Gain-of-function alterations in histone-modifying proteins drive multiple cancers and are now being targeted by compounds in the clinic,1 but loss-of-function mutations, which are more frequently found in chromatin remodelers, remain largely intractable. Separate teams from Japan, Boston and Novartis AG now have identified synthetic lethal interactions that could be exploited to help treat cancers with mutations in the SWI/SNF chromatin-remodeling complex.2-5

The key question is whether these targets-core regulatory proteins within the SWI/SNF (switch/sucrose nonfermentable) complex-can be selectively inhibited with an acceptable therapeutic index.

SWI/SNF is a large multiprotein complex that plays diverse roles in regulating transcription and DNA replication and repair by altering chromatin structure. Loss-of-function mutations in members of the complex have been found in several tumor types.

At the SciBX Summit on Innovation in Drug Discovery and Development last October, Charles Roberts said that these mutations were ripe for further functional exploration.

"Eight different subunits of this complex are currently mutated in cancer at quite high frequencies-the latest data suggest that at least 20% of all human cancers have a mutation of one or another SWI/SNF subunit. Further, the genes encoding these subunits are being validated as bona fide tumor suppressors using mouse models. Now, an important question is, what can we do about it therapeutically?" asked Roberts, who is an associate professor of pediatrics at Harvard Medical School and an associate professor of pediatric oncology at the Dana-Farber Cancer Institute.

Initial work suggested that uncovering new genetic dependencies in SWI/SNF-mutant cancers could point the way to actionable targets. For example, mutations in the SWI/SNF subunit SNF5 (SWI/SNF-related matrix-associated actin-dependent regulator of chromatin subfamily b member 1; SMARCB1) cause the rare pediatric cancer malignant rhabdoid sarcoma. Studies from Roberts6,7 and cancer epigenetics company Epizyme Inc.8 showed that these cancers require BRG1 (SWI/SNF-related matrix-associated actin-dependent regulator of chromatin subfamily a member 4; SMARCA4) and EZH2 (enhancer of zeste homolog 2) function.

BRG1 knockdown or EZH2 inhibition killed SNF5-mutant malignant rhabdoid sarcoma cells. Epizyme's EZH2 inhibitor E7438 is in Phase I/II trials to treat lymphoma.

However, SNF5 is only one of many SWI/SNF complex components mutated in cancer. The dependencies created by mutations in other subunits remained unknown.

Now, three separate teams including Roberts' have shown that loss-of-function mutations in BRG1 cause cells to become dependent on the highly related SWI/SNF component BRM (SMARCA2).2-4

An additional study also by Roberts' group homed in on mutations in the SWI/SNF complex member AT rich interactive domain 1A (ARID1A) and showed that cells with the mutation became dependent on the closely related protein ARID1B.5

BRGing in

The teams took distinct approaches to arrive at the same conclusion, namely, that BRM is required for viability in tumors with mutations in BRG1.

The first team to publish its results was led by Takashi Kohno, chief of the Division of Genome Biology at Japan's National Cancer Center Research Institute. Because BRM and BRG1 are highly homologous and play complementary roles, the group sought to test whether depletion of BRM might be synthetically lethal in BRG1-mutant cell lines.

Indeed, in a panel of BRG1-mutant non-small cell lung cancer (NSCLC) cell lines, siRNA against BRM significantly decreased viability compared with control siRNA, whereas BRG1-intact cells were unaffected.

Adding back wild-type BRG1 restored viability, whereas adding back a catalytically inactive mutant form of BRG1 did not. In a xenograft model of BRG1-mutant NSCLC, depletion of BRM decreased tumor volume compared with no depletion. Results were published in Cancer Research.4

A second study published in January by Roberts in collaboration with researchers at the Broad Institute of MIT and Harvard took an unbiased approach using Project Achilles, a genome-wide shRNA screening effort using genetically defined cancer cell lines.

When 8 cell lines with unambiguous inactivating alterations in BRG1 were compared to the remaining 157 cell lines, BRM emerged as the top essential gene. Results were published in Molecular and Cellular Biology.3

In a separate analysis of the Project Achilles data published in Nature Medicine, Roberts' lab also showed that mutations in another SWI/SNF complex member, ARID1A, sensitized cells to depletion of ARID1B.5

Finally, a publication on BRM dependency from a team led by Frank Stegmeier and Zainab Jagani at the Novartis Institutes for BioMedical Research (NIBR) described a high-coverage shRNA screen of chromatin regulators that pinpointed BRM as an essential gene in BRG1-mutant cancers. Results were published in the Proceedings of the National Academy of Sciences.2

Both Roberts' and Stegmeier's teams showed that knockdown of BRM did not completely disrupt the SWI/SNF complex. That finding suggested that the activity of the BRM-containing SWI/SNF complex was responsible for the oncogenic effects caused by the BRG1 mutations.

Stegmeier is director and head of Cambridge oncology target ID and validation at NIBR. Jagani is an investigator at NIBR.

The Novartis and Boston academic teams operated independently, although the teams communicated results before publication, and Jagani and Roberts are authors on each other's papers.

Chemical matters

The series of studies collectively makes the case for inhibiting BRM in the genetically defined subset of BRG1-mutant cancers. However, only one inhibitor of the target has been disclosed, and the molecule is nonselective. Thus, the safety of the approach remains an open question.

Roberts told SciBX, "A key question is whether there is a therapeutic window we can take advantage of."

Central to evaluating the therapeutic window will be the development of chemical matter that can selectively inhibit BRM or BRG1. The proteins consist of two potentially druggable domains: an ATPase domain required for chromatin remodeling and a bromodomain that binds to acetylated histone tails.

The Structural Genomics Consortium has developed a probe in collaboration with Pfizer Inc. The molecule, dubbed PFI-3, inhibits the bromodomains of BRM, BRG1 and polybromo 1 (PBRM1; PB1). Although the chemical properties of the compound are available on SGC's website, no studies detailing the function of the molecule have been published.

Stefan Knapp, a professor of structural biology at the University of Oxford and principal investigator of epigenetics chemical biology at SGC, told SciBX that the inhibitor has not yet been published because finding a biological effect has been challenging.

"The SWI/SNF complex contains many bromodomains, and predicting the outcome of chemical inhibition of a few of them is challenging. There is no effect of the inhibitor on cell proliferation, but we now see interesting phenotypes in developmental models. We hope that we will publish these data soon," said Knapp.

He added that, based on these findings, he does not expect that BRG1 or BRM bromodomain inhibitors would show the same effect as BRM knockdown experiments.

Patrick Trojer, senior director and head of biology at epigenetics company Constellation Pharmaceuticals Inc., said that it is still the early days when evaluating BRG1 and BRM as drug targets. "From these findings it is still a long way to go to a potentially successful therapy. It is not yet clear if one can engineer selectivity into small molecules inhibiting the ATPase domain of BRM and BRG1. The same question arises when considering drugging the BRM and BRG1 bromodomains," said Trojer.

He added, "It seems clear that inhibition of both BRM and BRG1 ATPase domains will be quite toxic and needs to be avoided. In general, ATPases have been difficult to drug and perhaps do not lend themselves to identifying potent small molecule inhibitors. But it is biological data like these that will get drug discoverers excited and perhaps initiate certain efforts to find out."

Stegmeier agreed that "additional studies are required to fully understand which domains and functions of BRM are critical for cancer dependency." He declined to disclose if Novartis is developing inhibitors of SWI/SNF complex members or whether any IP has been filed around this work.

Kohno and Roberts also did not disclose the IP status of their work.

Stegmeier said that future studies will focus on understanding the function of residual SWI/SNF complex activity in the mutant cancers and said that placing context around the role of the mutations will be important.

"Inactivating mutations in SWI/SNF co-occur in the context of other genetic lesions, and current models of inactivation are restricted to complete gene knockout in mouse models, making it difficult to clearly predict oncogenic effects as well as safety concerns," he said. "Such parameters would therefore have to be carefully studied and may differ depending on the mechanism of action of pharmacological inhibitors."

Roberts is further studying the role of residual SWI/SNF complex activity and did not disclose plans for development of inhibitors of BRG1, BRM or ARID1B. The last of the three targets could be the hardest to inhibit because it would require blocking protein-protein interactions.

Trojer wants to see additional studies elucidating the function of ARID1A and B in greater detail. He also noted that the PBRM1 subunit of the SWI/SNF complex is frequently mutated in cancer as well, so it would be interesting to look for synthetic lethal dependencies in PBRM1-mutant cells.

Trojer did not disclose whether Constellation is pursuing the targets.

Cain, C. SciBX 7(9); doi:10.1038/scibx.2014.245
Published online March 6, 2014

REFERENCES

1.   Lou, K.-J. SciBX 6(40); doi:10.1038/scibx.2013.1118

2.   Hoffman, G.R. et al. Proc. Natl. Acad. Sci. USA; published online Feb. 11, 2014; doi:10.1073/pnas.1316793111
Contact: Zainab Jagani., Novartis Institutes for BioMedical Research, Cambridge, Mass.
e-mail: zainab.jagani@novartis.com
Contact: Frank Stegmeier, same affiliation as above
e-mail: frank.stegmeier@novartis.com

3.   Wilson, B.G. et al. Mol. Cell. Biol.; published online Jan. 13, 2014; doi:10.1128/MCB.01372-13
Contact: Charles W.M. Roberts, Dana-Farber Cancer Institute, Boston, Mass.
e-mail: charles_roberts@dfci.harvard.edu

4.   Oike, T. et al. Cancer Res.; published online July 19, 2013; doi:10.1158/0008-5472.CAN-12-4593
Contact: Takashi Kohno, National Cancer Research Institute, Tokyo, Japan
e-mail: tkkohno@ncc.go.jp

5.   Helming, K.C. et al. Nat. Med.; published online Feb. 23, 2014; doi:10.1038/nm.3480
Contact: Charles W.M. Roberts, Dana-Farber Cancer Institute, Boston, Mass.
e-mail: charles_roberts@dfci.harvard.edu

6.   Wilson, B.G. et al. Cancer Cell 18, 316-328 (2010)

7.   Wang, X. et al. Cancer Res. 69, 8094-8101 (2009)

8.   Knutson, S.K. et al. Proc. Natl. Acad. Sci. USA 110, 7922-7927 (2013)

COMPANIES AND INSTITUTIONS MENTIONED

Broad Institute of MIT and Harvard, Cambridge, Mass.

Constellation Pharmaceuticals Inc., Cambridge, Mass.

Dana-Farber Cancer Institute, Boston, Mass.

Epizyme Inc. (NASDAQ:EPZM), Cambridge, Mass.

Harvard Medical School, Boston, Mass.

National Cancer Center Research Institute, Tokyo, Japan

Novartis AG (NYSE:NVS; SIX:NOVN), Basel, Switzerland

Novartis Institutes for BioMedical Research, Cambridge, Mass.

Pfizer Inc. (NYSE:PFE), New York, N.Y.

Structural Genomics Consortium, Oxford, U.K.

University of Oxford, Oxford, U.K.