response.

STING Agonist ADC Approach Could Address Administration Issues, Systemic Tolerability, and Activity

• Systemic administration with targeted delivery to the tumor

• Improved anti-tumor activity compared to free agonist

• Improved tolerability compared to free agonist

Intratumoral STING Agonist

Systemic Free STING Agonist

STING-Agonist ADC

Tumor with STING-Mediated Innate Immune Activation

Tumor, no immune activation

Systemic immune activation

STING Agonist ADCs Deliver a One-Two Punch, Activating STING in Myeloid and Tumor Cells2

Killing

Growth

Tumor Cells STING Knockout Tumor CellsTumor Cell Killing in Immune Cell (PBMC) Co-Culture

Can

cer c

ell c

onflu

ency

at 8

0 ho

urs

(nor

m. t

o t=

0)

-20

-15

-10

-5

0

5

10

15

20

Payload dose5 nM 25 nM -20

-15

-10

-5

0

5

10

15

20

Payload dose5 nM 25 nM

Figure 1. Tumor-targeted Fc silent ADC activity is diminished in STING knock out cancer:PBMC co-cultures. STING wt and ko cancer cells were co-cultured with PBMCs and cancer cell killing activity of theXMT-2056 surrogate Fc wt or Fc silent ADCs, and control ADC were assessed by IncuCyte analysis. XMT-2056 surrogate with wt Fc induced significant killing of both STING wt and ko cancer cells in PBMC co-cultures, whereas Fc silent XMT-2056 surrogate induced significant killing of only STING wt cells but hadno activity in STING ko cancer cell:PBMC co-cultures. These data demonstrate the contribution of tumor-intrinsic STING pathway activation to the anti-tumor activity of the tumor cell-targeted STING agonist ADCsin in vitro co-cultures of human cancer cell : primary immune cell co-cultures. For a more in-depthdiscussion on tumor intrinsic activity with a STING agonist ADC, please refer to abstract 1773.

Control ADCXMT-2056 surrogateFc-silent XMT-2056 surrogate

Highly Modular Approach Allowed for Optimal ADC Synthesis

Building ADCs with a modular approach:• Flexibility in design enables optimization of ADC for optimal pharmacological and

pharmacokinetic properties.• Modular components enable fine-tuning of drug-to-antibody ratio (DAR).• Amenable to many bioconjugation methods.• Ultimately the ADC is optimized for the target, antibody, and payload.

Precisely Optimized for a STING agonist

STING agonistLinkerScaffoldBioconjugationAntibody

Critical Attributes Matched to Antibody and Target

Conclusions• XMT-2056 delivers a STING agonist payload which binds to the closed conformation

of the protein and designed with properties suitable for ADC development.• XMT-2056 has potent in vitro STING activity with >100-fold improvement in activity

over the free payload, exhibiting a ‘one-two punch’ unique to STING agonist ADCs.• XMT-2056 shows excellent in vivo efficacy after a single IV dose, while having

minimal effect on systemic cytokines.• XMT-2056 has good exposure in NHPs and is well tolerated at a dose level ~10-fold

higher than required for sustained tumor regression in mice models.• These data support further development of XMT-2056 with expected FIH dose in

early 2022.

XMT-2056 Shows Good Exposure After Multiple Doses and is Well Tolerated in Non-Human Primates

References:1. Corrales et al. JCI 2016, 126: 2404-2411.2. Cetinbas et al. SITC 2020 poster; Cetinbas et al. AACR 2021 poster. Both posters available at

www.Mersana.com.3. Ramanjulu et al. Nature 2018, 564: 438-443.

NHP ResultsSingle-dose and repeat-dose studies at 9

mg/kg antibodyIntravenous administration

• No clinical signs, no mortality• High exposure, high ADC stability in

circulation• Transient elevation of 5 cytokines out of

24 tested• No adverse changes in clinical

pathology• No adverse findings in histopathology

• High stability as indicated by parallel curvesof antibody and conjugated drug

• Space between curves reflectshigher molecular weight of antibodyrelative to STING-agonist payload

• Comparable PK profiles after 1st and 2nd

dose

1 8 15 22 291

10

100

1000

10000

100000

1000000

Time (Days)

Plas

ma

Conc

entra

tion

(ng/

mL)

Quantitation Limits

Plasma Concentrations in NHP(Total Antibody and Conjugated STING agonist )

Plasma Concentrations in NHP vs. Mouse

(Conjugated STING agonist)

Exposure of XMT-2056 at a well-tolerated dose inNHPs is ~10-fold higher than the exposurerequired for sustained tumor regression in mouse

1 2 3 4 5 6 7 810

100

1000

10000

100000

Time (Days)

Plas

ma

Conc

entra

tion

(ng/

mL)

NHP,9 mg/kg

Mouse,1 mg/kg

Vehicle Control ADC (3/0.12 mg/kg)

XMT-2056 surrogate(3/0.09 mg/kg)

diABZi IV STING Agonist (0/5 mg/kg)*(mAb/Payload dose)

Figure 6. STING agonist ADC induces significantly lower levels of serum cytokines than the diABZISTING agonist administered intravenously. Luminex analysis of systemic cytokines in serum samples oftumor bearing mice treated with a single dose of XMT-2056 surrogate or control ADC and the diABZI IVSTING agonist. Systemically delivered diABZI STING agonist induced 6-100x higher systemic cytokines thantargeted STING agonist ADC despite the lack of sustained anti-tumor activity.

Doses of STING ADC that Result in Complete Tumor Regressions Induce Significantly Lower Levels of Serum

Cytokines than the IV STING AgonistCXCL10 IL-6

pg/m

L

0 20 40 60 800

1000

2000

3000

4000

Hours

0 20 40 60 800

5000

10000

15000

20000

25000

Hours0 20 40 60 80

0

2000

4000

6000

Hours

0 20 40 60 800

50

100

150

200

Hours

TNF-α

pg/m

L

IFNγ

Systemic Free STING Agonist

STING-Agonist ADC

XMT-2056 Demonstrates Excellent In Vivo Efficacy and PK

Figure 5. XMT-2056 induced durable and complete tumor regressions in xenograft mouse models atsignificantly lower doses by payload than the IV administered diABZI STING agonist3 anddemonstrated excellent in vivo PK properties. Single doses of XMT-2056 administered intravenouslyresulted in tumor regression in a dose dependent manner. Control ADC, unconjugated mAb, and freepayload at equivalent doses show little to no effect demonstrating the advantage of the ADC. The diABZI IVagonist (administered as described in reference 3) showed only modest effect on tumor growth inhibition. PKassessment in non-tumor bearing mice showed excellent exposure over the course of the study.

Identification of an Optimal Payload for ADC Delivery

XMT-2056 is a Potent STING Agonist In Vitro Demonstrating Significant Improvement Over the Free Payload

XMT-2056 Demonstrates Good Physicochemical Properties

*Ligand has been removed from structure

THP1 R232 (CXCL10) 8 nM

THP1 R232 (IFNβ) 16-40 nM

J774 mouse (IRF3) 27 nM

Human STING Knockout >10,000 nM

Cytotoxicity (THP1, EC50) >10,000 nM

Safety Screen 44 - Panlabs No activity at 1 µM

10-4 10-2 100 102 104 1060

5000

10000

15000

log[Agonist] nM

Luci

fera

se (R

LU)

THP1 CellsIRF3 Reporter

EC50 4 nM

0.0001 0.01 1 100 10000

0

10000

20000

30000

40000

50000

Concentration (nM)

RLU

IRF3

Rep

orte

r

XMT-2056 EC50 = 0.06 nM

Free Payload EC50 ~ 150 nM

Cancer cell /PBMC Co-Culture

CXC

L10

ELIS

A

Cancer cell/THP1 Co-Culture

XMT-2056 EC50 = 0.78 nM

Figure 2. Characterization of a STING agonist. A) THP1 cells engineered with a secreted luciferasereporter for IRF3 activity were incubated with compound for 2h before analysis showed dose-dependentactivity of the STING agonist. Additional characterization of the STING agonist shows similar activity inactivating mouse STING relative to hSTING activity and the high specificity of the compound. B) Highresolution co-crystal structure obtained demonstrating the agonist binds to the closed, or active,conformation of STING.

HIC SEC XMT-2056

Unconjugated mAb

Legend

A) B)

Figure 3. Analytical characterization of XMT-2056. HIC and SEC analysis of XMT-2056 compared tounconjugated mAb. Drug-to-antibody ratio (DAR) was determined by UV-Vis spectrometry. Amount ofunconjugated mAb was determined by HIC analysis. Amount of high molecular weight species (HMW)was determined by SEC analysis.

Figure 4. In vitro analysis of XMT-2056. Cancer cells were co-cultured with either engineered THP1 cells orPBMCs, treated with XMT-2056 and/or the free payload, and evaluated for STING activation by IRF3 orCXCL10 levels, respectively. XMT-2056 shows a significant approvement in activity compared to the freepayload. In mono-culture experiments, XMT-2056 showed no activity at the highest dose in both assays (datanot shown), demonstrating the necessity of both the cancer and immune cell for STING activation.

Attribute ValuemAb (mg/mL) >10.0

Unconjugated mAb (%) <1

DAR (UV-Vis method) ~8

HMW (%) <1

Figure 7. XMT-2056 shows good pharmacokinetics within NHP. Animals were treated with XMT-2056 at 9mg/kg on days 1 and 22 (q3w x 2) with a scheduled necropsy on day 29. Assessments included clinicalpathology, serum cytokines, and histopathology.

0 5 10 151

10

100

1000

10000

100000

Time (Days)

Plas

ma

Con

cent

ratio

n (n

g/m

l)

Plasma Concentrations in Mouse

Total Antibody – measuring total antibody in the plasma

Conjugated Drug – measuring drug levels which are attached to the antibodySpace between curves reflects higher molecular weight of antibody relative to STING-agonist payload

0.1 mg/kg by mAb

0.3 mg/kg by mAb

1 mg/kg by mAb3 mg/kg by mAb (0.128 mg/kg by payload)

VehicleFree payload (0/0.128 mg/kg)Unconjugated mAb (3/0 mg/kg)Non-Binding ADC Control (3.42/0.128 mg/kg)diABZI IV agonist (0/1.5 mg/kg q3d x 3)XMT-2056 (varied doses)

*(mAb/Payload dose)

Isotype ControlEC50 > 200 nM

Free Payload EC50 = 39 nM

Isotype ControlEC50 > 200 nM