Additivity or cooperativity: Which model can predict the influence of simultaneous incorporation of two or more functionalities in a ligand molecule?

Article abstract of DOI:10.1016/j.ejmech.2014.11.056

Predicting how binding affinity responds to ligand structural modifications in structure-activity relationship studies (SAR) is a major challenge in medicinal chemistry. This is particularly true when two or more of these modifications are carried out simultaneously. In this study, we present binding affinity data from several series of thermolysin inhibitors in which simultaneous structural modifications were investigated to determine whether they are cooperative or additive. Data revealed that, while additivity is at work in some cases, cooperativity is more commonly demonstrated. Cooperativity and additivity were then correlated with ligand descriptors, such as the spacing and the topological features of the modified groups, in a manner that may provide guidance as to when each model should be utilized. Cooperativity was particularly associated with contiguous groups and small unbranched hydrophobic side chain. Additivity, on the other hand, was associated with moderately distant hydrophobic group combinations and side chain branching. Such correlations can improve the predictability of SAR studies and can provide a starting point for additional investigations that may lead to further significant enhancements in the current scoring functions.

Full text of DOI:10.1016/j.ejmech.2014.11.056

Accepted Manuscript
Additivity or Cooperativity: Which Model Can Predict the Influence of Simultaneous
Incorporation of Two or More Functionalities in a Ligand Molecule?
Nader N. Nasief , David Hangauer
PII:
S0223-5234(14)01098-8
10.1016/j.ejmech.2014.11.056
EJMECH 7549
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 10 July 2014
Revised Date: 10 October 2014
Accepted Date: 27 November 2014
Please cite this article as: N.N Nasief, D. Hangauer, Additivity or Cooperativity: Which Model Can
Predict the Influence of Simultaneous Incorporation of Two or More Functionalities in a Ligand
Molecule?, European Journal of Medicinal Chemistry (2014), doi: 10.1016/j.ejmech.2014.11.056.
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.

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT
Additivity or Cooperativity: Which Model Can Predict the Influence of
Simultaneous Incorporation of Two or More Functionalities in a Ligand
Molecule?
*
*
Nader N Nasief and David Hangauer
Department of Chemistry, University at Buffalo, The State University of New York, Buffalo,
New York 14260
KEYWORDS: Drug design, Lead optimization, Structure-activity relationship, Protein-ligand
binding, Molecular recognition, Thermolysin
ABSTRACT:
Predicting how binding affinity responds to ligand structural modifications in structure-activity
relationship studies (SAR) is a major challenge in medicinal chemistry. This is particularly true
when two or more of these modifications are carried out simultaneously. In this study, we present
binding affinity data from several series of thermolysin inhibitors in which simultaneous
structural modifications were investigated to determine whether they are cooperative or additive.
Data revealed that, while additivity is at work in some cases, cooperativity is more commonly
demonstrated. Cooperativity and additivity were then correlated with ligand descriptors, such as
the spacing and the topological features of the modified groups, in a manner that may provide
guidance as to when each model should be utilized. Cooperativity was particularly associated
with contiguous groups and small unbranched hydrophobic side chain. Additivity, on the other
hand, was associated with moderately distant hydrophobic group combinations and side chain
branching. Such correlations can improve the predictability of SAR studies and can provide a
starting point for additional investigations that may lead to further significant enhancements in
the current scoring functions.
∗
Correspondences should be addressed to D H at hangauer@buffalo.edu or N N N at nnnasief@yahoo.com.
1

ACCEPTED MANUSCRIPT
1
. Introduction:
Lead optimization involves cycles of structural modifications which aim at improving the
lead’s binding affinity or enhancing its pharmacokinetic properties. A typical structural
modification might be the replacement of an H or a functional group with another. It is not
uncommon for a medicinal chemist to perform more than one structural modification at a time in
order to reduce the number of compounds to be synthesized. For example, the structural
modifications A→X and B→Y could be carried out individually (two compounds are
synthesized) and then combined in a third compound after evaluating whether these structural
modifications move the process towards the desired goal. Alternatively, a medicinal chemist may
opt to synthesize the third compound after evaluating only one or neither of these modifications
(i.e. one or two compounds are synthesized). It should be noted that each of these choices could
be misleading in one way or another. For instance, suppose the structural modifications A→X
and B→Y are carried out, and one of them is found to be disadvantageous. The medicinal
chemist might be discouraged from synthesizing the third compound that has both modifications,
even though this third compound, if synthesized and evaluated, might display what has
previously been termed “positive cooperativity” between the two modifications [1-3] and—as a
consequence—might be good. On the other hand, skipping the evaluation of the individual
modifications carries the risk of missing good modifications if the third compound is not good
because the two modifications are negatively cooperative (i.e. the individual modifications are
good, while the combination is bad). It is therefore crucial for medicinal chemists to be capable
of accurately predicting not only the impact of the individual structural modifications on the
binding affinity (or the pharmacokinetic property that is desired to be improved), but also the
correct model that is to be employed when two, or even more, structural modifications are
combined in a ligand.
<
Insert Figure 1>
Now, how can we identify cooperativity or additivity between two structural modifications in
a ligand? A typical analysis that has been commonly used to study cooperative phenomena is the
double mutant cycle analysis [4-8]. This analysis has been used to determine whether ligand
structural modifications are cooperative or additive with regard to their binding affinity/free
energy [1-2, 9]. To illustrate how this analysis works in general terms, consider Figure 1. In this
2

ACCEPTED MANUSCRIPT
figure, the relationship between the structural modifications H→X and H’→Y is evaluated by
comparing the binding free energy change (the differential binding energy)[10] occurring when
both groups exist in the ligand (∆∆G(H,H’→X,Y)) with the sum of the binding free energy changes
occurring when each group exists individually (∆∆G(H,H’→X,H’) + ∆∆G(H,H’→H,Y)). There are three
possible outcomes: (1) ∆∆G(H,H’→X,Y) = ∆∆G(H,H’→X,H’) + ∆∆G(H,H’→H,Y); (2) ∆∆G(H,H’→X,Y)
<
∆∆G(H,H’→X,H’) + ∆∆G(H,H’→H,Y); and (3) ∆∆G(H,H’→X,Y) > ∆∆G(H,H’→X,H’) + ∆∆G(H,H’→H,Y). In the
first case H→X and H’→Y demonstrate additivity, while in both the second and the third cases
H→X and H’→Y are cooperative (The second is a case of positive cooperativity, and the third is
a case of negative cooperativity). Alternatively, one can compare either the differential binding
free energy associated with the replacement of the ligand H with group X in the presence
(∆∆G(H,Y→X,Y)) and absence (∆∆G(H,H’→X,H’)) of group Y, or the differential binding energy
caused by the H’→Y replacement in the presence (∆∆G(X,H’→X,Y)) and absence (∆∆G(H,H’→H,Y)) of
group X. If, for example, ∆∆G(H,Y→X,Y) and ∆∆G(H,H’→X,H’) are equal, H→X and H’→Y are
deemed additive. On the other hand, a more negative and a more positive ∆∆G(H,Y→X,Y) values
indicate positive and negative cooperativities, respectively. Cooperativity may therefore be
defined as a variation in ∆∆G(H→X) which occurs when a second group Y is incorporated in the
ligand molecule.
1
.1. Additivity/Cooperativity and the Partitioning of the Differential Binding Energy
Figure 2 illustrates a “three-dimensional” Born-Haber cycle which can be used to partition the
binding of two ligands LH and LX to a biological target P [11-12]. These two ligands differ only
in that ligand LX has the functional group X replacing an H in ligand LH. The differential
binding energy caused by this functional group replacement is therefore represented by the free
energy difference (∆GLX − ∆G ). Because each of these free energy terms can be partitioned
LH
into basic components as illustrated by Eq. 1 which represents the partitioning of ∆G , the
LH
differential binding energy can be partitioned as well, and this partitioning is illustrated by Eq. 2.
This equation describes the partitioning of the differential free energy ∆∆G(H→X) into three major
components: the differential desolvation of the ligand (∆GLX-desolv − ∆GLH-desolv), the differential
ligand-protein association (∆GLX-assoc − ∆GLH-assoc), and the differential ligand-protein complex
resolvation (∆GLX-resolv − ∆GLH-resolv). It should be noted that both Eq. 1 and Eq. 2 should include
3

ACCEPTED MANUSCRIPT
other terms if conformational or ionization changes occur in either the ligands or the target
during the course of binding.
<
Insert Figure 2>
+ ∆GP-desolv + ∆GLH-assoc + ∆GLH-resolv
LH-desolv
∆
G = ∆G
(Eq. 1)
LH
∆∆G(H→X) = ∆GLX − ∆GLH = (∆GLX-desolv − ∆GLH-desolv) + (∆GLX-assoc − ∆GLH-assoc) + (∆GLX-resolv
−
∆
GLH-resolv (Eq. 2)
)
Given that additivity and cooperativity were previously defined in terms of variation in the
differential free energy, these phenomena can be explained through the differential free energy
partitioning. Additivity, for instance, exists when the differential free energy of a structural
modification (e.g. H→X) is the same, no matter whether the initial or the final group of the
second modification exists in the ligand (e.g. H’ or Y; Figure 1): ∆∆G(H,H’→X,H’) = ∆∆G(H,Y→X,Y)
.
This case could be obtained if (1) none of the differential free energy components illustrated in
Eq. 2 changes when the structural modification H→X is carried out in presence of the H’ or the
Y of the modification H’→Y; or (2) in the presence of Y vs. H’, more than one of these free
energy components change in opposite directions so that no net change in the differential free
energy is produced (e.g. (∆GLX-desolv − ∆GLH-desolv) and (∆GLX-assoc − ∆GLH-assoc) change in
opposite directions but with the same magnitude). On the other hand, cooperativity is obtained
when the differential binding free energy of the modification H→X differs based on the
existence of the initial or the final group of the second modification in the ligand (∆∆G(H,H’→X,H’)
≠
∆∆G(H,Y→X,Y)). Cooperativity is obtained when one or more of the differential free energy
components vary when group Y exists vs. the H’ (i.e. Y modulates one or more of the differential
free energy components). For example, (∆GLX-desolv − ∆GLH-desolv) and/or (∆GLX-assoc − ∆GLH-assoc
might become more negative (more favorable) in presence of group Y. As a consequence,
∆G(H,Y→X,Y) becomes more favorable and positive cooperativity is produced (more details are
)
∆
given in the supplementary materials). It is important to note that the way the differential free
energy is partitioned in Figure 2 is not the only way to partition this quantity; rather, we
previously utilized another partitioning scheme that bypasses the gas phase to explain the origin
of cooperativity [2].
4

ACCEPTED MANUSCRIPT
1
.2. Additivity or Cooperativity?
Deviations from the additivity principle occur frequently in SAR studies. For example, Patel et
al. analyzed eight nearly complete combinatorial libraries assayed on several different biological
responses and showed that only half exhibit clear additive behavior [13]. Furthermore, the
importance of the cooperativity (nonadditivity) principle in molecular recognition and ligand
binding studies was demonstrated by Muley and co-workers, who showed that hydrophobic
interactions and hydrogen bonding reinforce each other in thrombin inhibitors [1]. Cooperativity
was then regarded by Bissantz et al. as one of the important factors affecting molecular
interactions [14]. Given that both additivity and cooperativity can be obtained experimentally,
the major challenge we face is to determine when additivity is a valid assumption and when, on
the contrary, a cooperative model would more accurately describe the binding process. In order
to address this challenge, extensive studies of the relationships among functional groups in
multiple ligand-biological target systems are needed. These studies should focus on (1)
correlating additive/cooperative behaviors with the structural features of both the ligand and the
biological target; (2) unraveling the interwoven nature of the binding elements and understanding
when particular elements become more significant than others; and (3) identifying recurring
additivity/cooperativity patterns that can be utilized prospectively to predict the outcomes of
SAR studies.
1
.3. Designing a Study to Explore the Additivity/Cooperativity in Thermolysin Inhibitors
The study presented in the current contribution was intended to be an initial endeavor to
address the aforementioned challenge. The phosphonamidate-thermolysin system, which was the
subject of some of our recent studies, [2, 15-16] was deemed an appropriate ligand-protein
system for this study. We, therefore, aimed at identifying patterns of additivity and/or
cooperativity in series of phosphonamidate ligands with respect to their binding to thermolysin
(TLN). TLN is a thermostable bacterial zinc-metalloprotease that is obtained from Bacillus
thermoproteolyticus [17-19]. It shares common active site structural features with other zinc-
metalloproteases such as carboxypeptidase A [20] and angiotension-converting enzyme (ACE)
[21]. One of these common structural features is the presence of zinc ion which facilitates the
substrate peptide bond cleavage via coordinating the C=O group. Other important features of the
TLN active site are the S1’ pocket, which is a deep hydrophobic pocket that largely determines
5

ACCEPTED MANUSCRIPT
substrate specificity, [22] and the S2’ pocket, which is a shallow, flat, solvent-exposed pocket.
-
The phosphonamidate inhibitors chosen for this study coordinate with the zinc ion via the PO2
moiety (Figure 3). They also form several H-bonds with various residues in the active site. These
-
H-bonds include a charge-assisted H-bond formed between the inhibitor terminal COO group
and the Asn112 –C(=O)NH group (Figure 3). In addition, these inhibitors can have hydrophobic
2
1 2
side chains which bind both the S1’ and the S2’ pockets (e.g. R and R groups).
The study presented herein was designed to investigate the additive or cooperative relationship
-
between (1) the ligand terminal COO group and the R
1
hydrophobic side chain; (2) the ligand
-
terminal COO group and the R hydrophobic side chain; and (3) the ligand two major
2
hydrophobic side chains R and R . It is worth mentioning that, with regard to the second group
1
2
-
pair (i.e. the COO and the R
2
side chain), positive cooperativity has already been demonstrated
-
between the COO and the Me side chain [2]. The design of the current study furthermore
involved evaluating the correlations between the experimentally observed cooperative/additive
patterns and some of the structural features of the ligand and the protein. For example, questions
like the following were investigated: is the relationship between a polar group and a hydrophobic
-
side chain (the COO and R
1
or R
2
) different from the relationship between two hydrophobic side
chains (R and R ), in terms of the cooperative/additive behavior? Is the cooperative/additive
1
2
-
-
2 1
behavior distance-dependent (the COO and R vs. the COO and R )? Is this behavior dependent
on the nature of the protein pocket (S1’ vs. S2’)? Is this behavior dependent on hydrophobic side
chain descriptors such as the size, the degree of branching, and the aliphaticity/aromaticity of the
hydrophobic side chain? Figure 3 illustrates the group pairs that were studied and indicates some
of the characteristics that were correlated with the observed cooperative/additive behaviors.
<
Insert Figure 3>
-
The first studied group pair was the (R , COO ) pair. In order to study this pair, two series of
1
inhibitors were designed (Scheme 1: Series I and II). In both series, the hydrophobic side chain
-
was grown from Me to Et, n-Pr, and i-Bu. One of these series, however, lacks the COO group
-
(
i.e. Series I). The absence of the COO group in this series made it possible to evaluate how the
-
COO group influences the contributions of different R side chains to the binding affinity, and
1
thus double mutant cycles similar to the ones shown in Figure 1 could be constructed. Second, in
-
order to study the (R , COO ) group pair, two series, which differ only in the presence or absence
2
6

ACCEPTED MANUSCRIPT
-
2
of the COO group, were investigated (Series III and IV, Scheme 1). Initially, the R side chain in
both series was grown from H to Me, Et, n-Pr, and n-Bu. Later on, the scope of studying this pair
was expanded to include branched and aromatic side chains. Furthermore, the influence of
1
truncating the R side chain from i-Bu to Me on the cooperative/additive behavior of this group
pair was explored by testing two additional series having a varied R side chain and a Me as the
2
-
R
1
side chain, without and with the COO (Series V and VI, Scheme 1). Finally, the (R
1
, R
2
) pair
was studied by comparing the data of series V, which has a Me side chain as R , with the data of
1
some ligands from series III (series III-R, Scheme 1). This comparison involved varying the
hydrophobicity of R
1 2 2
and R individually and simultaneously. For example, the R side chain
was varied from Me to Et, i-Bu, and Bn, while the R side chain was either Me (series V) or i-Bu
1
(series III-R).
Details about the double mutant cycles that evaluate the cooperative/additive relationships in
the group pairs being studied herein are given later. In the next section, the synthesis of the
ligands that belong to these group pairs will be discussed in detail.
<Insert Scheme 1>
2
. Results and Discussions
2
.1. Chemistry [23]:
The synthesis of the intermediate 4 is illustrated in Scheme 2. Commercially available
benzylcarbamate was heated in an aqueous basic formaldehyde solution to give benzyl N-
hydroxymethyl) carbamate 1 [24]. The terminal hydroxyl group of intermediate 1 was
(
acetylated using acetic anhydride to give benzyl N-(acetoxymethyl) carbamate 2. Intermediate 2
was converted to dimethyl N-benzyloxycarbonyl aminomethylphosphonate 3 by refluxing with
trimethyphosphite. Intermediate 4 was obtained by the hydrolysis of one of the two methyl
phosphonate esters in 3 using 10% NaOH solution.
<Insert Scheme 2>
The synthesis of ligands 1-38 is shown in Scheme 3. First, the commercially available Boc-
protected amino acids were coupled to the hydrochloride salts of various amines or amino acid
esters to give the intermediates IA1-38. Either EDCI/HOBt or PyBop in anhydrous DMF were
7

ACCEPTED MANUSCRIPT
used effectively to achieve the coupling in presence of diisopropylethylamine. The intermediates
IB1-38 were then obtained as hydrochloride salts upon the removal of the Boc groups from IA1-
3
8. In order to remove the Boc groups, either 3 M HCl/MeOH solution was used, or HCl gas was
bubbled into an ethyl acetate solution of the Boc-protected intermediate to avoid
transesterification with MeOH.
<Insert Scheme 3>
Finally, intermediate 4 was coupled to each of the intermediates IB1-38 in anhydrous
dichloromethane using PyBop as the coupling reagent to give IC1-38. Compounds IC1-38 were
then hydrolyzed using lithium hydroxide to give the final compounds 1-38 either as dilithium
salts (when the carboxylate group is present) or monolithium salts (when the carboxylate group
is absent). All the final compounds were purified by reverse-phase HPLC to at least 95% purity.
The substitution patterns of the synthesized ligands and their intermediates are illustrated in
Table 1.
<Insert Table 1>
2
.2. Biochemical assay:
The inhibition constants (K
i
) for compounds 1-38 were determined in a standard TLN
biochemical assay using 2-furanacryloyl-Gly-Leu-NH as a substrate [25]. The assay was carried
2
out in a high salt concentration which improves the enzyme activity as well as the substrate
binding to the enzyme. More details about the assay conditions are given in the experimental
section. The K and the corresponding free energy of binding values for the TLN inhibitors 1-38
i
are provided in Table 2.
<Insert Table 2>
-
2
.3. The relationship between the hydrophobic side chain R and the terminal COO group
1
<
Insert Figure 4>
-
In order to investigate the relationship between the R side chain and the COO group, the
1
-
differential binding free energies of the Me→R
1
modifications in absence of the COO group
were compared to the differential binding energies of the same modifications in presence of the
-
COO . This comparison is illustrated in the double mutant cycle shown in Figure 4. In this
8

ACCEPTED MANUSCRIPT
double mutant cycle, ligand 1 is mutated to 2, 3, or 4, and the differential binding energy in
-
absence of the COO (∆∆G(Me,H→R1,H)) is obtained when the binding free energy of 1 is subtracted
from that of either 2, 3, or 4. This double mutant cycle also involves the mutation of ligand 5 to
6
, 7, or 8, which yields values for (∆∆G(Me,COO→R1,COO)). These values are obtained by subtracting
the free energy of 5 from that of 6, 7, or 8. The values of the differential free energies in both the
-
absence and the presence of the COO are given in Table 3. This table also includes values for
(∆∆G(Me,COO→R1,COO) − ∆∆G(Me,H→R1,H)), which is the term that indicates positive cooperativity if
negative, negative cooperativity if positive, and additivity if 0. As shown in Table 3, the
-
replacement of Me with Et is not influenced by the presence of the COO (i.e. additive). As the
side chain is grown, mild cooperativity starts to show up. For example, the Me→n-Pr and the
Me→i-Bu replacements are little more contributive to the binding free energy in presence of the
-
COO (by 2.1 kJ/mol in the former and 1.8 kJ/mol in the latter case: about 2X additional
improvement in Ki compared to what would be anticipated based on the additivity principle).
<Insert Table 3>
-
With regard to the mild cooperativity observed in the data of the group pair (R
several points should be noted.
1
, COO ),
<
Insert Figure 5>
-
(
1) It is not likely that the COO would influence the differential desolvation (i.e. ∆GLR1-desolv
−
∆GLMe-desolv: Eq. 2) of ligands in which the Me is grown to a larger side chain. For the differential
-
desolvation to be dependent on the presence of the COO , the water molecules hydrating the R
1
side chain (i.e. in the unbound state) should reorganize differently when the Me is grown to a
-
-
larger side chain in presence vs. in absence of the COO . This might require the COO to be in
direct contact with the hydration layer of the R side chain, but, most likely, this is not the case.
1
-
For example, Figure 5a reveals that the R
1
side chain and the COO group might be positioned on
-
opposite sides of the molecule, and therefore the required direct contact between the COO and
the R hydration layer might not be achievable. It should be noted that the conformation shown
1
in the figure is the bioactive conformation, which is not necessarily the same as the global
minimum of the unbound ligand (the conformation that needs to be investigated when ligand
desolvation is considered). However, the hydrophobic collapse between the two hydrophobic
side chains R and R in the bioactive conformation indicates that the spatial arrangement of the
1
2
9

ACCEPTED MANUSCRIPT
-
1 2
R , R , and the COO in this conformation is favorable in terms of free energy, and might
therefore be the same, or at least very close to the spatial arrangement of these groups in the
global minimum of the unbound ligand. Conclusions drawn from the bioactive conformation
regarding these groups might therefore be extrapolated to the unbound ligand.
(2) The differential resolvation (i.e. ∆GLR1-resolv − ∆GLMe-resolv: Eq. 2) of ligands in which the
-
Me is grown to a larger side chain is not likely influenced by the presence of the COO . In fact,
this differential resolvation term might be equivalent to zero both in presence and absence of the
-
COO . For this term to exist, the complex of a ligand with a Me side chain should be resolvated
differently from the complex of a ligand with a larger side chain. Given that both the Me and the
larger side chains are buried in the deep S1’ pocket (Figure 5b), the only way to achieve this
different resolvation is if the complex of the ligand with Me side chain traps water molecule(s) in
the unoccupied space of the S1’ pocket. This is however unlikely because of the large entropic
penalty of trapping such water molecule(s). Even if waters were trapped in the S1’ pocket of the
complex of a ligand with a Me side chain (and therefore ∆GLR1-resolv − ∆GLMe-resolv ≠ 0), this
water, being isolated to a great extent, would not be anticipated to sense the presence of the
-
-
COO , and therefore the differential resolvation term would not be influenced by the COO
group.
-
(
3) If the COO alters neither the differential desolvation nor the differential resolvation terms,
the mild cooperativity observed in Table 3 might be attributed to variation in the differential
-
association term (i.e. ∆GLR1-assoc − ∆GLMe-assoc: Eq. 2) caused by the COO group. The most
apparent cause for such variation is the mutual reinforcement of the direct interactions of the
-
-
COO and R groups with the protein (a H-bond with Asn112 in case of the COO , and dispersion
1
-
interactions in case of the R side chain; ∆G
shifts to the negative when the COO is
1
LR1-assoc
present, see supplementary materials section 1.1). It was previously mentioned that Muley and
coworkers described this kind of cooperative behavior in thrombin inhibitors [1]. The
cooperativity described by them, however, accounted for about an order of magnitude of activity
enhancement (i.e. strong cooperativity). It is important to note, however, that the functional
-
group pair studied by Muley et al., unlike the (R , COO ) pair, involves groups that are in close
1
proximity to each other. This might therefore suggest that the distance between the functional
groups involved in this type of cooperativity correlates with how much benefit can be obtained
10

ACCEPTED MANUSCRIPT
from the cooperative enhancement of non-covalent interactions (i.e. short and moderate distances
yield strong and mild cooperativity, respectively).
-
2
.4. The relationship between the hydrophobic side chain R
2
and the terminal COO group
<
<
Insert Figure 6>
Insert Figure 7>
-
In order to investigate the relationship between the R side chain and the COO group, the
2
differential binding free energies of the H→R modifications in absence and presence of the
2
-
COO group were compared. This comparison is illustrated by the double mutant cycle in Figure
6
. In this double mutant cycle, ligand 9 is mutated to 10-19, or 4; and the differential binding
-
energy in absence of the COO (∆∆G(H,H→R2,H)) is obtained when the binding free energy of 9 is
subtracted from that of any of ligands 10-19, and 4. This double mutant cycle also involves the
mutation of ligand 20 to 21-30, or 8, which yields values for (∆∆G(H,COO→R2,COO)). A
“∆∆G(H,COO→R2,COO)” value is obtained by subtracting the free energy of 20 from that of any of
ligands 21-30, and 8. The values of the differential free energies in both the absence and the
-
presence of the COO are given in Table 4. This table also includes values for
(∆∆G(H,COO→R2,COO) − ∆∆G(H,H→R2,H)).
<Insert Table 4>
The data presented in Table 4 reveal a complex cooperativity/additivity pattern between R
2
-
and the COO . This complex pattern is graphically illustrated in Figure 7 through plotting the
2
quantity “∆∆G(H,COO→R2,COO) − ∆∆G(H,H→R2,H)” against the R side chain modifications. Both
Table 4 and Figure 7 show that positive cooperativity is at maximum when the R side chain is
2
Me. This positive cooperativity, then, slightly diminishes when the R side chain is grown from
2
Me to Et, n-Pr and n-Bu (i.e. homologation: the blue line in Figure 7). The branching of the R
2
side chain reduces the positive cooperativity as well, but to a larger extent. For example, when
the Et side chain is branched to i-Pr and tert-Bu, positive cooperativity decreases from -4.7
kJ/mol to -3.1 and -1.3 kJ/mol, respectively. A larger decrease in positive cooperativity was
observed upon the branching of the n-Pr side chain to sec-Bu (-4.5 kJ/mol→ -0.1 kJ/mol: -0.1
kJ/mol indicates additivity), or to i-Bu and neopentyl (-4.5 kJ/mol→ -3.2 and 1.3 kJ/mol,
respectively). In case of the neopentyl, “∆∆G(H,COO→R2,COO) − ∆∆G(H,H→R2,H)” value, in fact,
11

ACCEPTED MANUSCRIPT
shows that some negative cooperativity is at work (i.e. indicated by the positive sign of the 1.3
kJ/mol). Positive cooperativity also diminishes when an aromatic moiety is introduced in the R2
side chain. For example, when the Me is replaced with either Bn or 2-thienylmethyl side chain,
positive cooperativity decreases from -5.3 kJ/mol to either -2.7 kJ/mol in case of the Bn or -3.7
kJ/mol in case of the 2-thienylmethyl side chain.
The data presented for series III and IV; therefore, indicate that there is a correlation between
the cooperativity/additivity pattern and the characteristics of the hydrophobic side chain being
investigated. Size, degree of branching, and aromaticity are some of these characteristics which
can be correlated with the cooperative behavior. For example, small and linear hydrophobic side
-
chains, like the Me, Et, and n-Bu are more synergistic with a nearby COO than a bulky,
branched, or aromatic side chains. It is worth mentioning that positive cooperativity, when
-
manifested in contiguous groups (i.e. R and the COO ), is more prominent than in moderately
2
-
distant ones (i.e. R
1
and the COO ). This confirms our earlier conclusion that the distance
between the correlated functional groups/side chains plays an important role in determining the
-
-
magnitude of cooperativity. Also, unlike the (R , COO ) pair, the close proximity of the COO
1
-
and the R
2
side chain likely causes the COO to strongly influence the structural and the
thermodynamic features of the water molecules hydrating the R , both in the unbound and the
2
2
complexed states. For example, water may reorganize differently when R is modified in
-
presence vs. absence of the COO group. This effect has been previously shown to be the most
likely cause for the dependency of the thermodynamic signature associated with the modification
-
-
2
of the R side chain on the COO group [16]. Furthermore, the influence of the COO on the
water reorganization that is associated with replacing the H with Me (9→10 and 20→21) in the
complexed state was demonstrated by X-ray crystallography and was correlated with the positive
-
cooperativity between the Me and the COO (water mediated cooperativity) [2]. This correlation
was suggested based on the analysis of the crystallographic data using the previously referenced
partitioning scheme, which bypasses the gas phase. Now, can this correlation be made if the
crystallographic data are analyzed using the partitioning scheme in Figure 2 which includes a gas
phase state?
<
Insert Figure 8>
12

ACCEPTED MANUSCRIPT
Simply put, this partitioning scheme can be equivalently used, and this is how. First, because
we need to examine the water networks of ligand-protein complexes, our focus should be on the
final step of the Born-Haber cycle in Figure 2 (the resolvation of the ligand-protein complex).
-
Second, we need to demonstrate that, when the COO exists, the differential resolvation term
(∆GLMe-resolv − ∆GLH-resolv) varies in a manner that produces the observed cooperativity. To do
this, Figure 8 that demonstrates the ligand-protein complex resolvation of 9 and 10 in
comparison with 20 and 21 can be consulted. This figure reveals that, in the presence of the
-
COO , the H-bond network of the LH-TLN resolvation shell is broken (e.g., ligand 20 vs ligand
9
). This can be translated into a less enthalpically and more entropically favorable resolvation, or
-
a ∆HLH-resolv and a –T∆SLH-resolv that are shifted in the presence of the COO toward the positive
-
and the negative, respectively. On the other hand, the presence of the COO supports the
resolvation network of the LMe-TLN complex. For instance, in case of ligand 21, the water
network is characterized by the presence of additional crystallographic waters that participate
into a more developed H-bond network. The resolvation of LMe-TLN is therefore more
-
enthalpically and less entropically favorable, and the presence of the COO shifts ∆H
and
LMe-resolv
–
T∆SLMe-resolv toward the negative and the positive, respectively. Taken together, a ∆HLMe-resolv
shifted toward the negative and a ∆HLH-resolv shifted toward the positive yield a (∆HLMe-resolv
HLH-resolv) shifted toward the negative; and a –T∆SLMe-resolv shifted toward the positive and a –
−
∆
T∆SLH-resolv shifted toward the negative yield a (−T∆SLMe-resolv – (−T∆SLH-resolv)) shifted toward the
positive (SI; Enthalpy: sections 1.2 # 3; Entropy: section 1.3 # 6). These shifts indeed explain the
experimentally observed enthalpic synergism and entropic antagonism [2]. The difficult
question, however, is whether the negative shift in the differential resolvation enthalpy can
overcome the positive shift in the differential resolvation entropy to produce a net negative shift
in differential resolvation energy (and, in turn, a positive cooperativity between the Me and the
-
COO ). Most likely, this is the case. Previously, we have shown that, for the phosphonamidate-
TLN system, the differential free energy correlates well with the differential enthalpy [16]. The
-
observed enthalpic synergism in the case of the (Me, COO ) pair is therefore anticipated to be
accompanied by a net free energy synergism. Now that we pointed out that the most likely cause
for the observed cooperativity is the differential resolvation, could there be any other source for
this cooperativity?
13

ACCEPTED MANUSCRIPT
Consider the ligand-protein association step. In this step, the additional Me group in LMe
contributes some additional dispersion interactions with the protein. The favorable impact of
these dispersion interactions on the differential enthalpy of association is not anticipated to
-
change, whether the COO group exists or not (i.e. (∆HLMe-assoc − ∆HLH-assoc) for ligands 21 and
2
0 is the same as for 10 and 9). Therefore, it is not likely for cooperativity between the Me and
-
the COO to be due to ligand-protein direct interaction. This hypothesis, however, is currently
under investigations using QM calculations. The other important consequence for the ligand-
protein association is the restriction of the ligand’s mobility upon binding. For example, the
ligand’s translational and rotational degrees of freedom are converted into vibrational degrees of
freedom, causing a significant loss in entropy (−T∆SL-assoc > 0). Additionally, torsional degrees
of freedom are restricted and consequently contribute to the entropic loss. Can this factor be
responsible for cooperativity?
In ref 2 we have pointed out that such factor is not responsible for the entropic negative
cooperativity. Using Born-Haber analysis (Figures 2 and 8), we can reach the same conclusion.
For instance, the number of the torsional degrees of freedom being restricted upon ligand binding
is the same in ligands 21 and 20. On the other hand, ligand 10, relative to ligand 9, has an
additional rotatable bond that gets restricted upon binding (Figure 8). -T∆SLMe-assoc, in case of the
1
0 and 9 ligand pair, is therefore significantly more positive (unfavorable) than -T∆SLH-assoc; and
consequently, the differential entropy (−T∆SLMe-assoc – (−T∆SLH-assoc)) is shifted to the positive
-
-
when the COO is absent. A positive shift in (-T∆S
– (−T∆SLH-assoc)) when the COO is
LMe-assoc
-
absent is by default a negative shift in this quantity when the COO is present, and this, in turn, is
translated into a negative shift in −T∆∆S(H→Me) (SI, section 1.3). This proposed negative shift in
−T∆∆S(H→Me) was not observed experimentally [2]; consequently, the entropic signature of the
ligand-protein association is not the factor responsible for the entropic negative cooperativity.
With regard to the free energy, however, the proposed negative shift in −T∆∆S(H→Me) can and
indeed did display itself as a negative shift in ∆∆G(H→Me) (i.e. a positive cooperativity; Note:
∆∆G(H→Me) = ∆∆H(H→Me) + (−T∆∆S(H→Me)), supplementary materials, Eq. S2). We only do not
know to what extent this factor participates in the free energy cooperativity, but we can make a
reasonable suggestion that aligns with what was learned about the binding thermodynamics in
this ligand-protein system. Based on our findings in ref 16, it can be hypothesized that 27% of
the enthalpic cooperativity that is originated from variability in the solvation-resolvation patterns
14

ACCEPTED MANUSCRIPT
of the ligand and/or the ligand-protein complex is demonstrated as free energy cooperativity.
This portion would be equivalent to 27% × 7.2 kJ/mol = 1.94 kJ/mol, or about 60% of the
cooperativity observed in the isothermal titration calorimetry (ITC) data [16] (Note: the
biochemical assay data presented in the current study and the ITC free energy data are not
identical but are correlated well, supplementary materials, section 2). What about the other 40%?
Most likely, this 40% comes from the restriction of the additional rotatable bond in ligand 10 and
the entropic consequence of this bond restriction.
2
1 2
.5. The influence of size reduction of the R side chain on the relationship between R and
-
the COO
1
In order to investigate the influence of reducing the size of the R side chain (from i-Bu to Me)
-
on the relationship between the R side chain and the COO group, the differential binding free
2
energies of each H→R modification in series V was compared with that of the same H→R
2
2
modification in series VI. This is illustrated by the double mutant cycle in Figure 9. In this
double mutant cycle, ligand 31 is mutated to 32, 33, 1, or 34; and the differential binding energy
(
∆∆G(H,H→R2,H)) is obtained when the binding free energy of 31 is subtracted from that of any of
ligands 32, 33, 1, and 34. Similarly, the mutation of ligand 35 to 36, 37, 5, or 38 yields values for
∆∆G(H,COO→R2,COO)) via subtracting the free energy of 35 from that of any of ligands 36, 37, 5,
(
and 38. The values of these differential free energies are given in Table 5. This table also
includes the values of (∆∆G(H,COO→R2,COO) − ∆∆G(H,H→R2,H)).
<
Insert Figure 9>
2
The data presented in Table 5 reveal that, in most cases, the cooperativity between the R side
-
chain and the COO is not significantly influenced by the reduction in the size of the R side
1
chain. For example, when R = Me, the modification of the R side chain from H to Me yields a -
1
2
5
.2 kJ/mol positive cooperativity, which is almost the same amount produced when R
1
= i-Bu
(
Table 4). The same trend is observed when R is modified from H to Et or i-Bu (within
2
experimental error).
<Insert Table 5>
When the H is replaced by Bn, however, a reduction in the positive cooperativity is observed
when the R side chain is Me (i.e. -0.9 kJ/mol vs. -2.7 kJ/mol). It could therefore be concluded
1
15

ACCEPTED MANUSCRIPT
that a large R
1
hydrophobic side chain might augment the positive cooperativity between the R
2
-
side chain and the COO , but only when the R side chain is modified into a bulky (and probably
2
aromatic) group. Because only one case of such influence is presented herein, and the difference
in the cooperativity indicator (i.e. ∆∆G(H,COO→R2,COO) - ∆∆G(H,H→R2,H)) is not very pronounced
(only 1.8 kJ/mol), this deviation from the main conclusion drawn from these series (i.e. that R1
-
does not influence the cooperativity between the R
2
and the COO ) should be viewed with
caution. More data that can test this hypothesis further might therefore be needed. Figure 10
graphically depicts the cooperative behavior in series V and VI (R = Me), in relation to the same
1
R
2
modifications done in series III and IV (R
Insert Figure 10>
.6. The relationship between the hydrophobic side chains R and R2
1
= i-Bu).
<
2
1
1 2
The relationship between the hydrophobic side chains R and R were investigated using series
V and some of series III ligands (series III-R). The double mutant cycle in Figure 11 reveals how
these series were used in this investigation. Specifically, the H→R replacements (H→Me/Et/i-
2
Bu/Bn) were carried out while R = Me (small, less hydrophobic side chain), and while R = i-Bu
1
1
(
large, more hydrophobic side chain). The differential free energy values for the H→R2
replacement while R = Me were then compared with the differential free energies of these
replacement while R = i-Bu (i.e. ∆∆G(H,Me→R2,Me) vs. ∆∆G(H,i-Bu→R2,i-Bu)), and the values of
1
2
∆∆G(H,i-Bu→R2,i-Bu) − ∆∆G(H,Me→R2,Me) were calculated and listed in Table 6. These data
demonstrate uncoupling between the R and the R side chain (i.e. they behave in an additive
1
2
manner). For example, there is no significant difference between any ∆∆G(H,Me→R2,Me) and its
corresponding ∆∆G(H,i-Bu→R2,i-Bu). In other words, the term ∆∆G(H,i-Bu→R2,i-Bu) - ∆∆G(H,Me→R2,Me) is
always close to zero.
<Insert Figure 11>
<Insert Table 6>
3
. Conclusions
The study presented herein reveals that the relationships among the ligand functional
groups/side chains can be complex. For example, in some instances, mild positive cooperativity
16

ACCEPTED MANUSCRIPT
1
is demonstrated. This is shown in the relationship between the R side chains and the terminal
-
COO group of the TLN phosphonamidate inhibitors. In other instances, strong positive
cooperativity is at work. This is clearly demonstrated in the relationship between most of the R2
-
side chains and the terminal COO . In some of the more branched R
2
side chains, however,
additivity or even negative cooperativity exists. The cooperative/additive behavior of this
-
particular ligand group pair (R
2
, COO ) did not show sufficient evidence for being influenced by
the change in the R side chain (Me vs. i-Bu). Additionally, investigating the relationship
1
between the side chains R and R demonstrates that these two side chains are independent, and
1
2
they show additive behavior. It follows that the first conclusion which can be drawn from this
study is that ligand functional groups can be additive, synergistic, or antagonistic. Given that the
cooperativity indicator described and utilized in this study is more likely to deviate from zero,
cooperativity, including both synergism and antagonism, is anticipated to be the more common
experimental finding, and this is what was observed in this study. One, however, should not
assume that one type of behavior will be always observed as is commonly assumed with
traditional scoring functions that completely ignore cooperativity. Also, the additivity principle
should not be completely dismissed; rather, we should attempt to determine the molecular basis
of each type of behavior, and in turn predict how the system would behave in each setting.
The second conclusion that can be drawn from this study is that the cooperative/additive
behavior of a ligand functional group pair may be correlated with the properties of the involved
groups. For example, a polar group and a hydrophobic side chain are likely to be synergistic (e.g.
-
the COO and the R
1
or R
2
side chain). The magnitude of this synergism is however influenced
by the distance between the group pair, and by the size or the degree of branching of the
hydrophobic moiety (e.g., a more branched R causes the synergism to be diminished). For
2
example, strong synergism, which amounts to 5.0 kJ/mol, can be observed when the two groups
-
are adjacent (the COO and the R side chain, which are separated by 2 single bonds), while
2
synergism accounts only for about 2.0 kJ/mol when the two groups are moderately distant (the
-
COO and the R , which are separated by 5 single, rotatable bonds). Synergism diminishes when
1
2
the side chain is branched (e.g., R side chain), most likely due to a reduction in the stability of
organized water arrangements that hydrates the hydrophobic side chain in either the unbound, the
complexed state, or both [16]. On the other hand, hydrophobic-hydrophobic side chains might
17

ACCEPTED MANUSCRIPT
show additivity when they are moderately distant (e.g. R
separated by 5 single bonds).
1 2
and R side chains, which are
It is important to note that the correlations made in the current study between the
additivity/cooperativity patterns and the structural/physicochemical characteristics of the ligand
functional groups represent a starting point for proposing SAR and QSAR models that
incorporate such patterns. The end result will be a substantial improvement in our ability to
predict the outcomes of lead optimizations with regard to the binding affinity. For example, a
preliminary algorithm, which provides guidance as to whether medicinal chemists would
encounter cooperativity or additivity in the course of simultaneous incorporation of two or more
functionalities in a ligand molecule, can be constructed based upon this study (Scheme 4).
Additional studies involving other ligand-protein model systems could result in further
refinements to this algorithm. An improvement in our ability to predict cooperativity vs.
additivity is also much needed in order to develop novel, more accurate scoring functions. A
notable recent advancement in this field was the development of ScorpionScore, an empirical
scoring function that goes beyond the additive treatment of non-covalent ligand-protein
interactions (i.e. it incorporates the cooperativity principle) [27]. The current and subsequent
related studies, which will be reported in due course, are likely to be very contributive to such
endeavors.
<Insert Scheme 4>
4
. Experimental Section
4
1
.1. Biochemical assay: The inhibition constants of the thermolysin phosphonamidate inhibitors
-38 were determined photometrically at 345 nm using 2-furanacryloyl-Gly-Leu-NH as a
2
substrate [25]. The assay was carried out on a Cary 100/300 UV/VIS spectrophotometer at 25.0
0.2 ºC. A 0.05 M Tris buffer containing 0.02 M CaCl , 2.5 M NaBr [10], and 1.25% DMF, was
±
2
used in all measurements. Buffer pH was adjusted to 7.3 ± 0.05 at room temperature prior to use.
The concentration of the enzyme stock solution was determined by UV absorbance at 280nm
-1
(
ε1% =17.65 cm ) [28]. The concentrations of the stock solutions of the substrate and the
inhibitors were determined from accurately weighed samples. The enzyme concentration in all
the final assay solutions was approximately 8 nM, the substrate concentration in the final assay
18

ACCEPTED MANUSCRIPT
i i
solutions was 0.8 M, and the inhibitors’ concentrations were in the range of 0.5K – 10K for
each inhibitor. The inhibition constant (K ) for the inhibitor was taken to be the average of at
i
least three K determinations, each of which was calculated from the experimentally determined
i
IC50 using Cheng-Prusoff equation [29] (K
m
= 3.9± 0.6 mM). The IC50 values were determined
from υ /υ vs. [I] plots [10, 30] for all of the inhibitors with inhibition constants above 30 nM
0
i
(υ
0
i
/υ = [I]/IC50 + 1), or Henderson plots [31] for inhibitors with inhibition constants below this.
At least six different inhibitor concentrations [I] were used to construct each plot.
4
4
.2. Chemistry
.2.1. General methods: Reagents were obtained from commercial suppliers and used without
further purification. Anhydrous solvents were purchased as sealed bottles from Aldrich and were
maintained under an argon atmosphere. Tetrahydrofuran (THF) was distilled from a
sodium/benzophenone still and used immediately. Dichloromethane (DCM) was distilled from a
calcium hydride still and used immediately. Solvent removal was performed on a rotary
evaporator equipped with a 20-60° C water bath and a self-contained aspirator. Thin-layer
chromatography (TLC) was performed on Analtech (Newark, DE) 200 micron Silica Gel F
coated on polyethylene sheets. Visualization was accomplished with 254 nm UV light or iodine
staining. The silica gel used in the flash chromatography was 40-75 ꢀm flash grade purchased
from Sorbent Technologies (Atlanta, GA). All amino acids used are L unless otherwise noted.
Proton, phosphorus and carbon nuclear magnetic resonance was performed in deuterated solvents
purchased from Cambridge Isotope Laboratories, Inc (Andover, MD) on one of the following
1
instruments: Varian Gemini 300 MHz, Varian Inova 400 MHz, or Varian Inova 500 MHz. H
NMR data is reported in the following format: chemical shift (ppm values in relation to TMS or
appropriate solvent peak), multiplicity (s = singlet, d = doublet, t = triplet, q = quartet, dd =
doublet of doublet, dt = doublet of triplet, dq = doublet of quartet, m = multiplet, brs = broad
singlet), coupling constant(s), and integration. Whenever fractions of chemically equivalent
protons appear at widely-spaced chemical shifts, like when the compound exists in multiple
conformations, the chemical shifts are reported; followed by the multiplicity(ies) preceded by the
number of peaks (e.g. 2 s, 2 d, etc.), the coupling constant(s), and the sum of the integrations of
these peaks. Low resolution ESI mass spectrometry was performed on a Thermo Finnigan LCQ
Advantage instrument using 60% methanol in water with 1% acetic acid or 60% acetonitrile in
19

ACCEPTED MANUSCRIPT
water with 0.1% trifluoroacetic acid as the mobile phase. Preparative and semi-preparative
HPLC instrumentation included a Milton Roy gm4000 gradient programmer, Milton Roy
Constametric I and III pumps, a Rheodyne 7125 injector with a 5.00 mL sample loop, and a
Knauer Variable Wavelength Detector set at either 218 nm or 254 nm with a preparative flow
cell. The HPLC column used was a Phenomenex LUNA C18(2), 5 ꢀm, 100A pore, 21 mm X
2
50 mm with Security Guard cartridge used with a flow rate of 8 mL/min. All final compounds
were at least 95% pure by HPLC analysis. The HPLC analysis of the final compounds involved
the use of acetonitrile/water as a mobile phase in a gradient elution method (10→90%
acetonitrile over 14 min). Given below are the detailed synthesis and the characterization of the
final compounds and their intermediates that were not previously reported. The detailed
synthesis and characterization of the other compounds are previously reported [16].
4
.2.2. Synthesis of benzyl N-(hydroxymethyl)-carbamate (1)
Benzyl carbamate (6.0 g, 40 mmol) was added to a solution of 37% formalin (4.4 g, 56 mmol)
and sodium carbonate (2.2 g, 20 mmol) in 65 mL water. The mixture was heated until all the
solids were dissolved, then cooled to room temperature and stirred overnight. The precipitated
solid was then filtered, dried, and redissolved in dichloromethane. The solution was dried using
anhydrous magnesium sulfate and the solvent was removed under vacuum to give the product as
1
a white solid which was used in the next step without further purification (5.4 g, 74%) H NMR
(
CDCl ) δ 4.10 (s, 1H), 4.71 (d, J = 6.5 Hz, 2H), 5.13 (s, 2H), 6.07 (s, 1H), 7.36 (s, 5H); m/z
3
+
(
LCMS, ESI): calc. for [M + H] 182.1; found 182.2
4
.2.3. Synthesis of benzyl N-(acetoxymethyl)-carbamate (2)
Compound 1 (3.6 g, 20 mmol) was dissolved in 25 mL anhydrous THF and was added slowly
to an ice-cooled stirred solution of 23 mL acetic anhydride and 6.5 mL anhydrous pyridine under
Argon. The mixture was stirred at r.t. for 2 h, then the solution was diluted with 150 mL ethyl
acetate and washed with 1 M HCl (3X 150 mL) and brine (2X 150 mL). The organic layer was
dried with anhydrous magnesium sulfate and the volatile materials were removed under vacuum
to give an oily residue which was purified with flash chromatography (3.0 g, 67%) of the pure
1
product. H NMR (CDCl ) δ 2.08 (s, 3H), 5.16 (s, 2H), 5.23 (d, J = 7.5 Hz, 2H), 6.08 (s, 1H),
3
+
7
.38 (s, 5H); m/z (LCMS, ESI): calc. for [M + Na] 246.1; found 246.0
20

ACCEPTED MANUSCRIPT
4
.2.4. Synthesis of dimethyl N-(benzyloxycarbonyl)-aminomethylphosphonate (3)
A mixture of compound 2 (2.9 g, 13 mmol) and trimethylphosphite (4.6 mL, 39 mmol) was
o
refluxed for 3 h. The volatile materials were removed by distillation at 60 C under reduced
pressure to give the product as an oily residue which was used in the next step without further
1
purification (3.4 g, 97%) H NMR (CDCl ) δ 3.62 (dd, J = 6.5 Hz, J = 11.0 Hz, 2H), 3.72 (d,
3
H-P
3
1
J
3
H-P = 11.0 Hz, 6H), 5.09 (s, 2H), 5.83 (s, 1H), 7.25-7.40 (m, 5H), P NMR (CDCl ) δ 25.32;
+
m/z (LCMS, ESI): calc. for [M + Na] 296.1; found 296.1
4
.2.5. Synthesis of methyl N-(benzyloxycarbonyl)-aminomethylphosphonate (4)
Compound 3 (3.3 g, 12 mmol) was shook vigorously with 10% NaOH (14.5 mL, 3 equiv.)
until it was completely dissolved. The mixture was stirred at room temperature for 2 h then
diluted with water, extracted with ethyl acetate (2X 30 mL), and acidified to pH 1 with 2 M HCl.
The aqueous solution was extracted with dichloromethane (2X 100 mL) and ethyl acetate (2X 50
mL). The dichloromethane layers were combined, washed with brine (2X 50 mL), and dried
using anhydrous magnesium sulfate. The ethyl acetate layers were also combined, washed with
brine (2X 25 mL), and dried with anhydrous magnesium sulfate. The two organic layers were
then combined and the volatile solvents were removed under high vacuum to give the product as
1
a pure white solid (2.3 g, 73%). H NMR (CDCl
3
) δ 3.64 (d, JH-P = 11.0 Hz, 2H), 3.71 (d, JH-P
=
3
1
1
2
1.0 Hz, 3H), 5.12 (s, 2H), 5.7 (brs, 1H), 7.28-7.42 (m, 5H), 11.8 (brs, 1H), P NMR (CDCl ) δ
3
+
4.12; m/z (LCMS, ESI): calc. for [M + Na] 282.1; found 282.1
4
.2.6. General procedure for amide coupling: To a cooled solution of Boc-
L-leucine (1.0
equiv.), the amine/α-aminoester HCl (1.2-1.5 equiv.), and PyBop 1.2 equiv. (or EDCI.HCl 1.2
equiv. and HOBt 1.2 equiv.) in anhydrous DMF was added diisopropylethylamine (3.3-4.0
equiv.) gradually. The reaction mixture was stirred at room temperature for 5 h to overnight,
diluted with ethyl acetate (50 mL for every 5 mL DMF), then extracted with 1 M HCl (3X),
saturated sodium bicarbonate (3X), and brine (2X). The organic layer was then dried with
anhydrous sodium sulfate and the solvent was evaporated under vacuum to give the products
which were purified by flash chromatography whenever needed.
21

ACCEPTED MANUSCRIPT
4
.2.6.1. (S)-2-(tert-butoxycarbonylamino)-N-isopentylpropanamide (IA1)
Following the general procedure for amide coupling, Boc-L-alanine (473 mg, 2.5 mmol) was
reacted with isoamylamine (262 mg, 3.0 mmol) in anhydrous DMF (10 mL), using PyBop as a
coupling reagent (1.56 g, 3.0 mmol) and diisopropylethylamine (1.07 g, 8.25 mmol) as a base.
1
4
19 mg of compound IA1 was obtained after purification with flash chromatography (65%). H
6
NMR (DMSO-d ) δ 0.85 (d, J = 6.4 Hz, 6H), 1.14 (d, J = 6.8 Hz, 3H), 1.29 (m, 2H), 1.37 (s, 9H),
1
.55 (m, 1H), 3.06 (m, 2H), 3.90 (m, 1H), 6.81 (d, J = 8.5 Hz, 1H), 7.66 (t, 1H); m/z (LCMS,
+
ESI): calc. for [M + H] 259.2; found 259.1
.2.6.2. (S)-2-(tert-butoxycarbonylamino)-N-isopentylbutanamide (IA2)
Following the general procedure for amide coupling, Boc- -α-aminobutyric acid (508 mg, 2.5
4
L
mmol) was reacted with isoamylamine (262 mg, 3.0 mmol) in anhydrous DMF (10 mL), using
PyBop as a coupling reagent (1.56 g, 3.0 mmol) and diisopropylethylamine as a base (1.07 g,
8
.25 mmol). 483 mg of compound IA2 was obtained after purification with flash
1
chromatography (71%). H NMR (DMSO-d ) 0.82 (t, J = 7.0 Hz, 3H), δ 0.85 (d, J = 6.5 Hz, 6H),
6
1
6
.28 (q, J = 7.5 Hz, 2H), 1.38 (s, 9H), 1.47 (m, 2H), 1.55 (m, 1H), 3.06 (m, 2H), 3.78 (m, 1H),
+
.72 (d, J = 8.0 Hz, 1H), 7.73 (t, J = 6.3 Hz, 1H); m/z (LCMS, ESI): calc. for [M + H] 273.2;
found 273.2
.2.6.3. (S)-2-(tert-butoxycarbonylamino)-N-isopentylpentanamide (IA3)
Following the general procedure for amide coupling, Boc- -norvaline (543 mg, 2.5 mmol) was
4
L
reacted with isoamylamine (262 mg, 3.0 mmol) in anhydrous DMF (10 mL) using PyBop as a
coupling reagent (1.56 g, 3.0 mmol) and diisopropylethylamine (1.07 g, 8.25 mmol). 493 mg of
1
compound IA3 was obtained after purification with flash chromatography (69%). H NMR
(
DMSO-d ) δ 0.79 (m, 9H), 1.29 (m, 4H), 1.30 (s, 9H), 1.44 (m, 3H), 2.99 (m, 2H), 3.77 (m,
6
+
1
3
H), 6.65 (d, J = 6.4 Hz, 1H), 7.64 (t, 1H); m/z (LCMS, ESI): calc. for [M + Na] 309.2; found
09.1
4
.2.6.4. ((S)-2-(tert-butoxycarbonylamino)propanoyl)-L-leucine methyl ester (IA5)
Following the general procedure for amide coupling, Boc-
L-alanine (473 mg, 2.5 mmol) was
reacted with -leucine methyl ester hydrochloride (545 mg, 3.0 mmol) in anhydrous DMF (10
L
22

ACCEPTED MANUSCRIPT
mL), using PyBop as a coupling reagent (1.56 g, 3.0 mmol) and diisopropylethylamine (1.07 g,
8
.25 mmol) as a base. 569 mg of compound IA5 was obtained after purification with flash
1
chromatography (72%). H NMR (DMSO-d ) δ 0.84 (d, J = 6.5 Hz, 3H), 0.90 (d, J = 6.5 Hz,
6
3
H), 1.17 (d, J = 7.0 Hz, 3H), 1.38 (s, 9H), 1.45-1.70 (m, 3H), 3.62 (s, 3H), 3.98 (m, 1H), 4.30
(m, 1H), 6.88 (d, J = 8.5 Hz, 1H), 8.07 (d, J = 7.5 Hz, 1H); m/z (LCMS, ESI): calc. for [M +
+
Na] 339.2; found 339.1
4
.2.6.5. ((S)-2-(tert-butoxycarbonylamino)butanoyl)-L-leucine methyl ester (IA6)
Following the general procedure for amide coupling, Boc-
mmol) was reacted with -leucine methyl ester hydrochloride (545 mg, 3.0 mmol) in anhydrous
DMF (10 mL), using PyBop as a coupling reagent (1.56 g, 3.0 mmol) and diisopropylethylamine
L-α-aminobutyric acid (508 mg, 2.5
L
(1.07 g, 8.25 mmol) as a base. 644 mg of compound IA6 was obtained after purification with
1
flash chromatography (78%). H NMR (DMSO-d ) δ 0.85 (m, 6H), 0.9 (d, J = 6.5 Hz, 3H), 1.38
6
(s, 9H), 1.45-1.70 (m, 5H), 3.62 (s, 3H), 3.87 (m, 1H), 4.31 (m, 1H), 6.76 (d, J = 8.5 Hz, 1H),
+
8
.22 (d, J = 7.5 Hz, 1H); m/z (LCMS, ESI): calc. for [M + H] 331.2; found 331.2
4
.2.6.6. ((S)-2-(tert-butoxycarbonylamino)pentanoyl)-L-leucine methyl ester (IA7)
Following the general procedure for amide coupling, Boc-
reacted with -leucine methyl ester hydrochloride (545 mg, 3.0 mmol) in anhydrous DMF (10
mL), using PyBop as a coupling reagent (1.56 g, 3.0 mmol) and diisopropylethylamine (1.07 g,
L-norvaline (543 mg, 2.5 mmol) was
L
8
.25 mmol) as a base. 645 mg of compound IA7 was obtained after purification with flash
1
chromatography (75%). H NMR (DMSO-d
6
) δ 0.83 (m, 9H), 1.29 (m, 2H), 1.35 (s, 9H), 1.38-
1
.68 (m, 5H), 3.59 (s, 3H), 3.90 (m, 1H), 4.17 (m, 1H), 6.75 (d, J = 8.4 Hz, 1H), 8.07 (d, J = 8.0
+
Hz, 1H); m/z (LCMS, ESI): calc. for [M + Na] 367.2; found 367.1
.2.6.7. (S)-2-(tert-butoxycarbonylamino)-N-ethylpropanamide (IA32)
Following the general procedure for amide coupling, Boc- -alanine (473 mg, 2.5 mmol) was
4
L
reacted with ethylamine hydrochloride (306 mg, 3.7 mmol) in anhydrous DMF (12 mL), using
EDCI.HCl (573 mg, 3.0 mmol) and HOBt (405 mg, 3.0 mmol) as coupling reagents and
diisopropylethylamine (1.29 g, 10.0 mmol) as a base. 437 mg of compound IA32 was obtained
1
and used without further purification (67%). H NMR (DMSO-d ) δ 1.00 (t, J = 7.5 Hz, 3H),
6
23

ACCEPTED MANUSCRIPT
1
.15 (d, J = 6.5 Hz, 3H), 1.38 (s, 9H), 3.05 (m, 2H), 3.90 (m, 1H), 6.82 (d, J = 7.5 Hz, 1H), 7.74
+
(
t, 1H); m/z (LCMS, ESI): calc. for [M + Na] 239.2; found 239.0
4
.2.6.8. (S)-2-(tert-butoxycarbonylamino)-N-propylpropanamide (IA33)
Following the general procedure for amide coupling, Boc- -alanine (473 mg, 2.5 mmol) was
L
reacted with propylamine hydrochloride (287 mg, 3.0 mmol) in anhydrous DMF (10 mL), using
EDCI.HCl (573 mg, 3.0 mmol) and HOBt (405 mg, 3.0 mmol) as coupling reagents and
diisopropylethylamine (1.07 g, 8.25 mmol) as a base. 403 mg of compound IA33 was obtained
1
and used without further purification (70%). H NMR (DMSO-d
6
) δ 0.83 (t, 3H), 1.15 (d, J = 7.5
Hz, 3H), 1.38 (s, 9H), 1.40 (m, 2H), 3.00 (m, 2H), 3.92 (m, 1H), 6.83 (d, J = 8.5 Hz, 1H), 7.71 (t,
+
1
H); m/z (LCMS, ESI): calc. for [M + Na] 253.2; found 253.1
4
.2.6.9. (S)-2-(tert-butoxycarbonylamino)-N-(2-phenylethyl)-propanamide (IA34)
Following the general procedure for amide coupling, Boc-L-alanine (473 mg, 2.5 mmol) was
reacted with 2-phenylethylamine hydrochloride (473 mg, 3.0 mmol) in anhydrous DMF (10 mL),
using EDCI.HCl (573 mg, 3.0 mmol) and HOBt (405 mg, 3.0 mmol) as coupling reagents and
diisopropylethylamine (1.07 g, 8.25 mmol) as a base. 504 mg of compound IA34 was obtained
1
after purification with flash chromatography (69%). H NMR (DMSO-d ) δ 1.12 (d, J = 7.0 Hz,
6
3
7
3
H), 1.38 (s, 9H), 2.51 (t, J = 7.5 Hz, 2H), 3.34 (m, 2H), 3.72 (m, 1H), 6.83 (d, J = 8.5 Hz, 1H),
+
.16-7.32 (m, 5H), 7.83 (t, J = 5.0 Hz, 1H); m/z (LCMS, ESI): calc. for [M + Na] 315.2; found
15.1
4
.2.6.10. (N-(tert-butoxycarbonyl)-
L
-alaninyl)-glycine ethyl ester (IA35)
-alanine (473 mg, 2.5 mmol) was
Following the general procedure for amide coupling, Boc-
L
reacted with glycine ethyl ester hydrochloride (419 mg, 3.0 mmol) in anhydrous DMF (10 mL),
using EDCI.HCl (573 mg, 3.0 mmol) and HOBt (405 mg, 3.0 mmol) as coupling reagents and
diisopropylethylamine (1.07 g, 8.25 mmol) as a base. 501 mg of compound IA35 was obtained
1
after purification with flash chromatography (73%). H NMR (DMSO-d
6
) δ 1.18 (m, 6H), 1.38
(s, 9H), 3.83 (m, 2H), 4.00 (m, 1H), 4.08 (q, J = 7.0 Hz, 2H), 6.94 (d, J = 7.5 Hz, 1H), 8.18 (t,
+
1
H); m/z (LCMS, ESI): calc. for [M + Na] 297.2; found 297.1
24

ACCEPTED MANUSCRIPT
4
.2.6.11. (N-(tert-butoxycarbonyl)-
L
-alaninyl)-
L
-alanine methyl ester (IA36)
-alanine (473 mg, 2.5 mmol) was
Following the general procedure for amide coupling, Boc-
L
reacted with alanine methyl ester hydrochloride (419 mg, 3.0 mmol) in anhydrous DMF (10
mL), using EDCI.HCl (573 mg, 3.0 mmol) and HOBt (405 mg, 3.0 mmol) as coupling reagents
and diisopropylethylamine (1.07 g, 8.25 mmol) as a base. 480 mg of compound IA36 was
1
obtained after purification with flash chromatography (70%). H NMR (DMSO-d
6
) δ 1.17 (d,
3
8
H), 1.28 (d, J = 7.5 Hz, 3H), 1.38 (s, 9H), 3.62 (s, 3H), 3.98 (m, 1H), 4.27 (m, 1H), 6.88 (d, J =
+
.0 Hz, 1H), 8.18 (d, J = 7.0 Hz, 1H); m/z (LCMS, ESI): calc. for [M + Na] 297.2; found 297.1
4
.2.6.12. Methyl (S)-N-(N-(tert-butoxycarbonyl)- -alaninyl)-2-aminobutanoate (IA37)
L
Following the general procedure for amide coupling, Boc-
L-alanine (473 mg, 2.5 mmol) was
reacted with methyl -α-aminobutyrate hydrochloride (461 mg, 3.0 mmol: synthesized in-house)
L
in anhydrous DMF (10 mL), using EDCI.HCl (573 mg, 3.0 mmol) and HOBt (405 mg, 3.0
mmol) as coupling reagents and diisopropylethylamine (1.07 g, 8.25 mmol) as a base. 547 mg of
compound IA37 was obtained after purification with flash chromatography (76%); m/z (LCMS,
+
ESI): calc. for [M + Na] 311.2; found 311.1
4
.2.6.13. (N-(tert-butoxycarbonyl)-L-alaninyl)-L-phenylalanine methyl ester (IA38)
Following the general procedure for amide coupling, Boc-
reacted with phenylalanine methyl ester hydrochloride (647 mg, 3.0 mmol) in anhydrous DMF
10 mL), using EDCI.HCl (573 mg, 3.0 mmol) and HOBt (405 mg, 3.0 mmol) as coupling
L-alanine (473 mg, 2.5 mmol) was
(
reagents and diisopropylethylamine (1.07 g, 8.25 mmol) as a base. 683 mg of compound IA38
1
was obtained after purification with flash chromatography (78%). H NMR (DMSO-d
6
) δ 1.12
(d, J = 7.5 Hz, 3H), 1.37 (s, 9H), 2.98 (m, 2H), 3.58 (s, 3H), 3.97 (m, 1H), 4.47 (m, 1H), 6.87 (d,
J = 8.5 Hz, 1H), 7.17-7.29 (m, 5H), 8.23 (d, J = 7.5 Hz, 1H); m/z (LCMS, ESI): calc. for [M +
+
Na] 373.2; found 373.1
4
.2.7. General procedure for Boc deprotection: The Boc-protected compound was dissolved
either in 3 M HCl/MeOH or in ethyl acetate. When the compound is dissolved into ethyl acetate,
hydrogen chloride gas generated from the reaction of sulfuric acid and sodium chloride was
bubbled into the solution at 0 °C. The solution was then stirred for 1.5-3 h at room temperature
when HCl/MeOH solution is used or at 0 °C when HCl is bubbled into the solution. The volatile
25

ACCEPTED MANUSCRIPT
materials were then removed under vacuum to give the product as hygroscopic solid which was
purified with reverse phase HPLC.
4
.2.7.1. (S)-2-amino-N-isopentylpropanamide hydrochloride (IB1)
Following the general procedure for Boc deprotection, compound IA1 (387 mg, 1.5 mmol)
was exposed to 3 M HCl/MeOH (3.0 mL) for 3 h. 276 mg of compound IB1 was obtained after
1
purification with reverse phase HPLC (95%). H NMR (DMSO-d ) δ 0.84 amd 0.85 (2 x d, J =
6
6
.5 Hz, 6H), 1.30 (q, J = 7.0 Hz, 2H), 1.34 (d, J = 7.0 Hz, 3H), 1.57 (m, 1H), 3.07 (m, 1H), 3.12
(m, 1H), 3.81 (m, 1H), 8.35 (brs, 3H), 8.69 (t, J = 5.5 Hz, 1H); m/z (LCMS, ESI): calc. for [M +
+
H] 159.2; found 159.1
4
.2.7.2. (S)-2-amino-N-isopentylbutanamide hydrochloride (IB2)
Following the general procedure for Boc deprotection, compound IA2 (408 mg, 1.5 mmol)
was exposed to 3 M HCl/MeOH (3.0 mL) for 3 h. 302 mg of compound IB2 was obtained after
1
purification with reverse phase HPLC (97%). H NMR (DMSO-d ) δ 0.86 (m, 9H), 1.30 (m,
6
2
H), 1.56 (m, 1H), 1.72 (m, 2H), 3.05 (m, 1H), 3.18 (m, 1H), 3.65 (t, 1H), 8.20 (brs, 3H), 8.55 (t,
+
J = 5.5 Hz, 1H); m/z (LCMS, ESI): calc. [M + H] 173.2; found 173.1
4
.2.7.3. (S)-2-amino-N-isopentylpentanamide hydrochloride (IB3)
Following the general procedure for Boc deprotection, compound IA3 (429 mg, 1.5 mmol)
was exposed to 3 M HCl/MeOH (3.0 mL) for 3 h. 315 mg of compound IB3 was obtained after
1
purification with reverse phase HPLC (95%). H NMR (DMSO-d ) δ 0.85 (m, 9H), 1.27 (m,
6
4
1
H), 1.62 (m, 3H), 3.05 (m, 1H), 3.17 (m, 1H), 3.68 (t, 1H), 8.25 (brs, 3H), 8.56 (t, J = 5.5 Hz,
+
H); m/z (LCMS, ESI): calc. for [M + Na] 209.2; found 209.1
4
.2.7.4. ((S)-2-aminopropanoyl)-L-leucine methyl ester hydrochloride (IB5)
Following the general procedure for Boc deprotection, compound IA5 (474 mg, 1.5 mmol)
was exposed to 3 M HCl/MeOH (3.0 mL) for 3 h. 370 mg of compound IB5 was obtained after
1
purification with reverse phase HPLC (98%). H NMR (CD OD) δ 0.93 (d, J = 6.4 Hz, 3H), 0.96
3
(d, J = 6.4 Hz, 3H), 1.52 (d, J = 6.8 Hz, 3H), 1.67 (m, 3H), 3.71 (s, 3H), 3.76 (m, 1H), 4.49 (m,
+
1
H); m/z (LCMS, ESI): calc. for [M + H] 217.2; found 217.1
26

ACCEPTED MANUSCRIPT
4
.2.7.5. ((S)-2-aminobutanoyl)-L-leucine methyl ester hydrochloride (IB6)
Following the general procedure for Boc deprotection, compound IA6 (495 mg, 1.5 mmol)
was exposed to 3 M HCl/MeOH (3.0 mL) for 3 h. 383 mg of compound IB6 was obtained after
1
purification with reverse phase HPLC (96%). H NMR (DMSO-d ) δ 0.88 (m, 9H), 1.46-1.86
6
(m, 5H), 3.61 (s, 3H), 3.76 (t, 1H), 4.30 (m, 1H), 8.26 (brs, 3H), 8.86 (d, J = 7.2 Hz, 1H); m/z
+
(
LCMS, ESI): calc. for [M + H] 231.2; found 231.2
4
.2.7.6. ((S)-2-aminopentanoyl))-L-leucine methyl ester hydrochloride (IB7)
Following the general procedure for Boc deprotection, compound IA7 (516 mg, 1.5 mmol)
was exposed to 3 M HCl/MeOH (3.0 mL) for 3 h. 407 mg of compound IB7 was obtained after
1
purification with reverse phase HPLC (97%). H NMR (DMSO-d ) δ 0.87 (m, 9H), 1.35 (m,
6
2
1
H), 1.50 (m, 5H), 3.61 (s, 3H), 3.78 (m, 1H), 4.29 (m, 1H), 8.23 (brs, 3H), 8.83 (d, J = 7.2 Hz,
+
H); m/z (LCMS, ESI): calc. for [M + H] 245.2; found 245.1
4
.2.7.7. (S)-2-amino-N-ethylpropanamide hydrochloride (IB32)
Following the general procedure for Boc deprotection, compound IA32 (324 mg, 1.5 mmol)
was exposed to 3 M HCl/MeOH (3.0 mL) for 3 h. 213 mg of compound IB32 was obtained after
1
purification with reverse phase HPLC (94%). H NMR (DMSO-d
6
) δ 1.03 (t, J = 7.4 Hz, 3H),
1
.32 (d, J = 6.8 Hz, 3H), 3.10 (m, 2H), 3.75 (q, 1H), 8.23 (brs, 3H), 8.52 (t, J = 5.5 Hz, 1H); m/z
+
(
LCMS, ESI): calc. for [2M + H] 233.2; found 233.1
4
.2.7.8. (S)-2-amino-N-propylpropanamide hydrochloride (IB33)
Following the general procedure for Boc deprotection, compound IA33 (345 mg, 1.5 mmol)
was exposed to 3 M HCl/MeOH (3.0 mL) for 3 h. 239 mg of compound IB33 was obtained after
1
purification with reverse phase HPLC (96%). H NMR (DMSO-d
6
) δ 0.84 (t, J = 7.4 Hz, 3H),
1
8
.33 (d, J = 6.8 Hz, 3H), 1.42 (m, 2H), 3.04 (m, 2H), 3.78 (q, J = 7.2 Hz, 1H), 8.25 (brs, 3H),
+
.54 (t, J = 5.5 Hz, 1H); m/z (LCMS, ESI): calc. for [M + H] 130.1; found 130.0
4
.2.7.9. (S)-2-amino-N-(2-phenylethyl)-propanamide hydrochloride (IB34)
Following the general procedure for Boc deprotection, compound IA34 (438 mg, 1.5 mmol)
was exposed to 3 M HCl/MeOH (3.0 mL) for 3 h. 331 mg of compound IB34 was obtained after
1
purification with reverse phase HPLC (97%). H NMR (DMSO-d ) δ 1.27 (d, J = 7.2 Hz, 3H),
6
27

ACCEPTED MANUSCRIPT
2
.73 (m, 2H), 3.26 (m, 1H), 3.38 (m, 1H), 3.73 (brs, 1H), 7.16-7.30 (m, 5H), 8.20 (brs, 3H), 8.57
+
(
t, J = 6.0 Hz, 1H); m/z (LCMS, ESI): calc. for [2M + H] 385.2; found 385.0
4
.2.7.10. L-alaninyl glycine ethyl ester hydrochloride (IB35)
Following the general procedure for Boc deprotection, an ethyl acetate solution of compound
IA35 (411 mg, 1.5 mmol) was exposed to hydrogen chloride gas bubbling for 3 h. 296 mg of
1
compound IB35 was obtained after purification with reverse phase HPLC (94%). H NMR
(
DMSO-d ) δ 1.20 (t, J = 7.0 Hz, 3H), 1.38 (d, J = 7.0 Hz, 3H), 3.92 (m, 3H), 4.10 (q, J = 7.0 Hz,
6
+
2
3
H), 8.24 (brs, 3H), 8.94 (t, J = 5.8 Hz, 1H); m/z (LCMS, ESI): calc. for [2M + H] 349.2; found
49.0
4
.2.7.11. L-alaninyl-L-alanine methyl ester hydrochloride (IB36)
Following the general procedure for Boc deprotection, compound IA36 (411 mg, 1.5 mmol)
was exposed to 3 M HCl/MeOH (3.0 mL) for 3 h. 304 mg of compound IB36 was obtained after
1
purification with reverse phase HPLC (97%). H NMR (DMSO-d ) δ 1.29 and 1.34 (2 d, J = 7.0
6
Hz, together 6H), 3.62 (s, 3H), 3.82 (q, 1H), 4.22 (m, 1H), 8.20 (brs, 3H), 8.70 (d, J = 7.0 Hz,
+
1
H); m/z (LCMS, ESI): calc. for [2M + H] 349.2; found 349.0
4
.2.7.12. Methyl (S)-N-(L-alaninyl)-2-aminobutanoate hydrochloride (IB37)
Following the general procedure for Boc deprotection, compound IA37 (432 mg, 1.5 mmol)
was exposed to 3 M HCl/MeOH (3.0 mL) for 3 h. 318 mg of compound IB37 was obtained after
+
purification with reverse phase HPLC (95%). m/z (LCMS, ESI): calc. for [M + H] 189.1; found
1
89.0
4
.2.7.13. L-alaninyl-L-phenylalanine methyl ester hydrochloride (IB38)
Following the general procedure for Boc deprotection, compound IA38 (525 mg, 1.5 mmol)
was exposed to 3 M HCl/MeOH (3.0 mL) for 3 h. 419 mg of compound IB38 was obtained after
+
purification with reverse phase HPLC (98%). m/z (LCMS, ESI): calc. for [M + H] 251.1; found
2
51.1
4
.2.8. General procedure for the synthesis of compounds 1-3, 5-7, and 31-38: To a cooled
solution of compound 4 (1 equiv.), any of compounds IB1-3, IB5-7 or IB31-38 (1.2-1.5 equiv.),
and PyBop (1.2 equiv.) in anhydrous DCM was added diisopropylethylamine (4 equiv.)
28