Discovery of Peptidomimetic Antibody-Drug Conjugate Linkers with Enhanced Protease Specificity

Article abstract of DOI:10.1021/acs.jmedchem.7b01430

Antibody-drug conjugates (ADCs) have become an important therapeutic modality for oncology, with three approved by the FDA and over 60 others in clinical trials. Despite the progress, improvements in ADC therapeutic index are desired. Peptide-based ADC linkers that are cleaved by lysosomal proteases have shown sufficient stability in serum and effective payload-release in targeted cells. If the linker can be preferentially hydrolyzed by tumor-specific proteases, safety margin may improve. However, the use of peptide-based linkers limits our ability to modulate protease specificity. Here we report the structure-guided discovery of novel, nonpeptidic ADC linkers. We show that a cyclobutane-1,1-dicarboxamide-containing linker is hydrolyzed predominantly by cathepsin B while the valine-citrulline dipeptide linker is not. ADCs bearing the nonpeptidic linker are as efficacious and stable in vivo as those with the dipeptide linker. Our results strongly support the application of the peptidomimetic linker and present new opportunities for improving the selectivity of ADCs.

Full text of DOI:10.1021/acs.jmedchem.7b01430

Subscriber access provided by READING UNIV
Article
Discovery of Peptidomimetic Antibody-Drug
Conjugate Linkers with Enhanced Protease Specificity
Binqing Wei, Janet Gunzner-Toste, Hui Yao, Tao Wang, Jing Wang, Zijin Xu, Jinhua Chen, John
Wai, Jim Nonomiya, Siao Ping Tsai, Josefa dela Cruz-Chuh, Katherine Ruth Kozak, Yichin Liu,
Shang-Fan Yu, Jeff Lau, Guangmin Li, Gail D. Lewis Phillips, Doug Leipold, Amrita Kamath, Dian
Su, Keyang Xu, Charles Eigenbrot, Stefan Steinbacher, Rachana Ohri, Helga Raab, Leanna R.
Staben, Guiling Zhao, John A. Flygare, Thomas H Pillow, Vishal Verma, and Brian Salvatore Safina
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.7b01430 • Publication Date (Web): 11 Dec 2017
Downloaded from http://pubs.acs.org on December 12, 2017
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.
Journal of Medicinal Chemistry 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 37
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
Discovery of Peptidomimetic AntibodyꢀDrug Conjugate Linkers with
Enhanced Protease Specificity
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
BinQing Wei,*,† Janet GunznerꢀToste,† Hui Yao, ‡ Tao Wang, ‡ Jing Wang, ‡ Zijin Xu, ‡ Jinhua Chen, ‡
John Wai, ‡ Jim Nonomiya,† Siao Ping Tsai,† Josefa Chuh,† Katherine R. Kozak,† Yichin Liu,† Shangꢀ
Fan Yu,† Jeff Lau,† Guangmin Li,† Gail D. Phillips,† Doug Leipold,† Amrita Kamath,† Dian Su,†
Keyang Xu,† Charles Eigenbrot,† Stefan Steinbacher,§ Rachana Ohri,† Helga Raab,† Leanna R Staben,†
Guiling Zhao,† John A. Flygare,† Thomas H. Pillow,† Vishal Verma,† Brian Safina*,†
†Genentech, Inc., 1 DNA Way, South San Francisco, CA 94080, USA
‡WuXi AppTec, 288 Fute Zhong Road, Waigaoqiao Free Trade Zone, Shanghai, 200131, China
§Proteros Biostructures GmbH, Bunsenstr. 7a, Dꢀ82152 Martinsried, Germany
* Correspondence: Wei.BinQing@gene.com , Safina.Brian@gene.com
ABSTRACT
AntibodyꢀDrug Conjugates (ADCs) have become an important therapeutic modality for
oncology, with three approved by the FDA and over 60 others in clinical trials. Despite the progress,
improvements in ADC therapeutic index are desired. Peptideꢀbased ADC linkers that are cleaved by
lysosomal proteases have shown sufficient stability in serum, and effective payloadꢀrelease in targeted
cells. If the linker can be preferentially hydrolyzed by tumorꢀspecific proteases, safety margin may
improve. However, the use of peptideꢀbased linkers limits our ability to modulate protease specificity.
Here we report the structureꢀguided discovery of novel, nonꢀpeptidic ADC linkers. We show that a
cyclobutaneꢀ1,1ꢀdicarboxamideꢀcontaining linker is hydrolyzed predominantly by cathepsin B while the
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 37
valineꢀcitrulline dipeptide linker is not. ADCs bearing the nonꢀpeptidic linker are as efficacious and
stable in vivo as those with the dipeptide linker. Our results strongly support the application of the
peptidomimetic linker and present new opportunities for improving the selectivity of ADCs.
1
2
3
4
5
6
7
8
9
Introduction
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
The use of antibody drug conjugates has emerged as a powerful approach to deliver cytotoxic
drugs to tumor cells. There are a number of clinical trials ongoing involving this class of therapeutics.1
However, the therapeutic window of existing clinicalꢀstage ADC’s is relatively narrow.2,3 Factors likely
contributing to this phenomenon include linker instability that results in premature release of the
attached cytotoxic payload during systemic circulation,4,5 and the release of payload in nonꢀcancerous
tissues due to nonꢀspecific uptake (pinocytosis) and/or low level of antigenꢀexpression on normal cells.6
Accordingly, current approaches to the design of new ADCs seek to improve on at least one of these
parameters in order to achieve enhanced clinical performance.7
The use of proteaseꢀcleavable peptide linkers, in particular a valineꢀcitrulline dipeptide linker
(ValꢀCit),8 has proved to be an effective strategy for ADC construction owing much to the clinical
success of brentuximab vedotin.9 Peptidic linkers have been shown to be stable in circulation and able
to effectively release attached payloads inside target cells10 and, when conjugated to noninternalizing
antibodies, outside target cells.11 They are the subject of active investigations.12,13 It is generally
believed that lysosomal cathepsin B is one of the proteases mainly responsible for the cleavage of the
ValꢀCit ADC linker inside target cells. Other proteases, including both close homologs of cathepsin B
(e.g. cathepsin X) as well as enzymes that are distantly related/unrelated (e.g. cathepsins L and D), may
also be involved.8 Importantly, the precise extent to which the activation of ValꢀCit ADC linker
depends on cathepsin B enzyme activity has not been reported.
Cathepsin B (EC 3.4.22.1) is a member of the papain family of cysteine proteases.14 It can
function as an exopeptidase (i.e. carboxydipeptidase) at acidic pH as well as an endopeptidase at neutral
pH, owing to a structural feature called the occluding loop.15,16 Cathepsin B is normally associated with
ACS Paragon Plus Environment

Page 3 of 37
Journal of Medicinal Chemistry
lysosomes involved in autophagy and immune response, but numerous studies have shown that its
overexpression is correlated with invasive and metastatic phenotypes in cancers.17,18 A major
mechanism that cathepsin B contributes to cancer progression is its involvement in degradation of the
extracellular matrix.19 A recent report suggested that the exopeptidase and endopeptidase activities of
cathepsin B both contribute to the proteolysis of extracellular matrix and tumor cell invasion.20
Therefore it is conceivable that, compared to the ValꢀCit linker, an ADC linker whose hydrolysis is
more dependent on cathepsin B may help target the release of the attached cytotoxic payloads more
specifically to tumor cells.
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
In an effort to further enhance the tumorꢀspecific release of ADC payload via proteaseꢀ
dependent cleavage, we sought to identify nonꢀpeptidic proteaseꢀsubstrates that could serve as effective
linkers for ADC. To the best of our knowledge, the exploration of such peptidomimetic ADC linkers
has not been previously reported. We anticipated that, due to a reduction in the number of hydrolysable
amide bonds, the new linker moieties would offer greater opportunities to modulate the recognition by
tumorꢀspecific/elevated proteases relative to the legacy peptide linkers. In this initial report, we
describe the identification of several peptidomimetic linkers with enhanced specificity toward cathepsin
B that efficiently release various attached payloads following intracellular ADCꢀmediated delivery.
Results and Discussion
To facilitate rapid in vitro biochemical testing of our new linker designs, we devised a simplified
molecular construct to assess the ability of various linkerꢀpayloads to undergo proteaseꢀmediated
cleavage (Scheme 1, compound 1). Using this system, hydrolysis of the anilineꢀderived amide bond
present in 1 can be readily detected through a norfloxacin payload released by subsequent immolation of
the liberated pꢀaminobenzylcarbonyl moiety. We anticipated that these simple molecules could be used
to model the ability of peptidomimetic linkers to undergo similar proteaseꢀeffected cleavage following
ADCꢀmediated intracellular delivery. We expect the model to be valid regardless of whether the linkerꢀ
payload is attached to an antibody or to just a cysteine, since the exact sequence of events between
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 37
linker cleavage and antibody degradation is not known.
1
2
3
4
5
6
7
8
9
To validate our approach of using this simplified molecular construct, we applied it to some
representative dipeptide linkers and measured their cleavage kinetic parameters using a constant
concentration of cathepsin B enzyme. MichaelisꢀMenten steadyꢀstate Vmax and Km data for the linkers
were compared as the P1 and P2 side chains with different charge, polarity and shape were explored
(Scheme 1, Table 1). A hydrogen bond donorꢀrich citrulline sidechain at the P1 position afforded a ~9
fold higher Vmax value than an alanine sidechain (1 vs. 2). A positively charged arginine sidechain at
this position further improved substrate binding affinity (>10 fold reduction of Km) compared to the
citrulline, making 3 the most preferred substrate in the biochemical assay. Despite its positive charge,
βꢀ(4ꢀpiperidinyl) alanine (4) caused a ~10 fold decrease in binding affinity than 3, suggesting that this
bulky sidechain may fit less well in the active site. At the P2 position, bulky aromatic sidechains gained
affinity (lower Km) at the cost of Vmax, resulting in no improvement in the ratio of Vmax/ Km ratio (1 vs. 5
or 6). These observed trends are consistent with previous reports of the substrate preference profile of
human cathepsin B determined using combinatorial peptide libraries,21,22 and support the use of this
approach. Additionally it is important to note that, despite having a Vmax/Km value 3ꢀfold lower than the
ValꢀCit linker, the valineꢀalanine linker (ValꢀAla) has been successfully used in ADCs currently in
clinical trials.10 Therefore we reasoned that a proteolytic efficiency level similar to that of the ValꢀAla
linker would be sufficient for an effective ADC.
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 guide the design of new, nonꢀpeptidic linkers cleavable by cathepsin B, we identified the
structure of the enzyme bound to a dipeptidyl nitrile inhibitor (PDB entry 1GMY).23 This structure was
particularly informative of the substrate recognition by the enzyme because the inhibitor has a peptide
scaffold while its nitrile moiety formed a thioimidate ester with the catalytic cysteine residue (Cys29),
mimicking the acyl enzyme intermediate of substrate hydrolysis (Figure 1A and 1B).23 The amide
backbone of the molecule extends along the active site, forming hydrogen bonds with Gly74 and
Gly198 residues known to be critical for substrate binding (Figure 1B).23 At the position corresponding
to the P1 residue of peptide substrate is a glycine moiety. Its Cα carbon atom is within 5 Å of the
ACS Paragon Plus Environment

Page 5 of 37
Journal of Medicinal Chemistry
carboxylate group of Glu122 residue, which has been proposed to confer a preference for cationic
sidechains at P1 position.15,23 At the position corresponding to the P2 residue is a phenylalanineꢀlike
moiety, fitting snugly in a hydrophobic pocket presumably occupied by those lipophilic sidechains
favored at this position (e.g. valine and phenylalanine). We hypothesized that these backbone and
sidechain interactions together hold the scissile amide bond in the correct position for nucleophilic
attack by Cys29. This led us to investigate two ways to design nonꢀpeptidic substrates: (1) to replace
the amide bond between the P1 and P2 residues, and (2) to replace the P2 residue.
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
We first sought to replace the amide moiety between the P1 and P2 residues while maintaining
the interaction of those sidechains with the protein. Guided by modeling based on the crystal structure
(Figure 1B, Supporting Information Figures S1A and S1B), we identified two chemical series that
showed significant activities as substrate: fluoro olefins (Table 2) and 1,2,3ꢀtriazoles (Table 3). With a
citrulline sidechain at the P1 position, the valineꢀfluoroolefinꢀcitrulline linker (7) showed a 6ꢀfold lower
Vmax value and a 7ꢀfold higher Km value than the ValꢀCit peptide linker (1), amounting to a nearly 40ꢀ
fold reduction in the Vmax/ Km ratio. Changing the P1 sidechain from citrulline to alanine decreased the
catalytic efficiency by approximately 6 fold (7 vs. 8), which is consistent with the trend observed from
the corresponding peptide linkers (1 vs. 2). In the triazole series, the valineꢀtriazoleꢀcitrulline linker (9)
showed a 210ꢀfold loss in catalytic efficiency relative to the ValꢀCit linker. Replacing the citrulline
sidechain with an arginine (10) or a lysine (11) sidechain at the P1 position improved the cleavage
efficiency of the triazole linker by 18 to 29 fold, though the substrates were still significantly less active
than the corresponding peptide linkers (10 vs. 3). Unlike the peptide linkers, incorporation of an
aromatic P2 sidechain reduced the cleavage efficiency (11 vs. 12, and 9 vs. 13). Based on modeling, we
had anticipated that for both these series, some of the hydrogen bonds between the linker’s peptide
backbone and the protein could be lost (Supporting Information Figure S1A and S1B). The reduction in
proteolytic efficiency of those new molecules relative to the peptides suggested to us that those
hydrogen bond interactions are critical.
Next we sought to replace the P2 amino acid of peptide linker while keeping its hydrogen bond
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 37
interactions with the residues Gly74 and Gly198 of the protease. Using a computational, shapeꢀ
similarity search, we identified the cyclobutaneꢀ1,1ꢀdicarboxamide series (“cBu”, Figure 1A, 1C and
1D). The cyclobutyl group was predicted to be optimal in size to fill the hydrophobic S2 pocket, while
the 1,1ꢀdicarboxamide moiety was expected to form all three hydrogen bonds to the protein like the
peptide linker. Furthermore, the bound conformation of cyclobutaneꢀ1,1ꢀdicarboxamide predicted by
modeling was found energetically favored. We tested those predictions by making a series of
compounds, varying the moieties on either side of the cyclobutyl ring as well as the ring itself (Table 4).
As the ring size ranged from 3 to 6 atoms, the Vmax and Vmax/ Km ratio both peaked when the size
reached 4 (compounds 14-17). The linker bearing a gemꢀdimethyl glycine (18) showed an activity too
low to be determined under our assay conditions. At the P1 position, compared to a citrulline sidechain
(15), an arginine sidechain increased the cleavage efficiency of the linker by about 8 fold (19), while a
lysine or dimethyl lysine sidechain only improved the Vmax/ Km ratio by up to about 3 fold (20, 21).
Acetylation of the lysine sidechain reduced the Vmax/ Km ratio by about 2 fold (22). More significantly,
removal of all hydrogen bond donors at this position rendered the cleavage activity too low to be
determined (23). In addition to the P1 and P2 positions, we also explored substitutions on the cBu
linker that correspond to the P3 sidechain of the peptide linker. Modeling suggested that an
aryl/heteroaryl ring with appropriate chirality may contribute to tighter binding affinity by stacking with
the backbone amide of Gly74 and interacting with Try75 residue. Indeed, compared to an isopropyl
group (24), a 3ꢀthiophenyl (25) or phenyl (27) moiety increased the Vmax/ Km ratio by 5ꢀ10 fold. That
improvement was more due to tighter substrateꢀbinding than better Vmax. Given that both ValꢀCit and
ValꢀAla dipeptides were shown effective linkers in vivo, it was not clear to us that a P3 substitution
maybe necessary, so we did not try to optimize the P3 moiety further.
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
To verify our computational model of the cBu linker, we obtained a crystal structure of human
cathepsin B in complex with an analog of the cBuꢀCit compound (Table 4, 28). Because 15 would be
hydrolyzed by the enzyme, we replaced the scissile amide bond with an electrophilic cyano group to
trap the hydrolysis product. The resulting 1.6 Åꢀresolution crystallographic structure showed 28
ACS Paragon Plus Environment

Page 7 of 37
Journal of Medicinal Chemistry
forming a thioimidate ester with the catalytic residue Cys29 (Figure 1D and 1E), like the PDB structure
that guided the modeling (Figure 1A). Gratifyingly, some of the predicted interactions were observed:
the dicarboxamide moiety is wellꢀpositioned to form three hydrogen bonds with residues Gly74 and
Gly198, and the cyclobutyl ring indeed occupies the S2 pocket. The electron density was sufficient to
determine the coordinates of the citrulline side chain only up to the Cβ atom. We speculated that, in the
absence of a substrate leaving group, the rest of the side chain could be rather mobile. Also, there was
not enough electron density to define the position of the terminal methyl or the chirality of 28. Overall,
the crystallographic structure as well as the data (Table 4) confirmed our computational model for the
cBu linker.
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
Among the new peptidomimetic linkers we identified, Vmax values for the cBu series were found
to be similar to those for the peptide linkers while the fluroolefin and the triazole series exhibited lower
Vmax values (Figure 2). In terms of Vmax/Km ratio, the cBu linker was about 15 fold lower than the
corresponding peptide linker (15 vs. 1, 19 vs. 3). We wondered whether/how this difference would
affect the payload release and efficacy of ADC.
Using those new, nonꢀpeptidic linkers, we prepared ADCs conjugated to different payloads
(Tables 5 and 6). We adopted a cysteineꢀengineering strategy that allows for the straightforward
conjugation of payloads to yield homogeneous antibodyꢀdrug conjugates.24 The drug/antibody ratios
(DAR) for all the conjugates were between 1.8 and 2 as determined by LC/MS analysis of the lysyl
endopeptidase digest of the ADCs. Here we report the data from ADCs conjugated at position 118 (EU
numbering) on the antibody heavy chain. The results are representative of our findings using ADCs
conjugated at different sites25 (data not shown). Figure 3A shows a representative data set obtained with
ADCs carrying monomethyl auristatin E26 (MMAE) payload. In the Her2ꢀoverexpressing SKꢀBRꢀ3
cells, antiꢀHer2 ADCs using either ValꢀCit (29) or cBuꢀCit (31) linkers were equally active in inhibiting
cell proliferation. As expected, a nonꢀtarget control antibody (antiꢀCD22) conjugated with MMAE
using the cBuꢀCit linker had little activity in SKꢀBRꢀ3 cells. Similarly, the antiꢀHer2 ADCs showed
very small effects against MCF7 cells which express only low levels of the Her2 antigen (data not
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 37
shown). Collectively, the data indicated that the two ADCs were equally active in inhibiting the
proliferation of the cancer cells and their activities were antigenꢀspecific. However, this equivalence in
activity does not necessarily mean that the rates of payload release inside target cells are similar for the
two linkers. This is because, for example, full release of free MMAE molecules may not be required for
the activity.
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
Given the differences in Vmax and Km values observed for the ValꢀCit and cBuꢀCit model
constructs (1 vs 15), we wondered whether the two ADCs (29 vs. 31) may differ in their ability to
release free MMAE in targeted cells. We measured the concentrations of MMAE inside SKꢀBRꢀ3 cells
after ADC treatment. A duration of 23 hours was found optimal for the assay as it was long enough to
allow ADCs to be substantially endocytosed and extensive cell killings or lysis had not started yet. We
tested ADC concentrations ranging from 0.08 to 10 µg/mL and found that the Her2ꢀmediated
endocytosis appeared to saturate at about the 2 µg/mL level. Surprisingly, no statistically significant
difference in the amounts of released MMAE was found between an ADC containing the cBuꢀVal linker
(31) or one bearing the ValꢀCit peptide linker (29, Figure 3B). This was observed at all ADC
concentrations tested regardless of whether the targetꢀmediated endocytosis reached saturation or not.
To test whether enzymatic or nonꢀenzymatic process underlie the release of MMAE, two ADCs
containing the respective (R)ꢀcitrulline linkers (30, 32) were included for comparison. Switching from
the Sꢀ to the Rꢀcitrulline is expected to render those linkers resistant to proteases. Indeed, both 30 and
32 afforded negligible release of MMAE at all concentration levels examined. This confirms that the
observed release resulted from protease hydrolysis instead of nonꢀenzymatic process. The apparent
discrepancy between these observations in the cells and the Vmax/ Km data of the corresponding model
compounds (1 vs. 15) can be explained by the fact that the cathepsin B assay does not fully capture how
the ADC linkers are degraded in cell lysosome. For example, intracellular concentrations of the
substrates (i.e. ADC) and/or the proteases may be substantially different (higher) than those in the
enzymatic assay. In addition, other proteases besides cathepsin B may contribute to intracellular payload
release. To further probe this latter possibility, we conducted additional experiments using the MMAE
ACS Paragon Plus Environment

Page 9 of 37
Journal of Medicinal Chemistry
conjugates.
1
2
3
4
5
6
7
8
9
Using a panel of commercially available protease inhibitors with varying degrees of specificity,
we assessed the impact of these reagents on the MMAE payload release from several ADCs bearing
peptide or nonꢀpeptidic linkers (29, 31, 33, 34, Figure 4). We tested CAꢀ074_Me27 (cathepsin B
inhibitor), CAA022528 (cathepsin L inhibitor), BML_24429 (cathepsin K inhibitor), pepstatin30 (inhibitor
of aspartyl proteases such as cathepsin D), and leupeptin31 (inhibitor of cysteine, serine and threonine
proteases). The inhibitors were applied at a concentration of 50µM in order to saturate the enzymes.
Interestingly, release of MMAE from the ValꢀCitꢀcontaining ADC (29) appeared resistant to all but the
broadꢀspectrum protease inhibitor leupeptin, which reduced the intracellular MMAE level by 50%. For
example, only a 15%ꢀdecrease in the MMAE level was observed upon the treatment with either the
aspartyl proteases inhibitor (not statistically significant) or the potent cathepsin B inhibitor. In contrast,
for the three ADCs bearing the peptidomimetic linkers, the release of MMAE was over 90% blocked by
leupeptin. More importantly, the payload release of those ADCs was more sensitive to the inhibition of
cathepsin B than that of the ValꢀCitꢀcontaining ADC. For the cBuꢀCitꢀcontaining ADC (31), the
MMAE release was over 75% suppressed while, for the triazole (34) and the fluoroolefinꢀcontaining
ADCs (33), the reductions were 60% and 32%, respectively. The additional decrease of MMAE level
upon the treatment with leupeptin compared to that with the cathepsin B inhibitor suggest that some
other enzyme(s) may contribute to the cleavage of these nonpeptidic linkers as well. We also observed
a 26% decrease in payload release by the triazoleꢀcontaining ADC after the treatment with the cathepsin
Lꢀspecific inhibitor. It is important to note that this comparison relies on the published specificity data
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
of those inhibitors.
Nevertheless, the observed differences strongly suggest that the new
peptidomimetic linkers, especially cBuꢀCit linker, have an enhanced protease specificity toward
cathepsin B compared to the ValꢀCit peptide linker.
We next compared CD22ꢀtargeting ADCs bearing the cBuꢀCit (31) or ValꢀCit linker (29) in their
efficacy and stability in mouse xenograft tumor models. In a CD22ꢀexpressing lymphoma model
(WSUꢀDLCL2), the two ADCs displayed equivalent, targetꢀdependent activity when administered at
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 37
two different doses (Figure 5A). No body weight loss was observed from either conjugate at up to 8
mg/kg dose level suggesting they were both wellꢀtolerated (Supporting Information, Figure S2). We
also tested the same ADCs in another CD22ꢀexpressing lymphoma model (Bjabꢀluc), and again
observed equivalent efficacies for the two conjugates (Figure 5B). To see if this observation was only
1
2
3
4
5
6
7
8
9
applicable to tubulinꢀtargeting payloads, we also tested
a
DNAꢀdamaging payload,
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
pyrrolobenzodiazepineꢀdimer32 (PBD) (Table 6). In a Napi2bꢀexpressing ovarian cancer model
(OVCAR3X2.1), Napi2bꢀtargeting ADCs containing either the ValꢀCit or cBuꢀCit liners were both
efficacious in inhibiting tumor growth when dosed at the 3 mg/kg level (Figure 5C). Interestingly, the
cBuꢀCit linked ADC appeared somewhat more potent and afforded tumor regression in this model. The
efficacy comparison suggested that, in the tumor cells, the proteolytic capacity of cathepsin B was
sufficient for effective release of payload from the cBuꢀCit linker. Contributions by other proteases that
can digest the ValꢀCit but not the cBuꢀCit linker are not required for the efficacy. Besides the efficacy,
we also monitored the pharmacokinetic profiles of the two ADCs in the same mouse xenograft study.
For example, for the 1mg/kg dose groups in the above Bjabꢀluc lymphoma study (Figure 5B, group 2
and 4), no significant difference was observed in either the halfꢀlife of the antibody (Figure 6A) or the
stability of the conjugate (Figure 6B) between the two ADCs bearing different linkers. This equivalence
in both antibody halfꢀlife and conjugate stability between the ValꢀCit and the cBuꢀCit linkers was also
demonstrated in mouse tumor studies using ADCs carrying the PBD payload (data not shown).
Conclusion
Structureꢀbased design led to the discovery of three series of nonꢀpeptidic ADC linkers,
including the cBu series which exhibited a level of proteolysis efficiency approaching that of the peptide
linkers. A crystal structure of an analogous covalent inhibitor bound to cathepsin B confirmed the
modeling predictions regarding how a cBuꢀCit linker is recognized by the protease active site for
cleavage. A comparison of the cBuꢀCit linker with the ValꢀCit linker in cancer cell lines found that they
exhibited equally potent antiꢀproliferation effects and displayed similar rates of intracellular payloadꢀ
ACS Paragon Plus Environment

Page 11 of 37
Journal of Medicinal Chemistry
release. Using selected protease inhibitors, we showed that the cleavage of the cBuꢀCit linker
predominantly depends on cathepsin B. In contrast, the degradation of the ValꢀCit peptide linker was
resistant to the inhibition of cathepsin B. Furthermore, ADCs containing the cBuꢀCit linker were
equally efficacious and stable in multiple mouse tumor models compared to the conjugates bearing the
ValꢀCit linker. This equivalence in in vivo efficacy and stability was consistently observed with ADCs
delivering MMAE drug (a tubulin polymerization blocker) as well as ADCs carrying PBD dimer
payload (a DNAꢀalkylator). Our data support the application of the new peptidomimetic linkers in the
construction of ADC therapeutics, and demonstrate the feasibility and opportunity to tailor the linker
moiety toward cleavage by tumorꢀspecific/enhanced proteases.
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
Experimental Section
Cathepsin B cleavage assay Cathepsin Bꢀcatalyzed linker cleavage activity was assessed by an
LC/MS, MRMꢀbased quantitation method. Briefly, varying concentrations of experimental linker (1.5ꢀ
fold serial dilution from 100 µM maximum down to 1.16 µM minimum) were incubated with 2 nM
human liver Cathepsin B (EMD Millipore, #219364), 25 nM Norfloxacinꢀd5 internal standard (CDN
Isotopes, #Dꢀ7196), 10 mM MES pH 6.0, and 1 mM DTT. The reaction was incubated for 1 hour at
37°C, followed by quenching with the addition of an equal volume of 2% formic acid in water. Free
Norfloxacin liberated by Cathepsin B and Norfloxacinꢀd5 internal standard were quantitated by an
LC/MS MRM method on a Sciex 5500 QTRAP mass spectrometer (Sciex, Framingham, MA) equipped
with a Waters Acquity Hꢀclass uHPLC. Norfloxacin and Norfloxacinꢀd5 were purified and desalted
using a Water BEHꢀC18 column, using a gradient of water/acetonitrile with 0.1% formic acid.
Measured Norfloxacin MRM AUC was normalized against Norfloxacinꢀd5 AUC, then plotted against
linker concentration using GraphPad Prism (GraphPad Software, La Jolla, CA). The resulting data was
fitted for Km and Vmax values using MichaelisꢀMenten equation. Each experiment was done in duplicate
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 37
for each linker concentration. MRM parameters for Norfloxacin: Q1=320, Q3=233.1, collision
energy=30; MRM parameters for Norfloxacinꢀd5: Q1=325, Q3=238.1, collision energy=30.
1
2
3
4
5
6
7
8
Computational modeling and crystallography Available human cathepsin B structures in Protein
Data Bank were analysed, and a structure bound covalently to a dipeptidyl nitrile inhibitor (PDB entry
1GMY) was selected to guide the design of nonꢀpeptidic, cathepsin Bꢀcleavable linkers. Specific
interactions between the inhibitor and cathepsin B observed from the crystallographic structure inspired
us to pursue two design strategies: 1) to replace the amide bond between the P1 and P2 moieties while
maintaining the positions/interactions of their side chains; 2) to replace the P2 moiety while keeping the
network of hydrogen bond interaction with residues Gly74 and Gly198 (Figure 1A and 1B). In both
cases, a “Link Multiple Fragments” search was performed in MOE (Chemical Computing Group,
v2012) using the default library to identify chemical structures that could connect the two fragments on
either side of the moiety being replaced. Using MOE, the hits were filtered by pharmacophores defined
by the observed interactions, energyꢀminimized in absence of the protein in order to identify highlyꢀ
strained conformations, and then visually inspected in the enzyme active site. For a selected subset of
candidate structures, the tetrahedral intermediate of linker hydrolysis with a pꢀamino benzyl alcohol
leaving group were built in MOE based on the reported catalytic mechanism of cathepsin B (Figure 1C
and 1D, Supporting Information Figure S1A and S1B).15,16 The Xꢀray crystallographic structure of 28
(crystallized and determined by Proteros Biostructures, Germany) was refined at 1.6Å resolution to
Rfree=17.6% in space group P21. The data were twinned (twin fraction = 48.5%, twin operator: ꢀH, ꢀK,
H+L). The coordinates were deposited under the accession code 6AY2 in Protein Data Bank.
Synthesis of chemicals Synthesis and characterization of compounds 1-36 are provided in
Supporting Information. For all biologically evaluated compounds, the purity was determined by HPLC
to be >95%.
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
Conjugation process and characterization by mass spectrometry Construction and production of
the THIOMABTM antibodies used were done as reported previously.24 Briefly, an alanineꢀtoꢀcysteine
ACS Paragon Plus Environment

Page 13 of 37
Journal of Medicinal Chemistry
mutation was made at 118 (EU numbering) position of the heavy chain of the antibody to produce the
corresponding THIOMABTM antibody. The THIOMABTM antibodies were conjugated to different linker
drugs as described previously.24 Briefly, the antibody was reduced in presence of fiftyꢀfold molar
excess DTT (Calbiochem) overnight. The reducing agent and the cysteine and glutathione blocks were
purified away using HiTrap SPꢀHP column (GE Healthcare). The antibody was reoxidized in presence
of fifteenꢀfold molar excess dhAA (MP Biomedical) for 2.5 hours. The formation of interchain disulfide
bonds was monitored by LC/MS. Linker drug in range of threefold to tenfold molar excess over protein
was incubated with the activated THIOMABTM antibodies for 3 or 16 hours. The antibody drug
conjugate was purified on HiTrap SPꢀHP column (GE Healthcare) to remove excess linker drug. The
number of conjugated linker drug molecules per THIOMABTM antibody was quantified by LC/MS
analysis. Purity was also assessed by size exclusion chromatography. LC/MS analysis was performed
on a 6530 AccurateꢀMass Quadrupole TimeꢀofꢀFlight (QꢀTOF) LC/MS (Agilent Technologies).
Samples were chromatographed on a PRLPꢀS, 1000 Å, 8 µm (50 mm 2.1 mm, Agilent Technologies)
heated to 80 °C. A linear gradient from 30–60% B in 4.3 minutes (solvent A, 0.05% TFA in water;
solvent B, 0.04% TFA in acetonitrile) was used and the eluent was directly ionized using the
electrospray source. Data was collected and deconvoluted using the Agilent Mass Hunter qualitative
analysis software. Before LC/MS analysis, antibody drug conjugate was treated with 15 mM DTT for
30 minutes at pH 8.0, and 37 °C to separate the HC and LC portion for ease of analysis.
Chromatographic conditions were chosen to achieve baseline resolution of LC, HC and HC+1 drug in
different peaks. The drug to antibody ratio (DAR) was calculated using the integrated peak area of the
UV chromatogram at 280 nm and orthogonally from the abundance of the ions present in LC/MS
deconvoluted results. The peaks were identified using LC/MS. All ADCs in this study employed
THIOMABTM antibodies and the DAR values were determined to be between 1.8 and 2.0.
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
Characterization of ADC activity in cell viability assay Cells were plated in blackꢀwalled 96ꢀwell
plates (4, 000 cells/100 µl for SKꢀBRꢀ3 and MCF7 cells) and allowed to adhere overnight at 37o C in a
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 37
humidified atmosphere of 5% CO2. Medium (Ham’s Fꢀ12: high glucose DMEM [50:50] supplemented
with 10% heatꢀinactivated fetal bovine serum and 2 mmol/L Lꢀglutamine) (all from Invitrogen Corp.)
was then removed and replaced by 100 µl fresh culture medium containing various concentrations of
each conjugate (conjugate stock [in 20 mM histidine acetate pH 5.5, 240 mM sucrose, 0.02% Tween 20]
diluted in medium). Cell TiterꢀGlo (Promega Corp.) was added to the wells at 5 days after drug
administration and the luminescent signal was measured using EnVision Multilabel Plate Reader
(PerkinElmer).
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
Quantitation of ADC payload release in cell lines and the effect of protease inhibitor treatment
Her2ꢀexpressing SKBR3 cells were seeded at were seeded at 2.0x105 cells/mL in 50uL DMEM culture
media supplemented with 10% fetal bovine serum and 2mM glutamine (Genentech Media Prep), in 96ꢀ
well Costar plates and incubated at 37°C overnight in a CO2 incubator. The next day, antibody drug
conjugates (ADC) with ValꢀCit or cBuꢀCit linkers were diluted with the culture media at 2X final
concentrations ranging from 0.16 ꢁg/ml to 20 ꢁg/ml were added to the plates, and incubated with cells
for 2 or 23 hrs. After incubation, close to 100ul of media was removed from each well and transferred to
a new deepꢀwell plate, then mixed with 300 µl of 100% acetonitrile per sample. Immediately after the
media removal, cell lysate samples were prepared with 200 µl of 75% acetonitrile to disrupt the cells
and precipitate the protein. All lysed samples were centrifuged at 4,000 x g for 15min at 4°C. After
centrifugation the 150uL supernant from cell lysate plate was transferred to a new plate, and 300uL
from media plate supernatants into fresh 96ꢀwell plate, both plates were placed in the Turbo Vap
concentration workstation to facilitate evaporation for and evaporated for 20 and 40min, respectively
(N2 gas flow rate 60 ft3/hr, heater temperature 50°C) using TurboVap® 96 (Biotage, Charlotte, NC).
Pellets were reconstituted in 120uL of water (Optima™ LC/MS, Fisher Chemical), and centrifuged at
4,000 x g for 15min at 4°C. The samples were then reconstituted with H2O. After centrifugation,
100uL of each supernatant was transferred to a new 96ꢀwell plate. All samples were than analyzed using
electrospray LCꢀMS/MS.
ACS Paragon Plus Environment

Page 15 of 37
Journal of Medicinal Chemistry
LCꢀMS/MS analysis protocol: 10 µL of the acidified samples were injected onto a Phenomenex Kinetex
50 x 2.1 mm i.d. XBꢀC18 reversed phase column coupled to an Sciex 6500 QTRAP mass spectrometer
system at a flow rate of 0.8 mL/min with the following gradient: 5% B (100% acetonitrile + 0.1%
formic acid) at 0ꢀ0.5ꢂmin; 70% B at 1ꢂmin; 95% B at 1.1ꢂmin. The mass spectrometer Turbo Spray
1
2
3
4
5
6
7
8
9
o
IonDrive Source temperature was 400 C and the ion spray voltage 3500 V. The source parameters
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
curtain gas, ion source gas 1 and ion source gas 2 were set at 35, 50 and 50 psi respectively. Multiple
reaction monitoring (MRM) transitions m/z 718.4 ꢀ 134.2 (CE 43 V, CXP 8V), 718.4 ꢀ 506.3 (CE 31
V, CXP 14V), and 718.4 ꢀ 686.3 (CE 39 V, CXP 18V) were used with dwell time 100 msec and
declustering potential 46 V. And for internal control Loperamide MRM m/z 477.15 ꢀ 266.2 (CE 35 V,
CXP 14V, declustering potential 116 V, dwell time 100 msec). Data was analyzed using MultiQuant
software, and graphs were generated using Microsoft excel and GraphPad Prism 6.
For the protease inhibitor treatment experiment, SKBR3 cells were seeded at 80,000 cells/well onto the
Costar plate as described above. ADC at 1 µg/ml was added onto the plate in the presence of 50 µM of
inhibitor. All ADCꢀinhibitor combinations were incubated with cells overnight. Cell lysates were
collected, prepared and then analyzed using electrospray LCꢀMS/MS as described above.
Inhibitors CAꢀ074_Me (catalog no. 205531), CAA0225 (catalog no. 2019502), and pepstatin (catalog
no. 516481) were purchased from EMD Millipore Corp, (Billerica, MA). BML_244 (catalog no.
ab141723) was purchased from Abcam (Cambridge, MA), and leupeptin (catalog no. 78435) was from
Fisher Scientific Co. (Pittsburgh, PA). The purities of the inhibitors were all above 95% by HPLC
according to the vendors.
Efficacy study in mouse xenograft models All animal studies were carried out in compliance with
National Institutes of Health guidelines for the care and use of laboratory animals and were approved by
the Institutional Animal Care and Use Committee at Genentech, Inc.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 37
The efficacy of antiꢀCD22 drug conjugates was evaluated in a mouse xenograft model of CD22ꢀ
expressing WSUꢀDLCL2 or Bjabꢀluc human nonꢀHodgkin lymphoma. The WSUꢀDLCL2 cell line was
obtained from DSMZ (German Collection of Microorganisms and Cell Cultures; Braunschweig,
Germany). The Bjabꢀluc cell line was obtained from Genentech cell line repository, and was
authenticated by short tandem repeat (STR) profiling using the Promega PowerPlex 16 System to
determine cell line ancestry.
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
To set up the WSUꢀDLCL2 or Bjabꢀluc xenograft model, tumor cells (20 million cells in 0.2mL
Hank’s Balanced Salt Solution; Hyclone) were inoculated subcutaneously into the flanks of female C.Bꢀ
17 SCID mice (Charles Rivers Laboratories). The efficacy of antiꢀNaPi2b drug conjugates was
investigated in a mouse xenograft model of NaPi2bꢀexpressing OVCAR3X2.1 human ovarian cancer.
The OVCAR3 cell line was obtained from ATCC (American Type Culture Collection; Manassas, VA)
and a subꢀline OVCAR3X2.1 was generated at Genentech for optimal growth in mice. To set up the
OVCAR3X2.1 xenograft model, tumor cells (10 million cells in 0.2mL Hank’s Balanced Salt Solution;
Hyclone) were inoculated in the thoracic mammary fat pad region of female C.Bꢀ17 SCIDꢀbeige mice
(Charles Rivers Laboratories). When the xenograft tumors reached the desired volume (day 0), animals
were divided into groups of 8 mice with similar mean tumor size and received a single intravenous
injection of antibody drug conjugates through the tail vein. The results were plotted as mean tumor
volume ± SEM of each group over time. Blood samples were collected via retroꢀorbital bleeds and used
to derive plasma for PK analyses.
Sparse PK and in vivo ADC stability Sparse pharmacokinetic sampling (collections at study days 1,
7 and 14 postꢀdose) was collected from efficacy study (12ꢀ3033A) to preliminarily assess exposure and
stability differences of these conjugates. The total antibody (TAB) is a measurement of the antibody,
and irrespective of the drug load, while the conjugated antibody (Conj. Ab) is measurement of the
antibody and stability of the linker. Both analytes were measured by ELISA as described previously.33
Total Antibody ELISA. Nunc® MaxiSorp™ 384ꢀwell plates (Nalge Nunc International, Rochester, NY)
ACS Paragon Plus Environment

Page 17 of 37
Journal of Medicinal Chemistry
were coated with 0.33 ꢁg/mL 6D3 antiꢀIdiotypic antibody (Genentech, South San Francisco, CA)
diluted in coat buffer (0.05 M carbonate/bicarbonate buffer pH 9.6) and incubated overnight at 4oC. The
plates were washed 3 times with wash buffer (0.5% Tweenꢀ20 in PBS buffer, pH 7.4) and treated with
block buffer (PBS/0.5% BSA/15 ppm Proclin, pH 7.4) for 1 to 2 hours. The plates were again washed 3
times with wash buffer and then samples diluted in sample diluent (PBS/0.5% BSA/0.05% Tween
20/5mM EDTA/0.25% CHAPS/ 0.35M NaCl/15 ppm Proclin, pH 7.4) were added to the wells and
incubated overnight at 4oC. The next day, the plates were brought to room temperature and then washed
6 times with wash buffer. A detection antibody, sheep antiꢀhuman IgG conjugated to horseradish
peroxidase (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA), diluted to 120 ng/mL in
assay buffer (PBS/0.5% BSA/15 ppm Proclin/0.05% Tween 20, pH7.4) was added to the wells and
incubated on a shaker for 1 hour at 4oC. The plates were washed 6 times with wash buffer and
developed using TMB peroxidase substrate (Moss Inc., Pasadena, Maryland) for 15 minutes followed
by 1M Phosphoric acid to stop the reaction. Absorbance was measured at 450 nm against a reference
wavelength of 620 nm. The assay range was 0.164 – 40 ng/mL with a minimum dilution of 1:100.
(LOD = 16.4ng/mL)
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
Supporting Information Available: Threeꢀdimensional coordinates of the models of the tetrahedral
intermediate of hydrolysis for compounds 7, 9 and 15, as well as the chain A of PDB entry 1GMY are
included as PDBꢀformatted files. Models for 7 and 9 are also shown in Figures S1A and S1B. Figure
S2 shows the body weight data for the efficacy study in the CD22ꢀexpressing lymphoma model WSUꢀ
DLCL2. Table S1 reports the crystallographic data statistics. Synthesis and characterization of
compounds 1-36 as well as a csv file of their SMILE strings are included.
Accession Codes: The Xꢀray crystallographic structure of 28 bound to human cathepsin B was
deposited in the PDB under the accession code 6AY2. Authors will release the atomic coordinates and
experimental data upon article publication.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 37
Corresponding Author Information: Tel: (650) 467ꢀ8283; fax: (650) 225ꢀ2061; email:
1
2
3
Wei.BinQing@gene.com
4
5
6
7
8
Acknowledgement: We thank Jack Sadowsky, Martine Darwish for helping in the preparation and
analysis of some materials used in the study, Susan Spencer and Genee Lee for the coordinating the
studies, and Peter S. Dragovich and Daniel F. Ortwine for reviewing the manuscript. Special thanks to
James Kiefer for twin refinement of the cathepsin B structure in complex with 28.
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
Abbreviations: ADC, antibodyꢀdrugꢀconjugate; ValꢀCit, valineꢀcitrulline; ValꢀAla, valineꢀalanine;
cBu, cyclobutaneꢀ1,1ꢀdicarboxamide; MMAE, monomethyl auristatin; PBD, pyrrolobenzodiazepine
dimer.
References:
(1)
(2)
(3)
(4)
Junutula, J. R.; Gerber, H.ꢀP. NextꢀGeneration AntibodyꢀDrug Conjugates (ADCs) for Cancer
Therapy. ACS Med. Chem. Lett. 2016, 7 (11), 972–973.
Saber, H.; Leighton, J. K. An FDA Oncology Analysis of AntibodyꢀDrug Conjugates. Regul.
Toxicol. Pharmacol. 2015, 71 (3), 444–452.
Donaghy, H. Effects of Antibody, Drug and Linker on the Preclinical and Clinical Toxicities of
AntibodyꢀDrug Conjugates. MAbs 2016, 8 (4), 659–671.
Shen, B.ꢀQ.; Xu, K.; Liu, L.; Raab, H.; Bhakta, S.; Kenrick, M.; ParsonsꢀReponte, K. L.; Tien, J.;
Yu, S.ꢀF.; Mai, E.; Li, D.; Tibbitts, J.; Baudys, J.; Saad, O. M.; Scales, S. J.; McDonald, P. J.;
Hass, P. E.; Eigenbrot, C.; Nguyen, T.; Solis, W. a; Fuji, R. N.; Flagella, K. M.; Patel, D.;
Spencer, S. D.; Khawli, L. a; Ebens, A.; Wong, W. L.; Vandlen, R.; Kaur, S.; Sliwkowski, M. X.;
Scheller, R. H.; Polakis, P.; Junutula, J. R. Conjugation Site Modulates the in Vivo Stability and
Therapeutic Activity of AntibodyꢀDrug Conjugates. Nat. Biotechnol. 2012, 30 (2), 184–189.
(5)
Dorywalska, M.; Strop, P.; MeltonꢀWitt, J. a.; HasaꢀMoreno, A.; Farias, S. E.; Galindo Casas,
M.; Delaria, K.; Lui, V.; Poulsen, K.; Loo, C.; Krimm, S.; Bolton, G.; Moine, L.; Dushin, R.;
Tran, T.ꢀT.; Liu, S.ꢀH.; Rickert, M.; Foletti, D.; Shelton, D. L.; Pons, J.; Rajpal, A. Effect of
Attachment Site on Stability of Cleavable Antibody Drug Conjugates. Bioconjug. Chem. 2015,
150220133437002.
(6)
(7)
Cohen, R.; Vugts, D. J.; Visser, G. W. M.; Stigterꢀvan Walsum, M.; Bolijn, M.; Spiga, M.;
Lazzari, P.; Shankar, S.; Sani, M.; Zanda, M.; van Dongen, G. A. M. S. Development of Novel
ADCs: Conjugation of Tubulysin Analogues to Trastuzumab Monitored by Dual Radiolabeling.
Cancer Res. 2014, 74 (20), 5700–5710.
Beck, A.; Wurch, T.; Bailly, C.; Corvaia, N. Strategies and Challenges for the next Generation of
Therapeutic Antibodies. Nat. Rev. Immunol. 2010, 10 (5), 345–352.
ACS Paragon Plus Environment

Page 19 of 37
Journal of Medicinal Chemistry
(8)
Dubowchik, G. M.; Firestone, R. A.; Padilla, L.; Willner, D.; Hofstead, S. J.; Mosure, K.; Knipe,
J. O.; Lasch, S. J.; Trail, P. A. Cathepsin BꢀLabile Dipeptide Linkers for Lysosomal Release of
Doxorubicin from Internalizing Immunoconjugates: Model Studies of Enzymatic Drug Release
and AntigenꢀSpecific in Vitro Anticancer Activity. Bioconjug. Chem. 2002, 13 (4), 855–869.
1
2
3
4
5
6
7
8
(9)
Senter, P. D.; Sievers, E. L. The Discovery and Development of Brentuximab Vedotin for Use in
Relapsed Hodgkin Lymphoma and Systemic Anaplastic Large Cell Lymphoma. Nat. Biotechnol.
2012, 30 (7), 631–637.
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
(10) Jain, N.; Smith, S. W.; Ghone, S.; Tomczuk, B. Current ADC Linker Chemistry. Pharmaceutical
Research. 2015, 3526–3540.
(11) Dal Corso, A.; Cazzamalli, S.; Gébleux, R.; Mattarella, M.; Neri, D. ProteaseꢀCleavable Linkers
Modulate the Anticancer Activity of Noninternalizing AntibodyꢀDrug Conjugates. Bioconjug.
Chem. 2017, 28 (7), 1826–1833.
(12) Singh, R.; Setiady, Y. Y.; Ponte, J.; Kovtun, Y. V; Lai, K. C.; Hong, E. E.; Fishkin, N.; Dong, L.;
Jones, G. E.; Coccia, J. A.; Lanieri, L.; Veale, K.; Costoplus, J. A.; Skaletskaya, A.; Gabriel, R.;
Salomon, P.; Wu, R.; Qiu, Q.; Erickson, H. K.; Lambert, J. M.; Chari, R. V; Widdison, W. C. A
New Triglycyl Peptide Linker for AntibodyꢀDrug Conjugates (ADCs) with Improved Targeted
Killing of Cancer Cells. Mol Cancer Ther 2016, 15 (6), 1311–1320.
(13) Puthenveetil, S.; He, H.; Loganzo, F.; Musto, S.; Teske, J.; Green, M.; Tan, X.; Hosselet, C.;
Lucas, J.; Tumey, L. N.; Sapra, P.; Subramanyam, C.; O’Donnell, C. J.; Graziani, E. I.
Multivalent Peptidic Linker Enables Identification of Preferred Sites of Conjugation for a Potent
Thialanstatin Antibody Drug Conjugate. PLoS One 2017, 12 (5), 1–16.
(14) Rawlings, N. D.; Barrett, A. J.; Finn, R. Twenty Years of the MEROPS Database of Proteolytic
Enzymes, Their Substrates and Inhibitors. Nucleic Acids Res. 2016, 44 (D1), D343–D350.
(15) Musil, D.; Zucic, D.; Turk, D.; Engh, R. a; Mayr, I.; Huber, R.; Popovic, T.; Turk, V.; Towatari,
T.; Katunuma, N. The Refined 2.15 A XꢀRay Crystal Structure of Human Liver Cathepsin B: The
Structural Basis for Its Specificity. EMBO J. 1991, 10 (9), 2321–2330.
(16) Illy, C.; Quraishi, O.; Wang, J.; Purisima, E.; Vernet, T.; Mort, J. S. Role of the Occluding Loop
in Cathepsin B Activity. J. Biol. Chem. 1997, 272 (2), 1197–1202.
(17) Mohamed, M. M.; Sloane, B. F. Cysteine Cathepsins: Multifunctional Enzymes in Cancer. Nat.
Rev. Cancer 2006, 6 (10), 764–775.
(18) Gondi, C. S.; Rao, J. S. Cathepsin B as a Cancer Target. Expert Opin. Ther. Targets 2013, 17 (3),
281–291.
(19) Roshy, S.; Sloane, B. F.; Moin, K. Pericellular Cathepsin B and Malignant Progression. Cancer
and Metastasis Reviews. 2003, 271–286.
(20) Mitrović, A.; Mirković, B.; Sosič, I.; Gobec, S.; Kos, J. Inhibition of Endopeptidase and
Exopeptidase Activity of Cathepsin B Impairs Extracellular Matrix Degradation and Tumour
Invasion. Biol. Chem. 2016, 397 (2), 164–174.
(21) Cezari, M. H. S.; Puzer, L.; Juliano, M. A.; Carmona, A. K.; Juliano, L. Cathepsin B
Carboxydipeptidase Specificity Analysis Using Internally Quenched Fluorescent Peptides.
Biochem. J. 2002, 368 (Pt 1), 365–369.
(22) Choe, Y.; Leonetti, F.; Greenbaum, D. C.; Lecaille, F.; Bogyo, M.; Brömme, D.; Ellman, J. A.;
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 37
Craik, C. S. Substrate Profiling of Cysteine Proteases Using a Combinatorial Peptide Library
Identifies Functionally Unique Specificities. J. Biol. Chem. 2006, 281 (18), 12824–12832.
1
2
3
4
5
6
7
8
9
(23) Greenspan, P. D.; Clark, K. L.; Tommasi, R. A.; Cowen, S. D.; McQuire, L. W.; Farley, D. L.;
Van Duzer, J. H.; Goldberg, R. L.; Zhou, H.; Du, Z.; Fitt, J. J.; Coppa, D. E.; Fang, Z.; Macchia,
W.; Zhu, L.; Capparelli, M. P.; Goldstein, R.; Wigg, A. M.; Doughty, J. R.; Bohacek, R. S.;
Knap, A. K. Identification of Dipeptidyl Nitriles as Potent and Selective Inhibitors of Cathepsin
B through StructureꢀBased Drug Design. J. Med. Chem. 2001, 44 (26), 4524–4534.
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
(24) Junutula, J. R.; Raab, H.; Clark, S.; Bhakta, S.; Leipold, D. D. D.; Weir, S.; Chen, Y.; Simpson,
M.; Tsai, S. P.; Dennis, M. S.; Lu, Y.; Meng, Y. G.; Ng, C.; Yang, J.; Lee, C. C.; Duenas, E.;
Gorrell, J.; Katta, V.; Kim, A.; McDorman, K.; Flagella, K.; Venook, R.; Ross, S.; Spencer, S.
D.; Lee Wong, W.; Lowman, H. B.; Vandlen, R.; Sliwkowski, M. X.; Scheller, R. H.; Polakis, P.;
Mallet, W. SiteꢀSpecific Conjugation of a Cytotoxic Drug to an Antibody Improves the
Therapeutic Index. Nat. Biotechnol. 2008, 26 (8), 925–932.
(25) Vollmar, B. S.; Wei, B.; Ohri, R.; Zhou, J.; He, J.; Yu, S.ꢀF.; Leipold, D.; Cosino, E.; Yee, S.;
FourieꢀO’Donohue, A.; Li, G.; Lewis Phillips, G. D.; Kozak, K. R.; Kamath, A.; Xu, K.; Lee, G.;
Lazar, G. A.; Erickson, H. K. Attachment Site Cysteine Thiol pKa Is a Key Driver for Siteꢀ
Dependent Stability of THIOMAB TM AntibodyꢀDrug Conjugates. Bioconjug. Chem. 2017, 28
(10), 2538ꢀ2548.
(26) Doronina, S. O.; Toki, B. E.; Torgov, M. Y.; Mendelsohn, B. a; Cerveny, C. G.; Chace, D. F.;
DeBlanc, R. L.; Gearing, R. P.; Bovee, T. D.; Siegall, C. B.; Francisco, J. a; Wahl, A. F.; Meyer,
D. L.; Senter, P. D. Development of Potent Monoclonal Antibody Auristatin Conjugates for
Cancer Therapy. Nat. Biotech. 2003, 21 (7), 778–784.
(27) Montaser, M.; Lalmanach, G.; Mach, L. CAꢀ074, but Not Its Methyl Ester CAꢀ074Me, Is a
Selective Inhibitor of Cathepsin B within Living Cells. Biol. Chem. 2002, 383 (7–8), 1305–1308.
(28) Takahashi, K.; Ueno, T.; Tanida, I.; MinematsuꢀIkeguchi, N.; Murata, M.; Kominami, E.
Characterization of CAA0225, a Novel Inhibitor Specific for Cathepsin L, as a Probe for
Autophagic Proteolysis. Biol. Pharm. Bull. 2009, 32 (3), 475–479.
(29) Desmarais, S.; Massé, F.; Percival, M. D. Pharmacological Inhibitors to Identify Roles of
Cathepsin K in CellꢀBased Studies: A Comparison of Available Tools. Biol. Chem. 2009, 390 (9),
941–948.
(30) Marciniszyn, J.; Hartsuck, J. A.; Tang, J. Mode of Inhibition of Acid Proteases by Pepstatin. J.
Biol. Chem. 1976, 251 (22), 7088–7094.
(31) Libby, P.; Goldberg, a L. Leupeptin, a Protease Inhibitor, Decreases Protein Degradation in
Normal and Diseased Muscles. Science (80-. ). 1978, 199 (4328), 534–536.
(32) Gregson, S. J.; Howard, P. W.; Hartley, J. A.; Brooks, N. A.; Adams, L. J.; Jenkins, T. C.;
Kelland, L. R.; Thurston, D. E. Design, Synthesis, and Evaluation of a Novel
Pyrrolobenzodiazepine DNAꢀInteractive Agent with Highly Efficient CrossꢀLinking Ability and
Potent Cytotoxicity. J. Med. Chem. 2001, 44 (5), 737–748.
(33) Kaur, S.; Xu, K.; Saad, O. M.; Dere, R. C.; CarrascoꢀTriguero, M. Bioanalytical Assay Strategies
for the Development of Antibody–drug Conjugate Biotherapeutics. Bioanalysis 2013, 5 (2), 201–
226.
ACS Paragon Plus Environment

Page 21 of 37
Journal of Medicinal Chemistry
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

Journal of Medicinal Chemistry
Page 22 of 37
Scheme 1. A simplified molecular construct for measuring the cleavage of ADC linker moieties by
protease. Reduction of ADC to the simplified construct is illustrated using the dipeptide linker ValꢀCit
(compound 1).
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

Page 23 of 37
Journal of Medicinal Chemistry
Table 1. Vmax and Km properties for the cleavage of representative dipeptide linkers by cathepsin
B measured using the simplified constructs.
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
b
aThe concentration of human cathepsin B enzyme was kept constant at 2nM. Vmax and Km values were
obtained by fitting the cleavage assay data (duplicate, n=1) using MichaelisꢀMenten equation. All
values are rounded to two significant digits (95% confidence intervals given in parentheses).
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 37
Table 2. Vmax and Km properties for the cleavage of fluoro-olefin linkers by cathepsin B measured
using the simplified constructs.
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
b
aThe concentration of human cathepsin B enzyme was kept constant at 2nM. Vmax and Km values were
obtained by fitting the cleavage assay data (duplicate, n=1) using MichaelisꢀMenten equation. All
values are rounded to two significant digits (95% confidence intervals given in parentheses).
ACS Paragon Plus Environment

Page 25 of 37
Journal of Medicinal Chemistry
Table 3. Vmax and Km properties for the cleavage of triazole linkers by cathepsin B measured
using the simplified constructs.
1
2
3
4
5
6
7
8
9
O
R2
O
O
O
N
O
N
N
F
N
O
H
N
N
N
H
N
O
R1
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
OH
Cathepsin B clevagea
Ex
9
R1
R2
b
b
Vmax/Km
Vmax
(pmol/min)
Km
(ꢁM)
1.3Eꢀ3 (8.1Eꢀ4, 1.9Eꢀ3)
0.020 (0.018, 0.022)
6.7 (2.8, 11)
3.5 (2.1, 4.9)
2.0Eꢀ4
5.7Eꢀ3
NH2
HN
O
10
NH2
NH
HN
0.031 (0.026, 0.036)
3.5Eꢀ3
1.1Eꢀ4
8.8 (4.1, 13)
31 (14, 48)
11
12
NH2
3.4Eꢀ3 (1.4Eꢀ3, 5.3Eꢀ3)
NH2
NDc
NDc
NDc
13
NH2
HN
O
aThe concentration of human cathepsin B enzyme was kept constant at 2nM. Vmax and Km values were
obtained by fitting the cleavage assay data (duplicate, n=1) using MichaelisꢀMenten equation. All
values are rounded to two significant digits (95% confidence intervals given in parentheses). cNot
determined because the cleavage activity was too low to be measured under the assay condition.
b
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 37
1
2
3
4
Table 4. Vmax and Km properties for the cleavage of cycloalkyl-1,1-dicarboxamide linkers by
cathepsin B measured using the simplified constructs.
aThe concentration of human cathepsin B enzyme was kept constant at 2nM. Vmax and Km values were
5
6
7
8
b
obtained by fitting the cleavage assay data (duplicate, n=1) using MichaelisꢀMenten equation. All
values are rounded to two significant digits (95% confidence intervals given in parentheses). cNot
d
determined because the cleavage activity was too low to be measured under the assay conditions. 25
9
and 26 are a pair of diastereomers with unknown stereochemistry. e28 is a single diastereomer with
unknown stereochemistry at one of the chiral centers. It was used in the crystallography study as a
mimic of the cBuꢀCit linker.
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

Page 27 of 37
Journal of Medicinal Chemistry
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

Journal of Medicinal Chemistry
Page 28 of 37
1
2
3
4
5
Table 5. Structures of the linker-MMAE molecules used to prepare the ADCs tested in biological
studies. All ADCs were conjugated via the maleimide group to an engineered cysteine at position 118
(EU numbering) on the two heavy chains of the respective antibodies.
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

Page 29 of 37
Journal of Medicinal Chemistry
Table 6. Structures of the linker-PBD molecules used to prepare the ADCs tested in biological
studies. All ADCs were conjugated via the maleimide group to an engineered cysteine at position 118
(EU numbering) on the two heavy chains of the respective antibodies.
1
2
3
4
5
6
7
O
O
O
O
O
O
N
N
R =
H
H
N
N
8
HO
9
Pyrrolobenzodiazepine dimer (PBD)
Structure
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
Ex
Alias
R
O
O
O
O
H
O
35 ValꢀCit PBD
N
N
N
H
N
H
O
O
HN
O
NH2
R
O
O
O
O
36 cBuꢀCit PBD
H
H
N
N
N
N
H
O
O
O
HN
O
NH2
ACS Paragon Plus Environment