Selective inhibition of prolyl 4-hydroxylases by bipyridinedicarboxylates

Article abstract of DOI:10.1016/j.bmc.2015.05.003

Abstract Collagen is the most abundant protein in animals. A variety of indications are associated with the overproduction of collagen, including fibrotic diseases and cancer metastasis. The stability of collagen relies on the posttranslational modification of proline residues to form (2S,4R)-4-hydroxyproline. This modification is catalyzed by collagen prolyl 4-hydroxylases (CP4Hs), which are Fe(II)- and α-ketoglutarate (AKG)-dependent dioxygenases located in the lumen of the endoplasmic reticulum. Human CP4Hs are validated targets for treatment of both fibrotic diseases and metastatic breast cancer. Herein, we report on 2,2′-bipyridinedicarboxylates as inhibitors of a human CP4H. Although most 2,2′-bipyridinedicarboxylates are capable of inhibition via iron sequestration, the 4,5′- and 5,5′-dicarboxylates were found to be potent competitive inhibitors of CP4H, and the 5,5′-dicarboxylate was selective in its inhibitory activity. Our findings clarify a strategy for developing CP4H inhibitors of clinical utility.

Full text of DOI:10.1016/j.bmc.2015.05.003

Accepted Manuscript
Selective inhibition of prolyl 4-hydroxylases by bipyridinedicarboxylates
James D. Vasta, Ronald T. Raines
PII:
S0968-0896(15)00401-0
DOI:
http://dx.doi.org/10.1016/j.bmc.2015.05.003
BMC 12297
Reference:
Bioorganic & Medicinal Chemistry
To appear in:
Received Date:
Revised Date:
Accepted Date:
24 March 2015
27 April 2015
3 May 2015
Please cite this article as: Vasta, J.D., Raines, R.T., Selective inhibition of prolyl 4-hydroxylases by
bipyridinedicarboxylates, Bioorganic & Medicinal Chemistry (2015), doi: http://dx.doi.org/10.1016/j.bmc.
2015.05.003
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Bioorganic & Medicinal Chemistry
BMC-D-15-00395
REVISED
Selective inhibition of prolyl 4-hydroxylases
by bipyridinedicarboxylates
James D. Vastaa and Ronald T. Rainesa,b,*
aDepartment of Biochemistry and bDepartment of Chemistry,
University of Wisconsin−Madison, Madison, Wisconsin 53706, United States
* Corresponding author. Tel.: 608 262 8588; fax: 608 890 2583
E-mail address: rtraines@wisc.edu (R. T. Raines)
Keywords:
Collagen
α-Ketoglutarate
Hydroxyproline
Hypoxia-inducible factor
Iron

Vasta & Raines
page 2
ABSTRACT
Collagen is the most abundant protein in animals. A variety of indications are associated with the
overproduction of collagen, including fibrotic diseases and cancer metastasis. The stability of
collagen relies on the posttranslational modification of proline residues to form
(2S,4R)-4-hydroxyproline. This modification is catalyzed by collagen prolyl 4-hydroxylases
(CP4Hs), which are Fe(II)- and α-ketoglutarate (AKG)-dependent dioxygenases located in the
lumen of the endoplasmic reticulum. Human CP4Hs are validated targets for treatment of both
fibrotic diseases and metastatic breast cancer. Herein, we report on 2,2′-bipyridinedicarboxylates
as inhibitors of a human CP4H. Although most 2,2′-bipyridinedicarboxylates are capable of
inhibition via iron sequestration, the 4,5′- and 5,5′-dicarboxylates were found to be potent
competitive inhibitors of CP4H, and the 5,5′-dicarboxylate was selective in its inhibitory activity.
Our findings clarify a strategy for developing CP4H inhibitors of clinical utility.

Vasta & Raines
page 3
1. Introduction
Collagen is the principle component of bone, connective tissues, and the extracellular matrix
in animals.1 A variety of diseases are associated with the overproduction of collagen, including
fibrotic diseases (such as pulmonary fibrosis, renal fibrosis, and hepatic cirrhosis)2,3 and
cancers.4-8 The stability of collagen relies on posttranslational modifications that occur
throughout the secretory pathway.9 The most prevalent of these modifications is the
hydroxylation of collagen strands by collagen prolyl 4-hydroxylases (CP4Hs10), which are
Fe(II)- and α-ketoglutarate (AKG)-dependent dioxygenases (FAKGDs) located in the lumen of
the endoplasmic reticulum.11 Catalysis by CP4Hs converts (2S)-proline (Pro) residues in
protocollagen strands into (2S,4R)-4-hydroxyproline (Hyp) residues (Figure 1), which are
essential for the conformational stability of mature collagen triple helices.12 Importantly, CP4Hs
are validated targets for treating both fibrotic diseases13 and metastatic breast cancer,7 which in
aggregate represent a massive unmet clinical need.
Like all enzymes of the FAKGD superfamily, catalysis by CP4Hs requires Fe(II), which is
bound by a conserved His-X-Asp/Glu…Xn…His motif, as well as the cosubstrates AKG and
dioxygen.14 AKG chelates to enzyme-bound Fe(II) using its C-1 carboxylate and C-2 keto groups,
while the C-5 carboxylate group engages in Coulombic interactions with a basic residue
(typically arginine or lysine) and additional hydrogen bonds. All FAKGDs are believed to effect
catalysis through a similar two-stage mechanism in which AKG is first oxidatively
decarboxylated to generate a highly reactive Fe(IV)=O species (ferryl ion), after which the ferryl
ion interacts with a hydrocarbon substrate and effects hydroxylation via a radical rebound
process.14

Vasta & Raines
page 4
In vertebrates, CP4Hs are known to exist as α2β2 tetramers. In these tetramers, the α-subunit
contains the catalytic and substrate-binding domains, and the β-subunit is protein disulfide
isomerase, which is a multifunctional protein that is responsible for maintaining the α-subunit in
a soluble and active conformation.11 Three isoforms of the α-subunit, α(I), α(II), and α(III), have
been identified in humans.11 All α-subunit isoforms form tetramers with the β-subunit, which we
refer to herein as the CP4H1, CP4H2, and CP4H3 holoenzymes. As the most prevalent of the
isoforms, the α(I)-subunit has been characterized extensively. Whereas the structure of the
tetrameric complex is unknown, those of individual domains of the α(I)-subunit have provided
insight into the manner in which CP4Hs interact with the protocollagen substrate, as well as the
means by which the α-subunits dimerize to facilitate formation of the tetramer.15-17
The development of CP4H inhibitors has been of keen interest since the mid 1970s. Like
many FAKGDs, human CP4Hs are inhibited by simple metal chelators, such as 2,2′-bipyridine
(1), as well as AKG mimics (Figure 2), such as N-oxalyl glycine (2), pyridine-2,4-dicarboxylic
acid (3a) and pyridine-2,5-dicarboxylic acid (3b).18 In general, simple metal chelators are
thought to inhibit CP4Hs by iron sequestration, as evidenced by IC50-values that are similar to
the iron concentration used in assays in vitro. In contrast, AKG mimics are thought to bind
competitively in the AKG binding site, using a chelating moiety to interact with the enzymic iron
and an additional carboxyl group to form favorable Coulombic interactions with Lys493.18
Compounds of this nature typically display values of IC50 or Ki that are less than the value of Km
for AKG (which is 20 µM for human CP4H1).19 There are also examples of electrophilic AKG
mimics such as coumalic acid20 and 4-oxo-5,6-epoxyhexanoic acid21 that appear to inhibit CP4H
via covalent modification of the active site. Compounds of this nature could have intriguing
chemical and biological utility.

Vasta & Raines
page 5
Of the simple AKG mimics, the most potent inhibitor of human CP4Hs reported to date is 3b,
which was found to have a Ki = 0.8 µM and display modest selectivity for human CP4Hs
compared to other FAKGDs, such as the prolyl hydroxylase domain-containing proteins (PHDs),
factor-inhibiting HIF (FIH), and Jumonji domain-containing 2E (JMJD2E).22,18 Thus, much work
has focused on modification of the 3b scaffold with the intent of generating a cell-active
inhibitor that is both potent and selective. Some examples include the 5-position
N-acylsulfonamide derivatives23 and 2-heterocyclic glycinamide derivatives24 of 3b (Figure 2),
though these compounds were not developed further due to insufficient activity in cultured cells
or intolerable cytotoxicity.
The most potent inhibitor of human CP4Hs identified to date is 2,2′-bipyridine-5,5′-
dicarboxylic acid (4e, Figure 3).25 Whereas the parent compound 2,2′-bipyridine-5-carboxylic
acid (5) was found to be a modest inhibitor (IC50 = 13 µM), 4e was reported to be substantially
more potent (Ki = 185 nM).25 The inhibitory mechanism of this compound is, however, not
known. 1 itself is well known to inhibit human CP4Hs via iron sequestration rather than enzymic
binding. Although structure–activity relationships (SARs) suggest that enzymic binding
contributes to inhibition by 4e,25 the role of iron sequestration is unclear.
Herein, we investigate 2,2′-bipyridinedicarboxylates as inhibitors of human FAKGDs, both
in terms of the relevant inhibitory mechanism(s) and optimum geometry for inhibition. We
access a library of seven 2,2′-bipyridinedicarboxylates and assess the ability of each to sequester
iron and bind to a CP4H. We find that both 2,2´-bipyridine-4,5´-dicarboxylate (4c) and 4e are
nanomolar inhibitors of a CP4H, though both also bind to free Fe(II). Significantly, we discover
that 4c, but not 4e, binds to a human PHD. These data provide a roadmap for developing potent,
selective inhibitors of human CP4Hs.

Vasta & Raines
page 6
2. Results and discussion
2.1. 2,2′-Bipyridinedicarboxylate library
We accessed a focused library of 2,2′-bipyridinedicarboxylates (4a–g) (Figure 3). The
symmetric members of the library, 2,2′-bipyridine-3,3′-dicarboxylic acid (4a), 2,2′-bipyridine-
4,4′-dicarboxylic acid (4b), 4e, and 2,2´-bipyridine-6,6′-dicarboxylic acid (4g), were obtained
from commercial vendors. The asymmetric members were synthesized by a 4-step route in which
the key 2,2′-bipyridine intermediates were accessed by the palladium-catalyzed 2-pyridylation of
the corresponding pyridine N-oxides (Scheme 1).26-28 This route afforded 4c, 2,2′-bipyridine-4,6′-
dicarboxylic acid (4d), and 2,2′-bipyridine-5,6′-dicarboxylic acid (4f) in overall yields of 53%,
36%, and 15%, respectively.
2.2. Iron sequestration
To begin our analysis, we evaluated 4a–g as iron chelators. We performed titration
experiments to determine the half-maximal concentration required to form a complex with 20
µM Fe(II) (Fe20-EC50) at pH 7.0, taking advantage of the strong absorption of complexes of 1
and related analogues with Fe(II) (Table 1). Along with an estimation of stoichiometry obtained
with Job’s method, the Fe20-EC50 value provides a comparative metric for iron affinity. With the
exception of 4a and 4g, all members of the library were capable of forming distinct red
complexes with Fe20-EC50 values in the micromolar range (see: Supporting Information).
Because 1 is already known to inhibit CP4Hs, these results suggest that most of these ligands
would inhibit CP4H1 by sequestering iron under iron-limiting conditions.

Vasta & Raines
page 7
2.3. CP4H binding
To identify compounds that inhibit human CP4H1 through enzymic binding rather than iron
sequestration, we next sought to determine the extent to which 1 and 4a–g form a complex with
Fe(II) under CP4H assay conditions. Typical in vitro assays for human CP4Hs employ a
tris(hydroxymethyl)aminomethane (Tris) buffer. Although Tris is known to form complexes with
metal ions,29,30 the use of Tris does not affect the observed hydroxylase activity of purified
enzyme.19 Still, to determine if 1 and 4a–g can form iron complexes under our assay conditions,
we incubated our CP4H assay buffer (50 mM Tris–HCl buffer, pH 7.8, containing 1% w/v
bovine serum albumin (BSA), 100 µg/mL catalase, 2 mM sodium ascorbate, 100 µM AKG,
100 µM dithiothreitol (DTT), and 50 µM FeSO4) at 30 °C in the presence or absence of 1
(150 µM), and determined the presence of a Fe(1)32+ complex using spectrophotometry. Under
these conditions, we observed the formation of the Fe(1)32+ complex rapidly and in virtually
identical abundance to that observed in unbuffered conditions. Moreover, similar results were
observed for complexes with 4b, 4c, and 4e ligands. In order of addition experiments, we found
2+
that Tris alone inhibits the formation of the Fe(1)3 , but that the subsequent addition of either
sodium ascorbate or DTT allowed for complex formation.
Given these results and the requirement of ascorbate for hydroxylase activity, the design of
assay conditions that preclude the formation of Fe(II) complexes with 1 and related analogues is
highly unlikely. Thus, we next sought to develop in vitro screening conditions for human CP4Hs
such that the inhibitory effect of iron sequestration would be minimal. We chose an initial
screening concentration of 10 µM, which is substantially below the concentration of FeSO4
(50 µM) used in the assay. Importantly, 1 showed virtually no inhibition under these screening
conditions whereas 3b showed significant inhibition (Figure 4A), which validates these

Vasta & Raines
page 8
conditions for the discovery of compounds where the primary inhibitory mechanism is other than
merely iron sequestration. Although most of the compounds screened under these conditions
showed little to no inhibition, both 4e and 4c were found to be potent inhibitors (more than 90%
reduction of CP4H activity) and 4b was found to be a modest inhibitor that is similar in potency
to 3b (Figure 4A). In subsequent dose–response experiments, the inhibition curves for 4e and 4c
were found to be sigmoidal (Figure 4B) with IC50 values in the low micromolar range (Table 2).
Yet, the inhibition curve for 4b was found to be non-sigmoidal (see: Supporting Information),
which suggests a mixed inhibitory mechanism wherein iron sequestration becomes a contributing
factor at higher concentrations (supported by the observation of a red color in the assay
solutions). These data suggest that CP4H1 is inhibited strongly by 2,2′-bipyridinedicarboxylates
of two different geometries with almost equal potency, and that the inhibition does not rely upon
iron sequestration.
To understand further the potent inhibition observed for 4e and 4c, we next probed the
kinetic mechanism of these compounds. Human CP4Hs have been shown previously to display
an ordered ter–ter mechanism in which AKG first binds to the CP4H·Fe(II) complex, after which
O2 and the peptide substrate bind in an ordered fashion.19 Lineweaver–Burke analyses confirmed
that 4e and 4c both function as competitive inhibitors with respect to AKG (Figure 5), which
suggests that they bind to CP4H1 in a manner that requires enzymic iron. Based upon the
observed competitive inhibition, we used the Cheng–Prussoff equation to estimate values of Ki
for these compounds (Table 2). We found that 4e and 4c inhibit human CP4H1 with Ki values in
the nanomolar range. These Ki values should be taken as upper limits, as it is difficult to take into
account the fraction of each that is present in complexes with free iron under assay conditions.
Although the competitive inhibition observed here should be interpreted with caution in the

Vasta & Raines
page 9
absence of structural information, these data are consistent with a mechanism in which 4e and 4c
inhibit human CP4H1 by binding in the AKG binding pocket. In addition, we tested the biphenyl
analogs of 4e and 4c and found that the pyridyl nitrogens were essential for detectable inhibition
(data not shown), providing further support for a mechanism in which coordination of the
enzymic iron is a requirement.
It is interesting to note the SARs observed above for the 2,2′-bipyridinedicarboxylates
parallel that of the pyridinedicarboxylates. For example, human CP4H1 is inhibited by both 3a
and 3b, where the geometry of 3b is preferred compared to that of 3a.18 Similarly, both 4e and 4c
(reminiscent of 3b and 3a, respectively) are accommodated in the active site, where the geometry
of 4e appears to be preferred compared to that of 4c. Yet, it is also important to note that 4e and
4c are substantially more potent than the corresponding pyridinedicarboxylates. For example,
comparing 3b and 4e and assuming energetically similar interactions with the assay buffer, the
difference in Ki corresponds to a difference in binding free energy of ∆G° = (1.7 0.1) kcal/mol,
which is typical for a hydrogen bond. These comparisons suggest that 4e and 4c likely bind to
human CP4H1 in a manner similar to the corresponding pyridinedicarboxylates, except that their
binding is stabilized further by additional enzymic interactions with the second carboxylate
group.
2.4. Inhibition of PHD2
We reasoned that the second carboxyl group of 4e and 4c could allow for the selective
inhibition of CP4Hs over other FAKGDs. This issue is, however, difficult to address in the
absence of structural information about the catalytic domain of any human or mammalian CP4H.
Although the structure of the N-terminal peptide binding domain of the α-subunit has been

Vasta & Raines
page 10
characterized with both X-ray crystallography16,17 and NMR spectroscopy,15 the structure of the
catalytic domain has remained elusive.
Unlike with CP4H, there are numerous crystal structures available for PHD2, including some
solved as a complex with AKG or AKG mimics. Moreover, preliminary SARs for PHD2 with
respect to simple AKG mimics such as 3a, 3b, and 2 have already been established, facilitating a
comparison between the SARs of CP4H1 and PHD2.18 For example, 2 and 3a inhibit PHD2 at
low micromolar concentrations, but 3b does not inhibit PHD2 even at high micromolar
concentrations. These and other data suggest differences in the active sites of these essential
FAKGDs that could be understood further using a combination of both chemical approaches and
existing structural information.
Toward this end, we screened our library at 10 µM as inhibitors of PHD2. We found that,
whereas 4c (and to a lesser extent 4b) inhibited PHD2, most other 2,2′-bipyridinedicarboxylates
including 4e showed virtually no inhibition (Figure 6). Moreover, the inhibition observed for 4c
was not greater than that observed for simple AKG mimics, such as 2 and 3a. These data are
consistent with the hypothesis that 4c is accommodated in the PHD2 active site, but without the
engagement of its 5′-carboxyl group in a favorable enzymic interaction. These data are also in
agreement with examinations of available crystal structures of PHD2 (PDB entries 3ouj, 3ouh,
2hqu, and 2g1m), wherein the active site cleft opposite the AKG binding site is exposed to
solvent. Using all extant data, we put forth schematic models of 2,2′-bipyridinedicarboxylates
with CP4H1 and PHD2 (Figure 7).

Vasta & Raines
page 11
2.5. Other Isoforms of CP4H
We note that our investigations have focused on the CP4H1 holoenzyme rather than the
CP4H2 and CP4H3 isoforms. This choice was largely due to the relative abundance of CP4H1 in
most tissues compared to that of the other two isoforms.11 Nonetheless, BLAST experiments
suggest high similarity between the three α-subunits, with the α(I)-subunit 78 and 56% similar to
the α(II)- and α(III)-subunits, respectively. Moreover, these similarities improve to >85 and
>63% when considering only the catalytic domains. Given these high similarities, the SARs
identified for CP4H1 (as well as the model put forth in Figure 7A) could also pertain to CP4H2
and CP4H3.
3. Conclusion
The need for selective inhibitors of human CP4Hs with therapeutic potential is evident from
the growing list of diseases associated with the overproduction of collagen. The lack of structural
information available for mammalian CP4Hs has made the development of such inhibitors
difficult, and the understanding of known inhibitors such as compounds 3b and 4e likewise
evasive. Our investigations of the 2,2′-bipyridinedicarboxylate family reveal new information.
For example, 3b was known to be a selective inhibitor of human CP4H compared to other
FAKGDs, such as the PHDs, FIH, and JMJD2E.18 These results, along with additional structure–
activity relationships of 3b analogues,23 led researchers to postulate that 3b could bind to CP4H
in a manner that is distinct from the conventional AKG binding mode, possibly taking advantage
of alternative coordination sites on the enzymic iron center. Our finding that 4c and 4e are
competitive inhibitors of human CP4H1 with almost equal potency contrasts with the previous
postulate and suggests that the AKG-binding pocket of this enzyme is likely able to

Vasta & Raines
page 12
accommodate two different inhibitor geometries (reminiscent of the geometries of either 3a or
3b) in the conventional AKG binding mode. Still, the exact structural and/or conformational
features that allow for this accommodation are not yet understood.
Although achieving selectivity for CP4Hs over other FAKGDs is ideal, it is especially
important to achieve selectivity over PHD2. For example, in hypoxic breast cancers, the
expression of the CP4H α(I)-and α(II)-subunits is upregulated in a PHD2-dependent manner,
which ultimately contributes to the metastatic potential of these tumors.7 More specifically, the
loss in PHD2 activity due to hypoxia allows the accumulation of the PHD2 substrate and
transcription factor HIF-1α, which ultimately enhances the expression of these metastatically
important CP4H α-subunits.7 Although treatment with a CP4H inhibitor could be a viable
strategy to disrupt breast cancer metastasis, that inhibitor must avoid inhibition of PHD2 so as to
prevent a consequent increase in the target CP4Hs. Indeed, the poor performance of many
previously characterized CP4H inhibitors in cells could be due, at least in part, to the off-target
effects associated with inhibiting PHD2. Our finding that both 4c and 4e are substantially more
potent towards human CP4H1 compared to simple AKG mimics, as well as the finding that this
increased potency does not occur when inhibiting PHD2, suggests the existence of additional
CP4H1 residues that can be used to achieve the desired selectivity. Although the identity of these
residues is unclear, previous investigations of the 4e scaffold showed that additional functional
groups such as anilides and sulfonamides are tolerated as replacements of the 5′-carboxylate,23,25
suggesting that the interaction(s) in this second site might not necessarily be Coulombic in nature.
Lastly, we note that our data reveal that both compound 4c and compound 4e are potent
inhibitors of CP4H1 and that inhibition arises largely through enzymic binding. Still, 4c and 4e

Vasta & Raines
page 13
do bind to free Fe(II), which could compromise therapeutic efficacy. Our future work will seek
to modify these scaffolds to limit Fe(II) binding while maintaining selective affinity for CP4Hs.
4. Experimental section
4.1. General experimental procedures
Compounds 1, 3a, 3b, 4a, and 4e were from Sigma–Aldrich (St. Louis, MO). Compound 4b
was from TCI America (Portland, OR). Compound 4g was from CMS Chemicals (Compton,
UK). Biphenyl-4,4′-dicarboxylic acid was from Acros Organics (Geel, Belgium), and biphenyl-
3,4′-dicarboxylic acid was from Combi-Blocks (San Diego, CA). Phosphine ligands and
phosphonium salts were from either Sigma–Aldrich or Strem (Newberryport, MA), stored in a
dessicator, and used without further purification. Pd(OAc)2 was from Sigma–Aldrich, stored in a
dessicator, and used without further purification. HIF-1α peptide556–574 was from AnaSpec
(Fremont, CA) and used without further purification. All other reagent chemicals were from
Sigma–Aldrich, Acros Organics, Combi-Blocks, Oakwood Products (West Columbia, SC),
Enamine (Monmouth Junction, NJ), Bachem (Bubendorf, Switzerland), or Calbiochem–
Novabiochem (San Diego, CA), and were used without further purification.
All glassware was flame- or oven-dried, and reactions were performed under N2(g) unless
indicated otherwise. Dichloromethane and toluene were dried over a column of alumina.
Dimethylformamide was dried over alumina and further purified through an isocyanate
scrubbing column. Other anhydrous solvents were obtained in septum-sealed bottles. Flash
chromatography was performed with columns of 40–63 Å silica gel, 230–400 mesh from
Silicycle (Québec City, Canada). Thin-layer chromatography (TLC) was performed on plates
250-µm silica 60-F254 from EMD Merck (Darmstadt, Germany) with visualization by UV light

Vasta & Raines
page 14
irradiation or staining with KMnO4. The phrase “concentrated under reduced pressure” refers to
the removal of solvents and other volatile materials using a rotary evaporator at water aspirator
pressure (<20 torr) while maintaining water-bath temperature below 40 °C. Residual solvent was
removed from samples at high vacuum (<0.1 torr). The term “high vacuum” refers to vacuum
achieved by a mechanical belt-drive oil pump. All reported yields are unoptimized. The purity of
all synthesized compounds was assessed by HPLC, and the purity of all final compounds was
found to be greater than 95%.
4.2. Instrumentation
Melting points were determined with a PMA160 digital melting apparatus from Stanford
Research Systems (Sunnyvale, CA). NMR spectra were acquired at ambient temperature with a
DMX-400 Avance spectrometer or a Avance 500i spectrometer from Bruker (Billerica, MA) at
the National Magnetic Resonance Facility at Madison (NMRFAM) and were referenced to TMS
or a residual protic solvent. Some compounds exist as either mixtures of rotamers or tautomers
that do not interconvert on the NMR timescale at ambient temperature and therefore exhibit
multiple sets of NMR signals (as indicated). Electrospray ionization (ESI) and electron
ionization (EI) mass spectrometry were performed with a Micromass LCT® and Micromass
AutoSpec® instruments, respectively, from Waters (Milford, MA) at the Mass Spectrometry
Facility in the Department of Chemistry at the University of Wisconsin–Madison. The progress
of reactions catalyzed by prolyl 4-hydroxylases was determined by analytical HPLC using a
system from Waters equipped with a 996 photodiode array detector and Empower 2 software.
Compound purity was also determined using the Waters HPLC system above. Preparative HPLC
was performed using a Prominence HPLC instrument from Shimadzu (Kyoto, Japan) equipped

Vasta & Raines
page 15
with two LC-20AP pumps, SPD-M20A photodiode array detector, and CTO-20A column oven.
Iron complexes with bipyridine ligands were analyzed by spectrophotometry using a Cary 60
UV–Vis spectrometer from Agilent Technologies (Santa Clara, CA). Protein concentrations were
calculated from their absorbance at 280 nm as measured with a NanoVue Plus spectrophotometer
from GE Healthcare using an extinction coefficient of 290,000 M–1cm–1 for human CP4H31 and
36,900 M–1cm–1 for human PHD2.32 IC50- and EC50-values were calculated from experimental
data with Prism version 6.0 software from GraphPad Software (La Jolla, CA).
4.3. Production of recombinant human CP4H1
Human CP4H containing the α(I) isoform was produced heterologously in Origami B(DE3)
Escherichia coli cells and purified as described previously.31
4.4. Assay of human CP4H1 activity in the presence of inhibitors
The catalytic activity of human CP4H1 was assayed as described previously.31 Briefly,
activity assays were carried out at 30 °C in 100 µL of Tris–HCl buffer, pH 7.8, containing
human CP4H1 (100 nM), inhibitor (0–500 µM), substrate (dansylGlyProProGlyOEt, 500 µM),
FeSO4 (50 µM), BSA (1 mg/mL), catalase (0.1 mg/mL), ascorbate (2 mM), DTT (100 µM), and
α-ketoglutarate (100 µM). Reaction mixtures were pre-incubated with or without inhibitor for
2 min at 30 °C, after which the reaction was initiated by the addition of α-ketoglutarate. After
15 min, reactions were quenched by boiling for 45 s and centrifuged at 10,000g. The supernatant
(20–50 µL) was injected into a Nucleodur® C18 Gravity reversed-phase column (4.6 × 250 mm,
5-µm particle size) from Macherey–Nagel (Bethlehem, PA). The column was eluted at 1 mL/min
with a gradient of aqueous acetonitrile (20–45% over 20 min) containing 0.1% v/v TFA. The

Vasta & Raines
page 16
absorbance of the eluent was monitored at 289 nm. All assays were performed in triplicate. Data
is reported as activity relative to control reactions lacking inhibitor, where activity is determined
from the percent conversion of substrate to product. Dose-response curves were generated for
each inhibitor by plotting the relative activity versus the log of the inhibitor concentration. IC50-
values for each inhibitor were interpolated from the dose-response curves by non-linear
regression using the sigmoidal dose-response function within the Prism software.
4.5. Production of human PHD2
A cDNA encoding human PHD2181-426 possessing an N-terminal hexahistidine (His6) tag
(NHis6-PHD2181-426) was cloned using the Gibson strategy,33 and the encoded protein was
produced and purified as described previously.32
4.6. Assay of human PHD2 activity in the presence of inhibitors
The catalytic activity of human PHD2 in the absence or presence of putative inhibitors was
assayed as described previously.32 Briefly, activity assays were carried out at 30 °C in 100 µL of
Tris–HCl buffer, pH 7.8, containing human NHis6-PHD2181–426 (5 µM), inhibitor (0–10 µM),
substrate (HIF-1α peptide556–574, 50 µM), FeSO4 (50 µM), BSA (1 mg/mL), catalase (0.3
mg/mL), ascorbate (2 mM), DTT (1 mM), and α-ketoglutarate (35 µM). Reactions were pre-
incubated with or without inhibitor for 2 min at 30 °C, after which the reaction was initiated by
the addition of α-ketoglutarate. After 10 min, reactions were quenched by boiling for 60 s and
centrifugation at 10,000g. The supernatant (50 µL) was injected into a Nucleodur® C18 Gravity
reversed-phase column (4.6 × 250 mm, 5 µm particle size) from Macherey–Nagel. The column
was eluted at 1 mL/min with a gradient of aqueous acetonitrile (5–56% over 34 min) containing

Vasta & Raines
page 17
0.1% v/v TFA. The absorbance of the eluent was monitored at 218 nm. All assays were
performed in triplicate. Data is reported as activity relative to control reactions lacking inhibitor,
where activity is determined from the percent conversion of substrate to product.
4.7. Assay of Fe(II)-affinity of bipyridine compounds
The affinity of bipyridine compounds for Fe(II) was determined comparatively by measuring
the half maximal concentration (EC50) required for binding 20 µM Fe(II) (Fe20-EC50) in sodium
phosphate buffer, pH 7.0. Stock solutions of ligands were prepared in water. Stock solutions of
FeSO4 were prepared in H2O and used within 3 h of preparation. Ligand solutions (3–500 µM
depending on affinity) were prepared in 10 mM sodium phosphate buffer, pH 7.0, after which
Fe(II) stock solution was added to initiate complex formation. Solutions were allowed to
equilibrate for 15 min, after which the absorbance was recorded at the λmax for the complex
under study. Absorbance values were corrected by subtracting the absorbance value in the
absence of ligand. Dose–response curves were generated for each ligand by plotting the
absorbance versus the log of the ligand concentration. Fe20-EC50 values for each ligand were
interpolated from the dose-response curves by non-linear regression using the sigmoidal dose-
response function in Prism. All experiments were performed in triplicate.
4.8. Determination of Fe(II) complex stoichiometry
The stoichiometry of complexes formed by bipyridine ligands and Fe(II) was estimated with
Job’s method.34,35 Briefly, solutions were prepared such that the concentration of ligand and
Fe(II) were kept at 0.4 mM, but the mole fraction of the ligand was varied from 0 to 1. Stock
solutions of ligands were prepared in water. Stock solutions of FeSO4 were prepared in water and

Vasta & Raines
page 18
used within 3 h of preparation. Mixtures were prepared in 10 mM sodium phosphate buffer, pH
7.0, and complex formation was initiated by the addition of the Fe(II) solution. Solutions were
allowed to equilibrate for 15 min, after which the absorbance was recorded at the appropriate
λmax for the complex under study. Absorbance values were corrected by subtracting the
absorbance value of a blank solution in the absence of FeSO4, after which the values were
normalized relative to the mixture with the highest absorbance value. All experiments were
performed at least in duplicate. Job plots were constructed by plotting the normalized absorbance
versus the mole fraction of bipyridine ligand, after which the stoichiometry of the complex was
estimated from the value on the abscissa of the point of intersection for the two best-fit lines
approaching the point of maximum absorbance.
4.9. Determination of compound purity
All synthesized compounds were analyzed by HPLC to determine their purity. The analyses
were performed on the Waters HPLC system using a Nucleodur® C18 Gravity reversed-phase
column (4.6 × 250 mm, 5-µm particle size) from Macherey–Nagel. Samples (50 µL) dissolved in
H2O were injected into the column and eluted at 1 mL/min with a gradient of aqueous
acetonitrile (5–95% v/v over 20 min) containing 0.1% v/v TFA. The maximal absorbance in the
range of 210–400 nm was used as the detection wavelength.
4.10. Synthesis
4.10.1. 3-Methoxycarbonylpyridine N-oxide
Methyl nicotinate (2.0 g, 14.6 mmoles) and m-chloroperoxybenzoic acid (6.5 g, 29.2
mmoles) were dissolved in dry dichloromethane (150 mL) in a flame-dried flask. This reaction

Vasta & Raines
page 19
mixture was stirred at room temperature for 24 h. The solution was concentrated under reduced
pressure, and the resulting residue was purified by chromatography on silica (5–20% MeOH in
acetone) to afford the title compound as a pale yellow solid (2.19 g, 98%) with mp 94.6–99.1 °C.
1H NMR (400 MHz, CDCl3, δ): 8.77 (s, 1 H), 8.34 (d, J = 6.8 Hz, 1 H), 7.86 (d, J = 8.0 Hz, 1
H), 7.38 (dd, J = 6.8, 8.0 Hz, 1 H), 3.98 (s, 3 H); 13C NMR (100 MHz, CDCl3, δ): 163.2, 142.5,
140.3, 129.9, 126.2, 125.8, 53.1; HRMS (ESI) m/z 154.0494 [calc’d for C7H8NO3 (M + H)+
154.0499].
4.10.2. 4-Methoxycarbonylpyridine N-Oxide
Methyl isonicotinate (1.8 mL, 15.2 mmoles) and m-chloroperoxybenzoic acid (6.8 g, 30.4
mmoles) were dissolved in dry dichloromethane (150 mL) in a flame-dried flask. This reaction
mixture was stirred at room temperature for 24 h. The solution was concentrated under reduced
pressure, and the resulting residue was purified by chromatography on silica (1–4% v/v MeOH
in acetone) to afford the title compound as a white solid (2.26 g, 97%) with mp 118.1–121.3 °C.
1H NMR (400 MHz, CDCl3, δ): 8.20 (d, J = 4.0 Hz, 2 H), 7.84 (d, J = 4.0 Hz, 2 H), 3.90 (s, 3
H); 13C NMR (100 MHz, CDCl3, δ): 163.7, 139.4, 126.5, 126.4, 52.8; HRMS (ESI) m/z
154.0492 [calc’d for C7H8NO3 (M + H)+ 154.0499].
4.10.3. 5-Methoxycarbonyl-2-(4-methoxycarbonylpyridin-2-yl)pyridine N-oxide
Pd(OAc)2 (36 mg, 0.16 mmoles), [P(t-Bu)3H]BF4 (142 mg, 0.49 mmoles), K2CO3 (903 mg,
6.5 mmoles), methyl 2-bromoisonicotinate (705 mg, 3.3 mmoles), and
3-methoxycarbonylpyridine N-oxide (2000 mg, 13 mmoles) were added to a dried flask. The
flask was fitted with a reflux condenser capped with a septum, evacuated, and purged with N2(g)

Vasta & Raines
page 20
(~5 times). Dry toluene (22 mL) was added via syringe, and the reaction mixture was stirred at
110 °C for 18 h. The reaction mixture was cooled to room temperature and filtered through
Celite®, and the filtrate was concentrated under reduced pressure. The crude product was then
purified by chromatography on silica (15% acetone in 1:1 DCM/hexanes) to afford the title
compound (537 mg, 57%) as a pale orange solid with mp 170.6–174.1 °C. 1H NMR (400 MHz,
CDCl3, δ): 9.49 (s, 1 H), 8.89 (s, 1 H), 8.88 (d, J = 5.2 Hz, 1 H), 8.30 (d, J = 8.4 Hz, 1 H), 7.94
(dd, J = 1.2, 5.2 Hz, 1 H), 7.90 (dd, J = 1.2, 8.4 Hz, 1 H), 3.98 (s, 3 H), 3.97 (s, 3 H); 13C NMR
(100 MHz, CDCl3, δ): 165.3, 163.4, 150.2, 149.9, 149.4, 141.9, 138.0, 128.9, 127.7, 125.7,
124.9, 124.1, 53.1, 52.8; HRMS (ESI) m/z 289.0826 [calc’d for C14H13N2O5 (M + H)+ 289.0819].
4.10.4. 4-Methoxycarbonyl-2-(6-methoxycarbonylpyridin-2-yl)pyridine N-oxide
Pd(OAc)2 (15.6 mg, 0.070 mmoles), [P(t-Bu)3H]BF4 (60.4 mg, 0.21 mmoles), K2CO3 (384
mg, 2.8 mmoles), methyl 2-bromopicolinate (300 mg, 1.4 mmoles), and 4-
methoxycarbonylpyridine N-oxide (858 mg, 5.6 mmoles) were added to a dried flask. The flask
was fitted with a reflux condenser capped with a septum, evacuated, and purged with N2(g)
(~5 times). Dry toluene (9.0 mL) was added via syringe, and the reaction mixture was stirred at
110 °C for 24 h. The reaction mixture was cooled to room temperature and filtered through
Celite®, and the filtrate was concentrated under reduced pressure. The crude product was purified
by chromatography on silica (1% v/v MeOH in EtOAc) to afford the title compound (266 mg,
66%) as a pale yellow solid with mp 193.3–197.0 °C. 1H NMR (400 MHz, CDCl3, δ): 8.88 (d, J
= 8.0 Hz, 1 H), 8.74 (d, J = 2.4 Hz, 1 H), 8.31 (d, J = 6.8 Hz, 1 H), 8.18, (d, J = 7.6 Hz, 1 H),
7.99 (t, J = 7.6 Hz, 1 H), 7.87 (dd, J = 2.4, 6.8 Hz, 1 H); 13C NMR (100 MHz, CDCl3, δ): 165.3,
163.4, 150.2, 149.9, 149.4, 141.9, 138.0, 128.9, 127.7, 125.7, 124.9, 124.1, 53.1, 52.8; HRMS

Vasta & Raines
page 21
(ESI) m/z 289.0818 [calc’d for C14H13N2O5 (M + H)+ 289.0819].
4.10.5. 5-Methoxycarbonyl-2-(4-methoxycarbonylpyridin-2-yl)pyridine N-oxide
Pd(OAc)2 (10.1 mg, 0.045 mmoles), [P(t-Bu)3H]BF4 (39.2 mg, 0.14 mmoles), K2CO3
(248 mg, 1.8 mmoles), methyl 2-bromopicolinate (195 mg, 0.9 mmoles), and
3-methoxycarbonylpyridine N-oxide (553 mg, 3.6 mmoles) were added to a dried flask. The
flask was fitted with a reflux condenser capped with a septum, evacuated, and purged with N2(g)
(~5 times). Dry toluene (6.0 mL) was added via syringe, and the reaction mixture was stirred at
110 °C for 24 h. The reaction mixture was cooled to room temperature and filtered through
Celite®, and the filtrate was concentrated under reduced pressure. The crude product was then
purified by chromatography on silica (EtOAc) to afford the title compound (135 mg, 52%) as a
1
pale orange solid with mp 159.7–162.0 °C. H NMR (400 MHz, CDCl3, δ): 9.10 (d, J = 8.0 Hz,
1 H), 8.82 (d, J = 0.8 Hz, 1 H), 8.34 (d, J = 8.4 Hz, 1 H), 8.13 (d, J = 7.6 Hz, 1 H), 7.96 (t, J =
8.0 Hz, 1 H), 7.86 (dd, J = 1.2, 8.4 Hz, 1 H), 3.97 (s, 3 H), 3.93 (s, 3 H); 13C NMR (100 MHz,
CDCl3, δ): 165.1, 163.3, 149.2, 149.0, 147.8, 141.7, 137.5, 128.9, 128.6, 128.0, 125.9, 125.9,
53.0, 52.9; HRMS (ESI) m/z 289.0815 [calc’d for C14H13N2O5 (M + H)+ 289.0819].
4.10.6. Dimethyl 2,2′-bipyridine-4,5′-dicarboxylate
4-Methoxycarbonyl-2-(5-methoxycarbonylpyridin-2-yl)pyridine N-oxide (75 mg,
0.26 mmoles) was dissolved in dry CHCl3 (8.7 mL), and PCl3 (30 µL, 0.31 mmoles) was added
to the resulting solution. The ensuing reaction mixture was stirred at 60 °C until the starting
material was consumed completely, as judged by TLC (6 h). The reaction mixture was quenched
by the dropwise addition of saturated aqueous Na2CO3 (10 mL) while stirring on ice. The

Vasta & Raines
page 22
product was extracted with CHCl3 (4 × 10 mL), and the combined organics were dried over
Na2SO4(s) and concentrated under reduced pressure to afford the title compound (69 mg, 98%)
as a pale yellow solid with mp 163.1–165.2 °C. 1H NMR (400 MHz, CDCl3, δ): 9.30 (d, J = 1.6
Hz, 1 H), 9.00 (s, 1 H), 8.85 (d, J = 5.2 Hz, 1 H), 8.52 (d, J = 8.0 Hz, 1 H), 8.43 (dd, J = 2.0, 8.4
Hz, 1 H), 7.92 (dd, J = 1.6, 5.2 Hz, 1 H), 4.01 (s, 3 H), 3.99 (s, 1 H); 13C NMR (100 MHz,
CDCl3, δ): 165.7, 165.5, 158.6, 156.2, 150.6, 150.1, 138.6, 138.1, 126.0, 123.5, 121.1, 120.6,
52.8, 52.5; HRMS (ESI) m/z 273.0882 [calc’d for C14H13N2O4 (M + H)+ 273.0870].
4.10.7. Dimethyl 2,2′-bipyridine-4,6′-dicarboxylate
4-Methoxycarbonyl-2-(6-methoxycarbonylpyridin-2-yl)pyridine N-oxide (200 mg, 0.69
mmoles) was dissolved in dry CHCl3 (23 mL). PCl3 (76 µL, 0.83 mmoles) was added to the
resulting solution. The ensuing reaction mixture was stirred at 60 °C until the starting material
was consumed completely, as judged by TLC. The reaction mixture was quenched by the
dropwise addition of saturated aqueous Na2CO3 (25 mL) while stirring on ice. The product was
extracted with CHCl3 (4 × 25 mL), and the combined organics were dried over Na2SO4(s) and
concentrated under reduced pressure to afford the title compound (145 mg, 77%) as a pale
yellow solid with mp 182.5–185.0 °C. 1H NMR (400 MHz, CDCl3, δ): 9.03 (s, 1 H), 8.86 (d, J
= 5.2 Hz, 1 H), 8.63 (d, J = 7.6 Hz, 1 H), 8.18 (d, J = 7.6 Hz, 1 H), 8.01 (t, J = 8.0 Hz, 1 H), 7.91
(d, J = 8.0 Hz, 1 H), 4.05 (s, 3 H), 4.01 (s, 3 H); 13C NMR (100 MHz, CDCl3, δ): 165.6, 165.3,
155.9, 149.3, 147.9, 139.4, 138.2, 125.6, 124.7, 123.6, 121.3, 52.9, 52.9; HRMS (ESI) m/z
273.0863 [calc’d for C14H13N2O4 (M + H)+ 273.0870].
4.10.8. Dimethyl 2,2′-bipyridine-5,6′-dicarboxylate

Vasta & Raines
page 23
5-Methoxycarbonyl-2-(6-methoxycarbonylpyridin-2-yl)pyridine N-oxide (127 mg,
0.44 mmoles) was dissolved in dry CHCl3 (14.7 mL), after which PCl3 (50 µL, 0.53 mmoles)
was added. The reaction mixture was stirred at 60 °C until the starting material was consumed
completely, as judged by TLC. The reaction was quenched by the dropwise addition of saturated
aqueous Na2CO3 (15 mL) while stirring on ice. The product was extracted with CHCl3 (5 ×
15 mL), and the combined organics were dried over Na2SO4(s) and concentrated under reduced
pressure to afford the title compound (61 mg, 53%) as a pale yellow solid with mp 170.9–
172.3 °C. 1H NMR (400 MHz, CDCl3, δ): 9.25 (d, J = 2.0 Hz, 1 H), 8.66 (d, J = 7.6 Hz, 1 H),
8.62 (d, J = 7.6 Hz, 1 H), 8.41 (dd, J = 2.0, 8.4 Hz, 1 H), 8.16 (d, J = 8.0 Hz, 1 H), 7.98 (t, J =
8.0 Hz, 1 H), 4.03 (s, 3 H), 3.97 (s, 3 H); 13C NMR (100 MHz, CDCl3, δ): 165.6, 165.6, 158.4,
155.2, 150.4, 147.7, 138.1, 138.1, 126.1, 125.7, 125.0, 121.1, 52.9, 52.4; HRMS (ESI) m/z
273.0858 [calc’d for C14H13N2O4 (M + H)+ 273.0870].
4.10.9. 2,2′-Bipyridine-4,5′-dicarboxylic acid (4c)
Dimethyl 2,2′-bipyridine-4,5′-dicarboxylate (482 mg, 1.8 mmoles) and KOH (460 mg, 7.1
mmoles) were added to a vial. MeOH (9.0 mL) was added to the vial, and the resulting reaction
mixture was heated to 60 °C until the starting material was consumed completely, as judged by
TLC. The reaction mixture was cooled to room temperature and concentrated under reduced
pressure. The crude product was dissolved in water (20 mL). The aqueous layer was washed with
EtOAc (1 × 20 mL), after which the product was precipitated from the aqueous layer by
adjusting to pH 3–4 with 1M HCl. After cooling to 4 °C, the product was filtered, washed with
water (3 × 5 mL), and dried in vacuo to afford 4c (412 mg, 95%) as an off-white solid with mp
>260 °C. 1H NMR (400 MHz, DMSO-d6, δ): 13.67 (bs, 2 H), 9.20 (s, 1 H), 8.92 (d, J = 4.8 Hz,

Vasta & Raines
page 24
1 H), 8.88 (s, 1 H), 8.54 (d, J = 8.0 Hz, 1 H), 8.44 (dd, J = 2.0, 8.4 Hz, 1 H), 7.93 (d, J = 4.8 Hz,
1 H); 13C NMR (100 MHz, DMSO-d6, δ): 166.5, 166.5, 158.0, 155.8 151.2, 150.8, 140.0, 139.0,
127.3, 124.2, 120.9, 120.7; HRMS (ESI) m/z 243.0405 [calc’d for C12H7N2O4 (M – H)–
243.0411].
4.10.10. 2,2′-Bipyridine-4,6′-dicarboxylic acid (4d)
Dimethyl 2,2′-bipyridine-4,6′-dicarboxylate (75 mg, 0.28 mmoles) and KOH (80 mg, 1.12
mmoles) were added to a vial. MeOH (8.0 mL) was added to the vial, and the resulting reaction
mixture was heated to 60 °C until the starting material was consumed completely, as judged by
TLC. The reaction mixture was cooled and concentrated under reduced pressure, after which the
crude product was dissolved in water (2 mL). The aqueous layer was washed with EtOAc (1 × 2
mL), after which the product was precipitated from the aqueous layer by adjusting to pH 3–4
with 1 M HCl. After cooling to 4 °C, the product was filtered, washed with water (3 × 1 mL),
and dried in vacuo to afford 4d (49 mg, 73%) as an off-white solid with mp >260 °C. 1H NMR
(400 MHz, DMSO-d6, δ): 13.66 (bs, 2 H), 8.94 (s, 1 H), 8.90 (d, J = 5.2 Hz, 1 H), 8.62 (dd, J =
1.6, 7.6 Hz, 1 H), 8.17 (t, J = 7.6 Hz, 1 H), 8.13 (dd, J = 1.6, 7.6 Hz, 1 H), 7.92 (dd, J = 1.6, 4.8
Hz, 1 H); 13C NMR (100 MHz, DMSO-d6, δ): 166.6, 166.4, 156.1, 155.0, 151.0, 148.7, 140.0,
139.4, 125.8, 124.1, 123.9, 120.4; HRMS (ESI) m/z 243.0404 [calc’d for C12H7N2O4 (M – H)–
243.0411].
4.10.11. 2,2′-Bipyridine-5,6′-dicarboxylic Acid (4f)
Dimethyl 2,2′-bipyridine-5,6′-dicarboxylate (39 mg, 0.14 mmoles) and KOH (40 mg, 0.57
mmoles) were added to a vial. MeOH (4.0 mL) was added to the vial, and the resulting reaction

Vasta & Raines
page 25
mixture was heated to 60 °C until the starting material was consumed completely, as judged by
TLC. The reaction mixture was cooled and concentrated under reduced pressure, after which the
crude product was dissolved in water (2 mL). The aqueous layer was washed with EtOAc (1 × 2
mL), after which the product was precipitated from the aqueous layer by adjusting to pH 3–4
with 1M HCl. After cooling to 4 °C, the product was filtered, washed with water (3 × 1 mL), and
dried in vacuo to afford 4f (20 mg, 57%) as an off-white solid with mp >260 °C. 1H NMR (400
MHz, DMSO-d6, δ): 13.47 (bs, 2 H), 9.18 (d, J = 1.6 Hz, 1 H), 8.68–8.64 (m, 2 H), 8.47 (dd, J =
2.0 , 8.0 Hz, 1 H), 8.20–8.14 (m, 2 H); 13C NMR (100 MHz, DMSO-d6, δ): 166.5, 166.3, 158.0,
154.6, 150.6, 148.7, 139.5, 138.8, 127.3, 126.0, 124.7, 121.3; HRMS (ESI) m/z 243.0414 [calc’d
for C12H7N2O4 (M – H)– 243.0411].
4.10.12. N-(Methoxyoxalyl)glycine ethyl ester
HGlyOEt·HCl (333 mg, 2.4 mmoles) and DMAP (29.1 mg, 0.24 mmoles) were added to a
round-bottom flask. The flask was capped with a septum and purged with N2(g) (~5 times).
DCM (6 mL) was added, and the reaction mixture was cooled in an ice bath. DIEA (750 µL,
4.3 mmoles) and monomethyl oxalyl chloride (200 µL, 2.2 mmoles) were added via syringe, and
the reaction was stirred and allowed to come to room temperature. After 6 h, the reaction mixture
was quenched on ice by the dropwise addition of a saturated aqueous solution of ammonium
chloride (10 mL). The organic layer was collected, and the aqueous layer was extracted with
EtOAc (3 × 10 mL). The combined organic extracts were dried over Na2SO4(s), and concentrated
under reduced pressure to afford a crude yellow oil. The crude product was then purified by
chromatography on silica (70% v/v EtOAc in hexanes) to afford the title compound (416 mg,
92%) as a white solid with mp 58.5–62.5 °C. 1H NMR (400 MHz, CDCl3, δ): 7.65 (bs, 1H),

Vasta & Raines
page 26
4.23 (q, J = 7.2 Hz, 2 H), 4.11 (d, J = 5.6 Hz, 2 H), 3.90 (s, 3 H), 1.28 (t, J = 7.2 Hz, 3 H); 13C
NMR (100 MHz, CDCl3, δ): 168.6, 160.4, 156.4, 61.9, 53.7, 41.5, 14.1; HRMS (ESI) m/z
212.0535 [calc’d for C7H11NO5Na (M + Na)+ 212.0530].
4.10.13. N-Oxalylglycine (2)
N-(Methoxyoxalyl)glycine ethyl ester (100 mg, 0.53 mmoles) and KOH (88 mg, 1.6 mmoles)
were added to a vial. MeOH (5.0 mL) was added to the vial, and the resulting reaction mixture
was heated to 60 °C until the starting material was consumed completely, as judged by TLC. The
reaction mixture was cooled to room temperature and concentrated under reduced pressure. The
crude product was dissolved in water (2 mL), and the pH of the resulting solution was adjusted to
3–4 with 1 M HCl. The aqueous solution was concentrated under reduced pressure, and the
residue was dissolved in boiling EtOAc. The solution was cooled, filtered, and concentrated
under reduced pressure to a viscous oil. The oil was dissolved in a minimal amount of DCM,
diluted with hexanes, and concentrated under reduced pressure. This procedure was repeated 3
times, after which the product was dried under high vacuum to afford 2 (77 mg, 99%) as a white
solid with mp 104.5–110.5 °C. 1H NMR (500 MHz, DMSO-d6, δ): 14.03 (bs, 1 H), 12.76 (bs, 1
H), 9.04 (t, J = 6.0 Hz, 1 H), 3.79 (d, J = 6.5 Hz, 2 H); 13C NMR (125 MHz, DMSO-d6, δ):
170.8, 162.1, 158.9, 41.4; HRMS (ESI) m/z 165.0506 [calc’d for C4H9N2O2 (M + NH4)+
165.0506].
4.10.14. N-Benzyloxycarbonyl-(2S)-prolyl-(2S)-prolylglycine ethyl ester
(CbzProProGlyOEt)
N-Benzyloxycarbonyl-(2S)-prolyl-(2S)-proline (CbzProProOH, 967 mg, 2.8 mmoles),

Vasta & Raines
page 27
HGlyOEt·HCl (428 mg, 3.1 mmoles), EDC·HCl (590 mg, 3.1 mmoles), and HOBT (415 mg, 3.1
mmoles) were added to a round-bottom flask. The flask was capped with a septum and purged
with N2(g) (~5 times). DMF (28 mL) and DIEA (1.9 mL, 11.2 mmoles) were added via syringe,
and the reaction mixture was stirred at room temperature until the reaction was judged to be
complete by TLC (approximately 48 h). The reaction mixture was concentrated under reduced
pressure, and the resulting residue dissolved in EtOAc (50 mL). The organic layer was washed
with 5% w/v KHCO3 (2 × 25 mL), 5% w/v KHSO4 (2 × 25 mL), and brine (1 × 25 mL). The
combined organics were dried over Na2SO4(s) and concentrated under reduced pressure to a
crude yellow oil. The crude product was then purified by chromatography on silica (0–2% v/v
MeOH in EtOAc) to yield the title compound (755 mg, 63%) as a pale yellow oil. 1H NMR (400
MHz, CDCl3, mixture of 3 or more rotamers, integrations are approximate, δ): 8.29 (t, J =
6.0 Hz, 0.7 H), 7.40–7.20 (m, 15.4 H + CHCl3), 7.15–7.11 (m, 0.4 H), 5.94 (t, J = 6.0 Hz, 0.1 H),
5.31 (d, J = 12.4 Hz, 0.1 H), 5.18–4.93 (m, 5.0 H), 4.77 (d, J = 12.4 Hz, 0.1 H), 4.69 (app dd, J =
2.4, 8.0 Hz, 1.0 H), 4.54 (app dd, J = 4.0, 4.0 Hz, 1.0 H), 4.49–4.20 (m, 3.4 H), 4.17–3.80 (m,
10.4 H), 3.76–3.39 (m, 10.0 H), 3.31–3.23 (m, 0.9 H), 2.99 (app dd, J = 6.0, 17.6 Hz , 0.1 H),
2.53 (app dd, J = 5.6, 12.4 Hz , 0.7 H), 2.35–1.65 (m, 21.4 H), 1.44–1.01 (m, 9.6 H); 13C NMR
(100 MHz, CDCl3, mixture of 3 or more rotamers, δ): 172.9, 172.5, 171.7, 171.6, 171.5, 171.4,
171.3, 169.7, 169.6, 169.3, 168.7, 155.2, 154.9, 154.0, 136.7, 136.5 (multiple signals), 129.0,
128.8, 128.6, 128.4 (multiple signals), 128.1 (2 signals), 127.9 (2 signals), 127.7 (multiple
signals), 125.3, 67.3, 67.2, 67.1, 66.9, 61.2 (multiple signals), 61.1, 61.0, 59.7, 59.4, 58.5, 58.3
(multiple signals), 57.4, 47.7, 47.3 (2 signals), 47.1 (multiple signals), 46.9 (multiple signals),
46.7 (multiple signals), 41.5, 41.2, 40.5, 33.9, 31.2, 31.8, 30.4 (multiple signals), 29.4 (2 signals),
26.9, 26.5 (multiple signals), 25.7, 25.2 (multiple signals), 25.0, 24.7, 24.3, 23.7, 23.4, 22.1, 22.0,

Vasta & Raines
page 28
14.1 (2 signals); HRMS (ESI) m/z 449.2374 [calc’d for C22H29N3O6Na (M + Na)+ 449.2395].
4.10.15. N-Dansylglycyl-(2S)-prolyl-(2S)-prolylglycine ethyl ester
(dansylGlyProProGlyOEt)
A suspension of CbzProProGlyOEt (755 mg, 1.8 mmoles) and Pd/C (76 mg) in MeOH
(18 mL) was stirred under an atmosphere of H2(g) for 3 h. The reaction mixture was filtered
through a pad of Celite® and then concentrated under reduced pressure to a pale yellow viscous
oil (553 mg). The oil was dissolved in DCM (5.9 mL) and cooled to 0 °C, after which
dansylGlyOH (586 mg, 2.0 mmoles) and PyBroP® (973 mg, 2.3 mmoles) were added. The flask
was capped with a septum and purged with N2(g). DIEA (0.63 mL) was added via syringe, and
the reaction mixture was stirred at 0 °C for 5 min. The reaction mixture was allowed to come to
room temperature and stirred for an additional 6 h. The reaction mixture was concentrated under
reduced pressure, and the crude product was purified by chromatography on silica (2% v/v
MeOH in EtOAc) and HPLC (25–55% v/v acetonitrile in H2O with 0.1% v/v TFA) to afford the
title compound (950 mg, 92%) as a yellow solid with mp 80.1–85.2 °C. 1H NMR (500 MHz,
CDCl3, mixture of 2 or more rotamers, integrations are approximate, δ): 8.59 (d, J = 8.5 Hz,
1.0 H), 8.37 (dd, J = 8.5, 14.5 Hz, 1.1 H), 8.25 (ddd, J = 1.0, 7.0, 9.5 Hz, 1.0 H), 8.20–8.14 (m,
0.7 H), 7.65–7.55 (m, 2.1 H), 7.26 (d, J = 7.5 Hz, 1.0 H), 7.18 (t, J = 5.0 Hz, 0.7 H), 5.76 (dd, J =
3.5, 5.0 Hz, 0.7 H), 5.67 (dd, J = 3.5, 5.0 Hz, 0.6 H), 4.58 (app dd, J = 2.5, 8.0 Hz, 1.2 H), 4.45 (t,
J = 6.5 Hz, 0.7 H), 4.31 (d, J = 8.5 Hz, 0.6 H), 4.24–4.13 (m, 2.1 H), 4.09–4.04 (m, 0.8 H), 3.94–
3.80 (m, 2.6 H), 3.74–3.61 (m, 2.3 H), 3.55–3.35 (m, 3.4 H), 2.97–2.95 (m, 4.9 H), 2.56 (app dd,
J = 6.0, 12.5 Hz, 0.7 H), 2.36–2.31 (m, 0.8 H), 2.24–1.80 (m, 11.9 H), 1.33–1.26 (m, 2.9 H); 13C
NMR (100 MHz, CDCl3, mixture of 2 or more rotamers, δ): 171.5, 171.4, 170.3, 169.7, 169.4,

Vasta & Raines
page 29
165.8, 165.4, 151.3, 150.9, 134.3, 134.2, 130.5, 130.4, 129.8 (2 signals), 129.7 (2 signals), 129.5,
129.4, 128.5, 128.4, 123.4, 123.3, 119.8 (multiple signals), 119.5 (multiple signals), 115.6 (2
signals), 61.4, 61.3, 61.1, 59.7, 59.1, 58.2, 47.2, 47.1, 46.6, 46.2, 45.6, 45.5, 44.6, 44.5, 41.5,
41.3, 31.8, 28.5, 28.4, 27.0, 25.2, 24.9, 24.5, 22.1, 14.2 (2 signals); HRMS (ESI) m/z 610.2325
[calc’d for C28H37N5O7SNa (M + Na)+ 610.2306].
Acknowledgments
This work was supported by Grant R01 AR044276 (NIH). J. D. V. was supported by
Molecular Biosciences Training Grant T32 GM007215 (NIH) and a fellowship from the
Department of Biochemistry at the University of Wisconsin–Madison. This study made use of
the National Magnetic Resonance Facility at Madison, which is supported by Grant P41
GM103399 (NIH). The Micromass LCT® mass spectrometer was obtained with support from
Grant CHE-9974839 (NSF).
Supplementary data
Compound characterization and additional binding data.