EP0769022A1

EP0769022A1 - O-glycosylated authentic igf-i and truncated variants thereof, a method of preparation thereof and pharmaceutical compositions

Info

Publication number: EP0769022A1
Application number: EP95925201A
Authority: EP
Inventors: Rolf Eriksson; Stig Johansson; Harriet RÖNNHOLM; Anna Skottner; Karin Sonesson
Original assignee: Pharmacia and Upjohn AB
Current assignee: Pfizer Health AB
Priority date: 1994-07-01
Filing date: 1995-06-22
Publication date: 1997-04-23
Also published as: WO1996001275A1; CA2193399A1; IL114270A0; SE9402332D0; JPH10502623A; AU2940695A

Abstract

The invention relates to O-glycosylated authentic IGF-I, having three or more mannose residues to the Ser69 amino acid of the IGF-polypeptide chain or being truncated and having two mannose residues to the Thr29 amino acid of the truncated IGF-polypeptide chain. The O-glycosylated IGF-I according to the invention have substantially no insulin receptor affinity, which makes them valuable for use in therapeutic treatment. The invention also relates to a method of obtaining IGF-I as defined above by expressing IGF-I in yeast cells and isolating the claimed O-glycosylated IGF-I from the medium and to pharmaceutical composition containing these IGF-I together with a pharmaceutically acceptable carrier, diluent or excipient. It also relates to the use of the O-glycosylated IGF-I in the preparation of a medicament for the treatment of e.g. growth deficiency and insulin resistant states.

Description

O-GLYCOSYLATED AUTHENTIC IGF-I AND TRUNCATED VARIANTS

THEREOF, A METHOD OF PREPARATION THEREOF AND

PHARMACEUTICAL COMPOSITIONS.

The invention relates to O-glycosylated authentic IGF-I, having three or more

mannose residues to the Ser69 amino acid of the IGF- polypeptide chain or

being truncated and having two mannose residues to the Thr29 amino acid of

the truncated IGF- polypeptide chain. The O-glycosylated IGF-I according to

the invention have substantially no insulin receptor affinity, which makes them

valuable for use in therapeutic treatment.

The invention also relates to a method of obtaining IGF-I as defined above by

expressing IGF-I in yeast cells and isolating the claimed O-glycosylated IGF-I

from the medium, and to pharmaceutical compositions which contain these

IGF-I together with a pharmaceutically acceptable carrier, diluent or excipient. It also relates to the use of the O-glycosylated IGF-I in the preparation of a

medicament for the treatment of e.g. growth deficiency and insulin resistant

states.

INTRODUCTION

Insulin-like growth factor- 1 (IGF-I) is a growth factor comprising 70 amino acid

as a single polypeptide chain, which displays relatively high homology with

proinsulin.

It is known to partly mediate the effect of growth hormone, in order to

promote cell growth and differentiation and also to display insulin-like

properties. The endogenous human form is non-glycosylated.

Recombinant human insulin-like growth factor I (rhIGF-I) can be produced by

expression in the yeast S. cerevisiae and is transported into the fermentation

medium by virtue of a secretion system. Expression of foreign proteins in yeast

can lead to post-translational modifications. One such modification that has been observed in proteins produced in yeast is glycosylation.

Protein glycosylation in yeast can occur both as N- and O-linked, but since IGF-

I lacks the recognition site for N-glycosylation (Asn-Xaa-Ser/Thr) only de novo

O-glycosylation is to be expected in IGF-I. One site for O-glycosylation at Thr29

has been determined, see WO 90/02198, Gellerfors, P.et al. (1989), "Isolation and characterization of a glycosylated variant of human insulin-like growth

factor I produced in Saccharomyces cerevisiae", J. Biol. Chem., 264, 11444-11449

and Elliott, S., et al. (1990), "Yeast-derived recombinant human insulin-like

growth factor I: Production, purification, and structural characterization", J.

Prot. Chem., 9, 95-104).

Yeast is known to contain carboxypeptidases (JONES, E. W., (1990), "Tackling

the protease problem in Saccharomyces cerevisi e", Methods EnzymoL, 194, 428-

453). Hence, the expression of foreign proteins in yeast may also lead to

production of C-terminally truncated variants.

There is known from WO 90/02198 a glycosylated IGF-I variant of IGF-I which

has two or more mannose residues attached to the Thr 29 amino acid of the

IGF-I polypeptide chain. It is suggested that this glycosylated variant of IGF-I

has a more pronounced effect in lowering the blood glucose than authentic

IGF-I.

Truncated variants of IGF-I are earlier known from WO 91/18621, which

discloses des (1-3-) - IGF-I as useful in the treatment of diabetes. The truncated

variant des(l-3) IGF-I is also known to reduce the severity of CNS damage.

(WO 93/02695).

WO 93/08826 discloses (4-70) IGF -I and (54-67)IGF-I, the truncated variants

of which promote the survival of retinal neuronal cells in ophthalmic

compositions. WO 87/01038 discloses peptide analogues of IGF-I in which 1-5 amino acid

residues are absent from the N-terminal. They are said to have increased

biological potency, which is useful in e.g. treating growth deficiency and

catabolic conditions.

The authentic IGF-I in which the glutamic acid at position 3 is replaced by

another amino acid or deleted is also known, see WO 91/10348.

We describe in this document the isolation and characterization of two other

variants of glycosylated authentic IGF-I (gIGF-I) and of two new variants of C-

terminally truncated IGF-I which have been isolated and characterized.

We have thus isolated an authentic O-glycosylated IGF-I, having four or five

mannose residues to the Ser69 amino acid of the IGF- polypeptide chain and an

O-glycosylated IGF-I, having three mannose residues to the Ser69 amino acid

of the IGF- polypeptide chain.

We have also isolated a variant lacking the C-terminal Ala70 residue and

glycosylated at Thr29. Another variant lacked the C-terminal tripeptide

Lys68Ser69Ala70, and furthermore glycosylated at Thr29.

These new O-glycosylated variants of IGF-I appear to be promising

therapeutical agents, especially in the view of the less adverse effects such as decreased insulin-like effects although having a normal, expected IGF-I effect .

Thus, they can be given in higher doses than human authentic IGF-I in the

treatment of growth disorders, possibly without causing problems with

hypoglycaemia. Furthermore, due to decreased binding to IGF binding protein

1 (IGFBP-1) they can be used for treatment of patients suffering from insulin

resistance following e.g. trauma, insulin receptor antibody production, poorly

controlled diabetes or intrauterine growth retardation (IUGR),

Figures 1-6 illustrate the invention.

Figure 1 shows the IGF-I peptides after digestion with S. aureus V8 protease

(SAP-V8). The peptides obtained are labelled according to their appearance in

the primary structure.

Figure 2 shows the purification step based on hydrophobic interaction of

rhIGF-I. Fraction 8 was collected for the isolation of the characterized new IGF-I

variants.

Figure 3 shows the elution profile of affinity-chromatography of the "fraction 8"

on a Concanavalin A column. Two pools (x and z) were collected from repeated

preparations.

Figure 4 shows the elution profile of reversed phase chromatography (RP-

HPLC) of the fraction x pool, cf Figure 3.

Figure 5 shows the peptide map of SAP V8 digests of the two main variants, z and x2, compared to authentic rhIGF-I (DSQ 93). Figure 6 shows the peptide map of SAP V8 digests of the x3, x4 and x5

variants, compared to authentic rhIGF-I (DSQ 93).

The invention relates to O-glycosylated authentic IGF-I, having three or more

mannose residues to the Ser69 amino acid of the IGF- polypeptide chain or

being truncated and having two mannose residues to the Thr29 amino acid of

the truncated IGF- polypeptide chain. The O-glycosylated IGF-I according to

the invention have substantially no insulin receptor affinity, which makes them

valuable for use in therapeutic treatment.

It relates especially to those having four and /or five mannose residues to the

Ser69 amino acid of the IGF- polypeptide chain or having three mannose

residues to the Serg9 amino acid.

It also relates especially to truncated variants of O-glycosylated IGF-I which

are O-glycosylated (1-67)IGF-I, having two mannose residues to the Thr29

amino acid of the IGF- polypeptide chain or O-glycosylated (1-69)IGF-I,

having two mannose residues to the Thr29 amino acid of the IGF- polypeptide

chain.

The invention also relates to a method of obtaining IGF-I as defined above by

expressing IGF-I in yeast cells, and isolating the claimed O-glycosylated IGF-I

from the medium. It also relates to pharmaceutical compositions containing these IGF-I together

with a pharmaceutically acceptable carrier, diluent or excipient and a method

of preparing a pharmaceutical composition by mixing the claimed IGF-I

variant with a pharmaceutically acceptable carrier, diluent or excipient.

The claimed variants of IGF-I could be useful in the preparation of a

medicament for the treatment of growth deficiency or for anabolic effects. .

Furthermore, due to deceased binding to IGF binding protein 1 (IGFBP-1) they

can be used for treatment of patients suffering from insulin resistant state

following e.g. following trauma, genetic defects in insulin receptor functions,

auto antibody production towards the insulin receptor and intrauterine growth

retardation (IUGR), and also for the treatment of patients with abnormally

elevated IGFBP levels, particularly IGFBP1.

The new glycosylated IGF variants may be administrated in doses ranging

from 20 μg/kg to 1 mg/kg, preferably between 20 and 250 μg/kg.

Isolation and characterization of glycosylated variants

The IGF-I gene was expressed in Saccharomyces cerevisiae using an alpha-

mating factor leader peptide-IGF-I expression plasmid, p539/12.

The process is described in detail in WO 90/02198 page 6 under Example to

page 7. The yeast fermentation medium contained authentic human IGF-I and

variants thereof. The first separation of the IGF-variants from authentic human

IGF-I, was achieved by hydrophobic interaction chromatography (HI-HPLC). The fraction before the peak of authentic rIGF-I (fraction 8) was collected for the

isolation of the glycosylated variants. Se Figure 2.

Chromatography on a concanavalin A (ConA) affinity column was an effective

way of separating glycosylated IGF-I from the native form. Figure 3 shows the

elution profile of affinity-chromatography of the "fraction 8" on a Concanavalin

A column. Two partially resolved fractions were collected as two pools, x and

z. These pools were collected from repeated preparations.

Sodium dodecylsulphate polyacrylamide gel electrophoresis (SDS-PAGE)

followed by ConA blotting indicated that z was relatively homogeneous while

x consisted of two main bands.

Carbohydrate analysis of the purified variants using GC-MS (KENNE, L., and

STROMBERG, S. (1990), "A method for the microanalysis of hexoses in

glycoproteins", Carbohydr. Res., 198, 173-179.) showed that both variants x and

z contained only one type of sugar, mannose, although in different relative

quantities (data not shown).

The x-fraction was further purified on RPV-HPLC and the main peak was

further characterized. Figure 4, showing the peaks x2, x3, x4 and x5. Immunob lotting with an antibody specific for the C-terminus of IGF-I, revealed

some further differences between the variants. Hence, modifications at the C-

terminus were to be expected.

The structural differences between the glycosylated IGF-I variants was further

delineated using a SAP V8 enzymatic mapping process.

This allowed changes to be assigned to specific peptides by comparison with

native peptides. Figure 5 shows the peptide map of SAP V8 digests of the two

main variants, z and x2, compared with authentic rhIGF-I (DSQ 93) and Figure

6 shows the peptide map of SAP V8 digests of the x3, x4 and x5 variants,

compared with authentic rhIGF-I (DSQ 93).

In the SAP V8 map, the only modified fragment of z was the C-terminal SAP 9

(Figure 1 for nomenclature). The amino acid composition of the SAP 9 fragment

was found to be correct (Table I), this Table presenting the results of amino acid

analysis of the SAP peptides in question), while the mass analysis gave a

molecular mass 484.1 units higher than the theoretical value 1279.6 (Table II,

which presents the results of mass spectrometry of the SAP peptides in

question.). This difference in mass fitted well with the suggestion of three

additional mannose residues, the molecular mass of mannose being 162.1. Mass

spectrometry of the intact polypeptide, Table III, presents the results of mass

spectrometry of the non-digested IGF-I variants and supports this suggestion. As with variant z, the only modified SAP fragment of x2 was SAP 9, see Figure

5. Furthermore, it was observed that the fragment was not pure. The amino acid

composition was unchanged also in the case of x2, (Table I). For the SAP 9

fragment of x2, the mass analysis gave a molecular mass 645.9 units higher than

the theoretical value 1279.6 (Table III). Hence, in this case the suggested number

of additional mannose residues would be 4. On the other hand, mass analysis of

the undigested polypeptide pointed rather to 5 mannose residues (Table III),

although this value was not entirely correct. An explanation might be that x2

was a mixture of tetra- and pentasubstituted IGF-I, a heterogeneity also

indicated both by SDS-PAGE and as a shoulder on the SAP 9 peak (Figure 5).

Different preparations may then have contained different relative amounts of

the two variants.

Regarding the amino acid sequence of the SAP 9 fragment, the only residue

known to enable O-glycosylation was Ser69, thus this was the proposed

glycosylation site.

In the SAP V8 map, the modified SAP 5 fragment common to x3, x4 and x5

(Figure 6) was caused by dimannose glycosylation at Thr 29. The respective C-

terminal SAP 9 fragments of x4 and x5 were also modified. The SAP V8 map of variant x4 revealed two modifications of the SAP 9

fragment, labelled SAP 9:1 and SAP 9:2 (Figure 1). According to the amino acid

sequence analysis (Table I), one Ala residue was missing in both fragments,

supposedly the C-terminal Ala70. Mass analysis of fragment SAP 9:1 supported

this modification. As for the SAP 9:2 fragment, mass analysis gave an additional

26.6 units (Table II). This discrepancy was consistent when mass analysis was

performed on the undigested polypeptide (Table III). No rational explanation

has been found for this difference in molecular mass.

The fragment corresponding to SAP 9 of variant x5 had a retention time slightly

longer than x4 SAP 9:2, see Figure 4. Amino acid analysis (Table I) of the

fragment revealed the lack of one lysine, one serine, and one alanine residue,

proposed a truncation of the C-terminal tripeptide Lys68-Ser69-Ala70. This suggested modification was confirmed by mass analysis of both the SAP 9

fragment (Table II), and the undigested polypeptide (Table III).

It has thus been confirmed that:

By x2 is meant O-glycosylated authentic IGF-I, having four or five mannose

residues to the Ser69 amino acid of the IGF- polypeptide chain.

By z is meant O-glycosylated authentic IGF-I, having three mannose residues

to the Ser69 amino acid of the IGF- polypeptide chain. By x4 is meant O-glycosylated (1-69)IGF-I, having two mannose residues to the

Thr29 amino acid of the IGF- polypeptide chain.

By x5 is meant O-glycosylated (1-67)IGF-I, having two mannose residues to the

Thr29 amino acid of the IGF- polypeptide chain.

Table I. Amino acid analysis of peptides obtained from SAP V8 digestion of

glycosylated IGF-I variants. Values are given as mol/mol. The theoretical

values as predicted from intact peptides + indicates trace amounts are given in

paran thesis.

Sample x4 x4 x4 x5 x5 z x2 peptide SAP 5 SAP 9:1 SAP 9:2 SAP 5 SAP 9 SAP9 SAP 9

Asx 1.0 (1) 1.0 (1)

Thr 1.0 (1) 0.9 (1)

Ser 1.0 (1) 0.9 (1) nd (1) 1.0 (1) 0.9(1)

Glx

Gly 3.1 (3) 3.0 (3)

Ala 2.1 (3) 2.0 (3) 1.9 (3) 3.2 (3) 2.9 (3)

Cys nd (1) nd (1) nd (1) nd (l) nd (l) Val

Met + (i) + (1) + (1) nd (l) + (1) lie

Leu 1.1 (1) 1.1 (1) 1.1 (1) 1.1 (1) 1.0 (1)

Tyr 1.9 (2) 0.9 (1) 1.0 (1) 2.0 (2) 1.1 (1) 0.9 (1) 1.0 (1) Phe 1.9 (2) 2.1 (2)

Lys 1.0 (1) 2.0 (2) 2.0 (2) 0.9 (1) 0.9 (2) 2.1 (2) 2.0 (2)

Arg 1.0 (1) 1.1 (1)

Pro 0.9 (1) 1.9 (2) 2.0 (2) 1.0 (1) 2.0 (2) 1.8 (2) 2.2 (2)

SAP 5, SAP 9:1, SAP 9:2, SAP 5 and SAP 9 are shown in Figures 5 and 6.

nd: not detected. Table II Plasma desorption mass spectrometry of SAP V8 derived peptides of

the characterized glycosylated IGF-I variants.

Theoretical mass:

SAP 5 1406.6

SAP 9 1279.6

Sample experimental mass suggested theor. mass

mass value difference modification difference

z SAP 9 1763.7 +484.1 +3 mannos +486.4

x2 SAP 9 1925.5 +645.9 +4 mannos +648.6 x4 SAP 5 1728.5 +321.9 +2 mannos +324.3

x4 SAP 9:1 1207.4 -72.2 -Ala -71.1

x4 SAP 9:2 1235.1 -44.5 -Ala, +? -71.1

X5 SAP 9 992.4 -287.2 -LysSerAla -286.3

Table III Plasma desorption mass spectrometry of intact molecules of the

characterized glycosylated IGF-I variants.

Theoretical mass:

IGF-I 7648.7

Sample experimental mass suggested theor. mass mass value difference modification difference

z 8131.7 +483.0 +3 mannos +486.4

x2 8435.4 +786.7 +4 mannos +648.6

+5 mannos +810.7

x4 7927.6 +278.9 +2 man,-Ala,+? +253.2

x5 7683.0 +34.3 +2 mannos +38.0

-LysSerAla

BIOLOGICAL ACTIVITY

Nuclear magnetic resonance data indicate that the core of IGF-I is structurally

very similar to insulin. Structural determinants for binding IGF-I to the Insulin

and Type I receptor in this region overlap. Consequently, in addition to

binding its own receptor IGF-I crossreacts with the insulin receptor, although in

most cases with a much lower affinity. The structure-function relationship of the IGF-I receptor is also similar to the

insulin receptor. Apart from the different binding affinities for their respective

ligands, the insulin receptor and the IGF-I receptor seem to be identical at least

with regard to the initial steps in signal transduction following ligand binding.

This suggests that bioeffect differences between IGF-I and insulin are mainly

due to differential expression of their receptors and signalling pathways in

different target cells. An important determinant for the target organ bioeffects

of IGF-I or IGF-I variants is therefore their affinity to, and the bioeffects

mediated by binding to the two receptor types.

Bioavailability and the actions of the IGFs are also influenced by their

association with a group of specific binding proteins (IGF BP 1-6). IGF BP3 is the major circulating IGF BP form in adults, and normally binds

more than 95% of IGF-I in serum in a 150 kDa ternary complex. The ternary

complex has a relatively long half-life and is unable to leave circulation intact.

Since only the free, uncomplexed, forms of IGF are considered to be

biologically active, it is necessary to provide specific mechanisms by means of

which IGF can be shuttled from the circulation to the target cells for receptor

binding. IGF BPl is believed to be important both for the transport of IGF-I to the target

cell and for regulating the IGF responsiveness of target cells. Studies support a

physiological role for IGFBP-1 in modulating IGF availability for glucose

homeostasis. IGFBP-1 has been shown to inhibit the insulin-like activities of IGF-I, both in vitro and in vivo. IGFBP-1 displays a pattern of regulation

characteristic for glucose counterregulators, and is the only IGFBP which

displays rapid fluctuations in serum with a pronounced diurnal rhythm. There

exists an inverse relationship between circulating insulin and IGFBP-1 and

insulin is known to regulate the production of hepatic IGFBP-1.

IGFBP-1 not only acutely regulates IGF action with respect to blood sugar

control (as well as other insulin-like actions), but also appears to be highly

important in controlling growth in the long term. Clinical observations point to

an inverse correlation between circulating IGFBP-1 levels and various growth

parameters. Low levels are found at puberty concomitant with peak IGF-I

levels and accelerated growth. Conversely, the stunted growth of poorly

controlled diabetes is accompanied by a decrease in IGFBP-3 and IGF-I while

IGFBP-1 serum levels are high. IGFBP-1 is further believed to control fetal

growth, and elevated levels correlate with low weight at birth.

It is therefore obvious that the distribution of IGF between the various IGFBP

forms is an important determinant for clearance, tissue distribution and

bioactivity of IGF. It follows that IGF variants with altered affinities for binding to the IGFBPs concomitant with changes in receptor binding, may demonstrate

dramatically different bioprofiles. The ability of the glycosylated IGF-I variants to bind to the IGF-I receptor was

tested in a receptor binding assay using a human placenta membrane

preparation (Hall K et al., A. J. Clin. Endocrinol. Metabol. 39:973-976, 1974).

The z and x2 variants both demonstrated a reduction of IGF-I receptor binding

(60 - 70%, compared to the non-glycosylated authentic IGF-I preparation.

The 1-69 dimannosyl Thr29 variant (x4) demonstrated a potency of 50% and

the 1-67 dimannosyl Thr29 variant (x5) demonstrated a potency of 30% relative

to authentic human IGF-I. See Table IV.

Thus, type I receptor binding was not significantly reduced by glycosylation in

pos Thr 29, but when combined with further C-terminal truncations produced a

graded and significant loss of receptor binding potency.

The in vitro bone growth stimulatory effects of the glycosylated rh IGF-I

variants were tested in an embryonal chick femur assay (Endo H et al, Nature

(London), 286:262-264, 1980). The stimulatory effects of IGF-I in this model are

primarily mediated by the chick IGF-I receptor. All glycosylated rhIGF-I

variants were found to be equipotent to the non-glycosylated authentic IGF-I

regarding bone growth stimulation. See Table IV.

This either suggest that the ability to bind the chick IGF-I receptor is intact, or

that the binding potency for this receptor is marginally affected, as for the

human IGF-I receptor, but that this effect is compensated for e.g. by an altered

affinity to the IGF-I binding proteins. Local binding protein for IGF-I in the

bone are identified. The growth promoting effect of the glycosylated variants was thus more or less

unchanged.

When the insulin-like properties of the IGF-I variants was tested in a lipogenic

assay using primary rat adipocytes (Small J et al., J. Biol. Chem. 262:11071-79,

1987) a major reduction of lipogenic activity was revealed. The lipogenic

potency of all variants was reduced to 30% or less compared to non-

glycosylated authentic rhIGF-I. See Table IV.

This indicates that the ability of these IGF-I analogues to bind the insulin

receptor is severely impaired.

Since it is known that primary rat adipocytes carry few, if any, functional IGF-I

receptors, it is suggested that O-glycosylation leads to a reduction in insulin-

receptor binding, and hence the reduced insulin-like potency. This is

advantageous to patients in need of IGF-I for growth effects.

Affinities to the large BP3 (BP = binding protein) and the small molecular

weight BPl was tested by a Biacore method. This method measures binding

of IGF to IGFBPs immobilized on a biosensor chip surface by a surface plasmon

resonance technique (Jonsson U, Bio Techniques 11:5, 1991). The z and x2

variants demonstrated a reduced association with the IGFBPl (approximately

60% of the non-glycosylated authentic IGF-I), while the IGFBP3 affinity of the

two variants was unchanged. See Table IV. The results suggest that administered Ser69 variants will form the BP complex

similar to authentic IGF-I in contrast to the earlier characterized 1-70 Thr29

variant, which will not. For all variants a reduced IGFBPl association suggests

a higher bioavailability and less inhibition of bioactivity.

In conclusion, in vitro data indicate an altered pharmacological profile resulting

from O-glycosylation on position Ser69 and from O-glycosylation on position

Thr29 when truncated. The reduced insulin-like activity together with the

altered IGFBP 1 affinity are expected to lead to changes in bioeffect selectivity

and availability that may be used to advantage in clinical situation.

TABLE IV

Comparison of Biological effects in vitro. The potency of gIGF-I variants are

given relative to the non-glycosylated authentic rhIGF-I.

BIOLOGICAL EFFECT

authentic 1-70 IGF x2 X4 X5

IGF-I Thr 29

IGF-I Receptor binding

(hplacenta) 1 0.9 0.6 0.7 0.5 0.3

Growth of chick embryo

femora 1 1 1 1 1 1

Lipogenesis in primary rat

adipocytes 1 0.3 0.3 0.3 <0.2* <0.3

IGF BPl binding 1 0.1 0.6 0.6 n.d. n.d.

IGF BP3 binding 1 0.25 ^' 1 1 n.d. n.d.

n.d.= not determined

Not statistically different from nonstimulated controls. CONCLUSIONS

A second site for O-glycosylation of rhIGF-I has been localized in fragment SAP

9, at residue Ser69. The degree of glycosylation appeared to range from three to

five mannose residues. The mannosylated(Ser69) rhIGF-I variants characterized

were not glycosylated at other sites.

The biological results show that all of the described variants of IGF-I have

normal, expected IGF-I effect on in vitro growth of chick embryo femora,

despite of IGF-I receptor binding being either similar to authentic IGF-I (1-70

Thr29 IGF) or reduced (z x2, x4, x5 variants). Although this was unexpected, it

is nevertheless of considerable interest, since it indicates that a loss of receptor

binding can be compensated for by other intrinsic properties of the variants e.g.

the demonstrated changes in IGFBP associations. The variants have an

increased selectivity for IGF-I receptors over insulin receptors as indicated by

the lipogenisis assay. This assay indicates that the variants can be selective for

bone and muscles but not for fat or liver, since the latter tissues are known to

contain low concentration of receptors for IGF-I and significant concentrations

of insulin receptors.

As the insulin-like effect is reduced, less adverse effects such as hypoglycaemia

are expected and the dose may be higher. All variants are of potential interest clinically as they

1/ offer a graded receptor selectivity which may be used to achieve selective

strengthening of the classical IGF-I effects on cellproliferation and

differentiation at the expense of insulin receptor mediated "insulin-like" effects,

2/ are likely to demonstrate differences in kinetics and bioavailability as a

result of either the presence or absence of glycosylation and due to changed

affinities to IGF Binding proteins (IGFBPs).

Both C-terminal truncation and glycosylation important. Decreased

lipogenesis versus authentic IGF-I has been shown for all variants.

The IGF BPl binding is lower than for authentic IGF-I but not as low as for the

glycosylated variant with mannose on Thr29-

The IGF BP3 binding is the same as for IGF-I indicating similarities in half-life.

The earlier identified glycosylated variant with mannose on Thr29, has a

much lower IGFBP3 binding.

Claims

1. O-glycosylated authentic IGF-I, having three or more mannose

residues to the Ser69 amino acid of the IGF- polypeptide chain or

being truncated and having two mannose residues to the Thr29

amino acid of the truncated IGF- polypeptide chain.

2. O-glycosylated IGF-I according to claim 1 having substantially

no insulin receptor affinity.

3. O-glycosylated IGF-I according to claim 1 or 2 having four or

five mannose residues to the Ser69 amino acid of the IGF-

polypeptide chain.

4. O-glycosylated IGF-I according to claim 1 or 2 having three

mannose residues to the Ser69 amino acid of the IGF- polypeptide

chain.

5. Truncated variant of O-glycosylated IGF-I according to claim 1

or 2, which is an O-glycosylated (1-67)IGF-I, having two mannose

residues to the Thr29 amino acid of the IGF- polypeptide chain.

6. Truncated variant of O-glycosylated IGF-I according to claim 1

or 2, which is an O-glycosylated (1-69)IGF-I, having two mannose

residues to the Thr29 amino acid of the IGF- polypeptide chain.

7. A method of obtaining O-glycosylated IGF-I according to any

of claims 1 to 6 by expressing IGF-I in yeast cells, and isolating the

claimed O-glycosylated IGF-I from the medium.

8. A pharmaceutical composition containing O-glycosylated IGF-I

according to any of claims 1 to 6 together with a pharmaceutically

acceptable carrier, diluent or excipient.

9. A method of preparing a pharmaceutical composition

comprising mixing O-glycosylated IGF-I according to any of

claims 1 to 6 with a pharmaceutically acceptable carrier, diluent

or excipient.

10. Use of O-glycosylated IGF-I according to any of claims 1-6 in

the preparation of a medicament for the treatment of growth

deficiency.

11. Use of O-glycosylated IGF-I according to any of claims 1-6 in

the preparation of a medicament for the treatment of insulin

resistant states e.g. following trauma, genetic defects in insulin

receptor functions, auto antibody production towards the insulin

receptor and intrauterine growth retardation (IUGR).

12. Use of O-glycosylated IGF-I according to any of claims 1-6 in

the preparation of a medicament for the treatment of patients with abnormally elevated IGFBP levels, particularly IGFBPl.