The advantage of biosensor analysis over enzyme inhibition studies for slow dissociating inhibitors-characterization of hydroxamate-based matrix metalloproteinase-12 inhibitors

Article abstract of DOI:10.1039/c2md20268a

The kinetic characteristics of hydroxamate-based inhibitors of matrix metalloproteinase (MMP)-12 were explored using an SPR biosensor-based assay and enzyme inhibition analysis. These high-affinity inhibitors were shown to dissociate very slowly from the enzyme-inhibitor complex while a carboxylate analogue had a much faster dissociation rate, verifying the importance of the hydroxamate group for the slow dissociation. Progress curve enzyme inhibition analysis confirmed that the hydroxamate compounds but not the carboxylate compound acted as time-dependent inhibitors. The slow dissociation excluded steady-state estimation of IC₅₀-values and K_i values but also made K_i values from progress curve analysis unreliable. Although a full characterization of the inhibitors using biosensor analysis was limited by slow dissociation, it provided kinetic and mechanistic information of relevance for MMP drug discovery and avoided some pitfalls of conventional enzyme inhibition assays. The Royal Society of Chemistry.

Full text of DOI:10.1039/c2md20268a

MedChemComm
View Article Online
View Journal | View Issue
CONCISE ARTICLE
The advantage of biosensor analysis over
enzyme inhibition studies for slow dissociating
inhibitors – characterization of hydroxamate-based
matrix metalloproteinase-12 inhibitors
Cite this: Med. Chem. Commun., 2013,
4, 432
a
a
b
c
¨
Thomas Gossas,† Helena Nordstrom, Ming-Hua Xu, Zhi-Hua Sun,‡
c
d
a
*
Guo-Qiang Lin, Hans Wallberg and U. Helena Danielson
The kinetic characteristics of hydroxamate-based inhibitors of matrix metalloproteinase (MMP)-12 were
explored using an SPR biosensor-based assay and enzyme inhibition analysis. These high-affinity
inhibitors were shown to dissociate very slowly from the enzyme–inhibitor complex while a carboxylate
analogue had a much faster dissociation rate, verifying the importance of the hydroxamate group for
the slow dissociation. Progress curve enzyme inhibition analysis confirmed that the hydroxamate
compounds but not the carboxylate compound acted as time-dependent inhibitors. The slow
dissociation excluded steady-state estimation of IC50-values and Ki values but also made Ki values from
progress curve analysis unreliable. Although a full characterization of the inhibitors using biosensor
analysis was limited by slow dissociation, it provided kinetic and mechanistic information of relevance
for MMP drug discovery and avoided some pitfalls of conventional enzyme inhibition assays.
Received 7th September 2012
Accepted 19th December 2012
DOI: 10.1039/c2md20268a
www.rsc.org/medchemcomm
inuences processes like inammation and cancer metastasis.
The MMP family has for that reason attracted attention from
Introduction
Our understanding of structure–activity relationships is still
very much founded on equilibrium-based data, e.g. affinity (KD)
or IC50, while the structural features that result in fast associ-
ation or slow dissociation are not well understood. It is difficult
to obtain such information with conventional enzyme inhibi-
tion assays and the fact that many lead compounds interact
with slow kinetics is oen not recognized. The advantage of a
time-resolved direct binding assay for lead characterization is
therefore emphasized in this study and discussed in compar-
ison with conventional inhibition assays which oen give
misleading results for inhibitors with slow kinetics.
many pharmaceutical and medical researchers. The specic
interest in MMP-12 as a drug target has emerged as a conse-
quence of its involvement in several major human diseases, for
example chronic obstructive pulmonary disease (COPD) and
rheumatoid arthritis.2,3 The degradation of the extracellular
matrix by this enzyme is an essential part of these diseases and
MMP-12 inhibitors have therefore been expected to be suitable
as drugs.
Potent MMP inhibitors can be designed by incorporating a
zinc-binding group (ZBG), most commonly a hydroxamic acid
moiety, into the inhibitor in a position where it makes a strong
interaction with the catalytic zinc ion. For additional affinity
and selectivity, interactions between the inhibitor and the S10–
S30 pockets are also exploited. This approach has produced
relatively small compounds with high affinity. Although several
MMP inhibitors of this type have reached clinical trials, they
have all failed due to severe side effects. Some of these effects
are most likely due to inhibition of other MMPs and other
metalloproteases, such as those of the ADAM-family.4,5 It has
been speculated that the ZBG contributes too much to the
affinity so that discrimination among the MMP isoenzymes is
very difficult to achieve.6 This has seriously hampered the
evolution of inhibitors to functioning drugs.7 However, in order
to fully understand the molecular basis of the poor specicity of
ZBG-containing inhibitors, information about the kinetics of
the interaction between these inhibitors and MMPs is required.
Matrix metalloproteinase (MMP)-12a is a metallo endopep-
tidase belonging to the matrixin subfamily M10A, according to
the terminology used in the MEROPS database.1 This family of
enzymes is involved in extracellular tissue remodelling and thus
aDepartment of Chemistry – BMC, Uppsala University, Box 576, SE-751 23, Uppsala,
Sweden. E-mail: helena.danielson@kemi.uu.se; Fax: +46-18-558431; Tel: +46-18-471
4545
bShanghai Institute of Materia Medica, Chinese Academy of Sciences, 555 Zuchongzhi
Road, Shanghai 201203, China
cShanghai Institute of Organic Chemistry, Chinese Academy of Sciences, 345 Lingling
Road, Shanghai 200032, China
dMedivir AB, P.O. Box 1086, S-141 22 Huddinge, Sweden
† Current address: Beactica AB, Box 567, SE-751 22 Uppsala, Sweden.
‡ Current address: College of Chemistry and Chemical Engineering, Shanghai
University of Engineering Science, Shanghai 201600, China.
432 | Med. Chem. Commun., 2013, 4, 432–442
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Concise Article
MedChemComm
For this purpose, we used a set of drug-like inhibitors of
MMPs (Fig. 1 and Scheme 1) and a time resolved SPR-based
biosensor assay. The assay was originally developed for the
identication of fragments binding to the active site of MMP-12
and used for screening of a fragment library.8 Here it was used
for the characterization of the interaction kinetics of effective
inhibitors.
The present study conrmed that the hydroxamate group
contributes signicantly to the overall interaction with MMP-12
and that compounds of this class can act essentially as irre-
versible inhibitors of the enzyme, at least on the time scale of
biochemical experiments. Although the slow dissociation can
be an advantage for clinical efficacy, oen discussed in terms of
residence time, there may be disadvantages with respect to
selectivity.
Results
Synthesis and selection of compounds
In order to have access to two compounds differing only in the
zinc binding group, a compound with a carboxylate zinc
binding group (1 (ref. 8)), and a hydroxamate analogue (2), were
synthesized according to the strategy outlined in Scheme 1. Two
structurally unrelated, but also hydroxamate containing
compounds, 3 (ref. 9) and 4 (ref. 10) were selected for compar-
ison (Fig. 1). GM 6001 (5),11,12 a hydroxamate-based MMP
inhibitor tested in clinical trials against several types of
diseases, was used as a well-known reference (Fig. 1).
Scheme 1 General procedure for the preparation of compounds 1 and 2.
Evaluation of inhibitory potency of studied compounds
As a starting point, the inhibitory potency of the compounds
selected for this study was determined by a steady-state based
enzyme inhibition assay using initial rate analysis (Table 1). It
was conrmed that all compounds were inhibitory and had
apparent affinities (Kai pp) ranging from 1 (2) to 50 nM (1). The
parameter has been denoted “apparent” throughout this paper
to point out that it is not a true equilibrium constant since the
inhibition is not measured under strict equilibrium conditions
(see below).
Table 1 Inhibition and interaction kinetic parameters for studied MMP-12
inhibitors
Initial rate analysisa
Progress curve analysisb
kon
koff
Kai pp
(nM)
Compound Kai pp (nM)
(105 Mꢀ1 sꢀ1
)
(10ꢀ3 sꢀ1
)
c
1
2
3
4
5
48
1.0
30
1.9
3.8
—
—
—
0.423
0.414
0.382
—
—
3.53
0.61
0.52
—
1.20
6.82
7.40
Development of a biosensor assay for characterization of
MMP-12 inhibitors
a
The Ki-values are estimates obtained by steady-state analysis using pre-
In order to get further insights into the kinetics of the interac-
tions and how they depended on the structural features of the
incubation of enzyme and inhibitor, as described in Experimental.
b
Obtained from the analysis of the enzyme inhibition data presented
in Fig. 6, using eqn (3) and (4). kon was determined from koff and Kiapp
,
using the relationship kon ¼ koff/Ki. The reaction was started by
c
addition of enzyme. —: not measured.
compounds, a previously developed biosensor assay for MMP-
12 (ref. 8) was adapted for kinetic characterization of inhibitors.
An amine-coupling procedure, optimised for immobilization of
MMP-12, was used to immobilize an amount of enzyme giving a
signal of 2000–2500 RU. Compound 1 was selected for evalua-
tion of the surface characteristics since it was found to bind
reversibly, albeit with high affinity, and could be removed by
mild regeneration conditions.
Fig. 1 Structures of compounds 3, 4 and 5.
This journal is ª The Royal Society of Chemistry 2013
Med. Chem. Commun., 2013, 4, 432–442 | 433

View Article Online
MedChemComm
Concise Article
By injection of compound 1 at saturating concentrations, the as possible surface densities and stabilization of the surface for
apparent binding capacity of the sensor surface was estimated one hour before experiments to avoid the initial rapid dri of
to be 40% of the theoretical maximum binding capacity. new surfaces. The remaining dri was compensated for in the
However, the sensor surface was not stable. This is illustrated in data analysis by subtraction of blank-injection responses from
Fig. 2 where a baseline dri can be observed in sensorgrams the analyte response.
before injection of compound 1, and aer dissociation of the
Immobilized MMP-12 was conrmed to be catalytically
compound. Still, the surface was stabilized with time, as illus- active by incubation of the substrate on a complete chip surface
trated by the difference in dri in the late dissociation phase for with an enzyme manually immobilized (i.e. outside the instru-
1 when injected over the same surface but with a one hour time ment). By docking the chip in the instrument before and aer
difference (Fig. 2a). The dri was higher at high surface densi- immobilization, it was determined that the signal increased by
ties and appeared to be correlated with the activity of MMP-12 2610 ꢁ 300 RU, averaged over all four ow cells. This corre-
as it was also found to be analyte-dependent, i.e. the injection of sponds to 130 ng enzyme immobilized on the whole chip
inhibitor caused a shi of the baseline corresponding to inter- surface. The specic activity of the immobilized protein was
rupted dri during the association phase (Fig. 2b).13 To ensure 0.07 mmol minꢀ1 mgꢀ1, while that of MMP-12 in solution was
that surfaces had sufficiently high binding capacities and 0.7 mmol minꢀ1 mgꢀ1
.
stabilities over time for the performed experiments, a number
of measures were taken, these include: (1) addition of 10 mM
CaCl2 to all reagents injected during the immobilisation
procedure to ensure adequate immobilization levels, (2) the use
of a cross-linking procedure and high concentrations of CaCl2
in the running buffer (200 mM) to minimize the dri, (3) as low
Experimental design
Once the hydroxamate containing compounds (2–5) were
tested, it was obvious that the original regeneration conditions
were inadequate. For the purpose of identifying efficient
regeneration procedures also for these compounds, a variety of
conditions were tested, including salts, acids, bases, organic
solvents, detergents, metal complexing agents and combina-
tions thereof. Unfortunately, the repertoire of regeneration
procedures that could be used was seriously restricted by the
requirement of CaCl2, combined with the high affinity of some
inhibitors. Suitable regeneration conditions could therefore not
be established for all compounds. This had serious practical
consequences since the conventional serial experimental
strategy using a single surface for a series of inhibitor injections
could not be used – compounds with very slow dissociation
rates could simply not be removed from the surface without
damaging the surface itself. Thus, injection of such compounds
blocked the surface and prohibited its further use. Although
single cycle kinetics is sometimes useful when robust regener-
ation conditions are lacking, it was not attempted since the
combination of very slow dissociation and analyte dependent
dri was considered to prohibit the accurate quantication of
rate constants also with that type of experimental design.14
Instead, a new enzyme surface was thus prepared for every
inhibitor used, and the resulting sensorgrams were qualitatively
compared by overlaying the sensorgrams, normalised with
respect to the reference compound 1. It limited the possibility of
obtaining quantitative data and resulted in a rather costly
consumption of sensor chips. But, importantly, it provided
useful data for qualitative comparative analysis of the different
compounds.
Interaction kinetic characteristics of inhibitors
All of the selected hydroxamate containing compounds had very
slow dissociation rates compared to the non-hydroxamate
containing compound (Fig. 3). In order to verify that the
hydroxamate-ZBG was the structural element responsible for
the slow dissociation, a structural analogue to 1 (compound 2)
differing only by having a hydroxamate-ZBG instead of a
carboxylate-ZBG (Fig. 1) was also tested. When injecting
Fig. 2 Analysis of the stability of the MMP-12 sensor surface. (a) Effect of time on
the baseline drift. Sensorgrams of 5 mM 1 injected over the same surface but with
one hour difference. The first injection was made one hour after immobilization.
(b) Effect of inhibitor on the baseline drift. Sensorgram of 5 mM 1. The dashed line
is an extrapolation of the baseline prior to injection. The solid line has the same
slope, but is horizontal during association.
434 | Med. Chem. Commun., 2013, 4, 432–442
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Concise Article
MedChemComm
Fig. 3 Qualitative comparison of sensorgrams for inhibitors interacting with
MMP-12. Sensorgrams of 3, 4 and 5 were obtained with different enzyme
surfaces, while one sensorgram of 1 was obtained for each surface and used to
normalize the binding capacities of the different surfaces. All inhibitors were
injected at a concentration of 5 mM. The sensorgrams are solvent corrected,
adjusted for differences in molecular weight of inhibitors and blank subtracted.
compounds 1 and 2 at the same concentration, it was clearly
shown that the hydroxamate compound had a much slower
dissociation rate than the carboxylate compound (Fig. 4). As a
result, the hydroxamate compound also had a higher steady-
state level of binding, indicating that larger amounts of the
complex were formed.
The sensorgrams for the studied compounds were distinctly
biphasic, in both the association phase and the dissociation
phase (Fig. 3), indicating an interaction mechanism involving
more than one equilibrium. In an attempt to establish the
nature of the complex interaction mechanism for the hydrox-
amate containing inhibitors, a more extensive experimental
design using a series of inhibitor concentrations was used for
compounds 1 and 5 (Fig. 5). For 5 the same strategy as used
above, i.e. with new enzyme surfaces for each concentration,
was exploited (Fig. 5a), while for compound 1, the same sensor
surface could be used for all concentrations since the inhibitor
dissociated completely (Fig. 5b). These sensorgram series
showed the biphasic characteristic for every concentration,
perhaps most visible in the dissociation phase where 5
Fig. 5 Sensorgrams illustrating the concentration dependence of inhibitors
interacting with MMP-12. (a) 5 at concentrations of 80 nM to 10 mM. (b) 1 at
concentrations of 0.062–45 mM. The sensorgrams are solvent corrected, blank
subtracted and adjusted for differences in binding capacity of the surfaces.
dissociates rapidly in the rst ten seconds and thereaer the
dissociation almost ceases. The highest concentrations of 5 also
completely stopped the baseline-dri. Although both
compounds 1 and 5 interacted with the enzyme in a concen-
tration dependent manner, the two sets of sensorgrams differed
qualitatively. None of the interactions could be satisfactorily
described by a binding equation corresponding to a simple
Langmuir or a 2-step interaction (e.g. induced t), suggesting
that the interaction is more complex. Other datasets (data not
shown) suggested that the interaction was heterogeneous and
composed of a high affinity and a low affinity component,
perhaps resulting from auto-hydrolysis on the surface. But it
was not possible to reliably determine the rate constants or
affinities for these interactions. We therefore chose to keep the
analysis qualitative, as it was sufficient to conclude for
the purpose of the present study that the hydroxamate (but not
the carboxylate) containing inhibitors dissociated very slowly
from MMP-12.
Analysis of time-dependent inhibition
As a control, an alternative assay, involving measurement of the
effect of the inhibitors on the catalytic activity of MMP-12 was
used to verify the kinetic features of the inhibitors also in the
solution. In such an assay the slow dissociation of the hydrox-
amate containing inhibitors from the enzyme (seen in Fig. 3–5)
should be manifested as time-dependent inhibition, unless the
inhibitor is pre-equilibrated with the enzyme for an extended
time.15 Therefore, the inhibition of MMP-12 by the three
Fig. 4 Comparison of sensorgrams of hydroxamate- and a carboxylate-based
inhibitor interacting with MMP-12 (compounds 1 and 2, respectively). The
inhibitors were injected at a concentration of 5 mM.
This journal is ª The Royal Society of Chemistry 2013
Med. Chem. Commun., 2013, 4, 432–442 | 435

View Article Online
MedChemComm
Concise Article
slow-dissociating inhibitors (3, 4 and 5) and the fast-dissoci-
ating carboxylate compound (1) was further investigated using
an enzyme inhibition assay where enzyme was added to start
the reaction (i.e. without pre-incubating the enzyme with the
inhibitor).
Correlation between interaction kinetics, inhibition and
structure
The three most inhibitory compounds (2, 4 and 5) had similar
apparent Ki-values (1.0, 1.9 and 3.8 nM, respectively) but
differed clearly in the maximal level of the formed complex
(Fig. 3). The two least effective inhibitors (1 and 3) had apparent
Ki-values in the same range (48 and 30 nM, respectively) but
quite different interaction proles, with 1 dissociating relatively
rapidly and 3 forming comparatively stable complexes, albeit
not as efficiently as, for example, compound 4. These compar-
isons show that there is no simple correlation between the
interaction kinetics and the inhibitory parameter that could be
determined for the compounds, indicating that the interaction
characteristics are not easily understood simply from inhibition
data, and vice versa. In contrast, the interaction kinetic proles
revealed a unique behaviour for the carboxylate compound, the
only one showing clear dissociation. This illustrates the
usefulness of time-resolved data and some of the problems in
using enzyme inhibition analysis for characterization of enzyme
inhibitor interactions.
These experiments showed that the three hydroxamate
compounds displayed time-dependent inhibition, whereas the
carboxylate compound did not (Fig. 6). An inhibitor concen-
tration equal to the enzyme concentration was enough for
complete inhibition by compounds 4 and 5, indicating that the
inhibition was due to an interaction with a 1 : 1 stoichiometry. A
model for time-dependent inhibition was tted to the data by
non-linear regression in order to identify the mechanism for the
observed time-dependent inhibition. The model for a simple
1 : 1 interaction (E + I # EI) resulted in satisfactory ts to the
progress curve data of the time dependent inhibitors and
allowed estimation of the kinetic parameters for compounds 3,
4 and 5 (Table 1). (This procedure could not be applied for
compound 1 since it was not time-dependent and other infor-
mation required for estimation of Ki was lacking.) These Ki-
values were lower than those initially determined by assuming
competitive inhibition and using the integrated form of the
Michaelis–Menten equation (Table 1). This reects the differ-
ences in values obtained by a procedure based on analysis of
complete progress curves and when pre-equilibrating the
inhibitor with the enzyme so that the measurements are
performed at a presumed steady-state.
Discussion
The original aim of the current study was to explore the kinetic
features of hydroxamate containing MMP-12 inhibitors in order
to better understand the relationship between the structure and
Fig. 6 Time dependence of MMP-12 inhibition. Progress curves with four different inhibitors: 1 at 0–100 nM, 3 at 0–100 nM, 4 at 0–10 nM and 5 at 0–10 nM. Each
curve is the result of two measurements, i.e. the apparent noise represents the difference between the two replicate experiments. The black lines represent a fit of a
model for time dependent inhibition, except for 1 which is a linear fit.
436 | Med. Chem. Commun., 2013, 4, 432–442
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Concise Article
MedChemComm
efficacy of inhibitors. A time-resolved biosensor assay was combination of a complex interaction, slow dissociation, and a
therefore used to study the interaction between MMP-12 and a limited experimental dataset did not allow for the establish-
small set of structurally diverse inhibitors, with and without a ment of the interaction mechanism or a reliable quantication
hydroxamate group. The assay was previously developed for of the interaction kinetic rate constants, but the qualitative
screening of a fragment library8 and required a slightly modi- analysis provided the required information for this study.
ed experimental design for the current experiments. The
The inhibition experiments conrmed the slow dissociation
baseline dri was initially a problem but it could be minimized of the hydroxamates, which resulted in progress curves dis-
by optimization of the experimental procedure and compen- playing characteristics typical of time dependent inhibition.
sated for in the data analysis by subtraction of blank injections Reliable quantication of the inhibition was not possible by this
from analyte injection sensorgrams. Instead, it was rather an method either, even if rate constants and affinities of slow-
interesting phenomenon that provided information about the binding inhibitors can in principle be determined by non-linear
characteristics of the immobilized enzyme. The dri was nearly regression analysis of progress curves obtained at a series of
abolished by injection of inhibitors of high affinity, suggesting inhibitor concentrations.19 It becomes more challenging when
that it was caused by autohydrolysis and that the immobilized the inhibitors are also tight-binding (see Morrison 1988 (ref. 20)
enzyme was highly active. The same phenomenon has previ- and Copeland 2005 (ref. 15)), which was the case here. (The
ously been observed when inhibitors bind to HIV-1 protease.13 established terminology is somewhat misleading since “tight-
Furthermore, it was reduced by high calcium concentrations, binding” is dened as Ki # [E] and depends on the assay setup
which is in agreement with a recent study that suggests that rather than the value of Ki, while “slow-binding inhibitors” are
calcium stabilizes the active form of the enzyme by preventing not slow-binding per se (slow kon), but the formation of complex
the partial unfolding necessary for autohydrolysis.16 The cross- (measured as time-dependence of inhibition) will appear to be
linking procedure used for stabilization of the surface also most slow when [I] is low since kon ꢂ [I] will be very small even if kon is
likely prevented the unfolding of the enzyme that is required for as fast as diffusion permits.) But the regression analysis of the
autohydrolysis. The 10-fold lower activity of immobilized inhibition data did not give satisfactory ts even when
enzyme was somewhat puzzling, considering that the apparent accounting for both slow and tight binding. This might be due
functionality determined by the use of an inhibitor was to a lower than expected, or decreasing, active enzyme
approximately 40%. But the characteristics may be explained by concentration during measurements. It is also possible that the
considering the design of this experiment, where the substrate complex interaction seen in the biosensor assay is not well
needs to be transported to the chip-surface in order to be modelled by these equations. It was however possible to t a
hydrolysed. The limited mass transport of substrate to the simpler equation to the data, not accounting for tight-binding,
immobilized enzyme in experiments performed with sensor which provided estimates of the parameters. Despite these
chips simply sitting on the bench with “gentle agitation”, i.e. limitations, this analysis provided more relevant inhibition
without guaranteed adequate mixing, may reduce the effective constants than the steady-state analysis initially used (only
concentration in the sensor matrix, thus reducing the observed resulting in Kai pp) and conrmed that the slow-dissociating
specic activity on the chip. Nevertheless, the experiment hydroxamate inhibitors were time dependent inhibitors,
showed that the immobilized enzyme was catalytically active whereas the carboxylate was not.
although some catalytic power may have been lost in the
immobilization procedure or aer auto-hydrolysis aer parameters of slowly dissociating inhibitors is challenging also
immobilization. with a biosensor-based approach, it allows a semi-quantitative
Although determination of the mechanism and kinetic
The interaction between the inhibitors and MMP-12 was analysis that is very informative. In our earlier studies we have
found to be complex and not well described by a simple 1 : 1 used different methods for elucidating the interaction mecha-
interaction mechanism. Inhibitors containing a hydroxamate nism and for the ranking of compounds on the basis of alter-
moiety were discovered to dissociate very slowly (if at all) from native parameters describing relevant kinetic features. For
the enzyme–inhibitor complex. The inability to establish a example, in the case of non-nucleoside inhibitors interacting
regeneration procedure for these inhibitors required an with HIV-1 reverse transcriptase (NNRTIs),21 a comparative
unusual experimental design with one inhibitor injection per analysis of different compounds and simulation of sensorgrams
immobilization. Time-dependent inhibition is oen observed provided approximations of the range of the kinetic constants
with zinc-chelating MMP inhibitors, and it is in fact a general and enabled an interpretation of the kinetic features of rele-
trait of reasonably potent inhibitors of MMPs,15 although this is vance for anti-replicative efficacy in the cell culture. This
not immediately apparent when studying the literature since it procedure could not be used here due to the uncharacterized
is generally not recognized. Based on inhibition experiments, complexity of the interaction (not shown).
the mode of binding has been suggested to follow a simple one
One of the advantages of time-resolved analysis of molecular
step interaction mechanism although there is evidence for a interactions is that it provides more direct information about
conformational change leading to a tighter enzyme–inhibitor the interaction between the inhibitor and its target than
complex.17,18 The biosensor-data shows that the tested hydrox- conventional enzyme inhibition assay methods used for lead
amate-based inhibitors had very slow dissociation rates when discovery. For example, the irreversible/slow dissociating nature
interacting with MMP-12, but the involvement of a rate limiting of lead compounds is readily observed by this biosensor-based
conformational change could not be conrmed. The approach, but can be missed or misinterpreted by conventional
This journal is ª The Royal Society of Chemistry 2013
Med. Chem. Commun., 2013, 4, 432–442 | 437

View Article Online
MedChemComm
Concise Article
procedures that need to be designed specically for detection of with slow dissociation rates. Such interactions can also ensure
this feature. When using a biosensor-based approach, there is that binding only occurs to the specic MMP of interest since
no risk that the slow-binding detected in an inhibition assay is hydroxamate-containing compounds can in principle interact
mistaken for slow association since it will be clearly seen as slow with all MMP isoenzymes and potentially with many other
dissociation. Moreover, the direct readout makes it indepen- enzymes that have catalytically essential zinc ions. Alternative
dent on substrate-related effects, which can be problematic strategies of achieving selectivity for compounds with strong
when comparing data for different enzymes. In the case of ZBGs have been suggested; where one postulates that the slow
hydroxamate-based MMP inhibitors, the lack of catalysis-inde- binding behaviour of these inhibitors can be utilized to gain
pendent inhibitor kinetic information may have led to selectivity.25 Since our data shows that the slow-binding
misleading interpretations of compound selectivity, one of the behaviour is an artifact caused by slow dissociation rather than
main problems associated with hydroxamate compounds. It being an effect of slow association, this strategy is not realistic.
will therefore briey be elaborated here:
Another line of development that is pursued, is to exclude the
The most common methods for evaluating structure activity ZBG altogether and focus on inhibitors binding in the
relationships and selectivity for MMP inhibitors are based on S10-specicity loop.26 This is difficult since the active site of
IC50- or Ki-values obtained by initial rate measurements of MMPs is exible and the S10-specicity loop is mobile.27 Hence,
enzyme inhibition using FRET-peptide substrates.22 There are whatever strategy is used, structural modelling (for example
several concerns with this approach. Firstly, it assumes that the using available crystal structures28,29) and design need the
enzyme and the inhibitor are in rapid equilibrium during the support of interaction kinetic information in order to analyse
measurements, or that they have reached equilibrium in a pre- specicity issues. However, this requires full insight into the
incubation step. It is not trivial to establish for inhibitors with interaction mechanism. In a recent study involving non-nucle-
very slow association and/or slow dissociation rates of inhibi- oside inhibitors of HIV-1 reverse transcriptase, we have shown
tion, or that follow a complex inhibition mechanism. Secondly, that slow dissociation as such is not the critical parameter for
analysis of selectivity with respect to target analogues (e.g. MMP efficacy, but that other kinetic features are also important.21
isoenzymes) requires that the assays and/or the data are Thus, it is not possible to use a simple and general rule-of-
normalized with respect to differences in the catalytic efficacy of thumb for the desirable kinetic features of inhibitors (such as
the enzymes, accounting for example for differences in the slow koff or long residence time30). The ideal features but must
KM-values for the enzymes and substrates used. Failing to be established on a case-by-case basis. The biosensor assay
recognize these issues can, for example, result in the common, presented herein can clearly provide the detailed kinetic infor-
but misleading, comparisons of IC50-values for different mation that can help guide the design of a new generation of
enzymes. A typical example is a study that intends to show the specic and efficient MMP inhibitors. We have already exploi-
selectivity prole of compound 5.23 However, it has not recog- ted the biosensor assay described herein for screening of a
nized the slow dissociation of this class of inhibitors, calling for fragment library and identied inhibitors that do not contain
an experimental design and analysis procedure that takes this zinc-chelating groups.8 Further development of these hits into
into consideration. In addition, it uses the same substrate and leads is currently being explored.
conditions for all MMPs studied, without accounting for the
different catalytic properties of the isoenzymes. The use of IC50
values for the “selectivity” analysis of compound 5 in the study
-
Conclusions
is therefore problematic. These considerations also explain why The present study provides new insights into the structural and
the apparent Ki-values obtained in the present study were not kinetic features of effective MMP-12 inhibitors and that
useful for description of the affinity of the studied compounds biosensor-based interaction analysis is very informative for lead
and why there was no simple correlation between the interac- characterization, especially for slowly dissociating inhibitors
tion kinetic and the inhibition data.
where enzyme inhibition-based assays have several drawbacks.
By interpreting the data into the context of selectivity we The data show that non-covalent, but slowly dissociating
illustrate that it is easy to misinterpret inhibition data without inhibitors may have the same problems as covalent modiers.
substrate-independent interaction kinetic information, i.e. This puts the recent focus on long residence time of drugs30 in a
association and dissociation rate constants. But more impor- new perspective and demonstrates the importance of charac-
tantly, without kinetic information it is not possible to fully terizing the mechanism and kinetics of inhibitors.
understand structure–activity relationships and to modify
inhibitor structures so that both affinity and residence time are
optimized.
Experimental
Inhibitors
The data supports the hydroxamate-based design of MMP
inhibitors, in the sense that this clearly results in very effective Compounds 3 (ref. 9) and 4 (ref. 10) were obtained from Cal-
inhibitors. The interactions between hydroxamates and zinc in biochem (EMD Chemicals, Gibbstown, NJ, USA) and compound
solution have been explored, providing a good description of 5 (ref. 11 and 12) (GM 6001, ilomastat) from Chemicon (Milli-
the interaction between this moiety and the active site of the pore, Solna, Sweden). The purity of all tested compounds was
enzyme.24 However, additional interactions, for example with $95%, as determined by analytical high-performance liquid
the S10-pocket, are also needed in order to result in inhibitors chromatography (HPLC).
438 | Med. Chem. Commun., 2013, 4, 432–442
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Concise Article
MedChemComm
under reduced pressure. The residue was diluted with EtOAc
and washed with brine. The organic layer was dried over
Na2SO4, and concentrated under reduced pressure. The crude
product was puried by silica gel column chromatography to
give 7 in 74% yield (764 mg).
Chemical procedures
General. NMR-spectra were recorded on a Varian 500 MHz
instrument using (CD3)2SO as a solvent. Tetramethylsilane
(TMS) was used as reference.
LC-MS purity measurements were performed with a Waters
Alliance 2695 HPLC instrument and a 50 ꢂ 3.0 mm ACE C8
column with 3 mm particles and two different methods. Chro-
matography system A. Mobile phases A: 10 mM NH4OAc, B: 10
mM NH4OAc in 90% CAN; gradient program: 20–100% B in 5
min followed by a wash for 2 min at 100% B; ow rate: 0.8 mL
minꢀ1; injection volume: 5 mL; sample concentration: 1 mg
mLꢀ1 (diluted in acetonitrile, ACN); detection: UV @ 210–400
nm and ESI-MS. Chromatography system B. Same as system A
except: mobile phases A: 0.2% formic acid; B: 0.18% formic acid
in ACN; sample concentration: 1 mg mLꢀ1 (diluted in MeOH).
The mass spectrometry analyses were performed on a Waters
ZQ instrument equipped with an electrospray interface. The
operating conditions for the electrospray interface and the mass
spectrometer were spray voltage 3300 V, cone voltage 30 V,
source temperature 120 ꢃC, desolvation gas temperature 300 ꢃC,
and desolvation gas ow 800 L hꢀ1. Full scan mass spectra of
positive ions were recorded for the mass range 300–900 Da
(300–700 Da for 2). The scan time was 1.0 s per spectrum. All
data were processed using the Masslynx soware version 4.1
from Waters.
LC-MS accurate mass measurements were performed using a
HDMS Synapt instrument from Waters (UK) equipped with a
lockspray interface, connected to a Waters Aquity system. The
acquisition range was m/z 100 to 1000 with an acquisition time
of 0.15 s (+ESI). Leucine enkephalin was used as lock mass. The
reversed phase column was an YMC-UltraHT Pro C18, 2.1 ꢂ 50
mm, 2 mm, 120 A from YMC (U.S.A) and the mobile phases were
based on water/acetonitrile containing 0.2% formic acid.
(S)-4-Benzyl-3-(4-(4-bromophenyl)butanoyl)oxazolidin-2-one
(6). Under nitrogen, to a solution of Evan's auxiliary (S)-4-ben-
zyloxazolidin-2-one (708 mg, 4.0 mmol) in dry tetrahydrofuran
(R)-2-(4-Bromophenethyl)-4-tert-butoxy-4-oxobutanoic acid
(8). To a solution of 7 (671 mg, 1.3 mmol) in THF–H2O (4/1, 10
ꢃ
mL) at 0 C was added hydrogen peroxide (30%, 0.7 mL). Aer
stirring for 10 min, lithium hydroxide (62 mg, 2.6 mmol) was
added. The reaction was stirred overnight and then quenched
with Na2SO3. THF was evaporated under reduced pressure. The
pH of aqueous solution was adjusted to 2 with HCl (1 N),
extracted with EtOAc and washed with brine. The organic layer
was dried over Na2SO4, and concentrated under reduced pres-
sure. The residue was puried by silica gel column chroma-
1
tography to afford the compound 8 in 85% yield (394 mg). H-
NMR (300 MHz, CDCl3): d 7.42 (d, J ¼ 8.4 Hz, 2H), 7.07 (d, J ¼ 8.4
Hz, 2H), 2.90–2.78 (m, 1H), 2.73–2.58 (m, 3H), 2.47–2.38 (m,
1H), 2.05–1.95 (m, 1H), 1.94–1.72 (m, 1H), 1.44 (s, 9H).
(R)-tert-Butyl 5-(4-bromophenyl)-3-((S)-1-(methylamino)-1-
oxo-3-phenylpropan-2-ylcarbamoyl)pentanoate (9). To a solu-
tion of 8 (357 mg, 1 mmol) in dimethylformamide (DMF) (5 mL)
was added hydroxybenzotriazole (HOBt) (270 mg, 2 mmol) at
ꢃ
ꢃ
0 C. The mixture was cooled to ꢀ15 C, and EDC (384 mg, 2
mmol) was added. Aer 30 min, the reaction was warmed to
room temperature, L-phenylalanine methylamide hydrochloride
(428 mg, 2 mmol) and N-methylmorpholine (NMM) (0.16 mL,
1.44 mmol) were added. The mixture was stirred overnight. The
solvent was removed and the residue was partitioned between
ethyl acetate (EtOAc) and saturated NH4Cl solution. The
aqueous layer was extracted with EtOAc. The combined organic
layers were dried and concentrated under reduced pressure.
The residue was puried by silica gel column chromatography
to afford 9 in 72% yield (372 mg). 1H-NMR (300 MHz, CDCl3): d
7.38 (d, J ¼ 8.7 Hz, 2H), 7.34–7.20 (m, 5H), 6.98 (d, J ¼ 8.1 Hz,
2H), 6.38 (d, J ¼ 8.1 Hz, 1H), 5.80 (br, s, 1H), 4.52 (dd, 1H), 3.12–
3.05 (m, 2H), 2.71 (d, J ¼ 2.4 Hz, 3H), 2.57–2.34 (m, 5H), 1.92–
1.82 (m, 1H), 1.43 (s, 9H).
ꢃ
(THF) (15 mL) at ꢀ78 C was added n-BuLi (2.5 mL, 1.6 M, 4.0
mmol). The mixture was stirred at ꢀ78 ꢃC for 30 min, where
aer 4-(4-bromophenyl)-butyryl chloride (877 mg, 4.05 mmol) in
THF (5 mL) was added slowly. Aer stirring for another 1 h, the
reaction was allowed to warm to room temperature, and stirred
for additional 6 h, and nally quenched with saturated aqueous
NH4Cl. The aqueous phase was extracted with CH2Cl2; the
combined organic layers was washed with aqueous NaHCO3
and brine, dried over Na2SO4, and concentrated under reduced
pressure. The residue was puried by silica gel column chro-
matography to provide the compound 6 in 83% yield (1335 mg).
(R)-tert-Butyl 3-((S)-4-benzyl-2-oxooxazolidine-3-carbonyl)-5-
(4-bromophenyl) pentanoate (7). Under nitrogen, to a solution
of 6 (804 mg, 2.0 mmol) in dry THF (10 mL) at ꢀ78 ꢃC was added
freshly prepared lithium diisopropylamide (LDA) in THF
(R)-tert-Butyl 3-((S)-1-(methylamino)-1-oxo-3-phenylpropan-
2-ylcarbamoyl)-5-(4-(pyridin-3-yl)phenyl)pentanoate (10). Under
nitrogen, the mixture of 9 (350 mg, 0.68 mmol), 3-pyridinebor-
onic acid (120 mg, 0.82 mmol), Pd(PPh3)4 (39 mg, 0.034 mmol),
2 M Na2CO3 (1 mL), toluene (5 mL), and EtOH (3 mL) was heated
to reux for 2 h. The EtOH was removed and the residue was
diluted with CH2Cl2, washed with brine, dried and concentrated
under reduced pressure. The crude product was puried by
silica gel column chromatography to give 10 in 74% yield
(260 mg).
(R)-3-((S)-1-(Methylamino)-1-oxo-3-phenylpropan-2-ylcarba-
moyl)-5-(4-(pyridin-3-yl)phenyl)pentanoic acid (1). To a solution
ꢃ
of 10 (258 mg, 0.5 mmol) in CH2Cl2 (5 mL) at 0 C were added
triuoroacetic acid (TFA) (2 mL) and water (3 drops). Aer
stirring for 3 h at room temperature, the mixture was diluted
with EtOAc and 10% NaOH was added to adjust the pH to 6. The
aqueous layer was extracted with EtOAc, the combined organic
phases were dried over anhydrous Na2SO4 and concentrated
under reduced pressure. The crude product was puried by
ꢃ
(2.1 mmol). The mixture was stirred at ꢀ78 C for 1 h, where
aer tert-butyl bromoacetate (429 mg, 2.2 mmol) was addeꢃ d.
The reaction was continued to stir for another 2 h at ꢀ78 C,
and then allowed to warm to room temperature overnight.
Saturated aqueous NH4Cl was added and THF was removed
This journal is ª The Royal Society of Chemistry 2013
Med. Chem. Commun., 2013, 4, 432–442 | 439

View Article Online
MedChemComm
Concise Article
8
¨
silica gel column chromatography to afford 1 in 92% yield Nordstrom et al. MMP-12 was immobilized on a CM5 chip
(211 mg).
using amine coupling. The ow rate was 10 mL minꢀ1 during the
1H-NMR (500 MHz, DMSO-d6) d 9.04 (s, 1H), 8.70 (d, 1H, J ¼ immobilization procedure. The chip was activated by an injec-
5.2 Hz), 8.43 (d, 1H, J ¼ 8.0 Hz), 8.11 (d, 1H, J ¼ 7.6 Hz), 7.82 (dd, tion of EDC/NHS for ten minutes. MMP-12, at a concentration
1H, J ¼ 4.2; 8.9 Hz), 7.77 (dd, 1H, J ¼ 5.2; 8.0 Hz), 7.70 (d, 2H, J ¼ of 60 mg mLꢀ1 in 10 mM maleate, pH 6.0, 10 mM CaCl2, was
8.4 Hz), 7.29 (d, 2H, J ¼ 8.4 Hz), 7.26–7.15 (m, 5H), 4.47 (td, 1H, subsequently injected for seven minutes. A cross-linking step32
J ¼ 6.2; 9.4; 14.6 Hz), 2.98 (dd, 1H, J ¼ 6.2; 14.7 Hz), 2.85 (dd, 1H, consisting of a three-minute injection of EDC/NHS, containing
J ¼ 10.2; 14.7 Hz), 2.68 (m, 1H), 2.58 (d, 3H, J ¼ 4.2 Hz), 2.50 (m, 10 mM CaCl2, followed by a deactivation injection during ten
2H), 2.41 (dd, 1H, J ¼ 8.0, 16.3 Hz), 2.25 (dd, 1H, J ¼ 6.7, 16.3 minutes of 1 M ethanolamine, pH 8.5, 10 mM CaCl2 concluded
Hz), 1.68 (m, 2H); 13C-NMR (125 MHz, DMSO-d6) d 173.9, 173.7, the immobilization procedure. The running buffer during
171.9, 145.0, 144.4, 143.5, 138.5, 138.5, 137.4, 133.4, 129.6 (2C), immobilization was 10 mM Hepes, pH 7.4, 0.2 M CaCl2, 0.1 M
128.5–126.6 (7C), 125.9, 54.5, 41.7, 37.9, 36.6, 34.3, 32.6, 26.0. NaCl with 0.005% surfactant P-20.
HRMS: (M + H)+ calculated: 460.2236, found; 460.2242. LC-UV
purity system A tR: 2.43 min, 98.8%.
The running buffer for the interaction experiments was 50
mM Tris–HCl, pH 7.4, 0.2 M CaCl2, 0.1 M NaCl, 0.005%
(R)-N4-Hydroxy-N1-((S)-1-(methylamino)-1-oxo-3-phenyl- surfactant P-20, 5% DMSO. Solvent correction solutions were
propan-2-yl)-2-(4-(pyridin-3-yl)phenethyl)succinamide (2). To a prepared from running buffer with 4.5% and 5.8% DMSO stock
solution of the above obtain compound 1 (150 mg, 0.33 mmol) solutions, as described in the Biacore S51 methodology hand-
in DMF (5 mL) was added N-methylmorpholine (NMM) (0.16 book (Biacore AB/GE Healthcare Biosciences, Uppsala Sweden).
mL, 1.38 mmol). The mixture was cooled to 0 ꢃC and benzo-
triazole-1-yl-oxy-tris-(dimethylamino)-phosphonium hexauoro-
phosphate (BOP) (165 mg, 0.38 mmol) was added. The reaction
mixture was stirred for 30 min at 0 ꢃC, where aer hydroxy-
lammonium chloride (48 mg, 0.69 mmol) was added. The
reaction was allowed to warm to room temperature and stirred
overnight. The reaction mixture was concentrated under
reduced pressure. The residue was diluted with EtOAc, washed
with 1 N HCl, saturated NaHCO3 and brine, dried over anhy-
drous Na2SO4, and concentrated under reduced pressure. The
residue was puried by silica gel column chromatography to
afford the product 2 in 68% yield (105 mg).
Inhibition assay
The enzyme activity was measured in 96-well plates (Corning/
Sigma Aldrich, Sweden) at a nal volume of 150 mL, where 10 mL
of 3 M CaCl2 dissolved in H2O, 115 mL of buffer (50 mM Tris, pH
7.5), 5 mL of inhibitor dissolved in DMSO, and 10 mL of the
internally quenched uorogenic peptide substrate Mca-Pro-Leu-
Gly-Leu-Dpa-Ala-Arg-NH2 (RnD Systems Europe Ltd) dissolved
in buffer with 12.5% DMSO to 750 mM, were added to the wells.
ꢃ
This was incubated at 30 C for 5 minutes before the reaction
was started by adding 10 mL of 3 mg mLꢀ1 MMP-12. The nal
concentration of CaCl2 was 200 mM, while the nal concen-
tration of DMSO was 4.2% (v/v) and that of MMP-12 and the
substrate was 10 nM and 50 mM respectively. The enzyme stock
solution (Tris buffer pH 7.5, 5 mM CaCl2) was diluted in assay
buffer just prior to starting the reaction. Plates were read on a
Fluoroskan Ascent (Thermo Labsystems Oy, Helsinki, Finland)
using lex ¼ 320 nm and lem ¼ 460 nm for 25 minutes.
Apparent Ki-values were determined using slightly different
conditions and the assay set-up: the assay was performed at
room temperature in 200 mM calcium acetate, 50 mM Tris–HCl,
pH 7.5, 1% DMSO, 50 mM substrate and 2 nM enzyme. The
addition of substrate started the reaction aer a pre-incubation
of enzyme and inhibitor for 10 minutes. The plates were read for
12 minutes.
1H-NMR (500 MHz, DMSO-d6) d 10.45 (s, 1H), 8.86 (d, 1H, J ¼
2.2 Hz), 8.77 (s, 1H), 8.54 (dd, 1H, J ¼ 1.5; 4.6 Hz), 8.15 (d, 1H,
J ¼ 8.3 Hz), 8.04 (dd, 1H, J ¼ 2.0; 8.0 Hz), 7.97 (q, 1H), 7.62 (d,
2H, J ¼ 8.0 Hz), 7.46 (dd, 1H, J ¼ 4.6; 8.0 Hz), 7.21 (d, 2H, J ¼ 8.0
Hz), 7.26–7.16 (m, 5H), 4.46 (td, 1H, J ¼ 5.2; 9.1; 14.3 Hz), 3.05
(dd, 1H, J ¼ 5.0; 13.7 Hz), 2.84 (dd, 1H, J ¼ 9.6; 13.7 Hz), 2.61 (m,
1H), 2.59 (d, 3H, J ¼ 4.6 Hz), 2.44–2.38 (m, 2H), 2.10 (dd, 1H, J ¼
7.2, 14.8 Hz), 2.01 (dd, 1H, J ¼ 8.2, 14.8 Hz), 1.65–1.57 (m, 2H);
13C-NMR (125 MHz, DMSO-d6) d 173.9, 172.0, 168.0, 148.7,
147.9, 142.5, 138.7, 135.9, 135.0, 134.3, 129.5 (2C), 128.5–126.6
(7C), 124.3, 54.6, 42.4, 37.7, 36.4, 33.9, 32.6, 26.1. HRMS: (M +
H)+ calculated: 475.2345, found: 475.2337. LC-UV purity system
B tR: 2.60 min, 99.0%.
In order to measure the activity of the immobilized enzyme, it
was immobilized manually to the entire surface of a CM5 chip
Enzyme
Pure recombinant human MMP-12 was produced essentially as outside the instrument, using the same procedure as described
described by Morales et al.31 The protein consisted of the cata- above with the S51-instrument. The substrate (60 mL of 50 mM in
lytic domain, comprising amino acids 100–263 of the entire running buffer) was subsequently pipetted as a droplet to the
naturally translated pre-protein.
surface. Aer 20 minutes under gentle agitation at room temper-
ature, the substrate droplet was transferred to a 96-well plate and
uorescence was measured, as described above. The amount of
immobilized protein was quantied according to Stenberg et al.33
Interaction experiments
A Biacore S51 instrument (Biacore AB/GE Healthcare Biosci-
ences, Uppsala, Sweden) was used for all experiments, except for
the characterisation of compound 1 interactions, which were
Data analysis
performed with a Biacore 2000 instrument. The assay was per- Biacore S51 soware and/or Biaevaluation 3.2 (Biacore AB/GE
formed according to the procedures previously described by Healthcare Biosciences, Uppsala, Sweden) were used for
440 | Med. Chem. Commun., 2013, 4, 432–442
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Concise Article
MedChemComm
¨
¨
¨ ¨
¨
evaluating the biosensor data. The response of a blank sample
injection following the analyte injection was subtracted from
the analyte response. The analyte response was then normalized
with respect to the response from a positive control.
Apparent Ki-values were determined from product vs. time
data (entire progress curves) by tting of the integrated
Michaelis–Menten equation:34
8 H. Nordstrom, T. Gossas, M. Hamalainen, P. Kallblad,
¨
S. Nystrom, H. Wallberg and U. H. Danielson, J. Med.
Chem., 2008, 51, 3449–3459.
9 J. I. Levin, J. Chen, M. Du, M. Hogan, S. Kincaid, F. C. Nelson,
A. M. Venkatesan, T. Wehr, A. Zask, J. DiJoseph, L. M. Killar,
S. Skala, A. Sung, M. Sharr, C. Roth, G. Jin, R. Cowling,
K. M. Mohler, R. A. Black, C. J. March and J. S. Skotnicki,
Bioorg. Med. Chem. Lett., 2001, 11, 2189–2192.
10 S. Pikul, K. L. McDow Dunham, N. G. Almstead, B. De,
M. G. Natchus, M. V. Anastasio, S. J. McPhail, C. E. Snider,
Y. O. Taiwo, T. Rydel, C. M. Dunaway, F. Gu and
G. E. Mieling, J. Med. Chem., 1998, 41, 3568–3571.
11 M. Yamamoto, H. Tsujishita, N. Hori, Y. Ohishi, S. Inoue,
S. Ikeda and Y. Okada, J. Med. Chem., 1998, 41, 1209–
1217.
12 D. E. Levy, F. Lapierre, W. Liang, W. Ye, C. W. Lange, X. Li,
D. Grobelny, M. Casabonne, D. Tyrrell, K. Holme,
A. Nadzan and R. E. Galardy, J. Med. Chem., 1998, 41, 199–
223.
13 P. O. Markgren, M. Hamalainen and U. H. Danielson, Anal.
Biochem., 2000, 279, 71–78.
14 R. Karlsson, P. S. Katsamba, H. Nordin, E. Pol and
D. G. Myszka, Anal. Biochem., 2006, 349, 136–147.
15 R. A. Copeland, Methods Biochem. Anal., 2005, 46, 1–265.
16 T. Gossas and U. H. Danielson, Biochem. J., 2006, 398, 393–
398.
17 G. Rosenblum, S. O. Meroueh, O. Kleifeld, S. Brown,
S. P. Singson, R. Fridman, S. Mobashery and I. Sagi, J. Biol.
Chem., 2003, 278, 27009–27015.
ꢀ
ꢁ
8
9
S0 ꢀ Vt
>
<
>
=
S0
KM
½Pꢄ ¼ S0 ꢀ KMW
e
(1)
>
:
>
;
KM
where W is a Lambert's Omega function which satises the
equations:
W(z) ¼ y
yey ¼ z
(2)
The Lambert Omega function was implemented in Sigma-
Plot 2000 using 5 iterations of the algorithm of Corless et al.35
The time dependence of inhibition was evaluated in BIAe-
valuation 3.0.2 by tting the following equation to product vs.
time data:
ꢀ
ꢁ
ꢂ
ꢃ
vi ꢀ vs
½Pꢄ ¼ vst þ
1 ꢀ eðꢀkobstÞ
(3)
kobs
where
ꢀ
ꢁ
I
18 F. J. Moy, P. K. Chanda, J. Chen, S. Cosmi, W. Edris,
J. I. Levin, T. S. Rush, J. Wilhelm and R. Powers, J. Am.
Chem. Soc., 2002, 124, 12658–12659.
kobs ¼ koff 1 þ
(4)
Ki
19 J. F. Morrison and C. T. Walsh, Adv. Enzymol. Relat. Areas
Mol. Biol., 1988, 61, 201–301.
20 J. F. Morrison, Biochim. Biophys. Acta, 1969, 185, 269–
286.
Acknowledgements
This work was supported by the Swedish Agency for Innovation
Systems (VINNOVA) and the Swedish Research Council (VR). We
would also like to thank Tatiana Agback (Medivir AB) and Kurt
Benkestock (Medivir AB) for help regarding the analyses of 1
and 2.
¨
21 M. Elinder, P. Selhorst, G. Vanham, B. Oberg, L. Vrang and
U. H. Danielson, Biochem. Pharmacol., 2010, 80, 1133–
1140.
22 C. Lombard, J. Saulnier and J. Wallach, Biochimie, 2005, 87,
265–272.
23 R. Johnson, A. G. Pavlovsky, D. F. Ortwine, F. Prior,
C. F. Man, D. A. Bornemeier, C. A. Banotai, W. T. Mueller,
P. McConnell, C. Yan, V. Baragi, C. Lesch, W. H. Roark,
M. Wilson, K. Datta, R. Guzman and H. K. Dyer, J. Biol.
Chem., 2007, 282, 27781–27791.
References
1 N. D. Rawlings, D. P. Tolle and A. J. Barrett, Nucleic Acids Res.,
2004, 32, D160–D164.
2 G. L. Snider, Am. J. Respir. Crit. Care Med., 2003, 167, 1045– 24 M. Tegoni, F. Dallavalle and M. Amelia Santos, J. Inorg.
1046. Biochem., 2004, 98, 209–218.
3 M. Liu, H. Sun, X. Wang, T. Koike, H. Mishima, K. Ikeda, 25 M. M. Bernardo, S. Brown, Z. H. Li, R. Fridman and
T. Watanabe, N. Ochiai and J. Fan, Arthritis Rheum., 2004,
50, 3112–3117.
4 B. Turk, Nat. Rev. Drug Discovery, 2006, 5, 785–799.
5 L. M. Coussens, B. Fingleton and L. M. Matrisian, Science,
2002, 295, 2387–2392.
S. Mobashery, J. Biol. Chem., 2002, 277, 11201–11207.
26 A. C. Dublanchet, P. Ducrot, C. Andrianjara, M. O'Gara,
R. Morales, D. Compere, A. Denis, S. Blais, P. Cluzeau,
K. Courte, J. Hamon, F. Moreau, M. L. Prunet and
A. Tertre, Bioorg. Med. Chem. Lett., 2005, 15, 3787–3790.
6 C. M. Overall and O. Kleifeld, Br. J. Cancer, 2006, 94, 941–946. 27 P. Cuniasse, L. Devel, A. Makaritis, F. Beau, D. Georgiadis,
7 C. M. Overall and O. Kleifeld, Nat. Rev. Cancer, 2006, 6, 227–
239.
M. Matziari, A. Yiotakis and V. Dive, Biochimie, 2005, 87,
393–402.
This journal is ª The Royal Society of Chemistry 2013
Med. Chem. Commun., 2013, 4, 432–442 | 441

View Article Online
MedChemComm
Concise Article
28 R. Lang, A. Kocourek, M. Braun, H. Tschesche, R. Huber, 32 P. O. Markgren, M. T. Lindgren, K. Gertow, R. Karlsson,
¨
¨ ¨
W. Bode and K. Maskos, J. Mol. Biol., 2001, 312, 731–742.
29 H. Nar, K. Werle, M. M. Bauer, H. Dollinger and B. Jung,
J. Mol. Biol., 2001, 312, 743–751.
30 R. A. Copeland, D. L. Pompliano and T. D. Meek, Nat. Rev.
Drug Discovery, 2006, 5, 730–739.
M. Hamalainen and U. H. Danielson, Anal. Biochem., 2001,
291, 207–218.
33 E. Stenberg, B. Persson, H. Roos and C. Urbaniczky, J. Colloid
Interface Sci., 1991, 143, 513–526.
34 C. T. Goudar, J. R. Sonnad and R. G. Duggleby, Biochim.
Biophys. Acta, 1999, 1429, 377–383.
31 R. Morales, S. Perrier, J. M. Florent, J. Beltra, S. Dufour, I. De
Mendez, P. Manceau, A. Tertre, F. Moreau, D. Compere, 35 R. M. Corless, G. H. Gonnett, D. E. G. Hare, D. J. Jeffrey
A. C. Dublanchet and M. O'Gara, J. Mol. Biol., 2004, 341,
1063–1076.
and D. E. Knuth, Adv. Comput. Math., 1996, 5, 329–
359.
442 | Med. Chem. Commun., 2013, 4, 432–442
This journal is ª The Royal Society of Chemistry 2013