(Received for publication, December 24, 1996, and in revised form, February 28, 1997)
From the Departments of Alanine scanning mutagenesis has been used to
identify specific side chains of insulin which strongly influence
binding to the insulin receptor. A total of 21 new insulin analog
constructs were made, and in addition 7 high pressure liquid
chromatography-purified analogs were tested, covering alanine
substitutions in positions B1, B2, B3, B4, B8, B9, B10, B11, B12, B13,
B16, B17, B18, B20, B21, B22, B26, A4, A8, A9, A12, A13, A14, A15, A16,
A17, A19, and A21. Binding data on the analogs revealed that the
alanine mutations that were most disruptive for binding were at
positions TyrA19, GlyB8, LeuB11, and GluB13, resulting in decreases in
affinity of 1,000-, 33-, 14-, and 8-fold, respectively, relative to
wild-type insulin. In contrast, alanine substitutions at positions
GlyB20, ArgB22, and SerA9 resulted in an increase in affinity for the insulin receptor. The most striking finding is that B20Ala insulin retains high affinity binding to the receptor. GlyB20 is conserved in
insulins from different species, and in the structure of the B-chain it
appears to be essential for the shift from the Insulin mediates its effects by binding to the insulin receptor in
the plasma membrane of target cells. The molecular mechanisms for
insulin interaction with the receptor are not fully understood. The
crystal structure of the insulin molecule has been known for more than
25 years (1), but it remains an open question whether the structure of
insulin that binds to the receptor is similar to the crystal structure.
Until the structure of bound insulin and the side chains that are
actually involved in binding is identified by co-crystallization of the
receptor and ligand, more information about the binding domain on
insulin can be obtained using mutational approaches.
The binding domain of the insulin molecule has been studied by
investigating receptor binding of a number of insulins from different
animal species as well as chemically modified and more recently
genetically engineered insulins (2-4). These studies have provided
experimental support for a model in which invariant residues from both
A and B chains form a surface that binds to the insulin receptor. The
putative binding domain comprises a number of residues overlapping the
dimer-forming surface (ValB12, TyrB16, GlyB23, PheB24, PheB25, TyrB26,
GlyA1, GlnA5, TyrA19, and AsnA21) and some of the residues buried
beneath the COOH terminus of the B-chain (IleA2, ValA3, GluA4) (2).
Cross-linking studies with an azidophenylalanine-substituted analog
have shown that one of these residues, PheB25, comes into close
proximity to the insulin receptor upon binding (5).
Recently, a second binding site has been proposed, involving residues
LeuA13 and LeuB17 (6, 7). A biphasic binding reaction involving this
second binding site could explain the negative cooperativity phenomenon
(8) as well as discrepancies between receptor binding data and
metabolic potencies found in hystricomorph insulins and synthetic
analogs with mutations at these positions (9, 10).
Although a large number of insulin analogs have been studied to date,
no comprehensive analysis of the insulin side chains involved in
receptor binding has been reported. We have applied alanine scanning
mutagenesis to elucidate further the role of individual amino acid
residues in receptor binding. A total of 21 new insulin constructs with
alanine substitutions were expressed as single chain insulin precursors
in the yeast Saccharomyces cerevisiae. Yeast culture medium
was treated with Achromobacter lyticus protease to yield
mature insulin and then used directly in the binding assay. The
receptor used for the binding assay is an
IR1-IgG fusion protein immobilized in
microtiter plates coated with protein A.
The binding affinity of 19 new analogs and 7 HPLC-purified alanine
analogs was measured. Compiling these data with data in the literature
on alanine analogs allows an extensive overview covering the effect of
single alanine substitutions at a total of 38 positions of the insulin
molecule. We conclude that insulin residue TyrA19 is essential for
binding to receptor, whereas single alanine mutations in the B20-B22
turn actually appear to enhance affinity for insulin receptor.
Insulin, A14-125I-insulin, and
A. lyticus protease were from Novo Nordisk. DNA restriction
enzymes and T4 ligase were from New England Biolabs. Taq
polymerase was from Perkin-Elmer. Staphylococcus aureus
protein A, Sepharose FF for immobilizing A. lyticus
protease, and the SP-Sepharose BigBeads were from Pharmacia Biotech
Inc.
Single alanine
mutations were inserted in the sequence encoding insulin precursor by
polymerase chain reaction or overlap polymerase chain reaction using
appropriate oligonucleotides (11). The polymerase chain reaction
fragment was purified and subcloned into the yeast expression vector
cPOT using standard procedures (12). The yeast expression system used
has been described in detail previously (13). Briefly, the insulin
precursor was coupled to a 41-amino acid synthetic prepropeptide
sequence in the configuration: prepropeptide-KR-extension-insulin
precursor, where KR is the potential Kex2 protease processing site, the
extension is the removable spacer peptide E(EA)3EPK, and
the insulin precursor is either miniproinsulin B(1-29)-A(1-21) or
B(1-29)-AAK-A(1-21) (14, 15). Thus the precursors are single chain
proinsulin-like peptides that either lack the C-peptide or have the
short synthetic C-peptide AAK.
The insulin precursor constructs were expressed in S. cerevisiae strain MT663 (16). Yeast cells were transformed and
selected on YP plates with 2% glucose according to standard methods
(17). Yeast cultures were grown in YP medium with 2% glucose for
72 h at 30 °C, and the yield of insulin precursor in the
fermentation supernatants was quantified by HPLC with human insulin as
external standard (18).
The single chain
insulin precursors were converted to des-B30-insulin using the
lysine-specific A. lyticus endoprotease that removes the
extension and the C-peptide of the precursor. Cell-free yeast culture
medium with precursor was adjusted to pH 9 with 0.1 volume 0.5 M glycine, pH 9, and applied to a column with A. lyticus protease immobilized on Sepharose. The column was
incubated for 2 h at 37 °C before cleaved precursor was eluted
from column into 0.1 volume of 1 M Tris, pH 7.0, and the
concentration of des-B30-insulin was quantified by HPLC (18).
For analogs that were expressed poorly the precursor was partially
purified by cation exchange chromatography prior to treatment with
A. lyticus protease. Briefly, the cell-free yeast medium was
adjusted to pH 3.0 with HCl and then batch treated with cation exchange
resin (SP-Sepharose BigBeads) for 16 h at room temperature to
adsorb peptides, which were subsequently eluted with 200 mM NaCl, 50 mM glycine, pH 9.0. The HPLC-purified analogs
(B2Ala, B8Ala, B17Ala, B18Ala, B26Ala, A8Ala, and A21Ala) were
fermented in yeast and purified as described (4, 14).
The insulin receptor used is
a soluble fusion protein consisting of the insulin receptor (exon 11+)
ectodomain fused to the Fc region of the IgG heavy chain (19). The
IR-IgG fusion was stably expressed in baby hamster kidney cells by
inserting the 3,532-base pair NotI/XbaI fragment
from the pBluescript II-HIRs-Fc vector (19) into the ZEM expression
vector and transfecting this into baby hamster kidney cells as
described previously (20, 21).
For binding experiments the IR-IgG fusion protein was immobilized on
protein A-coated microtiter plates. First the wells were coated with
protein A by incubating for 1 h with 50 µl of protein A (1 µg/ml) in TBS (150 mM NaCl, 10 mM Tris, pH
7.5); then the plates were washed three times with binding buffer (100 mM Hepes, pH 8.0, 100 mM NaCl, 10 mM MgCl2, 0.5% (w/v) bovine serum albumin, 0.025% (w/v) Triton X-100). The IR-IgG fusion receptor was immobilized by adding 50 µl of a 33-fold dilution of baby hamster kidney
supernatant in binding buffer to each well and incubating for 3 h
before washing the plates three times with binding buffer. Binding
experiments were performed by adding a total volume of 150 µl of
binding buffer with A14-125I-insulin (10 pM)
and varying dilutions of A. lyticus protease-treated insulin
analog. After 16 h at 4 °C unbound ligand was removed by
aspirating the buffer and washing once with 200 µl of cold binding
buffer, and the radioactivity in each well was counted. The
concentration of receptor in the well was adjusted to yield 10-15%
binding when no competing ligand was added in the assay.
The data from the competitive displacement experiments were fitted to a
two-site binding model using a nonlinear regression algorithm in
GraphPad Prism 2.01 (GraphPad Software, San Diego). All sets of binding
data fitted better to the two-site model than the one-site model.
For the alanine
scanning mutagenesis we made 21 new insulin analog constructs with
alanine substitutions in positions B1, B3, B4, B9, B10, B11, B12, B13,
B16, B20, B21, B22, A4, A9, A12, A13, A14, A15, A16, A17, and A19 (Fig.
1). We chose these 21 positions because they are located
primarily on the surface of the insulin molecule and therefore are more
likely to be candidates involved in receptor binding.
The insulin precursors were expressed in the yeast S. cerevisiae and fermented in 5-ml cultures. The yield of insulin
precursor in the fermentation supernatant was quantified on HPLC. The
yield of the alanine analog precursors relative to the wild-type
insulin precursor is shown in Fig. 2. Most of the
alanine substitutions result in some decrease in yield, probably
because the yeast expression system is optimized for expressing the
wild-type insulin precursor. Nevertheless, 15 of the analogs gave more
than 20% of the yield of the wild-type precursor, indicating that they
are processed and folded correctly by the yeast cell. The positions
where the alanine substitutions had the most deleterious effects on
expression levels were at the hydrophobic residues LeuB11, ValB12, and
LeuA16. Mutations in the region GlyB20, GluB21, and ArgB22 also
resulted in poor yields, indicating that the precursor is folded less
efficiently because of structural changes induced by alanine mutations
in this region.
Because of the low level of expression the B11Ala, B12Ala, and
A16Ala insulin analogs were partially purified using cation exchange
chromatography. The B20Ala and B22Ala analog data were obtained from
both partially purified and raw medium supernatants. Two of the insulin
analogs (B12Ala and A16Ala) never yielded sufficient material for
binding experiments.
The precursors were converted to des-B30-insulin by treating with
A. lyticus protease. HPLC chromatograms for the wild-type insulin precursor before and after A. lyticus protease
treatment are shown in Fig. 3. The A. lyticus
protease-treated precursor elutes later from the HPLC column than
untreated precursor because of increased hydrophobicity, and this
allowed precise quantification of the converted precursor.
For the binding assay an IR-IgG fusion
protein was immobilized in microtiter plates using protein A. This
binding assay allows efficient analyses of the high number of analogs,
and in addition this assay has lower background than precipitation
assays. The binding data obtained when displacing
A14-125I-insulin from the IR-IgG receptor with insulin
standard (Fig. 4A) fits to a two-site binding
model and yields binding affinities (ED50) of 0.012 and 1.4 nM for the two predicted sites. However, the 0.012 nM value for the high affinity site is not reliable because
it is similar to the concentration of tracer added (10 pM),
and therefore all binding data presented reflect binding to the low
affinity site, that is, affinities in the range of 0.3-200
nM. The binding affinity of each analog is based on at least three independent sets of binding curves.
Binding of the wild-type insulin precursor expressed in the yeast cells
before and after treatment with A. lyticus protease is
included in Fig. 4A. The affinity of the wild-type precursor after A. lyticus protease treatment (des-B30-insulin) is
similar to the affinity of insulin (ED50 0.016 and 1.5 nM), whereas the curve for the uncleaved insulin precursor
is shifted far to the right (Fig. 4A), demonstrating a very
poor affinity of the uncleaved precursor for the insulin receptor. This
is consistent with previous reports on the biological activity of the
miniproinsulin B(1-29)-A(1-21) (15). The assay was further verified
when measuring affinities of the HPLC-purified analogs. Five of these
have been investigated previously, and the relative affinities obtained
for the IR-IgG and the full-length insulin receptor are
similar2 (Table I). These
controls demonstrate that the in vitro IR-IgG binding assay
performed directly on A. lyticus protease-treated yeast
culture supernatants reproduces previous assays using the full-length
receptor.
Table I.
Alanine scanning receptor binding data
Insulin Research, ¶ Cell
Technology, and
Protein Chemistry, Novo Nordisk, 2880 Bagsvaerd,
Denmark, the ** Department of Medicine,
Department of Biochemistry and Molecular
Biology, and the §§ Howard Hughes Medical
Institute, University of Chicago, Chicago, Illinois 60637
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
FOOTNOTES
ACKNOWLEDGEMENTS
REFERENCES
-helix B8-B19 to the
-turn B20-B22. Thus, replacing GlyB20 with alanine most likely
modifies the structure of the B-chain in this region, but this
structural change appears to enhance binding to the insulin receptor.
Miscellaneous
Alanine Constructs and Expression in Yeast
Fig. 1.
Overview of insulin analogs with single
alanine substitution. Amino acid sequence of insulin with
positions of alanine substitutions indicated by A. New
analogs and purified analogs (HPLC-purified) are described under
"Materials and Methods." Literature analogs have been
described previously (1, 22-26).
[View Larger Version of this Image (31K GIF file)]
Fig. 2.
Yield of insulin analogs with alanine
mutation. The figure shows the yield of mutated precursor relative
to the yield of wild-type insulin precursor for each position of
alanine substitution. The yeast S. cerevisiae was
transformed with cPOT plasmid carrying cDNA encoding mutated
insulin precursor. The yield of precursors was quantified by
HPLC.
[View Larger Version of this Image (31K GIF file)]
Fig. 3.
HPLC analysis of insulin precursor.
Chromatograms showing the HPLC elution profile after applying culture
supernatant from yeast expressing wild-type insulin precursor. In
panel A the supernatant was untreated, and in panel
B the supernatant was converted to des-B30-insulin by treating
with A. lyticus protease (panel B).
[View Larger Version of this Image (14K GIF file)]
Fig. 4.
Competition curves for
125I-insulin binding to insulin receptor. IR-IgG
receptor immobilized in microtiter plates was incubated with
125I-insulin (10 pM) for 16 h at 4 °C
together with varying dilutions of unlabeled analog. The amount of
125I-insulin bound, as a percentage of
125I-insulin bound in the absence of unlabeled analog, is
plotted against the concentration of unlabeled analog. Panel
A, 125I-insulin displaced with insulin (), A. lyticus protease-treated wild-type insulin precursor (
), or
untreated wild-type insulin precursor (
). Panel B,
125I-insulin displaced with the following three A. lyticus protease-treated precursors: wild-type insulin (
),
B11Ala insulin (
), or A19Ala insulin (
).
[View Larger Version of this Image (14K GIF file)]
A-chain
B-chain
Position
Affinity experimentally determined
Affinity
reported in literature
Position
Affinity experimentally
determined
Affinity reported in literature
% of insulin
%
of insulin
% of insulin
%
of insulin
A1
G
12
(22)
B1
F
79
± 26
A2
I
0.6 (23)
B2
V
110 ± 21
A3
V
1.8 (23)
B3
N
134 ± 21
A4
E
139 ± 35
B4
Q
54 ± 16
A5
Q
49 (24)
B5
H
31 (24)
A6
C
B6
L
1.4 (25)
A7
C
B7
C
A8
T
87
± 13
78a
B8
G
3
± 2
A9
S
260 ± 87
B9
S
80 ± 23
A10
I
B10
H
80 ± 24
A11
C
B11
L
7 ± 1
A12
S
108 ± 28
B12
V
A13
L
30 ± 7
B13
E
12 ± 3
A14
Y
66
± 13
B14
A
110 (insulin)
A15
Q
122 ± 37
B15
L
A16
L
B16
Y
69 ± 15
A17
E
56 ± 20
B17
L
62
± 14
24a
A18
N
B18
V
88
± 17
35a
A19
Y
0.1
B19
C
A20
C
B20
G
270 ± 85
A21
N
66
± 5
140a
B21
E
230 ± 79
B22
R
405 ± 187
B23
G
3 (26)
B24
F
5 (26)
B25
F
10 (26)
B26
Y
36 ± 8
64a
B27
T
B28
P
B29
K
76 (34)
B30
T
102 (porcine insulin)
a
L. Schäffer, unpublished data.
In all of the binding experiments performed the purified insulin and the wild-type precursor were included as standards, and all data are expressed as the ED50 of analog relative to the ED50 of the wild-type insulin precursor.
Three analogs tested showed a marked decrease in affinity for the
receptor, supporting a role for these residues in receptor binding. The
most pronounced decrease occurred with TyrA19 Ala. We could not
obtain a full binding curve due to the poor affinity; the curve for the
A19Ala analog is shifted about three decades to the right compared with
the wild-type insulin curve (Fig. 4B), implying a 1,000-fold
decreased affinity for the insulin receptor. Other disruptive alanine
substitutions were found at LeuB11 and GluB13, resulting in a drop to 7 and 12%, respectively, in affinity for the receptor (Fig.
4B). One of the HPLC-purified analogs, B8Ala insulin, also
disrupted binding, with an approximate decrease of 30-fold in affinity
for the receptor. Although residues LeuA13 and LeuB17 have been
suggested to be part of a second binding site in the insulin molecule,
alanine substitutions at these positions had only modest effect on the
binding affinity (2-3-fold) (Table I).
In contrast to the disruptive mutations there were a number of alanine substitutions that did not alter affinity for the receptor, and at least three of the alanine analogs even showed an increase in receptor affinity. Alanine substitution for residues GlyB20, ArgB22, and SerA9 consistently resulted in a 2-4-fold increase in affinity for the insulin receptor (Table I).
The binding data are summarized in Table I and Fig. 5.
When including data on alanine analogs reported in the literature (1,
22-26) the effects of alanine substitutions at a total of 38 positions
of the insulin molecule are now available.
By implementing a new high volume screening assay for insulin analogs we have used the alanine scanning mutagenesis approach to identify amino acid side chains of insulin which strongly influence binding to the insulin receptor. A total of 21 single alanine mutants of insulin were expressed in yeast and converted to des-B30-insulin using A. lyticus endoprotease. The binding constants of the analogs for the IR-IgG fusion receptor were determined in a microtiter plate assay based on immobilization of receptor using protein A. This receptor retains high affinity for insulin (19), and immobilization via protein A does not influence binding affinity (data not shown).
In the present study the binding affinity of 19 new analogs and 7 HPLC-purified analogs was measured. Compiling these data with analogs described in the literature allows an extensive overview covering the effect of substituting alanine at 38 positions of the insulin molecule (Table I and Fig. 5).
Two types of information are obtained in this mutagenesis study. First, mutations that disrupt binding are of interest because they indicate that the residue of insulin either interacts directly with the receptor or, alternatively, that the residue supports a conformation required for receptor binding. Additionally, scanning mutagenesis provides a test to confirm the role of residues thought to be essential for receptor binding. If alanine substituted at these positions does not alter binding, the structure in this particular region does not appear to be important for binding.
In addition to binding data, comparison of analog yields in the S. cerevisiae expression system reveals differences in the biosynthesis of the analogs. Most of the alanine-substituted analogs were expressed efficiently; 15 of the analogs gave more than 20% of the yield of the wild-type precursor, indicating that these analogs were folded efficiently by the yeast cell (Fig. 2). The positions where the alanine substitution had the most deleterious effect on expression level were at the hydrophobic residues LeuB11, ValB12, and LeuA16. These amino acids are buried in the hydrophobic core of the insulin molecule and therefore might provide the free energy needed for efficient folding. Poor yields were also observed in analogs with mutations in the loop region B20-B22 (Fig. 2).
Binding data on the new analogs revealed that the alanine mutations that were most disruptive for binding were at positions TyrA19, LeuB11, and GluB13, resulting in decrease in affinity to 0.1, 7, and 12%, respectively, relative to wild-type insulin (Table I). The A19Ala analog is well expressed in the yeast expression system (Fig. 2), indicating that the folding is not altered by this mutation, pointing to a direct role of TyrA19 in interaction with the receptor. Three modifications of insulin at position A19 have been described previously (27-29). Replacing TyrA19 with leucine had a dramatic effect on receptor binding, decreasing the affinity 1,000-fold, whereas substitution with phenylalanine only resulted in a 4-5-fold decrease in affinity (27, 29). Additionally, iodination at TyrA19 had an intermediate effect on receptor binding (for review, see Ref. 28). Thus the substitution of alanine or leucine at position A19 disrupts receptor binding, whereas iodination or insertion of another aromatic ring at position A19 (Phe) is well tolerated, suggesting that an aromatic ring at position A19 is crucial for insulin binding to its receptor.
Including analogs reported in the literature (Table I) the alanine
mutations that cause the largest reduction in binding are found at
positions LeuB6, GlyB8, GlyB23, PheB24, IleA2, ValA3, and TyrA19.
Substitution of alanine at these positions reduced affinity for the
receptor by more than 20-fold (Table I). All of these residues are
evolutionary conserved in insulins from different species, supporting
the conclusion that they are important for structure and/or function.
Most of these disruptive alanine mutations are related to the classical
binding surface. The residues that are most likely to be directly
involved in binding are B23, B24, and A19, which play a role in dimer
formation, and A2 and A3, which are hydrophobic residues buried beneath
the COOH terminus of the B-chain (Fig. 6). We propose
that the residues that are essential for direct interaction with the
receptor are found in this patch, but there may also be structural
effects that can account for the disruption of binding when replacing
these residues with alanine. In particular, interaction between A2 and
A19 appears to stabilize the insulin conformation (30). Mutating one of these residues thus could distort the conformation of insulin that
binds to the receptor. On the other hand, the high expression yield of
the A19Ala analog argues against major conformational distortions in
this molecule.
The explanation for the large reduction in receptor affinity in insulin
analogs LeuB6Ala and GlyB8Ala is not clear. A glycine residue is not
likely to contribute free energy of binding, and in addition GlyB8 is
thought to be important for initiating the central -helix of the
B-chain. Therefore, mutating this residue probably has structural
consequences. The B6Ala insulin was described by Nakagawa and Tager
(25), and they also speculated on a structural role for the LeuB6 in
insulin.
Substituting alanine for other residues of the classical putative binding surface including GluA4, AsnA21, TyrB16, and TyrB26 only altered affinity for the insulin receptor moderately (less than 3-fold) (Table I), indicating that these residues are not part of the functional binding epitope.
These data are incorporated in the structural model shown in Fig. 6. The five residues GlyB23, PheB24, IleA2, ValA3, and TyrA19 form a patch in the insulin molecule. Substituting alanine for any of these five amino acids results in more than a 20-fold decrease in receptor affinity, and therefore we propose that this patch is essential for direct interaction with the insulin receptor. In the structure of insulin, part of this proposed binding patch is buried beneath the COOH terminus of the B-chain; if all of these residues interact directly with the receptor partial unfolding of the COOH-terminal B-chain would be required for exposing this patch. Such a mechanism would be in accordance with the hypothesis of Dodson and others that some degree of separation of COOH-terminal B-chain from NH2-terminal A-chain is required for interaction with the insulin receptor (15, 31).
Two recent papers have suggested that in addition to the classical binding site there is a second binding site in the hexamer-forming surface of the insulin molecule (6, 7). This second binding surface includes LeuA13 and LeuB17. In the present study alanine substitutions at these positions did not dramatically decrease affinity for the insulin receptor, the affinities of the A13Ala and B17Ala analogs were 30 and 52%, respectively (Table I). However, as we only look at the low affinity site of the IR-IgG fusion we may miss effects of A13 and B17 mutations. This would be in agreement with Schäffer (6), who reported that a characteristic of analogs mutated in positions A13 or B17 is that they bind with relative higher affinity to the soluble insulin receptor (nanomolar affinity site) than to the high affinity site of the holo-receptor (picomolar affinity) (6). Thus it would be expected that mutations in the second binding site would have less impact on binding to the low affinity site on the insulin receptor.
Three of the alanine substitutions resulted in a 2-4-fold increase in
affinity for the insulin receptor fusion. These were SerA9Ala,
ArgB22Ala, and most surprisingly GlyB20Ala (Table I). The expression
yield of B20Ala and B22Ala was very low (Fig. 2) indicating that
substituting alanine at these positions has structural consequences.
GlyB20 is conserved in insulins from different species, but apparently
it is not important for binding; thus it would be interesting to
investigate the effect of the GlyB20Ala mutation on folding of the
wild-type precursor proinsulin. In the structure of the insulin B-chain
GlyB20 appears to be essential for the shift from the -helix B8-B19
to the
-turn B20-B22 (Fig. 6). Analyzing the impact of alanine at
position B20 on the secondary structure of an insulin B-chain using the
Chou-Fasman algorithm (32) suggests that the
-helix B8-B19 will be
extended by an additional turn comprising B20Ala, GluB21, and ArgB22.
This does not take into account the interaction with the A-chain, but
substituting glycine at position B20 with alanine nevertheless is
expected to cause conformational changes in this region. What effect
this putative structural change could have on the COOH terminus of the
B-chain is uncertain. Earlier reports on the natural mutation of PheB24
to serine (insulin Los Angeles) suggested that unfolding of the B-chain
is compatible with high affinity receptor binding (31); it may be that
similar effects will be seen when the structure of the B20Ala insulin
is solved.
In conclusion, the present activity data on series of analogs will in combination with structure elucidation of selected analogs further characterize the functional properties of the insulin molecule.
We thank Durita Simonsen, Ulla M. Jørgensen, and Marianne Kallesøe for excellent technical assistance; Knud Vad for the B1Ala-insulin construct; and Per Balschmidt for the A. lyticus protease-Sepharose.