Structure of the epigenetic oncogene MMSET and inhibition by N-alkyl sinefungin derivatives

Article abstract of DOI:10.1021/acschembio.6b00308

The members of the NSD subfamily of lysine methyl transferases are compelling oncology targets due to the recent characterization of gain-of-function mutations and translocations in several hematological cancers. To date, these proteins have proven intrac

Full text of DOI:10.1021/acschembio.6b00308

Subscriber access provided by CORNELL UNIVERSITY LIBRARY
Article
Structure of the Epigenetic Oncogene MMSET
& Inhibition by N-Alkyl Sinefungin Derivatives
Dominic Tisi, Elisabetta Chiarparin, Emiliano Tamanini, Puja Pathuri, Joseph E. Coyle, Adam Hold,
Finn P. Holding, Agnes C. L. Martin, Sharna J. Rich, Valerio Berdini, Jeff Yon, Paul Acklam, Rosemary
Burke, Ludovic Drouin, Jenny Elizabeth Harmer, Fiona Jeganathan, Rob L.M. van Montfort, Yvette
Newbatt, Marcello Tortorici, Maura Westlake, Amy Wood, Swen Hoelder, and Tom D. Heightman
ACS Chem. Biol., Just Accepted Manuscript • DOI: 10.1021/acschembio.6b00308 • Publication Date (Web): 29 Aug 2016
Downloaded from http://pubs.acs.org on August 31, 2016
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
ACS Chemical Biology is published by the American Chemical Society. 1155 Sixteenth
Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 43
ACS Chemical Biology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment

ACS Chemical Biology
Page 2 of 43
1
2
3
4
5
6
7
8
Structure of the Epigenetic Oncogene MMSET &
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Inhibition by NꢀAlkyl Sinefungin Derivatives
Dominic Tisi1, Elisabetta Chiarparin1, Emiliano Tamanini1, Puja Pathuri1, Joseph E. Coyle1,
Adam Hold1, Finn P. Holding1, Agnes C.L. Martin1, Sharna J. Rich1, Valerio Berdini1, Jeff
Yon1, Paul Acklam2, Rosemary Burke2, Ludovic Drouin2, Jenny E. Harmer2, Fiona
Jeganathan2, Rob L.M. van Montfort2, Yvette Newbatt2, Marcello Tortorici2, Maura
Westlake2, Amy Wood2, Swen Hoelder2, and Tom D. Heightman1*
1Astex Pharmaceuticals, 436 Cambridge Science Park, Milton Road, Cambridge UK CB4
0QA
2CRUK Cancer Therapeutics Unit, Institute of Cancer Research, 15 Cotswold Road,
Belmont, Sutton, Surrey UK SM2 5NG
Corresponding Author: *To whom correspondence should be addressed.
T.D.H.: Tel +44 1223 226200; Fax: +44 1223 226201 Email: tom.heightman@astx.com.
1
ACS Paragon Plus Environment

Page 3 of 43
ACS Chemical Biology
1
2
3
4
ABSTRACT
5
6
7
8
9
The NSD subfamily of lysine methyl transferases are compelling oncology targets due to the
recent characterization of gainꢀofꢀfunction mutations and translocations in several
haematological cancers. To date these proteins have proven intractable to small molecule
inhibition. Here we present initial efforts to identify inhibitors of MMSET (aka NSD2 or
WHSC1) using solution phase and crystal structural methods. Based on 2D NMR
experiments comparing NSD1 and MMSET structural mobility, we designed an MMSET
construct with five point mutations in the Nꢀterminal helix of its SET domain for
crystallisation experiments and elucidated the structure of the mutant MMSET SET domain
at 2.1 Å resolution. Both NSD1 and MMSET crystal systems proved resistant to soaking or
coꢀcrystallography with inhibitors. However, use of the close homologue SETD2 as a
structural surrogate supported the design and characterization of Nꢀalkyl sinefungin
derivatives which showed low micromolar inhibition against both SETD2 and MMSET.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
2
ACS Paragon Plus Environment

ACS Chemical Biology
Page 4 of 43
1
2
3
4
INTRODUCTION
5
6
7
8
9
Over the past decade the large family of chromatin modifying enzymes have emerged as
promising targets for new anticancer therapy 1. Among them, several protein lysine
methyltransferases have been characterized as drivers in specific patient subgroups
harbouring activating mutations or translocations, spurring concerted drug discovery efforts 2.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Multiple myeloma (MM) is a genetically unstable malignancy of postgerminal center Bꢀ
lineage cells that frequently results in bone destruction, bone marrow failure, and death 3.
Some 15ꢀ20% of multiple myeloma (MM) patients possess a t(4;14) chromosomal
rearrangement that is are associated with a poor prognosis 4 5. This unique translocation
results in a simultaneous overexpression of the multiple myeloma SET domain (MMSET, aka
NSD2 or WHSC1) and fibroblast growth factor receptor 3 (FGFR3). The expression of
FGFR3 does not appear to influence the survival of the t(4;14)+ MM patients 6 7, suggesting
that MMSET plays the dominant role in the oncogenic transformation caused by the t(4;14)
translocation 8.
MMSET is expressed as several splice variants: MMSET type I (647 aa), MMSET type II
(1365 aa), and a third isotype (REꢀIIBP) with a 584 aa protein identical to the carboxyꢀ
teminal region of MMSET type II 4 9. Overexpression of MMSET in MM cells results in a
switch in methylation of histone H3 from K27 (repressive) to K36 (activating) 10 8, leading to
altered expression of a wide range of cell cycle, apoptosis, DNA repair and adhesion
regulators 11 12 13 14. Knockdown of MMSET using shRNA or active site directed
mutagenesis in KMS11 cells possessing the t(4;14) translocation caused a significant loss of
H3K36 methylation accompanied by antiproliferative effects, which were reversed by rescue
transfection with wildꢀtype MMSET II, confirming the driver role of MMSET in these cells
10
.
3
ACS Paragon Plus Environment

Page 5 of 43
ACS Chemical Biology
1
2
3
4
5
6
7
8
9
MMSET has been implicated in a range of other cancers beyond MM. Recently a E1099K
point mutation in MMSET was shown to enhance its methyltranferase activity and lead to
altered global chromatin methylation in lymphoid malignancies including pediatric acute
lymphoblastic leukemia 15, 16. MMSET overexpression has been reported in some solid
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
tumours including lung, prostate, neuroblastoma and bladder 17 18 19
.
In addition to MMSET, the closely related homologues NSD1 and NSD3 (aka WHSC1L) are
also subject to translocations in acute leukaemia resulting in fusion proteins 3, 20, and NSD3
amplifications occur in breast cancer 21. Taken together, these findings present a compelling
case for inhibitors of MMSET and its close homologues as potential personalized cancer
therapies.
Inspired by recent successes in identifying clinical candidates for the lysine
methyltransferases EZH2 22 and DOT1L 23 as well as chemical tool inhibitors for G9a 24,
SMYD2 25 and SET7/9 26, we set out to discover selective inhibitors of MMSET using a
structureꢀguided design approach. As to date no structure of MMSET has been reported, we
initiated efforts to elucidate its structure by Xꢀray crystallography, while in parallel exploring
the potential of recently published crystal structures of the homologues NSD1 and SETD2 as
surrogates for MMSET inhibitor design.
The NSD family predominantly monoꢀ and diꢀ methylate H3K36 in vivo 27, 28. The SET
domain in which the methyltransferase activity resides is located close to the Cꢀterminus in
all three family members, preceded by several Plant Homeodomain (PHD) domains and a
ProꢀTrpꢀTrpꢀPro (PWWP) domain 29. Crystal structure analyses of the SET domains of NSD1
and the closely related protein SETD2 have provided valuable insights into the architecture of
the cofactor and substrate binding sites, as well as potential conformational changes required
for substrate binding 30, 31. The catalytic core of the SET domain is sandwiched between a
4
ACS Paragon Plus Environment

ACS Chemical Biology
Page 6 of 43
1
2
3
4
5
6
7
8
9
preꢀSET region (also referred to as the AWS domain) and a postꢀSET domain. The postꢀSET
region (~20 residues) is composed of a conserved zinc binding motif and a soꢀcalled
autoregulatory loop whose conformation regulates accessibility to the substrate binding site, a
mechanism also characterized in the methyltransferase ASH1L 32. The structure of NSD1ꢀ
SET 30 shows the postꢀSET loop (PSL) in a closed conformation which occludes the substrate
binding site, and suggests a conformational rearrangement is required for substrate binding to
occur. The flexibility of the PSL is confirmed by the structural characterisation of SETD2 31:
depending on the presence of SAM or a derivative of the fungal metabolite sinefungin in the
cofactor binding site, the PSL adopt a respective inactive or active conformation. In addition
to the design of cofactor binding site inhibitors, the observation that the PSL of SETD2 can
adopt open, active conformations upon ligand binding provides an additional opportunity to
target the substrate binding site using structure based drug design (SBDD) methods, which is
not available using the published NSD1 structure. Whether access to the substrate binding
site of other SET domains is also regulated by conformational flexibility of the PSL is still to
be determined.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
5
ACS Paragon Plus Environment

Page 7 of 43
ACS Chemical Biology
1
2
3
4
RESULTS AND DISCUSSION
5
6
7
Solution Phase Characterisation and Construct Design
8
9
Despite high sequence homology between MMSET and NSD1, our initial efforts to
crystallize MMSET using SET domain constructs analogous to the one used to determine the
NSD1 structure 30 were unsuccessful. To characterize solution state differences in the
behaviour of the two proteins, we carried out a backꢀbone assignment of the NSD1 and
MMSET SET domains (NSD1ꢀSET, residues 1852ꢀ2082; and MMSETꢀSET, residues 973ꢀ
1203) in the presence of the coꢀfactor SAM 33. The NMR derived secondary structure
assignment of NSD1ꢀSET was consistent with the published crystal structure of NSD1. The
backbone assignment of MMSETꢀSET revealed a secondary structure very similar to that of
NSD1, with a core SET domain motif composed of three beta sheets with an alphaꢀhelix
adjacent to the second beta sheet (Figure S1) 33. This core SET domain is flanked by an Nꢀ
terminal helix adjacent to beta sheet 1 and the AWS domain at the other end, adjacent to beta
sheet 3.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1HꢀNMR and 1H,15NꢀNMR fingerprints of MMSETꢀSET demonstrated a large degree of
ordering upon addition of coꢀfactor SAM (Figure S3). LC/MS experiments showed that
during the purification of the NSD1ꢀSET and MMSETꢀSET proteins, extensive washing
reduced the SAM content in MMSETꢀSET, but not in NSD1ꢀSET, consistent with a higher
degree of conformational mobility in MMSET allowing for a more facile removal of SAM.
Moreover, native MS experiments on MMSET SET domain also suggested a highly
conformationally mobile structure evidenced by a broad distribution of charge states (Figure
S2). The differences in solution phase dynamics between NSD1ꢀSET and MMSETꢀSET upon
coꢀfactor binding prompted us to identify specific regions of conformational mobility which
could be modified to generate crystallisable MMSET constructs.
6
ACS Paragon Plus Environment

ACS Chemical Biology
Page 8 of 43
1
2
3
4
5
6
7
8
9
In the search for early tool inhibitors, we conducted a screen of recently published inhibitors
of other lysine methyltransferases that might show offꢀtarget activity against MMSET. We
observed weak inhibition of MMSET (77% inhibition at 300 µM) by the G9a/GLP
methyltransferase inhibitor UNC0638 34, while interestingly no inhibition of NSD1 could be
measured. ITC measurements allowed determination of a Kd of 134 µM for UNC0638 vs
MMSETꢀSET, while the Kd value vs NSD1ꢀSET was very weak (>1mM) and could not be
calculated accurately (Fig. 1a). This differential binding of UNC0638 towards MMSET
compared with NSD1 prompted us to use further 2DꢀNMR experiments to explore potential
differences in dynamics between the two proteins upon binding of UNC0638.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Analysis of the 2DꢀNMR spectrum of MMSETꢀSET in the presence of 1 mM UNC0638
identified regions of the protein which are perturbed upon ligand binding (Figure 1b). In
particular, the Nꢀterminal helix of MMSETꢀSET (residues L974ꢀE982) was observed to
undergo significant changes upon UNC0638 binding, as evidenced by the chemical shift
changes and/or intensity changes for resonances assigned to Q975, R979, T981, R986, and
K987 on the helix itself, and W1075, T1115 and F1139 which are close to the helix based on
equivalent residues in NSD1. By contrast, changes in this region were not observed for
NSD1ꢀSET on incubation with 1 mM UNC0638 (Figure S4), consistent with the lack of
interaction observed by bioassay and ITC experiments. Structural analysis of this region in
NSD1 identified a cluster of leucine residues (Lꢀ1585, Lꢀ1588 and Lꢀ1681) which forms a
local hydrophobic core stabilising the interaction between the Nꢀterminal helix and the loop
comprising residues T1680ꢀW1685 that forms one side of the SAM binding site and precedes
the highly conserved GlyꢀTrpꢀGly motif found in many SET domains (Figure 2a). Sequence
analysis confirmed that none of the three Leucine residues residues are conserved in
MMSET, where the corresponding residues are hydrophilic and not able to form a similar
hydrophobic interaction (Figure 2b). In addition in NSD1 two residues Q1682 and R1683 in
7
ACS Paragon Plus Environment

Page 9 of 43
ACS Chemical Biology
1
2
3
4
5
6
7
8
9
the T1680ꢀW1685 loop which engage in polar interactions with the body of the protein are
not conserved in MMSET. We hypothesised that these five residues could play a role in the
observed differences between NSD1ꢀSET and MMSETꢀSET in UNC0638 binding and in
local conformational flexibility. Accordingly, to reduce conformational flexibility in this
region of the protein and encourage crystallization, an MMSETꢀSET construct incorporating
five point mutations was made to replace the wild type MMSET amino acids with the
corresponding residues found in NSD1 (MMSET numbering ꢀ Q975L, A978L, D1071L,
G1072Q, K1073R). This construct was used for crystallisation studies and biophysical
characterisation.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
MMSET Crystal Structure
Post SET loop and SAM binding site
The crystal structure of mutant MMSETꢀSET was solved at 2.1 Å resolution and shares the
common SET domain architecture observed in the structures of NSD1 and SETD2 (Fig. 3).
The PSL of MMSET is in a closed conformation similar to the autoꢀinhibited conformation
observed in NSD1, albeit with a slight deviation around residues 1180ꢀ1182 (Fig. 4(a)). The
binding site of the SAM coꢀfactor is largely conserved between all three proteins, with the
majority of hydrogenꢀbonding interactions with SAM mediated by the protein backbone. The
adenine ring of SAM interacts via hydrogen bonds between N1 and the mainꢀchain NH of
Arg1192, between the 6ꢀNH2 and the carbonyl of His1142, and between N7 and the mainꢀ
chain NH of His1142 (Fig. 4 (a)). In addition the face of the adenine ring forms close van der
Waals contacts with the side chain of Leu1202. The ribose O2 oxygen interacts with the main
chain carbonyl of Thr1115, and with the side chain of His1116, while the ribose O3 oxygen
interacts with the main chain C=O of Phe1117 and the side chain of Asn1186, which serves
8
ACS Paragon Plus Environment

ACS Chemical Biology
Page 10 of 43
1
2
3
4
5
6
to anchor the end of the PSL. The amino acid group of SAM interacts with the main chain
NH and carbonyl of Trp1075 and with the phenolic OH side chain of the conserved Tyr1118.
7
8
9
Comparison of the SAM binding pocket of MMSET with NSD1 shows a high degree of
conservation: His 1116 represents the only change in an amino acid side chain directly
contacting the SAM, replaced by Asn1186 in NSD1 which undergoes very similar interaction
with the ribose O2 oxygen (Fig. 4c)). The majority of MMSET SAM site interactions are also
conserved in SETD2: in this case the only amino acid side chain change is Asn1186, replaced
by Ala1675 in SETD2 – here the key interaction with the ribose O3 is mediated by the
adjacent Gln1676 side chain.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Comparison of the channel leading from the SAM methyl group to the substrate site also
shows high sequence and structural conservation between MMSET and NSD1: all five
hydrophobic amino acid side chains lining the channel are identical (Fig. 4e). In SETD2 two
amino acid side chains in this channel are different: Leu1120 is replaced by Met1607 in
SETD2, while Leu1163 is replaced by the more bulky Phe1650 in SETD2. This feature is
somewhat counterintuitive given that unlike MMSET and NSD1, SETD2 is able to methylate
H3K36Me2 to generate the more bulky H3K36Me3. It is possible that methylation state
specificity arises not through steric selection, but due to the two key residue changes in
SETD2 being better able to stabilize the cationic quaternary ammonium group of H3K36Me3
through Πꢀcation and sulfurꢀcation interactions with Phe1650 and Met1607 respectively,
which are not possible with the two Leu residues present in these positions in MMSET and
NSD1. However it is also possible that the static view provided by these structures masks a
difference in plasticity which might allow the substrate channel to open more widely in
SETD2 to accommodate its larger substrate (see below).
9
ACS Paragon Plus Environment

Page 11 of 43
ACS Chemical Biology
1
2
3
4
N-terminal helix region
5
6
7
8
9
Although the mutation strategy led to successful crystallization of MMSET, and the overall
structures of mutant MMSETꢀSET and NSD1ꢀSET are similar, the Nꢀterminal helix
unexpectedly adopts a significantly different orientation (Figure 5a), forming a short helix
located ~18Å above the SAM binding site which is not engaged in intraꢀmolecular
interactions with the body of the protein. The electron density for this helix is sufficiently
resolved (particularly molecule A) to place main chain atoms; however, exact side chain
conformations for some residues within the helix are not clearly defined. Analysis of the
crystal packing shows that this helix is involved in crystal contacts within the crystal lattice
unlike the corresponding regions of NSD1 and SETD2, which make intraꢀmolecular
interactions but no crystal contacts within their respective crystal systems. Given the
significant differences in the solution phase mobility of this helix between NSD1 and
MMSET and its different solid phase orientations for all three proteins, it is tempting to
speculate that the helix may play a functional role in the activation of the protein. However it
should be noted that the available structures of all three proteins comprise significantly
truncated SETꢀdomain constructs which may not represent the endogenous, full length
situation. In particular, these constructs lack a number of additional chromatin binding
domains which are likely to contribute to the activation mechanism.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
SETD2ꢀInhibitor Coꢀcrystal Structures
Despite the postꢀSET loop of MMSET appearing unencumbered by crystal contacts, we were
unable to obtain coꢀstructures of sinefungin derivatives by coꢀcrystallization or soaking,
hampering structureꢀbased design efforts targeting the coꢀfactor binding site. We were also
unable to obtain coꢀstructures with the close homologue NSD1 which could be envisaged as a
10
ACS Paragon Plus Environment

ACS Chemical Biology
Page 12 of 43
1
2
3
4
5
6
7
8
9
surrogate for MMSET. The closed conformation of the substrate binding site in the structures
of both proteins also prevented structureꢀbased design or in silico docking of potential
inhibitors binding in this site. Since SETD2 is the next closest structural homologue of
MMSET, we developed conditions to evaluate the use of SETD2 as a surrogate for the design
of MMSET inhibitors. SETD2ꢀsinefungin coꢀcrystals were reproduced according to the
published conditions 31. We then solved SETD2 coꢀstructures with several Nꢀalkyl sinefungin
derivatives by either backꢀsoaking out sinefungin from SETD2ꢀsinefungin coꢀcrystals or by
coꢀcrystallizing SETD2 with an Nꢀalkyl sinefungin derivative. Back soaking sinefungin coꢀ
crystals with an Nꢀalkyl sinefungin derivative maintained the same crystal packing observed
in the SETD2ꢀsinefungin coꢀcrystal structure. However, coꢀcrystallization of apo SETD2
with Nꢀalkyl sinefungin derivatives resulted in a different crystal packing not previously
observed with the SETD2 crystal system. We generated the SETD2ꢀNꢀpropyl sinefungin
derivative crystal structure in both crystal forms and an overlay of the two crystal forms
revealed no significant conformational differences in residues lining the SAM binding site.
This gave us confidence that either crystal form would be suitable for an inꢀdepth analysis of
inhibitors bound in the SAM site of SETD2.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Conformational analysis of the Post SET Loop
High resolution structures of SETD2 with both sinefungin and Nꢀpropyl sinefungin (3)
showed a similar overall protein structure; however, key differences between the two proteinꢀ
inhibitor structures were observed in the PSL. In the SETD2ꢀsinefungin structure the PSL is
open and disordered: electron density was particularly weak for residues Gln1667ꢀGlu1674;
while in the SETD2ꢀNꢀpropyl sinefungin structure the PSL is open and ordered. An overlay
of the two structures revealed different orientations for residues Asp1665 and Tyr1666 at the
11
ACS Paragon Plus Environment

Page 13 of 43
ACS Chemical Biology
1
2
3
4
5
6
7
8
9
beginning of the PSL, which may have an impact on the ordering of the loop. Although our
SETD2ꢀNꢀpropyl sinefungin structure is very similar to that published by Zheng et al., we
observed a significant difference in the orientation of Arg1670, a residue shown by
mutational analysis to play a pivotal role in enzyme catalysis and substrate recognition 31.
Zheng et al., report Arg1670 pointing into the SAM binding site in the SETD2ꢀSAH coꢀ
structure (Fig. S5(a)); in their Nꢀpropyl sinefungin coꢀstructure Arg1670 is shifted by ~15 Å
to point out of the substrate groove, presumably in order to accommodate the Nꢀpropyl
moiety (Fig. S5(b)). However, in our Nꢀpropyl sinefungin coꢀstructure we observed clear
electron density showing Arg1670 some 10 Å away from the Zheng et al. structure, pointing
into the substrate binding site and holding the PSL in an open and ordered conformation
through interactions with the backbone carbonyls of Asn1601 and Ile1602 (Fig. S5(c)). These
observations highlight the dependence of flexibility of the SETD2 PSL on the chemical
structure of the ligand bound in the SAM binding site. Additional SETD2 structures in
complex with various Nꢀalkyl sinefungin derivatives further demonstrate the mobility of the
SETD2 PSL (see below).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Sinefungin analogues as MMSET inhibitors
The natural product sinefungin (2) has been widely reported as a nonꢀselective inhibitor of
SETꢀdomain containing KMTs 35. Sinefungin is a close structural analogue of the coꢀfactor
SAM, in which the sulfonium atom is replaced by a carbon atom and the reactive methyl
group by a basic NH2 group which, when protonated, mimics the incipient positive charge on
the methyl group as it is transferred to the lysine εꢀnitrogen atom of its substrate. The position
of the nitrogen appears important: isosteric analogues in which the nitrogen is in the position
of the sulfonium atom of SAM are generally much weaker inhibitors of SETꢀdomain
12
ACS Paragon Plus Environment

ACS Chemical Biology
Page 14 of 43
1
2
3
4
5
containing KMTs ꢀ see for example 36, 37 (although for protein arginine methyltransferases
(PRMTs) and the phylogenetically related KMT DOT1L this isostere appears favourable 23).
6
7
8
9
A number of groups have pursued analogues of sinefungin as KMT inhibitors, including
Zheng et al who explored Nꢀsubstituted sinefungin derivatives as inhibitors of SETD2 31.
These authors proposed that the Nꢀalkyl substituent could mimic the substrate lysine side
chain, potentially bridging between the coꢀfactor and substrate binding sites on opposite sides
of the protein. This led to the characterization of inhibitors such as Nꢀpropyl sinefungin (3)
and Nꢀbenzyl sinefungin (9) which showed subꢀmicromolar inhibition of SETD2 and,
somewhat surprisingly, promising selectivityagainst a panel of 13 other KMTs.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Noting the relatively close homology between SETD2 and MMSET, we explored the
potential of Nꢀsubstituted sinefungin analogues to inhibit MMSET. This required
establishment of a synthetic route allowing lateꢀstage functionalization of the amino group
with a manageable number of steps. An assessment of published syntheses of sinefungin
suggested that of Geze et al 38 to be most suited for adaptation to Nꢀsubstitution (Scheme 1).
The readily available starting material adenosine acetonide (21) was protected by N,Nꢀ
dibenzoylation, followed by conversion to the 5ꢀiodoꢀ5ꢀdeoxy derivative 22 via the tosylate.
Displacement of the iodide by diethyl cyanomethylphosphonate led to a mixture of the
monobenzoylated phosphonate ester 24 and its dibenzoylated analogue – both could be
transformed to the desired monobenzoylated alkene 27 by Wittig reaction with the aldehyde
26, itself prepared by DessꢀMartin oxidation of NꢀBoc homoserine methyl ester (25).
Magnesium reduction of the conjugated alkene in 27 followed by partial hydrolysis of the
nitrile afforded the primary carboxamide 28, which was transformed to the primary amine 29
using iodosobenzene bisꢀtrifluoroacetate. Acidic deprotection of 29 afforded sinefungin as a
mixture of 6’ꢀepimers; alternatively, reductive amination with aldehydes and NaBH3CN
followed by deprotection gave the Nꢀsubstituted sinefungin derivatives 3ꢀ9 as mixtures of 6’ꢀ
13
ACS Paragon Plus Environment

Page 15 of 43
ACS Chemical Biology
1
2
3
4
5
6
7
8
9
epimers which were used for SAR and structural studies. The corresponding analogues of Nꢀ
propyl sinefungin 3 lacking either the terminal carboxyl (10) or amino groups (11) were
prepared by condensation of the cyanophosphonate 24 with the corresponding aldehydes,
followed by a similar sequence of nitrile hydrolysis, degradation to the primary amine,
reductive amination and deprotection (Scheme 2).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
To determine inhibitory structureꢀactivity relationships for compounds against MMSET and
SETD2, we used assays in which the catalytic methylation of H3K36 by each enzyme is
measured by selective antibody detection. For SETD2 a short peptide containing the
H3K36Me2 was used as substrate with H3K36Me3 product detection in ELISA format. The
assay for MMSET was performed using a protein construct containing additional Cꢀterminal
residues (aa 941ꢀ1240), with recombinant nucleosomes containing unmodified H3K36 as
substrate and a H3K36Me2 antibody for product detection by AlphaScreen. Inhibition and
binding constant data are summarized in Table 1. Sinefungin (2) shows a modest IC50 of 26
µM vs MMSET, consistent with the previously published value of 30 µM using a 3HꢀSAM
turnover assay 35; a similar level of inhibition was measured against SETD2. The Nꢀpropyl
derivative 3 showed more potent inhibition of both enzymes: an IC50 of 3.3 ꢁM was
determined in the MMSET bioassay, consistent with a Kd of 1.6 µM as determined by ITC
(Table 1; Figure 6). Encouraged by these results, we explored the effects of growing further
into the substrate binding channel. The structure of 3 in complex with SETD2 suggested
opportunities for modification of the propyl group in several ways. The propyl group inserts
into the channel formed by hydrophobic side chain residues from the PSL (Y1666, F1664)
and the body of the enzyme (residues M1607, Y1579, F1650) (Figure 7a). The presence of a
small pocket adjacent to the propyl group suggested branching to give the secꢀbutyl
derivative 4: this resulted in a modest increase in affinity against both enzymes (Table 1). The
structure of 4 in complex with SETD2 shows good steric complementarity as the small
14
ACS Paragon Plus Environment

ACS Chemical Biology
Page 16 of 43
1
2
3
4
5
6
7
8
9
pocket is filled by the additional methyl group (Figure 7bꢀc). The inhibitors were further
elaborated to target the hydroxyl groups of the nearby Y1579 and Y1666. Substitution of the
nꢀpropyl with a hydroxyl group at the 2ꢀposition to give the secondary alcohol 5 resulted in a
significant loss in potency, despite a clear hydrogen bond to the hydroxyl group of Y1579 in
the SETD2 complex structure (Figure 7d). Similarly the 3ꢀsubstituted primary alcohol 6
incurred a significant potency loss despite three new Hꢀbonds observed between the inhibitor
hydroxyl group and the hydroxyl groups of Y1579 and Y1666 and the backbone carbonyl of
F1606 (Figure 7e). Incorporation of a basic group in this region of the molecule to give
secondary amine 7 also resulted in a significant loss of potency. These results suggested that
this region of the pocket prefers hydrophobic groups. Indeed, replacement of the nꢀpropyl
with nꢀpentyl (8) was tolerated, with only a modest loss of potency against both enzymes.
The SETD2 structure in complex with 8 shows the extended alkyl chain tucked under the aryl
ring of Tyr1579 in a similar region to the branched alkyl group of the secꢀbutyl derivative 4.
In contrast, replacement of the propyl with the more bulky benzyl group (9) was tolerated in
SETD2 but not in MMSET. The structure of 9 in complex with SETD2 shows the benzyl
group filling the lipophilic pocket created by the side chains of Tyr1579 and 1666 and by
Phe1650 and 1664, with the side chain of Met1607 forming a lid (Figure 7f). This complex
shows only modest structural rearrangements compared with the SETD2 structure in complex
with the nꢀpropyl derivative 3. By contrast, in the SAMꢀbound MMSET structure, the side
chain of Tyr1092, the corresponding residue of Tyr1579 in SETD2, is shifted to partially
occupy this pocket, reflecting differences between the two proteins in the positioning of the
PSL and the βꢀsheet comprising residues 1575ꢀ1585 (SETD2 numbering; 1088ꢀ1098 in
MMSET). These findings suggest that the SETD2 substrate channel overall is more
conformationally flexible, provide a rationale for its ability to accommodate this more bulky
derivative compared with the less flexible channel in MMSET.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
15
ACS Paragon Plus Environment

Page 17 of 43
ACS Chemical Biology
1
2
3
4
5
6
7
8
9
Finally we explored the effects of changes to the amino acid terminus of 3 which mimics the
methionine moiety of SAM/SAH. Removing either the carboxy group to give 10 or the amino
group to give 11 resulted in a dramatic loss in affinity vs MMSET, underscoring the
importance of these groups in the molecular recognition of the coꢀfactor.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
CONCLUSIONS
The crystallization of the MMSET SET domain proved highly challenging despite the
precedence of structures of the closely related homologues NSD1 and SETD2. By comparing
NSD1 and MMSET SET domain solution structures and dynamics using 2DꢀNMR
techniques, we identified regions of increased conformational mobility in MMSET. We
developed a hypothesis for a cluster of amino acid residues underlying these conformational
differences, and expressed mutant constructs designed to restrict conformational mobility in
MMSET. This work led to the identification of a construct of MMSET SET domain which
crystallized and allowed structure determination at 2.1 Å resolution. In parallel, we used
biochemical and biophysical assays to characterize Nꢀpropyl sinefungin as a low micromolar
inhibitor of MMSET. Despite the postꢀSET loop appearing unencumbered by crystal
contacts, we were unable to obtain a coꢀstructure of Nꢀpropyl sinefungin by coꢀcrystallization
or soaking. The similarity of the MMSET structure with that of SETD2 indicated that the
latter protein might be a valid surrogate for structureꢀbased design of inhibitors. We
developed robust conditions which allowed soaking of Nꢀalkyl sinefungin derivatives into
SETD2 crystals, allowing the structural characterization of a number of derivatives which
inhibited both SETD2 and MMSET with varying degrees of selectivity. The data indicated
that there is some useful overlap between the two proteins in terms of inhibitor design, but
also some divergence arising both from amino acid differences in the substrate binding
channel and also apparent differences in the plasticity of the postꢀSET loop and SET domains
16
ACS Paragon Plus Environment

ACS Chemical Biology
Page 18 of 43
1
2
3
4
5
6
7
8
9
of the two proteins, such that SETD2 can accommodate larger groups in this region.
Nonetheless we believe our findings provide a helpful foundation for further hypothesisꢀ
driven design of selective inhibitors of MMSET and/or SETD2, to allow deeper exploration
of the roles of these epigenetic modulatory proteins in diseases including cancer.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
17
ACS Paragon Plus Environment

Page 19 of 43
ACS Chemical Biology
1
2
3
4
EXPERIMENTAL
5
6
7
Protein expression and purification of MMSET
8
9
The gene encoding the SET domain of MMSET (residues 973ꢀ1203) was cloned into pET28b
(Clontech). Site directed mutagenesis was performed using the QuikChange siteꢀdirected
mutagenesis system (Agilent Technologies) to incorporate point mutations at the following
positions ꢀ Q975L, A978L, D1071L, G1072Q, K1073R. For protein expression, transformed
BL21 DE3 E.coli cells were grown in 2XYT media at 37°C until mid log phase. The
temperature was reduced to 18°C and expression was induced by the addition of Isopropyl βꢀ
Dꢀ1ꢀthiogalactopyranoside (IPTG) to a final concentration of 1 mM. Cells were incubated at
18°C overnight. Cell pellets were harvested by centrifugation at 4000rpm. Cell pellets were
resuspended in lysis buffer (50 mM Tris pH 8.0, 300 mM NaCl) and lysed by sonication.
Lysate was clarified by centrifugation at 20000 rpm for 30 mins. Clarified lysate was applied
to a 5 ml HisTrap FF Crude column (GE Healthcare) attached to an AKTA PURE
chromatography system (GE Healthcare). The column was washed with lysis buffer + 50 mM
imidazole and fractions containing MMSET were eluted with lysis buffer + 250 mM
imidazole. To generate SAM free MMSET for biophysical characterisation, MMSET bound
to the HisTrap column was washed with 1L of lysis buffer + 50 mM imidazole at a flow rate
of 1ml/min prior to elution. The vector derived Nꢀterminal hexahis tag was removed by
incubating the protein with Thrombin (Sigma) overnight at 4°C whilst dialysing against 50
mM Tris pH 8.0, 300 mM NaCl, 5 mM DTT. The protein sample was then applied to a
Superdex75 16/60 gel filtration column preꢀequilibrated in 50 mM Tris pH 8.0, 300 mM
NaCl, 5 mM DTT. Fractions containing MMSET were pooled and used for crystallisation
studies.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
18
ACS Paragon Plus Environment

ACS Chemical Biology
Page 20 of 43
1
2
3
4
Isothermal titration calorimetry
5
6
7
8
9
ITC experiments were performed on a Microcal VPꢀITC (Malvern Instruments Ltd) at 25°C
in a buffer comprising 50 mM Hepes, 100 mM NaCl, 1 mM TCEP and 5% DMSO at pH 7.5.
The protein construct used corresponded to human MMSET residues 973 – 1203 or human
NSD1 residues 1852ꢀ2082. For all experiments compound was in the syringe and protein was
in the sample cell. Data analysis was performed using Origin 7.0 software. Heats of dilution
were estimated using the final injections of each individual titration and subtracted before
data analysis. 1mM UNC0638 was titrated into 34 µM MMSET in the presence and absence
of a 10ꢀfold molar excess of SAM. In the absence of SAM the binding isotherm revealed that
UNC0638 was binding but the data quality prevented unambiguous interpretation of the
UNC0638 binding affinity. The ITC data quality was improved in the presence of SAM
allowing an estimate of the UNC0638 binding affinity for the MMSET:SAM complex.
Identical conditions were used for titration of 1mM UNC638 into 40 µM NSD1. The titration
for nꢀpropyl sinefungin 3 was performed with 400µM 3 in the syringe and 20µM MMSET
was in the cell.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Expression and purification of 15N labelled MMSET
15N labelled MMSET (residues 973ꢀ1203) was produced using the isotope labelling method
described by Li et al 39. Several colonies of transformed BL21 DE3 E.coli were transferred
into 50 ml of nonꢀinducing media and incubated overnight at 30°C. The overnight culture
was used to inoculate the inducing media with a starting OD600 of 0.02, incubated at 25°C
for 4ꢀ5 hr and left at 18°C for another 48 hr. The final OD600 reached 14. Cells were then
harvested, resuspended in lysis buffer (50 mM Hepes, 300 mM NaCl, 200 ꢁM PMSF at pH
8) and lysed in a cell disruptor at 30 psi (2 passes). Lysate was centrifuged at 100,000 g and
19
ACS Paragon Plus Environment

Page 21 of 43
ACS Chemical Biology
1
2
3
4
5
6
7
8
9
the supernatant was supplemented with 10mM imidazole and batch bound to NiNTA resin
overnight at 4°C. After extensively washing the resin in lysis buffer containing 20 mM
imidazole, MMSET was eluted in lysis buffer containing 250 mM imidazole. MMSET
fractions were pooled and dialysed overnight at 4°C into gel filtration buffer (50 mM Hepes,
150 mM NaCl, 2 mM DTT at pH 7.5) in the presence of thrombin (Sigma) to remove the Nꢀ
terminal histidine tag. Protein was then applied to a Superdex75 26/60 gel filtration column
equilibrated in gel filtration buffer. MMSET containing fractions were pooled and
concentrated to ~ 10 mg/ml. The protein was then applied to a desalting column prior to
loading onto a Qꢀsepharose column equilibrated in 50 mM sodium phosphate and 10 mM
NaCl at pH 7.5. The Qꢀsepharose bound protein was extensively washed in the running buffer
to facilitate the removal of any residual SAM associated with MMSET. The protein was
eluted from the Qꢀsepharose column by applying a 10ꢀ500 mM NaCl gradient. MMSET
containing fractions were pooled, concentrated and snap frozen in liquid nitrogen.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
NMR spectroscopy
All NMR experiments were carried out at 310 K, using a Bruker AvanceIII 500 MHz
spectrometer equipped with a cryoprobe. NMR spectra were processed using TOPSPIN and
analysed using TOPSPIN 3.1. Each NMR sample at a protein concentration of 200 µM and
saturating concentration of the coꢀfactor SAM of 2 mM was made up to a volume of 160 ꢁl in
a 2.5 mm capillary which was placed in a standard 5 mm NMR tube. The NMR buffer was
sodium phosphate 50 mM pH 7, NaCl 450 mM, TCEP 2 mM, SAM 2 mM, NaN3 0.01% w/v,
D2O 10% v/v. Binding of UNC0638 was measured by monitoring chemical shift and shape
changes of 1H and 15N resonances as a function of compound concentration. Each sample
20
ACS Paragon Plus Environment

ACS Chemical Biology
Page 22 of 43
1
2
3
4
contained 200 ꢁM MMSETꢀSET protein and increasing concentrations of UNC0638 up to
5
6
1000 µM.
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Crystallisation and structure determination of MMSET
Prior to crystallisation, MMSET protein was incubated with 1mM SAM (Sigma) dissolved in
gel filtration buffer and the pH readjusted to pH7.5. Protein was then concentrated to
~7mg/ml. Broad crystallisation screens were set up including the JCSG Core Suite (Qiagen).
Crystals of MMSET grew from conditions containing 16ꢀ20% PEG3350, 0.2M ammonium
chloride at 20°C. Plateꢀlike crystals grew overnight and were cryoꢀprotected in 30%
PEG3350 prior to being flash frozen in liquid nitrogen. Data were collected at 100K on
beamline ID29 at ESRF and processed using XDS 40. Data was scaled and merged using
SCALA 41 from the CCP4i suite of programs 42. The structure of MMSET was solved by
molecular replacement using AMORE (CCP4i) 43 using NSD1 as the template model (PDB:
3OOI). The protein structure was rebuilt and refined using COOT 43 and REFMAC 44
respectively.
MMSET crystallised in space group P212121 with two molecules in the asymmetric unit. All
residues present in the protein construct were clearly resolved in electron density for
molecule A. In molecule B, there are breaks in electron density for residues Gln982ꢀArg986
and Cys1018ꢀPro1025. Side chains where specific rotamer conformations were ambiguous
due to poor electron density were trimmed back to the carbonꢀβ atom position.
Protein expression and purification of SETD2
21
ACS Paragon Plus Environment

Page 23 of 43
ACS Chemical Biology
1
2
3
4
5
6
7
8
The gene encoding the SET domain of SETD2 (residues 1433ꢀ1711) was a gift from
Structural Genomics Consortium, Toronto. Protein expression and purification of SETD2
broadly followed the published methods 31.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Typically, protein expression was performed using transformed BL21 DE3 Codon plus RIL
E.coli cells (Agilent Technologies) grown at 37°C in Terrific Broth (Sigma) until mid log
phase. The temperature was lowered to 18°C and expression was induced by the addition of
isopropyl βꢀDꢀ1ꢀthiogalactopyranoside (IPTG) to a final concentration of 1 mM. Cells were
incubated over night at 18°C prior to harvesting by centrifugation at 4000 rpm. Cells were
resuspended in Lysis buffer (50 mM Hepes, pH 8.0, 500 mM NaCl, 1.6 mM βꢀ
mercaptoethanol) and lysed by sonication. Lysate was clarified by centrifugation at 14,000
rpm for 30 mins. The supernatant was incubated with ~10 ml of Nickel NTA resin (Qiagen)
and incubated at 4°C for 1 hr. The resin was washed with ~200 ml of lysis buffer and SETD2
protein was eluted in ~50 ml of lysis buffer + 200 mM imidazole. The Nꢀterminal hexahis tag
was removed by incubating SETD2 with rTEV protease (Sigma) whilst dialysing against 20
mM Hepes pH7.3, 5 mM DTT. This sample was then applied to a 5 ml HiTrap SP FF ion
exchange column (GE Healthcare) equilibrated in 20 mM Hepes pH 7.3, 5 mM DTT. A
linear salt gradient was applied over 20 column volumes using 20 mM Hepes pH 7.3, 500
mM NaCl, 5 mM DTT. Fractions containing SETD2 were pooled and applied to a Hiprep Sꢀ
200 26/60 gel filtration column (GE Healthcare) equilibrated in 20 mM Hepes pH 7.3, 150
mM NaCl, 5 mM DTT. Fractions containing SETD2 were pooled and concentrated to 1
mg/ml.
Crystallisation of SETD2
22
ACS Paragon Plus Environment

ACS Chemical Biology
Page 24 of 43
1
2
3
4
5
6
7
8
9
For coꢀcrystallisation of SETD2, protein at 1 mg/ml was incubated with sinefungin or the
compound of interest to a final concentration of 1.5 mM. The protein was then concentrated
to ~10 mg/ml for crystallisation. 2 µl of protein solution was mixed with 2 µl of reservoir
solution containing 0.1 M Hepes pH 7.3, 0.1 M KSCN, 25ꢀ30% MPEG2000 at 20°C.
Crystals grew using the hanging drop vapour diffusion method, appeared overnight and took
~1 week to grow to full size. For the SETD2–compound complexes obtained via compound
exchange, SETD2ꢀsinefungin coꢀcrystals were soaked in a solution containing 5 mM of the
compound of interest for 24 h at 20°C to exchange sinefungin for the compound of interest.
All crystals were cryoprotected in a solution containing 35% PEG3350, 0.1 M KSCN, 0.1 M
Hepes pH 7.3 prior to flash freezing in liquid nitrogen for data collection.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
MMSET Bioassay
MMSET activity was measured using an AlphaScreen assay. Each well contained a final
concentration of 10 nM MMSET enzyme (MMSET, aa 953ꢀ1240), 12.5 µM nucleosomes
(New England Biolabs, NEB), 0.5 µM Sꢀadenosylꢀmethionine (NEB) and 3% v/v DMSO in
reaction buffer (50 mM Tris pH 8.5, 5 mM MgCl2, 1 mM DTT, 0.01% Tweenꢀ20). Initial
reactions were performed for 60 mins at room temperature, protected from light. 1 nM Bioꢀ
antiH3 (PerkinElmer), 20 µg/ml mouse IgG beads, 1 nM H3K36Me2 antibody (Active Motif)
and 20 µg/ml Strepavidin donor beads (PerkinElmer antiꢀmouse IgG kit) in detection buffer
(50 mM Tris pH 7.5, 900 mM NaCl, 0.01% Tweenꢀ20 and 0.001% PolyꢀLꢀlysine) was then
added and incubated overnight in the dark at room temperature before reading using the
Pherastar FS (BMG), Ex680 nm/Em615 nm. IC50 values were determined by fitting the data
to a four parameter logistic fit using GraphPad Prism 6.0.
23
ACS Paragon Plus Environment

Page 25 of 43
ACS Chemical Biology
1
2
3
4
SETD2 Bioassay
5
6
7
8
9
SETD2 activity was measured using an AlphaScreen assay. Reactions were set up using a
SETD2 chemiluminescent kit (BPS Bioscience). The method was as described in the product
protocol with the exception that SAM from NEB was used. Luminescence was measured
using a Pherastar FS (BMG). IC50 values were determined by fitting the data to a four
parameter logistic fit using GraphPad Prism 6.0.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
SAM quantitation by LC/MS
SAM content in MMSET was quantified by LC/MS using an HP1100 (Agilent Technologies)
with diode array UV detection connected to a Bruker Esquire 3000 plus Ion Trap MS (Bruker
Corporation). LC conditions: column ꢀ Hypercarb 100x3 mm, 3 µm (Thermo Scientific, part
number 35005ꢀ103030) at 40 °C, aqueous solvent (A) ꢀ 100% water / 0.1% formic acid,
organic solvent (B) ꢀ 95% acetonitrile / 5% water / 0.1% formic acid. Solvent gradient – 0
min (5% B), 0.5 min (5%B), 2 min (95% B), 4 min (95% B), 4.1 min (5% B), 7 min (5% B),
flow rate = 1 ml/min with all flow to the UV detector then to the MS. SAM levels were
quantitated from the UV data (summed 265 nm to 350 nm) and the extracted ion
chromatogram (XIC) for the (M+H)+ of SAM (m/z 399.1). SAM standards for UV and MS
calibration curves were prepared from 100 mM SAM in 100 mM ammonium acetate, pH 7.4
and serially diluted in 80:20 1 M KOH:TFA to give 250 µM, 12.5 µM and 0.625 µM stock
solutions. The average peak areas from duplicate injections of the standard solutions for m/z
399.1 and the corresponding UV peak areas were plotted against amount injected in nMoles
to give calibration plots; from the MS data of 6.25 pM (LOQ) to 0.125 nM (R2 = 0.9995) and
from UV absorption data of 25 pM (LOQ) to 0.5 nM (R2 = 0.9995). Preparation of SAM
from MMSET samples was according to the following procedure: 5 µL of MMSET protein
24
ACS Paragon Plus Environment

ACS Chemical Biology
Page 26 of 43
1
2
3
4
5
6
7
8
9
was diluted with 10 µL of solvent A and 2 µL of 100% TFA was added to precipitate the
protein, the sample was aspirated a number of times and centrifuged for 5 minutes at 13000
rpm to pellet the protein. The supernatant containing SAM was removed (approximately 17
µL) and 8 µL of 1 M KOH added giving a 1 in 5 dilution of the original sample at a final
volume of 25 µL. Injection volumes were between 0.5 µL to 10 µL depending on the original
protein concentration.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accession Codes
MMSETꢀSAM: tbc; SETD2ꢀsinefungin: tbc; SETD2ꢀ3: tbc; SETD2ꢀ4: tbc; SETD2ꢀ5: tbc;
SETD2ꢀ6: tbc; SETD2ꢀ9: tbc.
25
ACS Paragon Plus Environment

Page 27 of 43
ACS Chemical Biology
1
2
3
4
ASSOCIATED CONTENT
5
6
7
8
9
Supporting Information. Figure S1 – Secondary structure assignment of NSD1 and
MMSETꢀSET domains. Native MS procedures for MMSETꢀSET. Figure S2 – Native MS for
MMSETꢀSET domain. Figure S3 – NMR observation of MMSET ordering upon addition of
coꢀfactor SAM. Figure S4 ꢀ 15Nꢀ1H HSQC 2DꢀNMR spectrum of labelled NSD1ꢀSET
domain showing NSD1 signals in presence (blue) or absence (red) of UNC0638. Figure S5 –
PostꢀSET loop conformations for SETD2 in complex with SAM analogues. Table S1 – Xꢀray
collection statistics for MMSET and SETD2 structures. Synthetic procedures for Nꢀpropyl
sinefungin (3) and analogues. This material is available free of charge via the Internet at
http://pubs.acs.org.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
PDB ID Codes
Coordinates and structure factors for the MMSET and SETD2 ligand complex have been
deposited with PDB with accession codes: tbc.
AUTHOR INFORMATION
Corresponding Author
T.D.H.: Astex Pharmaceuticals, 436 Cambridge Science Park, Cambridge CB4 0QA,
tom.heightman@astx.com.
Author Contributions
The manuscript was written through contributions of all authors. All authors have given
approval to the final version of the manuscript.
ACKNOWLEDGMENTS
26
ACS Paragon Plus Environment

ACS Chemical Biology
Page 28 of 43
1
2
3
4
5
6
7
8
9
We gratefully acknowledge H. Wu, Structural Genomics Consortium, University of Toronto
for helpful discussions on the crystallography of SETD2. This work was supported by Cancer
Research UK grant number C309/A11566. We acknowledge NHS funding to the NIHR
Biomedical Research Centre at The Institute of Cancer Research and the Royal Marsden
Hospital. We thank the staff of the European Synchrotron Radiation facility for their support
during data collection.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ABBREVIATIONS
Boc2O, diꢀtertꢀbutyl dicarbonate; BzCl, benzoyl chloride; DCM, dichloromethane; DMF,
dimethylformamide; DMSO, dimethylsulfoxide; DOT1L, DOT1ꢀlike histone H3K79
methyltransferase; DTT, dithiothreitol; EZH2, enhancer of zeste 2; IPTG, isopropyl βꢀDꢀ1ꢀ
thiogalactopyranoside; ITC, isothermal calorimetry; KMT, protein lysine methyltransferase;
MM, multiple myeloma; MMSET, multiple myeloma SETꢀdomain containing; NSD1,
nuclear receptor binding SET domain protein 1; PHD, plant homeodomain; PMSF,
phenylmethanesulfonyl fluoride; PRMT, protein arginine methyltransferase; PSL, postꢀSET
loop; SAM, Sꢀadenosyl methionine; SBDD, structureꢀbased drug design; SET, Su(var)3ꢀ9
Enhancer of zeste Trithorax; SETD2, SET domain containing 2; SET7/9, SET domain
containing 7/9; SMYD2, SET and MYND domain containing; TCEP, tris(2ꢀ
carboxyethyl)phosphine; TEA, trimethylamine; TFA, trifluoroacetic acid; TLC, thin layer
chromatography; TMSCl, trimethylsilyl chloride; TsCl, tosyl chloride.
27
ACS Paragon Plus Environment

Page 29 of 43
ACS Chemical Biology
1
2
3
4
REFERENCES
5
6
7
8
1.
2.
3.
4.
Dawson, Mark A., and Kouzarides, T. (2012) Cancer Epigenetics: From Mechanism to
Therapy, Cell 150, 12ꢀ27.
Copeland, R. A. (2013) Molecular Pathways: Protein Methyltransferases in Cancer, Clin.
Cancer Res. 19, 6344ꢀ6350.
Chesi, M., and Bergsagel, P. L. (2011) Many Multiple Myelomas: Making More of the
Molecular Mayhem, ASH Education Program Book 2011, 344ꢀ353.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Chesi, M., Nardini, E., Lim, R. S. C., Smith, K. D., Kuehl, W. M., and Bergsagel, P. L.
(1998) The t(4;14) Translocation in Myeloma Dysregulates Both FGFR3and a Novel Gene,
MMSET, Resulting in IgH/MMSET Hybrid Transcripts, Blood 92, 3025ꢀ34.
Keats, J. J., Reiman, T., Maxwell, C. A., Taylor, B. J., Larratt, L. M., Mant, M. J., Belch, A.
R., and Pilarski, L. M. (2003) In multiple myeloma, t(4;14)(p16;q32) is an adverse prognostic
factor irrespective of FGFR3 expression, Blood 101, 1520ꢀ9.
Rasmussen, T., Hudlebusch, H. R., Knudsen, L. M., and Johnsen, H. E. (2002) FGFR3
dysregulation in multiple myeloma: frequency and prognostic relevance, Br. J. Haematol.
117, 626ꢀ628.
Keats, J. J., Maxwell, C. A., Taylor, B. J., Hendzel, M. J., Chesi, M., Bergsagel, P. L., Larratt,
L. M., Mant, M. J., Reiman, T., Belch, A. R., and Pilarski, L. M. (2005) Overexpression of
transcripts originating from the MMSET locus characterizes all t(4;14)(p16;q32)ꢀpositive
multiple myeloma patients, Blood 105, 4060–4069.
Mirabella, F., Wu, P., Wardell, C. P., Kaiser, M. F., Walker, B. A., Johnson, D. C., and
Morgan, G. J. (2013) MMSET is the key molecular target in t(4;14) myeloma, Blood Cancer
J. 3, e114.
5.
6.
7.
8.
9.
Garlisi, C. G., Uss, A. S., Xiao, H., Tian, F., Sheridan, K. E., Wang, L., Motasim Billah, M.,
Egan, R. W., Stranick, K. S., and Umland, S. P. (2001) A Unique mRNA Initiated within a
Middle Intron of WHSC1/MMSET Encodes a DNA Binding Protein That Suppresses Human
ILꢀ5 Transcription, Am. J. Respir. Cell Mol. Biol. 24, 90ꢀ98.
10.
MartinezꢀGarcia, E., Popovic, R., Min, D.ꢀJ., Sweet, S. M. M., Thomas, P. M., Zamdborg, L.,
Heffner, A., Will, C., Lamy, L., Staudt, L. M., Levens, D. L., Kelleher, N. L., and Licht, J. D.
(2011) The MMSET histone methyl transferase switches global histone methylation and alters
gene expression in t(4;14) multiple myeloma cells, Blood 117, 211ꢀ20.
Lauring, J., Abukhdeir, A. M., Konishi, H., Garay, J. P., Gustin, J. P., Wang, Q., Arceci, R. J.,
Matsui, W., and Park, B. H. (2008) The multiple myeloma–associated MMSET gene
contributes to cellular adhesion, clonogenic growth, and tumorigenicity, Blood 111, 856ꢀ64.
Brito, J. L. R., Walker, B., Jenner, M., Dickens, N. J., Brown, N. J. M., Ross, F. M.,
Avramidou, A., Irving, J. A. E., Gonzalez, D., Davies, F. E., and Morgan, G. J. (2009)
MMSET deregulation affects cell cycle progression and adhesion regulons in t(4;14)
myeloma plasma cells, Haematologica 94, 78ꢀ86.
11.
12.
13.
Kuo, Alex J., Cheung, P., Chen, K., Zee, Barry M., Kioi, M., Lauring, J., Xi, Y., Park,
Ben H., Shi, X., Garcia, Benjamin A., Li, W., and Gozani, O. (2011) NSD2 Links
Dimethylation of Histone H3 at Lysine 36 to Oncogenic Programming, Mol. Cell 44, 609ꢀ
620.
14.
15.
Walker, B. A., Wardell, C. P., Chiecchio, L., Smith, E. M., Boyd, K. D., Neri, A., Davies, F.
E., Ross, F. M., and Morgan, G. J. (2011) Aberrant global methylation patterns affect the
molecular pathogenesis and prognosis of multiple myeloma, Blood 117, 553ꢀ62.
Jaffe, J. D., Wang, Y., Chan, H. M., Zhang, J., Huether, R., Kryukov, G. V., Bhang, H.ꢀe. C.,
Taylor, J. E., Hu, M., Englund, N. P., Yan, F., Wang, Z., Robert McDonald Iii, E., Wei, L.,
Ma, J., Easton, J., Yu, Z., deBeaumount, R., Gibaja, V., Venkatesan, K., Schlegel, R., Sellers,
W. R., Keen, N., Liu, J., Caponigro, G., Barretina, J., Cooke, V. G., Mullighan, C., Carr, S.
A., Downing, J. R., Garraway, L. A., and Stegmeier, F. (2013) Global chromatin profiling
reveals NSD2 mutations in pediatric acute lymphoblastic leukemia, Nat. Genet. 45, 1386ꢀ
1391.
16.
Oyer, J. A., Huang, X., Zheng, Y., Shim, J., Ezponda, T., Carpenter, Z., Allegretta, M., Okotꢀ
Kotber, C. I., Patel, J. P., Melnick, A., Levine, R. L., Ferrando, A., MacKerell, A. D.,
28
ACS Paragon Plus Environment