Note: Descriptions are shown in the official language in which they were submitted.
CA 02334859 2000-12-12
WO 99/65941 PCT/GB98/01722
1
INSULIN ANALOGUE
The present invention relates to novel insulin
analogues which are covalent conjugates of an insulin
molecule and a derivative of the hormone thyroxine,
3,3',5'triiodothyronine.
In WO-A-95/05187 we described novel insulin conjugates
with hormones, specifically with tetraiodothyroxine
(3,3',5,5'tetraiodothyronine, T4), which were
hepatoselective. The hepatoselectivity was believed to be
due to the fact that, when introduced percutaneously, the
size of the molecule (about 155 higher molecular weight
than insulin itself) allows it to diffuse through the
capillary endothelium into the circulation. In the
circulation it is believed to bind reversibly the
circulating proteins having an affinity for the thyroxine
moiety, namely throxine binding globulin, thyroxine binding
prealbumin and albumin, collectively known as thyroxine
binding proteins (THP). These higher molecular weight
complexes are then unable to diffuse back through capillary
endothelium, but are able to diffuse through the relatively
larger pores of the hepatic endothelium. The conjugate is
found to retain insulin activity. The hepatoselectivity
ensures that insulin is directed to the site where its
activity is required.
In WO-A-95/07931 hydrophobically modified insulin
analogues are described. The insulin is generally
derivatised by acylation of the pendant amino group of
lysine at B29 with a fatty acid. However there is also an
example of derivatising that residue with thyroxine, or
with tetraiidothyroacetic acid. The analogues are alleged
to have a protracted profile of action, although the
mechanism by which this takes place is not elucidated.
One potential problem with the T4-insulin conjugate is
that it may retain thyroxine activity. The present
invention seeks to solve this problem while providing a
conjugate which retains its hepatoselectivity, insulin
activity and circulating protein affinity.
CA 02334859 2000-12-12
WO 99/65941 PCT/GB98/01722
2
A new compound according to the invention comprises an
insulin molecule covalently bound to 3,3',5'-
triiodothyronine.
The 3,3',5'triiodothyronine molecule is not a
naturally occurring compound. It is an isomer of 3,5,3'
triiodothyronine (T'3) and is consequently known as reverse
T3, rT3. It has insignificant activity on thyroxine
receptor, but thyroxine binding proteins have an affinity
for the molecule. Thus the compound of the invention
should have affinity for TBP's and, it is believed,
consequential hepatoselectivity whilst the compound and its
metabolites should not stimulate thyroxine activity.
The rT3 moiety should be conjugated to a residue of
the insulin molecule such that insulin activity is not
adversely affected. As in WO-A-95/05187, conjugation is
preferably through the B1 residue of insulin.
Alternatively the B29 residue may be linked to rT3. 1n WO-
95/07931, the B29 residue may be derivatised and the
methods of conjugating a carboxylic acid-containing
compound to the B29 residue as disclosed in that reference
may be used in the present invention.
The insulin may be made by recombinant DNA techniques
or may be isolated from natural sources, human or animal.
Recombinant insulin may have deleted residues as desired,
for instance the B29 residue may be deleted. Other
residues of naturally occurring insulin may be substituted,
usually by conservative substitutions. For instance in WO-
A-95/07931, analogues in which the B3 and/or A21 residues
are other than those of naturally occurring insulin.
The rT3 molecule is conjugated to the insulin using
conventional biochemical techniques in which pendant groups
on the appropriate residue of the insulin molecule are
covalently bonded to rT3, through the carboxylate group.
The pendant group :is usually the e-amino group of a .Lysine
residue. Any other lysine residues may be rendered
unreactive by protecting the e-amine groups using
CA 02334859 2000-12-12
WO 99/65941 PCT/GB98/01722
3
conventional techniques. Protecting groups are removed
after conjugation to the rT3 molecule.
The phenolic OH group of rT3 is protected during the
process, also.
Either or both of the amine group and the carboxylate
group may be activated prior to contact of the insulin with
the rT3. Conventional techniques for generation of amide
linkages may be used, for instance using known reagents.
A spacer may be included between the insulin molecule
and the rT3 molecule. A spacer may, for instance, improve
retention of insulin activity and/or TBP-binding. A spacer
may also be used to control in vivo cleavage and metabolism
of the conjugate compound, and consequently its insulin
activity. A spacer may, for instance include a chain
comprising 2 to 22 carbon and/or heteroatoms, such as a 4-
10 atom chain, preferably comprising an alkylene group and
carbonyl and/or amino groups, amido groups and or oxygen
atoms in ester or ether linkages.
The inventors have found that the insulin-rT3
conjugate has a similar potency relative to human insulin
itself. This is in contrast to T~-insulin, which appears
to have a greater potency than human insulin. In the
presence of binding proteins, especially thyroxin binding
proteins, the potency of T4-insulin is reduced, whereas
these proteins do not affect the potency of rT3-insulin.
These data indicate, that the conjugate is likely to have
similar effects as insulin in vivo.
Further tests in which the ED50 of the conjugates as
compared to insulin, in the presence and absence of binding
proteins (human serum albumin and thyroxin binding globulin
and transthyretin) show that each conjugate on its own has
a similar ED50 to human insulin itself. The ED50's of the
T4-insulin conjugate are significantly increased by the
presence of TBG, whilst the ED50's of the rT3-insulin are
~ not effected to a significant degree.
We have also conducted competitive binding assays of
the insulin analogues compared to human insulin with
CA 02334859 2000-12-12
WO 99/65941 PCT/GB98/01722
4
~zs-Insulin to insulin receptors on liver plasma membrane
(LPM). Insulin is known to inhibit the binding of
~zs-Insulin to these receptors . We have found that TBP does
not affect this ability. rT3 behaves in a similar way to
human insulin in that it inhibits binding of lzs-Insulin to
the receptors on LPM and this is not affected by the
presence of TBP. T4 insulin itself does inhibit 'zs-Insulin
binding to these receptors. In contrast, however, TBP
significantly affects this inhibition.
The novel compound is suitable for use in a method of
treatment of the human or animal, for instance to replace
insulin in a method of insulin replacement therapy. The
invention thus comprehends novel compositions containing
the compound as well as pharmaceutical compositions
containing the compound and a pharmaceutically acceptable
excipient. The composition is formulated so as to be
suitable for administration by the usual routes, generally
by subcutaneous injection. Accordingly the carrier is
generally aqueous. The invention comprehends also a new
use of the compound in the manufacture of a medicament for
use in a method of treatment of the human or animal body.
The following examples illustrate the invention.
Example 1
Preparation of frT3(Na-B1)l-insulin
1.1 Synthesis of Msc-rT3
50.0 mg rT3 (76.8 umol, 651.0 g/mol)
20.4 mg Msc-OSu (76.9 umol, 265.24 g/mol)
50.0 mg rT3 were suspended in 400 ul dimethylformamide
and 20.4 mg Msc-OSu, dissolved in 100 ul dimethylformamide,
were added. 4 ul of triethylamine were pipetted into the
solution and the mixture was stirred overnight at
room temperature.
CA 02334859 2000-12-12
WO 99/65941 PCT/GB98/01722
1.2 Synthesis of Msc-rT3-OSu
16.6 mg DCC (80.6 umol, 206.3 g/mol)
16.6 and DCC were dissolved in 50 ul dimethylformamide
5 and added to the above reaction mixture. The activation is
complete after 3 h at room temperature.
1.3 Synthesis of ~rT3(Na-B1)~-insulin
230 mg A1,B29-(Msc)2-insulin (6078 g/mol, 38 umol)
synthesised according to Schizttler A and Brandenburg D,
Hoppe-Seyler's Z. F~hysiol.Chem. 360, 1721-1725 (1979) were
dissolved in 3 ml d.imethylformamide with the addition of 4
ul triethylamine and then reacted with 69 ug Msc-rT3-OSu
(898 g/mol, 76 umol, two-fold excess with respect to
insulin derivative). After stirring for 3 h at room
temperature the acylation was stopped by addition of 50 ul
acetic acid. The solution was dialysed overnight against
distilled water and lyophilised. For cleavage of Msc
protecting groups the protein material was diluted in a
mixture of 1 ml dimethylformamide, 1.5 ml methanol and 1.5
ml water. The solution was cooled to 0°C and addition of
0.5 ml of ice-cold 2 M sodium hydroxide solution started
the cleavage reaction. The reaction was stopped by
acidification with 1 ml of 10~ (v/v) acetic acid. The
protein was precipitated by pipetting the reaction solution
into a mixture of 250 ml of ice-cold ether and 20 ml
methanol and stirring for 1 h. The ether was decantated
from the precipitated protein and the protein dried in
vacuo.
Purification of the raw material was performed by use
of RP-MPLC. Fractions were collected and lyophilised.
Chromatographic conditions:
Column: RP20C18, 2.5 x 250 mm, 122 ml total volume,
Gradient: 25-~40~ (v/v)
2-propanol in water containing 0.1~ trifluoro acetic
acid, total gradient volume 1.5 1; flow rate 20 ml / 3 min.
CA 02334859 2000-12-12
WO 99/65941 PCT/GB98/01722
6
Yield: 27 mg (10~ of theory, based on A1,B29-(Msc)2-
insulin)
Molecular mass: 6437 a ( calc. 6436.6 u)
Purity (RP-HPLC): 93 ~ (Absorption at 215 nm)
1.4 Mass spectrometry
MS-TOF spectrometer VG TofSpec, Fisons
Ionisation: Ar-laser, MCP Volts, : 1750, 337 nm, Linear
modus Acceleration: 20 kV
Standard: bovine insulin 5731 a (calc. 5731 u),
vasointestinal peptide 1424 a (talc. 1426 u) [rT3(Na-B1)]-
insulin: 6437 (calc. 6437)
Examt~le 2 - Effects of Binding Proteins on Receptor
Bindinct
The rT3-insulin conjugate made in Example 1 is used in
various tests to determine the binding potencies of the
analogues on liver plasma membrane. izs-Insulin is used as
the labelled insulin. It is known that insulin itself
inhibits binding of 12s-Insulin.
Results
Equilibrium binding curves
The equilibrium binding curves of average normalised
bound against the log-concentration of insulin or analogue
(nmol/1) with or without the presence of THBP were
generated. The trends initially illustrated by the curves
were:
H-Ins, rT3-Ins and T4-Ins appear similar in their
positions, i.e. there is no difference between them in
their ability to inhibit the binding of 12s-Insulin to
insulin receptors on LPM.
The presence of THBP does not appear to affect the
ability of H-Ins to inhibit the binding of 12s-Insulin to
insulin receptors on LPM.
CA 02334859 2000-12-12
WO 99/65941 PCT/GB98/01722
7
The presence of THBP does not appear to affect the
ability of rT3-Ins to inhibit the binding of lzs-Insulin to
insulin receptors on LPM.
The presence of THBP does appear to affect the ability
of T4-Ins to inhibit the binding of lzs-Insulin to insulin
receptors on LPM as shown by the shift in the T4-Ins+THHP
curves to the right . TBG seems to have the greatest effect
on T4-Ins, i.e. causes the greatest shift.
ED50
The ED50's as calculated by the G-PIP software were
inverse logged because the concentrations entered in G-PIP
had to be entered as the log of the concentrations. The
average (nmol/1)~ SEM of the ED50's was then calculated.
The results are shown in Table 1. These give a
quantitative idea of the shift, if any in the equilibrium
binding curves.
TABLE 1
Average of ED50 SEM
Average SEM n=
H-Ins 1.966 0.43 5
rT3-Ins 2.455 0.35 6
0.5% HSA 2.48 0.478 4
1% HSA 3.24 0.379 3
2.5% HSA 2.76 2
Transthyretin 1.805 0.55 4
0.135~.mo1/1 TBG I 3.147 I 0.35 3
T4-Ins 1.316 .034 5
0.5% HSA* 3.715 2
1% HSA* 5.823 2.108 3
2.5% HSA* 4.81 2
Transthyretin* 2.935 0.32 4
0.135~.mo1/1 TBG* 21.67 2.258 3
0.27~.mol/1 TBG* 36.55 2
CA 02334859 2000-12-12
WO 99/b5941 PCT/GB98/01722
8
* Fisher's test also performed.
Statistical analysis of the ED50's
From the statistical analysis it was found that the
ED50's of rT3-Ins and T4-Ins were not significantly
different from that of H-Ins. The ED50's of rT3-Ins with
THBP were not significantly different from those of rT3-Ins
without THBP present as determined by ANOVA. On the other
hand, the ED50's of T4-Ins without THBP present (p<0.05} as
determined by Fisher's least squares test (see Table 1*).
Potency estimates
The potency estimates of the analogues relative to H
Ins and the analogues in the presence of THBP relative to
the analogues in the absence of THBP are shown in Table 2
with their fiducial limits. This demonstrates that rT3-Ins
has a similar potency relative to H-Ins. T4-Ins seems to
have a greater potency relative to H-Ins. The presence of
THBP seems to have no effect on the binding potency
estimates of rT3-Ins binding to insulin receptors relative
to rT3-Ins without THBP present. However the presence of
THBP present. However the presence of THBP greatly reduces
the T4-Ins binding potency estimates relative to T4-Ins
binding to insulin receptors without THBP present (Table
2) .
CA 02334859 2000-12-12
WO 99/65941 PCT/GB98/01722
9
TABLE 2
Potency Estimates
Potency 95~ fiducial
limits
H-Ins 100%
rT3-Ins 94% 56-157
T4-Ins 184% 111-318
rT3-Ins 100%
0.5% HSA 122% 87-173
1% HSA 87% 58-129
2.5% HSA 119% 80-178
0.135~mo1/1 TBG 76% 54-107
Transthyretin 183% 111-306
T4-Ins 100%
0.5% HSA 27% 15-46
1% HSA 31% 16-54
2.5% HSA ' 35% 19-60
0.135~Cmo1/1 TBG 5% 2-9
Transthyretin X33% X20-54
Scatchard Plots
The Scatchard plot of H-Ins demonstrates the
characteristic curvilinear shape of negative co-operativity
that should be exhibited by human insulin. It may be seen
from the Scatchard plots of rT3-Tns and T4-Ins that these
analogues also exhibit negative co-operativity due to their
curvilinear shape.
Reference Example - Synthesis of Insulin - T4
The T4 insulin is B1-thyroxyl-insulin made according
to the technique described in WO-A-95/05187, Example 1.
