A cassette-dosing approach for improvement of oral bioavailability of dual TACE/MMP inhibitors

Article abstract of DOI:10.1016/j.bmcl.2006.02.042

The structural features contributing to the different pharmacokinetic properties of the TACE/MMP inhibitors TNF484 and Trocade^TM were analyzed using an in vivo cassette-dosing approach in rats. This enabled us to identify a new lead compound with excellent pharmacokinetic properties, but weaker activity on the biological targets. Directed structural modifications maintained oral bioavailability and restored biological activity, leading to a novel compound almost equipotent to TNF484 in vivo, but with a more than tenfold higher oral bioavailability.

Full text of DOI:10.1016/j.bmcl.2006.02.042

Bioorganic & Medicinal Chemistry Letters 16 (2006) 2632–2636
A cassette-dosing approach for improvement of oral
bioavailability of dual TACE/MMP inhibitors
Philipp Janser,* Ulf Neumann, Wolfgang Miltz, Roland Feifel and Thomas Buhl
Novartis Institutes for BioMedical Research, CH-4002 Basel, Switzerland
Received 23 January 2006; revised 14 February 2006; accepted 14 February 2006
Available online 3 March 2006
Abstract—The structural features contributing to the different pharmacokinetic properties of the TACE/MMP inhibitors TNF484
and TrocadeTM were analyzed using an in vivo cassette-dosing approach in rats. This enabled us to identify a new lead compound
with excellent pharmacokinetic properties, but weaker activity on the biological targets. Directed structural modifications main-
tained oral bioavailability and restored biological activity, leading to a novel compound almost equipotent to TNF484 in vivo,
but with a more than tenfold higher oral bioavailability.
Ó 2006 Elsevier Ltd. All rights reserved.
Matrix metalloproteases (MMPs) and the closely related
family of a Disintegrin and Metalloprotease (ADAM)
enzymes have attracted considerable interest as drug
targets for the treatment of arthritis, periodontal dis-
ease, tumor metastasis, tumor growth, aneurysm, and
atherosclerosis.1–3 We recently described a series of b-ar-
yl-succinic acid hydroxamates as highly potent dual
inhibitors of TACE and matrix metalloproteinases.4
Notably the key compound (2S,3R)-N4-((S)-2,2-dimethyl-
1-methylcarbamoyl-propyl)-N1-hydroxy-2-hydroxy-
out to determine structural features that had either ben-
eficial or deleterious effects on the pharmacokinetic
properties of TrocadeTM and TNF484, respectively.
In an effort to establish a structure–PK relationship be-
tween these two compounds, we designed the three chi-
meras 1, 2, and 3 of TNF484 and TrocadeTM. The
synthesis of TrocadeTM was published by Broadhurst
et al.10 and our group has reported the process develop-
ment of TNF484.11 Accordingly, all our inhibitors were
prepared following this methodology. The key step was
the diastereoselective Lewis acid promoted Claisen–Ire-
land rearrangement12 shown in the general Scheme 1.
methyl-3-(4-methoxy-phenyl)-succinamide
(TNF484)
potently inhibited the release of the pro-inflammatory
cytokine TNFa from cells as well as the LPS-induced
systemic TNFa release in rats. This compound also
showed in vivo activity in models of airway inflamma-
tion and pneumococcal meningitis.5,6 Despite this
impressive in vivo activity, its poor pharmacokinetic
(PK) profile with an oral bioavailability (F) of approxi-
mately 3% in rats remained a major concern. Several
other hydroxamate-type inhibitors, for example,
Marimastat or TrocadeTM, underwent clinical trials for
the indications of cancer and rheumatoid arthritis.7,8
TrocadeTM is structurally rather distinct from TNF484
(see Fig. 1) and has a reported absolute oral bioavail-
ability of 26% in rats, as well as good tolerability and
pharmacokinetics in arthritis patients.9 Thus, we set
The Claisen–Ireland rearrangement produced the race-
mic mixtures with erythro/threo ratios of typically 10:1
or better. The enantiomerically pure acids were obtained
by crystallization with S-(À)-1-phenyl-ethylamine. After
the coupling with appropriate amines, the olefins were
subjected to ozonolysis followed by reductive workup
with dimethylsufide. The crude aldehydes were used
without further purification and oxidized to the carbox-
ylic acids by sodium chlorite. Finally, the acids were
activated with morpholinoethyl isocyanide (MEI) and
2-hydroxy-pyridine-N-oxide (HOPO) and then coupled
with TMS-protected hydroxylamine. Simple hydrolysis
with water gave the desired hydroxamic acids in good
yields.
Keywords: TNFa; TACE; MMP; Metalloproteinase; Dual; Inhibitor;
Pharmacokinetics; Cassette-dosing; Bioavailability; Oral; Clearance;
Absorption; Rheumatoid arthritis; Trocade; TNF484; Claisen–Ireland
rearrangement.
Assessing the pharmacokinetics in a timely fashion and
with a high throughput13 was of key importance for this
project. In our hands, cassette-dosing14 in conscious,
*
Corresponding author. E-mail: philipp.janser@novartis.com
0960-894X/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2006.02.042

P. Janser et al. / Bioorg. Med. Chem. Lett. 16 (2006) 2632–2636
2633
O
O
N
HO
N
H
O
O
O
O
O
H
N
O
H
N
HO
N
N
H
HO
N
H
O
N
H
N
H
N
OH
O
O
HO
Marimastat
TrocadeTM
TNF484
O
O
O
O
O
O
H
N
HO
N
HO
N
H
N
N
H
H
O
O
O
O
O
HO
N
N
N
N
H
O
N
O
N
O
HO
1
2
3
O
Cl
O
O
O
O
HO
N
HO
HO
N
N
O
N
N
H
N
H
H
O
O
O
O
NH
O
NH
HO
NH
5
6
4
Figure 1. Chemical structures of TACE/MMP-inhibitors.
permanently cannulated rats15 proved to be a very use-
ful and reliable tool which guided our structural modifi-
cations toward inhibitors with considerably improved
pharmacokinetic properties.
the higher bioavailability of TrocadeTM did not lead to
a much higher maximal blood concentration (Cmax
)
which can be explained by the relatively high clearance
(CL). Apart from the oral bioavailability, the most pro-
nounced differences between TrocadeTM and TNF484
were the low clearance and the small volume of distribu-
tion (VSS) of the latter. This is typical for polar and
hydrophilic compounds like TNF484 which have a high
polar surface area (PSA) and a low clogP. The combi-
nation of a low clearance with low oral bioavailability
suggests that the latter is limited by a poor absorption
rate. Replacing the hydroxymethyl group in TNF484
by the hydantoin residue in 1 further increased the polar
surface area together with a slight increase in clogP.
This did not significantly affect the bioavailability, but
it led to a further reduction of Cmax, mainly due to a
higher clearance. Replacement of the tert-butylglycine
amide group by piperidine in 2 had a quite dramatic
influence on the overall pharmacokinetic profile. Due
to the higher lipophilicity, both clearance and volume
of distribution were significantly higher but the absorp-
tion rate improved, which resulted in a better bioavail-
ability of 13%. An even bigger effect was observed for
3 (F = 36%), where the hydantoin residue was replaced
with the original hydroxymethyl group present in
TNF484. We attribute this substantial improvement
mainly to a good absorption rate, combined with a
slightly lower clogP and polar surface area, leading to
a more than threefold reduction in clearance compared
to TrocadeTM. This is also mirrored in the high Cmax
of 253 nM (dose-normalized to 1 mg/kg). To our
The results of our first cassette-dosing experiment are
summarized in Table 1. For comparison, we have also
included the pharmacokinetic profile of TNF484 when
administered as single compound (last entry). The com-
parable outcome supports again the reliability of
cassette-dosing within a series of similar compounds.
We could confirm both, the low (3%) and the medium
(32%) oral bioavailabilitiy (F) of TNF484 and
TrocadeTM, respectively. Interestingly, we found that
X
Cl
O
(a)
(b)
R
1
R
N
O
OH
X
2
R
O
O
R
R
X
X
(d)
(c)
1
1
O
O
R
N
R
N
HO
2
2
HO
N
R
R
H
O
O
R
R
Scheme 1. Synthesis of succinic hydroxamates. Reagents: (a)
LiHMDS, TMS-Cl, TiCl4, THF; (b) HNR1R2, BOP-Cl, NEt3,
CH2Cl2; (c) i—ozone, MeOH; ii—Me2S; iii—NaClO2, NaH2PO4, t-
BuOH, water; (d) i—MEI, HOPO, CH2Cl2; ii—TMSONH2;
iii—water.

2634
P. Janser et al. / Bioorg. Med. Chem. Lett. 16 (2006) 2632–2636
Table 1. Pharmacokinetic parameters (means SEM) calculated from plasma levels after iv (1 mg/kg each) and po (3 mg/kg each) administration of
a cocktail of five compounds to conscious rats (n = 4)
Compound
CL (mL/min/kg)
VSS (L/kg)
Cmax poa (nM)
F (%)
clogP
PSA
TNF484
TrocadeTM
9.2 ( 0.3)
67.8 ( 3.5)
18.5 ( 0.9)
84.9 ( 7.6)
18.8 ( 0.4)
10.0 ( 1.1)
0.37 ( 0.02)
2.31 ( 0.19)
0.48 ( 0.04)
2.45 ( 0.28)
1.08 ( 0.03)
0.34 ( 0.03)
31.7 ( 6.7)
58.3 ( 10.3)
19.0 ( 7.0)
28.3 ( 8.0)
253.3 ( 44.3)
42.7 ( 18.3)
2.8 ( 0.4)
31.6 ( 6.8)
3.2 ( 0.8)
13.0 ( 2.0)
35.9 ( 5.2)
3.2 ( 0.7)
0.62
2.12
0.85
1.51
1.36
0.62
137.0
110.3
157.4
119.5
99.1
1
2
3
TNF484b
137.0
Results from an identically designed experiment where TNF484 was administered as a single compound are shown for reasons of comparison.
a Mean from individual animals, dose-normalized to 1 mg/kg.
b Dosed as a single compound.
surprise, the anticipation that the hydroxymethyl group
would be a metabolic weak point (e.g., oxidation,
glucuronidation, sulfation, etc.) was not confirmed,
and thus, we could identify 3 as a new lead compound
with excellent pharmacokinetic properties, even better
than TrocadeTM (cf. Fig. 2).
Table 2. In vitro potencies (measured in duplicate) for compounds
Compound MMP-1a MMP-3a MMP-13a TACEa HPBMCb
Ki (nM) Ki (nM) Ki (nM)
Ki (nM) IC50 (nM)
TNF484
TrocadeTM
1
0.9
56
0.5
1
0.6
230
48
>10,000
484
0.3
0.7
1
2
3
4
5
6
12
187
300
3
19
57
321
66
8
83
>10,000
>10,000
>10,000
836
Unfortunately, 3 was much less potent than TNF484 at
inhibiting both the recombinant enzymes as well as the
cellular TNFa release from human peripheral blood
mononuclear cells (HPBMCs), as can be seen in
Table 2. By comparison of the two pairs 3/TNF484
and 1/2 it is obvious that the removal of the tert-butyl-
glycine amide unit leads to improved pharmacokinetic
properties, but at the same time is responsible for the
loss of activity. Therefore, our further efforts were aimed
at restoring the amide group to recover the activity,
while maintaining the favorable PK properties.
213
23
55
10
7
n.d.
25
11
5
8
0.7
2
5
316
a Determined by fluorimetric assay using recombinant enzymes.16
b Human peripheral blood mononuclear cells, stimulated with LPS/c-
interferon.4
Not unexpectedly, the quite polar inhibitor 4 still had an
unfavorable pharmacokinetic profile, again indicating
poor absorption. In contrast, the more lipophilic ethyl
derivative 5 resembled more the profile of 3, but this
structural modification also increased the clearance
which could be responsible for the lower bioavailability
of only 20%.
First, we replaced the piperidine in 3 with pipecolic
amides (cf. 4). Despite a tenfold increase in potency on
TACE, the cellular activity was still >10 lM. Introduc-
tion of an ethyl substituent instead of the hydroxymeth-
yl group in 5 did not change TACE or MMP inhibition
but improved the IC50 to 836 nM in the cellular TNFa
release. Finally, we replaced the methoxy-phenyl
substituent by chloro-phenyl and obtained with
compound 6 a TACE inhibitor with a cellular IC50 of
316 nM. The PK properties of these compounds are
summarized in Table 3.
A further increase in lipophilicity and a reduction of po-
lar surface area, accompanied with the elimination of a
possible metabolic weak point by replacing the p-meth-
oxy group in 5 with a p-chloro substituent in 6, proved
to be highly beneficial. Compound 6 had a well-bal-
anced pharmacokinetic profile, that is, moderate clear-
ance, medium volume of distribution, and a good
bioavailability of 65.7%. A further interesting aspect
was the high maximal blood concentration of 452 nM
(dose-normalized to 1 mg/kg). This result motivated us
to perform a single compound administration experi-
ment with compound 6 (last entry in Table 3). With
the exception of the higher volume of distribution (indi-
cating good tissue penetration) we could confirm the
cassette-dosing study. A comparable oral bioavailability
of 47% and a maximal blood concentration of 389 nM
were found.
0.8
0.6
0.4
0.2
0.0
With this profile, we decided to test 6 in the LPS-in-
duced systemic TNFa release model in rats (see
Fig. 3).17
0
30 60
120
240
360
420
Time (min)
At the four doses tested (1, 3, 10, and 30 mg/kg) it inhib-
ited TNFa release into plasma by 29%, 53%, 70%, and
94%, respectively. From these data we calculated an
Figure 2. Plasma level time courses in rats (n = 4) for: s, TNF484; j,
TrocadeTM; h, 3. Shown are means SEM.

P. Janser et al. / Bioorg. Med. Chem. Lett. 16 (2006) 2632–2636
2635
Table 3. Pharmacokinetic parameters (means SEM) calculated from plasma levels after iv (1 mg/kg each) and po (3 mg/kg each) administration to
conscious rats (n = 4)
Compound
CL (mL/min/kg)
VSS (L/kg)
Cmax poa (nM)
F (%)
clogP
PSA
4b
5
23.2 ( 1.0)
68.0 ( 2.6)
43.3 ( 2.3)
28.1 ( 2.6)
1.06 ( 0.05)
5.95 ( 0.41)
2.25 ( 0.13)
16.67 ( 2.37)
38.1 ( 11.0)
48.0 ( 15.0)
452 ( 67)
8.4 ( 1.7)
19.9 ( 2.7)
65.7 ( 9.6)
47.1 ( 8.3)
0.75
1.29
2.08
2.08
128.2
108.0
98.7
6
6b
388.6 ( 60.7)
98.7
Inhibitors 5 and 6 were dosed as a cocktail of five compounds, whereas 4 was administered as single compound. Results from an identically designed
experiment where 6 was administered as a single compound are shown for reasons of comparison.
a Mean from individual animals, dose-normalized to 1 mg/kg.
b Dosed as a single compound.
120
Acknowledgments
100
80
60
40
20
0
n.s.
We thank R. Carton, V. Stadelmann, H. Kaiser,
D. Pralet, I. Brzak, B. Meyer, R. Nagele, J.-L. Runser,
R. Endres, D. Lehmann, and H. Zihlmann for the skill-
ful technical assistance in compound synthesis, in vitro
and in vivo biology, and HPLC/MS analytics.
Supplementary data
Supplementary data associated with this article can be
found, in the online version, at doi:10.1016/
j.bmcl.2006.02.042.
control
1
3
10
30
Dose of (mg/kg)
Figure 3. Release of TNFa into plasma of rats (n = 4–7) treated orally
with four different doses of 6 compared to vehicle alone (=control).
Shown are means SEM; n.s., not significant; *p < 0.05; **p < 0.01;
***p < 0.001.
References and notes
1. Risto, A.; Veli-Matti, K. Biochimie 2005, 87, 273.
2. Handsley, M. M.; Edwards, D. R. Int. J. Cancer 2005,
115, 849.
3. Tayebjee, M. H.; Lip, G. Y. H.; MacFadyen, R. J. Curr.
Med. Chem. 2005, 12, 917.
ED50 of 1.5 mg/kg, which is almost equipotent with
TNF484 which had an ED50 of 1 mg/kg in the same
model.4
4. Kottirsch, G.; Koch, G.; Feifel, R.; Neumann, U. J. Med.
Chem. 2002, 45, 2289.
Poor oral bioavailability and its associated high var-
iability in exposure is a frequent cause for failure
in clinical drug development. We have shown that
cassette-dosing was a useful and reliable tool to as-
sess the pharmacokinetics of a series of TACE/
MMP-inhibitors in the rat. We successfully applied
this technique to identify the key structural features
that are responsible for the different pharmacokinetic
properties of TNF484 and TrocadeTM. These findings
proved to be instrumental in our efforts to improve
the poor absolute oral bioavailability (3%) of the po-
5. Trifilieff, A.; Walker, C.; Keller, T.; Kottirsch, G.;
Neumann, U. Br. J. Pharmacol. 2002, 135, 1655.
6. Meli, D. N.; Loeffler, J. M.; Baumann, P.; Neumann, U.;
Buhl, T.; Leppert, D.; Leib, S. L. J. Neuroimmunol. 2004,
151, 6.
7. Hemmings, F. J.; Farhan, M.; Rowland, J.; Banken, L.;
Jain, R. Rheumatology 2001, 40, 537.
8. Millar, A. W.; Brown, P. D.; Moore, J.; Galloway, W. A.;
Cornish, A. G.; Lenehan, T. J.; Lynch, K. P. Br. J. Clin.
Pharmacol. 1998, 45, 21.
9. Lewis, E. J.; Bishop, J.; Bottomley, K. M. K.; Bradshaw,
D.; Brewster, M.; Broadhurst, M. J.; Brown, P. A.; Budd,
J. M.; Elliott, L.; Greenham, A. K.; Johnson, W. H.;
Nixon, J. S.; Rose, F.; Sutton, B.; Wilson, K. Br. J.
Pharmacol. 1997, 121, 540.
10. Broadhurst, M. J.; Brown, P. A.; Lawton, G.; Ballantyne,
N.; Borkakoti, N.; Bottomley, K. M. K.; Cooper, M. I.;
Eatherton, A. J.; Kilford, I. R.; Malsher, P. J.; Nixon, J.
S.; Lewis, E. J.; Sutton, B. M.; Johnson, W. H. Bioorg.
Med. Chem. Lett. 1997, 7, 2299.
tent TACE/MMP-inhibitor TNF484.
A new lead
compound (3) could be identified with excellent phar-
macokinetic properties, but significantly weaker
potency in vitro. Further structural modifications
guided by parallel screening for good in vitro potency
and superior biopharmaceutical and pharmacokinetic
properties resulted in the new inhibitor 6 which is
equipotent to TNF484 in vivo, but has a more than
tenfold higher absolute oral bioavailability (47%) in
rats and thus a reduced variability in exposure. Con-
sidering that rat pharmacokinetics is often representa-
tive for the human situation, we anticipate that
inhibitor 6 should have a smaller risk of clinical
failure.
11. Koch, G.; Kottirsch, G.; Wietfeld, B.; Kuster, E. Org.
Process Res. Dev. 2002, 6, 652.
¨
12. Koch, G.; Janser, P.; Kottirsch, G.; Romero-Giron, E.
Tetrahedron Lett. 2002, 43, 4837.
13. Roberts, S. A. Xenobiotica 2001, 31, 557.
14. Manitpisitkul, P.; White, R. E. Drug Discovery Today
2004, 9, 652.

2636
P. Janser et al. / Bioorg. Med. Chem. Lett. 16 (2006) 2632–2636
15. PK cassette-dosing in rats. The experiment was per-
formed in conscious, fed, permanently cannulated rats
kept under standard conditions. For oral administration
by gavage, a mixture of ethanol (3%) and cornoil was
used as a vehicle (5 mL/kg), whereas for the intravenous
route compounds were administered into the femoral
vein as a solution in N-methyl-2-pyrrolidone (30%) in
polyethylene glycol 200 (0.5 mL/kg). Plasma samples of
approximately 70 lL were collected from the femoral
artery for 24 h after iv administration and for 7 h after
oral dosing. iv and po administration was in the same
animals with a 48 h washout period between adminis-
trations. After acetonitrile precipitation of plasma sam-
ples (25 lL), parent drug concentrations in extracts were
measured using a specific HPLC/MS method. Complete
details of the procedure are given as supporting
information.
16. Neumann, U.; Kubota, H.; Frei, K.; Ganu, V.; Leppert,
D. Anal. Biochem. 2004, 328, 66.
17. Rat LPS challenge. Conscious male Sprague–Dawley rats,
kept under standard conditions, were orally dosed by
gavage with aqueous solutions of 6 at doses between 1 and
30 mg/kg (4–7 rats per group; controls receiving vehicle
only were included in the study). The rats were anesthe-
tized 3 h later for the remaining experimental procedure.
The carotid arteries (for blood sampling) and jugular veins
(for LPS injections) were cannulated, and one hour
thereafter a single dose (500 lg/kg) of LPS was given.
One hour later, arterial blood was taken, and plasma was
prepared and, subsequently, was tested for TNFa con-
centrations (ELISA). TNFa inhibition of the compound-
treated groups was calculated as a percentage of controls
treated with vehicle only. Additionally, an ED50 was
determined.