Intracellular Trapping of the Selective Phosphoglycerate Dehydrogenase (PHGDH) Inhibitor BI-4924 Disrupts Serine Biosynthesis

Article abstract of DOI:10.1021/acs.jmedchem.9b00718

Phosphoglycerate dehydrogenase (PHGDH) is known to be the rate-limiting enzyme in the serine synthesis pathway in humans. It converts glycolysis-derived 3-phosphoglycerate to 3-phosphopyruvate in a co-factor-dependent oxidation reaction. Herein, we report the discovery of BI-4916, a prodrug of the co-factor nicotinamide adenine dinucleotide (NADH/NAD⁺)-competitive PHGDH inhibitor BI-4924, which has shown high selectivity against the majority of other dehydrogenase targets. Starting with a fragment-based screening, a subsequent hit optimization using structure-based drug design was conducted to deliver a single-digit nanomolar lead series and to improve potency by 6 orders of magnitude. To this end, an intracellular ester cleavage mechanism of the ester prodrug was utilized to achieve intracellular enrichment of the actual carboxylic acid based drug and thus overcome high cytosolic levels of the competitive cofactors NADH/NAD⁺

Full text of DOI:10.1021/acs.jmedchem.9b00718

Subscriber access provided by UNIV OF SOUTHERN INDIANA
Article
Intracellular Trapping of the Selective Phosphoglycerate
Dehydrogenase (PHGDH) Inhibitor BI-4924 Disrupts Serine Biosynthesis
Harald Weinstabl, Matthias Treu, Jörg Rinnenthal, Stephan Zahn, Peter Ettmayer, Gerd Bader, Georg
Dahmann, Dirk Kessler, Klaus Rumpel, Nikolai Mischerikow, Fabio Savarese, Thomas Gerstberger,
Moriz Mayer, Andreas Zoephel, Renate Schnitzer, Wolfgang Sommergruber, Paola Martinelli, Heribert
Arnhof, Biljana Peric Simov, Karin Stephanie Hofbauer, Géraldine Garavel, Yvonne Scherbantin, Sophie
Mitzner, Thomas Fett, Guido Scholz, Jens Bruchhaus, Michelle Burkard, Roland Kousek, Tuncay Ciftci,
Bernadette Sharps, Andreas Schrenk, Christoph Harrer, Daniela Haering, Bernhard Wolkerstorfer, Xuechun
Zhang, Xiaobing Lv, Alicia Du, Dongyang Li, Yali Li, Jens Quant, Mark Pearson, and Darryl B. McConnell
J. Med. Chem., Just Accepted Manuscript • Publication Date (Web): 31 Jul 2019
Downloaded from pubs.acs.org on July 31, 2019
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 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 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.
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 130
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Lv, Xiaobing; ChemPartner
Du, Alicia; ChemPartner
Li, Dongyang; ChemPartner
Li, Yali; ChemPartner
Quant, Jens; Boehringer Ingelheim RCV GmbH & Co KG
Pearson, Mark; Boehringer Ingelheim RCV GmbH und Co KG
McConnell, Darryl; Boehringer Ingelheim RCV GmbH und Co KG
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 130
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Intracellular Trapping of the Selective
Phosphoglycerate Dehydrogenase (PHGDH)
Inhibitor BI-4924 Disrupts Serine Biosynthesis
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Harald Weinstabl*, §,‡, Matthias Treu , Joerg Rinnenthal , Stephan K. Zahn , Peter
§, ‡
§
§
§
§
$
§
§
Ettmayer , Gerd Bader , Georg Dahmann , Dirk Kessler , Klaus Rumpel , Nikolai
§
§
§
§
Mischerikow , Fabio Savarese , Thomas Gerstberger , Moriz Mayer , Andreas
§
§
§
§
Zoephel , Renate Schnitzer , Wolfgang Sommergruber , Paola Martinelli ,Heribert
§
§
§
§
Arnhof , Biljana Peric-Simov , Karin S. Hofbauer , Géraldine Garavel , Yvonne
§
§
§
§
§
Scherbantin , Sophie Mitzner , Thomas N. Fett , Guido Scholz , Jens Bruchhaus ,
§
§
$
§
Michelle Burkard , Roland Kousek , Tuncay Ciftci , Bernadette Sharps , Andreas
§
§
§
§
Schrenk , Christoph Harrer , Daniela Haering , Bernhard Wolkerstorfer , Xuechun
#
#
#
#
#
§
§
Zhang , Xiaobing Lv , Alicia Du , Dongyang Li , Yali Li , Jens Quant , Mark Pearson ,
Darryl B. McConnell§
ACS Paragon Plus Environment
1

Page 3 of 130
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
§Boehringer Ingelheim RCV GmbH & Co KG; Dr.-Boehringer-Gasse 5-11, 1121 Vienna,
Austria
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
$Boehringer Ingelheim Pharma GmbH & Co. KG; Birkendorfer Str. 65, 88400 Biberach
an der Riß, Germany
#Shanghai ChemPartner Co., LTD., No. 5 Building, 998 Halei Road, Zhangjiang Hi-Tech
Park Pudong New Area, Shanghai China, 201203
+
PHGDH, inhibitor, NAD competitive, (tumor) metabolism, cancer, in vitro PK-PD
ABSTRACT
Phosphoglycerate dehydrogenase (PHGDH) is known to be the rate limiting enzyme in
the serine synthesis pathway (SSP) in humans. It converts glycolysis derived 3-
+
phosphoglycerate to 3-phosphopyruvate in a NADH/NAD -dependent oxidation
ACS Paragon Plus Environment
2

Journal of Medicinal Chemistry
Page 4 of 130
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
reaction. Herein we report the discovery of BI-4916, a prodrug of the NADH/NAD+-
competitive PHGDH inhibitor BI-4924 which has shown high selectivity against the
majority of other dehydrogenase targets. Starting with a fragment-based screening
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(FBS) a subsequent hit optimization using structure based drug design (SBDD) was
conducted to deliver a single digit nanomolar lead series and to improve potency by six
orders of magnitude. To this end, an intracellular ester cleavage mechanism of the ester
prodrug was utilized to achieve intracellular enrichment of the actual carboxylic acid
based drug and thus overcome high cytosolic levels of the competitive cofactors
NADH/NAD+.
INTRODUCTION
Cancer cells can rewire their metabolism to support fast growth and proliferation, and
indeed the “deregulation of cellular energetics” has been included among the hallmarks
1
of cancer. Metabolic enzymes have therefore raised interest as potential targets for
therapies aimed at “fasting” cancer cells.
ACS Paragon Plus Environment
3

Page 5 of 130
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Phosphoglycerate dehydrogenase (PHGDH), an enzyme in the serine synthesis
2
3,
pathway, has been found to be involved in numerous cancers such as breast cancer,
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
4
5
6
7, 8
9
melanoma, pancreatic cancer, kidney cancer, and non-small cell lung cancer, as
10
well as in pulmonary fibrosis. PHGDH, the gene coding for phosphoglycerate
dehydrogenase, was found to be focally amplified or gained in melanoma and breast
cancer,3, 4 where it is broadly overexpressed in about 70% of triple-negative cases even
in the absence of amplification. Cells with PHGDH amplification and overexpression
were shown to be highly sensitive to its depletion. A subgroup of PHGDH-
overexpressing tumors was also observed among non-small cell lung cancer and lung
adenocarcinoma cases and also in this case PHGDH knock-down selectively impaired
the growth of overexpressing cells, suggesting an oncogenic function of PHGDH.9, 11
In more detail, a negative-selection RNAi screening using an orthotopic human breast
cancer xenograft model was developed by Possemato et al. in 2011 identifying PHGDH
3
as a gene required for in vivo tumorigenesis. in addition, PHGDH was shown to be
localized in a genomic region of recurrent copy number gain in breast cancer
ACS Paragon Plus Environment
4

Journal of Medicinal Chemistry
Page 6 of 130
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
1
2
particularly in ER-negative or triple negative breast cancers. Similarly, PHGDH is
recurrently amplified in a genomic region of focal copy number gain most commonly
found in melanoma. Knockdown of PHGDH selectively inhibited the growth of
melanoma cells that exhibit PHGDH amplification versus those that lack this
amplification.4, 5 PHGDH protein expression was recently correlated with advanced TNM
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
13
stage, larger tumor size and poor outcomes in colorectal cancer patients. Studies
utilizing a large panel of NSCLC cell lines revealed that NRF2 controls the expression of
the PHGDH gene and that PHGDH confers poor prognosis in human NSCLC.9
Recently, a strong antiproliferative effect was shown in a pancreatic xenograft model by
inhibition of PHGDH with IoxA, a natural withanolide.14
PHGDH is the rate limiting enzyme in the serine synthesis pathway in humans which
diverts 3-phosphoglycerate from glycolytic serine biosynthesis by oxidizing it to 3-
+
2
phosphopyruvate in a NADH/NAD -dependent manner (Figure 1). Subsequently, 3-
phosphopyruvate is converted to serine by phosphoserine transaminase and
phosphoserine hydrolase. Serine is regarded as a non-essential amino acid, however, it
ACS Paragon Plus Environment
5

Page 7 of 130
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
is indispensable and plays a critical role in several cellular processes such as the
conversion to glycine supporting the backbone of central carbon metabolism (carbon
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
15
unit for purine synthesis), the generation of sphingosine via reaction with palmitoyl-
16
17
CoA, the regulation of redox potential (production of NADPH) and as a precursor of
16
other amino acids (e.g. cysteine) . It is apparent that cells with PHGDH amplification
and/or high expression are not sensitive to serine depletion, indicating the loss of
proliferation upon knockdown of PHGDH is not simply due to reduction in serine levels
3
per se. Although a plethora of published data supports the contribution of the serine
synthesis pathway to tumorigenesis, in particular the role of the key enzyme PHGDH,
only recently the precise molecular mechanism of PHGDH in oncogenesis has been
characterized in detail. Inhibition of PHGDH alters nucleotide metabolism in a way that
is independent of serine utilization and redox maintenance. Mechanistically, it rather
results in disruption of mass balance within central carbon metabolism leading to
alterations in both the pentose phosphate pathway (PPP) and the tri-carboxylic acid
18
cycle (TCA). Indeed, restoration of the TCA cycle and PPP through addition of ribose
and cell-permeable α-ketoglutarate to PHGDH-inhibitor treated tumor cells was
ACS Paragon Plus Environment
6

Journal of Medicinal Chemistry
Page 8 of 130
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
sufficient to rescue the antiproliferative effect generated by PHGDH inhibition. In their
study, Reid and colleagues clearly showed that the mass balance and the requirement
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
of maintaining flux balance is sufficient to regulate the flux into other neighboring
anabolic pathways such as glycolysis and the TCA cycle.18
The lack of serine derived glutathione (GSH) has been reported to underlie the
increased sensitivity towards reactive oxygen species (ROS) of tumor cells lacking
19
PHGDH. PHGDH has been further demonstrated to be able to use α-ketoglutarate as
a substrate to produce the oncometabolite D-2-hydroxyglutarate (D-2-HG), perhaps
linking PHGDH activity with changes in the epigenetic program of amplified tumor
20
cells. However, as the PHGDH enzymatic activity towards 2-hydroxyglutarate
production is extremely low compared to 3-phosphohydroxypyruvate it remains to be
determined if those levels of 2-hydroxyglutarate produced by PHGDH do play a similar
role in the epigenetic re-programming as it was shown for the mutated isocitrate-
dehydrogenase 1 and 2 in glioma and AML.21, 22 (Figure 1)
ACS Paragon Plus Environment
7

Page 9 of 130
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 1: Known roles of PHGDH in tumor metabolism and tumor epigenetics
In preclinical models, breast and pancreatic adenocarcinomas have been shown to be
dependent on the SSP suggesting that PHGDH may represent a therapeutic
2
intervention point. Developing highly potent and specific PHGDH inhibitors will promote
understanding of the molecular mechanisms triggering the dependency of certain tumor
(sub)types on this pathway but also opens a novel therapeutic opportunity to treat those
tumors. This emphasizes the need for compounds exhibiting low nanomolar cellular
potency and exquisite specificity to enable better understanding of the target in a
cellular context.
ACS Paragon Plus Environment
8

Journal of Medicinal Chemistry
Page 10 of 130
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
The challenge when targeting metabolic enzymes in a competitive fashion is
+
intrinsically tied to competing with millimolar intracellular levels of the cofactor NAD and
NADH accompanied by the fact that modulation of metabolic targets often need high
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
23
levels of target occupancy to achieve efficacy. Preceding in house target validation
experiments and data published by Chen et. al. strengthen this hypothesis. The authors
showed that > 90% PHGDH knockdown over 14 days was anti-proliferative in PHGDH
high breast cancer cell lines and in line with this observation < 90% knockdown in vivo
was not associated with tumor growth inhibition – indeed defining the level and duration
of inhibition of a given target required for efficacy is essential in order to design
24
appropriate drugs. Furthermore, PHGDH exhibits a polar shallow cofactor/substrate
binding pocket which makes the druggability of PHGDH extremely challenging.
Several compounds are reported in literature to inhibit PHGDH which can be divided
25
26
in two groups: allosteric inhibitors, such as CBR-5884 (2), NCT-503 (3), “compound
34” (5) and PKUMDL-WQ-2201 (7) display a micromolar potency and are not
dependent on intracellular cofactor levels.27, 28 On the other hand, orthosteric inhibitors,
ACS Paragon Plus Environment
9

Page 11 of 130
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
such as Astra Zeneca’s indole-2-carboxamide 1 and fragment 6 are validated
adenosine pocket binders thus underlying the competitive binding of the natural cofactor
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
NADH/ NAD+.29, 30 Compound 4 discovered by RAZE pharmaceuticals is likely to share
the same binding mode due to structural similarity.31
+
Herein we report on the discovery of a prodrug-based NADH/NAD -competitive
PHGDH inhibitor with high selectivity (based on proteomics ANOVA analysis) against
the majority of other dehydrogenase targets. An intracellular ester cleavage mechanism
of the permeable ester prodrug BI-4916 was utilized to achieve intracellular enrichment
of the actual carboxylic acid based poorly permeable drug BI-4924 and thus to
+
overcome high intracellular levels of the competitive cofactor NADH/NAD . Compound 1
+
and BI-4924 are X-ray validated NAD competitive binders. Compounds 3 (NCT-503;
thiourea) and 5 (“compound 34”; thioamide) and 7 (thiosemicarbazide) exhibit
potentially reactive PAINS motives.25-33
ACS Paragon Plus Environment
10

Journal of Medicinal Chemistry
Page 12 of 130
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
N
Previous work:
HO
S
S
F
Cl
O
HN
O
F
F
O
H
N
N
H
N
N
HN
S
NH
HO
N
O
O
O
OH
O
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
O
2
3
Fuller et. al. (ref. 29):
published AZ PHGDH inhibitor
Kd, PHGDH = 180 nM
Mullarky et. al. (ref. 25):
CBR-5884
IC50, PHGDH = 33 µM
Pacold et. al. (ref. 26):
NCT-503
IC50, PHGDH = 2.5 µM
H
N
HO
H
N
O
Cl
O
O
O
Cl
N
O
OH
Cl
O
HN
N
NH
2
S
Cl
HN
N
S
O
O
S
F
O
O
2
N
N
O
4
5
6
7
Mainolfi et. al. (ref. 31):
published RAZE PHGDH inhibitor
0.01 µM < IC50, PHGDH (500 µM NAD) < 1.5 µM
Ravez et. al. (ref. 27):
compound 34
IC50, PHGDH = 35.1 µM
Unterlass et. al. (ref. 30):
compound 5
Kd, PHGDH = 1.5 mM
Wang et. al. (ref. 28):
PKUMDL-WQ-2201
IC50, PHGDH = 35.7 µM
This work:
HO
HO
Cl
Cl
intracellular
hydrolysis
Cl
HN
O
Cl
HN
O
N
O
N
O
OH
O
S
S
BI-4916
BI-4924
O
O
O
O
IC50, PHGDH = 2 nM
Figure 2: Overview of selected published PHGDH inhibitors and reported
binding/inhibition data.
RESULTS AND DISCUSSION
Screening:
ACS Paragon Plus Environment
11

Page 13 of 130
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Screening was performed in two different hit finding approaches (Figure 3). First, a
high throughput screen (HTS) was performed based on a pool consisting of > 1,000,000
molecular entities. Within this approach, no boundaries were set and allosteric as well
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
+
as orthosteric binders were evaluated with a biochemical NAD -dependent screening
cascade. Second, a fragment based screen (FBS) was conducted in parallel.
ACS Paragon Plus Environment
12

Journal of Medicinal Chemistry
Page 14 of 130
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 3: Schematic representation of the hit finding process. Saturation Transfer
Difference NMR (STD-NMR) guided based FBS screening on the right (blue) and
luciferase based HTS screening on the left (grey).
High Throughput Screen (HTS): HTS yielded 27,000 primary hits (2.2% hit rate). The
confirmed hits (15,200; 56% of primary hits) were profiled in a shortcut assay containing
ACS Paragon Plus Environment
13

Page 15 of 130
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
no PHGDH, and dual hitters were removed as false positives (11,250 compounds
remaining; 74% of confirmed hits). These hits were triaged to 4,750 (42%) hits by
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
33
removal of unwanted substructures (e.g. PAINS) and compounds with MW ≥ 600 and
clogP ≥ 5.5. For these triaged hits, dose response curves were measured in PHGDH,
lactate dehydrogenase (subtype A; LDHA) and shortcut assays. Thus, 2,041 potent and
selective PHGDH inhibitors were progressed to cluster analysis. 323 representative hits
from 9 classes and singletons of interest were prioritized. The purity of 72 out of 104
class representatives was confirmed by LC/MS using freshly prepared DMSO stock
solutions. 43 out of these 72 were confirmed as selective PHGDH inhibitors (> 10-fold
selectivity towards LDHA) with an IC < 100 µM. Due to the fact that many of these hits
50
(triazoles, aryl-pyrazoles) have been synthesized employing transition metal catalysis
34
and that dehydrogenases often show a high sensitivity towards transition metals, a
selection of metal ions was tested in the biochemical assays. It was discovered that
transition metal ions (e.g. Hg, Cu, Pt, Pd) are potent inhibitors of PHGDH and indeed,
for 5 out of 12 selected HTS hits Hg and Cu impurities were quantified (68 elements
screened by ICP-MS) in sufficient concentrations to explain the observed PHGDH
ACS Paragon Plus Environment
14

Journal of Medicinal Chemistry
Page 16 of 130
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
inhibition. In line with this finding, it was not surprising that 7 out of 10 resynthesized
and purified HTS hits lost their PHGDH inhibition.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
+
As a consequence, STD-NMR with and without NAD and SPR were applied for
orthogonal hit validation. The latter was found to tolerate metal impurities (only Pd2+
shows a measureable K of significance) and the paramagnetic nature of copper leads
d
35
to line broadening in STD-NMR. As a next step, we went back to the 4,750 confirmed
actives and performed a cluster analysis to selected 3,230 (68%) hits for re-screening at
50 µM for specific binding to PHGDH in the SPR assay. From the identified 323 (10%)
PHGDH binders, 77 had a K < 300 µM which were again clustered according to their
d
chemical structure. This orthogonal HTS hit validation strategy translated in >80 % SPR
K confirmation for the re-synthesized cluster representatives and resulted in three
d
chemical series with X-ray derived binding mode information among which compound 8
(indole-2-carboxamide) is a representative example.
Fragment based screening (FBS): A truncated version of human PHGDH according to
a previously published structure (PDB code: 2G76,
ACS Paragon Plus Environment
15

Page 17 of 130
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
https://www.thesgc.org/structures/2g76) was designed (A4-V315; see supporting
information), expressed and purified in gram quantities. In contrast to the full length
protein (tetramer), it was found to be active as a dimer in solution and was also used for
the biochemical assay as well as crystallographic analysis. FBS aimed at finding NAD-
orthosteric binders, thus a NAD stripping protocol was developed for the preparation of
PHGDH apo-protein. An in-house library of 1,536 fragments was complemented by a
customized library of 324 soluble (> 100 µM aqueous solubility) based on fragments
selected based on learnings from a LDHA fragment based screening campaign.34
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
A total of 1,860 fragments were screened in quadruplex mixtures by STD-NMR (250
µM fragment and 20 µM apo-protein) to reduce the time requirements of the screening
campaign. STD-NMR spectra for the individual hits from 10 mixtures with > 40% STD
signal and 59 mixtures with > 20% STD signal were recorded with and without NAD (2
mM). 60 hits (3.2% hit rate) were identified that bind preferentially to the apo-protein and
in a less pronounced manner to the NAD bound protein. SPR-K -values for PHGDH,
d
ACS Paragon Plus Environment
16

Journal of Medicinal Chemistry
Page 18 of 130
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
lactate dehydrogenase, and carbonic anhydrase II were measured for these 60 compounds,
resulting in the identification of 19 specific PHGDH binders (32% hit rate).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
The development of a low temperature co-crystallization protocol allowed the X-ray-
validation of an adenine pocket binder 9 out of the phenylpyrazole-5-carboxamide
series (2.15 Å resolution). This fragment shows a strong interaction with D174 and thus
+
exhibits suitable vectors along the NAD -binding site (Figure 4).
ACS Paragon Plus Environment
17

Page 19 of 130
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Figure 4: Validated hit from the STD-NMR FBS screen: phenylpyrazole-5-carboxamide
9
6
co-crystallized with PHGDH showing the main polar interaction to D174 (PDB code:
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
RIH).
All hits identified bind orthosterically to the lipophilic adenine pocket and form a H-
bond to D174.
Structural Analysis of PHGDH in Complex with its natural Cofactor and Substrate
Co-crystallization of PHGDH with its natural substrate 3-PG was performed to gain
more insight into the catalytic mechanism of the enzyme and to enable the structure
+
based design of transition state analogues. The binding mode of the cofactor NAD was
found to be identical to previously published structures of human PHGDH (Figure 5).
The sequence of murine PHGDH shows a high degree of similarity (98.5%) as well as
identity (94.6%). Briefly, the nicotinamide (NA) interacts with the backbone of C233 and
A285 and the sidechain of D259. The ribose of the nicotinamide part as well as the
ACS Paragon Plus Environment
18

Journal of Medicinal Chemistry
Page 20 of 130
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
linking phosphates form numerous direct and water mediated hydrogen bonds with the
protein. The main directed interaction of the adenosine moiety is the hydrogen bonding
of the ribose hydroxyl groups with the side chain of D174. In a recent publication, L-
tartrate, derived from the crystallization buffer, bound in one of the two active sites and
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
36
was discussed as an analogue of the natural substrate. In our structure (PDB code:
6CWA), 3-PG indeed binds to the same positon as the L-tartrate. The phosphate group
forms extensive interactions with the side chains of R53, S54 and R74 of monomer A as
well as R134 of monomer B. Also, the rest of the 3-PG moiety is part of an extensive
hydrogen bonding network with the active site of PHGDH.
ACS Paragon Plus Environment
19

Page 21 of 130
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
+
Figure 5: Electron density map of ternary complex of cofactor NAD and substrate 3-PG
in PHGDH (PDB code: 6CWA)
Structure Based Compound Optimization
ACS Paragon Plus Environment
20

Journal of Medicinal Chemistry
Page 22 of 130
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Scheme 1: General Overview of the hit optimization
O
OH
long range
electrostatic
interactions (LRE)
1,3-allylic strain
H
N
HN
O
H
N
(R)
N
H
N
N
N
(R)
N
N
O
N
N
O
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
O
HTS hit
compound 8
KD, SPR, PHGDH = 144 µM
11
FBS hit
compound 9
KD, SPR, PHGDH = 105 µM
15
XRAY available
OH
O
3rd sphere
acid isosters
knowledge transfer
from FBS series
Cl
adenosine pocket optimization
+ tosyl acetate introduction
H
N
HN
O
(R)
Cl
HN
O
(R)
core switch
H
N
tosyl acetate introduction
N
(R)
N
(R)
N
N
N
N
O
O
40
XRAY available
OH
58
XRAY available
S
4
4
OH
O
43
O
S
O
O
NH
O
O
S
O
O
O
OH
chiral methyl
optimization
O
bidentate D174
chelation
HO
HO
Cl
Cl
Cl
HN
O
(S)
Cl
HN
(S)
prodrug strategy
N
O
N
O
O
O
OH
S
BI-4916
O
BI-4924
XRAY available
S
O
O
O
Double tracked strategy: Merging learnings from HTS and FBS leads to BI-4924 and BI-
4916
Both hit classes – (phenylpyrazole-5-carboxamides from the FBS and indole-2-
carboxamides from the HTS (Scheme 1) – form a H-bond to the side-chain of D174 and
the amide NH of the ligand. This acts as a polar anchor to place a lipophilic aromatic
motif in the corresponding hydrophobic hotspot, referred to as the adenine pocket. The
phenyl pyrazole and the indole motif occupy the same part of the adenine pocket with
ACS Paragon Plus Environment
21

Page 23 of 130
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
similar orientation of the corresponding N-methyl groups and the carboxamide NH
setting the stage for SAR transfer between both hit series.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
The medicinal chemistry challenge during hit optimization was the kinked shape of the
binding pocket harboring both apolar (adenosine binding region) and polar (phosphate
binding region) sites. Early synthesized analogs had the tendency to point out of the
binding pocket towards the bulk water, thus showing no gain in affinity towards PHGDH.
Due to more favorable physicochemical properties (low clogP, high solubility in all
relevant assays) as well as the commercial availability of suitable starting materials, our
initial efforts were focused on the phenylpyrazole-5-carboxamides series.
A detailed conformational analysis of the crystal structure of the screening hit 9 (K =
d
105 µM) confirmed that the observed binding confirmation is in alignment with the
minimum energy confirmation induced by 1,3-allylic strain.37, 38 The remaining proton of
the cyclopropyl ring is in planar alignment with the carbonyl group of the corresponding
phenylpyrazole-5-carboxamide and thus locking the conformation of the cyclopropyl
moiety. Therefore, commercially available 2-methyl-5-phenyl-2H-pyrazole-3-carboxylic
ACS Paragon Plus Environment
22

Journal of Medicinal Chemistry
Page 24 of 130
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
acid (10) (K not measurable) was chosen for amide library synthesis because a clear
d
eutomer/distomer and vectors into the phosphate channel were expected based on the
key learnings from the X-ray analysis. Phenethylamides from the library such as 11
showed a potency of 22 µM in the biochemical assay, whereas the distomer 12 and the
methylene analog 13 were not measureable within the detection range of up to 100 µM
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(Table 1).
ACS Paragon Plus Environment
23

Page 25 of 130
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Table 1: Effect of chiral methyl group on binding affinity.
H
R
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
N
N
N
O
Compound
R
Biochemical
PHGDH (low
NAD+)
IC [µM]
50
(
R)-11
S)-12
Me
Me
H
22
(
> 100
> 100
1
3
To gain insight into the structure activity relationship (SAR) of the phosphate channel of
PHGDH a carboxylic acid was installed in the para-position of the phenyl core thus
ACS Paragon Plus Environment
24

Journal of Medicinal Chemistry
Page 26 of 130
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
offering diversification opportunities. Installation of an acidic motif resulted in a dramatic
boost in potency (17). Subsequently, these findings were independently confirmed by
Frédérick et. al. as well as Fuller et. al.29, 39 (Table 2 and Figure 2)
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Having now single digit micromolar compounds in hand, it was decided to re-evaluate
the contribution of the allylic strain to the binding affinity. The significant loss in potency
observed with the geminal dimethyl analogue 14 illustrated the contribution of the 1,3-
allylic strain to the overall binding affinity and is in good alignment with quantum
mechanical geometry optimization for hydrogen atoms and the chiral methyl group at
B3LYP/6-31G* level based on ligand conformation extracted from X-ray of 15 ((R)-
pyrazolo-derivative). The R-isomer is favored by -4.35 kcal/mol over the S-isomer in the
given conformation (Table 2 and Figure 6).
ACS Paragon Plus Environment
25

Page 27 of 130
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Table 2: The influence of allylic strain on binding affinities
H
N
R1
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
R2
N
N
O
OH
O
Compound
R1
R2
Biochemical
PHGDH (low
NAD+)
IC [µM]
50
1
1
1
4
5
6
Me
Me
Me
H
1.3
0.4
5.7
0.9
H
H
rac-17
Me/H H/Me
ACS Paragon Plus Environment
26

Journal of Medicinal Chemistry
Page 28 of 130
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 6: X-ray structure confirmation of compound 15(PDB code: 6RJ3).
29
As previously shown by Fuller et al., the introduction of an acidic functionality led to a
> 50-fold increase in potency. Crystallographic analysis in conjunction with the
biochemical assay data revealed that moving the acidic moiety away from the para
position led to a significant decrease in potency. However, the para-position was also
judged to be the most suitable one for further growing into the NA-site. Thus, it was
ACS Paragon Plus Environment
27

Page 29 of 130
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
envisioned to establish appropriate replacements with additional vectors enabling
building the molecule towards the substrate binding site. A summary of these efforts is
shown in Table 3. Disappointingly, neither neutral, nor basic motifs resulted in the
desired improvement in potency.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment
28