Abstract # 1042 Development of AEB1102, an Engineered Human Arginase 1 for Patients with Solid Tumors S W Rowlinson, S E Alters, G Agnello, J Tyler, A Lowe, M Okamoto-Kearney, D Johnson, E M Stone*, G Georgiou*, and D G

Development of AEB1102, an Engineered Human Arginase 1 for Patients with Solid Tumors

S W Rowlinson, S E Alters, G Agnello, J Tyler, A Lowe, M Okamoto-Kearney, D Johnson, E M Stone*, G Georgiou*, and D G Lowe Aeglea Biotherapeutics 901 S MO PAC EXPY, Barton

Oaks Plaza One, Suite 250 Austin, TX 78746-5759 aegleabio.com *The University of Texas, Austin

Normal cells make their own supply of arginine using the enzymes ornithine transcarbamylase (OTC), argininosuccinate synthase (ASS1) and argininosuccinate lyase (ASL). In

many tumor cells, silencing one or more of these enzymes disables arginine synthesis, making tumor cells dependent on extracellular arginine uptake for survival (Fig 1 and 2). This makes tumors potentially vulnerable to arginine depletion by

AEB1102, an engineered form of human arginase 1. This optimized form of native human arginase 1, generated by substituting the manganese cofactor for cobalt results in a clinical candidate molecule with significantly improved

catalytic activity and stability (Fig 3). The goal of Aeglea Biotherapeutics is to perform all non-clinical and chemistry, manufacturing & controls (CMC) activities and initiate clinical development in both solid tumors and hematologic

A product development CPRIT Grant (TX)

provided insight into non-clinical animal pharmacology and manufacturing for AEB1102. Using AEB1102 from this prior grant, IND-enabling in vitro and in vivo non-clinical oncology studies were performed with the A375 melanoma model.

Additional in vivo studies were also performed using patient derived xenograft (PDx) models, with one of these models being derived from a patient with the B-RAF V600E mutation. A pilot dose range finding study with AEB1102

was performed in monkeys to identify doses to be utilized in subsequent GLP toxicology studies. Bioanalytical assays to determine PK and PD were developed and validated. AEB1102 CMC activities that were optimized included producing

material to support GLP toxicology studies. A Phase 1 clinical trial is now enrolling solid tumor patients at START in San Antonio and the

TUMOR ARGININE DEPENDENCE

Many tumors can t make Arg and

depend on an external supply for survival

LETHAL to cancer cells lacking

Silencing by methylation or transcriptional repression makes these

genes potential predictive biomarkers of AEB1102 efficacy

OTC: ornithine transcarbamoylase

ASS: argininosuccinate synthase

ASL: Arginosuccinate lyase

CLINICAL DEVELOPMENT OPPORTUNITIES FOR AEB1102

% of Patient Tumor Samples with Biomarkers of Arginine Dependence

295 183 55 20 51 71 40 20

119 20 27 214 54 47 88 13 98 83 21 701 16 # of Patient Tumor Samples

Predicted % Arginine Dependence

Data derived from multiple human tumor sample studies published in

scientific and medical literature.

Loss of OTC, ASS and ASL is due to transcriptional repression or epigenetic modification.

AEB1102 HAS IMPROVED PROPERTIES

Mn-Arg I = wild type arginase 1

The substitution of manganese for cobalt improves both the catalytic and stability

properties of AB1102 relative to wild type Arginase 1 (Fig 3). Completed CMC development activities include: a) optimized expression in bacteria, b) PEGylation procedures that ensure reproducibility, and c) buffer formulation conditions

that maintain product integrity and stability. To assess if these enhanced properties associated with cobalt substitution translate to improved in vivo efficacy, bioanalytical assays measuring arginine levels and AEB1102 activity were developed

for cynomolgus monkey plasma. Dosing of an equivalent amount of manganese and cobalt containing pegylated-Arginase 1 in monkeys confirms that AEB1102 has enhanced in vivo efficacy relative to wild type pegylated-Arginase 1 (Fig

4). Additional dose range finding studies showed a dose response effect on the duration of arginine reduction in the monkey with AEB1102 (Fig 5). These data were used to establish doses used in the GLP toxicology studies that

identified an NOAEL in both species as well as providing data that suggests the PK properties of AEB1102 are compatible with once per week dosing in the clinic.

AEB1102 VERSUS WILD-TYPE ARGINASE 1 IN CYNOMOLGUS MONKEYS

Arginine Concentration (M)

Single IV Dose in Primates

Wild Type-PEG 1 mg/kg

Lowest limit of detection

AEB1102 PRIMATE DOSE RANGE FINDING STUDY

Cynomolgus monkeys received i.v doses on days 1 and 8

Co-Arg1-PEG Conc. (nM)

0 48 96 144 192 240 288 336

Arg levels <2uM for 3 days

ARG Concentration (M)

Co-Arg1-PEG Conc. (g/mL)

0 48 96 144 192 240 288 336

ARG Concentration (M)

Arg levels <2uM for >7 days

To confirm the literature data that proposes certain tumor types as being candidates for responsiveness to arginine deprivation we performed quantitative in situ hybridization

for OTC, ASS1 and ASL on clinical samples from melanoma (Fig 6 A-C) and HCC (Fig 7 A-C) These histologies have been reported as having 100% penetrance of arginine dependence (Fig 2). Immunohistochemistry of ASS1 was also performed to confirm

the correlation of message and protein expression (Fig 6D and 7D). Our data confirms previous findings that melanoma is a cancer type that should be sensitive to arginine deprivation therapy owing to the loss of ASS1 expression in a significant

percentage of patient samples. In contrast ASS1 expression in HCC was not reduced to the same extent as that observed in melanoma and as such HCC was de-prioritized for further investigation.

To confirm efficacy of AEB1102 as an anti-tumor therapeutic the enzyme was tested against a panel of 10 xenograft cell lines with the A375 melanoma cell line being the most

sensitive to arginine depletion. In vivo efficacy with AEB1102 administered once weekly significantly delayed A375 tumor growth in mice and yielded a survival benefit (Fig 8). Additional in vivo studies were performed in melanoma PDx

models as these are more predictive of clinical efficacy. Like the A375 model, AEB1102 was effective at delaying tumor growth, doing so in both a B-RAF wild type (Fig 9A) and B-RAF V600E (Fig 9B) model.

QUANTITATIVE ISH AND IHC OF MELANOMA AND CONTROL SKIN

A OTC In Situ Hybridization

B ASS1 In Situ Hybridization

C ASL In Situ Hybridization

D ASS1 Immuno Histochemistry

QUANTITATIVE ISH AND IHC OF HCC AND CONTROL LIVER

A OTC In Situ Hybridization

B ASS1 In Situ Hybridization

C ASL In Situ Hybridization

D ASS1 Immuno Histochemistry

IN VIVO EFFICACY OF AEB1102 IN THE A375 MELANOMA MODEL

Efficacy of AEB1102 in A375 Xenograft

AEB1102 2mg/kg 1xwkly

AEB1102 4mg/kg 2xwkly

AEB1102 6mg/kg 3xwkly

IN VIVO EFFICACY OF AEB1102

IN MELANOMA PDX MODELS

A ST585 Melanoma PDx model

(Mean SEM of animal groups)

Vehicle (PBS) AEB1102 (6mg/kg, 1xwkly)

Tumor Volume (Cubic Millimeters)

ASS1 in situ score =0

B ME1154 Melanoma PDx model

(Mean SEM of animal groups)

Vehicle (PBS) AEB1102 (4mg/kg, Q7dx4)

Tumor volume (Cubic Millimeters)

0 5 10 15 20 25 30 35

ASS1 (log10FPKM) score = 0.91

Aeglea Biotherapeutics has successfully executed on all non-clinical and CMC

activities necessary to support the clinical development of AEB1102. Expression profiling of melanoma and HCC clinical samples indicates melanoma is a more arginine dependent histology owing to the greater loss of ASS1 expression and as such is

a candidate histology to pursue in future clinical studies. In support of this in vivo studies using melanoma pre-clinical models confirm the sensitivity of this histology to arginine deprivation with AEB1102. The Phase 1 open-label dose

escalation clinical study in solid tumors initiated patient dosing in October 2015. A second Phase 1 study in hematologic malignancies is planned for 2016.