Science - USA (2022-04-08)

PHOTO: JOHNS HOPKINS PATHOLOGY PHOTOGRAPHY LAB

science.org SCIENCE

INSIGHTS

PRIZE ESSAY

INNOVATION

Mutation-guided therapeutics

Development of bispecific antibodies to target

mutant peptides in cancer

By Jacqueline Douglass

A

t the start of my PhD, my advisers

tasked me with the challenge of using

antibodies to target driver mutation-

derived mutant peptides that are

presented on the surface of cancer

cells. I was new to the field of tumor

immunology and did not appreciate how

wild a concept this was or all the hurdles

that would await. My naïvety allowed me

to believe that this approach might work

against cancer cells in a Petri dish or even

in a mouse model. Yet I never imagined that

this challenge would blossom into a project

encompassing numerous team members

and multiple laboratories, resulting within

a few short years in several licensed patents

and a start-up company that develops clini-

cal-grade products.

Cancer development is driven by driver

mutations in critical oncogenes and tumor

suppressor genes. These mutations are spe-

cific to cancer cells; required for initiation

and maintenance of the cancerous state,

thus unlikely to be lost during the evolution

of cancer cells; and often shared among

many patients. These characteristics make

driver mutations preferred targets for can-

cer therapy.

The general concept of using antibodies

to target mutant peptides seemed straight-

forward enough. Mutations in the DNA of

cancer cells translate to mutant proteins.

When mutant proteins are degraded, some

of the resulting peptides that contain the

mutated amino acids can be presented on

the cancer cell surface by a type of protein

called human leukocyte antigen (HLA), re-

sembling a hot dog (the 8– to 11–amino acid

peptide) in a hot dog bun (the HLA protein)

( 1 – 3 ). A mutant peptide HLA (pHLA) on the

surface of a cancer cell would be an ideal

therapeutic target; its mutant amino acid

sets it apart from its wild-type counter-

part. The location of the mutant pHLA on

the cell surface would make these protein

complexes amenable to targeting through a

variety of therapeutic modalities, including

antibody-based therapies.

We began by developing a method to

identify antibodies specific to these mutant

pHLAs. Traditional hybridoma technology

failed to generate antibodies able to dis-

criminate between wild-type and mutant

peptides, which differ by a single amino

acid. We then designed and built two di-

verse phage display libraries to screen dif-

ferent bacteriophages—viruses that only

infect bacteria—that each express a distinct

antibody fragment for antibody clones spe-

cific to the target protein of interest ( 4 , 5 ).

We then needed to decide which mutant

pHLA complexes to target. Our primary

criterion was mutant peptides derived

from common driver mutations in can-

cers. However, we needed to assess which

peptides are presented on the cell surface

by HLA. We initially relied on in silico al-

gorithms to predict binding of peptides to

HLA proteins but soon found that the pre-

dictions could be misleading. Therefore,

we developed a mass spectrometry–based

method that was quantitative to determine

which peptides are presented as pHLA,

which revealed that these pHLA were typi-

cally present at very low numbers on the

cell surface ( 6 ).

After identifying mutant pHLAs to tar-

get, we had to convert our antibodies into

a format capable of killing cancer cells pre-

senting these complexes. However, there

was no precedent for an antibody-based

therapy that targets such a low number

of pHLAs on a cancer cell. We assessed

a range of modalities, settling on bispe-

cific antibodies: small bivalent molecules

that redirect T cells to kill cancer cells.

We tested dozens of bispecific antibody

formats before identifying one with the

sensitivity required to target the very low-

density mutant pHLAs ( 5 , 7 ).

We then had to demonstrate that the

bispecific antibodies were effective and

specific therapeutic agents. We used gene-

editing techniques to produce cancer cell

lines that differed by only a single mutation

of interest. These tailored cell lines allowed

us to show that our bispecific antibodies

could redirect T cells to kill only cancer

cells that harbor the mutant protein in in

vitro and in vivo mouse models (5, 7). To

better understand the determinants of the

antibodies’ specificities, we collaborated

with biophysicists to solve the structures of

our antibodies in complex with the mutant

pHLAs ( 7 – 9 ).

FINALIST

Jacqueline Douglass

Jacqueline

Douglass

received her

undergraduate

degree from the

Massachusetts

Institute of

Technology and an MD-PhD

from Johns Hopkins Univer-

sity School of Medicine. She is

currently an internal medicine

resident at Johns Hopkins and

plans to pursue medical oncol-

ogy training. http://www.science.org/

doi/10.1126/science.abo4237

Johns Hopkins School of Medicine, Johns Hopkins University,

Baltimore, MD, USA. Email: [email protected]

147-B 8 APRIL 2022 • VOL 376 ISSUE 6589

0408PrizeEssay_15347997.indd 150 4/1/22 4:36 PM