Application of enzymatically stable dipeptides for enhancement of intestinal permeability. Synthesis and in vitro evaluation of dipeptide-coupled compounds

Article abstract of DOI:10.1016/S0968-0896(01)00066-9

Transport across the intestinal barrier of compounds with low permeability may be facilitated by targeting the human oligopeptide transporter, hPepT1. A flexible synthetic pathway for attaching compounds to dipeptides through ester or amide bonds was deve

Full text of DOI:10.1016/S0968-0896(01)00066-9

Bioorganic & Medicinal Chemistry9 (2001) 2625–2632
Application of Enzymatically Stable Dipeptides for Enhancement
of Intestinal Permeability. Synthesis and In Vitro Evaluation
of Dipeptide-Coupled Compounds
Gerda M. Friedrichsen,a Palle Jakobsen,b Mitchell Taubc and Mikael Begtrupa,*
aDepartment of Medicinal Chemistry, The Royal Danish School of Pharmacy,
Universitetsparken 2, DK-2100 Copenhagen, Denmark
bMedicinal Chemistry Research, Novo Nordisk A/S, Novo Nordisk Park, DK-2760Maaloev, Denmark
cDepartment of Drug Metabolism, Novo Nordisk A/S, Novo Nordisk Park, DK-2760Maaloev, Denmark
Received 17 January2001; accepted 5 March 2001
Abstract—Transport across the intestinal barrier of compounds with low permeabilitymaybe facilitated bytargeting the human
oligopeptide transporter, hPepT1. A flexible synthetic pathway for attaching compounds to dipeptides through ester or amide
bonds was developed. Furthermore, a synthetic approach to functionalize model drugs from one key intermediate was generated
and applied to a glucose-6-phosphatase active model drug. The model drug was coupled to d-Glu-Ala through various linkers, and
the G-6-Pase activityas well as the aqueous solubilityand transport properties of these prodrugs, as compared to those of the
parent drugs, were examined. None of the peptide-coupled compounds seemed to be transported byhPepT1, though one of the
peptide-coupled compounds had affinityfor hPepT1. Interestingly, in one case the parent drug was activelyeffluxed, while the
corresponding peptide-coupled prodrug was not. The low aqueous solubilityof the parent compounds was not increased after
attachment to a dipeptide. This suggests that onlycompounds with a certain intrinsic aqueous solubilityshould be targeted to
hPepT1 byattachment to a dipeptide. Important information about the design of peptide-coupled drugs targeted for hPepT1 is
presented. # 2001 Elsevier Science Ltd. All rights reserved.
Introduction
The overall purpose of the present work was to examine
if the above strategyto increase the permeabilityacross
the intestine could be applied to larger biologically
active molecules with low intestinal permeability. By
attaching the target compound to a dipeptide, which is
recognized byhPepT1, the compound maybe trans-
ported bycarrier-mediated mechanisms across the
intestinal epithelium. Furthermore, attachment of a
dipeptide maylead to increased aqueous solubilityof
the target compound, thus contributing to an overall
increase in absorption. As model compounds we selec-
ted glucose-6-phosphatase (G-6-Pase) inhibitors of the
4-(4-chlorophenyl)-4,5,6,7-tetrahydrothieno3,2-cpyr-
idine-5-yl-phenylmethanone type.7 These compounds
were found to be active as inhibitors of G-6-Pase in in
vitro assays using disrupted microsomes as well as
hepatocytes, but they lacked in vivo activity. The cata-
lytic site of the G-6-Pase enzyme is situated inside the
microsomes, meaning that the compounds have to pass
two membranes to reach their target. Investigations of
the model drugs in the Caco-2 cell line revealed that the
compounds had verylow or no permeability. This may
The human oligopeptide transporter, hPepT1, situated
in the small intestine, is involved in the absorption of
nutrient oligopeptides and transports numerous di- and
tripeptides and peptidomimetics.1,2 Previous studies
have shown that model drugs such as benzyl alcohols
can be coupled to an enzymatically stable dipeptide
which is recognized byhPepT1 and transported across
the intestine via hPepT1.3,4 Benzyl alcohol has been
attached to the side-chain carboxylic acid of dipeptides
such as d-Asp-Ala and d-Glu-Ala, and these model
prodrugs were shown to be relativelyenzmy atically
stable and cleaved during or after transport across the
intestinal epithelium by pH-dependent hydrolysis.5,6 To
our knowledge, no model drugs significantlylarger than
benzyl alcohol have been attached to enzymatically
stable dipeptides with the purpose of targeting hPepT1.
*Corresponding author. Tel.: +45-35-30-60-00; fax: +45-35-30-60-40;
e-mail: begtrup@medchem.dfh.dk
0968-0896/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PII: S0968-0896(01)00066-9

2626
G. M. Friedrichsen et al. / Bioorg. Med. Chem. 9 (2001) 2625–2632
be due to the inherent poor aqueous solubilityof the
compounds or other physicochemical characteristics,
which prevent the compounds from permeating cell
membranes and undergo passsive transcellular trans-
port. Consequently, it seemed worthwhile to investigate
if linkage to a dipeptide could increase permeabilityas
seen with peptide-coupled benzyl alcohol.
were prepared with the purpose of controlling the glu-
cose output from the liver in diabetic patients.7 A gen-
eral structure is presented in Scheme 2. Structure–
activityexperiments had previouslymapped the area
least sensitive to structural changes with respect to the
pharmacological action, and this site, the para position
in the phenyl group attached to carbonyl, was used for
functionalization. The 4-iodo keyintermediate 2 was
synthesized from 1 as outlined in Scheme 2.
We therefore developed a flexible method for introduc-
tion of various linkers into one keystructure of the
model drug. The functionalized model drug was then
attached to the enzymatically stable dipeptide d-Glu-
Ala through ester or amide bonds, and the G-6-Pase
activityas well as the aqueous solubilityand transport
properties of these prodrugs, as compared to those of
the parent drugs, were examined.
Using 2 as a starting point, various linkers were inserted
byhalogen–metal exchange, followed byaddition of an
electrophile or cross-coupling protocols. Halogen–metal
exchange was first attempted using the corresponding 4-
bromo derivative, but lithiation and addition of elec-
trophile gave a complex mixture. In contrast, the 4-iodo
derivative 2 was converted smoothlyto the formyl deri-
vative 3 via a magnesium intermediate,9 and further
reactions provided a series of functionalized compounds
originating from one keystructure ( 2) (Scheme 3).
Results and Discussion
Synthesis of D-Glu-Ala
From the formyl derivative 3 two different compounds
were prepared, to which subsequent coupling to the
dipeptide was possible: A compound with a hydroxy-
methyl linker (4) and a compound with an aminomethyl
linker (7). A derivative with an amino linker (9) was
synthesized from the acetamino derivative 8, which was
obtained in a similar wayas the 4-iodo derivative 2.
Based on previous studies, d-Asp-Ala was the first
dipeptide chosen for attachment to the model drug.3
However, d-Asp-Ala gave rise to problems during
synthesis of the prodrugs, probably due to formation of
an intramolecular cyclic imide.8 Extending the side-
chain in Asp with one methylene group minimized
cyclization, and therefore d-Asp-Ala was replaced by d-
Glu-Ala. This substitution seemed to exert a general
decrease in affinityfor hPepT1, but affinitywas still
retained in the 5 mM range.4
Synthesis of model prodrugs
The functionalized derivatives with a hydroxymethyl, an
aminomethyl or an amino linker were attached to d-
Glu-Ala through amide or ester bonds, respectively. To
minimize the risk of racemization, the coupling reac-
tions were performed without addition of base. The
synthesis of the model prodrugs is outlined in Scheme 4.
d-Glu-Ala was synthesized from Boc-d-Glu(OFm)-OH
and H-Ala-O-t-Bu HCl as illustrated in Scheme 1. After
.
HOBt and EDAC catalyzed coupling, the Fm-group
protecting the side-chain carboxylic acid could be
removed selectivelywith triethylamine.
Glucose-6-phosphatase activity and transport properties
Boc-d-Glu-Ala-O-t-Bu was purified byion exchange
and coupled to the target model drug. The product was
finallydeprotected bytrifluoroacetic acid in dichloro-
methane to give the trifluoroacetate salts. This synthetic
approach provides a flexible method for attachment of
compounds to the side-chain carboxylic acid of a
dipeptide through ester or amide bonds.
Since the linkers introduced in keystructure 2 were not
removable, the functionalized compounds, as well as the
model prodrugs, were tested with respect to G-6-Pase
inhibitoryactivityaccording to previouslydescribed
procedures.7 The hydroxymethyl linker compound 4
showed the highest G-6-Pase inhibitoryactivit,y fol-
lowed bythe amino linker compound 9. The amino-
methyl linker compound 7 did not show significant
activity(Fig. 1).
Introduction of linkers in the model drug
Compounds of the [4-(4-chlorophenyl)-4,5,6,7-tetra-
hydrothieno[3,2-c]pyridine-5-yl]-phenylmethanone type
As expected, the corresponding prodrugs 13, 14, and 15
all showed decreased G-6-Pase activity, presumably due
Scheme 1. Synthesis of Boc-d-Glu-Ala-O-t-Bu. (i) HOBt, EDAC, DIPEA, DMF, rt, 3 days; (ii) TEA, CH2Cl2, rt, 3 days.

G. M. Friedrichsen et al. / Bioorg. Med. Chem. 9 (2001) 2625–2632
2627
to the bulkiness of the peptide-coupled compounds,
which are no longer recognized bythe G-6-Pase
enzyme. Based on previous observations of cleavage of
this kind of prodrugs after transport, the active model
drug is expected to be released bypH-dependent
hydrolysis after transport to the active site.3ꢀ5
The apical to basolateral transport was not significantly
larger than the basolateral to apical transport in anyof
the compounds. This suggests that none of the com-
pounds are transported into the cell bycarrier-mediated
mechanisms. Interestingly, the parent compound 4 with
the hydroxymethyl linker was substrate for an efflux
transporter such as P-glycoprotein, while the peptide-
coupled compound 13 was not (efflux transporter score
for compound 4=3.3 and for prodrug 13=1.5). This
suggests that active efflux can be reduced byattachment
to a dipeptide. A similar observation was made with the
antiherpes drug Acyclovir, which is actively effluxed
from the cell. Attachment of valine to Acyclovir results
in a hPepT1 substrate, Valacyclovir, which is no longer
effluxed from the cell.10 The two other prodrugs, the
aminomethyl linked prodrug 14 (not shown) and the
amino linked prodrug 15 had low permeabilityin the
Caco-2 cells, probablydue to the poor aqueous solubi-
lityof these compounds. Compound 9 with the amino
linker was transported, but was also activelyeffluxed as
seen with compound 4. The aqueous solubilityof com-
pound 7 with the aminomethyl linker was too low to
perform the transport study.
The transport properties of the six compounds were mea-
sured in Caco-2 cells, and the permeabilityis presented in
Figure 2. Verapamil was used as a control for monitoring
the function of P-glycoprotein. The presence of P-glyco-
protein confirms that the monolayers express transporter
characteristics of a properlymatured Caco-2 phenotype.
Scheme 2. Synthesis of the 4-iodo key intermediate 2. (i) TEA,
CH2Cl2, 0 ꢁC, 10 min then R.T., 3 h.
Scheme 3. Synthetic pathway for functionalised compounds 4 and 7 derived from 2 and for the amino derivative 9. (i) i-PrMgBr, ꢀ15 ꢁC, 1 h; (ii)
DMF, ꢀ15 ꢁC, 3 h; (iii) NaBH4, 0 ꢁC to rt; (iv) SOCl2, 0 ꢁC, 1 h; (v) NaN3, rt, 24 h; (vi) Pd/C, rt, 3 h; (vii) 2 M HCl, 80 ꢁC, 15 h.
Scheme 4. Synthesis of model prodrugs. (i) HOBt, EDAC, DMF, rt, 15 h; (ii) TFA, CH2Cl2, rt, 3 h.

2628
G. M. Friedrichsen et al. / Bioorg. Med. Chem. 9 (2001) 2625–2632
hPepT1. A general synthetic pathway for attaching
compounds to dipeptides through ester or amide bonds
was developed. Furthermore, a synthetic approach to
functionalize model drugs from one keyintermediate
was generated. None of the peptide-coupled compounds
were transported byhPepT1. One of three peptide-cou-
pled compounds seemed to have a certain affinityfor
hPepT1, though. Interestingly, in one case the parent
compound was activelyeffluxed from the cell, while the
corresponding peptide-coupled compound was not. This
suggests the possibilityto reduce active efflux by
attachment to dipeptides. Since the aqueous solubility
of the parent compounds was not improved significantly
when a dipeptide was attached, this did not contribute
to an overall increase in permeability.
This studyimplies restrictions towards the size of com-
pounds that can be transported via hPepT1 byattach-
ment to d-Glu-Ala. Furthermore, the compounds
should possess a certain intrinsic solubilityto be able to
explore the oligopeptide transporter bythe dipeptide
prodrug approach. Thus, important information about
rational design of dipeptide-coupled prodrugs targeted
for hPepT1 was obtained.
Figure 1. Glucose-6-phosphatase activity(IC 50).
Experimental
Chemistry
Materials. All reactions involving air-sensitive reagents
were performed under a nitrogen atmosphere using syr-
inge-septum cap techniques, and glassware was flame-
dried prior to use. N-(3-Dimethylaminopropyl)-N0-
ethylcarbodiimide hydrochloride (EDAC) was handled
under a nitrogen atmosphere. Flash chromatography
was performed using silica gel (Merck 60, 70–230 mesh).
Melting points are uncorrected. NMR spectra were
recorded on a 300 MHz instrument (Varian Gemini).
NMR signals were assigned using APT spectra and
comparing with the spectrum of the general structure
shown in Scheme 2 (X¼H), which was throughoutly
assigned using C-H correlated and HMBC spectra. Ele-
mental analyses were performed by Microanalytical
Laboratory, Department of Physical Chemistry, Uni-
versityof Vienna, Austria.
Figure 2. Permeabilityof compounds 4 and 9 and the corresponding
model prodrugs 13 and 15 (all 50 mM). AP: Apical side, BL: Baso-
lateral side.
The hPepT1-affinityof prodrug 13 was tested in the
presence of Gly-Sar to clarify if the compound was a
hPepT1 substrate. Due to the limited aqueous solubility,
concentrations above 0.1 mM were not obtained, and it
was not possible to calculate the Ki value. Some inhibi-
tion of hPepT1 with respect to Gly-Sar uptake at the
Caco-2 monolayers was observed, though. Recently,
several compounds have been reported to interact with
hPepT1 without being transported.11,12 Since further
transport studies could not be performed due to low
aqueous solubility, no conclusion about hPepT1-medi-
ated transport of 13 could be made.
.
Boc-d-Glu(OFm)-OH and H-Ala-O-t-Bu HCl were
purchased from Bachem, Bubendorf, Switzerland. 2-
Thienylethyl amine was purchased from Avocado
Research Chemicals Ltd, Karlsruhe, Germany. All
other reagents and solvents were obtained from Fluka
or Aldrich and used without further purification except
dimethyl formamide (DMF) and triethylamine (TEA),
which were dried and stored over 3A molecular sieves.
i-PrMgBr was titrated (1.42 M in THF) prior to use.13
The turbidimetric solubilityof the three functionalized
compounds and the corresponding prodrugs was mea-
sured bynephrelometr,y which revealed that the
aqueous solubilityof the parent compounds and the
prodrugs was verylow (from 40–100 mg/mL). Thus,
coupling to d-Glu-Ala did not increase the aqueous
solubilitysignificantly.
[4-(4-Chlorophenyl)-4,5,6,7-tetrahydrothieno[3,2-c]pyri-
dine-5-yl]-(4-iodophenyl)-methanone (2). To a stirred
solution of 1 (1.00 g, 4.03 mmol) in dichloromethane
(50 mL) was added dryTEA (1.1 mL, 8.06 mmol). The
solution was cooled to 0 ꢁC, 4-iodobenzoyl chloride
(1.60 g, 6.05 mmol) was added, and stirring was con-
Conclusions
In conclusion, this studypresents the first attempt to
transport dipeptide-coupled active drug substances via

G. M. Friedrichsen et al. / Bioorg. Med. Chem. 9 (2001) 2625–2632
2629
tinued for 10 min at 0 ꢁC, The mixture was then allowed
to warm up to rt and stirred for additional 3 h. After
addition of saturated aqueous NaHCO3 (50 mL), the
solution was stirred overnight at rt. The organic phase
was separated and washed with HCl (1 M, 2Â50 mL),
NaOH (2 M, 2Â50 mL), and water (2Â50 mL). Residues
of benzoic acid anhydride were converted to benzamide by
addition of NH3 (25% in water, 25 mL) to a dichloro-
methane solution. After stirring for 1 h at rt, the water
phase was removed, and the organic phase was evapo-
rated in vacuo. Flash chromatography(CH 2Cl2) yielded
3.30 g (86%) as colorless crystals of 2; Rf=0.79 (EtOAc
1:1); mp 220–221 C; H NMR (CDCl3) d 7.77 (2H, d,
J=8.4 Hz, 2ÂH-20), 7.30 (4H, m, 2ÂH-200+2ÂH-300),
7.20 (1H, d, J=5.1 Hz, H-2), 7.11 (2H, d, J=8.4 Hz,
2ÂH-30), 6.95 (1H, broad s, H-4), 6.72 (1H, d,
J=5.1 Hz, H-3), 3.71 (1H, m, H-6A), 3.28 (1H, m, H-
6B), 2.96 (1H, m, H-7A), 2.85 (1H, m, H-7B). Anal. calcd
for C20H15NOSClI: C, 50.07; H, 3.15; N, 2.92. Found:
C, 49.81; H, 3.29; N, 3.13. The racemate was used in the
study.
[4-(4-Chlorophenyl)-4,5,6,7-tetrahydrothieno[3,2-c]pyri-
dine-5-yl]-(4-chloromethyl-phenyl)-methanone (5).
A
solution of 4 (100 mg, 0.26 mmol) in dichloromethane
(3 mL) was treated with thionyl chloride (0.12 mL,
1.62 mmol) and stirred for 1 h at 0 ꢁC. The mixture was
concentrated, co-evaporated three times with toluene
and dried in vacuo to give 104 mg (99%) of 5 as a col-
orless solid, which was used without further purifica-
tion. For analysis, the compound was purified by flash
chromatography(CH 2Cl2); mp 129–131 ꢁC; Rf=0.72
(EtOAc 1:1); 1H NMR (CDCl3) d 7.44 (2H, d,
J=8.4 Hz, 2ÂH-30), 7.37 (2H, d, J=8.4 Hz, 2ÂH-20),
7.34–7.27 (m, 2ÂH-200+2ÂH-300), 7.21 (1H, d,
J=5.1 Hz, H-2), 6.97 (1H, broad s, H-4), 6.73 (1H, d,
J=4.5 Hz, H-3), 4.60 (2H, s, CH2Cl), 3.75 (1H, m, H-
6A), 3.28 (1H, m, H-6B), 2.98 (1H, m, H-7A), 2.84 (1H,
m, H-7B). Anal. calcd for C21H17NOSCl2: C, 62.69; H,
4.26; N, 3.48. Found: C, 62.92; H, 4.36; N, 3.52.
ꢁ
1
(4-Aminomethyl-phenyl)-4-(4-chloro-phenyl)-6,7-dihydro-
4H-thieno[3,2-c]pyridine-5-yl-methanone (7). Sodium
azide (162 mg, 2.49 mmol), DMF (10 mL), and 5
(1000 mg, 2.49 mmol) were stirred for 24 h at rt. After
addition of water (50 mL), the product was extracted
with dichloromethane (3Â50 mL). The solvent was
removed in vacuo, and the product was dissolved in
ethyl acetate (20 mL) and washed with saturated aqu-
eous CaCl2 (3Â10 mL) to remove DMF. Removal of the
solvent in vacuo gave 1.02 g (100%) of the azide 6 as a
colorless solid.
[4-(4-Chlorophenyl)-4,5,6,7-tetrahydrothieno[3,2-c]pyri-
dine-5-yl]-(4-formylphenyl)-methanone (3).
2 (1.00 g,
2.09 mmol) was dissolved in dryTHF (20 mL) byheat-
ing. The solution was cooled to ꢀ15 ꢁC, then i-PrMgBr
(1.42 M) (1.76 mL, 2.50 mmol) was added over a 2 min
period. Stirring was continued for 1 h at ꢀ15 ꢁC, DMF
(800 mL, 10.4 mmol) was added, and the solution was
stirred for further 3 h at ꢀ15 ꢁC. After addition of HCl
(1 M, 50 mL), the solution was allowed to warm up to rt
in 1 h. A colorless precipitate appeared, which was iso-
lated byfiltration and dried in vacuo. This gave 779 mg
The azide 6 (965 mg) was dissolved in MeOH (50 mL)
and cooled to ꢀ78 ꢁC. This solution was added to Pd/C
(10%) (50 mg) and stirred under hydrogen for 3 h at rt.
After filtration through Celite, the solvent was evapo-
(97%) of 3; mp 190–191 ꢁC; H NMR (CDCl3) d 10.06
1
(1H, s, CHO), 7.95 (2H, d, J=8.4 Hz, 2ÂH-20), 7.53
(2H, d, J=8.1 Hz, 2ÂH-30), 7.32 (4H, m, 2ÂH-
200+2ÂH-300), 7.22 (1H, d, J=5.1 Hz, H-2), 6.98 (1H,
broad s, H-4), 6.75 (1H, d, J=5.1 Hz, H-3), 3.63 (1H,
m, H-6A), 3.31 (1H, m, H-6B), 2.98 (1H, m, H-7A), 2.86
(1H, m, H-7B). Anal. calcd for C21H16O2NSCl: C,
66.05; H, 4.22; N, 3.67. Found: C, 65.78; H, 4.48; N,
3.53. The compound should be stored under a nitrogen
atmosphere at ꢀ18 ꢁC.
rated in vacuo. Flash chromatography(CH Cl2-MeOH
2
9:1) gave 583 mg (64%) as colorless crystals of 7; mp
68–72 ꢁC; H NMR (CDCl3) d 7.40–7.26 (8H, m, aro-
1
matic protons), 7.20 (1H, d, J=5.1 Hz, H-2), 6.97 (1H,
broad s, H-4), 6.73 (1H, d, J=5.1 Hz, H-3), 3.91 (2H, s,
CH2NH2), 3.76 (1H, m, H-6A), 3.26 (1H, m, H-6B), 2.98
(1H, m, H-7A), 2.82 (1H, m, H-7B), 1.95 (2H, s, NH2).
Anal. calcd for C21H19N2OSCl+1.4 mol H2O: C, 61.80;
H, 5.38; N, 6.86. Found: C, 61.69; H, 5.01; N, 6.72.
[4-(4-Chlorophenyl)-4,5,6,7-tetrahydrothieno[3,2-c]pyri-
dine-5-yl]-(4-hydroxymethyl-phenyl)-methanone (4).
3
(4-Acetamino-phenyl)-[4-(4-chloro-phenyl)-6,7-dihydro-
4H-thieno[3,2-c]pyridine-5-yl]-methanone (8). 1-Hydroxy-
benzotriazole hydrate (649 mg, 4.81 mmol) was added to
a stirred solution of 4-acetaminobenzoic acid (717 mg,
4.00 mmol) in dryDMF (5 mL) under nitrogen atmo-
sphere at rt. After addition of EDAC (921 mg,
4.81 mmol), the solution was stirred for 30 min, then 1
(1000 mg, 4.00 mmol) was added. The solution was stir-
red overnight at rt. Addition of water (30 mL) gave a
colorless precipitate. The solution was filtered, and the
residue was dissolved in dichloromethane (30 mL). The
organic solution was washed with HCl (1 N, 10 mL) and
NaOH (1 N, 3Â10 mL), then dried (MgSO4), filtered,
and evaporated in vacuo. Flash chromatography
(EtOAc) yielded 1490 mg (91%) of 8 as colorless crys-
tals; Rf=0.50 (EtOAc); mp 162–163 ꢁC; 1H NMR
(CDCl3) d 7.86 (1H, s, NH), 7.48 (2H, d, J=8.4 Hz, H-
20), 7.30 (4H, m, 2ÂH-200+2ÂH-300), 7.28 (1H, d,
(500 mg, 1.31 mmol) was suspended in methanol
(10 mL) and cooled to 0 ꢁC. To this solution was added
NaBH4 (198 mg, 5.24 mmol) over a 5 min period. Stir-
ring was continued for 20 min while the solution was
allowed to warm up to rt. The crystals formed were
washed with water (10 mL, acidified with 4 M HCl) and
with further 2Â10 mL water until neutral. Drying
(MgSO4) gave 503 mg (100%) of the crude product.
Recrystallization from ethyl acetate gave 440 mg (87%)
as colorless crystals of 4; mp 122.5–123.5 ꢁC; H NMR
1
(CDCl3) d 7.44–7.26 (8H, m, 2ÂH-20+2ÂH-30+2ÂH-
200+2ÂH-300), 7.20 (1H, d, J=5.4 Hz, H-2), 6.98 (1H,
broad s, H-4), 6.74 (1H, d, J=5.4 Hz, H-3), 4.73 (2H, d,
J=5.4 Hz, CH2O), 3.75 (1H, m, H-6A), 3.27 (1H, m, H-
6B), 2.98 (1H, m, H-7A), 2.83 (1H, m, H-7B), 1.93 (1H, t,
J=5.7 Hz, OH). Anal. calcd for C21H18NO2SCl: C, 65.70;
H, 4.73; N, 3.65. Found: C, 65.44; H, 4.91; N, 3.53.

2630
G. M. Friedrichsen et al. / Bioorg. Med. Chem. 9 (2001) 2625–2632
J=8.4 Hz, 2ÂH-30), 7.20 (1H, d, J=5.1 Hz, H-2), 6.94
(1H, broad s, H-4), 6.71 (1H, d, J=4.5 Hz, H-3), 3.80
(1H, m, H-6A), 3.27 (1H, m, H-6B), 2.99 (1H, m, H-7A),
2.84 (1H, m, H-7B), 2.15 (3H, s, CH3).
(50 mL) at the H+-form. The column was eluted with
further 50 mL of MeOH-water [1:1]. Removal of the
solvent in vacuo gave 2.39 g (88%) of Boc-d-Glu-Ala-O-t-
Bu as colorless crystals; mp 135–136ꢁC; 1H NMR
(CDCl3) d 7.20 (1H, d, J=6.9 Hz, NH), 5.49 (1H, d,
J=7.8 Hz, NH), 4.60–4.24 (2H, m, CH in Glu+CH in
Ala), 2.60–1.80 (4H, m, 2ÂCH2 in Glu), 1.44 (9H, s, Boc),
1.42 (9H, s, COO-t-Bu), 1.37 (3H, d, J=7.2Hz, CH3 in
Ala); 13C NMR (DMSO) d 176.3 (COOH), 172.2/171.5
(COO-t-Bu), 156.1 (CONHR), 82.1/80.4 (C(CH3)3), 53.2
(CH), 48.7 (CH), 29.8 (CH2COOH, CH2CH), 18.1 (CH3).
(4-Amino-phenyl)-[4-(4-chloro-phenyl)-6,7-dihydro-4H-
thieno[3,2-c]pyridine-5-yl]-methanone (9).
8 (600 mg,
1.46 mmol) was dissolved in HCl (2 M, aq) (7.3 mL,
14.6 mmol) and MeOH (50 mL). The solution was stir-
red overnight at 80 ꢁC, evaporated in vacuo to dryness,
and co-evaporated twice with methanol (2Â10 mL) to
remove HCl. After addition of water (10 mL), the pro-
duct was extracted with dichloromethane (5Â20 mL).
Evaporation in vacuo afforded 536 mg (99%) of 9 as
colorless crystals. Recrystallization from 2-propanol
gave one batch containing 9 as the hydrochloride
(249 mg, 46%, mp 235–238 ꢁC) and one batch contain-
ing the free amine (269 mg, 33%); mp 169–170 ꢁC;
4-tert-Butoxycarbonylamino-4-(1-tert-butoxycarbonyl-
ethylcarbamoyl)-butyric acid 4-[4-(4-chloro-phenyl)-6,7-
dihydro-4H-thieno[3,2-c]pyridine-5-carbonyl]-benzyl ester
(10), Standard procedure. To a stirred solution of Boc-
d-Glu-Ala-O-t-Bu (300 mg, 0.80 mmol) in dryDMF
(5 mL) was added 1-hydroxybenzotriazole hydrate
(162 mg, 1.20 mmol) under a nitrogen atmosphere at rt.
After addition of EDAC (338 mg, 1.76 mmol), the solu-
tion was stirred for 30 min, then 4 (615 mg, 1.60 mmol)
was added. The solution was stirred for 15 h at rt. The
solvent was evaporated in vacuo, and the residue was
dissolved in dichloromethane (30 mL), washed with
water (2Â10 mL) and saturated aqueous NaHCO3
(4Â10 mL), dried (MgSO4), filtered, and evaporated in
vacuo. Flash chromatography(EtOAc 2:1 !EtOAc 1:1)
yielded 452 mg (76%) of the protected compound 10 as
a colorless solid with a broad melting point; Rf=0.37
1
Rf=0.26 (EtOAc 1:1); H NMR (CDCl3) d 7.30–7.25
(4H, m, 2ÂH-200+2ÂH-300), 7.23 (2H, dt, J=8.7 Hz,
J=2.4 Hz, 2ÂH-30), 7.18 (1H, d, J=5.4 Hz, H-2), 6.71
(1H, d, J=5.4 Hz, H-3), 6.66 (2H, dt, J=8.4 Hz,
J=2.40 Hz, 2ÂH-20), 3.85 (2H, s, NH2), 3.25 (1H, m, H-
6A), 3.03 (1H, m, H-6B), 2.85 (1H, m, H-7A), 2.80 (1H,
m, H-7B). Anal. calcd for C20H17N2OSCl: C, 65.12; H,
4.64; N, 7.59; Cl, 9.61. Found: C, 64.83; H, 4.74; N,
7.34; Cl, 9.65.
Boc-D-Glu(OFm)-Ala-O-t-Bu.
Boc-d-Glu(OFm)-OH
(P.E.0EtOAc [1:1]); H NMR (DMSO) d 7.44–7.26 (8H,
1
(5.00 g, 11.8 mmol) and 1-hydroxybenzotriazole hydrate
(1.80 g, 11.8 mmol) were dissolved in dryDMF
(100 mL) under nitrogen atmosphere. After 2–3 min, N-
ethyldiisopropylamine (1.90 mL, 12.9 mmol) was added.
H-Ala-O-t-BuHCl (1.71 g, 11.8 mmol) and EDAC
(3.38 g, 17.6 mmol) were added, and the solution was
stirred under nitrogen atmosphere for 3 days at rt. The
solution was evaporated to dryness in vacuo, and the
residue was dissolved in ethyl acetate (100 mL). The
organic solution was washed with saturated aqueous
NaHCO3 (3Â20 mL) and 5% aqueous acetic acid
(3Â30 mL). Drying (MgSO4), filtration, and removal of
EtOAc in vacuo afforded the crude product as a yellow
oil. Flash chromatography(CH 2Cl2!CH2Cl2-MeOH
9:1) yielded 3.96 g (61%) as colorless crystals; mp 52–
m, 2ÂH-20+2ÂH-30+2ÂH-200+2ÂH-300), 7.21 (1H, d,
J=4.8 Hz, H-2), 6.98 (1H, broad s, H-4), 6.73 (1H, d,
J=4.8 Hz, H-3), 6.67 (1H, d, J=6.9 Hz, NH), 5.20 (1H,
m, NH), 5.16 (2H, s, CH2O), 4.42 (1H, m, CH in Ala),
4.20 (1H, m, CH in Glu), 3.76 (1H, m, H-6A), 3.28 (1H,
m, H-6B), 3.00 (1H, m, H-7A), 2.84 (1H, m, H-7B), 2.52
(2H, m, CH2 in Glu), 2.26–1.86 (2H, m, CH2 in Glu),
1.46 (9H, s, Boc), 1.44 (9H, s, COO-t-Bu), 1.37 (3H, d,
J=6.9 Hz, CH3).
1-(1-Carboxy-ethylcarbamoyl)-3-{4-[4-(4-chlorophenyl)-
4,5,6,7-tetrahydro-4H-thieno[3,2-c]pyridine-5-carbonyl]-
benzyloxycarbonyl}-propyl-ammonium trifluoroacetate
(13), Standard procedure. The protected conjugate 10
(452 mg) was dissolved in dichloromethane (20 mL).
After addition of TFA (10 mL), the solution was stirred
for 3 h at rt. The solvent was removed byevaporation in
vacuo, and TFA was co-evaporated with ether twice.
This produced 426 mg (100%) of 13 as a colorless solid
with a broad melting point (60–75 ꢁC); 1H NMR
(DMSO) d 8.81 (1H, d, J=7.5 Hz, NH), 8.19 (3H,
broad s, NH3), 7.50–7.20 (9H, m, aromatic protons+H-
2), 6.85 (1H, d, J=5.1 Hz, H-3), 6.76 (1H, broad s, H-
4), 5.13 (2H, s, CH2O), 4.25 (1H, m, CH in Ala), 3.84
(1H, m, CH in Glu), 3.61 (1H, m, H-6A), 3.22 (1H, m,
H-6B), 2.97 (1H, m, H-7A), 2.86 (1H, m, H-7B), 2.5 (2H,
m, CH2 in Glu), 2.02 (2H, m, CH2 in Glu), 1.28 (3H,
2Âd, J=7.5 Hz, CH3); 13C NMR (DMSO) d 174.1
(COOH), 172.1 (COOCH2), 169.8 (CONR2), 168.2
(CONHR), 158.9 (q, J=36 Hz, CF3CO), 140.6 (C-400),
138.1 (C-40), 136.2 (C-8), 134.9 (C-10), 133.6 (C-100),
132.9 (C-9), 130.4 (C-300), 129.0 (C-20), 128.5 (C-200),
127.1 (C-30), 127.0 (C-3), 124.7 (C-2), 116.3 (q,
54 ꢁC; Rf=0.60 (CH2Cl2); H NMR (CDCl3) d 7.80–
1
7.28 (8H, m, aromatic Fm-protons), 6.69 (1H, d,
J=6.0 Hz, NH), 5.25 (1H, d, J=6.3 Hz, NH), 4.50–4.32
(3H, m, CH in Glu+CH in Ala+CH in Fm-group),
4.22 (2H, t, J=6.9 Hz, CH2 in Fm-group), 2.65–2.40
(2H, m, CH2 in Glu), 2.25–1.85 (2H, m, CH2 in Glu),
1.46 (9H, s, Boc), 1.43 (9H, s, COO-t-Bu), 1.37 (3H, d,
J=7.2 Hz, CH3). Anal. calcd for C31H40N2O7: C, 67.37;
H, 7.30; N, 5.07. Found: C, 67.44; H, 7.27; N, 5.01.
Boc-D-Glu-Ala-O-t-Bu. Boc-d-Glu(OFm)-Ala-O-t-Bu
(4.00 g) was dissolved in dichloromethane (30 mL), dry
TEA (5 mL) was added, and the solution was stirred for
3 days at rt. The solvent was removed in vacuo, and the
residue was dissolved in water (50 mL). The solution
was filtered, and the residue was extracted with further
3Â10 mL of water. MeOH (80 mL) was added to the
combined aqueous solutions. This solution was passed
through a column with ion exchanger (Amberlite IR-120

G. M. Friedrichsen et al. / Bioorg. Med. Chem. 9 (2001) 2625–2632
2631
J=293 Hz, CF3), 66.1 (PhCH2O), 53.1 (C-4), 51.7 (CH),
48.2 (CH), 41.4 (C-6), 29.1 (CH2CO), 26.6 (CH2CH),
25.4 (C-7), 17.6 (CH3); MS (EI) 584 (M+), calcd 584 (M).
The standard procedure yielded 100% of 15 as a colorless
solid with a broad melting point (115–135 ꢁC); 1H NMR
(DMSO) d 10.22 (1H, s, NH), 8.83 (1H, d, J=7.5 Hz,
NH), 8.20 (3H, d, J=4.2 Hz, NH3), 7.65 (2H, d,
J=8.7 Hz, 2ÂH-20), 7.41 (2H, d, J=8.7 Hz, 2ÂH-30),
7.46–7.20 (5H, m, 2ÂH-200+2ÂH-300+H-2), 6.83 (1H, d,
J=5.1 Hz, H-3), 6.73 (1H, broad s, H-4), 4.28 (1H, m, CH
in Ala), 3.87 (1H, m, CH in Glu), 3.70 (1H, m, H-6A),
3.20 (1H, m, H-6B), 2.99 (1H, m, H-7A), 2.87 (1H, m, H-
7B), 2.43 (2H, m, CH2 in Glu), 2.03 (2H, m, CH2 in
Glu), 1.31 (3H, d, J=7.2 Hz, CH3); 13C NMR (DMSO)
d 174.1 (COOH), 170.7 (CONHR), 170.0 (CONR2),
168.4 (CONHR), 158.7 (q, J=40 Hz, CF3CO), 140.9
(C-400), 140.6 (C-40), 135.0 (C-8), 133.7 (C-100), 132.8 (C-
10), 130.9 (C-9), 130.4 (C-300), 129.0 (C-20), 128.0 (C-200),
127.0 (C-3), 124.7 (C-2), 119.2 (C-30), 116.7 (q,
J=295 Hz, CF3), 53.0 (C-4), 51.9 (CH), 48.2 (CH), 40.7
(C-6), 31.5 (CH2CO), 26.9 (CH2CH), 25.2 (C-7), 17.7
(CH3); MS (EI) 569 (M+), calcd 569 (M).
2-(2-tert-Butoxycarbonylamino-4-{4-[4-(4-chloro-phenyl)
-6,7-dihydro-4H-thieno[3,2-c]pyridine-5-carbonyl]-benzyl-
carbamoyl}-butyrylamino-propionic acid tert-butyl ester
(11). The standard procedure yielded 76% of the pro-
tected compound 11 as a colorless solid with a broad
1
melting point; Rf=0.47 (EtOAc); H NMR (DMSO) d
7.37–7.27 (8H, m, aromatic protons), 7.20 (1H, d,
J=5.4 Hz, H-2), 6.95 (1H, broad s, H-4), 6.74 (1H, d,
J=5.7 Hz, H-3), 6.74 (1H, NH), 5.61 (1H, d, J=6.0 Hz,
NH), 4.49 (2H, m, CH2Ph), 4.40 (1H, m, CH in Ala),
4.13 (1H, m, CH in Glu), 3.75 (1H, m, H-6A), 3.26 (1H,
m, H-6B), 2.97 (1H, m, H-7A), 2.83 (1H, m, H-7B), 2.35
(2H, m, CH2 in Glu), 2.20–1.92 (2H, m, CH2 in Glu),
1.44 (9H, s, Boc), 1.42 (9H, s, COO-t-Bu), 1.32 (3H, d,
J=7.2 Hz, CH3).
1-(1-Carboxy-ethylcarbamoyl)-3-{4-[4-(4-chlorophenyl)-
4,5,6,7-tetrahydro-4H-thieno[3,2-c]pyridine-5-carbonyl]-
Transport assay
Materials
benzylcarbamoyl}-propyl-ammonium
trifluoroacetate
(14). The standard procedure yielded 100% of 14 as a
yellow solid with a broad melting point (100–120 ꢁC);
1H NMR (DMSO) d 8.80 (1H, d, J=7.5 Hz, NH), 8.54
(1H, t, J=6.0 Hz, NH in linker), 8.18 (3H, broad s,
NH3), 7.47–7.22 (9H, m, H-2+aromatic protons), 6.85
(1H, d, J=5.1 Hz, H-3), 6.76 (1H, broad s, H-4), 4.30
(2H, m, CH2Ph), 4.28 (1H, m, CH in Ala), 3.84 (1H, m,
CH in Glu), 3.63 (1H, m, H-6A), 3.20 (1H, m, H-6B),
2.98 (1H, m, H-7A), 2.86 (1H, m, H-7B), 2.26 (2H, m,
CH2 in Glu), 1.97 (2H, m, CH2 in Glu), 1.31 (3H, d,
J=7.2 Hz, CH3); 13C NMR (DMSO) d 174.1 (COOH),
171.6 (CONHR), 170.0 (CONR2), 168.4 (CONHR),
158.7 (q, J=34 Hz, CF3CO), 141.6 (C-400), 140.6 (C-40),
135.1 (C-8), 134.9 (C-10), 133.6 (C-100), 132.9 (C-9),
130.4 (C-300), 129.0 (C-20), 127.9 (C-200), 127.5 (C-3),
127.0 (C-30), 124.7 (C-2), 116.8 (q, J=294 Hz, CF3),
52.1 (CH), 48.1 (CH), 42.2 (CH2NH), 41.4 (C-6), 30.7
(CH2CO), 27.4 (CH2CH), 25.4 (C-7), 17.7 (CH3); MS
(EI) 583 (M+), calcd 583 (M).
Caco-2 human colon carcinoma cells were obtained
from the ATCC (Rockville, MD, USA). [3H]Gly-
cylsarcosine ([3H]Gly-Sar, 2.1 Ci/mmol) was purchased
from Moravek Biochemicals (Brea, CA, USA). 2-(N-
Morpholino)-ethanesulfonic acid (Mes) was purchased
from Sigma (St. Louis, MO, USA) and N-2-hydro-
xyethylpiprazine-N0-2-ethanesulfonic acid (Hepes) was
purchased from Biological Industries (Kibbutz Beit
Haemek, Israel). All analytical grade solvents used for
HPLC analysis were obtained from Merck (Darmstadt,
Germany), and Ultima Gold scintillation fluid was pur-
chased from Packard (Groningen, The Netherlands).
Cell culture and pretreatment
Caco-2 cell culture was maintained according to a pre-
3,13
viouslypublished method;
cells were used between
passages 62 and 70. Transwell1 monolayers reached a
transepithelial electrical resistance (TEER) of between
600–800 ohmsÂcm2, and the total amount of protein on
each confluent Transwell1 filter was calculated to be
0.42 mg/cm2 using the Lowrymethod.
2-(2-tert-Butoxycarbonylamino-4-{4-[4-(4-chloro-phenyl)
-6,7-dihydro-4H-thieno[3,2-c]pyridine-5-carbonyl]-phenyl-
carbamoyl}-butyrylamino-propionic acid tert-butyl ester
(12). The standard procedure yielded 56% of the pro-
tected compound 12 as a colorless solid with a broad
1
melting point; Rf=0.63 (EtOAc); H NMR (CDCl3) d
IC50 Determination experiments
9.11 (1H, broad s, NH), 7.65 (2H, d, J=8.4 Hz, 2ÂH-
30), 7.36 (2H, d, J=8.4 Hz, 2ÂH-20), 7.30 (4H, broad s,
2ÂH-200+2ÂH-300), 7.20 (1H, d, J=5.1 Hz, H-2), 6.95
(1H, broad s, H-4), 6.73 (1H, d, J=5.1 Hz, H-3), 6.56
(1H, d, J=6.9 Hz, NH), 5.51 (1H, s, NH), 4.43 (1H, m,
CH in Ala), 4.19 (1H, m, CH in Glu), 3.82 (1H, m, H-
6A), 3.28 (1H, m, H-6B), 3.00 (1H, m, H-7A), 2.85 (1H,
m, H-7B), 2.47 (2H, t, J=6.3 Hz, CH2 in Glu), 2.24–1.96
(2H, m, CH2 in Glu), 1.46 (9H, s, Boc), 1.44 (9H, s,
COO-t-Bu), 1.35 (3H, d, J=7.2 Hz, CH3).
[3H]Gly-Sar displacement experiments were performed
as has been described previously.3,13 Briefly, Caco-2
monolayers were first rinsed and then incubated with
HBSS (apical media=10 mM Mes, pH 6.0; basal med-
ia=10 mM Hepes, pH. 7.4) for 15 min at 37 ꢁC under a
5% CO2 atmosphere in order to equilibrate the cells to
the change in pH gradient. Next, [3H]Gly-Sar (0.25 mCi),
and in certain wells, dipeptide-modified compounds of
various concentrations were added concomitantlyto the
apical media of the Caco-2 Transwells. Following a 15-
min incubation period, buffer was removed from both
the apical and basal chambers and the cells were washed
four times with ice-cold HBSS, pH 7. Following this
1-(1-Carboxy-ethylcarbamoyl)-3-{4-[4-(4-chlorophenyl)-
4,5,6,7-tetrahydro-4H-thieno[3,2-c]pyridine-5-carbonyl]-
phenylcarbamoyl}-propyl-ammonium trifluoroacetate (15).

2632
G. M. Friedrichsen et al. / Bioorg. Med. Chem. 9 (2001) 2625–2632
washing step, the entire polycarbonate membrane was
cut from the Transwell support and placed into a scin-
tillation vial, scintillation fluid was added, and the cell-
associated radioactivitywas counted via liquid scintilla-
tion spectrometry.
Acknowledgements
The authors are grateful to Niels Westergaard and
Mads O. N. Jensen for performing the biological testing
and to Jane Lundbeck for general support during the
synthetic work. Furthermore, Bertel Rudinger is
acknowledged for technical assistance.
Transport of test compounds and Gly-Sar across Caco-2
monolayers
Transport via the apical oligopeptide transporter was
performed as has been described previously.13 Briefly,
the apical-to-basal transport of each compound was
measured hourlyover the course of a 2-h experiment.
For all experiments, confluent Caco-2 monolayers were
pretreated as described in ‘Cell culture and pretreat-
ment’, then either the dipeptide-modified prodrug
(50 mM) or [3H]Gly-Sar was added to the apical media.
The TEER was monitored periodicallythroughout the
course of the experiment. Samples taken from the basal
compartment were prepared for HPLC analysis as has
been described previously.14
References
1. Adibi, S. A. Gastroenterology 1997, 113, 332.
2. Doring, F.; Will, J.; Amasheh, S.; Clauss, W.; Ahlbrecht,
H.; Daniel, H. J. Biol. Chem. 1998, 273, 23211.
3. Taub, M.; Moss, B. A.; Steffansen, B.; Frokjaer, S. Int. J.
Pharmaceutics 1997, 156, 219.
4. Taub, M.; Moss, B. A.; Steffansen, B.; Frokjaer, S. Int. J.
Pharmaceutics 1998, 174, 223.
5. Lepist, E.-I.; Kusk, T.; Larsen, D. H.; Andersen, D.;
Frokjaer, S.; Taub, M. E.; Veski, P.; Lennernas, H.; Frie-
drichsen, G.; Steffansen, B. Eur. J. Pharm. Sci. 2000, 11,
43.
6. Steffansen, B.; Lepist, E.-I.; Taub, M. E.; Larsen,
B. D.; Frokjaer, S.; Lennernas, H. Eur. J. Pharm. Sci.
1999, 8, 67.
The apparent permeabilitycoefficients were calculated
using eq (1):
7. Madsen, P.; Lundbeck, J. M.; Jakobsen, P.; Varming,
A. R.; Westergaard, N. Bioorg. Med. Chem. 2000, 8, 2277.
8. Goolcharran, C.; Khossravi, M.; Borchardt, R. T. In
Pharmaceutical Formulation Development of Peptides and Pro-
teins; Frokjaer, S., Hovgaard, L., Eds.; Taylor and Francis,
London, 2000, pp 70–88.
9. Boymond, L.; Rottlander, M.; Cahiez, G.; Knochel, P.
Angew. Chem. Int. Ed. 1998, 37, 1701.
10. Lee, C.-P.; de Vrueh, R. L. A.; Smith, P. L. Adv. Drug
Del. Rev. 1997, 23, 47.
Papp ¼ V  dC=A  Co  dt
ð1Þ
where V Â ðdC=dtÞ represents the steady-state rate of
transport of the dipeptide from the apical to the baso-
lateral chamber following initial lag time; Co represents
the initial concentration of the dipeptide in the apical
chamber; and A represents the area of the Transwell
membrane used for these experiments (1 cm2).
11. Terada, T.; Sawada, K.; Saito, H.; Hashimoto, Y.; Inui,
K. Eur. J. Pharm 2000, 392, 11.
12. Schoenmakers, R. G.; Stehouwer, M. C.; Tukker, J. J.
Pharm. Res. 1999, 16, 62.
The efflux transporter score was calculated using eq (2):
13. Lin, H.-S.; Paquette, L. A. Synth. Comm. 1994, 24,
2503.
PappðBL toAPÞ : PappðAP toBLÞ;
ð2Þ
14. Taub, M. E.; Larsen, B. D.; Steffansen, B.; Frokjaer, S.
Int. J. Pharmaceutics 1997, 146, 205.
where AP is apical side and BL is basolateral side.