USP7 small-molecule inhibitors interfere with ubiquitin binding

LEttER

doi:10.1038/nature24006

USP7 small-molecule inhibitors interfere with

ubiquitin binding

1

,2

3

4

5

Lꢀrꢁa Kaꢂegaya *, Paꢀla Di Lellꢀ *, Liꢀꢁel Rꢀugé *, Ricꢃard Pasꢂꢀr , Keviꢁ R. Clark , Jasꢀꢁ Drummꢀꢁd , tracy Kleiꢁꢃeiꢁz ,

1

1,2

6

Eva Liꢁ , Jꢀꢃꢁ-Paul Upꢂꢀꢁ , Sumiꢂ Prakasꢃ , Jꢀꢃaꢁꢁa heideker , Mark McClelaꢁd , Maria Sꢂella Riꢂꢀrꢂꢀ , Dariꢀ R. Alessi ,

7

8

9

10

11

Maꢂꢂꢃias trꢀsꢂ , travis W. Baiꢁbridge , Micꢃael C. M. Kwꢀk , taylur P. Ma , Zacꢃary Sꢂiffler , Bradley Brasꢃer , Yiꢁyaꢁ taꢁg ,

11

4

9

12

Priyadarsꢃiꢁi Jaisꢃaꢁkar , Briaꢁ R. hearꢁ , Adam R. Reꢁslꢀ , Micꢃelle R. Arkiꢁ , Frederick Cꢀꢃeꢁ , Kebiꢁg Yu , Fraꢁk Peale ,

1

3

13

14

1

13

Flꢀriaꢁ Gꢁad , Maꢂꢂꢃew t. Cꢃaꢁg , Cꢃrisꢂiaaꢁ Klijꢁ , Elizabeꢂꢃ Blackwꢀꢀd , Scꢀꢂꢂ E. Marꢂiꢁ , William F. Fꢀrresꢂ ,

8

4

5

4

10

James A. Erꢁsꢂ , Cꢃudi ndubaku , Xiaꢀjiꢁg Waꢁg , Maureeꢁ h. Beresiꢁi , Vickie tsui , Carsꢂeꢁ Scꢃwerdꢂfeger ,

5

3

1,2

Rꢀberꢂ A. Blake , Jeremy Murray , till Maurer & Iꢁgrid E. Werꢂz

The ubiquitin system regulates essential cellular processes in cellular physiology (Fig. 1a)³. To identify USP7 inhibitors, we esta

eukaryotes. Ubiquitin is ligated to substrate proteins as monomers blished an activitybased highthroughput screening cascade (Extended

or chains and the topology of ubiquitin modifications regulates Data Fig. 1a). In parallel we developed a nuclear magnetic resonance

substrate interactions with specific proteins. Thus ubiquitination (NMR) screen to identify USP7 catalytic domainbinding fragments

directs a variety of substrate fates including proteasomal (Extended Data Fig. 1b). A hittolead selection cascade was established

,7–11

1

degradation . Deubiquitinase enzymes cleave ubiquitin from to guide optimization of highthroughput screening and NMR hits into

2

substrates and are implicated in disease ; for example, ubiquitin- more potent USP7 inhibitors. Selectivity assays and structural biology

specific protease-7 (USP7) regulates stability of the p53 tumour directed medicinal chemistry design at the apex of the cascade. Assays

3

suppressor and other proteins critical for tumour cell survival . to measure total and ubiquitinatedMDM2 levels were implemented

However, developing selective deubiquitinase inhibitors has as proximal indicators of USP7 activity, followed by viability assays

4

been challenging and no co-crystal structures have been solved using inactive compounds from matched chemical series as negative

with small-molecule inhibitors. Here, using nuclear magnetic controls (Fig. 1b).

resonance-based screening and structure-based design, we describe

Progression of highthroughput screening and NMR screen hits

the development of selective USP7 inhibitors GNE-6640 and GNE- through the hittolead selection cascade identified a series of fragments

6

776. These compounds induce tumour cell death and enhance having favourable properties (Extended Data Figs 1c–e, 2a–e).

cytotoxicity with chemotherapeutic agents and targeted compounds, Optimization of the parent fragment GNE2916 yielded GNE6640 and

including PIM kinase inhibitors. Structural studies reveal that GNE- GNE6776 (Extended Data Fig. 3a). These USP7 inhibitors enhanced

6

640 and GNE-6776 non-covalently target USP7 12Å distant from endogenous MDM2 ubiquitination while achieving selectivity

the catalytic cysteine. The compounds attenuate ubiquitin binding over USP47 (which has high USP7homology) and USP5, a highly active

12

and thus inhibit USP7 deubiquitinase activity. GNE-6640 and GNE- deubiquitinase . Both compounds accelerated endogenous MDM2

6

776 interact with acidic residues that mediate hydrogen-bond turnover with cycloheximide, a necessary cotreatment as MDM2 is a

5

interactions with the ubiquitin Lys48 side chain , suggesting that p53 target gene (Fig. 1a, c and Extended Data Fig. 3b). Accordingly, both

USP7 preferentially interacts with and cleaves ubiquitin moieties GNE6640 and GNE6776, but not controls GNE2118 and GNE6641,

that have free Lys48 side chains. We investigated this idea by promoted endogenous MDM2 ubiquitination with Lys48 (K48)linked

13

engineering di-ubiquitin chains containing differential proximal polyubiquitin chains, which direct proteasomal degradation (Fig. 1d

and distal isotopic labels and measuring USP7 binding by nuclear and Extended Data Fig. 3c). Active USP7 inhibitors stabilized p53 and

magnetic resonance. This preferential binding protracted the upregulated p21 in wildtype cells, but not in p53null or USP7null

depolymerization kinetics of Lys48-linked ubiquitin chains relative cells (Fig. 1e and Extended Data Fig. 3b). These data were corroborated

to Lys63-linked chains. In summary, engineering compounds that by significantly decreased viability of wildtype, but not USP7

inhibit USP7 activity by attenuating ubiquitin binding suggests null cells, treated with active inhibitors (Extended Data Fig. 3d, e).

opportunities for developing other deubiquitinase inhibitors and Thus GNE6640 and GNE6776 target cellular USP7, MDM2, and p53

may be a strategy more broadly applicable to inhibiting proteins that signalling pathways.

require ubiquitin binding for full functional activity.

Previously reported deubiquitinase inhibitors are neither potent nor

4

USP7 is a genetically validated deubiquitinase for MDM2, a ubi selective when evaluated in a massspectrometrybased activity assay .

quitin ligase that promotes degradation of the tumour suppressor p53. In this assay, GNE6640 and GNE6776 selectively inhibited recombi

USP7 depletion enhances ubiquitination of USP7 substrates, including nant USP7 relative to 36 other deubiquitinases (Fig. 2a and Extended

MDM2. Subsequent MDM2 proteasomal degradation stabilizes p53, Data Fig. 4a). GNE6776 remained selective even at 100μM, a more

3

,6,7

thus promoting cell cycle arrest and apoptosis . USP7 inhibition than sixfold higher concentration than used in cellular assays (Fig. 2a

enhances ubiquitination of additional substrates that also regulate and Extended Data Fig. 4b). Because deubiquitinases are regulated

¹Department of Discovery Oncology, Genentech, South San Francisco, California 94080, USA. ²Department of Early Discovery Biochemistry, Genentech, South San Francisco, California 94080,

USA. ³Department of Structural Biology, Genentech, South San Francisco, California 94080, USA. ⁴Department of Discovery Chemistry, Genentech, South San Francisco, California 94080, USA.

⁵Department of Biochemical and Cellular Pharmacology, Genentech, South San Francisco, California 94080, USA. ⁶MRC Protein Phosphorylation and Ubiquitylation Unit, School of Life Sciences,

University of Dundee, Dundee DD1 5EH, UK. ⁷Institute for Cell and Molecular Biosciences, Newcastle University, Newcastle upon Tyne NE2 1HH, UK. ⁸Department of Protein Chemistry, Genentech,

South San Francisco, California 94080, USA. ⁹Department of Microchemistry, Proteomics, and Lipidomics, Genentech, South San Francisco, California 94080, USA. ¹⁰Boston Biochem,

40 Memorial Drive, Cambridge, Massachusetts 02139, USA. ¹Department of Pharmaceutical Chemistry and Small Molecule Discovery Center, University of California, San Francisco,

1

8

San Francisco, California 94143, USA. ¹²Department of Research Pathology, Genentech, South San Francisco, California 94080, USA. ¹³Department of Bioinformatics and Computational Biology,

Genentech, South San Francisco, California 94080, USA. ¹⁴Department of Translational Oncology, Genentech, South San Francisco, California 94080, USA.

*

These authors contributed equally to this work.

0

0 M o n t h 2 0 1 7 | V o L 0 0 0 | n A t U R E | 1

©

reSeArCH Letter

a

b

Fig. 4e–i). Although efficacious exposure was only transiently achieved,

GNE6776 caused modest, although significant, EOL1 xenograft

growth delay (Extended Data Fig. 4j). Developing USP7 inhibitors that

have improved druglike properties will be necessary to comprehen

sively evaluate USP7 inhibition in vivo.

USP7

HTS and NMR hits

Other

inhibition

substrates

PTEN

N-Myc

HSV-ICP0

Foxp3

Endosomal

Epigenetic

Unknown

USP7

USP7 inhibition

USP5 and USP47

counterscreens

Structural biology

and MedChem

Ub

USP7 and MDM2 regulate proteins other than p53 (refs 3, 7–11, 15)

(Fig. 1a). Having confirmed the selectivity and cellular activity of GNE

Ub

≥10× selectivity

Ub

Protease selectivity

Cathespin-B

and caspases

MDM2

p53

6

640 and GNE6776, we prescreened 441 cell lines in 3day viability

Cellular MDM2 assays

Total MDM2 (IF)

and Ub-MDM2 (MSD)

assays to enable an unbiased analysis of cellular factors that regulate

USP7 inhibitor sensitivity (Extended Data Fig. 5a). GNE6641 was

inactive in all cell lines whereas GNE6640 decreased viability of 108

cell lines with halfmaximal inhibitory concentration (IC50)≤10μM

Biophysical analysis

DLS, covalent

Ub-MDM2 >

total MDM2

modiꢀcation

Cell viability

Compare active and

inactive compounds

(

data not shown). Next, we selected a subset of 181 cell lines displaying

p21

Caspase

cell cycle activation

arrest) (apoptosis)

(

a range of sensitivities for additional analysis. GNE6641 remained

inactive in 5day viability assays, whereas GNE6640 and GNE6776

≥2× selectivity

Lead compounds

WB: actin

decreased viability of 54 and 6 cell lines, respectively, with IC50 ≤ 1 0μM

(Extended Data Fig. 5b and Supplementary Table 1). Viability data and

c

WB: MDM2

d _W_i_l_d_-_t_y_p_e_H_C_T_-₁₁₆

CHX (h):

DMSO

0

0.5

1

3

5

7

0

0.5

1

3

5

7

exome sequencing analysis were integrated to identify features associated

with compound sensitivity. Acute myeloid leukaemia (AML) cell lines

had increased sensitivity to GNE6640 (Extended Data Fig. 5c, e). Loss

offunctionmutant TP53 lines were, in general, less responsive than

wildtype TP53 lines, but the trend was not significant (Extended Data

Fig. 5d). Indeed, the TP53 R175 hotspot was associated with increased

sensitivity, whereas the Y220 hotspot was associated with decreased sen

sitivity (Extended Data Fig. 5c, e and Supplementary Table 1). Although

both compounds decreased proliferation and activated caspases in TP53

wildtype cell lines, these effects were reduced, but not abolished, in

TP53null cell lines (Extended Data Fig. 6a–c). Combining GNE6640

with doxorubicin or cisplatin, DNAdamaging agents that activate the

7

5-

50-

GNE-6641

GNE-6640

GNE-6776

7

5-

2

1

50-

48-

1

0

.2

.0

.8

.6

.4

.2

.0

DMSO

Ub-

MDM2

GNE-6441

GNE-6440

GNE-6776

IP: K48-Ub

WB: MDM2

9

6

8-

4-

98-

WB: MDM2

1

48-

WB: USP7

WB: actin

50-

0

1

2

3

4

5

6

7

16

p53 response , enhanced USP7 inhibitor efficacy (Extended Data

CHX (h)

Fig. 6d, e). Thus the MDM2 or p53 pathway is a mediator of GNE6640

and GNE6776 cellular activity but is not the only determinant. These

e

Wild-type HCT-116 USP7 null HCT-116 p53 null HCT-116

−

/−

findings corroborate genetic studies that have crossed TP53

and

USP7⁻mice, which minimally rescues USP7 embryonic lethality .

The enhanced efficacy of GNE6640 in combination with DNA

damaging agents suggested a strategy to identify signalling pathways

that intersect with USP7 inhibition. We therefore screened EOL1 AML

cells with GNE6640 or GNE6776 in combination with 589 chemo

therapeutic or targeted agents. Top hits included PI3K and PIM kinase

inhibitors (Extended Data Fig. 6f, g and Supplementary Table 2). PIM

kinases share substrates with PI3K pathway kinases including TSC1/2

/−

−/−

7

WB: p53 50-

WB: p21 22-

1

48-

WB: USP7

WB: MDM2 98-

WB: tubulin

5

0-

1

7

Figure 1 | Identification and characterization of USP7 inhibitors.

a, Schematic of USP7regulated signalling. b, USP7 inhibitor hitto

and Bad (Extended Data Fig. 7a) . Given the role of PIM2 in hae

17

matological malignancies , we confirmed combination efficacy with

lead selection cascade. See text for more details. HTS, highthroughput

screening; IF, immunofluorescence; MSD, mesoscale discovery; DLS,

dynamic light scattering. c, Immunoblot analysis of MDM2 turnover on

USP7 inhibition, with actin loading controls. See also Extended Data

Fig. 3b. Actinnormalized MDM2 levels are plotted from three biological

replicates. DMSO, dimethylsulfoxide; IP, immunoprecipitation; WB,

western blot. d, Analysis of endogenous MDM2 polyubiquitinated

with K48linked chains. Denatured lysates from cells, treated with the

indicated compounds, were immunoprecipitated with K48polyubiquitin

linkagespecific antibody, and immunocomplexes or cell lysates were

immunoblotted as indicated. e, Immunoblot analysis of p53pathway

proteins in isogenic cell lines treated with USP7 inhibitors or inactive

controls. d, e, Representative of three experimental replicates. Gel source

data: Supplementary Fig. 1.

Bliss analysis using PIM inhibitors GDC0339 (ref. 18) or GDC0570

(

Fig. 2c and Extended Data Fig. 7b–d). Both inhibitors share struc

tural similarities and are selectively potent against PIM kinases (X.W.

et al., manuscripts in preparation). In mechanistic studies, GNE6776

promoted a modest elevation of p21 levels in AML cell lines whereas

PIM2 levels were reduced. The GNE6776induced decrease in PIM2

levels was rescued by inhibition of the ubiquitinactivating enzyme

1

3

UAE1 or the proteasome inhibitor bortezomib . PIM2 levels were

lower in USP7null cells, which was rescued by proteasome inhibition,

collectively indicating that USP7 stabilizes PIM2 levels via a ubiquitin

and proteasomedependent mechanism (Extended Data Fig. 7e, f). A

treatment time course with GNE6776, GDC0570 or both compounds

17

revealed that, as previously reported , GDC0570 reduced phospho

rylation of Bad and S6 and decreased expression of shortlived proteins

14

by numerous cellular mechanisms , we developed an activitybased p21 and Mcl1. GNE6776 decreased PIM2 levels and, remarkably, also

probe assay to evaluate effects on endogenous deubiquitinases reduced S6 and Bad phosphorylation and total Mcl1 levels, similar

(

J.H. et al., manuscript in preparation). GNE6640 or GNE6776 to PIM inhibition. In combination, GNE6776 and GDC0570 treat

significantly inhibited USP7 while remaining selective against 44–47 ments enhanced cleavage of poly(ADPribose) polymerase (PARP) and

other detected deubiquitinases (Fig. 2b and Extended Data Fig. 4c, d). caspase3, mechanistically confirming the cell viability studies (Fig. 2d).

Given the favourable features of these inhibitors, we investigated their We therefore hypothesized that PIM2 is a USP7 substrate. In support

efficacy in animal models. Pharmacodynamic and pharmacokinetic of this idea, endogenous USP7 and PIM2 coimmunoprecipitated

studies indicated that GNE6776 is orally bioavailable and promotes and GNE6776 treatment enhanced PIM2 ubiquitination (Fig. 2e

ontarget pathway modulation in human xenografts (Extended Data and Extended Data Fig. 7g). Recombinant USP7, but not the C223S

2

| n A t U R E | V o L 0 0 0 | 0 0 M o n t h 2 0 1 7

©

Letter reSeArCH

a

b

c

DUB class

USP

UCH

OTU

Josephin

MPN+

MINDY

SENP

Inhibition (%)

Bliss excess

20

GNE-

^D^U^B⁽^c^oⁿ^c^eⁿ^t^r^a^tⁱ^oⁿ^;^U^b2 linkage) 6640

GNE-

6641

GNE-

6776

3

2

1

0

20 97 99100100100100100100100100

6.67 55 75 93 98 98 99100100100100

2.22 ₁₃₁₇₁₈₆₇₉₂₉₇₉₉₉₉₉₉₉₉

3

5

6

5

2

0

3

5

3

1

0

1

0

USP7

–

1

USP1 (20 ng μl ; K63)

6.67 -2 16 34 38 31 14

2.22

–

1

USP2 (20 ng μl ; K63)

2

4

2

3

2

-1

-3

2

1

0

49 60 27

-1 -1

–

1

USP4 (20 ng μl ; K48)

USP7

0

0.25

0.082

.74

4

2

5

2

7

5

15 61 91 99 99 99 99

0.74

0.25

7

2

1

37 24

12 12

–

1

USP5 (2 ng μl ; K63)

–

1

9

8

2

6

3

8

35 78 97 99 100100

0

USP6 (2 ng μl ; K63)

1

–

1

25 69 95 99 100100 0.082

18 66 94 99 100100 0.027 -3 -8 -6 -5 -5

22 65 94 99 100100 0.009

21 67 94 99 100100

22 66 94 99 100100

2

3

0

USP7 (2 ng μl ; K11)

–

1

0.027 -4 -4 -1

0.009

0.003 _-₄_-₄_-₁

USP8 (20 ng μl ; K63)

44

47

–

1

USP9x (20 ng μl ; K11)

Control

0

3

1

-4 -2 -1 -1 -1

–

1

USP10 (20 ng μl ; K11)

−2 −1

0

1

2

−1

0

1

0.003 _-₃_-₈_-₅_-₄_-₂

1

–

1

USP15 (20 ng μl ; K29)

log (fold change in activity)

2

0

-2

2

5

0

-3

0

-1 -1

0

–

1

USP16 (2 ng μl ; K48)

3

GNE-6776

+ GDC-0570

–

1

USP20 (2 ng μl l; K63)

d

Compounds:

Time (h):

GNE-6776

GDC-0570

18 24 40

3

.

–

1

USP21 (2 ng μl ; K11)

.05

–

1

0

4

8

18 24 40

4

8

4 8 18 24 40

USP25 (2 ng μl ; K11)

[GDC-0570] (μM)

–

1

USP27x (100 ng μl ; K11)

WB: p21 22-

–

1

USP28 (2 ng μl ; K11)

–

1

e

GNE-6776

12

GNE-6776

12

USP36 (100 ng μl ; K11)

–

1

WB: p53

CYLD (20 ng μl ; K63)

50-

–

1

Time (h):

0

3

6

Time (h):

50-

0

3

6

Otulin (0.2 ng μl ; M1)

–

1

WB: MDM2

250-

OTUB1 (2 ng μl ; K48)

–

1

64-

48-

WB: PIM2

OTUB2 (2 ng μl ; K63)

1

148-

–

1

OTUD1 (20 ng μl ; K11)

WB: USP7

WB: PIM2

WB: P-S6

IP: K48-Ub

98-

–

1

Ub-

PIM2

OTUD3 (20 ng μl ; K11)

WB: MDM2

WB: USP7

–

1

5

0-

WB: PIM2 98-

OTUD5 (200 ng μl ; K63)

–

1

148-

A20 (20 ng μl ; K48)

64-

50-

36-

–

1

36-

Cezanne (0.2 ng μl ; K11)

_*IgG

–

1

TRABID (100 ng μl ; K33)

–

1

WB: tubulin 50-

vOTU (0.2 ng μl ; K11)

–

1

WB: total S6

WB: P-Bad

VCPIP (100 ng μl ; K48)

–

1

AMSH (20 ng μl ; K63)

–

1

250-

AMSH-LP (20 ng μl ; K63)

2

2-

–

1

IP: K48-Ub

148-

Ub-

BRCC3 (20 ng μl ; K63)

–

1

MDM2

USP47 (2 ng μl ; K48)

N/D

WB: total Bad

WB: Mcl-1

WB: MDM2

98-

–

1

22-

50-

UCHL1 (200 ng μl ; Ub-E2)

–

1

UCHL3 (20 ng μl ; Ub-E2)

–

1

UCHL5 (2 ng μl ; Ub-E2)

64-

–

1

BAP1 (20 ng μl ; Ub-E2)

WB: PARP

**

0

Inhibition (%)

100

98-

*

WB: cl. casp-3

WB: tubulin

1

5

6-

0-

Figure 2 | Selectivity of USP7 inhibitors and synergy with PIM kinase

inhibition. a, Activity heatmap of recombinant deubiquitinases (DUB)

incubated with USP7 inhibitors, representative of experimental duplicates.

Diubiquitin cleavage activity was detected by matrixassisted laser

desorption/ionization–time of flight (MALDI–TOF) mass spectrometry.

b, Volcano plots of endogenous deubiquitinase activity changes in

HEK293T cells. Data from experimental triplicates (left) or quadruplicates

inhibitor GDC0570 and GNE6776 in EOL1 cells. Left: inhibition scores:

0, no effect; 100, no viable cells. Right: Bliss analysis of observed versus

predicted viability. Positive values: greater than predicted viability

decrease. Synergy score =4.11. d, Immunoblot analysis of EOL1

wholecell lysates treated as indicated. Asterisks * and ** indicate

cleaved and uncleaved PARP, respectively. e, Analysis of endogenous

PIM2 polyubiquitinated with K48linked chains, as in Fig. 1d.

d, e, Representative of three experimental replicates. Gel source data:

Supplementary Fig. 1.

(

right) are plotted versus Benjamini–Hochberg false discovery rate

adjusted P values. Numbers indicate detected deubiquitinases relative to

the 20% false discovery rate line. c, Combination treatments with PIM

catalytic mutant, deubiquitinated PIM2 in vitro (Extended Data crosstalk between USP7 and PIM signalling and expand the repertoire

Fig. 7h). Simultaneous PIM kinase inhibition and USP7 inhibitor of USP7regulated oncogenic pathways.

mediated protein degradation thus antagonize PIM activity and cell

Given the selectivity and efficacy of GNE6640 and GNE

viability more than does either treatment individually. These data reveal 6776, understanding the molecular mechanisms by which these

a₁₂₂

b

GNE-6776

1

24

26

28

30

32

Q287

E371

USP7 apo

USP7 + ubiquitin

+

GNE-6776

USP7 + GNE-6776

9

.6 9.4 9.2 9.0

H G (p.p.m.)

9.6 9.4 9.2 9.0

H G (p.p.m.)

9.6 9.4 9.2 9.0

H G (p.p.m.)

1

Catalytic triad

c

α5 helix

α6 helix

E308

Pyridine-

Phenol-

E308

D305

carboxamide aminopyridine

β2 sheet

N

^N^H2

K48

D305

R301

H

N

β3

sheet

OH

N

R301

O

GNE-6776

Figure 3 | USP7 inhibitors compete with ubiquitin binding to USP7.

GNE6776: yellow sticks; catalytic triad: green sticks, with CPK colouring.

c, Left: USP7 catalytic domain (green) complexed with ubiquitin (cyan)

(PDB accession number 1NBF). UbiquitinK48 and USP7D305/E308

side chains are represented as sticks. Dashed lines indicate binding

interactions. Right: structure of USP7 and GNE6776 (orange cartoon

and yellow sticks, respectively) with ubiquitin (cyan) modelled from PDB

accession number 1NBF, showing GNE6776 steric blockade of USP7/

ubiquitinbinding.

1

15

a, Twodimensional H/ N transverse relaxation optimized spectroscopy

TROSY) spectra overlays of labelled USP7 catalytic domain (orange)

highlighting changes induced by ubiquitinbinding in the absence

(

left, blue) and in the presence of GNE6776 (middle, dark grey). Right:

GNE6776induced changes without ubiquitin (light grey). Representative

of three experimental replicates. b, Crystal structure of USP7 catalytic

domain in complex with GNE6776. USP7: orange cartoon;

0

0 M o n t h 2 0 1 7 | V o L 0 0 0 | n A t U R E | 3

©

reSeArCH Letter

a

K63 T66

Figure 4 | USP7 preferentially binds and

cleaves ubiquitin moieties with free K48 side

chains. a, Top row: schematics of distal (Dist.)

K48

T14* Dist.

T12*

T14* Dist.

T12*

K63

15

NThr or proximal (Prox.) NLeulabelled

C-terminus

K48

L15*

diubiquitins (see also Extended Data

Fig. 10b, c). Middle/bottom rows: two

dimensional bandselective optimized

flipangle short transient (SOFAST) spectra

overlays and onedimensional traces of labelled

Prox.

One-

dimensional

traces

Prox.

One-

dimensional

traces

L8*

L15* K63

L8* C-terminus

T12

T14

T12

T14

T12

1

21

22

23

24

25

26

+

–

USP7

+ USP7

L8

T14

L8

T14

15

diubiquitin. Left/middle: NThr or NLeu

peaks from K48linked diubiquitin without

–

USP7

8

.8 8.68.8 8.6

8.8 8.6 8.8 8.6

L15 L8

(

green) and with (orange) USP7. Right/middle:

L15

L8

15

L43

L15

L43

L15

NThr or NLeu peaks from K63linked

+

–

USP7

+ USP7

diubiquitin without (purple) and with (orange)

15

USP7. Bottom row: NThr or NLeu peaks

from K48linked diubiquitin without (green)

and with (orange) indicated USP7 mutants.

b, Time course of TAMRA/peptideconjugated

tetraubiquitin cleavage by fulllength

USP7 (top); corresponding densitometry

plots (bottom). a, b, Representative of four

experimental replicates. Gel source data:

Supplementary Fig. 1.

L50

.2 9.1 9.0 8.9 8.8 8.7 8.6 8.5 8.9 8.79.3 9.1

T12 T14

– USP7

9

9.2 9.1 9.0 8.9 8.8 8.7 8.6 8.5 8.9 8.7 9.3 9.1

T12 T14

T12

+ D305A

E308A

USP7

1

21

22

23

24

25

26

+ D305A

USP7

L8

T14

–

USP7

–

USP7

8

.8 8.6 8.8 8.6

L15 L8

8.8 8.6 8.8 8.6

L15 L8

+

D305A

L43

+ D305A

USP7

L43

E308A

USP7

L15

L50

– USP7

–

USP7

9

.2 9.1 9.0 8.9 8.8 8.7 8.6 8.5 8.9 8.7 9.3 9.1

9.2 9.1 9.0 8.9 8.8 8.7 8.6 8.5 8.9 8.7 9.3 9.1

1

H chemical shift (p.p.m.)

^T^A^M^R^A^/^p^e^p^tⁱ^d^e^K⁴⁸^U^b4

^T^A^M^R^A^/^p^e^p^tⁱ^d^e^K⁶³^U^b4

b

Reaction

time (min):

0

5

10 15 20 25 30 35

0

5 10 15 20 25 30 35

5

3

0-

6-

Peptide-K48 Ub₄

*

^P^e^p^tⁱ^d^e^-^K⁶³^U^b4

Peptide-K48 Ub₃

2

2-

Peptide-K48 Ub₂

1

6-

6

4

-

Free peptide

*

1

4

2

0

8

6

4

2

0

14

12

10

8

6

4

2

0

Free peptide

Peptide-Ub₄

*

^P^e^p^tⁱ^d^e^-^U^b3

Peptide-Ub₂

0

5

10 15 20 25 30 35

0

5

10 15 20 25 30 35

Reaction time (min)

tool compounds inhibit USP7 was of interest. Enzymatic studies 1NBF) reveals that the compounds sterically inhibit ubiquitin binding

verified inhibition of intrinsic USP7 activity (turnover rate, kcat) and prevent transition of the USP7 catalytic domain α5 helix to the

5

achieved by attenuating substrate binding (Michaelis constant, active conformation (Fig. 3c and Extended Data Fig. 8d). However,

Km) (Extended Data Fig. 8a). NMR analysis confirmed that given the extensive interactions between USP7 catalytic domain

5

USP7 inhibitors weaken USP7 catalytic domain interactions with and ubiquitin , we investigated the importance of the USP7 residues

native ubiquitin (Fig. 3a and Extended Data Fig. 8b, c). Analysis that interact with the ubiquitinK48 side chain. The ubiquitin K48A

of cocrystal structures revealed that both compounds bind into mutant has reduced affinity for USP7 catalytic domain (Extended Data

a pocket approximately 12Å from the catalytic triad that is located Fig. 9a); similarly, recombinant D305A/E308A USP7 catalytic domain

at the interface of the USP7 catalytic domain palm, fingers, and failed to cleave tetraK48 or tetraK63linked polyubiquitin conju

1

9

thumb subdomains (Fig. 3b and Extended Data Fig. 8d, e). gated to a TAMRAlabelled peptide . This apparent inactivity was

The phenol rings of the phenolaminopyridine moieties are buried in due to a significant increase in Km (ref. 20) (Extended Data Fig. 9b, c).

the pocket and the hydroxyl groups are engaged in hydrogenbond Expression of a D305A/E308A USP7fulllength mutant increased

interactions with USP7H403. The 2aminopyridine is directed out K48linked polyubiquitination on endogenous MDM2, comparable

of the pocket towards solvent with the 2amino group engaging to expression of catalytically inactive USP7 (Extended Data Fig. 9d).

USP7D349 (Extended Data Fig. 8d). The substituents at the 5position Thus ubiquitinK48 and USP7D305/E308 residues are critical for

of the aminopyridine—the pyridinecarboxamide (GNE6776) and USP7–ubiquitin binding and substrate deubiquitination.

indazole (GNE6640)—are largely solventexposed and lie between

The GNE6640 and GNE6776 binding modes suggested that

the α5 and α6 helices. The NH atoms of the GNE6776 carboxamide USP7 may preferentially associate with ubiquitin moieties that have

moiety engage USP7D305 on the α5 helix, whereas the indazole of free K48 side chains. Mechanistically, this could direct USP7 to dis

GNE6640 makes extensive van der Waals contacts in the crevice tal subunits of K48linked polyubiquitin, resulting in sequential dis

between α5 and α6 (Fig. 3c and Extended Data Fig. 8d). Comparison taltoproximal chain depolymerization (exocleavage). However, in

of the cocrystal structures of GNE6776 and GNE6640 with the nonK48linked chains, USP7 may bind and cleave proximally to any

USP7–ubiquitin complex (Protein Data Bank (PDB) accession number ubiquitin moieties having accessible K48 side chains (exo, endo, or

4

| n A t U R E | V o L 0 0 0 | 0 0 M o n t h 2 0 1 7

©

Letter reSeArCH

basecleavage) (Extended Data Fig. 10a). To evaluate this idea, we engi 5. Hu, M. et al. Crystal structure of a UBP-family deubiquitinating enzyme in

isolation and in complex with ubiquitin aldehyde. Cell 111, 1041–1054 (2002).

Cummins, J. M. et al. Tumour suppression: disruption of HAUSP gene stabilizes

p53. Nature 428, 486–487 (2004).

neered K48 and K63linked diubiquitin with differentially labelled

6

.

distal and proximalubiquitin subunits using spectroscopically dis

tinct amino acids. To select ubiquitin residues for isotopic labelling, 7. Kon, N. et al. Inactivation of HAUSP in vivo modulates p53 function. Oncogene

29, 1270–1279 (2010).

we evaluated chemical shift perturbations induced by USP7 binding

Extended Data Fig. 10b). The cluster of residues L8, T12, T14, and L15

broaden upon USP7 catalytic domain binding and are well resolved

8.

Pozhidaeva, A. K. et al. Structural characterization of interaction between

human ubiquitin-speciﬁc protease 7 and immediate-early protein ICP0 of

herpes simplex virus-1. J. Biol. Chem. 290, 22907–22918 (2015).

Cheng, J. et al. Molecular mechanism for USP7-mediated DNMT1 stabilization

by acetylation. Nat. Commun. 6, 7023 (2015).

(

1

5

9

1

.

in the spectra; thus Leu residues were Nlabelled in the proximal

15

diubiquitin and Thr residues were Nlabelled in the distal subunit

0. Wang, L. et al. Ubiquitin-speciﬁc protease-7 inhibition impairs Tip60-

dependent Foxp3 T-regulatory cell function and promotes antitumor

(

Fig. 4a and Extended Data Fig. 10b, c). NMR analysis showed preferen

+

tial USP7 catalytic domain binding to the distal subunit of K48linked

diubiquitin, whereas USP7 catalytic domain binding was similar to

both proximal and distal K63linked diubiquitin subunits (Fig. 4a and

immunity. EBioMedicine 13, 99–112 (2016).

1

1. Hao, Y. H. et al. USP7 acts as a molecular rheostat to promote WASH-

dependent endosomal protein recycling and is mutated in a human

neurodevelopmental disorder. Mol. Cell 59, 956–969 (2015).

Extended Data Fig. 10c). Importantly, both D305A USP7 and D305A/ 12. Clague, M. J. et al. Deubiquitylases from genes to organism. Physiol. Rev. 93,

1289–1315 (2013).

E308A USP7 catalytic domain mutants demonstrated similar binding

to both proximal and distal K48linked diubiquitin subunits (Fig. 4a).

1

3. Deshaies, R. J. Proteotoxic crisis, the ubiquitin-proteasome system, and cancer

therapy. BMC Biol. 12, 94 (2014).

On the basis of these data, we evaluated whether differential di 14. Heideker, J. & Wertz, I. E. DUBs, the regulation of cell identity and disease.

Biochem. J. 467, 191 (2015).

ubiquitin binding affects the dynamics of ubiquitin chain depolyme

rization by USP7. A previous study reported that USP7 cleaves K6,

1

5. Fåhraeus, R. & Olivares-Illana, V. MDM2’s social network. Oncogene 33,

4365–4376 (2014).

K11, K33, K48, and K63linked diubiquitins with similar kcat and 16. Williams, A. B. & Schumacher, B. p53 in the DNA-damage-repair process. Cold

Spring Harb. Perspect. Med. 6, a026070 (2016).

m

K values, indicating that different isopeptide bond conformations do

20

17. Nawijn, M. C., Alendar, A. & Berns, A. For better or for worse: the role of Pim

not substantially impact cleavage efficiency . Remarkably, our studies

oncogenes in tumorigenesis. Nat. Rev. Cancer 11, 23–34 (2011).

reveal that USP7 sequentially depolymerizes tetraK48linked ubiqui 18. Hodges, A. J. M., Matteucci, M., Sharpe, A., Sun, M., Wang, X. & Tsui, V. H.

Pyrazol-4-yl-heterocyclyl-carboxamide compounds as Pim kinase inhibitors

tin from the distal residue (exocleavage) whereas tetraK63ubiquitin

chains are cleaved via a combination of exo, endo, or basecleavage,

as nearly complete depolymerization occurs within 4min (Fig. 4b and

Extended Data Fig. 10d, e). Thus USP7 preferentially binds ubiquitin

moieties that have free K48 side chains, resulting in protracted and

sequential depolymerization of K48linked polyubiquitin.

and their preparation. US patent 8, 614,206 B2 (2013).

1

9. Geurink, P. P., El Oualid, F., Jonker, A., Hameed, D. S. & Ovaa, H. A general

chemical ligation approach towards isopeptide-linked ubiquitin and

ubiquitin-like assay reagents. ChemBioChem 13, 293–297 (2012).

0. Faesen, A. C. et al. The diﬀerential modulation of USP activity by internal

regulatory domains, interactors and eight ubiquitin chain types. Chem. Biol. 18,

2

1550–1561 (2011).

Herein we describe GNE6640 and GNE6776, selective USP7 21. Huether, R. et al. The landscape of somatic mutations in epigenetic regulators

across 1,000 paediatric cancer genomes. Nat. Commun. 5, 3630 (2014).

inhibitors that possess a structurally defined mechanism of inhibi

tion. Establishing stringent screening cascades was critical for selecting

2

2. Schaefer, J. B. & Morgan, D. O. Protein-linked ubiquitin chain structure restricts

activity of deubiquitinating enzymes. J. Biol. Chem. 286, 45186–45196 (2011).

and optimizing ontarget inhibitors. Combination studies revealed 23. Husnjak, K. & Dikic, I. Ubiquitin-binding proteins: decoders of ubiquitin-

mediated cellular functions. Annu. Rev. Biochem. 81, 291–322 (2012).

a previously undescribed intersection between USP7 deubiquitinase

activity and PIM kinases in regulating cell viability. Cocrystal struc

Supplementary Information is available in the online version of the paper.

tures of GNE6640 or GNE6776 pointed to the importance of the

Acknowledgements We thank W. Fairbrother, L. Frick, S. Fong, E. Helgason,

complementary charged interactions between USP7D305/E308 and

T. Hunsaker, C. Lam, B. Liederer, M. Merchant, J. Nonomiya, J. Peng, T. Pham,

ubiquitinK48 side chains, which we confirmed using mutational L. Rangell, R. Rodriguez, U. Segal, R. Tong, L. Wang, R. Gennis, T. Iwasaki, the

analysis. Notably, D305G has been identified as a somatic lossof Genentech Protein Expression, Cell Central, gCSI, and Sequencing groups, and

2

1

the Boston Biochem team for reagents and collaborations. Use of the Stanford

Synchrotron Radiation Lightsource, SLAC National Accelerator Laboratory, was

supported by the US Department of Energy, Office of Science, Office of Basic

function mutant in patients with acute lymphoblastic leukaemia .

NMR analysis of USP7 binding to native monoubiquitin and differ

entially labelled diubiquitins revealed that USP7 preferentially inter Energy Sciences under contract number DE-AC02-76SF00515. The SSRl

Structural Molecular Biology Program was supported by the DOE Office of

acts with ubiquitin moieties having free K48 side chains. It has been

Biological and Environmental Research, and by the National Institutes of Health,

proposed that the inefficiency of some deubiquitinases to depolymerize

National Institute of General Medical Sciences (including P41GM103393).

longer substrateconjugated K48linked chains enables a threshold for The contents of this publication are solely the responsibility of the authors and

22

do not necessarily represent the official views of National Institute of General

Medical Sciences or the National Institutes of Health.

proteasometargeting polyubiquitination ; our studies substantiate this

idea and provide a biophysical mechanism. Numerous proteins, includ

ing other deubiquitinases, ubiquitin ligases, DNA repair and endocyto Author Contributions L.K., E.L., J.-P.U., S.P., F.P., S.E.M., and I.E.W. performed cell

viability and signalling studies, protein turnover and ubiquitination experiments,

sis machinery, and epigenetic regulators are functionally dependent on

and independently replicated cellular data. P.D.L. and T.M. performed NMR

23

ubiquitin binding . Developing selective inhibitors that attenuate ubiq

studies; L.R. and J.M. performed X-ray crystallography and cross-validated data.

uitin binding is an effective strategy for USP7 inhibition. Our studies J.D., T.K., T.W.B., M.C.M.K., Z.S., B.B., Y.T., M.R.A., J.A.E., C.S., and R.A.B. performed

in vitro and cellular biochemical studies and independently replicated data. J.H.,

demonstrate the feasibility of this approach, which may have broader

applications for inhibiting other classes of ubiquitinbinding proteins.

M.S.R., T.P.M., D.R.A., M.T., and K.Y. performed deubiquitinase selectivity profiling

studies, and M.M. and E.B. performed in vivo studies and cross-validated

data. K.R.C. and M.H.B. developed assays, and performed and analysed

Online Content Methods, along with any additional Extended Data display items and

Source Data, are available in the online version of the paper; references unique to

these sections appear only in the online paper.

high-throughput screening studies. R.P., P.J., B.R.H., A.R.R., F.C., C.N., X.W.,

and V.T. performed compound design/synthesis. F.G., M.T.C., C.K., and W.F.F.

performed bioinformatics/statistical analysis and replicated data. T.M. and

I.E.W. coordinated studies. J.M., T.M., and I.E.W. wrote the manuscript. D.R.A. is

supported by the UK Medical Research Council (grant MC_UU_12016/2).

ꢀeceived 8 February; accepted 16 August 2017.

Published online 18 October 2017.

Author Information Reprints and permissions information is available at

www.nature.com/reprints . The authors declare competing financial interests:

details are available in the online version of the paper. Readers are welcome to

comment on the online version of the paper. Publisher’s note: Springer Nature

remains neutral with regard to jurisdictional claims in published maps and

institutional affiliations. Correspondence and r equests for materials should be

addressed to T.M. (maurer.till@gene.com) or I.E.W. (ingrid@gene.com).

1.

2.

3.

4.

Komander, D. & Rape, M. The ubiquitin code. Annu. Rev. Biochem. 81, 203–229

2012).

Clague, M. J., Heride, C. & Urbé, S. The demographics of the ubiquitin system.

Trends Cell Biol. 25, 417–426 (2015).

Nicholson, B. & Suresh Kumar, K. G. The multifaceted roles of USP7: new

therapeutic opportunities. Cell Biochem. Biophys. 60, 61–68 (2011).

Ritorto, M. S. et al. Screening of DUB activity and speciﬁcity by MALDI-TOF

mass spectrometry. Nat. Commun. 5, 4763 (2014).

(

ꢀeviewer Information Nature thanks M. Rolfe, S. Scherer and the other

anonymous reviewer(s) for their contribution to the peer review of this work.

0

0 M o n t h 2 0 1 7 | V o L 0 0 0 | n A t U R E | 5

©

reSeArCH Letter

MꢁꢂꢃOꢄS

by monitoring the conversion to ubiquitin by mass spectrometry. Compounds were

High-throughput screen and counterscreen assays. USP7 UbA10 time-resolved dispensed into 384well polypropylene plates (Greiner BioOne), and 10μl USP7

fluorescence energy transfer (TR-FRET) activity assay. Potential inhibitors of were added and allowed to incubate for 10min. The reaction was started by the

USP7 were identified in a TRFRETbased enzyme activity assay with UbA10 as addition of 10μl of K48linked diubiquitin substrate (UC200, Boston Biochem)

substrate. UbA10 is a fragment of the naturally occurring ubiquitin precursor and allowed to progress for 70min; then it was stopped by the addition of 20μl 2%

Ub52; it retains the tenaminoacid segment of Ub52 that extends beyond the formic acid. The assay plates were stored frozen at −80°C until analysis by mass

ubiquitin C terminus. The primary screening assay is a novel TRFRETbased spectrometry at Agilent Technologies. Before quantification, the enzyme reaction

activity assay that measures cleavage by fulllength USP7 of a doubly tagged peptide was passed over a RapidFire cartridge to remove buffer components. Both di

substrate. The peptide is tagged with glutathione Stransferase (GST) on the N ubiquitin substrate consumption and ubiquitin product formation were monitored

terminus and with eight histidine residues on the C terminus (GST–UbA10His). by mass spectrometry using multiple reaction monitoring (MRM) in positive ion

The tags are detected by antiGSHd2 (TRFRET acceptor) and antiHis mode with parent ion/daughter ion transitions of 1142.1/260.1 and 770.8/817.7,

europium (TRFRET donor), respectively. Cleavage of this substrate by USP7 respectively. The percentage conversion of substrate was plotted as a function of

at the ubiquitin C terminus results in separation of these two tags and loss of compound concentration to generate the IC50 values as indicated above.

TRFRET signal. Compounds were dispensed into 1,536well black plates (MaKO, NMR screen and binding studies. All NMR spectra were recorded on Bruker

Aurora Microplates) followed by 2μl fulllength recombinant USP7 in assay buffer Avance 600 MHz and 800 MHz spectrometers operating at 14.1 and 18.8 T

(

50mM HEPES pH 7.5, 0.1% Prionex (Pentapharm), 0.01% Triton X100, and using tripleresonance cryogenic probes optimized for proton detection. All

0mM dithiothreitol (DTT)). After a 10min incubation, the reaction was started twodimensional spectra were acquired with a spectral width of 16p.p.m. and

by the addition of 2μl GST–UbA10His substrate in assay buffer. After 75min 2,048 (TROSY) or 954 (SOFAST) data points in the direct proton dimension

1

15

of reaction, 2μl of a detection antibody reagent, containing antiHiseuropium and 28p.p.m. and 192 sample points in the N dimension with echo/antiecho

Life Technologies) and antiGSTd2 (CISbio) in assay buffer were added. After (TROSY) or StatesTPPI (SOFAST) type selection. The resulting free induction

0min, the fluorescence at 618nm and 671nm with excitation at 340nm was decay resolution was 12.52 and 19.05Hz per data point for the TROSY and 20.08

(

6

read on a ViewLux reader (PerkinElmer). Cleavage of the doubly tagged substrate and 17.10Hz per data point for the SOFAST spectra, respectively. All spectra

resulted in loss of TRFRET signal, whereas inhibition of USP7 by compound were recorded at 300K. The pH was adjusted to 7.2 without correction owing

restored the signal. The TRFRET ratio was calculated as fluorescence at 671nm to isotope shifts. For data processing, NMRPipe/NMRDraw and the BRUKER

divided by fluorescence at 618nm. TRFRET ratios were normalized to controls software package TOPSPIN 3.2 were used. All data evaluation was done in NMR

to determine the percentage inhibition for the single concentration screen. For view and CCPN. Visualization and presentation of the threedimensional tertiary

confirmation in concentration–response mode (ten points with n= 2 ), percentage USP7 as well as related protein structures from the RSCB Protein Database were

inhibition was plotted against compound concentration, and the data were fitted done in PyMOL (PyMOL Molecular Graphics System, version 1.7.4, Schrödinger).

to a fourparameter curve with Screener Assay Analyzer (Genedata) to determine All samples contained 137mM NaCl, 10mM Na2HPO4, 27mM KCl, and 1.8mM

2

IC50 values. The assay buffer was optimized to maintain enzyme stability and KH

2

PO

4

adjusted to a pH of 7.2 and contained 7% (w/w) H

2

O and 0.5μM NaN

3

.

to maximize assay signal to background: Triton X100 was included to prevent Ubiquitin titration experiments were done by addition of purified bovine ubiquitin

nonspecific adsorption of the enzyme and/or substrate to the assay plate and/or (from erythrocytes, Sigma) from a stock solution of 20mM in the same buffer. All

15

to compound aggregates, Prionex carrier protein was included to help stabilize the proton chemical shifts were referenced to internal DSS (50μM) and N referenced

1

enzyme and to prevent nonspecific adsorption to container and tubing surfaces indirectly using the H chemical shift of the methyl group in DSS by multiplication

as well as to minimize nonspecific inhibition by library compounds, and DTT with a factor of 0.101329118.

served to maintain good USP7 activity and minimize the impact of inhibitors that NMR screening conditions. The primary fragment screen was performed on 4,871

act through redox cycling. Reagent concentrations were optimized for good assay fragments in mixtures of 5. The individual compound concentration was 500μM

performance at approximately 50% substrate conversion in the signal decrease and the concentration of unlabelled USP7 was 7μM. Binders were identified by

assay with a key aim of minimizing the required concentration of USP7. The GST– the presence of signals stemming from the individual ligand in the proton satura

UbA10His concentration was adjusted to maximize assay signal, antiGSTd2 tion transfer difference spectra recorded at 284K. The criterion used to identify a

concentration was adjusted approximately in parallel with that of the substrate, and binder was a signaltonoise ratio of above 5. All primary binders were remeasured

antiHiseuropium was used at a concentration that represented a minimum that as a single compound to confirm binding. Confirmed binders were defined as

was compatible with robust detection on the ViewLux plate reader. Timecourse compounds with a saturation transfer difference signaltonoise ratio of above

evaluations were conducted to confirm that the enzyme concentration (10nM) and 10. All confirmed binders were tested again at 2–2.5mM, by proteinobserved

1

15

reaction time (75min) were in the linear range of enzyme activity. Additionally,

H/ N TROSY experiments in the presence of Nlabelled USP7 at 220–340μM.

extended time courses for the detection reaction were used to demonstrate that NMR binding studies. The concentrationdependent NMR shift perturbations

the 60min incubation was sufficient to reach equilibrium. Under the final assay caused by the interaction of unlabelled ubiquitin with ¹⁵Nlabelled USP7 in the

conditions, reagent stability studies indicated greater than 20h of acceptable absence and presence of USP7 inhibitors were fitted to the function for a twostate

performance. Over the course of the screen, Z′ values averaged 0.76.

fast exchanging equilibrium based on concentrationdependent chemical shift

USP7 di-ubiquitin FRET activity assay. Potential USP7 inhibitors were confirmed in perturbations:

an orthogonal activity assay with an internally quenched K63linked diubiquitin

substrate (U310, Boston Biochem). Conditions were similar to those used for

Kd + [P] + [L] − (Kd + [P] + [L])²− 4[P][L]

Δδ = Δδmax

the UbA10 TRFRET activity assay. Compounds dispensed into 1,536well plates

were preincubated with fulllength recombinant USP7 in assay buffer for 10min,

2[P]

and the reaction was started by the addition of 2μl of the diubiquitin substrate. where Δδ is the chemical shift change at various protein/ligand ratios, Δδmax is the

d

During the 60min incubation, cleavage of the substrate by USP7 resulted in the chemical shift change at saturation, K is the dissociation constant, and [L] and [P]

release of quenching of the TAMRA tag by the QXl tag and thus an increase in are the ligand and protein concentrations, respectively. The chemical shift change

1

15

fluorescence. The fluorescence intensity was read with excitation at 540nm and is the root mean square of the H and N values scaled by 1 and 0.15, respectively.

emission at 585nm. Concentration–response assay methods and data analyses Seven ubiquitin concentrations were measured (0, 100, 250, 500, 750, 1,000,

were conducted as for the UbA10 TRFRET assay.

2,500μM) for n=8 independent chemical shift perturbations analysed, with USP7

USP7 Ubiquitin/Rho110 activity assay. Potential USP7 inhibitors were also at 250μM and each inhibitor at 1mM. The mean values and s.d. were calculated by

confirmed in an orthogonal activity assay with ubiquitin/rhodamine110 as averaging the values obtained for the eight wellresolved cross peaks: Y339, S341,

substrate. Compounds were preincubated for 10min with 2μl fulllength recom D342, G375, A381, G382, D412, and I419. Data were visualized using GraphPad

binant USP7 in assay buffer, and the reaction was started by the addition of 2μl Prism (GraphPad Software). All experiments involving selectively labelled di

of ubiquitin/rhodamine110 (U555, Boston Biochem). After a 60min reaction ubiquitin and unlabelled USP7 were performed with the following concentra

in which USP7 cleaved the rhodamine110 from the ubiquitin and thus increases tions: K48linked diubiquitin 70μM:USP7C223A 70μM, K63linked diubiquitin

fluorescence, the fluorescence intensity was read with excitation at 485nm and 65μM: USP7C223A 98μM, K48linked diubiquitin 60μM:USP7C223A–D305A

emission at 535nm. General assay conditions and data analyses were as described and USP7C223A–D305A–E308A 120μM.

above.

NMR assignments. The USP7 NMR assignments have been deposited in the

USP7 di-ubiquitin cleavage assay with mass spectrometric detection. To Biological Magnetic Resonance Data Bank under accession number 26766.

further validate compounds that met the initial highthroughput screen confirma Ubiquitin assignments were derived from Biological Magnetic Resonance Data

tion criteria, the ability of compounds to inhibit diubiquitin cleavage was assessed Bank accession number 17439.

©

Letter reSeArCH

Compound synthesis. Synthesis of GNE-6640 is representative of the syntheses for 7.10–7.04 (m, 2H), 6.92–6.86 (m, 2H), 4.94 (s, 2H), 2.26 (q, J=7.4Hz, 2H), 0.60

13

6

GNE-2118, GNE-6776, GNE-6641, GNE-2931, GNE-2917, and GNE-2916 and (t, J=7.4Hz, 3H). C NMR (150MHz, DMSOd ) δ 15.0, 24.0, 111.3, 117.4, 122.2,

GNE-9603.

122.3, 124.0, 127.8, 129.3, 130.2, 132.2, 135.3, 140.2, 147.2, 151.8, 157.2, 157.4,

+

3

,5Dibromo4ethylpyridin2amine: into a 500ml threenecked round highresolution mass spectrometry (HRMS) m/z 331.1633. (M+ H , C20H19ON

4

bottom flask purged and maintained with an inert atmosphere of nitrogen were requires 331.1653).

placed 4ethylpyridin2amine (10g, 81.85mM), tetrahydrofuran (200ml), and

5[6Amino4ethyl5(4hydroxyphenyl)3pyridyl]Nmethylpyridine2

1

NBS (29 g, 162.94 mmol) at 0 °C. The resulting solution was stirred at room carboxamide (GNE6776): H NMR (400MHz, DMSOd

6

) δ 9.56 (s, 1H), 8.80

temperature for 15min and then concentrated under vacuum. The residue was (q, J=4.7Hz, 1H), 8.59 (dd, J=2.3, 0.9Hz, 1H), 8.07 (dd, J=8.0, 0.9Hz, 1H), 7.96

purified on a silica gel column eluting with DCM/MeOH (100:1 to 20:1), to afford (dd, J=8.0, 2.2Hz, 1H), 7.79 (s, 1H), 7.11–7.02 (m, 2H), 6.94–6.86 (m, 2H), 5.19

1

7

8g (79%) of the title compound. H NMR (400MHz, CDCl3) δ 8.04 (s, 1H), (s, 2H), 2.84 (d, J=4.8Hz, 3H), 2.26 (q, J=7.4Hz, 2H), 0.62 (t, J=7.4Hz, 3H).

13

.33–7.23 (m, 0H), 4.93 (s, 2H), 2.93 (q, J = 7.5 Hz, 2H), 1.60 (s, 0H), 1.17

C NMR (150MHz, DMSOd ) δ 29.1, 38.8, 41.4, 131.6, 136.6, 137.1, 138.8, 141.5,

6

+

(

t, J=7.5Hz, 3H). Liquid chromatography–mass spectrometry (LC–MS) (electro 146.4, 152.5, 154.1, 161.1, 163.1, 164.2, 171.3. HRMS m/z 349.1659. (M+H ,

+

spray ionization (ESI) m/z): 264.1 (M+H ).

Bromo4ethylpyridin2amine: into a 500ml threenecked roundbottom

flask purged and maintained with an inert atmosphere of nitrogen was placed 2118): H NMR (400MHz, DMSOd6) δ 9.49 (s, 1H), 7.69 (s, 1H), 7.51 (d, J=8.1Hz,

C20H21O2N4 requires 349.1650).

3

4(6Amino4ethyl5(4(methoxymethyl)phenyl)pyridin3yl)phenol (GNE

1

3

,5dibromo4ethylpyridin2amine (18 g, 64.29 mmol) in tetrahydrofuran 2H), 7.31 (d, J=8.0Hz, 2H), 7.17 (d, J=8.0Hz, 2H), 6.86 (d, J=8Hz, 2H), 4.99

(

300 ml). To this was added a solution of nbutyllithium (in hexane) (58 ml, (s, 2H), 4.55 (s, 2H), 3.40 (d, J = 27.5 Hz, 4H), 2.29 (q, J = 7.4 Hz, 2H), 0.68

−

1

13

2

.2 mol l ) at −78 °C. The resulting solution was stirred at −78 °C for 1 h, (t, J=7.5Hz, 3H). C NMR (150MHz, DMSOd

6

) δ 14.1, 26.6, 57.7, 73.4, 115.0,

quenched by the addition of 450ml NH

4

Cl and then extracted with ethyl acetate 119.5, 127.0, 128.3, 129.4, 129.7, 130.5, 135.7, 137.5, 147.0, 148.4, 156.1, 156.2,

+

(

2×500ml). The combined organic layers were washed with brine (2× 500ml), HRMS m/z 331.1750 (M+H , C21H22O2N2 requires 331.1754).

dried over anhydrous sodium sulfate, and concentrated under vacuum. The crude

4[2Amino4ethyl5(2methylindazol6yl)3pyridyl]phenol (GNE6641):

1

product was purified by flash and preparative HPLC to afford 12g (93%) of the title

compound as a white solid. H NMR (400MHz, CDCl

HNMR (400 MHz, DMSOd6) δ 9.54 (s, 1H), 8.34 (s, 1H), 7.76 (s, 1H), 7.71

1

3

) δ 7.89 (d, J=5.1Hz, 1H), (d, J=9Hz, 1H), 7.45 (d, J=0.9Hz, 1H), 7.09 (d, J=9Hz, 2H), 6.99 (d, J=1.2Hz,

6.60–6.48 (m, 1H), 4.93 (s, 2H), 2.69 (q, J=7.6Hz, 2H), 1.33–1.14 (m, 5H). LC–MS 1H), 6.96 (d, J=1.2Hz, 2H) 4.99 (s, 2H), 4.17 (s, 3H), 2.51 (q, J=1.8Hz, 2H), 0.60

+

13

(ESI m/z): 264.1 (M+H ).

(t, J=7.5Hz, 3H). C NMR (150MHz, DMSOd

6

) δ 14.3, 22.8, 116.2, 116.7, 120.0,

4

Ethyl3(4methoxyphenyl)pyridn2amine: into a 500ml threenecked 120.2, 120.6, 124.0, 124.7, 126.7, 127.8, 131.0, 136.3, 146.7, 148.9, 156.8, 157.0.

+

roundbottom flask purged and maintained with an inert atmosphere of nitrogen HRMS m/z 345.1710. (M+H , C21H21ON4 =345.1650).

was placed a solution of 3bromo4ethylpyridin2amine (12g, 59.68mmol)

in CH3CN (100 ml), (4methoxyphenyl)boronic acid (11 g, 72.39 mmol),

4[2Amino4ethyl5(2methoxyphenyl)3pyridyl]phenol (GNE2931):

H NMR (400MHz, DMSOd6) δ 9.54 (s, 1H), 7.58 (s, 1H), 7.34 (ddd, J=8.2,

1

Na2CO3(120ml, sat.), and Pd(dppf)Cl2 (1.2g, 1.64mmol). The resulting solution 7.4, and 1.8Hz, 1H), 7.17–6.94 (m, 5H), 6.88 (d, J=7.7Hz, 2H), 4.89 (s, 2H), 3.71

13

was stirred at 110 °C for 1 h, diluted with of 500 ml of ethyl acetate, and then (s, 3H), 2.09 (q, J=7.2Hz, 2H), 0.55 (t, J=7.5Hz, 3H). C NMR (150MHz, DMSOd6)

extracted with ethyl acetate (2×500ml). The combined organic layers were washed δ 13.8, 23.1, 55.2, 111.0, 116.0, 119.6, 120.2, 123.6, 126.7, 127.6, 128.8, 130.8, 131.6,

+

with brine (3×200ml), dried over anhydrous sodium sulfate, and concentrated 146.7, 149.5. 156.7, 156.8, 156.9. HRMS m/z 321.1598 (M+H , C20H21O2N2

under vacuum. The residue was purified on a silica gel column eluting with ethyl requires 321.1650).

acetate/petroleum ether (1:100 to 1:10) to afford 10g (73%) of the title compound.

3[6Amino4ethyl5(4hydroxyphenyl)3pyridyl]benzamide (GNE2917):

H NMR (400MHz, DMSOd6) δ 9.52 (s, 1H), 7.99 (s, 1H), 7.84 (dtd, J=7.1,

1

H NMR (400MHz, CDCl

3

) δ 8.06 (d, J=5.1Hz, 1H), 7.21–7.11 (m, 2H), 7.06–6.97

(

1

m, 2H), 6.64 (d, J=5.4Hz, 1H), 4.48 (s, 2H), 3.86 (s, 3H), 2.31 (q, J=7.6Hz, 2H), 3.6, 1.8Hz, 2H), 7.73 (s, 1H), 7.54–7.43 (m, 2H), 7.35 (s, 1H), 7.10–6.94 (m, 2H),

+

.11–0.88 (m, 3H). LC–MS (ESI m/z): 264.1 (M+H ).

6.93–6.74 (m, 2H), 5.02 (s, 2H), 2.23 (q, J=4.4Hz, 2H), 0.61 (t, J=4.4Hz, 3H).

1

3

5Bromo4ethyl3(4methoxyphenyl)pyridin2amine: into a 250ml three

C NMR (150MHz, DMSOd6) δ 14.1, 22.6, 115.9, 116.0, 119.8, 125.9, 126.4,

necked roundbottom flask purged and maintained with an inert atmosphere 126.5, 128.2, 128.5, 130.7, 130.8, 132.3, 133.2, 134.2, 139.1, 146.7, 148.4, 156.7,

of nitrogen were placed 4ethyl3(4methoxyphenyl)pyridn2amine (10 g, 157.2. HRMS m/z 334.1550 (M+H , C20H20O2N3 requires 334.1700).

+

4

3.80mmol), tetrahydrofuran (THF) (100ml), followed by NBS (7.8g, 43.83mmol)

4[6Amino4ethyl5(4hydroxyphenyl)3pyridyl]benzamide (GNE2916):

H NMR (400MHz, DMSOd ) δ 9.53 (s, 1H), 7.98 (d, J=6.1Hz,1H), 7.95–7.87

6

1

at 0°C. The resulting solution was stirred at room temperature for 15min, diluted

with 500ml EtOAc and 500ml H2O. The resulting solution was extracted with (m, 2H), 7.72 (s, 1H), 7.43–7.31 (m, 3H), 7.14–6.95 (m, 2H), 6.93–6.74 (m, 2H),

13

ethyl acetate (2×500ml). The organic layers were combined, washed with brine 5.04 (s, 2H), 2.33 (q, J=7.4Hz, 2H), 0.61 (t, J=7.5Hz, 3H). C NMR (150MHz,

2×500ml), and concentrated under vacuum. The residue was purified on a silica DMSOd6) proton correlated carbon chemical shifts δ 14.9, 42.6, 113.9, 116.7,

(

gel column eluting with ethyl acetate/petroleum ether (1:20 to 1:10) to afford 8g 116.7, 129.0, 131.3, 131.4, (15.0, 25.9, 29.0. impurities). LC–MS (ESI) m/z 334.2

1

+

(

59%) of the title compound. H NMR (400MHz, CDCl

3

) δ 8.09 (s, 1H), 7.22–7.09 [M+H , C20H19N3O2].

m, 2H), 7.09–6.98 (m, 2H), 4.55 (s, 2H), 2.47 (q, J = 7.5 Hz, 2H), 1.10–0.84

5(3Acetamidophenyl)2amino4ethyl[3,4′bipyridine]1′oxide (GNE

+

1

m, 3H). LC–MS (ESI m/z): 264.1 (M+H ).

9603): H NMR (400MHz, DMSOd6) δ 9.53 (s, 1H), 7.98 (d, J=6.1Hz, 1H),

4(2Amino5bromo4ethylpyridin3yl)phenol: into a 250ml threenecked 7.95–7.87 (m, 2H), 7.72 (s, 1H), 7.43–7.31 (m, 3H), 7.14–6.95 (m, 2H), 6.93–6.74

roundbottom flask purged and maintained with an inert atmosphere of nitrogen (m, 2H), 5.04 (s, 2H), 2.33 (q, J=7.4Hz, 2H), 2.06 (s, 2H), 0.61 (t, J=7.5Hz,

was placed 5bromo4ethyl3(4methoxyphenyl)pyridin2amine as a white solid 3H). ¹³C NMR (150MHz, DMSOd

6

) δ 13.7, 23.9, 24.0, 120.0, 121.2, 125.6, 127.4,

(

8g, 26.04mmol), dichloromethane (100ml), followed by tribromoborane (19.6g, 129.2, 129.7, 130.2, 134.7, 136.9, 140.1, 140.3, 157.5, 168.8. HRMS m/z 349.1655.

+

7

8.24mmol) at 0°C. The resulting solution was stirred at room temperature for (M+H , C20H21O2N4 requires 349.1659).

1

h and then quenched by the addition of 100ml of NaHCO3 (1M) at 0°C. The Synthesis of GNE-6831 is representative of the syntheses for GNE-0956, GNE-2090,

solids were collected by filtration and then washed with 100ml H

2

O and 300ml GNE-2143, and GNE-2148.

of ethyl acetate/petroleum ether (1:1) to afford 6.3g (83%) of the title compound

Synthesis of 9chloroN(5chloro2,4dimethoxyphenyl)N(cyanomethyl)

1

as a white solid. H NMR (400MHz, CDCl

3

) δ 8.09 (s, 1H), 7.22–7.09 (m, 2H), 5,6,7,8tetrahydroacridine3carboxamide (GNE6831): 9Chloro5,6,7,8

7

.09–6.98 (m, 2H), 4.55 (s, 2H), 2.47 (q, J=7.5Hz, 2H), 1.10–0.84 (m, 3H). LC–MS tetrahydroacridine3carboxylic acid (100mg, 0.38mmol) was dissolved in 2ml of

+

(ESI m/z): 264.1 (M+H ).

N,Ndimethylformamide (DMF) and charged with 1[bis(dimethylamino)methy

4[2Amino4ethyl5(1Hindazol5yl)3pyridyl]phenol (GNE6640): into lene]1H1,2,3triazolo[4,5b]pyridinium 3oxide hexafluorophosphate (HATU)

a 250ml threenecked roundbottom flask purged and maintained with an inert (144mg, 0.38mmol). After stirring at room temperature for 30min, the mixture

atmosphere of nitrogen were placed 4(2amino5bromo4ethylpyridin3yl) was then charged with 3((5chloro2,4dimethoxyphenyl)amino)propanenitrile

phenol (1.0g, 3.41mmol,), 6(tetramethyl1,3,2dioxaborolan2yl)2Hindazole (91mg, 0.38mmol). The resulting solution was stirred at 110°C for 1h, diluted

(

880mg, 3.41mmol,), potassium carbonate (3.3g, 23.88mmol), water (30ml), with of 50ml of ethyl acetate, and then extracted with ethyl acetate (2×100ml).

,4dioxane (25ml), and Pd(dppf)Cl2 (200mg, 0.3mmol). The resulting solu The combined organic layers were washed with brine (3×200ml), dried over

1

2

tion was stirred at 80°C for 16h, and diluted with 500ml H O and 500ml ethyl anhydrous sodium sulfate, and concentrated under vacuum. The residue was puri

acetate. The organic layer was washed twice with brine (2×250ml) and concen fied on a silica gel column eluting with ethyl acetate/petroleum ether (1:100 to 1:10)

trated under vacuum. The residue was purified on a silica gel column eluting with to afford the title compound (110mg, 0.30mmol, 78% yield) of the title compound.

1

DCM/CH3OH (20:1–10:1) to afford the title compound. H NMR (400MHz,

6

H NMR (400MHz, DMSOd ) δ 8.01 (d, J=8.7Hz, 1H), 7.82 (d, J=1.7Hz, 1H),

DMSOd6) δ 13.07 (s, 1H), 9.52 (s, 1H), 8.07 (d, J=1.0Hz, 1H), 7.74 (s, 1H), 7.65 7.62–7.59 (m, 2H), 6.67 (s, 1H), 4.90–4.72 (m, 2H), 3.79 (s, 3H), 3.73 (s, 3H),

1

3

(dd, J=1.6, 0.8Hz, 1H), 7.56 (dt, J=8.6, 0.9Hz, 1H), 7.28 (dd, J=8.5, 1.6Hz, 1H), 2.98–2.90 (m, 4H), 1.85 (m, 4H). C NMR (150MHz, DMSOd6) δ 21.8, 21.8,

©

reSeArCH Letter

2

7.0, 33.5, 37.4, 56.4, 56.4, 98.1, 111.7, 116.3, 122.7, 123.0, 125.1, 125.8, 127.8, DMSO (650ml) was heated at 75°C for 21h. On cooling, the mixture was poured

129.7, 130.3, 135.6, 139.8, 144.8, 154.5, 155.3, 160.6, 170.0. HRMS m/z 470.1033 into water (1,500ml), extracted with ethyl acetate (4×500ml), and the combined

+

(M+H , C24H22O3N3Cl2, requires 470.1029).

organics washed with water (2×400ml) and brine (300ml) then dried (MgSO4).

9

ChloroN(5chloro2,4dimethoxyphenyl)N((3methyloxetan3yl)methyl) The solvent was removed under reduced pressure to afford (Z)1(1methyl4

,6,7,8tetrahydroacridine3carboxamide (GNE0956): H NMR (400 MHz, nitro1Hpyrazol5yl)2,3,6,7tetrahydro1Hazepine as a light brown solid

1

5

1

DMSOd6) 1H NMR (400 MHz, DMSOd6) δ 7.96 (d, J = 8.2 Hz, 1H), 7.71 (50.74g; 99%). HNMR (400MHz, CDCl

3

) δ 8.00 (s, 1H), 5.95–5.85 (m, 2H), 3.80

+

(

d, J=1.4Hz, 1H), 7.56 7.48 (m, 1H), 7.51 (s, 1H), 6.57 (s, 1H), 4.21 (d, J=5.9Hz, (s, 3H), 3.30–3.20 (m, 4H), 2.45–2.35 (m, 4H). LC–MS (ESI m/z): 223.1 [M+ H ].

2

H), 4.12–3.90 (m, 2H), 3.74 (d, J=13.5Hz, 6H), 3.37 (s, 2H), 2.99 (s, 2H), 2.91

4(1Methyl4nitro1Hpyrazol5yl)8oxa4azabicyclo[5.1.0]octane: seventy

) δ 21.8, seven per cent metachloroperbenzoic acid (77 g; 0.343 mol) was added

13

(s, 2H), 1.88–1.77 (m, 3H), 1.42 (s, 3H). C NMR (150MHz, DMSOd

6

2

1

5

1.8, 22.2, 27.0, 33.5, 40.2, 54.6, 56.1, 56.3, 80.0, 80.2, 97.8, 111.4, 122.7, 124.3, 124.4, portionwise over 10min to a stirred, icecooled solution of (Z)1(1methyl

25.8, 126.7, 129.6, 129.8, 137.7, 139.8, 145.0, 154.2, 154.7, 160.3, 170.2. HRMS m/z 4nitro1Hpyrazol5yl)2,3,6,7tetrahydro1Hazepine (50.74g; 0.23mol) in

15.1499 (M+H+, C27H27O4N2Cl2 requires 515.1400).

dichloromethane (1,000ml). An ice bath was used to control the minor exotherm

9

ChloroN(5chloro2,4dimethoxyphenyl)N(2,3dihydroxypropyl)5,6,7, observed during a smallerscale reaction. On complete addition, the ice bath

1

8

tetrahydroacridine3carboxamide (GNE2090): H NMR (400MHz, DMSOd

6

)

was removed and stirring continued at room temperature for 18h. The reaction

δ 7.96 (dd, J = 8.6, 3.4 Hz, 1H), 7.77–7.68 (m,1H), 7.66–7.47 (m, 2H), 6.56 mixture was washed with saturated sodium hydrogen carbonate (750ml), 1N

(

d, J=14.2Hz, 1H), 4.99 (d, J=5.2Hz, 1H), 4.54 (dt, J=15.4, 5.8Hz, 2H), 4.21 sodium hydroxide (2× 500ml), and brine (350ml). The organics were dried

(

4

dd, J=13.6, 4.3Hz, 1H), 3.93 (s, 1H), 3.80–3.63 (m, 5H), 3.46–3.33 (m, 2H), 2.99 (MgSO ) and the solvent removed under reduced pressure to afford 4(1methyl

(

s, 2H), 2.91 (s, 2H), 1.85 (s, 1H). ¹³C NMR (150MHz, DMSOd

6

) proton correlated 4nitro1Hpyrazol5yl)8oxa4azabicyclo[5.1.0]octane as a pale yellow

1

carbon chemical shifts δ: 22.4, 22.4, 27.5, 34.1, 56.6, 56.7, 56.8, 64.7, 98.1, 98.3, 123.2, solid (55.6g). HNMR (400MHz, CDCl3) δ 7.98 (s, 1H), 3.75 (s, 3H), 3.50–3.35

+

126.4, 127.4, 129.6, 129.8. LC–MS (ESI) m/z 505.2 [M+H ].

(m, 2H), 3.30–3.20 (m, 2H), 2.95–2.80 (m, 2H), 2.35–2.15 (m, 4H). LC–MS

+

9

ChloroN(2hydroxyethyl)Nmethyl5,6,7,8tetrahydroacridine3carbox (ESI m/z): 239.2 [M+H ].

1

amide (GNE2143): H NMR (400MHz, DMSOd

6

) δ 8.16 (d, J=8.2Hz, 1H), 7.95

Rel(4R,5R)5azido1(1methyl4nitro1Hpyrazol5yl)azepan4ol: to a

(

s, 1H), 7.62 (d, J=8.6Hz, 1H), 4.79 (s, 1H), 3.66 (s, 1H), 3.57 (s, 1H), 3.48 (s, 1H), stirred solution of 4(1methyl4nitro1Hpyrazol5yl)8oxa4azabicyclo[5.1.0]

13

3

.27 (s, 2H), 3.10–2.94 (m, 7H), 1.90 (td, J=3.8, 1.8Hz, 4H). C NMR (150MHz, octane (29.28g) in methanol (350ml) and water (90ml) was added ammonium

DMSOd6) δ 22.0, 27.0, 33.5, 49.7, 52.6, 57.9, 58.4, 123.5, 125.5, 125.9, 126.5, chloride (16.5 g; 0.308 mol) followed by sodium azide (20 g; 0.307 mol). The

+

1

37.9, 139.9, 145.5, 160.4, 170.0. HRMS m/z 319.1208 (M+H , C17

H

20

O

2

N

2

Cl mixture was heated behind a blast screen at 70°C for 22h, cooled, then concen

trated to 100ml under reduced pressure at 40°C. The concentrated solution was

requires 319.1500).

9

ChloroNmethylN(2morpholino2oxoethyl)5,6,7,8tetrahydroacridine poured into water (1,300ml), extracted with dichloromethane (4×400ml), and

1

3

8

7

3

2

1

4

carboxamide (GNE2148): H NMR (400MHz, DMSOd

6

) δ 8.18 (dd, J=22.2, the combined organics dried (MgSO4). Evaporation under reduced pressure at

.6Hz, 2H), 7.93 (d, J=1.6Hz, 1H), 7.81 (s, 1H), 7.65 (dd, J=8.6, 1.6Hz, 1H), 35°C gave rel(4R,5R)5azido1(1methyl4nitro1Hpyrazol5yl)azepan4ol

.54 (d, J=8.7Hz, 1H), 4.40 (s, 2H), 4.18 (s, 2H), 3.67–3.58 (m, 4H), 3.15 (s, 2H), (antiisomer and a racemic mixture) as a pale yellow oil (34.5g; 96% over two

13

1

.02 (t, 20.9Hz, 4H), 1.94–1.86 (m, 4H). C NMR (150MHz, DMSOd6) δ: 21.9, steps). HNMR (400 MHz, CDCl3) δ 8.03 (s, 1H), 3.85–3.77 (m, 1H), 3.77

7.0, 33.6, 34.5, 44.5, 48.5, 52.3, 66.0, 123.9, 125.4, 126.0, 126.5, 129.8, 137.0, 140.0, (s, 3H), 3.65–3.55 (m, 1H), 3.45–3.15 (m, 4H), 2.85–2.70 (m, 1H), 2.25–2.10

+

45.4, 160.6, 166,0 169.7. HRMS m/z 402.1579 (M+H , C21H25O3N3Cl requires (m, 2H), 2.05–1.85 (m, 2H). LC–MS (ESI m/z): 282.1 [M+H ].

02.1725).

Rel(4R,5R)4azido5fluoro1(1methyl4nitro1Hpyrazol5yl)azepane:

fifty per cent deoxofluor in THF (111ml; 0.307mol) was added slowly over 20min

Synthesis of GNE-0300 is representative of the syntheses for GNE-8735.

Synthesis of (1S)1(5bromo1Hindol2yl)N(2phenylethyl)ethanamine to a stirred, icecooled solution of rel(4R,5R)5azido1(1methyl4nitro1H

GNE0300): 1(5bromo1Hindol2yl)ethan1one (90mg, 0.38mmol) was dis pyrazol5yl)azepan4ol (32.5g; 0.115mol) in dichloromethane (500ml). On

(

solved in 2ml of DMF and charged with (S)phenethylamine (46mg, 0.38mmol). complete addition, the ice bath was removed and stirring continued at room

After stirring at room temperature for 30min, the mixture was diluted with 50ml temperature for 20 h. The reaction mixture was recooled in an ice bath and

ethyl acetate and 50ml water. The mixture was partitioned and the organic was saturated sodium hydrogen carbonate (400ml) added dropwise (note this reac

collected. The aqueous layer was then extracted with ethyl acetate (2×100ml). tion is effervescent). After stirring for 30min, the layers were separated and the

The combined organic layers were washed with brine (3×200ml), dried over aqueous layer extracted with dichloromethane (2×500ml). Pooled organics were

4

anhydrous sodium sulfate and concentrated under vacuum. The residue was puri dried (MgSO ) and the solvent removed under reduced pressure. Flash column

fied on a silica gel column eluting with ethyl acetate/petroleum ether (1:100 to 1:10) chromatography on silica eluting with 0–100% ethyl acetate in isohexane gradient

1

to afford the title compound (120mg, 0.35mmol, 92% yield). H NMR (400MHz, afforded rel(4R,5R)4azido5fluoro1(1methyl4nitro1Hpyrazol5yl)aze

1

DMSOd

6

) δ 11.05 (s, 1H), 7.59 (d, J=1.9Hz, 1H), 7.30–7.19 (m, 3H), 7.20–7.06 pane as a pale orange oil (26g; 80%). HNMR (400MHz, CDCl

3

) δ 8.03 (s, 1H),

(

m, 6H), 6.21 (d, J=1.8Hz, 1H), 3.93 (q, J=6.6Hz, 1H), 2.77–2.56 (m, 4H), 1.36 4.90–4.65 (m, 1H), 4.00–3.85 (m, 1H), 3.77 (s, 3H), 3.40–3.10 (m, 4H), 2.35–2.05

d, J=6.6Hz, 3H). ¹³C NMR (150MHz, DMSOd

) δ 22.5, 36.5, 49.3, 51.6, 98.0, (m, 3H), 1.95–1.75 (m, 1H). LC–MS (ESI m/z): 284.3 [M+H ].

+

(

6

1

11.5, 113.4, 122.0, 123.1, 126.3, 128.7, 129.0, 130.3, 135.1, 140.8, 146.1. HRMS m/z

43.0900, 345.0882 (M+H+, C18H22N2Br requires 343.0850, 345.0911).

Rel(4R,5R)5fluoro1(1methyl4nitro1Hpyrazol5yl)azepan4amine:

a mixture of rel(4R,5R)4azido5fluoro1(1methyl4nitro1Hpyrazol5yl)

3

(

1S)1(5bromo1Hindol2yl)N(2phenylethyl)ethanamine (GNE8735): azepane (26g; 91.8mmol) and triphenylphosphine (24.1g; 92mmol) in tetrahy

1

H NMR (400MHz, DMSOd

6

) δ 10.83 (s, 1H), 7.53 (s, 1H), 7.24–7.12 (m, 6H), drofuran (400ml) and water (80ml) was heated at 60°C for 20h, cooled and con

7

1

2

1

3

.05–7.03 (m, 1H), 3.88 (d, J=4Hz, 1H), 2.81–2.60 (m, 6H), 2.10–1.75 (m, 2H), centrated to approximately 80ml under reduced pressure. Ethyl acetate (500ml)

13

6

.75–1.60 (m, 2H), 1.50–1.30 (m, 2H). C NMR (150MHz, DMSOd ) δ 23.4, was added and the mixture extracted with 1N HCl (4×125ml). Pooled acidic

5.7, 28.3, 32.6, 36.0, 49.3, 55.6, 110.7, 112.8, 119.7, 122.1, 125.8, 128.2, 128.6, extracts were washed with ethyl acetate (500ml), basified to pH 14 with 6N NaOH,

+

24 2

30.3, 132.5, 140.5. HRMS m/z 385.1126, 383.1114 (M+ H , C21H N Br requires and extracted with dichloromethane (3×400ml). Combined extracts were dried

83.1117, 385.1130).

4

(MgSO ) and the solvent removed under reduced pressure to give rel(4R,5R)5

Synthesis of GDC-0570 followed similar procedures as described for GDC-0339 as fluoro1(1methyl4nitro1Hpyrazol5yl)azepan4amine as a pale yellow oil

1

below.

(22g; 93%). HNMR (400MHz, CDCl

3

) δ 8.03 (s, 1H), 4.60–4.40 (m, 1H), 3.77

(

Z)2,3,6,7tetrahydro1Hazepine hydrochloride: 4 N hydrogen chloride in (s, 3H), 3.45–3.10 (m, 5H), 2.35–1.90 (m, 3H), 1.80–1.65 (m, 1H), 1.60 (br s, 2H).

+

1

,4dioxane (250ml; 1mol) was added over 5min to a stirred, icecooled solution LC–MS (ESI m/z): 258.3 [M+H ].

of (Z)tertbutyl 2,3,6,7tetrahydro1Hazepine1carboxylate (50g; 0.254mol) in tertButyl (4R,5R)5fluoro1(1methyl4nitro1Hpyrazol5yl)azepan4

methanol (250ml). On complete addition, the ice bath was removed and stirring ylcarbamate: ditertbutyl dicarbonate (28g; 128.3mmol) was added to a stirred,

continued at room temperature for 3.75h. Volatiles were removed under reduced icecooled solution of rel(4R,5R)5fluoro1(1methyl4nitro1Hpyrazol

pressure and the residue triturated twice with diethyl ether (300ml) to afford 5yl)azepan4amine (22 g; 85.6 mmol) and diisopropylethylamine (22.4 ml;

(

Z)2,3,6,7tetrahydro1Hazepine hydrochloride as a pale pink solid (32.3g; 95%). 128.6mmol) in dichloromethane (600ml). On complete addition, the ice bath

1

HNMR (DMSOd6, 400MHz) δ 9.54 (br s, 2H), 6.40–6.25 (m, 2H), 3.1–3.05 was removed and stirring continued at room temperature for 20h. The reaction

(

m, 4H), 2.55–2.40 (m, 4H).

mixture was washed with saturated sodium hydrogen carbonate (500ml) and the

(

Z)1(1methyl4nitro1Hpyrazol5yl)2,3,6,7tetrahydro1Hazepine: aqueous layer reextracted with dichloromethane (2×300ml). Pooled organics

a mixture of (Z)2,3,6,7tetrahydro1Hazepine hydrochloride (32.3g; 0.24mol), were dried (MgSO4) and the solvent removed under reduced pressure. Flash

chloro1methyl4nitro1Hpyrazole (37.2g; 0.23mol), potassium fluoride column chromatography on silica eluting with 0–100% ethyl acetate in isohexane

56.24g; 0.96mol), and diisopropylethylamine (64ml; 0.362mol) in anhydrous gradient afforded a pale yellow solid (29.7g; 97%). Chiral separation of the racemic

5

(

©

Letter reSeArCH

mixture by supercritical fluid chromatography using a Chiralpak IA column with micromass ZQ single quadropole mass detector, and an XBridge C18 column

an isocratic mobile phase of 15% methanol (with 0.1% NH

4

OH) in carbon dioxide (5μm, 4.6mm×50mm).

1

gave the desired product (second peak) as a single enantiomer. HNMR (400MHz,

tertButyl (3R)3[(2S)2{[(benzyloxy)carbonyl]amino}3phenylpropan

CDCl

3

) δ 8.04 (s, 1H), 5.05 (m, 1H), 4.80–4.55 (m, 1H), 4.15–4.05 (m, 1H), 3.79 amido]pyrrolidine1carboxylate: to a solution of Cbzꢀphenylalanine (0.9g,

(

s, 3H), 3.45–3.10 (m, 4H), 2.35–2.05 (m, 3H), 1.95–1.80 (m, 1H), 1.47 (s, 9H). 3.00mmol) in dichloromethane (25ml) were sequentially added HOBt monohy

+

LC–MS (ESI m/z): 358.3 [M+H ].

drate (0.46g, 3.00mmol) and EDAC (0.58g, 3mmol). To this mixture was added a

tertButyl (4R,5R)5fluoro1(4amino1methyl1Hpyrazol5yl)azepan solution of tertbutyl (3R)3aminopyrrolidine1carboxylate (0.56g, 3.00mmol).

ylcarbamate: to a stirred solution of tertbutyl (4R,5R)5fluoro1(1methyl The reaction was stirred at ambient temperature for 18h, then partitioned between

nitro1Hpyrazol5yl)azepan4ylcarbamate (15.0g; 42 mmol) in ethanol dichloromethane and 5% aqueous citric acid. The layers were separated, and the

4

(

1,000ml) and water (100ml) was added ammonium chloride (11.34g; 210mmol) organic phase was washed with aqueous saturated NaHCO

3

and brine, dried over

and iron powder (9.4g; 168mmol). The mixture was heated at 99°C for 2.75h, MgSO

4

, filtered, and concentrated. The crude concentrate was purified by flash

cooled, filtered through Celite and evaporated to approximately 100 ml. The silica gel chromatography (25g Silicycle column), eluting with ethyl acetate in

concentrate was diluted with water (1,000ml) and extracted with ethyl acetate hexanes (20–80%) to provide 1.27g (91%) of the title compound.

(

2×500ml). The pooled extracts were washed with water (200ml), dried (MgSO

4

)

(2S)N[(3R)1cyanopyrrolidin3yl]3phenyl2(3phenylpropanamido)

and the solvent removed under reduced pressure to give tertbutyl (4R,5R)5 propanamide (GNE3086): a solution of tertbutyl (3R)3[(2S)2{[(benzyloxy)

fluoro1(4amino1methyl1Hpyrazol5yl)azepan4ylcarbamate as a pale carbonyl]amino}3phenylpropanamido]pyrrolidine1carboxylate (1.15 g,

1

brown solid (12.6g; 91%). HNMR (400MHz, CDCl

3

) δ 7.13 (s, 1H), 6.35–6.20 2.46mmol) in ethanol (50ml) was passed through 10% Pd/C catalyst at a rate

−

1

(

m, 1H), 4.85–4.65 (m, 1H), 4.35–4.15 (m, 1H), 3.66 (s, 3H), 3.45–3.30 (m, 2H), of 1ml min in the HCube reactor at 10 bar and 20°C until the reaction was

3

.152.90 (m, 2H), 2.65 (s, 2H), 2.30–2.15 (m, 1H), 2.15–1.95 (m, 2H), 1.90–1.80 judged complete by LC–MS. The reaction mixture was concentrated under

+

(

m, 1H), 1.45 (s, 9H). LC–MS (ESI m/z): 327.2 [M+H ].

reduced pressure to obtain 0.725g of crude tertbutyl (3R)3[(2S)2amino3

tertButyl (4R,5R)1(4(5tertbutoxycarbonylamino2(2,6difluorophenyl) phenylpropanamido]pyrrolidine1carboxylate as an offwhite, oily, solid that was

+

thiazole4carboxamido)1methyl1Hpyrazol5yl)5fluoroazepan4ylcarbamate: used without further purification. LC–MS: m/z=334.18 [M+H] .

a mixture of tertbutyl (4R,5R)5fluoro1(4amino1methyl1Hpyrazol5yl)

To a solution of tertbutyl (3R)3[(2S)2amino3phenylpropanamido]

azepan4ylcarbamate (12.6g, 38.5mmol), 5(tertbutoxycarbonylamino)2(2,6 pyrrolidine1carboxylate (0.036 g, 0.108) in tetrahydrofuran (0.5 ml) were

difluorophenyl)thiazole4carboxylic acid (14.4g, 40.4mmol), diisopropyleth added 3phenylpropanoyl chloride (0.016ml, 0.108mmol) and DIPEA (0.02ml,

ylamine (13.4ml, 77mmol), and PyBOP (26.1g, 50mmol) in dichloromethane 0.108mmol). The reaction was stirred at ambient temperature for 90min and

(

400ml) was stirred at room temperature for 48h. Saturated sodium hydrogen concentrated under reduced pressure to obtain crude tertbutyl (3R)3[(2S)3

carbonate (600ml) was added and stirring continued for 30 min. The mixture was phenyl2(3phenylpropanamido)propanamido]pyrrolidine1carboxylate that

filtered, the layers separated and the aqueous layer extracted with dichlorometh was used without further purification.

ane (500ml). The pooled organics were dried (MgSO4) and the solvent removed

To the crude tertbutyl (3R)3[(2S)3phenyl2(3phenylpropanamido)

under reduced pressure. Flash column chromatography on silica eluting with propanamido]pyrrolidine1carboxylate in dichloromethane (1.0ml), were added

–100% ethyl acetate in isohexane gradient afforded tertbutyl (4R,5R)1 triethylsilane (1ml), water (3ml), and trifluoroacetic acid (0.5ml). The reaction

4(5tertbutoxycarbonylamino2(2,6difluorophenyl)thiazole4carboxamido)1 was stirred at ambient temperature for 90 min, concentrated under reduced

methyl1Hpyrazol5yl)5fluoroazepan4ylcarbamate as a pale yellow solid pressure, and triturated with hexanes:diethyl ether (1:1) to obtain (2S)3phenyl

2.5g, 9.8%). Further elution with 0–10% methanol in ethyl acetate, then 10% 2(3phenylpropanamido)N[(3R)pyrrolidin3yl]propanamide as a trifluoro

0

(

methanol in dichloromethane, evaporation of relevant fractions under reduced acetic acid salt.

pressure, and trituration of the residue with cold diethyl ether gave a further 7.61g

To (2S)3phenyl2(3phenylpropanamido)N[(3R)pyrrolidin3yl]propan

1

(

30%) of product. HNMR (400MHz, CDCl3) δ 10.34 (s, 1H), 8.75 (s, 1H), 7.89 amide trifluoroacetic acid in acetonitrile (0.7ml) were sequentially added DIPEA

(

s, 1H), 7.40–7.25 (m, 1H), 7.15–7.00 (m, 2H), 4.95–4.85 (m, 1H), 4.85–4.65 (m, 1H), (0.07ml, 0.4mmol) and cyanogen bromide (2.0M in MeCN, 0.05ml, 0.1mmol).

4

.15–4.00 (m, 1H), 3.77 (s, 3H), 3.45–3.30 (m, 2H), 3.25–3.00 (m, 2H), 2.35–2.10 The reaction was stirred at ambient temperature for 30min and then diluted with

(

m, 3H), 1.95–1.75 (m, 1H), 1.55 (s, 9H), 1.43 (s, 9H).

dichloromethane. The diluted reaction was washed sequentially with water, 10%

5

AminoN(5((4R,5R)4amino5fluoroazepan1yl)1methyl1Hpyrazol aqueous KHSO4, and brine. The organic layer was dried over MgSO4, filtered,

4

yl)2(2,6difluorophenyl)thiazole4carboxamide (GDC0339): tertbutyl and concentrated. The crude concentrate was purified by flash silica gel chroma

4R,5R)1(4(5tertbutoxycarbonylamino2(2,6difluorophenyl)thiazole4 tography (4g Silicycle column), eluting with acetone/dichloromethane to obtain

(

1

carboxamido)1methyl1Hpyrazol5yl)5fluoroazepan4ylcarbamate (10.11g, 25 mg (64%) of the title compound. H NMR (400 MHz, CDCl3) δ 7.14–7.31

1

5.2mmol) in 4N HCl in dioxane (200ml) and methanol (200ml) was stirred at room (m, 10 H), 5.93 (d, J=6.6Hz, 1 H), 5.88 (br. s., 1H), 4.52 (d, J=7.1Hz, 1 H), 4.28

temperature for 20h. Evaporation under reduced pressure afforded a pale brown solid (d, J=6.1Hz, 1 H), 3.49 (dd, J=10.2, 6Hz, 1 H), 3.33 (dd, J=8.6, 5.1Hz, 1 H),

which was dissolved in 50% methanol in dichloromethane and added to a strong cation 3.17 (d, J=9.5Hz, 1 H), 3.03–3.08 (m, 2 H), 2.83–2.94 (m, 3 H), 2.50 (q, J=7.6Hz,

1

3

exchange (SCX) chromatography cartridge. After washing with dichloromethane 2 H), 1.94–1.98 (m, 1 H), 1.49–1.51 (m, 1 H). C NMR (150MHz, DMSOd6)

and methanol, elution with 1N ammonia in methanol and evaporation of the eluent δ 30.9, 31.0, 36.7, 38.2, 48.3, 48.7, 53.7, 54.5, 117.2, 125.8, 126.3, 128.0, 128.1, 128.2,

under reduced pressure afforded 5aminoN(5((4R,5R)4amino5fluoroazepan 129.2, 137.5, 141.3, 171.0, 171.1. LC–MS: m/z=391.2 [M+H] .

+

1

yl)1methyl1Hpyrazol4yl)2(2,6difluorophenyl)thiazole4carboxamide as Hit-to-lead selection cascade assays. Biochemical deubiquitinase assays.

the free base (6.3g, 89% yield). Purity by a 30min LC–MS at ultraviolet 254nm: 98.00%; Biochemical deubiquitinase assays used ubiquitinRho110 as a substrate to

D

3

−1

−3

melting point: 185.73°C; [α] 20 (in deg cm g dm )=

+6.312 (c=

0 .00301 g cm

enable kinetic monitoring of reactions for USP7, USP7 catalytic domain, USP5,

1

in methanol); HNMR (400MHz, DMSOd6) δ 8.84 (s, 1H), 7.57–7.45 (m, 1H), and USP47. The deubiquitinase proteins and their concentrations that were used

7

.53 (s, 1H), 7.49 (s, 2H), 7.32–7.22 (m, 2H), 4.42 (dtd, J=3.6, 8.3, 47.9Hz, 1H), 3.64 in biochemical reactions were as follows. USP7, fulllength Nterminal Histag,

(

s, 3H), 3.24–3.02 (m, 5H), 2.18–2.01 (m, 1H), 2.01–1.73 (m, 4H), 1.65–1.52 (m, 1H); native C terminus: 0.18nM (Genentech, Hs_USP7 21102); USP7 catalytic domain

13

C NMR (101MHz, DMSOd6) δ 32.7, 33.6, 35.9, 46.4, 48.9, 55.5, 98.4, 111.4, 112.7, Nterminal Histag: 40nM (Genentech, Hs_USP7.K208K554); USP5, fulllength:

1

12.9 113.1, 122.7, 131.6, 132.6, 133.9, 142.0, 158.7, 159.8, 162.5. ¹⁹FNMR (564MHz, 0.5nM (Boston Biochem catalogue number E322, lot 02010210); USP47, full

DMSOd

6

) δ 111.2 (dd, 2F), −171.7 (m, 1F). HRMS (m/z): [M+1]⁺calculated for length Nterminal Histag:0.8nM (Genentech Hs_USP47.M1D1287). Substrate

C20H23F3N7OS, 466.1559; found, 466.1626;

Synthesis of peptidomimetic GNE-3086.

K

m

values for USP proteins were USP7 fulllength, 2μM; USP5 fulllength, 0.33μM;

and USP47 fulllength, 0.175μM. The final assay conditions were as follows. The

General methods: unless otherwise noted, reactions were conducted using reaction buffer consisted of 50mM Tris (pH 7.5), 0.01%(v/v) Triton X100, 2.5mM

commercially available reagents and solvents, which were used as received. dithiothreitol, and 0.1% (w/v) bovine gamma globulin (Sigma catalogue number

Hydrogenation reactions were performed in a ThalesNano HCube reactor G500925G). The final substrate ubiquitinRho110 (Boston Biochem catalogue

using 30mm 10% Pd/C catalyst cartridges. Silica gel chromatography was per number U555) concentration used for reactions was 1μM. Reactions were per

formed using a Biotage Isolera Four flash purification system with Silicycle formed for 1h at room temperature, in black 20μlvolume polystyrene ProxiPlate

1

SiliaSep cartridges. H NMR spectra were recorded on a Varian INOVA400 400 384 F Plus (PerkinElmer catalogue number 6008260). Test compounds, including

MHz spectrometer. Chemical shifts are reported in δ units (p.p.m.) relative to a control USP7 inhibitor (Ubaldehyde, Boston Biochem catalogue number

residual solvent peak. Coupling constants (J) are reported in hertz (Hz). LC–MS U201), were serially diluted in DMSO, in 384well clear Vbottom polypropylene

analyses were performed using a Waters 2795 Separations Module equipped with plates (Greiner catalogue number 781280). Compounds in DMSO were diluted

a Waters 2996 photodiode array detector, a Waters 2424 ELS detector, a Waters tenfold into reaction buffer, to achieve threefold the final desired concentration.

©

reSeArCH Letter

The substrate, ubiquitinRho110 (Boston Biochem catalogue number U555), was Bioapplication ‘Compartmental Analysis ’. Channel 1 was used to define the nuclear

prepared at 3μM (threefold the final concentration) and 5μl was dispensed into region. Measurements were made of ‘Mean_CircAvgIntCh2 ’, which was the Alexa

the reaction plate. Five microlitres of the compounds (diluted in reaction buffer at Fluor 555 fluorescence intensity (MDM2) within the nuclear region measured on

threefold the final concentration) were transferred to the reaction plate. Five micro a percell basis and averaged over all the measured cells. Data analysis and EC50

litres of DUB protein, which was diluted in reaction buffer at threefold the final calculation were performed in GraphPad Prism using nonlinear fourparameter

concentration, was transferred to the reaction plate to initiate the reaction. After 1h curve fitting.

incubation at room temperature, the reaction was quenched by the addition of 5μl Cathepsin-B protease assay. CathepsinB proteolytic activity was quantified by an

of 400mM acetic acid, in the case of an endpoint reaction. The enzymatic product LC–MS, MRMbased detection method. In brief, varying concentrations of experi

was measured by quantifying the fluorescence signal of cleaved Rhodamine110 mental compound were incubated with 0.5nM human liver cathepsinB (EMD

using excitation at 485nm and emission at 535nm. When preincubation of DUB Millipore, 219364) and benzyloxycarbonylArgArg7amino4methylcoumarin

with compounds was required, the order of addition of reagents was modified to (CbzRRAMC) fluorogenic substrate in buffer containing 10mM MES pH 6.0 and

premix the compounds with DUB (with a 1h incubation period), before the addi 1mM DTT. The reaction was incubated for 5min at room temperature, followed

tion of the substrate and the initiation of the reaction period. Percentage inhibition by quenching with the addition of an equal volume of 2% formic acid in water.

values were calculated relative to a noenzyme control and an uninhibited enzyme Free AMC liberated by cathepsinB was quantified on a Sciex 5500 QTRAP mass

control. Curve fitting and IC50 calculations were performed using Genedata spectrometer (Sciex) equipped with a Biocius Rapidfire highthroughput LC

Screener software. Conversion between IC50 and K

the Cheng–Prusoff equation.

i

values was performed using system (Agilent). AMC product was captured on an Agilent C18 Rapidfire

cartridge, desalted with a 0.1% formic acid wash, and then eluted with 80%

Cellular ubiquitin–MDM2 assay. HCT116 colon cancer cells or SJSA1 osteo acetonitrile, 0.1% formic acid. Measured MRM areaundercurve for AMC was

sarcoma cells were seeded at a density of 150,000 cells per well in 90μl RPMI plotted against compound concentration using GraphPad Prism and fitted for

1

640 medium, 10% FBS (or 0.5% FBS for low serum conditions), 1× GlutaMAX IC50 using fourparameter fitting. MRM parameters for AMC were as follows: ESI

(

Gibco) in 96well black clear bottom, tissueculturetreated microplates (Greiner, positive mode, Q1=233.1, Q3=175.2, collision energy=16.

catalogue number 655090), and incubated for 2h at 37°C, 5% CO2 in a tissue Caspase-3 protease assay. Caspase3 proteolytic activity was quantified by an LC–

culture incubator. Compounds were prepared in a serial dilution in DMSO at 200× MS, MRMbased detection method. In brief, varying concentrations of exper

the final desired concentration in a 96well polypropylene Vbottom microplate imental compound were incubated with 0.025nM of recombinant caspase3

(

Greiner, catalogue number 651261), then diluted 1:20 in RPMI tissue culture (cloned and purified inhouse) and 1 μM (ZDEVD)2Rho110 fluorogenic

medium and 10μl transferred to each well of the cell plate. Cell plates were incu substrate in buffer containing 25 mM HEPES, pH 7.2, and 5 mM DTT. The

2

bated overnight for 20h, at 37°C, 5% CO . Twenty microlitres of a 120μM stock (in reaction was incubated for 20min at room temperature, followed by quench

RPMI) of the proteasome inhibitor MG132 (Cayman Chemical, catalogue num ing with the addition of an equal volume of 2% formic acid in water. Free

ber 10012628) was added to each well. Cells were incubated for 1h at 37°C, 5% ZDEVD liberated by caspase3 was quantified on a Sciex 5500 QTRAP mass

2

CO . Quantification of ubiquitin–MDM2 was performed using a Ub/total MDM2 spectrometer (Sciex) equipped with a Biocius Rapidfire highthroughput LC

wholecell lysate kit (MSD, catalogue number K15168D2). Cells were lysed by system (Agilent). ZDEVD product was captured on an Agilent C18 Rapidfire

adding 15μl of 5× MSD lysis buffer (containing additives: 10mM NaF, 10mM cartridge, desalted with a 0.1% formic acid wash, then eluted with 80% acetonitrile,

βglycerophosphate, 1.5mM Na3VO4, protease inhibitor cocktail (Sigma, P8340)) to 0.1% formic acid. Measured MRM areaundercurve for ZDEVD was plotted

each well and incubated at 4°C for 30min with shaking. One hundred microlitres of against compound concentration using GraphPad Prism and fitted for IC50 using

lysate were transferred to each well of the MSD 96well plate and incubated at room fourparameter fitting. MRM parameters for ZDEVD were as follows: ESI negative

temperature for 1h while shaking (650r.p.m.) in the dark. The MSD plates were mode, Q1 =609.2, Q3 =440.2, collision energy=−35.

washed three times in Trisbuffered saline (50mM TrisCl, pH 7.5, 150mM NaCl) USP7 aggregation analysis. Aggregation of fulllength USP7 was confirmed by

using a Biotek EL405 plate washer. Three millilitres of detection antibody solution dynamic light scattering using a Wyatt DynaPro Plate Reader. Dynamic light

were prepared per plate (1ml of block buffer A, 1.82 ml 1× Tris wash buffer, 150μl scattering data were acquired at 37°C, with Dynamics V7 software, with a 10s

2

% Blocker DM, 30μl 10% Blocker DR, 60μl 50× antitotal MDM2 antibody). acquisition time, ten acquisitions per measurement in autoattenuation mode.

Twentyfive microlitres of detection antibody solution were added per well Compounds were present at 100 μM in a buffer containing 50 mM HEPES

and incubated for 1h at room temperature (650r.p.m.) in the dark. Plates were pH 7.2, 150mM NaCl, 0.01% Triton X100, 1mM TCEP, 0.1% DMSO. Fulllength

−1

washed three times in Trisbuffered saline using a Biotek EL405 plate washer. MSD USP7 was present at 1mg ml with all compounds except rottlerin. Aggregate is

read buffer was prepared according to the manufacturer’s instructions and 150μl defined here as having a hydrodynamic radius greater than 10nm.

added per well. Plates were read using a MSD Sector Reader. The final meas USP7 covalent modification and LC–MS analysis. To evaluate potential covalent

urement was the ratio of ubiquitinated MDM2 to total MDM2. The percentage modification of the proteins, fulllength USP7 was incubated with excess com

increase in ubiquitin–MDM2 was calculated relative to DMSO controls using pound at room temperature overnight and covalent modification was evaluated

Genedata Screener software.

by LC–MS using standard methods.

Total MDM2 immunofluorescence. HCT116 cells were seeded at a density of 40,000 Cell culture and cell treatments. HCT116 parental, USP7null (HD R02028),

cells per well in 384well tissue culture plates (Greiner 781091) in 50μl per well of and HCT116 p53null (HD 104001) cell lines were purchased from Horizon.

RPMI, 10% FBS, 2mM ꢀglutamine, and incubated overnight. Test compounds All other cell lines were obtained from Genentech’s repository and were cultured

were prepared in a 20point serial dilution (one to twofold) in DMSO using a in standard conditions in RPMI medium containing 10% FBS (solid tumour cell

Biomek FX in a 384well Labcyteapproved polypropylene plate (Labcyte P05525). lines) or 20% FBS (haematopoietic cell lines), 1% penicillin/streptomycin, and

Compounds were acoustically dispensed into the cell plates using a Labcyte Echo 1% ꢀglutamine. All cell lines were mycoplasmatested and cell line identity was

(

final total volume transferred was 50nl) (DMSO final concentration was 0.1% confirmed by short tandem repeat and single nucleotide polymorphism profiling

2

4

v/v). Cell plates were incubated at 37°C, 5% CO2, for 24h. Cells were fixed by as described . For studies evaluating cellular effects of compound treatments,

addition of 15μl of 16% paraformaldehyde (Electron Microscopy Sciences 15710 2,500–5,000 cells were seeded in one well of a 96well plate (Corning 3904). The

S) directly to the 50μl cell culture medium in each well. Plates were incubated for following day, the medium was changed from normal (10%) to low serum (0.5%)

3

0min at room temperature. The well contents were aspirated using a Biotek EL406 containing vehicle (DMSO) or compounds. Sixteen to 24h later, FBS was added

and 50μl per well of permeabilization/block buffer added (phosphatebuffered to each well to bring serum levels back to 10%. Cells were then allowed to grow for

saline (PBS, pH 7.5), Triton X100 0.5% (v/v), BSA 0.5% (w/v), proclin 15 p.p.m.). 2 additional days. All treatments were done in triplicate. For degradation rescue

Plates were incubated for 30min then washed three times with 100μl per well of studies, the proteasome inhibitor bortezomib (Selleckchem) or the UAE inhibitor

PBS. Twentyfive microlitres per well of antiMDM2 (rabbit polyclonal, AbCam MLN7243 (Active Biochem) was added at 5μM 30–45min before collection.

ab58530 diluted 1:500 in PBS, BSA 0.5% (w/v), Triton X100 0.1% (v/v)) were Cycloheximide chase studies. MCF7 cells were treated for a total of 7h with

dispensed into each well. Plates were incubated for 2h at room temperature then DMSO or 15 μM of the indicated compounds. During the 7 h of compound

washed four times with 100 μl per well of PBS using a Biotek EL406. Twenty treatment, 50 μM cycloheximide was added for the indicated times before

five microlitres per well of Alexafluor 555 conjugated antirabbit IgG (Life collection. Cells were subsequently collected and lysates processed for western

−

1

Technologies A31572, diluted 1:1,000 and Hoechst 33342 1μg ml diluted in blot analysis.

PBS, BSA 0.5% (w/v), Triton X100 0.1% (v/v)) were dispensed into each well. Antibodies and reagents. Antibodies to the indicated proteins were purchased

Plates were incubated for 2h at room temperature then washed four times with from the following vendors: USP7 (ab84098), USP5 (ab84695) (AbCam) or (4833)

1

00μl per well of PBS using a Biotek EL406. Fluorescence images of the samples (Cell Signaling Technology); MDM2 (Santa Cruz sc965 or EMD Millipore 07575);

(

channel 1: 38623_BGRFRN_ BGRFRN (DNA); channel 2: 54915_BGRFRN_ tubulin (LICOR 92642211); tubulinHRP (5346), actinHRP (12620), GAPdH

BGRFRN (MDM2)) were acquired using a Cellomics XTI Arrayscan with the HRP (8884), caspase3 (9662), PARP (9441), total S6 (2217), phosphoS235/236S6

©

Letter reSeArCH

(

2945), Mcl1 (4572), PIM2 (2730), cleavedcaspase3 (2664), phosphoS112Bad of fixed doses of GNE6776 (0 nm, 125 nM, 250 nM, 500 nM, 1,000 nM, and

5284), p21 (2947) (Cell Signaling Technology); K48 polyubiquitin (Genentech); 2,000nM) or GNE6640 (400nM). In brief, 5,000 EOL1 cells were seeded into

(

USP47 (Bethyl A301190A); UCHL1 (Invitrogen 381000); HAHRP (clone HA7) 384well plates, and compound was added 24h later. Cell viability was deter

and Flag (A8592) (Sigma); p53 (Thermo Scientific MS738P1); p21 (Millipore mined 120h after compound addition (CellTiterGlo). Curves were fitted, and

0

5655), Bad (AF819) (R&D). Secondary antibodies were purchased from LICOR both IC50 and mean viability metrics were calculated. The IC50 was the dose at

Biosciences or Jackson ImmunoResearch. Fulllength human USP7 was cloned into which inhibition was 50% relative to untreated wells. The mean viability was the

a pRK5 mammalian expression construct with a Cterminal Flagtag and muta average of the fitted viabilities at each tested dose. Mean viability was equivalent to

tions were introduced by QuickChange sitedirected mutagenesis as instructed by the area under the logdose/viability curve divided by the total number of tested

the manufacturer (Agilent).

doses. Mean viability values were used for the analysis described in Extended Data

Western blotting, quantification, immunoprecipitations, and deubiquitinase Fig. 6g. All data were fitted using Genedata Screener software.

assays. Cells were treated with compounds as detailed above before lysis for western Primary combination screen analysis. Normalized mean viabilities were determined

blot analysis. Primary antibodies were diluted 1:1,000 and incubated for 1 h at room in the EOL1 cell line for 574 compounds that have known protein or mecha

temperature or 4°C overnight. Secondary antibodies were diluted 1:10,000 and nistic targets, in the presence of DMSO or increasing concentrations of GNE

incubated for 30–60min at room temperature. Blots were either imaged using a 6776 (100nM, 250nM, 500nM, 1,000nM or 2,000nM) or 400nM of GNE6640.

chemiluminescent reagent (Pierce) or were scanned and bands quantified using For each compound, we assessed the difference in mean viability between USP7

LICOR Odyssey instrumentation and software, respectively. Immunoprecipitations inhibitor treatment and the DMSO treatment. For targets targeted by three or

with ubiquitinspecific antibodies were performed as described²⁵, and immuno more compounds, we calculated the enrichment of high mean viability difference

precipitates and corresponding cell lysates were analysed by immunoblot analysis for each concentration of USP7 inhibitor by using a Wilcoxon ranksum test. For

25

as previously described using the antibodies detailed above. For in vitro deubi visualization purposes, we combined the results of all concentrations by taking the

quitination of PIM2, ubiquitinated PIM2 was captured by treating MV411 cells mean of the −log10(transformed P values) for each target.

for 45min with the proteasome inhibitor MG132 (SelleckChem). Cells were Bliss analysis. PIM inhibitors were tested in a ninepoint dose–response matrix

washed, lysed in a 6M urea lysis buffer, and antiK48 polyubiquitin immuno with GNE6676 in the same manner as described for the compound library screen.

precipirates were washed and deubiquitinated with 250nM wildtype or C223S Inhibitors were screened in threefold dilution using a top concentration of 10μM

fulllength recombinant USP7 following a similar protocol as described²⁵

.

for PIM inhibitors and 20μM for GNE6676. Bliss calculations were performed

Cell viability studies and analysis. IncuCyte live cell analysis. One day after cell using Genedata Screener software and the synergy score was calculated as

seeding, 10% cell culture medium was changed to 0.5% serum medium containing previously described²⁸

μM CellEvent Caspase 3/7 reagent (Life Technologies C10423) and compounds. Exome sequencing pipeline. Exome sequencing and analysis of the cell lines

.

2

The plates were placed in an IncuCyte live cell imager and scanning was started was performed using standard protocols and analysed by methods previously

within 15–20min. Twentyfour hours later, 10% serum was added back to the cell described²⁹. FASTQ reads were aligned to the human reference genome (GRCh38)

cultures. Images were taken every 2h for 68–72h, using a 10× objective. Phase using GSNAP version ‘20131010’ with the following parameters: M 2 n 10 B

contrast was used to measure cell confluency/density and green fluorescence 2 i 1–pairmaxdna=1,000–terminalthreshold=1,000–gmapmode=none–clip

was used to measure caspase activity. The images were analysed using IncuCyte overlap. Duplicate reads in the resulting BAM file were marked using PicardTools,

software (Basic Analysis parameters) and a ratio of caspase activity to cell density/ and indels realigned using the GATK IndelRealigner tool. Variations were called

count was determined. For combination experiments, MCF7 cells were treated with using the Bioconductor package VariantTools version 1.9.4. with default options

1

5μM USP7 inhibitors or 0.1μM doxorubicin alone or in combination. Similarly, except for two exceptions: (1) no variants were called in repeat regions as defined

U2OS cells were treated with 15μM USP7 inhibitors or 1μM cisplatin alone or in by the annotation Dust, Satellite repeats, and Tandem repeats in EnsEMBl 77;

combination.

and (2) the avgNborCount post filter was configured using all single nucleotide

CellTiter-Glo assays with wild-type and USP7-null HCT-116 cells. Cells were polymorphisms from dbSNP version 138.

plated and cultured as described for IncuCyte studies. Seventytwo hours after Deubiquitinase selectivity analysis. Recombinant deubiquitinase di-ubiquitin mass

compounds were added, CellTiterGlo (Promega) reagent was added following spectrometry cleavage assay. The MALDI–TOF DUB assay was performed using the

the vendor’s protocol. Three times more USP7null HCT116 cells were plated indicated concentrations of recombinant deubiquitinases, diubiquitin substrates,

4

per well than wildtype HCT116 cells (7,500 versus 2,500 cells), given the slower and USP7 inhibitor compounds as described previously . The inhibition efficiency

proliferation of USP7null cells.

for GNE6640 and GNE6776 against the UCHl family members was monitored

Tumour cell-line panel viability. GNE6640 and GNE6641 were profiled for on UbUbe2W (UbE2), an alternative substrate to diubiquitin.

days across 441 cell lines, and GNE6776, GNE6640, and GNE6641 were Endogenous deubiquitinase activity-based probe assay. HEK293T cells at 80%

3

26

profiled for 5 days across a subset of 185 cell lines as previously described . In confluency were collected by rinsing the plate once with 10ml PBS followed

brief, compounds were screened in ninepoint dose–response using a threefold by scraping. Cells were cleared by spinning them for 3min at 350g at 4°C and

dilution. Cells were seeded into 384well plates 24h before compound addition. cell pellets were flashfrozen in liquid nitrogen and stored at −80°C until lysis.

Cells were then incubated with compound for 72h or 120h before assaying viability Frozen pellets were lysed by quickly rethawing them in buffer A (50mM TrisHCl

(

CellTiterGlo, Promega). Assays were performed in biological triplicate. Cells pH 7.5, 250mM sucrose, 1× phosphatase STOP, 2.5mM TCEP, 2mM ATP, 50μM

were incubated (37°C, 5% CO2) in RPMI1640, 2.5% FBS (72h assay) or 5% FBS phenylmethylsulfonyl fluoride (PMSF), 120mM NaCl, 5mM MgCl ) and the lysate

2

(

120h assay), and 2mM glutamine throughout the assay. The reported IC50 and cleared by centrifugation by spinning at 18,000g at 4°C. The protein concentration

mean viability metrics were as follows: IC50 was the dose at which the estimated was adjusted to 5mg ml⁻, and 5mg (1ml) each of this cell lysate was incubated

1

inhibition was 50% relative to untreated wells (that is, absolute IC50). The mean with indicated compounds or DMSO at the indicated concentrations for 20min at

26

viability was calculated as previously described

.

900r.p.m., 25°C in a Thermomixer (Eppendorf AG). After compound incubation,

Pharmacogenomics analysis. We determined genomic and nongenomic 300ng of recombinant viral DUB was added and lysates were incubated with

features that were associated with differences in compound sensitivity. Mean 6.6μg ml⁻of the indicated activitybased DUB probe, at 1,250r.p.m. and 25°C,

viability was calculated as the arithmetic average of the fitted viabilities at each for the indicated times. The reaction was terminated by adding a 20% SDS solution

1

26

tested dose, as described . For this analysis, normalized mean viabilities of GNE to a final concentration of 0.4% for at least 30min at room temperature, rotating.

640 and GNE6446 were determined by normalizing the mean viabilities of each Subsequently, the lysate was diluted ten times with buffer B (50mM TrisHCl

6

compound by the mean viability of GNE6641. Cancer type, cell histology, loss pH 7.5, 150 mM NaCl, 5 mM EDTA, protease inhibitor cocktail EDTAfree

offunction, and hotspot mutations were assessed for statistical association with (Roche), 50 μM PMSF, 0.5 % NP40) into 15 ml conical tubes. About 120 μl

changes in normalized mean viability. Lossoffunction protein coding mutations slurry of preequilibrated antiHA affinity matrix (Roche) was added and

included (1) insertions, (2) deletions, and (3) substitutions resulting in predicted HAtagged proteins were immunopurified with this matrix by rotating the

truncating, splice site, translational start site, or nonstop mutations present in samples overnight at 4°C. The beads were then washed 1ml of icecold buffer

more than 90% of reads sequenced (>90% variant allele frequency). Hotspot B (three times), buffer B without NP40 (once), and TEAB (15mM triethylam

missense mutations assessed here were previously reported as significantly moniabicarbonate, pH 8.5) (three times). Spins between washes were performed

27

recurrent mutations in a pancancer analysis . Features present in at least three at 2,000g at 4 °C. To elute the immunopurified material, 330 μl of buffer C

−

1

cell lines were assessed. Statistical significance was determined using a twosided (1 mg ml HA peptide (Thermo Scientific), 15 mM TEAB, 0.02 % Rapigest

Student’s ttest. Q values were determined by correcting resulting P values for (Waters)) was added and the samples were incubated at 37 °C for 30 min at

multiple hypothesis testing using the Benjamini–Hochberg procedure.

1,100r.p.m. shaking in a Thermomixer (Eppendorf AG). The eluted material

Primary combination screen. A compound library comprising 589 compounds was cleared from the beads by spinning at 2,600g. The cleared material was

arrayed in ninepoint dose–response was screened in the absence or presence stored at −80°C until mass spectrometry preparation and analysis. For DUB

©

reSeArCH Letter

identifications, the eluted proteins were digested with trypsin using filteraided per time point). No randomization was used for DMPK studies. At 0.5, 1, 2, 4, 8

sample preparation. Eluents were added to Microcon30K filtration devices and 24h postdose, blood samples were collected by terminal cardiac puncture

(

Millipore) and briefly washed by 0.2ml of 8M urea in 200mM TEAB, pH 8.5. into anticoagulant tubes (EDTA). Clarified plasma was then transferred to a fresh

Between each step, liquid was cleared by centrifugation at 14,000g, except if noted. tube and snap frozen. GNE6776 plasma concentrations were determined by

Proteins were reduced by DTT for 20min at 60°C and subsequently alkylated by LC–MS/MS.

iodoacetamide for 15min in the dark. The membrane was further washed by In vivo pharmacodynamic response. For EOL1 AML xenograft studies, immu

8

M urea once and 200mM TEAB three times. Trypsin (Promega) was added nodeficient C.B-17 SCID mice (Charles River Laboratories), aged 12–16 weeks,

at 1:40 enzyme/substrate ratio. Devices were briefly centrifuged at 100g for 30s were inoculated subcutaneously on the right flank with five million cells in a 50:50

and incubated overnight at 37°C. Tryptic peptides were recovered by centrifu suspension of HBSS:Matrigel (BD Biosciences; 100μl). When tumour volumes

3

gation at 14,000g for 4min. An additional 60μl of 200mM TEAB, pH 8.5, was reached between approximately 285 and 500mm , mice were distributed into

added to the devices and centrifuged as an additional elution step. This eluent volumematched cohorts (n=4). For MCF7 breastcancer xenograft studies,

was combined with the previous ones for further processing. Ten per cent of the immunodeficient nu/nu mice (Charles River Laboratories), aged 6–8 weeks, were

eluents were dried in a SpeedVac and desalted with a C18 STAGE tip (Proxeon, implanted with 0.36mg oestrogen pellets (Innovative Research of America) via

Thermo Fisher) before being injected onto the mass spectrometer for LC–MS/MS trochar 1–3 days before tumour cell inoculation. Ten million MCF7Ser cells, an

acquisition.

in vivooptimized MCF7 variant, were injected orthotopically into the 2/3

Deubiquitinase identification by mass spectrometry. Peptides were loaded onto mammary fat pad of each mouse in a 50:50 suspension of HBSS:Matrigel (BD

an Acuity UPLC BEH130 C18 column (1.7 μm, 12 Å, 100 μm × 100 mm) at a Biosciences) in a total volume of 100μl. When tumour volumes reached between

flow rate of 1.5μl min⁻¹in solvent A (98% water/2% MeCN/0.1% formic acid) approximately 285 and 450mm , mice were distributed into volumematched

using a NanoAcquity UPLC system (Waters). Separation was achieved with a cohorts (n = 4). GNE6776 was formulated as a suspension in 0.5% methyl

3

−1

linear gradient of 2% solvent B (98% MeCN/2% water/0.1% formic acid) to 25% cellulose/0.2% Tween80 and administered at 200mgkg (body weight) by oral

solvent B over 45min. Eluted peptides were injected onto an Orbitrap Elite mass gavage at 0 and 4h. 0.5% Methylcellulose/0.2% Tween80 control or GNE6776

spectrometer (Thermo Fisher) using an Advance CaptiveSpray source (Bruker) at a treated samples were collected at 8h after the first dose and excised tumours were

voltage of 1.2kV. Full MS scans were collected in the Orbitrap at 60,000 resolution, flashfrozen on dry ice. Tumours were lysed in RIPA buffer containing protease

the top 15 most abundant ions were selected in a datadependent mode and frag inhibitors (Roche) and 300mM NaCl using a Qiagen TissueLyser. Samples were

mented with collisioninduced dissociation, and MS/MS were collected in the ion incubated on ice for 15min and then centrifuged at 20,000g at 4°C for 10min.

trap. MS/MS spectra were searched using Mascot (version 2.3.02) against a human Protein levels in clarified lysates were quantified using a Pierce BCA assay kit and

proteome database (Uniprot, December 2011) with known contaminants along concentrations were normalized with sample buffer. Samples were run on gels, and

with all decoy sequences. Search parameters included trypsin cleavage allowing proteins were transferred to membranes and western blotted as described above.

up to two missed cleavage events, a precursor ion tolerance of 50p.p.m., and a In vivo efficacy study. For EOL1 AML xenograft studies, immunodeficient

fragment ion tolerance of 0.8 Da. Searches also allowed variable modifications of C.B-17 SCID mice (Charles River Laboratories), aged 12–16 weeks, were inoculated

methionine oxidation (+15.9949 Da), and two cysteine modifications of either subcutaneously on the right flank with five million cells in a 50:50 suspension

(

+57.0215) for carbamidomethylation or (+192.0569 Da) for reacted DUB probe of HBSS:Matrigel (BD Biosciences; 100μl). When tumours became established

3

remnant. Peptide spectra matches were filtered with a false discovery rate of 5% (150–300mm ), mice were distributed into tumourvolumematched cohorts

3

on the peptide level and subsequently at 2% on the protein level. Each identified (n = 7, mean tumour volume ~250 mm ). GNE6776 was formulated as a

peptidespectrum match was quantified using areaundercurve, and the suspension in 0.5% methylcellulose/0.2% Tween80 and was administered at 100

same quantification event was extended to other runs where such peptide or 200mgkg⁻(body weight) by oral gavage on a once or twice daily schedule.

was not identified by an MS/MS spectrum, based on exact precursor m/z and Tumour volume measurements, body weight and body condition data were

1

retentiontime matching.

collected two or three times per week. The maximum tumour volume limit of

3

Statistical analysis of deubiquitinase activity data. Detected and quantified area 2,000mm was not reached in any animal.

undercurve measures from labelfree mass spectrometry were logarithmically USP7 enzymatic analysis. Michaelis–Menten kinetic measurements with full

(

base 2) rescaled. Changes between inhibitor and controltreated lysate were length USP7 were performed using 1nM USP7 with a series of ubiquitin–AMC

compared by a linear mixedeffects model with fixed effects for treatments and substrate titrations. The initial rate of substrate hydrolysis was determined using

random effects for peptide species and massspectrometry run. Computations the Magellan software on a Tecan Safire2 plate reader, and kinetic parameters

were performed in R, version 3.3.0, using the R package lme4, with error degrees modelled using nonlinear regression analysis with GraphPad Prism software.

of freedom estimated by the Kenward–Roger method using the R package pbkrtest. Standard error was calculated from three technical replicates. For studies using

For each protein, log

2

(fold change) in activity was normalized to that observed in a the USP7 D305/E308 mutant, samples were reacted in a buffer consisting of 50 mM

spikedin viral DUB or BSA control protein. Statistical significance of the observed HEPES (pH 7.5), 100mM NaCl, 2.5mM dithiothreitol, and 0.1% (w/v) bovine

‘

inhibitorvscontrol’ differences in log2(fold change) between each DUB and the gamma globulin (Sigma catalogue number G500925G). The starting substrate

control protein was then assessed by a tstatistic with Satterthwaiteestimated concentration of ubiquitinRho110 (Boston Biochem catalogue number U555)

degrees of freedom, and corrected for multiplicity by the Benjamini–Hochberg used for the Michaelis–Menten analysis was 100μM serial diluted to 781nM.

false discovery rate method set at 0.10.

Reactions were performed for 1h at room temperature with a final enzyme con

Animal use and care. All animal work followed the recommendations of the Guide centration of 100nM (three independent experiments, see symbols in plots), in

for Care and Use of Laboratory Animals with respect to restraint, husbandry, black 100μl volume 96well half area plates (Corning catalogue number 3993).

surgical procedures, feed and fluid regulation, and veterinary care. The animal care The enzymatic activity was calculated by fitting the data using the initial velocity

and use programme at Genentech is accredited by the Association for Assessment with the linear V values measured by analysing the fluorescence signal of cleaved

0

and Accreditation of Laboratory Animal Care International, which assures com Rho110 using excitation at 485nm and emission at 535nm.

pliance with accepted standards for the care and use of laboratory animals. Studies X-ray crystallography. Crystals were grown by the hangingdrop method by

were tailored to minimize the number of animals used, yet allow sufficient numbers mixing the USP7 catalytic domain (residues 208–554) at 15 mg ml⁻with an

to address any variability in drug exposure or biomarker response. For efficacy equal volume of reservoir solution containing 100mM TrisHCl, pH 7.0, and 20%

1

studies where growth rate differences introduce some variability, n= 1 0 engrafted PEG1000 (v/v). Costructures with compounds were obtained by soaking crystals

animals per treatment cohort followed recommended housing (five mice per cage) with 1mM of compound overnight. Crystals were cryoprotected with reservoir

and provided sufficient statistical power for a wellbehaved xenograft model. solution supplemented with 20% glycerol (v/v) and flashfrozen in liquid nitrogen.

For PD biomarker experiments, a lower number (n=4) was sufficient to note a Data collection and refinement statistics are detailed in Extended Data Fig. 8e.

consistent response. Extra animals were engrafted to ensure sufficient animals had Generation of isotopically labelled di-ubiquitins. In both K48 and K63linked

15

the desired range of tumour volumes at randomization, and cohorts were enrolled diubiquitin, the distal ubiquitin has isotopically labelled NThr residues and

15

that had a balanced distribution of tumour volumes. Owing to the need to monitor the proximal ubiquitin has NLeu residues. These are marked with asterisks in

potential adverse effects with first inlife assessment of novel compounds, no study Fig. 4a.

was conducted under blinded conditions.

The selectively labelled ubiquitin monomers were generated and purified

DMPK analysis. In vitro DMPK studies were performed using standard as follows: ubiquitin was cloned into a PET3a vector and transformed into the

protocols. GNE6776 was formulated as a suspension in 0.5% methylcellulose/0.2% auxotrophic strains RF2 and ML2 (a gift from R. Gennis and T. Iwasaki) and

Tween80 and was administered at 200mgkg⁻¹(body weight) by oral gavage to expressed with the following modifications. Bacterial cultures were grown at 37°C

female C.B-17 SCID mice, aged 12–16 weeks (Charles River Laboratories; n=3 in M9 medium supplemented with NH

4

Cl (2 g l⁻) and C

1

6

glucose (4 g l ). Strain

−1

©

Letter reSeArCH

15

RF2 was additionally supplemented with 6mM [ N]LThr (Cambridge Isoptopics TAMRA-peptide/tetra-ubiquitin conjugate depolymerization studies. For

15

catalogue number NLM142PK) and ML2 was supplemented with 6mM [ N] depolymerization assays, 100nM USP7 (catalogue number E519, lot 09939314)

LLeu (Cambridge Isoptopics catalogue number NLM1421) and 6mM unlabelled or 100nM USP7 catalytic domain Nterminal Histag (Genentech, Hs_USP7.

Ile and Tyr (SigmaAldrich catalogue number I2752 and catalogue number T3754). K208K554) was diluted in 1× PBS buffer (pH 7.5) containing 5mM DTT to

After reaching an absorbance of 0.6, the cells were induced with 1mM IPTG and generate 10× stock solutions in respect of the final concentration and preincubated

further grown for 6h. Protein purification was performed at room temperature. at room temperature for 10min. In a 90μl reaction, 9μg (2.7μM) of 5TAMRA

Cells were collected and lysed in lysis buffer (50mM HEPES 7.0). The cleared peptide/tetraubiquitin (K63 or K48linked) was mixed with 9 μl of diluted

lysate was subjected to affinity chromatography using DEAE Sepharose Fast Flow enzyme in 1× PBS buffer (pH 7.5). Aliquots of 10μl of the reaction were mixed

(

GE catalogue number 17070901). Ubiquitin was collected in the flow through with 4μl 2× SDS loading buffer at the time points indicated to stop the reaction.

and dialysed overnight into NaOAc, pH 4.5. Dialysed material was clarified by Samples (14μl) were subjected to SDS gelelectrophoresis using precast BioRad

centrifugation at 35,000g and ubiquitin was subjected to ion exchange chromato Criterion TGX 10–20% gels (catalogue number 5671114). Fluorescence was

graphy using a MonoS column (GE Healthcare, catalogue number 17516901). analysed using a FluorChem imager from Protein Simple according to the user’s

The following enzymes were obtained from Boston Biochem: UBE1 (catalogue manual. Densitometry values were analysed using the software AlphaViewSA,

number E305), UBE2K (catalogue number E2602) and UBE2N/UBE2V1 ProteinSimple.

complex (catalogue number E2664).

Data and code availability. The 120 cell line exome dataset for Extended Data Fig. 5

K63 and K48linked diubiquitin chains were generated and purified as follows: and Supplementary Table 1 is available as part of a larger cell line exome study

in separate reactions incubating 250nM E1 enzyme, 5μM UBE2K (K48 linked), (S. Seshagiri et al., manuscript in preparation) from the European Genome

or 5μM UBE2N/UBE2V1 complex (K63 linked) with equal molar ratios of 1mM phenome Archive under accession number EGAS00001002554. Atomic coordi

ubiquitin and 1mM ubiquitin, 10mM ATP, 50mM HEPES (pH 8.0), 10mM MgCl2 nates and structures have been deposited in the PDB under accession numbers

in a 10ml reaction at 37°C. After 2h, the reaction was acidified with 2ml of 17.4M 5UQV (USP7/GNE6640) and 5UQX (USP7/GNE6776). The USP7 NMR assign

glacial acetic acid. Obtained diubiquitins (K63 or K48linked) were purified ments in Figs 3a and 4a and Extended Data Figs 8c, 9a and 10b are deposited in

by cation exchange using a MonoS column (GE Healthcare, catalogue number the Biological Magnetic Resonance Data Bank under accession number 26766.

17516901). All purified diubiquitin chains were buffer exchanged into 1× PBS The authors declare that all other data supporting the findings of this study are

buffer and proteins were flashfrozen in liquid nitrogen before storage at −80°C. available within the paper and its Supplementary Information files. Source data

Generation of 5-TAMRA-peptide/tetra-ubiquitin conjugates. K63 and K48 are provided with the paper as Supplementary Figs 1 and 2 and Extended Data

linked tetraubiquitin chains were obtained from Boston Biochem, K63linked Fig. 4g, j linked source files, and Supplementary Tables 1 and 2. R code written

(

catalogue number UC310B), K48linked (catalogue number UC210B). The to run the linear mixed models and data analysis for endogenous deubiquitinase

TAMRA (5carboxytetramethylrhodamine) peptide was generated by CPC activity changes in Fig. 2b are available from the corresponding authors upon

5

Scientific consisting of the sequence 5TAMRAYPYDVPDYAIREIVSRNKRR request (W.F.F. et al., manuscript in preparation). R code used in Extended Data

1

9

YQEDG . K63 or K48 tetraubiquitin chains were conjugated to the peptide Fig. 5 can be accessed at https://github.com/changmt/Kategaya_et_al_USP7.

as follows. (1) Generation of tetraubiquitinMESNA: incubating 250nM E1,

1

0 mM MgCl2, 10 mM MgATP, 1 mM tetraubiquitin, and 100 mM MESNA

2

4. Yu, M. et al. A resource for cell line authentication, annotation and quality

(

SigmaAldrich, catalogue number 63705), in 20mM Na HPO at pH 8.0 at 37°C

2

4

control. Nature 520, 307–311 (2015).

overnight. This was then dialysed into 0.4% TFA, and tetraubiquitinMESNA was

lyophilized. (2) Lyophilized tetraubiquitin was dissolved in DMSO at a concen

tration of 0.5mM and 2mg of peptide added until all components were dissolved;

reaction volume 1ml. The reaction was initiated by adding NHS 27.5mM (final

25. Wertz, I. E. et al. Phosphorylation and linear ubiquitin direct A20 inhibition of

inﬂammation. Nature 528, 370–375 (2015).

26. Haverty, P. M. et al. Reproducible pharmacogenomic proﬁling of cancer cell line

panels. Nature 533, 333–337 (2016).

2

7. Chang, M. T. et al. Identifying recurrent mutations in cancer reveals widespread

lineage diversity and mutational speciﬁcity. Nat. Biotechnol. 34, 155–163

(2016).

concentration), AgNO

bated at room temperature overnight. The reaction was diluted ten times with

doubledistilled H O and desalted into PBS (pH 7.5). Nonconjugated peptide

3

3.3mM (final concentration), and 22μl DIPEA, and incu

28. Lehár, J. et al. Synergistic drug combinations tend to improve therapeutically

relevant selectivity. Nat. Biotechnol. 27, 659–666 (2009).

2

was removed by sizeexclusion chromatography. In a second step, nonconjugated

tetraubiquitin was removed by ion exchange chromatography as described above

and buffer exchanged into 1× PBS (pH 7.5). The concentration of the purified

final conjugate was determined by absorbance using an extinction coefficient for

2

9. Bueno, R. et al. Comprehensive genomic analysis of malignant pleural

mesothelioma identiﬁes recurrent mutations, gene fusions and splicing

alterations. Nat. Genet. 48, 407–416 (2016).

30. Reverdy, C. et al. Discovery of speciﬁc inhibitors of human USP7/HAUSP

deubiquitinating enzyme. Chem. Biol. 19, 467–477 (2012).

−1

5

TAMRA at 80,000cm mol .

©

reSeArCH Letter

a.

b.

Primary Screen

STD NMR with

Primary Screen

1.37 million

4

871

UbA10 TR-FRET

library compounds

library fragments

inhibition 40 – 150%

5-compound mixtures

normalized F615 -170 – 80%

primary structural filters

S/N >5

IC50 confirmation

primary confirmation

STD NMR with

UbA10 TR-FRET

USP7 diUb FRET

Ub-Rho110

3064

810

initial hits

preliminary hits

single compounds

confirmed IC50 in ≥2 assays

S/N >10

no Rho-110 interference

Hill slope < 2 in UbA10 TR-FRET

secondary confirmation 478

secondary confirmation 1315

titration with labeled USP7 confirmed binders

di-Ub cleavage assay preliminary actives

chemical shift

perturbation clusters

IC50 < 50µM

hit-to-lead candidates

8

3 crystallography

1

01

X-ray crystallography

crystal structure solved

hit-to-lead candidates

candidates

confirmed actives

Indole tricyclic

GNE-0300 GNE-8735

Peptidomimetic Tetrahydroacridine

Fragment

c.

Compound

GNE-3086 GNE-2090 GNE-6831 GNE-2916

Full Length

1

.03

0.47±0.3ꢀ

0.50±0.ꢀ1

>200

8.80±5.ꢁ0

1ꢀ.80

4.64±1.38

41.7±ꢀ0.6

>200

0.75±1.30

>200

ꢁ.63±0.43

1.40±0.18

>200

USP7 IC50(µM)

USP7 Catalytic

GNE-0300

GNE-8735

1.40

Domain IC50 (µM)

USP7 inhibition

and selectivity

Full Length

1

60

40

>100

N/D

USP5 IC50(µM)

GNE-3086

Full Length

1

>200

7.0

6ꢀ.ꢀ

>200

USP47 IC50(µM)

Total MDM2

none

increase

2.9µM

none

increase (IF)

>20µM

cellular MDM2

assays

Ub-MDM2

MSD (µM)

not done

>50.00

9.65±ꢀ.18

ꢀ.55±1.91

ꢀ.20±2.69

7.60±0.57

Cathepsin-B

>

40.00

6.12

not done

pass

not done

pass

> 40.00

not done

pass

not done

Inhibition (µM)

protease

GNE-2090

GNE-6831

selectivity

Caspase-ꢀ

apparent

Inhibition

precipitator

Dynamic Light

Scattering

GNE-2916

pass

biophysical

analysis

Covalent

not done

yes

not done

yes

Modiﬁcation

d.

e.

compound

ratio Ub-MDM2/total MDM2

Ub-MDM2

total MDM2

and EC50

6

4

2

00

0

600

400

200

0

600

400

200

0

GNE-8735

EC >50µM

vehicle

GNE-8735

50

-

6

-5

10^-⁴

10

-6

10

-5

10^-⁴

-6

-5

10^-⁴

1

0

10

1

8

6

4

2

00

concentration [M]

0

GNE-8735

6

4

00

600

400

200

0

600

400

200

0

GNE-3086

200

0

^E^C50 = 9.65±3.18µM

GNE-2916

-5

0

1 10^-

6

[

1 10

-4

10-6

10-5

10-4

10-6

10-5

10-4

10-6

10-5

10-4

GNE-2916

compound] (M)

concentration [M]

6

4

2

00

0

600

400

200

0

600

400

200

0

GNE-2090

EC₅₀= 3.55±1.91µM

-

6

-5

₁₀-4

-6

-5

₁₀-4

10-6

10-5

10-4

1

0

10

concentration [M]

6

4

2

00

0

600

400

200

0

600

400

200

0

GNE-6831

EC₅₀= 3.2±2.69µM

-

6

-5

10^-⁴

-6

-5

10^-⁴

-6

-5

10^-⁴

1

0

10

concentration [M]

6

4

00

600

400

200

0

600

400

200

0

GNE-2916

EC50 = 7.6±0.57µM

200

0

-

6

-5

10^-⁴

-6

-5

10^-⁴

-6

-5

10^-⁴

1

0

10

concentration [M]

Extended Data Figure 1 | See next page for caption.

©

Letter reSeArCH

Extended Data Figure 1 | Screening cascades and hit-to-lead assay

data for USP7 inhibitors. a, Highthroughput activitybased screening

cascade to identify USP7 inhibitors. Screening stages are identified in

bold print. Numbers of compounds at each stage are provided to the right

of each box. Criteria for progression to the next stage are highlighted

in italics to the left of each arrow. b, Fragment NMR screen diagram.

Screening stages are identified in bold print. Numbers of compounds at

each stage are provided to the right of each box. Criteria for progression

to the next stage are highlighted in italics to the left of each arrow.

Protein saturation transfer difference (STD) experiments were performed

at 284 K. Primary USP7 catalytic domain binders were selected on the

basis of the signaltonoise ratio of the respective compound with a cut

off of greater than 5. All primary binders were remeasured as single

compounds under otherwise identical conditions and confirmed binders

selected having a signaltonoise ratio of greater than 10. Hits were further

tested for specific binding to USP7 catalytic domain by measurement

of H/ N TROSY spectra. Positive hits were defined as compounds that

induced chemical shift perturbations. Perturbations were classified by

the chemical shift patterns and selected compounds passed onto Xray

soaking experiments. c, Table summarizing the hittolead assay results

of the lead compounds identified by the highthroughput screening and

NMR fragment screening campaigns. Compound series are grouped in

columns and hittolead selection assay data are listed in the indicated

rows. At least two experimental replicates were performed;

averages are shown ± s.e.m. See Fig. 1b and text for more details.

d, Endogenous MDM2 immunofluorescence studies. HCT116 human

colorectal carcinoma cells were treated with a range of concentrations of

the indicated USP7 inhibitors or DMSO vehicle for 24 h and endogenous

MDM2 protein levels were detected by immunofluorescence imaging.

Representative images show cells treated with 10 μM of the indicated

compounds or DMSO vehicle control. Scale bar, 20 μm. The graph shows

data from experimental duplicates of the quantified mean nuclear MDM2

protein levels per cell over a range of concentrations of GNE8735 and

GNE2916. The halfmaximal effective concentration (EC50) for the

elevation in MDM2 caused by GNE8735 was 2.9 μM. Image source data:

Supplementary Fig. 1. e, Quantification of total and ubiquitinatedMDM2

(UbMDM2) in USP7 inhibitortreated cells. SJSA1 human osteosarcoma

cells were treated with a range of concentrations of the indicated USP7

inhibitors or DMSO vehicle control for 24 h and the level of ubiquitinated

MDM2 and total MDM2 were measured using a multiplexed mesoscale

immunoassay. Graphs show experimental duplicates of the quantified level

of either total MDM2 (right column, red), ubiquitinated MDM2 (central

column, blue), or the percentage change in the ratio of the ubiquitinated

MDM2 signal and the total MDM2 signal (left column, orange). All data

are shown as percentage change in each value, relative to DMSO vehicle

treated samples. The maximal extent of the increase in the ubiquitinated

MDM2 to total MDM2 ratio varied between compounds; therefore to

compare the potency of the increase in the ratio of ubiquitinatedMDM2

to total MDM2 between compounds, the top level was universally set

to 100%. This normalization allowed calculation of the EC50 of the

percentage change in this ratio relative to DMSO (left column). Indole

tricyclic compounds including GNE8735 increased total MDM2 levels

and inhibited cathepsinB, indicating poor selectivity and induction of

general cell stress by this chemical series (d, e). Indole tricyclics also

precipitated caspase3, although they passed dynamic light scattering

analysis (see also Extended Data Fig. 2a). The peptidomimetic compounds

had weaker biochemical potency, poor selectivity, and covalently modified

USP7 cysteine (Cys) residues other than the catalytic Cys (data not

shown). Given these data, and because optimization of indole tricyclic

and peptidomimetic compounds proved challenging, these series were

discontinued. The tetrahydroacridine and fragment compounds were

relatively potent, selective, and enhanced cellular MDM2 ubiquitination

without significantly increasing total MDM2 (d, e). Tetrahydroacridine

compounds passed cathepsinB inhibition assays, demonstrating

additional protease selectivity. Neither tetrahydroacridine nor fragment

compounds showed evidence of USP7 aggregation in dynamic light

scattering or in NMR studies (data not shown; see Extended Data

Fig. 2a). Tetrahydroacridine compounds, including GNE6831, covalently

modified USP7, consistent with a previous report describing a similar

series³⁰(see also Extended Data Fig. 2b).

1

15

©

reSeArCH Letter

a. 1.25

b.

USP7

full length 129366.66

+

DMSO

1

.20

.15

1

29200 129400 129600 129800

USP7

129359.97

full length

1

29790.18

1.10

+ GNE-6831

4

30.2

Rottlerin = 100% aggregate

USP7 + GNE-8735 = 16%

USP7 + GNE-2090 = 47%

USP7 + DMSO = 23%

1

.05

.00

129200 129400 129600 129800

Mass (Da)

10.0

1.0E+3

1.0E+5

1.0E+7

time (microseconds)

c.

active compounds inactive control compounds

GNE-6831 GNE-2090 GNE-2148 GNE-2143 GNE-0956

Compound

Full Length

0.75±0.13 ꢀ.6ꢀ±1.38

>200

>20

196

>200

N/D

>20

USP7 IC (µM)

n = 4

n = 2

GNE-6831

GNE-2090

GNE-2143

GNE-2148

50

USP7 Catalytic

>100

ꢀ1.7±30.6

>63

Domain IC (µM)

n = 2

50

Full Length

>200

>63

USPꢀ7 IC (µM)

n = 3

50

Full Length

N/D

>200

USP5 IC (µM)

50

Ub-MDM2

30.8±11.1 2ꢀ.2±12.3

N/D

MSD (µM)

n = 5

n = 6

GNE-0956

active compound

inactive control compounds

d.

Compound

GNE-2916 GNE-2931 GNE-9603 GNE-2917

Full Length

2.63±0.ꢀ3

>200

109

11ꢀ

USP7 IC (µM)

n = 2

5

0

USP7 Catalytic 1.ꢀ0±0.1ꢀ

N/D

66.8

GNE-2916

GNE-2931

GNE-2917

Domain IC (µM)

n = 2

>200

50

Full Length

>63.3

>200

>20

USPꢀ7 IC (µM)

n = 2

50

Full Length

>200

N/D

>200

N/D

USP5 IC (µM)

50

Ub-MDM2

7.60±0.57

N/D

GNE-9603

MSD (µM)

n = 2

e.

AMO-1 cells

f.

KMS12-PE cells

1

25

00

125

GNE-6831

100

75

50

25

0

GNE-2090

GNE-0956

7

5

2

5

0

5

0

GNE-21ꢀ8

GNE-21ꢀ3

GNE-2090

1

5

10

2

10

20

concentration (µM)

1

25

00

125

100

75

50

25

0

GNE-2917

GNE-2916

7

5

2

5

0

5

0

GNE-2931

GNE-9603

GNE-2916

5

10

2

10

20

concentration (µM)

Extended Data Figure 2 | See next page for caption.

©

Letter reSeArCH

Extended Data Figure 2 | Biophysical and cellular characterization of

selected USP7 inhibitors. a, The dynamic light scattering autocorrelation

functions are shown for 100 μM rottlerin (red), fulllength USP7

with 100 μM GNE8735 (blue), fulllength USP7 with 100 μM GNE

n =1 experimental replicate, if not specified. See Fig. 1b and text for more

details. e, Cell viability assays in AMO1 cells treated as indicated with the

tetrahydroacridine compounds (top graph: purple lines, inactive controls;

green line, active compound) and fragment compounds (bottom graph:

red lines, inactive controls; blue line, active compound). Experimental

triplicate data normalized to vehicle control are plotted as a function

of compound concentration. f, Cell viability assays in KMS12PE cells

treated as indicated with the tetrahydroacridine compounds (top graph:

2

090 (brown), and fulllength USP7 with 0.1% DMSO vehicle control

(

black). The percentage aggregate by mass is shown in the legend. Two

experimental replicates were performed. b, USP7 fulllength protein was

incubated overnight with excess of GNE6831 and analysed by liquid

chromatography–mass spectrometry. Unmodified and covalently modified purple line, inactive control; green lines, active compounds) and fragment

USP7 are represented in the top and bottom panels, respectively. Three

experimental replicates were performed. c, Table of active and inactive

tetrahydroacridine compounds with structures. Hittolead selection assay

data are listed in the indicated rows. Averages are shown ± s.e.m., where

applicable, with the number of biologically independent experiments

indicated below. n =1 experimental replicate, if not specified. See

Fig. 1b and text for more details. d, Table of active and inactive fragment

compounds with structures. Hittolead selection assay data are listed in

the indicated rows. Averages are shown ± s.e.m., where applicable, with

the number of biologically independent experiments indicated below.

compounds (bottom graph: red line, inactive control; blue line, active

compound). Experimental triplicate data normalized to vehicle control

are plotted as a function of compound concentration. Tetrahydroacridine

compounds GNE6831 and GNE2090 decreased viability of KMS12PE

and AMO1 multiple myeloma cell lines but this activity was not

differentiated from control compounds GNE0956, GNE2143, and

GNE2148. By contrast, the fragment compound GNE2916 decreased

multiple myeloma cell viability significantly more than control compounds

GNE2917, GNE2931, and GNE9603. Thus work on tetrahydroacridine

series was discontinued and the fragment series was further optimized.

©

reSeArCH Letter

active compounds

inactive control compounds

a.

Compound

GNE-2916 GNE-6640 GNE-6776 GNE-2118 GNE-6641

Full Length

2.63±0.43 0.75±0.37 1.34±0.24

n = 2 n = 5 n = 5

>63.3

n = 2

>63.3

n = 2

USP7 IC (µM)

50

GNE-6640

USP7 Catalytic 1.40±0.14 0.43±0.07 0.61±0.15

>200

n = 2

>63.3

n = 2

Domain IC (µM)

n = 2

n = 3

n= 3

50

Full Length

>200

20.3±2.2

>200

GNE-2916

GNE-6776

GNE-6641

USP47 IC50(µM)

n = 2

n = 3

n= 2

Full Length

>200

n = 2

>200

n = 4

>200

n = 3

>200

n = 2

USP5 IC (µM)

50

Ub-MDM2

MSD (µM)

7.60±0.57 0.23±0.31 2.58±1.10

n = 2 n = 2 n = 2

>50

n = 2

>50

GNE-2118

WB: USP7

WB: p53

WB: p21

b.

c.

CHX (hours):

0

0.5

1

3

5

7

0

0.5

1

3

5

7

0

0.5

1

3

5

7

1

50-

DMSO

GNE-6641

GNE-6640

GNE-6776

5

0

-

20-

50-

0-

wild-type MCF-7

d. 120

e. 120

100

*

80

**

80

*

–

2

1

50-

**

*

P<0.05

60

**

6

0

*

*P<0.01

*

****

*

4

0

*

P<0.05

**

Ub-

MDM2

40

48-

**P<0.01

IP: K48-Ub

WB: MDM2

USP7 null HCT-116 (GNE-6641)

wild-type HCT-116 (GNE-6641)

USP7 null HCT-116 (GNE-6640)

wild-type HCT-116 (GNE-6640)

20

0

USP7 null HCT-116 (GNE-6776)

wild-type HCT-116 (GNE-6776)

9

8-

0

2

3

4

5

6 7 8 910

20

1

2

3

4

5 6 7 89 10

20

concentration (µM)

6

4-

1

20

00

120

100

WB: MDM2

1

48-

WB: USP7

WB: tubulin

80

60

40

20

0

8

0

5

0-

6

4

2

USP7 null HCT-116 (GNE-6641)

wild-type HCT-116 (GNE-6641)

USP7 null HCT-116 (GNE-6640)

wild-type HCT-116 (GNE-6640)

USP7 null HCT-116 (GNE-6776)

wild-type HCT-116 (GNE-6776)

0

2

3

4

5

6 7 8 910

20

1

2

3

4

5 6 7 89 10

20

concentration (µM)

Extended Data Figure 3 | Deubiquitinase inhibition and cellular

activity of optimized fragment compounds. a, Table summarizing

deubiquitinase biochemical assay data and ubiquitin–MDM2 assay data

from optimized fragment compounds and inactive controls. Averages

are shown ± s.e.m, where applicable, with the number of biologically

independent experiments indicated below. n =1 experimental replicate,

if not specified. Fragment compound structures are shown to the right.

b, Representative western blot analysis of USP7, p53, and p21 levels from

the cycloheximide (CHX)chase study of MDM2 turnover shown in

Fig. 1c, performed in experimental triplicate. c. Analysis of endogenous

MDM2 polyubiquitinated with K48linked chains. Top: denatured lysates

from MCF7 cells treated for 8 h with the indicated compounds were

immunoprecipitated with a K48 polyubiquitin linkagespecific antibody

and immunocomplexes were western blotted with an antiMDM2

antibody. Western blot analysis of wholecell lysates for the indicated

proteins are shown below. Representative of two experimental replicates.

d, Cell viability of wildtype and USP7null HCT116 colon

adenocarcinoma cells, treated as indicated, and analysed with a

CellTiterGlo assay. Experimental triplicates normalized to vehicle

control are plotted as a function of compound concentration. Top:

averages ± s.e.m. Twosided ttests were used to calculate P values

between wildtype HCT116 and USP7null cells treated with GNE6640:

7.5 μM P =0.01, 10 μM P =0.041, 12.5 μM P =0.009, 15 μM P =0.011,

20 μM P =0.017. Bottom: experimental triplicate data are shown. e. Cell

viability of wildtype and USP7null HCT116 colon adenocarcinoma cells,

treated and analysed as in d with the indicated doses of GNE6776. Top:

averages are shown ± s.e.m. Twosided ttests were used to calculate P

values between wildtype HCT116 and USP7null cells: 1 μM P =0.023,

2.5 μM P =0.003, 5 μM P =0.001, 7.5 μM P =0.003, 10 μM P =0.007,

12.5 μM P =0.001, 15 μM P =0.007, 17.5 μM P =0.001, 20 μM P =0.008.

Bottom: experimental triplicate data are shown. Gel source data:

Supplementary Fig. 1.

©

Letter reSeArCH

a.

Experiment # 1

Experiment # 2

Experiment # 3

b.

c.

GNE-6641 GNE-6640

5

4

3

2

1

0

µM compound:

HA-Ub-VS:

0

–

0

+

1.7

5

+

15 1.7

5

+

15

+

0

–

100

GNE-6640 (100 µM)

+

148-

WB: USP7

**

*

**

*

50

WB: USP47

K11 Ub2

IC50 = 2.55 ± 0.14 µM

.1 10 100 1000

148-

WB: USP5

**

0

1

0

9

8-

*

1

20

00

8

6

4

2

**

0

WB: UCHL1

WB: actin

36-

50-

100

GNE-6641 (100 µM )

*

K33 Ub2

IC50 = 2.90 ± 0.39 µM

.1 10 100 1000

50

0

1

8

6

4

2

0

2

1

50-

48-

0

K48 Ub2

IC50 = 2.44 ± 0.47 µM

.1 10 100 1000

1

00

0

1

GNE-6776 (100 µM )

98-

2

1

0

6

2

8

4

0

WB: HA

50

64-

50-

36-

K63 Ub2

IC50 = 2.08 ± 0.21 µM

.1 10 100 1000

GNE-6776] (µM)

0

1

[

0

.1%

10µM

15µM

30µM

0.1%

d.

e.

DMSO GNE-6776 GNE-6776 GNE-6776 GNE-6641 DMSO

GNE-6640

GNE-6776

time (min):

HA-Ub-VS:

0

–

15 30 45 15 30 45 15 30 45 15 30 45 45

MW/LogD7.4/ tPSA

LM CLhep (h/r/m/d/c)

Hep CLhep (h/r/m/d/c)

PPB % (h/r/m)

330.38/4.40/87.0

14/46/75/24/33

16/45/68/24/39

99.6/98.7/99.0

348.40/3.30/3.4/101

4.1/15/37/14/12

9.4/39/44/18/33

97.3/93.4/93.4

+

1

48-

WB: USP7

**

*

WB: USP47

**

*

1

48-

WB: USP5

MDCK Pappratio (B to A, A to B) 0.83 (12.17, 14.7)

*

0.75 (12.06, 16.10)

9

3

8-

6-

*

**

WB: UCHL1

WB: GAPdH

*

3

6-

f. 150

EOL-1 IC₅₀= 1.54µM

g. 100

100

10

1

0

1

00

1

0

.0

.1

1.0

0.1

5

0

GNE-6776 EOL-1 target exposure

200 mg/kg GNE-6776

100 mg/kg GNE-6776

2

00 mg/kg GNE-6776

1

0

0.01

-

8

-7

-6

-5

0

4

8

12

16

20

24

0

4

8

12

16

20

24

[

GNE-6776] log₁₀(M)

time (hours)

2

000

500

000

500

0

j.

*

1

vehicle

2

GNE-6776

vehicle

2

GNE-6776

h. MCF7

i.

EOL-1

xenograft:

1

3

1

2

3

1

3

1

2

3

WB: p53

WB: p53 50-

5

2

0-

2-

2

3

2-

8-

WB: p21

Vehicle (MCT), PO QD

GNE-6776, 200mg/kg PO QD

GNE-6776, 100mg/kg PO BID

full length

caspase-3

WB: MDM2

WB: actin

6

5

4-

0-

28-

WB:

caspase-3

0

2

4

6

8

10

Day

2000

1

7-

4-

cleaved

caspase-3

1

500

000

WB: MDM2

WB: actin

6

5

4-

0-

5

00

0

Vehicle (MCT), PO QD

GNE-6776, 200mg/kg PO QD

GNE-6776, 100mg/kg PO BID

0

2

4

6

8

10

Day

Extended Data Figure 4 | See next page for caption.

©

reSeArCH Letter

Extended Data Figure 4 | Selectivity of USP7 inhibitors, in vitro and

in vivo drug metabolism and pharmacokinetic (DMPK) profiling,

and xenograft growth inhibition studies with GNE-6776. a, GNE6776

dosedependent cleavage inhibition by USP7 of the indicated diubiquitin

chains as measured by MALDI–TOF. Data from experimental triplicates

are shown. b, Percentage inhibition of the indicated deubiquitinases for

cleaving diubiquitins after treatment with 100 μM of the indicated USP7

inhibitor compounds, representative of one experimental duplicate.

Deubiquitinase concentrations and diubiquitin substrates are as in

Fig. 2a. c. Supporting western blots for Fig. 2b, left. HEK293T cell

lysates were treated with the indicated USP7 inhibitors (0.1% DMSO

control =0 μM compound) and endogenous deubiquitinases were reacted

with the HA–ubiquitin–vinylsulfone activitybased probe (HA–Ub–VS).

Reacted cell lysates were immunoblotted with the indicated antibodies.

to basolateral (A to B) directions. f, EOL1 cell line viability in response

to GNE6776 as measured in a 5day CellTiterGlo assay performed in

experimental triplicates. g, In vivo pharmacokinetic analysis of GNE6776.

−

1

Mice (three per group) were dosed PO with 100 mg kg (body weight) or

−

1

200 mg kg (body weight) of GNE6776. Plasma concentrations of GNE

6776 were measured at the indicated time points and the mean ± s.d. (left)

or individual data points (right) plotted as a function of time. Exposure

metrics relating to the free fraction EC50 for EOL1 cells are also indicated,

where target exposure =(EOL1 IC50)/(1 − percentage plasma protein

binding) =1.54 μM/0.066 =23.33 μM. h, Western blot analysis of MCF7

Ser xenografted tumours. Mice bearing MCF7Ser xenograft tumours were

−

1

were dosed by mouth with vehicle or 200 mg kg (body weight) GNE

6776 at 0 and 4 h; 8 h after the initial treatment, tumours were excised and

the indicated proteins were examined by immunoblotting tumour lysates.

Data show three biological replicates. i, Western blot analysis of EOL1

xenografted tumours. Mice bearing EOL1 xenograft tumours were dosed

*

Unreacted deubiquitinases; **probereacted deubiquitinases; arrowhead

points to a band identified by antiHA immunoblotting that runs at

the expected molecular mass of USP7 and is diminished in lysates

treated with GNE6640. Representative of one experimental duplicate.

d, Supporting western blots for Fig. 2b, right. HEK293T cell lysates were

treated with the indicated USP7 inhibitors, endogenous deubiquitinases

were reacted with the HA–Ub–VS activitybased probe, and reacted cell

lysates were immunoblotted with the indicated antibodies. *Unreacted

deubiquitinases; **probereacted deubiquitinases. Representative of one

experimental duplicate. e, In vitro pharmacokinetic assessment of USP7

inhibitors. Calculated drug properties are indicated: molecular mass

−

1

by mouth with vehicle or 200 mg kg (body weight) GNE6776 at 0 and

4 h; 8 h after the initial treatment, tumours were excised and the indicated

proteins were examined by immunoblotting tumour lysates. Data

show three biological replicates. j, EOL1 xenograft growth inhibition

study of mice administered vehicle or the indicated doses of GNE6776

by mouth with n =7 mice per group. The mean ± s.e.m. (top) or

individual data points (bottom) are plotted as a function of time. The

P values were calculated using Dunnett’s multiple comparison test.

Asterisks indicate significant growth inhibition relative to vehicletreated

−

1

−1

(

MW), lipophilicity at pH 7.4 (logD7.4), total polar surface area (tPSA),

mice. Day 4 100 mg kg (body weight) P =0.0163, day 4 200 mg kg

−1

stability in hepatic microsomes (LM CLhep), or hepatocytes (Hep CLhep)

from human/rat/mouse/dog/cynomolgus monkey (h/r/m/d/c) species,

percentage plasma protein binding (PPB %), and permeability across

an MDCK cell monolayer from basolateral to apical (B to A) or apical

(body weight) P =0.0138, day 6 200 mg kg (body weight) P =0.0344.

Graph source data for g and j can be accessed in the Supplementary

Information. Gel source data: Supplementary Fig. 1.

©

Letter reSeArCH

a.

primary screen

secondary screen

pharmacogenomic

analysis

mean viability

cancer type

4

41 cell lines

181 cell lines

6 indications

5-day viability

cell histology

loss-of-function

mutations

15 indications

-day viability

3

hotspot mutations

b. 200

200

150

100

50

200

GNE-6640

GNE-6776

GNE-6641

181

156

150

100

50

108

100

2

5

50

3 19

12

1

7

3

6

1

1 1

2

1 1

0

Not determined 4.5 5.0 5.5 6.0 6.5 Not determined 4.5 5.0 5.5 6.0 6.5 Not determined 4.5 5.0 5.5 6.0 6.5

Activity (-log₁₀(IC Molar))

50

Activity (-log₁₀(IC Molar))

50

c.

GNE-6640

d.

GNE-6640

TP53 LOF TP53 WT

GNE-6776

Number of samples

●

TP53 LOF

TP53 WT

●

5

Blood | AML

●

10

0.2

0.1

0.0

0.1

0.2

−0.3

−0.4

0.2

0.1

3

2

1

0

.0

2

0

0.2

.0

0.2

0.0

●

50

0

0.0

TP53 Y220

●

−

−0.1

−0.2

−0.3

−0.4

−

0.2

0.4

−0.2

−0.4

−0.6

●

●●

●● TP53R175

●

●● ●

●

●●

●

● ●

−0.6

●

●●

●

●●●

●

●●

●

●●

●

●●

●

● ●● ●

TRUE FALSE TRUE FALSE

●

●●●

●

n=20 n=156 n=80 n=96

−

0.3 −0.2 −0.1 0.0 0.1 0.2 0.3

P=0.164

q=0.676

P=0.466

q=N/D

P=0.401

q=0.772

P=0.100

q=N/D

Normalized Mean Viability Ratio

(

log fold change)

95%CI: [-0.12, 0.02] 95%CI: [-0.03, 0.06]

95%CI: [-0.06, 0.02]

95%CI: [0, 0.05]

2

Less sensitivity

More sensitivity

with feature

e.

GNE-6640

TP53 Y220

GNE-6776

TP53 Y220

AML

TP53 R175

AML

TP53 R175

0

0.0

0.1

0.2

0.3

0.4

.2

.1

0.2

0.1

0.0

−0.1

−0.2

−0.3

−0.4

0.2

0.1

0.0

−0.1

−0.2

−0.3

−0.4

0

.2

0.2

0.0

0.2

0.0

−

0.2

0.4

0.6

−0.2

−0.4

−0.6

−0.2

−0.4

−0.6

TRUE FALSE TRUE FALSE

TRUE FALSE

n=5 n=171

P=0.215

TRUE FALSE TRUE FALSE TRUE FALSE

n=8 n=168

P=0.00028

q=0.019

n=3 n=173

P=0.012

n=8 n=168

P=0.114

n=3 n=173

P=0.182

n=5 n=171

P=0.312

q=0.279

q=0.677

q=0.750

q=0.772

9

5%CI: [0.18, 0.39] 95%CI: [-0.32, -0.10] 95%CI: [-0.11, 0.31]

95%CI: [-0.03, 0.22] 95%CI: [-0.23, 0.07] 95%CI: [-0.09, 0.20]

Extended Data Figure 5 | Bioinformatics analysis of USP7 inhibitor

identified hotspot mutations in TP53 (ref. 27). The number of samples,

minima, 25th percentile, centre, 75th percentile, and maxima for each

boxplot are as follows: GNE6640 TP53 LOF TRUE (20, −0.261, −0.119,

−0.071, 0.017, 0.304), GNE6640 TP53 LOF FALSE (156, −0.623,0.186,

−0.084, −0.008, 0.294), GNE6640 TP53 WT TRUE(80, −0.623, −0.182,

−0.085, −0.014, 0.293), GNE6640 TP53 WT FALSE (96, −0.425, −0.190,

−0.083, 0.012, 0.304), GNE6776 TP53 LOF TRUE (20, −0.166, −0.082,

−0.003, 0.029, 0.105), GNE6776 TP53 LOF FALSE(156, −0.434, −0.087,

−0.032, 0.015, 0.225), GNE6776 TP53 WT TRUE (80, −0.434, −0.098,

−0.036, 0.008, 0.225), GNE6776 TP53 WT FALSE(96, −0.304, −0.077,

−0.021, 0.023, 0.113), GNE6640 AMl TRUE(8, −0.623, −0.394, −0.340,

−0.271, −0.222), GNE6640 AMl FALSE(168, −0.488, −0.173, −0.079,

0.001, 0.304), GNE6640 TP53 Y220 TRUE(3, 0.070, 0.091, 0.112, 0.142,

0.171), GNE6640 TP53 Y220 FALSE(173, −0.623, −0.184, −0.085,

−0.010, 0.304), GNE6640 TP53 R175 TRUE(5, −0.425, −0.261, −0.155,

−0.119, 0.028), GNE6640 TP53 R175 FALSE(171, −0.623, −0.183,

−0.082, −0.004, 0.304), GNE6776 AMl TRUE(8, −0.434, −0.216,

−0.061, −0.036, −0.009), GNE6776 AMl FALSE(168, −0.363, −0.085,

−0.030, 0.021, 0.225), GNE6776 TP53 Y220 TRUE(3, −0.033, 0.010, 0.052,

0.070, 0.088), GNE6776 TP53 Y220 FALSE(173, −0.434, −0.087, −0.031,

0.018, 0.225), GNE6776 TP53 R175 TRUE(5, −0.304, −0.081, −0.040,

−0.034, −0.012), GNE6776 TP53 R175 FALSE(171, −0.434, −0.087,

−0.031, 0.019, 0.225). See Supplementary Table 1 for additional information.

cell viability screens. a, Schematic of the cellular viability assay workflow

and bioinformatics analysis. The six tumour cell line indications included

leukaemias, lymphomas, lung carcinomas, and breast, colon, and prostate

adenocarcinomas. b, Histogram of IC50 values of GNE6640, GNE6446,

and GNE6641 in 181 cell lines. Mean viability is calculated as the

arithmetic average of the fitted viabilities at each tested dose of GNE6640

or GNE6446 normalized to the mean viability of GNE6641. See also

Supplementary Table 1. c, Univariate analysis of features associated

with viability differences. The x axis represents the fold change (log2) in

normalized mean viability between cell lines present or absent for a given

feature. The y axis represents the nominal P value (−log10 scale). Features

with q values less than 0.05 and absolute log2(fold change) greater than 0.1

are coloured in red. Features with only absolute log2(fold change) greater

than 0.1 are coloured in gold. Other analysed features that did not reach

significance are indicated in grey. P values were determined using the two

sided Student’s ttest, and q values were determined by correcting resulting

P values for multiple hypothesis testing using the Benjamini–Hochberg

approach. The size of each point corresponds to the number of cell lines

present with the feature. d, e, Boxplots of selected features and their

respective associations with normalized mean viability. The respective

P and q values are indicated below and were calculated as outlined in the

Methods. ‘TP53 LOF’ indicates lossoffunction mutations and previously

©

reSeArCH Letter

a.

8

6

4

2

HCT-116 (WT p53)

HCT-116 (p53 null)

1

00

GNE-6641

100

80

60

40

20

0

GNE-6640

100

80

60

40

20

0

GNE-6776

80

GNE-6641

80

60

40

20

0

GNE-6640

80

60

40

20

0

GNE-6776

0

60

40

20

0

DMSO

5

µM

1

0µM

15µM

0

10 20 30 40 50 60 70

0 10 20 30 40 50 60 70

5

.0

0

5.0

4.0

3.0

2.0

1.0

0

5.0

4.0

3.0

2.0

1.0

0

5.0

4.0

3.0

2.0

1.0

0

5.0

4.0

3.0

2.0

1.0

0

5.0

4.0

3.0

2.0

1.0

0

4

3

2

1

10 20 30 40 50 60 70

0 10 20 30 40 50 60 70

time (hours)

c.

U2OS (WT p53)

SaOS (p53 null)

b.

8

6

4

2

MCF7 (WT p53)

MDA-MB157 (p53 null)

1

00

GNE-6641

100

80

60

40

20

0

GNE-6640

70

GNE-6641

70

GNE-6640

1

00

GNE-6641

100

80

60

40

20

0

GNE-6640

100

GNE-6641

100

80

60

40

20

0

GNE-6640

6

0

60

DMSO

5µM

8

6

4

2

0

80

60

40

20

0

5

50

40

30

20

10

0

7

.5µM

40

30

20

10

0

15µM

DMSO

µM

7.5µM

5µM

5

1

0

10203040506070 0 10203040506070

0.8

0 10203040506070 0 10203040506070

0

10203040506070 0 10203040506070

2.0

0 10203040506070 0 10203040506070

2

.0

.5

2.0

1.5

2.0

0.8

0.6

0.4

0.2

0

3.0

1

0

1.5

1.0

0.5

1.5

1.0

0.5

0.6

0.4

0.2

0

2

.0

2.0

1.0

.0

.5

0

1.0

0.5

0

1

0

10203040506070 0 10203040506070

0 10203040506070 0 10203040506070

time (hours)

d.

MCF7 (WT p53)

e.

U2OS (WT p53)

7

0

GNE-6641

70

GNE-6640

100

GNE-6641

100

80

60

40

20

0

GNE-6640

6

0

60

80

5

0

50

40

4

60

3

2

1

30

20

10

0

4

2

0

10203040506070 0 10203040506070

0 10203040506070

4

3

2

1

.0

0

4.0

2.0

1.5

1.0

0.5

0

2.0

1.5

1.0

0.5

0

3.0

2.0

1.0

0

10203040506070 0 10203040506070

0 10203040506070

time (hours)

0 10203040506070

time (hours)

DMSO

doxorubicin

cisplatin

GNE-6641

GNE-6640

GNE-6641

GNE-6640

GNE-6641 + doxorubicin

GNE-6640 + doxorubicin

GNE-6641 + cisplatin

GNE-6640 + cisplatin

3

2

.0-

.5-

PI3K/PIM pathways

RTK/MAPK pathways

epigenetic regulators

DNA damage

f.

g.

Transporter

Bromodomain

Protein Deacetylase Ubiquitination

Aminopeptidase GPCR Histone Lysine

Protein Arginine

Tx

Demethylase

Methyltransferase

factor

Undefined

2.0

1.5

1.0

0.5

Phospholipase

GEF GTPase

Histone Lysine

Protease

DNA

Methyltransferase

Phosphodiesterase

Phosphatase

Ion Channel Kinase

(

Lipid)

Other

non-

Golgi

Antioxidant

Mitochondrial

Decoupler

Enzyme

Other Enzyme

Kinase (Ser/Thr)

NHR

Kinase (Tyr)

0

.0

Lipase

Histone Deacetylase

Histone Arginine

Methyltransferase

Histone

Acetyltransferase

Extended Data Figure 6 | See next page for caption.

©

Letter reSeArCH

Extended Data Figure 6 | Live cell imaging of USP7 inhibitor-treated

cells and combination studies with chemotherapeutic and targeted

agents. a–e, Graphs showing cell confluence as a function of time (top

rows) and normalized caspase activity (bottom rows) in cells treated

with the indicated USP7 inhibitors. Data are representative of three

biologically independent experiments and each data point represents the

average ± s.e.m. of 12 technical replicates. a, TP53 wildtype or TP53null

HCT116 colon adenocarcinoma cells. b, TP53 wildtype MCF7 or TP53

null MDAMB157 breast adenocarcinoma cells. c, TP53 wildtype U2OS

cells or TP53null SaOS osteosarcoma cells. At least three experimental

replicates were performed for all experiments. d, Graphs showing cell

confluence as a function of time (top rows) and normalized caspase

activity (bottom rows) in MCF7 breast adenocarcinoma cells treated with

GNE6640 or doxorubicin alone or in combination. e, Graphs showing

cell confluence as a function of time (top rows) and normalized caspase

activity (bottom rows) in U2OS osteosarcoma cells treated with GNE6640

or cisplatin alone or in combination. f, Pie chart illustrating the

distribution of compound classes in the Genentech Chemical Genomics

Compound library, comprising 589 compounds. NHR, nuclear hormone

receptor; GEF, guanine nucleotide exchange factor; DNA, DNAdamaging

agent. Graph source data for a–e can be accessed in the Supplementary

Information. g, Bar plot visualizing the −log10(transformed P value)

from the Wilcoxon ranksum test evaluating the enrichment of a given

compound target over all concentrations of USP7 inhibitors versus DMSO

experiments in EOL1 cells. Only compound targets with three or more

compounds in the screen were visualized; the distribution results are

indicated by the individual data points from each compound. Higher

values indicate synergy with USP7 inhibitors and were further evaluated

by Bliss analysis (Fig. 2 and Extended Data Fig. 7). Compounds common

to certain signalling pathways including PI3K/PIM, RTK/MAPK,

epigenetic regulation, and DNA damage are colourcoded as indicated.

See Supplementary Table 2 for additional information.

©

reSeArCH Letter

receptor

Tyr kinase

GDC-0339

GNE-6676 (uM)

a.

b.

c.

125

00

0

1

0.003

PIM

0.009

0.027

0.082

75

PIM

PI3K

50

P

0.247

0.741

AKT

Bad

25

0

P

2

6

.22

.67

TSC1/2

GDC-0339

-9

-8

[

-7

-6

-5

mTOR

20.00

GDC-0339] log₁₀(M)

P

125

GDC-0570

S6

S6K

100

eIF4B

75

5

2

0

5

0

translation

-

9

-8

-7

-6

-5

d.

Percent Inhibition

Bliss Excess

[GDC-0570] log₁₀(M)

2

0

95 98100100100100100100 98100

.67 43 73 94 97 99100100100 98100

20

5

7

9

8

5

2

1

3

6

5

3

1

0

2

-2

0

1

0

2

3

2

6

2

0

6.67 -3 27 47 46 32 11

.22

.74

6

17 31 72 93 98100100 98100

2.22

0.74

0.25

2

4

2

11 24 59 51 18

3

0

3

6

3

28 73 94 99 98 97 99

10 47 87 96 96 97 99

4

1

6

4

3

21 35 16

3

.25 -2 -1

.082 -2 -1

.027 -6 -5 -2

.009 -5 -4 -1

4

1

10

1

8

3

1

0

1

0

8

3

5

39 82 95 95 96 98 0.082

-1

-2

0

33 80 94 94 96 98 0.027 -1 -2 -1 -4 -5

36 79 94 94 96 98 0.009 -1 -2

0

1

2

-1 -1

.003 -5 -4

0

6

7

36 79 94 94 96 98 0.003

36 79 94 94 96 98

0

-1

-1 -1

0

-3 -2

1

0

0 -1

[GDC-0339] (µM)

e.

cell line:

EOL-1

MV-4-11

f.

cell line: HCT-116

USP7: WT USP7-null

proteasome inh.:

g.

cell line:

EOL-1

IP Ab

MV-4-11

GNE-6776:

+

IP Ab

UAE1 inh.:

+

proteasome inh.:

+

50-

WB: PIM2

1

48-

WB: USP7

WB: p21 22-

WB: p53

98-

0-

36-

*^I^g^G

PIM2

5

WB: PIM2

WB: p53

50-

5

9

0-

8-

WB: MDM2

WB: USP7

6

2-

h. recomb.

USP7: C223S

reaction time

(minutes):

1

96-

148-

wild-type

WB: USP7

WB: tubulin

250-

50-

148-

–

9

6

8-

4-

148-

98-

Ub-

PIM2

5

3

0-

6-

WB: ubiquitin

IP: poly-Ub

WB: PIM2

64-

2

1

2-

IgG

*

–

5

3

0-

6-

148-

WB: tubulin

WB: USP7

50-

98-

Extended Data Figure 7 | See next page for caption.

©

Letter reSeArCH

Extended Data Figure 7 | Mechanism of action studies with USP7

inhibitor and PIM inhibitor combinations. a, Schematic of the PI3K

signalling pathway and regulation by PIM kinases. PIM and AKT kinases

regulate Bad and TSC1/2 phosphorylation status. Phosphoproteins

highlighted in yellow were profiled in cellular studies shown in Fig. 2d.

b, Compound structure of GDC0339. c, PIM inhibitor viability curves

at the indicated fixed doses of GNE6676 in EOL1 cells (technical

duplicates). d, Bliss analysis of 9 ×9 dose–response matrix with PIM

inhibitor GDC0339 and GNE6776 in EOL1 cells, representative of

technical duplicates. Left: curvefitted viability values at each dose across

the matrix. Zero represents no effect whereas 100 indicates complete loss

of viability. Right: the difference in observed versus predicted values using

the Bliss independence model. Positive values indicate a greater than

predicted decrease in viability. Calculated synergy score =4.84.

e, Immunoblot analysis of cell lysates from the indicated cell lines

treated with GNE6776 (2 μM for 18 h), either alone or in combination

with a UAE1 inhibitor MLN7243 (5 μM for 45 min) or the proteasome

inhibitor bortezomib (5 μM for 45 min) as indicated. f, Immunoblot

analysis of cell lysates from the indicated cell lines, either untreated or

treated with the proteasome inhibitor bortezomib (5 μM for 45 min).

g, Immunoprecipitation of cellular lysates using an antiUSP7 antibody

or an isotypematched control antibody followed by western blot

analysis of lysates or immunoprecipitates with the indicated antibodies.

h, Wildtype recombinant USP7, but not catalytically inactive USP7

C223S, deubiquitinates endogenous polyubiquitinated PIM2 that was

immunoprecipitated from proteasome inhibitortreated MV411 cells.

Data in e–h are representative of at least two experimental replicates. Gel

source data: Supplementary Fig. 1.

©

reSeArCH Letter

a.

K_c_a_t(s^-¹)

b.

USP7

inhibitor

K ( M)

K

/ K

M

USP7

inhibitor

USP7/ubiquitin

µM

M

cat

(M^-s )

1

-1

K

(

)

D

GNE-6640

GNE-6776

GNE-6641

no inhibitor

ꢀ.0 ± 0.ꢀ

ꢁꢀ.ꢁ ± 0.7

3.6 ± 0.ꢁ

3.4 ± 0.ꢁ

0.22 ± 0.03

4.4 ± 0.7 ꢂ ꢁ0⁴

GNE-6640

GNE-6776

GNE-6641

no inhibitor

372.7 ± ꢀ5.ꢁ

246.0 ± 20.3

ꢁꢀ6.3 ± ꢁ2.5

ꢁ33.4 ± 7.ꢁ

0.06ꢁ ± 0.007 4.0 ± 0.ꢀ ꢂ ꢁ0³

ꢁ.36 ± 0.06

ꢁ.ꢀ4 ± 0.02

3.8 ± 0.2 ꢂ ꢁ0^ꢀ

4.ꢀ ± 0.ꢁ ꢂ ꢁ0^ꢀ

c. ꢁ22

USP7 apo

USP7 + ubiquitin

ꢁ

24

26

28

30

32

+

GNE-6640

E371

5.6

5.4

5.2

5.0

5.6

5.4

5.2

5.0

ꢁ

H δ (ppm)

d.

e.

phenol-

indazole aminopyridine

GNE-6640

PDB code 5UQV) (PDB code 5UQX)

GNE-6776

(

Data collection

Space group

Cell dimensions

a, b, c (Å)

P21

GNE-6640

GNE-6776

75.79 68.44 76.74 76.36 6 7. .53 76.86

90 95.45 90

76.4 -2.84

,

(°)

90 97 90

pyridine-

carboxamide aminopyridine

phenol-

Resolution (Å)

67.8 - 2.23)

(2.24 -2.23)

0.18 (0.94)

9.1 (3.8)

0.6 (0.016)

96.8 (96.0)

3.8 (4.0)

a

(

2.85 -2.84)

Rmerge

I/ (I)

CC1/2

Completeness (%)

Redundancy

0.052 (0.491)

15.7 (2.9)

0.999 (0.87)

96.3 (98.5)

3.6 (3.8)

α5 helix

Refinement

Resolution (Å)

No. reflections

Rwork/Rfree

No. atoms

43.5-2.84

17844

0.221/0.278

38.2-2.23

37714

0.19/0.223

α6 helix

E308

Y348

Protein

Ligand/ion

Water

5384

52

2

5535

52

221

β2 sheet

D305

B factors

Protein

Ligand/ion

Water

74

66

54

41

48

H403

Q297

R301

R.m.s. deviations

Bond lengths (Å)

Bond angles (°)

D349

0.001

0.5

0.009

1.0

β3

sheet

a

Values in parentheses are for highest resolution shell

Molprobity score for GNE-6640 = 2.ꢀ3

Molprobity score for GNE-6776 = 2.42

Extended Data Figure 8 | See next page for caption.

©

Letter reSeArCH

Extended Data Figure 8 | Enzymatic analysis and supporting structural

biology data for USP7 inhibitors and USP7. a, Michaelis–Menten kinetic

analysis of USP7 and a series of ubiquitinAMC substrate titrations

with the indicated USP7 inhibitors. Initial rate of substrate hydrolysis

was determined using the Magellan software on a Tecan Safire2 plate

reader and kinetic parameters were modelled using nonlinear regression

analysis with GraphPad Prism software. Averages with standard errors

were calculated from three technical replicates and are representative

of two (GNE6776) or three (GNE6640, GNE6641, and no inhibitor)

biologically independent experiments. b, Affinity values of ubiquitin

binding to USP7 catalytic domain in the absence and presence of USP7

inhibitors. The values were determined by titration of unlabelled ubiquitin

to labelled USP7 catalytic domain and the NMR chemical shift changes

were fitted as described in the Methods. The mean values and standard

deviations are representative of three experimental replicates and were

calculated by averaging the values obtained for eight independent

chemical shift perturbations of the wellresolved cross peaks from Y339,

S341, D342, G375, A381, G382, D412, and I419. c, Overlay of a region

1

15

of the twodimensional H/ N TROSY spectrum of the USP7 catalytic

domain (orange) highlighting changes induced by binding of ubiquitin

in the absence (left, blue) and presence of GNE6640 (right, black).

Individual peaks stemming from residues E371 and Q387 are highlighted.

Data are representative of three experimental replicates. d, Comparison

of the crystal structure of USP7 catalytic domain in complex with GNE

6640 (cyan) and GNE6776 (yellow). The catalytic domain is shown as

an orange cartoon and the side chains of the residues in proximity to the

inhibitor binding sites are shown as orange sticks. GNE6640 and GNE

6776 compound structures are indicated above. e, PDB accession numbers,

data collection, and refinement statistics for GNE6640 and GNE6776

crystal structures with the USP7 catalytic domain.

©

reSeArCH Letter

a.

USP7 + WT ubiquitin

USP7 + K48A ubiquitin

0

.20

.15

.10

.05

0

0.20

0.15

0.10

0.05

0

K = 130±15ꢀM

K = 560±60µM

D

E371

Q387

E371

A381

Q387

A381

D342

Y339

0

500 1000 1500 2000 2500

0

500 1000 1500 2000 2500

[ubiquitin] (µM)

b.

USP7 catalytic domain D305A/E308A mutant

TAMRA-peptide K48 Ub conjugate TAMRA-peptide K63 Ub conjugate

4

reaction

time (min):

0

5

10 15 20 25 30 35

0

5

10 15 20 25 30 35

5

3

0-

6-

peptide-K48 Ub₄

peptide-K63 Ub₄

22-

16-

6-

4-

c.

D305A

d.

FLAG-USP7:

wild-type

C223A + E308A

4

3

2

1

0

–

experiment #1

experiment #2

experiment #3

2

50-

IP: K48-Ub

Ub-

WB: MDM2 148-

MDM2

9

8-

-1

64-

48-

K

= 0.68 sec

cat

1

FLAG

= 1120 µM

m

WB: FLAG

WB: MDM2

-USP7

0

20

40

60

80

100

9

8-

[

Ub-Rh110] (µM)

WB: GAPdH 36-

Extended Data Figure 9 | Analysis of the functional significance

of the interactions between ubiquitin-K48 and USP7-D305, USP7-E308

residues. a, Titration curves showing the effect of unlabelled wildtype

D305A/E308A mutant. Data are representative of three experimental

replicates. c, Michaelis–Menten analysis of USP7 catalytic domain

D305A/E308A mutant showing the results of three experimental

replicates. d, Evaluation of endogenous MDM2 ubiquitination status upon

expression of wildtype, C223A, or D305A/E308A fulllength USP7.

Top: denatured lysates from MCF7 cells transfected with the indicated

USP7 expression constructs were immunoprecipitated with a K48 linkage

specific antibody and immunocomplexes were blotted with an MDM2

antibody. Western blot analysis of wholecell lysates for the indicated

proteins are shown below. Data represent three experimental replicates.

Gel source data: Supplementary Fig. 1.

2

15

13

ubiquitin (left) and ubiquitin K48A (right) addition to [ H/ N/ C]

1

15

USP7 catalytic domain. The weighted combined H/ N chemical shift

change is plotted against the ubiquitin concentration for five wellresolved

peaks stemming from E371, Q387, A381, D342, and Y339 residues in the

1

15

H/ N TROSY spectrum and fitted as described in the Supplementary

Information. The standard error was calculated on the basis of the average

values over the five fits. Single data points are representative of three

experimental replicates. b, Timecourse analysis of peptideconjugated

tetraubiquitin chains reacted with the USP7 catalytic domain

©

Letter reSeArCH

Kategaya et al. Extended Data Figure 10

USP7

K48d

a.

b.

c.

T66

USP7

exo-cleavage Ubd K48d

K63

L15

T14

exo-cleavage Ub_d

K48c

K48

K63c

USP7

Ubc

K48b

endo-cleavage Ubc K48_c

K63b

T12

USP7

K63

L15

T14

T12

T66

L8

Ub_b

K48a

endo-cleavage Ubb K48b

K63a

base-cleavage Uba K48a

K48

Ub

L8

C-term

K63

USP7

^U^ba

C-term

T66

substrate

T66

K63

K48

T14

T12

T9

T14

T12

105

110

115

120

125

130

G75

K63

K48

C-term

L8

K48

E64

G76 S65

T7

T66

I3

L15

M1

K63 Q2

T12

T14

L8

L43 L50

R74

K48

K63

L8

T66*

dist.

T66*

dist.

L15

K48

K63

K48

V70

L67

L15

C-term

I13

T14*

T12*

T14*

T12*

K48

K63

ubiquitin alone

C-term

L8*

K48

A46

ubiquitin + USP7

prox.

L15*

L15* prox.

9

.5 9.0 8.5 8.0 7.5 7.0 6.5

L8*

1

H chemical shift (ppm)

K63

C-term

d.

USP7 catalytic domain

e.

full-length USP7

TAMRA-peptide K63 Ub₄

2 3 4 5 6 7

TAMRA-peptide K48 Ub4

TAMRA-peptide K63 Ub4

reaction

time (min):

0

5 10 15 20 25 30 35

0

5

10 15 20 25 30 35

0

1

5

3

2

0-

6-

2-

50-

36-

22-

16-

6

4

-

6-

4-

1

4

2

0

8

6

4

2

0

1

4

2

0

8

6

4

2

0

free peptide

^p^e^p^tⁱ^d^e^-^U^b4

^p^e^p^tⁱ^d^e^-^U^b3

^p^e^p^tⁱ^d^e^-^U^b2

0

5

10 15 20 25 30 35

reaction time (min)

0

5

10 15 20 25 30 35

0

1

2

3

4

5

6

7

reaction time (min)

Extended Data Figure 10 | See next page for caption.

©

reSeArCH Letter

Extended Data Figure 10 | Analysis of differential USP7 binding to

K48- and K63-linked poly-ubiquitin and depolymerization kinetics.

a, Schematic diagrams of substratebound K48linked polyubiquitin

chains (left) and K63linked chains (right), and proposed USP7

interactions. Dashed lines indicate residue side chains; filled circle

indicates an isopeptide bond between the ubiquitin C terminus and

a Lys residue side chain. The proximal ubiquitin subunits (ligated to

the substrate) and their K48 or K63 side chains are indicated with ‘a’

subscripts, the next most distal ubiquitin subunits and side chains are

indicated with ‘b’ subscripts, and so on. Preferential USP7 binding to

free K48 side chains would direct USP7 to the distal ubiquitin subunit of

K48 polyubiquitin and promote sequential exocleavage, whereas USP7

would bind all subunits of K63polyubiquitin and promote exo, endo,

replicates. c, Structure depictions by ribbon diagrams of the covalent

complex between USP7 catalytic domain (orange) with ubiquitin (cyan);

(PDB accession number 1NBF) (top), K48 linked diubiquitin (green;

PDB accession number 2KDE) (lower left), and K63 linked diubiquitin

(purple; PDB accession number 2RR9) (lower right). In all diagrams,

highlighted spheres are the residues in ubiquitin that are broadened in

1

15

1

15

the H/ N SOFAST spectrum of H/ Nlabelled protein by addition of

unlabelled USP7 catalytic domain (see b for more details). Leu or Thr

residues coloured in blue show wellresolved peaks and were amenable

1

5

to selective N labelling of diubiquitin, highlighted with asterisks in

the corresponding schematic diagrams. Lysine side chains of K48 and

K63 are indicated as sticks in red. d, Timecourse analysis of TAMRA/

peptideconjugated tetraubiquitin chains cleaved by the USP7 catalytic

domain (above) and corresponding densitometry plots (below). Left: time

course of USP7 catalytic domainmediated depolymerization of K48

linked tetraubiquitin conjugated to a TAMRAlabelled peptide. Right:

time course of USP7 catalytic domainmediated depolymerization of

K63linked tetraubiquitin conjugated to a TAMRAlabelled peptide. Data

are representative of three experimental replicates. e, A shorter 0–7 min

timecourse analysis of TAMRA peptideK63 tetraubiquitin conjugate

depolymerization by fulllength USP7 (top gel) and the corresponding

densitometry plot (bottom), representative of three experimental

replicates. Gel source data: Supplementary Fig. 1.

1

15

and basecleavage. b, The H/ N SOFAST spectrum of labelled ubiquitin

at 100 μM (cyan), superimposed with the spectrum of a 1:1 molar ratio

of labelled ubiquitin in the presence of unlabelled USP7 catalytic domain

(

orange). Ubiquitin residues affected by the USP7 interaction, which

results in exchange broadening of the residue cross peaks, are labelled with

grey or blue text and correspond to the residues depicted in c. The K48

and K63 residues as well as the N and Cterminal residues are labelled in

red type. The L43 and L50 residues labelled in green type do not broaden

upon USP7 binding and serve as internal controls. The box highlights the

region depicted in Fig. 4a. Data are representative of three experimental

©

Article Doi

USP7 small-molecule inhibitors interfere with ubiquitin binding

DOI: 10.1038/nature24006

Source and publish data:

Authors:

Article abstract of DOI:10.1038/nature24006

Full text of DOI:10.1038/nature24006

Products guided by the article

R&D Labs maybe for 2009273-71-4

Relevant to this article

Hot Product