Microcystin analogues comprised only of Adda and a single additional amino acid retain moderate activity as PP1/PP2A inhibitors

Article abstract of DOI:10.1016/S0960-894X(03)00588-2

A series of greatly simplified microcystin analogues comprised only of Adda (the β-amino acid common to the microcystins, nodularins, and motuporin,) and a single additional amino acid residue was synthesized and screened for inhibition of the protein phosphatases 1 and 2A. Several of the analogues were shown to be mid-nanomolar inhibitors of the enzymes.

Full text of DOI:10.1016/S0960-894X(03)00588-2

Bioorganic & Medicinal Chemistry Letters 13 (2003) 2907–2911
Microcystin Analogues Comprised Only of Adda and a Single
Additional Amino Acid Retain Moderate Activity as PP1/PP2A
Inhibitors
Brian M. Gulledge, James B. Aggen, Hugo Eng, Khuloud Sweimeh
and A. Richard Chamberlin*
Department of Chemistry, University of California at Irvine, Irvine, CA 92697, USA
Received 6December 2002; accepted 2 May 2003
Abstract—A series of greatly simplified microcystin analogues comprised only of Adda (the b-amino acid common to the micro-
cystins, nodularins, and motuporin,) and a single additional amino acid residue was synthesized and screened for inhibition of the
protein phosphatases 1 and 2A. Several of the analogues were shown to be mid-nanomolar inhibitors of the enzymes.
# 2003 Published by Elsevier Ltd.
The protein phosphatases are a ubiquitous enzymes that
regulate many crucial biological functions, including
neurotransmission, learning and memory, and cell divi-
sion.1 Of the serine-threonine phosphatases, two of the
most prevalent are protein phosphatase 1 and 2A (PP1
and PP2A). Understanding the roles of these two
enzymes in cellular signal transduction has benefited
greatly from the availability of natural toxin inhibitors
such as the microcystins, tautomycin, okadaic acid, and
others;2 however, their respective pharmacological
functions of PP1 and PP2A can be difficult to distin-
guish because of the structural similarity between the
two and the resultant scarcity of selective inhibitors,
particularly those that more strongly inhibit PP1.2
structural features of these toxins that are known to be
essential for potent inhibition of PP1/PP2A by micro-
cystin, the other being a free carboxylate.5 Although
Adda is necessary for inhibition, it is not alone suffi-
cient, as the parent amino acid and its N-acylated deri-
vatives 2, 3, and 4 are only very weakly active at best.6
Our group previously synthesized 17 and several ‘full-
sized’ analogues that were selective for PP1,3 albeit
modestly (up to 7:1). Due to the substantial synthetic
effort required to prepare these macrocyclic ‘non-nat-
ural’ heptapeptides, we have subsequently investigated
the possibility that much simpler derivatives, while they
The microcystins (illustrated by microcystin-LA, 1, Fig.
1) illustrate the point. They are a group of more than 60
cyclic heptapeptides that potently inhibit PP1 and PP2A
relative to other serine-threonine phosphatases but that
do not distinguish between the two (IC50=0.3 nM for
both PP1 and PP2A).3 One structural feature shared by
the microcystins, nodularins, and motuporin—and
which is essential for activity---is the unusual b-amino acid
(2S,3S,8S,9S,4E,6E)-3-amino-9-methoxy-2,6,8-trimethyl-
10-phenyl-4,6-decadienoic acid, 2, generally referred to
by its acronym ‘Adda.’4 This actually is one of two
*Corresponding author. Tel.: +1-949-824-7089; fax: +1-949-824-
8571; e-mail: archambe@uci.edu
Figure 1. Microcystin (1), adda (2), N-Ac-adda (3), and N-Boc-adda (4).
0960-894X/03/$ - see front matter # 2003 Published by Elsevier Ltd.
doi:10.1016/S0960-894X(03)00588-2

2908
B. M. Gulledge et al. / Bioorg. Med. Chem. Lett. 13 (2003) 2907–2911
may be expected to be less potent then the natural tox-
ins, might nonetheless have desirable properties such as
enhanced selectivity. In preliminary experiments toward
this goal, we have found—despite the rather un-encour-
aging precedent for Adda itself—that simple, acyclic ana-
logues containing only N-acetylated Adda and a single
additional amino acid—a surrogate for the iso-glutamyl
residue of microcystin—retain mid-nanomolar inhibi-
tory potency.8 We also find that further simplification, in
the Adda side chain, produces only inactive compounds.
With these considerations in mind, three Adda surro-
gates were designed. The first, 9 (Fig. 3), lacks both the
methoxy and methyl substituent of the parent structure.
Since neither of these substituents is thought to contact
the enzymes, their absence might have minimal effect on
inhibition, unless they play an important conforma-
tional role. This Adda analogue was to be incorporated
into a dipeptide (11) with d-alanine to give a simplified
dipeptide for direct comparison with 510 (Fig. 3). The
other two simplified analogues, 13 and 15 (Fig. 4), are
intended to constrain the side chain to a conformation
similar to that observed in the PP1-bound state of 1, in
which the methyl and methoxy groups are approxi-
mately coplanar.11 All three of the simplified Adda
analogues were synthesized12 as previously described for
Adda itself but employing the modified aldehydes 8, 12,
and 14. The route employed was patterned after on
developed by Rinehart et al., which is not convergent
but is high yielding and convenient. This linear
approach provides the diene after two sequential Wittig
olefinations, and sets the two stereocenters via a dia-
stereoselective aldol reaction. Each of these simplified
Adda analogues was then converted into the corre-
sponding dipeptides 11, 13, and 15, respectively.
Design and Synthesis
In order to determine the minimum structural require-
ments for inhibition by simplified analogues at our tar-
get levels (sub-micromolar IC50), we envisioned simply
linking the two structural elements known to be
required for activity in 1, namely Adda and a carboxylic
acid, via a peptide bond. The resultant dipeptides,
derived from N-acetyl-Adda (3) and d-alanine, glycine,
and l-alanine (5–7, respectively in Fig. 2),4 retain only
the Adda group and a remnant of the adjacent d-iso-
glutamyl residue (or its l-epimer) that have been, in
effect, ‘excised’ from the parent macrocyclic heptapep-
tide. In practice, the three were prepared by coupling
alanine or glycine to N-Ac-Adda with HATU, then
sapponifying the esters.6
Another set of analogues was designed to test combi-
nations of Adda itself with alternative, slightly more
complex iso-glutamyl surrogates. These derivatives were
Further simplification of these dipeptides, specifically in
the Adda side chain, would be highly desirable because
obtaining quantities of Adda itself requires a substantial
multi-step synthetic effort. Although we were mindful
that double bond isomerization, hydrogenation, or
ozonolysis of the Adda side chain renders 1 inactive,9
some structural variation is known to be tolerated. For
example, in microcystin itself a free alcohol or acetate
may substitute for the methyl ether with little effect on
activity; thus, modifications were limited to this part of
the side chain4 of our simplified analogues, leaving the
diene substitution pattern and stereochemistry intact.
Figure 3. Synthesis of simplified Adda analogues 11.
Figure 2. Syntheses of Adda coupled to d-alanine (5), glycine (6), and
l-alanine (7).
Figure 4. Simplified Adda analogues 13 and 15.

B. M. Gulledge et al. / Bioorg. Med. Chem. Lett. 13 (2003) 2907–2911
2909
designed to test the effect of incorporating a scaffold for
Biological Assays
(later) attachment of substituents that might alter PP1/
PP2A selectivity. Proline was chosen as a test scaffold,
bearing in mind that modified prolyl amide bond con-
formational preferences unfavorable to binding could
derail this plan.
All of the target analogues were screened either in a
phosphorylase-a assay16 (Table 2) or a pNPP assay
assay17 (Table 3) to determine IC50 values against PP1
and PP2A. Two additional compounds, 32 and 33, the
methyl ester precursors to 17 and 21, respectively, were
tested as negative controls. While the IC50 values deter-
mined in a given assay under standard conditions are
directly comparable, ‘cross-assay’ comparisons are not.
However, the values obtained for 5, which was screened
in both assays, show that the pNPP assay under these
conditions gives IC50 values that are a factor of 25–50
higher than those determined by the phosphorylase-a
procedure. This difference is also consistent with the 3.0
nM IC50 determined for microcystin-LR in the pNPP
assay (Table 3), which is 10- to 30-fold higher than the
0.1–0.3 nM literature values in the phosphorylase-a
assay.2
To test whether the scaffold backbone alone is tolerated,
the dipeptides 17 and 21 were synthesized (Table 1) by
coupling 3 to both enantiomers of proline (d- and l-
proline, 1713 and 20 respectively). The pyrrolidine-based
analogue 16, a control that was not expected to show
any activity because it lacks the carboxylate group that
plays the essential role of phosphate mimic, was also
prepared in a similar manner. As a preliminary test of
functionalized scaffolds, several 4-substituted proline
dipeptides were prepared, including the hydroxyproline
derivative 18 and the proline dicarboxylates, 19, 20, and
22. Additionally, several 5-substituted proline dicar-
boxylates were chosen because inspection of the micro-
cystin-PP1 X-ray structure11 suggested that a second
carboxylate in this position on the pyrrolidine ring
could possibly interact with Arg96and/or Tyr134 in the
active site of PP1. We prepared this series of dipeptides
from the corresponding amines (23, 24, and 29), which
were in turn synthesized from pyroglutamatic acid by
modifying previously published routes,14,15 to yield the
2,5-diacid analogues 27 and 28 (Fig. 5), with the corre-
sponding acyclic diacid 31, prepared analogously, as a
control (Fig. 6).
Discussion
Not surprisingly, the dipeptides 5–7 inhibited both PP1
and PP2A with a substantial loss in activity compared
to the extremely potent microcystin-LA (1, IC50=0.3
nM for both PP1 and PP2A in the phosphorylase-a
assay3). However, it is rather remarkable that 5, which
lacks five of the seven amino acids present in the parent
microcystin (and obviously also the macrocyclic ring) is
still a 160 nM inhibitor of PP2A in the phosphorylase-a
assay (7 mM in the pNPP assay). This, the most potent
of these three analogues, is the (R)-alanyl derivative and
therefore directly analogous to the natural (R)-iso-glu-
tamyl residue found in 1. Interestingly, of the three it is
also the most microcystin-like in its PP1/PP2A selectiv-
ity, at 1:6in favor of PP2A, compared to 1:1 for
microcystin itself. The glycine-based analogue 6 was
both less active and more selective for PP2A (1:11), and
Table 1. Syntheses of 4-substituted proline analogues
Table 2. Inhibitory activity (IC50’s) against PP1 and PP2A for
analogues
Compd
R1
R2
R3
R4
16
17
18
19
20
21
22
H
H
H
H
H
H
H
H
OTIPS
CO2H
H
H
H
H
H
CO2H
H
Analogue IC50 (mM) PP1a IC50 (mM) PP2Aa Ratio PP1: PP2A
CO2H
CO2H
CO2H
CO2H
H
1
5
6
7
0.0003
1.0
5.60.5
0.0003
0.16
1:1
1:6
1:11
170
7.8
1:22
CO2H
CO2H
H
CO2H
H
H
aValues determined using the phosphorylase-a assay.
Figure 5. Syntheses of 5-substituted proline dicarboxylates.

2910
B. M. Gulledge et al. / Bioorg. Med. Chem. Lett. 13 (2003) 2907–2911
potencies. Incorporation of a silyl ether (OTIPS, 18)
was the best tolerated of the substituents, resulting in
only a 2-fold loss in activity relative to 17. Incor-
poration of either a cis (19) or trans (20) carboxylate
resulted in at least a 4-fold reduction in potency.
Table 3. Inhibitory activity (IC50’s) against PP1 and PP2A for
analogues
Analogue IC50 (mM) PP1a IC50 (mM) PP2Aa Ratio PP1: PP2A
MC-LR
5
11
13
15
16
17
18
19
20
21
22
27
28
31
32
33
0.003
25
0.003
7.0
>100
>100
>100
>100
25
1:1
1:4
—
—
—
>100
>100
>100
>100
55
>100
>100
>100
>100
>100
>50
In direct contrast to the deleterious effects of carboxyl-
substituents at the 4-position, a 5-carboxyl group
enhances activity. The cis-2,5-dicarboxylate 27 was
almost 3-fold more active for PP2A than its simple d-
prolyl parent 17 and in fact was comparable in activity
to the most potent PP2A inhibitor tested, the d-alanyl
dipeptide 5. The pyrrolidine heterocycle apparently
plays a role in this enhanced potency—perhaps appro-
priately preorganizing the two carboxyl groups of 27—
since its the acyclic cousin, 31, was not active.
—
1:2
>1:2
—
—
—
63
>100
>100
>100
>100
10
—
>1:5
>1:1.3
—
—
—
>50
39
>100
>100
>100
>100
>100
>100
Among the Adda dipeptide examples discussed thus far,
a d-amino acid is essential for maximal activity. It was
therefore initially surprising that the l-prolyl
dicarboxylate 28 is only 4-fold less potent than 27, in
contrast to the other l-a-analogues that were invariably
much less potent than their d-diastereomers (e.g., IC50
for PP2A of 7> >5 and 21> >17) This specific 5-car-
boxyl substituent clearly ‘rescues’ the activity of this l-
carboxyproline analogue, perhaps by forming the
aforementioned postulated additional contacts (with
Arg96and/or Tyr134) in the enzyme active site that
compensates for the sub-optimal l-a-carboxy stereo-
chemistry. This conjecture is also consistent with the
increase in potency (discussed above) observed for 27,
which also has an (S)-5-carboxyl substituent on the
proline ring.
aValues determined using the pNPP assay.
Figure 6. Synthesis of acyclic diacid 31.
the (S)-alanyl analogue 7 was the least potent of the
three but with redoubled selectivity for PP2A (1:22).
Taken as a whole, these results suggest an as-yet-unex-
plained role of the iso-glutamyl residue of microcystin in
modulating PP1/PP2A selectivity. They also provide a
benchmark dipeptide, 5, against which to compare
activities of the other analogues.
We conclude that greatly simplifying the macrocyclic
heptapeptide structure of microcystin is possible with-
out unacceptable sacrifices in activity. While the active
dipeptides reported in this Letter do not approach the
sub-nanomolar potency of the parent toxins, they retain
remarkable levels of activity considering that the mac-
rocyclic ring and more than two-thirds of the amino
acid residues of microcystin are absent. The iso-gluta-
myl surrogates tested demonstrate that additional steric
bulk in the phosphate mimic is tolerated, and that a (S)-
5-carboxyprolyl substituent enhances activity in several
analogues, consistent with the proposed hypothesis of
interactions between this substituent and several active
site residues. Finally, all of the active dipeptides inhibit
PP2A with varying degrees of selectivity (PP1/PP2A
IC50 ratios ranging from 1:1.3 to 1:22), unlike the parent
toxin, which is non-selective. This consistent trend sug-
gests PP1/PP2A selectivity is in part a function of the
specific structure of the phosphate mimic of the ana-
logue. Further studies are underway to clarify this issue
and to prepare second generation dipeptides with
enhanced potency and selectivity.
The analogues 11, 13, and 15 were designed as (further)
simplified versions of 5, with modifications in the ter-
minus of the Adda side chain intended to simplify the
synthesis as discussed above. However, unlike 5, all
three compounds were inactive. Although disappoint-
ing, this result further reinforces the established notion
that activity is very sensitive to minor changes in the
structure of Adda.
The prolyl analogue 17, designed to test the effect of an
alternative, pyrrolidine-based surrogate for the iso-glu-
tamyl, proved to be only 3-fold less active than 5. Evi-
dence that both 5 and 17 bind in an analogous manner
is that neither the pyrrolidine heterocycle alone (16), nor
the esterified precursor (32) were active, suggesting the
carboxylate still forms a key contact. As with d-alanyl
analogue 5 and epimeric l-alanyl analogue 7, the
l-prolyl analogues (21 and 22) were significantly less
active than d-prolyl 17, with one notable exception (see
below).
Acknowledgements
Further functionalization of the prolyl heterocycle, at
the 4-position (18–20), resulted in somewhat decreased
We gratefully acknowledge the financial support of the
National Institutes of Health (GM-57550).

B. M. Gulledge et al. / Bioorg. Med. Chem. Lett. 13 (2003) 2907–2911
2911
References and Notes
J=5.0, 14.0 Hz, 1H), 3.16(s, 3H), 3.22 (ddd, J=5.5, 5.5, 6.5
Hz, 1H), 4.19 (dq, J=7.0, 7.5 Hz, 1H), 4.40 (m, 1H), 5.38 (d, J
10.0 Hz, 1H), 5.44 (dd, J=6.5, 16.0 Hz, 1H), 6.05 (d, J=16.0
Hz, 1H), 7.17 (d, J=7.5 Hz, 3H), 7.25 (t, J=7.5 Hz, 2H), 7.60
(d, J=9.0 Hz, 1H), 8.01 (d, J=7.0 Hz, 1H).
11. Goldberg, J.; Huang, H. B.; Kwon, Y. G.; Greengard, P.;
Nairn, A. C.; Kuriyan, J. Nature 1995, 376, 745.
12. Pearson, C.; Rinehart, K. L.; Sugano, M.; Costerison,
J. R. Org. Lett. 2000, 2, 2901.
13. 17: 1H NMR (500 MHz, DMSO-d6, 90 ꢀC) d 0.97 (d,
J=6.7 Hz, 3H), 1.00 (d, J=6.5 Hz, 3H), 1.55 (s, 3H), 1.83 (s,
3H), 1.85–1.97 (m, 3H), 2.09 (m, 1H), 2.57 (m, 1H), 2.66 (dd,
J=6.9, 14.4 Hz, 1H), 2.74 (dd, J=8.4, 14.4 Hz, 1H), 3.01 (m,
1H), 3.18 (s, 3H), 3.24 (m, 1H), 3.52 (m, 1H), 3.68 (m, 1H),
4.24 (br s, 1H), 4.44 (br s, 1H), 5.37 (d, J=9.3 Hz, 1H), 5.51
(dd, J=5.2, 15.5 Hz, 1H), 6.08 (d, J=15.5 Hz, 1H), 7.15–7.24
(m, 5H), 7.52 (br s, 1H).
1. McCluskey, A.; Sim, A. T. R.; Sakoff, J. A. J. Med. Chem.
2002, 45, 1151.
2. Sheppeck, J. E.; Gauss, C. M.; Chamberlin, A. R. Bioorg.
Med. Chem. 1997, 5, 1739.
3. Aggen, J. B.; Humphrey, J. M.; Gauss, C. M.; Huang, H.-B.;
Nairn, A. C.; Chamberlin, A. R. Bioorg. Med. Chem. 1999, 7, 543.
4. For two exceptions, see: Namikoshi, M.; Rinehart, K. L.;
Sakai, R.; Stotts, R. R.; Dahlem, A. M.; Beasley, V. R.; Car-
michael, W. W.; Evans, W. R. J. Org. Chem. 1992, 57, 866.
5. Gauss, C. M.; Sheppeck, J. E.; Nairn, A. C.; Chamberlin,
A. R. Bioorg. Med. Chem. 1997, 5, 1751.
6. For a review, see: Gulledge, B. M.; Aggen, J. B.; Huang, H.-B.;
Nairn, A. C.; Chamberlin, A. R. Curr. Med. Chem. 2002, 9, 1991.
7. Humphrey, J. M.; Aggen, J. B.; Chamberlin, A. R. J. Am.
Chem. Soc. 1996, 118, 11759.
8. Gulledge, B. M.; Aggen, J. B.; Chamberlin, A. R. Bioorg.
Med. Chem. Lett. See preceding paper. doi: 10.1016/S0960-
894X(03)00589-4.
14. Ohta, T.; Hosoi, A.; Kimura, T.; Nozoe, S. Chem. Lett.
1987, 2091.
15. Bridges, R. J.; Stanley, M. S.; Anderson, M. W.; Cotman,
C. W.; Chamberlin, A. R. J. Med. Chem. 1991, 34, 717.
16. Cohen, P.; Alemany, S.; Hemmings, B. A.; Resink, T. J.;
9. Sivonen, K.; Carmichael, W. W.; Namikoshi, M.; Rinehart,
K. L.; Dahlem, A. M.; Nielmla, S. I. Appl. Enviror. Microbiol.
1990, 56, 2650.
Stralfors, P.; Tung, H. Y. L. Methods Enzymol. 1988, 159, 390.
˚
1
10. 5: H NMR (500 MHz, DMSO-d6) d 0.94 (d, J=7.0 Hz,
3H), 0.96(d, J=7.0 Hz, 3H), 1.19 (d, J=7.0 Hz, 3H), 1.52 (s,
3H), 1.82 (s, 3H), 2.63 (dd, J=7.0, 14.0 Hz, 1H), 2.73 (dd,
17. Experiments verified linear behavior of PP1 in the pNPP
assay, which has previously reported to be problematic in
some cases: Takai, A.; Mieskes, G. Biochem. J. 1991, 275, 233.