C-3 benzoic acid derivatives of C-3 deoxybetulinic acid and deoxybetulin as HIV-1 maturation inhibitors

Article abstract of DOI:10.1016/j.bmc.2016.03.001

A series of C-3 phenyl- and heterocycle-substituted derivatives of C-3 deoxybetulinic acid and C-3 deoxybetulin was designed and synthesized as HIV-1 maturation inhibitors (MIs) and evaluated for their antiviral activity and cytotoxicity in cell culture. A 4-subsituted benzoic acid moiety was identified as an advantageous replacement for the 3′3′-dimethylsuccinate moiety present in previously disclosed MIs that illuminates new aspects of the topography of the pharmacophore. The new analogs exhibit excellent in vitro antiviral activity against wild-type (wt) virus and a lower serum shift when compared with the prototypical HIV-1 MI bevirimat (1, BVM), the first MI to be evaluated in clinical studies. Compound 9a exhibits comparable cell culture potency toward wt virus as 1 (WT EC₅₀ = 16 nM for 9a compared to 10 nM for 1). However, the potency of 9a is less affected by the presence of human serum, while the compound displays a similar pharmacokinetic profile in rats to 1. Hence 9a, the 4-benzoic acid derivative of deoxybetulinic acid, represents a new starting point from which to explore the design of a 2nd generation MI.

Full text of DOI:10.1016/j.bmc.2016.03.001

Accepted Manuscript
C-3 Benzoic Acid Derivatives of C-3 Deoxybetulinic Acid and Deoxybetulin
as HIV-1 Maturation Inhibitors
Zheng Liu, Jacob J. Swidorski, Beata Nowicka-Sans, Brian Terry, Tricia
Protack, Zeyu Lin, Himadri Samanta, Sharon Zhang, Zhufang Li, Dawn D.
Parker, Sandhya Rahematpura, Susan Jenkins, Brett R. Beno, Mark Krystal,
Nicholas A. Meanwell, Ira B. Dicker, Alicia Regueiro-Ren
PII:
S0968-0896(16)30138-9
DOI:
http://dx.doi.org/10.1016/j.bmc.2016.03.001
BMC 12850
Reference:
Bioorganic & Medicinal Chemistry
To appear in:
Received Date:
Revised Date:
Accepted Date:
11 January 2016
21 February 2016
1 March 2016
Please cite this article as: Liu, Z., Swidorski, J.J., Nowicka-Sans, B., Terry, B., Protack, T., Lin, Z., Samanta, H.,
Zhang, S., Li, Z., Parker, D.D., Rahematpura, S., Jenkins, S., Beno, B.R., Krystal, M., Meanwell, N.A., Dicker, I.B.,
Regueiro-Ren, A., C-3 Benzoic Acid Derivatives of C-3 Deoxybetulinic Acid and Deoxybetulin as HIV-1
Maturation Inhibitors, Bioorganic & Medicinal Chemistry (2016), doi: http://dx.doi.org/10.1016/j.bmc.2016.03.001
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.

C-3 Benzoic Acid Derivatives of C-3 Deoxybetulinic Acid and Deoxybetulin as HIV-1 Maturation
Inhibitors
Zheng Liu,1,* Jacob J. Swidorski,1 Beata Nowicka-Sans,2 Brian Terry,2 Tricia Protack,2 Zeyu Lin,2
Himadri Samanta,2 Sharon Zhang,2 Zhufang Li,2 Dawn D. Parker,3 Sandhya Rahematpura,3 Susan
Jenkins,3 Brett R. Beno,4 Mark Krystal,2 Nicholas A. Meanwell,1 Ira B. Dicker,2 and Alicia Regueiro-
Ren1,*
1Department of Discovery Chemistry, 2Virology, 3Pharmaceutical Candidate Optimization, 4Computer-Assisted
Drug Design
Bristol-Myers Squibb Research and Development
5 Research Parkway
Wallingford, CT 06492, USA
ABSTRACT: A series of C-3 phenyl- and heterocycle-substituted derivatives of C-3 deoxybetulinic
acid and C-3 deoxybetulin was designed and synthesized as HIV-1 maturation inhibitors (MIs) and
evaluated for their antiviral activity and cytotoxicity in cell culture. A 4-subsituted benzoic acid moiety
was identified as an advantageous replacement for the 3’3’-dimethylsuccinate moiety present in
previously disclosed MIs that illuminates new aspects of the topography of the pharmacophore. The
new analogs exhibit excellent in vitro antiviral activity against wild-type (wt) virus and a lower serum
shift when compared with the prototypical HIV-1 MI bevirimat (1, BVM), the first MI to be evaluated in
clinical studies. Compound 9a exhibits comparable cell culture potency toward wt virus as 1 (WT EC50
= 16 nM for 9a compared to 10 nM for 1). However, the potency of 9a is less affected by the presence
1

of human serum, while the compound displays a similar pharmacokinetic profile in rats to 1. Hence 9a,
the 4-benzoic acid derivative of deoxybetulinic acid, represents a new starting point from which to
explore the design of a 2nd generation MI.
Keywords: HIV-1, maturation inhibitors, betulinic acid, triterpene, antiviral.
INTRODUCTION
Combination antiretroviral therapy (cART) has transformed HIV-1 infection into a more
manageable disease for the majority of patients.1,2 The combination regimes consist of three
antiretroviral drugs that are generally selected from four classes based upon their mechanism of action -
nucleoside and nucleotide reverse transcriptase inhibitors, protease inhibitors, and integrase inhibitors,
with a fifth class, entry inhibitors, used less frequently.3 However, issues associated with long term
toxicity,4 therapeutic compliance5 and the development of resistance6 remain as major causes of
treatment failure. In order to address these problems, the discovery of antiretroviral drugs targeting new
mechanisms of action (MOAs) with improved safety profiles remains a high priority. Virion maturation
is a key step in the virus life cycle and therapeutic intervention at this stage of the replication process has
attracted the attention of several groups.7-11 During this phase of the life cycle, the viral protease cleaves
the Gag polyprotein sequentially in five different positions in a process that is carefully choreographed
to release the structural proteins MA, CA, NC and p6, as well as two spacer peptides, SP1 and SP2
(Figure 1).12 The last of the proteolytic cleavages is the rate-limiting step of this process and takes place
at the CA-SP1 junction and is the critical step in the maturation of the virion into the conical shape that
is characteristic of infectious virus.13 Maturation inhibitors (MIs) bind to this segment of the Gag
polyprotein, preventing the protease from converting p25 Gag (CA/SP1) to the mature p24 (CA) protein,
consequently leading to the production of immature HIV-1 particles which are not infectious.12,14,15
2

Figure 1. 1a) Protease processing sites within the HIV-1 Gag polyprotein. MIs prevent the last cleavage
of p25 Gag (CA/SP1) to the mature p24 (CA). The order of the cleavage events is noted in red. 1b)
Amino acid residues in the proximity of the CA/SP1 cleavage site. Amino acid residues that contribute
to reduced susceptibility to 1 are marked in blue.
Bevirimat (1) was the first MI to be advanced into clinical trials and achieved proof of concept
for this MOA by reducing viral load in HIV-1 infected patients.16 However, subsequent studies in
treatment-experienced patients receiving the drug by oral administration for 10 days revealed that only
45% of the subjects responded to therapy.17 A careful analysis of the data differentiated responders
from non-responders based on baseline polymorphisms occurring proximal to the CA/SP1 cleavage site,
specifically at residues 362 and 369-371 in the Gag protein.18 In addition, 1 exhibited high protein
binding (>99.9%), which reduced the in vitro antiviral activity by>100-fold, necessitating a high dose in
order to achieve clinical efficacy, since target plasma levels were defined as multiples of the protein
binding-adjusted EC50 in cell culture.19 Furthermore, the poor solubility of 1 translated to lower oral
exposure than expected as a consequence of solubility- and dissolution-limited absorption,20 further
complicating the development path for 1 as an anti-HIV-1 agent, and this compound was eventually
abandoned.21 These shortcomings provide a clear objective for the pre-clinical profile of a second
3

generation MI in terms of improved polymorphic virus coverage, reduced protein binding and better
pharmaceutical properties.
Structurally, 1 is the 3-O-(3’3-dimethyl succinyl) ester derivative of the natural product betulinic
acid (3). Although 3 possesses weak anti-HIV-1 activity, it is the dimethyl succinate side chain that
decorates the core C-3 hydroxyl group that confers the unique maturation inhibitory activity associated
with 1, exemplified by an EC50 value of 0.35 nM in acutely infected H9 lymphocytes.22 However, in
correlation with the clinical results, where the efficacy of 1 was highly diminished in patients with pre-
existing variations at Gag positions 369-371 of the SP1 peptide, the in vitro activity of 1 toward the most
prevalent polymorphic variant (V370A; ~12.4% of isolates listed in the Los Alamos National
Laboratory HIV sequence database) was reduced by more than 100-fold when compared with its
potency toward wt virus.23 More recently, it has been demonstrated that modifications at the C-28
position of 1 can, to some extent, improve polymorphic coverage and protein binding.23,24
4

In the absence of information on the structure of the MI binding site, extensive work has been
directed towards the optimization of the C-3 side chain. The terminal carboxylic acid, proper linker
length, gem-dimethyl substituent and ester functionality are deemed to be critical for potent inhibition of
virus containing the most common QVT sequence at the 369-371 positions.25 The vital importance of
the gem-dimethyl group and the carboxylic acid moiety for potency points towards a conformational
preference for a specific positioning of the carboxylic acid presumably aided, in part, by the Thorpe-
Ingold effect.26 The only latitude for structural change that has been described in the literature is with
the homologated analog 3-O-(3’3-dimethyl glutaryl) ester 2 which exhibits an EC50 value of 2.3 nM
toward wt virus, 6.5-fold less potent than 1.27 The rigidity of the triterpenoid core defines the relative
spatial disposition between the C-3 acyl side chain and the C-28 functionality, and even minor variations
in this distance and vector can result in a significant loss of potency.28
The use of conformationally restricted 3-O-acyl derivatives as a means to improve antiviral
activity by constraining the position of the C-3 pharmacophore was unsuccessful and resulted in
compounds with considerably diminished potency.29 We were interested in exploring the concept of
conformational constraint in the context of non-ester-based C-3 side chains, with a view to providing
additional illumination of the spatial relationships between the acid moiety and the triterpene core. With
the appropriate design, it was anticipated that the positioning of the carboxylic acid could be carefully
controlled, potentially leading to improved antiviral and physical properties. In order to place the
carboxylic acid at a similar distance from the core as in 1, the use of a six membered ring spacer directly
attached to the C-3 position of BA was considered, with the introduction of a planar phenyl ring as the
spacer specifically contemplated as a means of further modulating conformational freedom.
Accordingly, we set out to prepare a series of benzoic derivatives of both C-3 deoxybetulinic acid and
C3-deoxybetulin, with the objective of identifying a basic platform for the development of second
generation MIs that would show equivalent or, possibly, improved coverage of naturally occurring
5

polymorphisms with reduced protein binding. The results of the first steps of that process are reported
herein.30
CHEMISTRY
The synthesis of the modified C-3 betulinic acid analogs 9a-9u that are compiled in Table 1 is
outlined in Scheme 1 and began with commercially available betulinic acid.31 The C-28 carboxylic acid
of 3 was protected with benzyl bromide in the presence of potassium carbonate to give 5. Oxidation of
the hydroxyl group using pyridinium chlorochromate provided ketone 6 which, after treatment with
1,1,1-trifluoro-N-phenyl-N-(trifluoromethylsulfonyl)methanesulfonamide in the presence of potassium
bis(trimethylsilyl)amide at -78oC, afforded the enol triflate 7, a key intermediate. A Suzuki coupling
reaction of 7 with various aryl and heteroaryl boronic acids bearing an unprotected carboxylic acid
functionality, provided the corresponding cross-coupled intermediates. Careful control of the catalyst
loading (0.1-1 equiv.) and reaction time (1-24 h) facilitated the selective hydrogenolytic removal of the
benzyl group with no reduction of the two double bonds present in the core to afford target compounds
9a-9i, and 9k-9u.
Alternatively, triflate 7 was reacted with 4-carboxylphenylboronic acid followed by treatment
with thionyl chloride to yield the acid chloride 8. Conversion of 8 to the acylsulfonamide 9j was
accomplished by treatment with the corresponding sulfonamide in the presence of Hunig’s base,
followed by ester deprotection under the hydrogenolytic conditions described above. The synthesis of
9r also started from triflate 7 which was subjected to a Suzuki coupling with (1H-pyrazol-5-yl)boronic
acid to afford the pyrazole derivative 10. Subsequent alkylation of 10 with ethyl bromoacetate in the
presence of potassium carbonate yielded 11 which was selectively debenzylated using triethylamine and
t-butyldimethylsilane with the help of catalytic amount of palladium acetate to provide 12. Finally,
6

hydrolysis of ethyl ester 12 using aqueous NaOH or LiOH as the base generated 9r. It should be noted
that this route was initially approached using the C-28 methyl ester derivative instead of the benzyl
ester; however, the final unmasking of the carboxylic acid required harsh conditions (microwave at
240oC/8 h or heating 100oC/2 days) that resulted in loss of material and/or poor yields. The use of a
benzyl ester as the protecting group allowed for milder reaction conditions for removal and more
convenient workups which led to improved yields.
Scheme 1. Synthesis of C-3 deoxybetulinic acid derivatives 9a-9u.
7

8

Reagents and conditions: (a) K2CO3, BnBr, 60oC, 3.5 h, 97%; (b) PCC, CH2Cl2, 15 h, 98%; (c)
KHMDS, PhNTf2, THF, -78oC, 1 h, 86%; (d) (i) R1-B(OH)2, Na2CO3, Pd(Ph3P)4, DME, H2O, 100oC,
1.5-3 h. (ii) Pd/C, H2, 1 atm, EtOAc, MeOH, rt 3-24 h, 2-52% over two steps; (e) (i) (4-PhCOOH)-
B(OH)2, Na2CO3, Pd(PPh3)4, DME, H2O, 100oC, 2 h (ii) SOCl2, CH2Cl2, reflux, 17 h, 41% over two
steps; (f) (i) RSO2NH2, DMAP, Hunig's Base, DME, rt, 16 h. (ii) 10% Pd/C, H2, 40 psi, EtOAc, MeOH,
rt., 17 h, 7-26% over two steps; (g) 1H-pyrazol-5-ylboronic acid, Pd(PPh3)4, Na2CO3·H2O, dioxane,
water, 85oC, 5.5 h, 61%; (h) BrCH2COOEt, K2CO3, DMF, 75°C, 17 h, 73%; (i) TBDMSH, Pd(OAc)2,
TEA, DCE, 60°C, 7.7 h, 55%; (j) NaOH, 1,4-dioxane, 85°C, 17.5 h, 77%.
The C-3 deoxybetulin analogs 16a-16q were prepared in a similar manner, as outlined in Scheme
2. Betulin (4) was used as the starting material and the C-28 primary hydroxyl group was protected
using benzoic anhydride to afford ester 13. Preparation of triflate 15 followed by Suzuki coupling with
either aryl carboxylic acid or ester boronic acids or a Stille coupling reaction was performed as
described above for the betulinic acid analogs. Ultimately, removal of the benzoyl protecting group in
the C-28 position and hydrolysis of the aryl esters, when present, were accomplished in one step using
LiOH or NaOH in a mixture of dioxane/water to yield the corresponding target C-28 alcohols 16a-16q
(Table 1). The yields for the reactions described in Schemes 1 and 2 were not optimized.
Scheme 2. Synthesis C-3 deoxybetulin derivatives 16a-16q.
9

Reagents and conditions: (a) (PhCO)2O, DMAP, pyridine, 60oC, 3 h; (b) PCC, CH2Cl2, rt, 2 h; (c)
KHMDS, PhNTf2, THF, -78oC, 2 h, 60% over 3 steps; (d) (i) R1-B(OH)2, Na2CO3, K2CO3 or K3PO4,
Pd(PPh3)4, dioxane or DME, H2O, 85oC-100oC, 3-24 h; (ii) LiOH or NaOH, dioxane, H2O, 75oC, 3-15 h,
2-38% over 2 steps or (i) R1-SnBu3, LiCl, Pd(Ph3P)4, dioxane, 85oC, 15.5 h; (ii) LiOH or NaOH,
dioxane, H2O, 3.5-18 h, 6-47% over 2 steps.
RESULTS AND DISCUSSION
The in vitro antiviral activity and cytotoxicity associated with the target compounds are
assembled in Table 1. The C-3 modified betulinic acid and betulin analogs were tested for inhibitory
activity toward wt HIV-1 infection in a multiple cycle assay in MT-2 cells using variants of the
NLRepRlucP373S virus, an engineered laboratory strain derived from NL4-3 that expresses the Renilla
10

luciferase gene for use as a marker for virus growth. The effect of human serum albumin on compound
potency was measured by comparison of inhibitory activities using NLRepRlucP373S virus-infected
MT-2 cells in 10% fetal bovine serum (FBS) to that determined using 10% FBS plus 15 mg/mL human
serum albumin (HSA). In complete media (10% FBS + 40% human serum + 27 mg/mL HSA), which
more closely matches the physiological concentration of HSA (45 mg/mL), the EC50 value for 9a was
only 2-fold higher (0.31 µM), when compared to that in 15 mg/mL HSA. Therefore, 15 mg/mL HSA
was used as the screening paradigm to rank the effects of the most important serum component
(albumin) shown to reduce the antiviral activity of 1 in serum.32 The ratio between the cell culture
potency towards wt virus in the presence and absence of human serum is noted in Table 1 as fold-change
WT (HSA) EC50 / WT EC50. The V370A virus, which encodes reduced susceptibility to 1, was identical
to the wild type virus except for this single amino acid substitution and represents the most common
polymorphic variation in subtype B HIV-1 at position 370 in the Gag protein (~12.4%).33 The
cytotoxicity of the compounds was determined in parallel in MT-2 cells using a redox dye assay (3-(4,5-
dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2Htetrazolium)
with
data
reported as the half maximal concentration associated with cell viability.34
Table 1. Antiviral activity and cytotoxicity for compounds 1, 2, 9a-9u and 16a-16q
11

WT HIV-1
inhibition in
the presence
of 15 mg/mL
HSA
V370A
HIV-1
WT HIV-1
inhibition
EC50 (M)a
CC50
Fold change
EC50WT(HSA)
/EC50WT
R1
R2
(M)
inhibition
EC50 (M) a
a
EC50 (M) a
0.010±0.011
(n = 69)a
0.974±0.71
(n = 15)a
0.553±0.633
(n = 34)a
COOH
97.4
16.8
1
COOH
COOH
COOH
0.004
0.909
227
9.4
-
-
4.25
27.2
13.8
2
0.016±0.013
(n = 25)a
0.150±0.099
(n = 4)a
0.233±0.305
(n = 10)a
9a
9b
>8
>8
-
-
-
-
COOH
-
10.5
9c
MeHNO2S
COOH
COOH
1.23
2.13
-
-
-
-
-
-
13
9d
9e
12.7
COOH
COOH
2.52
-
-
-
-
-
9.72
13.4
9f
0.308
>8
9g
12

O
COOH
COOH
1.60
-
-
-
-
-
-
22.6
4.45
9h
9i
MeHN
>1.60
COOH
0.124
0.275
2.2
0.937
8.68
9j
0.215±0.257
(n = 4)a
COOH
COOH
-
-
-
-
0.633
-
>8
9k
9l
0.682
6.55
COOH
COOH
0.734
0.571
-
-
-
-
-
-
>8.0
1.83
9m
9n
HOOC
COOH
COOH
3.2
-
-
-
-
-
5.03
7.73
9o
9p
HOOC
COOH
HN
1.47
1.57
O
COOH
COOH
3.6
-
-
-
-
-
15.9
>30
9q
9r
0.167
>3
13

COOH
COOH
0.059
-
-
>2
0.79
3.08
9s
9t
0.021±0.025
(n = 5)a
0.331
15
0.589
0.018±0.023
(n = 5)a
COOH
0.067
3.7
0.036
10.1
9u
0.011±0.007
(n = 12)a
0.027±0.011
(n = 8)a
0.212±0.177
(n = 8)a
CH2OH
CH2OH
2.5
-
9.65
34.3
16a
16b
3.4
-
>4.0
>1.6
Me
O S O
CH2OH
CH2OH
>1.6
>1.6
-
-
8.82
20
16c
16d
HN
>4
>4
>4
CH2OH
CH2OH
CH2OH
1.08
0.503
0.061
1.08
1
-
>4
>2
22.3
4.30
0.57
16e
16f
16g
-
-
-
0.348
CH2OH
0.039
0.295
7.5
>2
12.1
16h
0.38±0.035
(n = 4)a
CH2OH
CH2OH
-
-
>2
14.1
1.97
16i
16j
0.024
0.094
3.9
0.262
14

CH2OH
CH2OH
0.046
0.081
0.185
0.065
4
0.309
2.78
18.8
9.76
16k
16l
0.8
16
m
CH2OH
CH2OH
CH2OH
0.513
0.038
0.055
1.186
-
2.3
-
>4
0.212
1.4
28.7
>20
16.5
16n
16o
0.050
1
CH2OH
CH2OH
0.114
0.613
0.117
0.660
1
1
>4
17.3
7.02
16p
16q
3.88
aThe data provided are the means of at least triplicate values with a mean CV (confidence value) of 83
+/- 50 % for all of the experiments, and all CV values <200 %. Where experiments were performed 4 or
more times, the corresponding standard deviations are noted.
We first elected to prepare the 4-benzoic acid 9a and the 3-benzoic acid 9b derivatives, which
seemed to offer a good approximation of the spatial disposition for the carboxylic acid group when
compared with the conformationally more flexible dimethylsuccinate side chain of 1. Preparation of the
2-benzoic acid analog was not considered at this time for that reason and also because that topological
relationship afforded poorly active compounds when examined in the context of saturated ring
15

systems.29 Compound 9a exhibited excellent cell culture potency toward the wt virus with an EC50 = 16
nM, comparable to that of 1, which demonstrated an EC50 = 10 nM. However, in the presence of human
serum, 9a was 6.5-fold more potent than 1, reflective of a much reduced serum binding, (9.4-fold for 9a
compared to 129-fold for 1).32 Remarkably, the 3-benzoic analog 9b was over 450-fold less active as an
inhibitor of wt virus when compared with 9a, demonstrating the critical importance of the topographical
disposition of the carboxylic acid moiety with respect to the core in the context of this kind of
conformational constraint.
An overlay of compounds 1, 9a, and 9b is depicted in Figure 2. The model was generated from
low energy conformers identified via conformational searches performed using BatchMin (Version 9.6,
Schrödinger, LLC), the OPLS_2005 force-field, and an implicit water solvation model.35 The lowest
energy conformer of 9a, a potent antiviral agent in which the disposition of the carboxylate with respect
to the core is the most restricted, was used for fitting conformers of 1 and 9b based on molecular shape
and pharmacophore features. The energy of the conformer of 1 included in the model was < 0.2
kcal/mol above that of the lowest energy conformer identified for the compound. The conformer of 9a
shown is the lowest energy conformer found. In both 9a and 9b, the aromatic ring and the triterpenoid
core show an almost perpendicular disposition, induced in part by the adjacent gem-dimethyl group in
the A-ring of the core. The dimethylsuccinate side chain of 1 and the 4-benzoic acid of 9a overlap quite
well to project the respective carboxylic acid moieties in very close spatial proximity. However, in the
3-substituted analog 9b, the carboxylic acid is clearly displaced and could not readily overlap with the
carboxylic acid of 1. Although no structural information regarding how 1 or any other MI specifically
interacts with the binding determinants of the assembled Gag polyprotein is available, these results
complement previous reports regarding the importance of the precise location of the carboxylic acid in
order to achieve antiviral potency, while providing a more definitive understanding of the topographical
relationship with the core.27
16

Figure 2. Overlay of compounds 1 (green), 9a (orange), and 9b (magenta).a
aImage generated with the PyMOL Molecular Graphics System (Version 1.7.6, Schrödinger, LLC).
The 4-benzoic acid deoxy-betulin analog 16a exhibited comparable antiviral potency toward wt
and V370A viruses as those of its BA analog 9a, while in the presence of HSA 16a was 3-fold more
potent than 9a. Therefore, both C-17 acids (9) and C-28 alcohols (16) were used as vehicles to further
explore the SAR associated with novel C-3 side chains, with some elements of the survey restricted to
one or the other chemotype based on synthetic accessibility. In the initial step of this survey, a variety
of functionalities that replaced the carboxylic acid moiety in either the 3- or 4-position of the phenyl ring
were explored in the context of compounds 9c-9j and 16b-16d. It has been shown that under some
circumstances, tetrazoles are effective acid isosteres, since they are ionized at physiological pH and
maintain a planar topology similar to a carboxylic acid.36 Although this heterocycle is larger than a
simple carboxylic acid,36c,d we anticipated it might be an acceptable substitute since the 3-O-(3’3-
dimethyl glutaryl) ester homologue 2 shows similar potency to 1. However, 16b showed a significant
loss of antiviral potency, while deployment of a tetrazole at C-3 (9c) failed to improve upon 9b,
demonstrating that this kind of bioisosterism is sensitive to context. Similarly, analogs incorporating
other potential acid isosteres, such as sulfonamides 9d, 9e and 16c and nitro-derivative 16d, were also
17

much less potent than the corresponding carboxylic acids.36,37 Only phenol 9g and acylsulfonamide 9j
maintained sub-micromolar inhibitory potency towards wt virus, but these were 18- and 7-fold less
potent than 9a against the wt virus, respectively.36,38 Of the acid isosteres examined, the
acylsulfonamide 9j offered the best profile towards wt virus. Perhaps not surprisingly, given the
absence of overt acidic functionality, both amide 9h and nitrile 9i were also less potent antiviral agents.
The effect of a spacer between the phenyl ring and the carboxylic acid was probed by
introducing linkers with different lengths and degrees of flexibility designed to map out a range of
topographical opportunity, as exemplified by analogs 9k-9q. The results revealed a substantial loss of
potency with this series, with only 9k-9m, which incorporate the shortest side chains and most closely
resemble the positioning of the carboxylic acid relative to 9a, displaying sub-micromolar inhibitory
potencies. The effect of replacement of the phenyl ring with a series of 5- and 6-membered heterocycles
was explored since these present vectors that probe additional topological relationships, a survey
represented by analogs 16e-16i and 9r-9s. Interestingly, the six-membered ring heterocyclic analogs
16e and 16f showed considerably weaker antiviral activity (EC50 >0.5 M) toward the wt virus than
compounds 9r, 9s and 16g-16i which incorporate five-membered ring heterocycles. Isoxazole 16h and
pyrazole 9r offered the best profile within this series, exhibiting EC50 values of 39 nM and 167 nM,
respectively, toward wt virus with cytotoxicities comparable to that of 9a. Both, pyrazole 16g and
thiazole 9s appeared to very potent against wt virus, with EC50 values of 61 and 59 nM, respectively, but
these compounds had low therapeutic indices (EC50/CC50), which casts doubt on the authenticity of the
antiviral effect.
Consequently, attention returned to the 4-benzoic acid analogs 9a and 16a and the SARs were
further explored by introducing substituents at positions 2 and 3 of the phenyl ring. These modifications
were designed to influence conformational relationships between the aromatic ring and the core, as well
as the carboxylic acid and the phenyl ring. Substitution meta- to the carboxylic acid moiety, exemplified
18

by compounds 16o-16q, resulted in reduced antiviral potency toward the wt virus, with the EC50 values
inversely correlated with the size of the substituent, such that the larger the substituent, the lower the
potency. This result seems to indicate that the orientation of the phenyl ring with respect to the core
may be affected by steric interactions between the meta-substituent and the A-ring gem-dimethyl
moiety, resulting in a preferred conformation that inadequately complements the binding site. While the
smaller meta-F, which has an A-value = 0.15, would allow the phenyl ring to adopt an angle with regard
to the core that resembles more closely that observed in Figure 2 for 9a, the larger Cl and Me
substituents, with A-values = 0.43 and 1.7, respectively, are forecasted to have larger steric effects that
might distort the orientation of the ring to a more perpendicular disposition with respect to the core,
which presumably is detrimental for potency.39
Substitution ortho- to the carboxylic acid, as in 9t-9u and 16j-16n, was better tolerated. The
ortho-F substituted compound 9u showed similar antiviral activity toward wt virus as 9a. However, the
effect of the ortho-F substitution did not translate to the alcohol series, as illustrated by compound 16k,
which is 4-fold weaker than 16a toward wt virus. The ortho-Cl analogs 9t and 16j maintained the
antiviral activity of the prototypes 9a and 16a, respectively. Introduction of other substituents in the
ortho-position, including hydroxyl (16l) and methoxy (16m), further diminished potency. Interestingly,
a second carboxylic acid was remarkably well tolerated, as 16n is just 3-fold weaker than 16a toward the
wt screening virus.
In addition to 9a and 16a, select compounds were tested for inhibitory potency toward wt virus
in the presence of HSA and the results compiled in Table 1. Phenol 9g experienced more than a 4-fold
serum shift, while the potency of the acylsulfonamide 9j was shifted by just 2-fold. Isoxazole 16h
experienced a similar serum shift (7.5-fold) to 9a, while pyrimidine 16e displayed no change in potency
in the presence of HSA. The serum shift for the ortho-F analogs 9u and 16k was the same (4-fold),
while the ortho-Cl compounds 9t and 16j exhibited a serum shift of 15- and 4-fold, respectively,
19

recapitulating the larger serum shift associated with the betulinic acid analog than the betulin analogs,
observed for 9a and 16a. Within the betulin series, the ortho- analogs 16j and 16k displayed a 4-fold
larger serum shift than the corresponding meta- analogs 16o and 16p, neither of which suffered a loss of
potency in the presence of HSA. Thus, for this series of C-3 benzoic acid derivatives the effect of serum
is generally modest and less than that observed for 1.
In terms of activity toward the highly prevalent BVM-resistant polymorphism, V370A, the EC50s
of 9a and 16a were both = 0.2 M, only slightly improved compared to 1, EC50 = 0.55 M. None of the
modifications introduced into 9a explored in analogs 9j-9k, 9p, 9r-t, 16b-16i, 16l-16q significantly
affected the potency toward the V370A polymorphic virus. Only the ortho-F analog 9u showed better
potency than 9a, although in the betulin series the same modification (16k) resulted in a loss of potency
when compared with 16a. Notably, in both cases, this modification adversely affected the solubility of
the compounds, making them sufficiently difficult to formulate that they could not be evaluated in vivo
to determine their pharmacokinetic profile.
The mechanism of the antiviral activity of 9a was studied to allow a comparison with 1. To that
end, 9a was evaluated in an in vitro assay designed to assess inhibition of the final step of CA/SP1 Gag
cleavage, which results in the conversion of the capsid precursor protein p25 to the capsid protein p24,
concomitant with the release of the SP1 spacer peptide,40 the final and key step in the triggering of HIV-
1 maturation.41 Assembled HIV-1 viral-like particles (VLPs) produced in HEK 293T cells using an
optimized codon expression vector (1_pcDNAGagOpt) and the VLPs were used as a substrate for in
vitro HIV-1 protease cleavage of p25 capsid/spacer peptide 1 (CA/SP1).42 As shown in the Western blot
in Figure 3, 9a at 2 M (lane 10) inhibits p24 production by 1.6-fold, compared with an average value of
1.6-fold for 1 at 3 and 1 µM (lanes 3 and 4, respectively), consistent with the assignment of 9a as an
HIV-1 maturation inhibitor specific for the inhibition of p25 cleavage. Analysis of the entire Western
20

blot (supplementary Figure 1) shows that 9a is specific for inhibiting p25 cleavage into p24, as there was
no indication of inhibition of cleavage at any other sites of the protein.
Figure 3. Inhibition of HIV-1 CA-SP1 processing by 1 and 9a. Lanes are numbered 1-8 (top).
Compound 1 was tested at 30 through 0.03 M (3-fold dilutions, lanes 1-8); compound 9a was tested at
2 M (lane 10); C (control, lanes 8 and 9) show samples with buffer only control added. All samples
were run on the identical gel, but lane 10 is offset from the others in the Figure due to gel effects.
Additionally, 9a was evaluated in a competitive binding assay. This assay measures the ability
of a target compound to compete with a radiolabeled derivative of 1 (the [3H]-dihydro 1)43 for binding to
purified VLPs. The results show that 9a binds competitively to wt VLPs with an IC50 of 27 nM (Figure
4).
21

Figure 4. Competitive binding of 9a to wt VLPs as measured by displacement of [3H]-dihydro 1.
The betulinic acid-derived analog 9a and the betulin-derived analog 16a were advanced into in
vitro assays designed to forecast ADME (absorption, distribution, metabolism and excretion) properties.
Both 9a and 16a are predicted to be low clearance compounds in the rat and in humans based on their
stability upon incubation with rat and human liver microsomes in the presence of NADPH (t1/2 >120 min
for both compounds in both species). An in vitro assessment of membrane permeability could not be
conducted due to the low solubility of these compounds. Nevertheless, both 9a and 16a were selected
for in vivo pharmacokinetic evaluation in snapshot experiments conducted in rats where comparison was
made with 1 as the reference standard. The results are presented in Table 2.
22

Table 2. In vivo pharmacokinetic properties of compounds 1, 9a and 16a in rats.a
CL
Cmpd F(%) AUC 6h (nM*hr)b Cmax (nM)b
tmax (h)
Vss IV (L/Kg)
iIV(mL/min/kg)
1
18
4
6,390
5,724
2,136
2,587
1,314
590
0.3
4
4.0
1.0
2.1
0.1
0.2
0.5
9a
16a
4
5
a
Vehicle: polyethylene glycol 300 (PEG 300), ethanol and tween-80 (TW80)(89:10:1 v/v/v), n=2;
Dose 5 mg/kg PO; 1 mg/kg iv; b Oral Cmax and AUC adjusted to 5 mg/kg
In the rat PK study, 9a displayed lower clearance (1.0 vs. 4.0 mL/min/kg), a smaller volume of
distribution (Vss = 0.1 vs. 0.2) and a longer tmax (4 vs. 0.3 h) when compared to 1. The oral exposure of
both compounds was comparable (AUC = 5,724 vs. 6,390 nM*hr), although 9a had lower oral
bioavailability (F= 4% compared to 18% for 1). In contrast, the less polar betulin-based derivative 16a
demonstrated improved clearance (2.1 mL/min/kg) compared to 1 but lower oral exposure than either 1
or 9a. The Cmax for both 9a and 16a were lower than that of 1 (1,314 and 590 nM, respectively,
compared to 2,587 nM for 1).
CONCLUSION
Overall, the antiviral potency and pharmacokinetic profile of 9a was comparable that of the
prototype 1, demonstrating that the 4-benzoic acid motif introduced at the C-3 position of BA acts as a
suitable replacement for the dimethyl succinic acid side chain present in previous BA-derived MIs. This
23

new C-3 substituent illuminates topographical aspects of the HIV-1 MI pharmacophore by constraining
the spatial relationship between the aryl carboxylic acid and the triterpenoid core and, by extension, the
C-28 substituent, which can be further modified. The new analogs, in general, exhibit excellent in vitro
antiviral activity against wt virus, with more modest potency changes in the presence of HSA (0.8 to
9.4-fold) than 1, indicative of reduced serum albumin binding. However, no significant improvement in
potency against Gag V370A and/or solubility were observed for the C-3 benzoic acid series when
compared with 1, with the sole exception of 9u. Although 1 was advanced into Phase IIb clinical trials,
its development was halted due to a poor antiviral responses in a significant percentage of the patient
population, a liability traced to a polymorphic variations proximal to the purported binding site of the
drug. In addition, high protein binding and difficulties in formulating 1 further complicated its
development.21 A successful second generation MI will need to be able to achieve efficacy against the
polymorphic viruses that show diminished antiviral susceptibility to 1, experience a reduced effect of
serum on potency and demonstrate improved physicochemical properties so as to better facilitate
improved pharmacokinetic profiles. Compound 9a exemplifies an advance toward that objective,
providing a highly potent chemotype that exhibits a lower serum shift in the presence of HSA, and
represents the initial effort towards finding a second generation MI.32,44 Further developments from this
novel 4-benzoic acid deoxybetulinic acid chemotype that show enhanced activity toward polymorphic
viruses, and improved physicochemical properties as the result of modification of the C-28 carboxylic
acid moiety, will be reported in due course.32
EXPERIMENTAL SECTION
Unless otherwise stated, solvents and reagents were used directly as obtained from commercial sources,
and reactions were performed under a nitrogen atmosphere. Flash chromatography was conducted on
24

Silica gel 60 (0.040-0.063 particle size; EM Science supply). Purification was accomplished by medium
pressure liquid chromatography on a Biotage flash chromatography with normal phase silica gel, or by
preparative reverse phase (rp) HPLC using MeCN or MeOH-H2O-TFA or NH4OAc as mobile phases.
Preparative rp HPLC was performed using two Shimadzu LC-8A pumps, a SPD-10AV UV-vis detector
set at 220 nm and an ELSD are used. HPLC purity analyses were performed using one of the following
conditions: Method A: column, Ascentis 2.7 m C18, 4.6 mm × 50 mm; 90% (100% H2O/10 mM
NH4OAc to 95% (100% MeCN). Method B: column, Supelco Acentis 2.7 m C18, 4.6 x 75 mm; 70%
(100% H2O/10 mM NH4OAc) to 95% (100% MeCN/10 mM NH4OAc); 6 min gradient; 1.5 mL/min
flow rate. Method C: column, Supelco Acentis 2.7 m C18, 4.6 x 50 mm; 70% (100% H2O/10 mM
NH4OAc) to 95% (100% MeCN/10 mM NH4OAc); 5 min gradient; 2 mL/min flow rate. Method D:
column, Supelco Acentis 2.7 m C18, 4.6 x 50 mm; 70% (100% H2O/10 mM NH4OAc) to 95% (100%
MeCN/10 mM NH4OAc); 6 min gradient; 2 mL/min flow rate. Method E: column, Waters Xbridge 5
m C18, 4.6 mm × 50 mm; 30% (100% H2O/10 mM NH4OAc) to 95% (100% MeOH/10 mM
NH4OAc); 5 min gradient; 1.2 mL/min flow rate. Method F: column, Phenomenex Luna 5 m C5, 4.6
mm × 150 mm; 80% (100% H2O/10 mM NH4OAc) to 95% (100% MeOH/10 mM NH4OAc); 12 min
gradient; 1 mL/min flow rate. Method G: column, Sunfire 5 m C18, 4.6 mm × 30 mm; 95% H2O/5%
MeOH/10 mM NH4OAc to 5% H2O/95% MeOH/10 mM NH4OAc; 15 min gradient; 4 mL/min flow
rate. Method H: column, Sunfire 3.5 m C18, 4.6 mm × 150 mm; 90% (95% H2O/5% MeCN/0.1%
TFA) to 100% (5% H2O/95% MeCN/0.1 TFA); 15 minute gradient; 2 mL/min flow rate. Method I:
column, Xbridge Phenyl 3.5 m, 4.6 mm × 150 mm; 90% (95% H2O/5% MeCN/0.1% TFA) to 100%
(5% H2O/95% MeCN/0.1 TFA); 15 minute gradient; 2 mL/min flow rate. Method J: column, Xbridge
C18 3.5 m, 4.6 mm × 150 mm; 90% (95% H2O/5% MeOH/10 mM NH4HCO3) to 100% (5% H2O/95%
MeOH/10 mM NH4HCO3); 15 minute gradient; 2 mL/min flow rate. Method K: column, Xbridge
Phenyl 3.5 m, 3.0 mm × 150 mm; 90% (95% H2O/5% MeCN/0.1% TFA) to 100% (5% H2O/95%
25

MeCN/0.1 TFA); 15 minute gradient; 1 mL/min flow rate. Method L: column, Xbridge C18 3.5 m, 3.0
mm × 150 mm; 90% (95% H2O/5% MeOH/10 mM NH4HCO3) to 100% (5% H2O/95% MeOH/10 mM
NH4HCO3); 15 minute gradient; 1 mL/min flow rate. All Liquid Chromatography (LC) data were
recorded on a Shimadzu LC-10AS liquid chromatograph using a SPD-10AV UV-Vis detector with Mass
Spectrometry (MS) data determined using a Micromass Platform for LC in electrospray mode. LCMS
analyses were performed using one of the following conditions: Method 1: column, Phenomenex Luna
S10, 3.0 mm × 50 mm; 95% H2O/5% MeCN/10 mM NH4OAc to 5% H2O/95% MeCN/10 mM
NH4OAc; 4 min gradient; 2 mL/min flow rate. Method 2: column, Supelco Acentis 2.7 m C18, 4.6 x
50 mm; 70% (100% H2O/10 mM NH4OAc) to 95% (100% MeCN/10 mM NH4OAc); 5 min gradient; 2
mL/min flow rate. Method 3: column, Supelco Acentis 2.7 m C18, 4.6 x 75 mm; 70% (100% H2O/10
mM NH4OAc) to 95% (100% MeCN/10 mM NH4OAc); 6 min gradient; 1.5 mL/min flow rate. Method
4: column, Supelco Acentis 2.7 m C18, 4.6 x 50 mm; 70% (100% H2O/10 mM NH4OAc) to 95%
(100% MeCN/10 mM NH4OAc); 6 min gradient; 2 mL/min flow rate. Method 5: column, Phenomenex
Luna 3 m C18, 3.0 mm × 50 mm; 90% H2O/10% MeOH/0.1% TFA to 10% water/90% MeOH/0.1%
TFA; 2 min gradient; 1 mL/min flow rate. Method 6: column, Ascentis 2.7 m C18, 4.6 mm × 50 mm;
90% (100% H2O/10 mM NH4OAc to 95% (100% MeCN); 7 min gradient; 2 mL/min flow rate. Method
7: column, Phenomenex Luna S10, 3.0 mm × 50 mm; 95% H2O/5% MeOH/10 mM NH4OAc to 5%
H2O/95% MeOH/10 mM NH4OAc; 2 min gradient; 4 mL/min flow rate. Method 8: column,
Phenomenex Luna 5 m C5, 4.6 mm × 150 mm; 80% (100% H2O/10 mM NH4OAc) to 95% (100%
MeOH/10 mM NH4OAc); 12 min gradient; 1 mL/min flow rate. Method 9: column, Waters Xbridge 5
m C18, 4.6 mm × 50 mm; 30% (100% H2O/10 mM NH4OAc) to 95% (100% MeOH/10 mM
NH4OAc); 5 min gradient; 1.2 mL/min flow rate. Method 10: column, Phenomenex Luna 3 m C18,
3.0 mm × 50 mm; 90% H2O/10% MeCN/0.1% TFA to 10% H2O/90% MeCN/0.1% TFA; 2 min
gradient; 1 mL/min flow rate. Method 11: column, Waters Xbridge Phenyl S5, 4.6 mm × 50 mm; 80%
26

(90% H2O/10%MeOH/0.1%TFA) to 10% H2O/90% MeOH/0.1% TFA; 1 min gradient; 4 mL/min flow
rate. High-resolution mass spectra were recorded on one of the following mass spectrometers: a LCT,
1
TOF, or a LTQ-Orbitrap instrument. H NMR spectra were recorded on Bruker DRX-500f at 500 MHz
(or Bruker AV 400 MHz, Bruker DPX-300B or Varian Gemini 300 at 300 MHz as stated). The
chemical shifts were reported in ppm on the  scale relative to  TMS = 0. The following internal
references were used for the residual protons in the following solvents: CDCl3 (H 7.26), CD3OD (H
3.30), Acetic-d4 (Acetic Acid d4) (H 11.6, 2.07), DMSOmix or DMSO-D6_CDCl3 ((H 2.50 and 8.25)
(ratio 75%:25%)), and DMSO-D6 (H 2.50). Standard acronyms were employed to describe the
multiplicity patterns: s (singlet), br. s (broad singlet), d (doublet), t (triplet), q (quartet), m (multiplet), b
(broad), app (apparent). The coupling constant (J) is in Hertz.
(1R,3aS,5aR,5bR,7aR,9S,11aR,11bR,13aR,13bR)-Benzyl
9-hydroxy-5a,5b,8,8,11a-pentamethyl-1-
(prop-1-en-2-yl)icosahydro-1H-cyclopenta[a]chrysene-3a-carboxylate (5). To a suspension of 3 (12 g,
26.3 mmol) and K2CO3 (7.26 g, 52.6 mmol) in DMF (150 mL) was added BnBr (3.28 mL, 27.6 mmol).
The mixture was heated to 60 °C for 3.5 h, and cooled to rt. A solid started to precipitate upon cooling.
The mixture was diluted with H2O (200 mL) and the solid formed was collected by filtration to give 5 as
1
a white solid (13.92 g, 97 %). H NMR (500 MHz, CDCl3)  7.48 - 7.27 (m, 5H), 5.26 - 5.03 (m, 2H),
4.71 (d, J = 1.83 Hz, 1H), 4.58 (s, 1H), 3.17 (dt, J = 11.4, 5.6 Hz, 1H), 3.03 (td, J = 10.99, 4.88 Hz, 1H),
2.27 (ddd, J = 12.36, 3.20, 3.05 Hz, 1H), 2.17 (td, J = 12.1, 3.5 Hz, 1H), 1.96 - 1.78 (m, 2H), 1.73 – 0.96
(m, 18H), 1.68 (s, 3H), 0.96 (s, 3H), 0.95 (s, 3H), 0.88 (td, J = 13.0, 4.1 Hz, 1H), 0.80 (s, 3H), 0.77 (s,
3H), 0.76 (s, 3H), 0.65 (d, J = 9.2 Hz, 1H).
(1R,3aS,5aR,5bR,7aR,11aR,11bR,13aR,13bR)-Benzyl 5a,5b,8,8,11a-pentamethyl-9-oxo-1-(prop-1-en-
2-yl)icosahydro-1H-cyclopenta[a]chrysene-3a-carboxylate (6). To a solution of 5 (7.1 g, 12.98 mmol)
in CH2Cl2 (100 mL) was added PCC (4.20 g, 19.48 mmol). After stirring for five minutes, the mixture
turned a deep crimson color. The stirring was continued for 5.5 h. The mixture was filtered through a
27

pad of Celite and silica gel which was washed with CH2Cl2 and then a 1:1 mixture of EtOAc:hexanes.
1
The filtrate was concentrated under reduced pressure to give 6 as a white foam (6.92 g, 98 %). H NMR
(500 MHz, CDCl3)  7.41 - 7.28 (m, 5H), 5.20 - 5.06 (m, 2H), 4.72 (d, J = 2.1 Hz, 1H), 4.59 (d, J = 1.2
Hz, 1H), 3.01 (td, J = 10.99, 4.88 Hz, 1H), 2.51 - 2.43 (m, 1H), 2.42 - 2.34 (m, 1H), 2.28 (dt, J = 12.59,
3.17 Hz, 1H), 2.21 (td, J = 12.28, 3.51 Hz, 1H), 1.94 - 1.82 (m, 3H), 1.74 -1.69 (m, 1H), 1.50 – 1.17 (m,
14H), 1.69 (s, 3H), 1.11 (dt, J = 13.4, 2.9 Hz, 1H), 1.07 (s, 3H), 1.03 - 0.97 (m, 1H). 1.02 (s, 3H), 0.96
(s, 3H), 0.91 (s, 3H), 0.80 (s, 3H).
(1R,3aS,5aR,5bR,7aR,11aR,11bR,13aR,13bR)-Benzyl 5a,5b,8,8,11a-pentamethyl-1-(prop-1-en-2-yl)-
9-(trifluoromethylsulfonyloxy)-2,3,3a,4,5,5a,5b,6,7,7a,8,11,11a,11b,12,13,13a,13b-octadecahydro-1H-
cyclopenta[a]chrysene-3a-carboxylate (7). To a solution of 6 (6.9 g, 12.67 mmol) and 1,1,1-trifluoro-
N-phenyl-N-(trifluoromethylsulfonyl)methanesulfonamide (9.05 g, 25.3 mmol) in THF (200 mL) at -78
°C was added KHMDS (50.7 mL, 25.3 mmol) slowly. The reaction mixture was stirred for 1 h at -78 °C.
TLC indicated starting material was consumed and desired product was formed. The reaction mixture
was quenched with brine (100 mL) and extracted with Et2O (3 x 100 mL). The extracts were dried over
Na2SO4, filtered and concentrated under reduced pressure. The residue was dissolved in toluene and
purified by flash chromatography using 2-10% toluene/hexanes and 5-10% ethyl EtOAc/hexanes to
provide the desired product 7 as a white solid (7.36 g, 86%). 1H NMR (500 MHz, CDCl3)  7.45 - 7.32
(m, 5H), 5.56 (dd, J = 6.71, 1.53 Hz, 1H), 5.22 - 5.06 (m, 1H), 5.12 - 5.05 (m, 1H), 4.74 (d, J = 2.14 Hz,
1H), 4.61 (dd, J = 1.37, 2.29 Hz, 1H), 3.04 (td, J = 10.99, 4.58 Hz, 1H), 2.31 (dt, J = 12.7, 3.2 Hz, 1H),
2.22 (td, J = 12.21, 3.36 Hz, 1H), 2.17 (dd, J = 17.09, 6.71 Hz, 1H), 2.00 - 1.81 (m, 2H), 1.80 - 1.72 (m,
2H), 1.69 (s, 3H), 1.52 - 1.25 (m, 12H), 1.25 - 1.19 (m, 1H), 1.16 (t, J = 3.1 Hz, 1H), 1.12 (s, 3H), 1.01
(s, 3H), 1.08 - 1.03 (m, 1H), 0.96 (s, 3H), 0.89 (s, 3H), 0.79 (s, 3H). 13C NMR (127 MHz, CDCl3) :
175.5, 155.1, 150.2, 136.2, 128.2, 128.0, 127.8, 113.5, 109.4, 65.5, 56.3, 53.0, 49.1, 48.7, 46.7, 42.1,
28

40.3, 40.0, 38.0, 37.6, 36.6, 36.0, 33.0, 31.8, 30.3, 29.3, 27.2, 25.3, 21.2, 19.2, 19.1, 18.7, 15.9, 15.2,
14.4.
General method for the preparation of compounds 9a-9i, 9k-9q and 9s-9u.
Step 1: Suzuki coupling. To a sealable vial containing 7 (1.0 equiv.) was added the corresponding
boronic acid (1.1-1.5 equiv.), Na2CO3 (5 equiv.) and Pd(PPh3)4 (0.03-0.1 equiv.). The mixture was
diluted with either a mixture of DME:water (1:0.5-1.0), or a mixture of 1,4-dioxane:water (1:0.5-1.0) to
a concentration of 0.015 M-0.074 M. The vial was flushed with N2, sealed, and heated to 85 °C-100 °C.
After 1.5 -24 h of heating, the mixture was cooled to rt. The mixture was diluted with either sat. NH4Cl,
1N HCl, or water and extracted with CH2Cl2 or EtOAc. The combined organic layers were dried with
sodium sulfate. The drying agent was removed by filtration and the filtrate was concentrated under
reduced pressure. The residue was either used directly in the next step, or was purified by flash
chromatography to afford the expected cross-coupling product which was used in the next step.
Step 2: Deprotection of carboxylic acid (9a-9i, 9k-9q and 9s-9u). A mixture of the corresponding
Suzuki coupling product from the previous step (1.0 equiv.) and 10% Pd/C (0.1-1 equiv.) in either
EtOAc or a mixture of EtOAc: MeOH (1:0.5-1.0) to a concentration of 0.01M - 0.06 M was stirred
under 1 atm of H2 for 1-24 h. The reaction mixture was filtered and the white solid formed was
collected. The solids were purified by prep. HPLC to provide the desired products 9a-9i, 9k-9q and 9s-
9u as white solids.
(1R,3aS,5aR,5bR,7aR,11aS,13aR,13bR)-9-(4-Carboxyphenyl)-5a,5b,8,8,11a-pentamethyl-1-(prop-1-
en-2-yl)-2,3,3a,4,5,5a,5b,6,7,7a,8,11,11a,11b,12,13,13a,13b-octadecahydro-1H-
cyclopenta[a]chrysene-3a-carboxylic acid (9a). The title compound 9a was prepared following general
method for the preparation of compounds 9a-9i, 9k-9q and 9s-9u described above using 4-
boronobenzoic acid as the reactant boronic acid. The product was isolated as a white solid (10 mg,
29