WO1994028917A1

WO1994028917A1 - Pharmaceutical formulations

Info

Publication number: WO1994028917A1
Application number: PCT/GB1994/001285
Authority: WO
Inventors: Stewart Craig; Shilpa Patel
Original assignee: British Biotech Pharmaceuticals Limited
Priority date: 1993-06-15
Filing date: 1994-06-15
Publication date: 1994-12-22
Also published as: AU6933794A; GB9312344D0

Abstract

Metal ions such as Cu2+ tend to promote multimerisation of stem cell inhibitors (SCIs) such as wild-type and variant macrophage inflammatory promoter-1α (MIP-1α). High order multimers are unwanted in both in processing and in pharmaceutical formulations. Therefore, in pharmaceutically useful and other formulations of SCIs, monomer formation is promoted by removing metal ions, such as by the use of EDTA or another chelating agent.

Description

PHARMACEUTICAL FORMULATIONS

This invention relates to pharmaceutical formulations. In particular, the invention relates to formulations of stem cell inhibitor (SCI) proteins including murine macrophage inflammatory protein-lα (muMIP-lα) , the human equivalent LD78 (huMIP-lα) and other proteins having the same or similar activity.

The haemopoietic system is composed of a range of diverse cell.-types and cell functions. All of the various cell lineages originate from the same multipotential stem cell population in bone marrow through a process of sequential division and differentiation. A proteinaceous factor which inhibited the proliferation of stem cells was originally isolated from murine bone marrow (Lord et al . , Bri t . J. Haematol . , 34, 441, (1976)) and subsequently shown to be the 8kDa protein, macrophage inflammatory protein-lot (MlP-lα) (Graham et al . , Nature (London), 344, 442, (1990)), a member of the intercrine family of cytokines. On the basis of sequence homologies, it was suggested (Schall, Cytok±ne, 3, 165, (1991)) and subsequently shown (Dunlop et al . , Blood, 79, 2221, (1992)) that the human tonsillar lymphocyte-derived 78 (LD78) gene product (Obaru eϋ al . , J.Biochem. , 99 , 885,

(1986)) was the human equivalent to muMIP-lα. The properties of these MlP-lα stem cell inhibitor proteins includes protection of progenitor stem cell populations from the toxic effects of cell cycle specific cytotoxic agents (Lord & Wright, Blood Cells, 6, 581, (1980) ; Lord et al . , Blood, 79, 2605, (1992); Dunlop et al . , Blood, 79, 2221, (1992)) . Stem cell inhibitor molecules, therefore, have enormous clinical potential in protecting the stem cells from chemotherapy (or radiotherapy) regimes used in the treatment of tumours. Reduction or prevention of the neutropaenia induced by administration of the cytotoxic agent would allow more aggressive and more frequent chemotherapy and result in more successful destruction of tumours and promotion of neutrophilia.

A major problem shared by the homologous proteins muMIP- lα and huMIP-lα, which limits their potential clinical utility, is that in physiologically relevant buffers they form large, soluble, heterogeneous multimeric complexes

(Wolpa et al . , J. Exp. Med. , 167, 570, (1988); Graham &

Pragnell, Dev. Biol . , 151, 377, (1992); Patel et al . ,

Biochemistry, 32, in press, (1993)). The soluble multimeric complexes show a broad heterogeneous mixture of molecular mass ranging from 8kDa up to 2000kDa. Self- association occurs via an equilibrium pathway of:

Monomer — Dimer — Tetramer — Dodecamer t— Higher Order Multimer

where the tetramer is the basic unit for higher order association. A proposed pathway for the higher order multimerisation constructed from hydrodynamic studies on wild-type and variant huMIP-lα is shown in Figure 1.

The principal consequence of the self-association behaviour is that clinical administration of the protein as a heterogeneous preparation could lead to varying efficacy, impaired tissue penetration and enhanced immunogenicity. For clinical administration, therefore, a homogeneous SCI preparation of defined molecular mass in a physiologically acceptable formulation is likely to be a prerequisite.

Approaches which may be taken to try and control the self-association behaviour of SCI's include: (a) Formulation of the protein molecule in an appropriate buffer or solution;

(b) Modification of one or more of the amino acid residues at the protein surface involved in promoting and/or stabilising the association of the components of a tetrameric or higher order complex; and

(c) Addition of an excipient to inhibit association of -^•protein units at a particular interface.

It has previously been shown that the self-association behaviour of wild-type SCI can be manipulated using certain solutions. In 150mM phosphate buffered saline

(PBS) pH7.4, wild-type huMIP-lα. exists as a heterogeneous population of molecules ranging in mass from 7.8kDa to

»250kDa (Patel et al . , Biochemistry (1993)) . In 30% acetonitrile, 0.1% TFA, wild-type huMIP-lα exists as a homogeneous population of 7.8kDa monomers (Mantel et al . , Proc. Nat 'l . Acad. Sci . USA, 90 2232 (1993)); Patel et al . , Biochemistry, (1993)) . In lOmM acetic acid pH3.2 or lOmM MES, 500mM NaCl pH6.4, wild-type huMIP-lα exists as a homogeneous population of 31.8kDa tetramers (Patel et al . , Biochemistry (1993)) . None of these solutions is physiologically acceptable for clinical administration and in fact acetonitrile is highly toxic.

It has been claimed (Mantel et al. , Proc. Nat 'l . Acad. Sci . USA, 90 2232 (1993)) that the monomeric species is the active form of the SCI protein and that solutions of monomer are 1000-fold more active than solutions containing multimers. In the in vi tro assay systems of the applicants (as will be described in relation to Example 1 and Figure 2) , the monomeric form of wild-type huMIP-lα is 3- to 4-fold more active than the multimeric form. It may, therefore, be clinically advantageous to administer SCI as a monomer to reduce the amount of protein required for an effective dose and in addition, reduce the cost.

The self-association properties of MIP-lo; can be manipulated by site-directed mutagenesis (as described in International Patent Application No. PCT/GB92/02390, a copy of which is annexed hereto and incorporated in this application) to give active variants of the molecule with defined quaternary structure in PBS. International Patent Application No. PCT/GB92/02390 describes the presence of a high affinity Cu²⁺ binding site in huMIP-lα, but does not suggest that the presence of that site can be exploited in any useful way.

It has now been discovered that metal ions promote the formation of high order multimers of natural and variant SCIs. The discovery arose as follows.

Size exclusion chro atography (SEC) of wild-type huMIP- lα in 12.5mM Tris, 1.0M glycine pH8.3 (as shown in Figure 3 and as described in International Patent Application No. PCT/GB92/02390) revealed that a significant proportion (§>70%) of protein eluted with a retention time equivalent to that expected for a monomer. By comparison, in 150mM PBS pH7.4, <1% of the wild-type protein elutes as a monomer in size exclusion chromatography. The quaternary structure of SCI appeared to be very sensitive either to pH > 8 or to a specific component of the Tris/glycine buffer. The high concentration of amino acid glycine does not give a high ionic strength, so dissociation must be related directly to an effect of the buffer components. Metal ions classically chelate to free amino and carboxyl groups of amino acids ("Chemistry of the Amino Acids", Volume 1, Chapter 6, Krieger Publishing, Florida, 1984, Greenstein & Winitz, eds.). Only a small population of dimeric huMIP-lα is evident in the SEC profile and no tetra eric species are present. It was postulated, therefore, that metal ions play a key role in stabilization of both the huMIP-lα tetramer and dimer units.

Size exclusion chromatography in the presence of EDTA and other chelators of multivalent metal ions was carried out and the results confirmed that multivalent metal ions were involved in stabilising the self- association of SCI.

It follows that the removal of metal ions, by chelating agents or other suitable means, enables the manipulation of the equilibrium of self-association in favour of, or at least towards, monomeric species of wild-type or variant stem cell inhibitors.

According to a first aspect of the invention, there is provided a pharmaceutical formulation comprising a stem cell inhibitor, the concentration of metal ions in the formulation being sufficiently low to avoid unwanted multimerisation of the stem cell inhibitor. Stem cell inhibitors may therefore be delivered systemically as a defined low molecular weight preparation.

Metal ions have been shown to be involved in some specific protein self-associations; for example, Zn²⁺ has been implicated in insulin multimerisation

(Hodgkin, Nature 255, 103 (1975)). Addition of l-3mM lysine and 5mM EDTA minimised aggregation of insulin in solution (Quinn and Andrade, J. Pharm. Sci , 72 1472

(1983) . However, the involvement of metal ions in the association/stabilisation of individual protein multimers cannot be predicted.

Other studies on huMIP-lα (Patel et al . , Biochemistry,

(1993)) show that ionic interactions are responsible for higher order association of tetramers, and that hydrophobic interactions (and inevitably hydrogen bonding) stabilise the tetramer unit itself. Removal of metal ion may, for example, primarily destabilise association of dimers (resulting in dissociation of the tetramer) or may destabilise both the dimer and tetramer association interfaces.

The stem cell inhibitor will generally be proteinaceous and can be wild-type murine or human MlP-lα, which may be prepared as described in WO-A-9104274 or WO-A- 9205198. The preferred wild-type molecule is a 69 amino acid form of LD78 described by Obaru et al J. Biochem. 99 885-894 (1986) . More preferable, however, is the use of a variant of murine or human MlP-lα containing one or more amino acid substitutions to control the higher order association of the molecule (as described in International Patent Application No. PCT/GB92/02390) . In this case formation of a lower order multimer can be achieved using a lower concentration of chelator, and/or a shorter incubation time; this is a significant advantage of the use of variants. Such variants can be formulated efficiently to monomers (at pharmacologically acceptable concentrations of a metal chelator) . Under these same conditions, only a partial change in the self- association of wild-type huMIP-lα can be achieved.

The term "variant" (or its synonym for present purposes "analogue") is used, broadly, in a functional sense. As a practical matter, though, most variants will have a high degree of homology with the prototype molecule if biological activity is to be substantially preserved. It will be realised that the nature of changes from the prototype molecule is more important than the number of them. As guidance, though, at the amino acid level, it may be that (in increasing order of preference) at least 40, 50, 60, 65, 67 or 68 of the residues will be the same as the prototype molecule; at the nucleic acid level, nucleic acid coding for an analogue may for example hybridise under stringent conditions (such as at approximately 35°C to 65°C in a salt solution of approximately 0.9 molar) to nucleic acid coding for the prototype molecule, or would do so but for the degeneracy of the genetic code.

The structures and preparation of preferred variants are as detailed in PCT/GB92/02390. There are in principle four stages in the association mechanism at which it is possible to prevent the formation of large multimers (and therefore aggregates) of SCIs. Inhibition of each of these stages could be influenced by a mutation in a different region of the SCI molecule.

First, further association of tetramers can be inhibited. Secondly, if the SCI dimers are prevented from associating to tetramers, then further multimerisation will be inhibited. Thirdly, SCI monomers may be prevented from dimerising. Fourthly, further association of dodecamers to higher order multimers can be inhibited. Any of these options can be implemented by specific mutation of residues involved in promoting and/or stabilising the association events. A further option would be to use a combination of mutations simultaneously to block two or all of the association events.

The following amino acid residues are preferred for modification:

(i) amino acid residues which could be involved in stabilising the interaction between two dimers; and

(ii) amino acid residues at surface regions, on the external faces of the tetramer, which could act as sites for higher order association.

Radical mutation of individual or combinations of key residues stabilising the association of dimers into tetramers will yield a dimeric recombinant SCI variant or analogue molecule. Similarly, mutation of residues at the sites of association of tetramers to multimers will yield a tetrameric SCI variant or analogue molecule. The amino acid modification preferably involves a substitution, although deletions and additions are contemplated within the scope of the invention.

The types of mutation preferred for producing the desired effects are:

(i) charge repulsions (successfully used to produce monomeric insulin; Dodson, Prospects in Protein Engineering Meeting- Abstracts, 49-53,

(1989)) ;

(ii) hydrophobic to hydrophilic changes;

(iii) neutral/hydrophobic to charged.

It is generally better not to substitute very hydrophobic residues into the protein in order to avoid contributing to the hydrophobic effect in association. Equally, it is preferred to avoid mutations which significantly disrupt secondary structural elements of the protein: so, for example, known ^-breakers are preferably not introduced into /3-sheet regions.

Certain types of mutation are most effective in producing desirable changes within the SCI molecule. These are : charge reversal; charged residue to neutral; hydrophobic to hydrophilic.

For optimum results substitutions should be made at particular sites within the molecule. The residues which should be altered are dependent on the level of multimerisation which is to be prevented.

The following discussion of preferred sites for mutation deals primarily with huMIP-lα (LD78) , the proposed structure of which is shown in Figure lb of PCT/GB92/02390. In Figure lb of that application, the ribbon traces the predicted path of backbone atoms for the huMIP-lα monomer. The labelled residues define the putative secondary structure elements. /3-sheet strand 1 runs from Phe23 to Thr30; /3-sheet strand 2 runs from Lys35 to Thr43; /3-sheet strand 3 runs from Ser46 to Pro53; and the C-terminal helix runs from Trp57 to Ala69. Analogous secondary structural elements may be inferred for other SCIs, including MIP-lo., for example using the amino acid alignment shown in Figure la of PCT/GB92/02390.

It is apparent that some faces of the monomer are involved in more than one part of the multimerisation pathway. The extent of disruption/inhibition of self- association in those faces is related to the nature of the amino acid substitution.

Inhibition of monomer to dimer formation can be achieved by one or more mutations, for example at residue 19 (He) or 39 (Val) . Either residue may be changed to Ala. Dimer to tetramer formation is affected by mutations in residues projecting away from the surface of the dimer in strand 1 of the β sheet, and/or in the turn between strands 2 and 3 of the sheet. Examples of the first region are amino acids 24-29 of huMIP-lo. (LD78) and of the second region are amino acids 43-47 of huMIP-lα (LD78) . In particular, Ile24>Asn, Tyr27>Asn, Phe28>Glu, Glu29>Arg, Lys44>Glu (especially with Arg45>Gln) and Arg 45>Glu are preferred.

Tetramer to dodecamer formation can be inhibited or disrupted by mutations of the nature described above in either the residues which form a chain N-terminal to the turn into strand 1 of the sheet (where two changes are preferred) , particularly residues 16-21, especially 17-19 or at position 4, 12, 26, 44, 48 or 66 of huMIP- lα (LD78) . In particular, Ala4>Glu, Phel2>Asp, Argl7>Ser, Asp26>Ala (especially with Glnl8>Glu) , Argl7>Glu (again especially with Glnl8>Glu) , Asp26>Ala, Lys44>Ser, Gln48>Glu (especially with Phe28>Glu) and Glu66 Ser are preferred.

Dodecamer to higher order multimer formation is prevented or disrupted by mutations at positions 12 to 21, especially positions 12, 18 and 21, of huMIP-lα.

(LD78) , or at position 65. In particular, Phel2>Gln,

Glnl8>Glu, Gln21>Ser and Leu65>Ala are preferred.

Generally preferred huMIP-lα analogues of the invention include molecules which comprise a sequence substantially corresponding to huMIP-lα, but with a mutation at one or more (but preferably no more than two) of the following amino acid residues: Serl, Leu2, Ala3, Ala4, Asp5, Thr6, Ala9, Phel2, Serl3, Tyrl4,

Serl6, Argl7, Glnlδ, Ilel9, Pro20, Gln21, Phe23, Ile24,

Asp26, Tyr27, Phe28, Glu29, Ser31, Ser32, Gln33, Ser35,

Lys36, Pro37, Gly38, Val39, Ile40, Leu42, Thr43, Lys44, Arg45, Ser46, Arg47, Gln48, Asp52, Glu55, Glu56, Gln59,

Lys60, Tyr61, Val62, Asp64, Leu65, Leu67, Glu66, Ser68, and Ala69.

Preferred huMIp-lα analogues in accordance with the invention include Lys44>Glu (with Arg45>Gln) , Arg4 >Glu, Phe28>Glu, Phe28>Glu (with Gln48>Glu) , Phe28>Glu (with Arg47>Glu) , Argl7>Ser (with Glnl8>Glu) , Phel2>Ala, Val39>Ala, Ile40>Ala, Asp26>Ala (with Glu29>Arg and Arg47>Glu) . More preferred huMIP-lα analogues in accordance with the invention include Argl7>Ser, Glu29>Arg, Glnl8>Glu, Asp26>Ser, Gln48>Ser, Thrl5>Ala, Gln21>Ser, Phe23>Ala, Ser32>Ala, Ala51>Ser, Ala4>Glu, Phel2>Asp, Asp26>Gln, Lys36>Glu, Lys44>Glu, Arg45>Glu, Glu66>Gln. The most preferred huMIP-lα analogues in accordance with the invention are

Phel2>Gln, Lys44>Ser, Argl7>Glu (with Glnl8>Glu) and, especially, Asp26>Ala and Glu66>Ser.

Generally preferred muMIP-lα analogues of the invention include molecules which comprise a sequence substantially corresponding to muMIP-lα, but with a mutation at one or more (but preferably not more than two) of the following amino acid residues: Alal, Pro2, Tyr3, Gly4, Ala5, Asp6, Thr7, AlalO, Phel3, Serl4, Tyrl5, Serl6, Argl7, Lyslδ, Ilel9, Pro20,Arg21, Phe23,

Ile24, Asp26, Phe28, Glu29, Ser31, Ser32, Glu33, Ser35, Gln36, Pro37, Gly38, Val39, Ile40, Leu42, Thr43, Lys44, Arg45, Asn46, Arg47, Gln48, Asp52, Glu55, Thr56, Gln59, Glu60, Tyrβl, Ile62, Asp64, Leu65, Glu66, Leu67, Asn68 and Ala69.

Preferred muMIP-lα: analogues of the invention correspond to the preferred huMIP-lα analogues described above.

Molecules for use in accordance with the invention will for preference be free of N-terminal extensions preceding Ser-1 (in the case of huMIP-lα) or Ala-1 (in the case of muMIP-lα) .

Engineered or natural variants containing an amino acid substitution at one or more of the aspartic acid or glutamic acid side-chains are especially preferred in the present invention. Examples include huMIP-lθ!(Asp26>Ala) , huMIP-lα(Glu56>Ser) , huMIP- lα (Phel2>Gln) , huMIP-lα (Argl7>Ser) , huMIP- lα(Glu66>Ser) , huMIP-lα(Asp26>Ser) and huMIP- lα(Phe23>Ala) .

The amount of stem cell inhibitor in the formulation will be sufficient to give an effective dose when administered as directed. The exact amount will depend on the nature and activity of the stem cell inhibitor present and in any event the amount administered will be under the control of the physician or clinician.

It may be desirable to prepare low metal ion or substantially metal ion-free formulations in accordance with the invention by a stage in the preparation of the formulation. For example, the stem cell inhibitor (and optionally some or all other ingredients of the formulation) may be brought into contact with means for removing or reducing the concentration of available metal ions; but those means need not form part of the formulation itself. Contact will last for a sufficiently long time for the metal ions to be sufficiently removed. In important embodiments of the invention, though, the formulation itself does contain means for removing or reducing the concentration of available metal ions.

Metal ions can be removed by a number of different ways. For example, metal ions can be precipitated as insoluble salts by the addition of appropriate anions. The anion will be chosen by reference to its acceptability in the context of the invention (for example it should not denature the active protein) and the solubility product of the salt formed with the metal ions in question.

'More convenient, however, is the use of chelating agents. There are a number of different chelators of metal ions which are permitted (at different concentrations per dose) for in vivo human administration. Of these, ethylenediaminetetraacetic acid (otherwise known as EDTA or edetic acid) is commonly regarded as an efficient chelator of multivalent metal ions such as for example Cu ⁺, Mg²⁺, Zn²⁺, Ca²⁺, Fe³⁺ & Pb²⁺. A bolus administration up to a concentration of 0.3% is permitted. Administration may be either intravenous or subcutaneous.

Other substituted carboxylic acids (particularly acetic acids) which can function as multidentate ligands with both nitrogen and oxygen bound to the metal may also be used in the invention. Examples include EGTA, EDBA (2,2' -ethylenediiminodibutyric acid) and ethylene glycol-Jbis- (3-aminoethyl ester) -N,N,N' ,N' -tetraacetic acid. EDBA is preferred as it is already clinically licensed; it has a higher affinity for zinc than copper.

Metal ions are used to chelate to free amino and carboxyl groups of amino acids to enable chemical modification of side chains in reaction mixtures (Chemistry of the Amino Acids, Volume 1, Chapter 6, Krieger Publishing, Florida, Greenstein & Winitz, eds. ) . It is feasible, therefore, that solutions of amino acid will be capable of complexing with metal and effectively removing it from solution. So a formulation of a suitable concentration of amino acid in an acceptable buffer could act as a chelator to dissociate SCI multimers.

The buffering agent citrate is commonly used in pharmaceutical formulations, as an anticoagulant (by chelation of Ca²⁺ ions) and for urological irrigations. Human MlP-lα has been used in vitro in isotonic citrate/NaCl (Rot et al . , J. Exp . Med. , 176, 1489,

(1992) ) , but there has been no suggestion that citrate may prevent higher order multimer formation of this or any other SCI; further, there was no suggestion of the sterility of the formulation that would be necessary for parenteral administration. A combination of citrate with a free amino acid or with EDTA could, therefore, provide a suitable metal chelating formulation for SCI. A combination of citric acid and arginine hydrochloride (Duffy et al . , US-A-4898826) has been shown to provide a beneficial formulation for the administration of tissue plasminogen activator (tPA) . No explanation or theory of beneficial effect is espoused in this patent; however, it is unlikely that the effect originates from a metal chelation event.

In the present invention, one or more chelating agents (or other metal ion removers) may be present at concentrations which are high enough to be effective but low enough not to have deleterious physiological effects. The optimal concentration to be used will depend on the nature of the SCI contained in the formulation, the nature of the chelating agent and the nature of the metal ion or ions whose removal from availability is sought. In the case of EDTA, a suitable concentration is ≤ lOmM. Usually, the concentration will be ≥ 3mM.

If chelating agents or other metal ion removers are not to form part of the desired formulation, alternative approaches within the invention can be adopted. As an illustration, a formulation of a stem cell inhibitor may be brought into contact with a stationary or solid phase capable of effectively removing metal ions for a suitable period of time. A specific example would be an EDTA-substituted resin over which the formulation, or ingredients of it, would pass, prior to packaging. The "suitable period of time" would depend on various factors including the nature of the metal ion removing agent, the nature of the metal ion, the nature of the protein and the temperature at which the operation is carried out. Although the optimun time may be determined by routine experiment, it seems that 36 hours may constitute a useful minimum guideline at room temperature in many cases.

If a stem cell inhibitor protein is distributed to users in a lyophilised or other non-aqueous state, directions may be provided to allow the SCI to be brought into contact with a metal ion removing agent, again for a suitable period of time, as discussed above.

Metal ions whose effective removal from availability is sought by means of the invention will generally be multivalent (that is to say divalent or higher) . Divalent and trivalent metal ions will be the multivalent ions likely to be encountered. Divalent metal ions include Cu²⁺, Mg ⁺, Zn²⁺, Ca²⁺ and Pb²⁺. Trivalent metal ions include Fe³⁺.

Formulations in accordance with the invention will generally be aqueous, with the SCI being dissolved in the water. Formulations intended for parenteral administration will be sterile. Water in the formulations may be provided as water for injections, physiological saline, phosphate buffered saline or any other buffer suitable for clinical administration

(which, if parenteral, may be intravenous, intramuscular or subcutaneous) . 150mM phosphate buffered saline, pH7.4, is generally regarded as equivalent to physiological ionic strength conditions and the buffer consists approximately of 137mM NaCl, 3mM KC1, lOmM phosphate. Ideally for systemic administration of an SCI-containing formulation, the ionic strength should be as near as possible to physiological. A formulation of ≤ lOmM EDTA, 150mM phosphate buffered saline pH7.4 would fulfil the requirements for formulation and administration of an SCI variant as a monomer.

Although formulations in accordance with the invention may primarily be for pharmaceutical use (which would include use on non-human animals) , non-pharmaceutical formulations also form part of the invention. This is because the use of metal chelators or other metal ion- removing means during production (or more specifically purification) of a stem cell inhibitor will give more homogeneous product, leading to better chromatography and making end formulation easier. Specifically, this may be applied for chromatographic separations such as size exclusion.

According to a second aspect of the invention, there is provided a process for preparing a stem cell inhibitor, the process comprising harvesting the stem cell inhibitor from a stem cell inhibitor-producing cell, removing metal ions from the stem cell inhibitor or from a medium in which it is contained and purifying the stem cell inhibitor, for example by chromatography.

According to a third aspect of the present invention, there is provided a citrate-free composition comprising a stem cell inhibitor, the concentration of metal ions in the composition being sufficiently low to avoid unwanted multimerisation of the stem cell inhibitor. Compositions in accordance with the third aspect do not have to be prepared to sterility or even to pharmaceutical grade in general; citrate is excluded because of the disclosure that human MlP-lα has been used in vi tro in isotonic citrate/NaCl (Rot et al . , J. Exp . Med. , 176, 1489, (1992)).

Preferred features of the second and third aspects of the invention are as for the first aspect, mutatis mutandis.

The invention will now be illustrated by the following examples. The examples refer to the accompanying drawings, in which:

FIGURE 1 shows a theoretical pathway of higher order association of SCI based on hydrodynamic studies of wild-type and variant huMIP-lα.

FIGURE 2 shows a comparison of the potency of monomeric and polymeric wild-type huMIP-lα in a calcium mobilisation assay. Monomer samples were reconstituted and diluted in 30% acetonitrile, 0.1%TFA. Polymer samples were reconstituted and diluted in 150mM PBS pH7.4. Samples were added to a final concentration of 3, 10, 30 or lOOng/ml in the FURA-2 assay described in Example 2.

FIGURE 3 demonstrates the presence of a significant population of monomeric huMIP-lα in the 12.5mM Tris, 1M glycine pH8.3 buffer. FIGURE 4 shows the size-exclusion elution profile of wild-type huMIP-lα in the presence and absence of lOmM EDTA in 150mM PBS pH7.4.

FIGURE 5 shows the size-exclusion elution profile of huMIP-lα(Asp26>Ala) in the presence and absence of lOmM EDTA in 150mM PBS pH7.4.

EXAMPLE 1 - Methods for the production of pure SCI molecules

Wild-type and variant SCI molecules may conveniently be obtained using synthetic genes, transfected into a recombinant host to express the protein and subsequently purified to homogeneity as described previously by Clements et al . , Cytokine, 4, 76 (1992); Patel et al . , Biochemistry,32, in press (1993) and International Patent Application No. PCT/GB92/02390. In the purification protocols described, SCI protein is eluted from the reverse-phase HPLC column in acetonitrile/trifluoracetic acid. The protein is subsequently lyophilized and stored at -70°C. In the examples below, freeze-dried aliquots of SCI protein are reconstituted in the relevant buffer systems detailed in the text.

EXAMPLE 2 - Potency assay for SCI molecules based on the mobilisation of intracellular calcium in response to receptor stimulation

The THP-1 cells required for the assay were grown in RPMI 1640 medium, 2mM glutamine, 50units/ml penicillin/ streptomycin, 0.05mM 3-mercaptoethanol, 10% heat inactivated (56°C for 30min.) foetal calf serum (Hyclone) . Cells were incubated at 37°C, 5% C0₂ in a Gallenkamp C0₂ incubator. Unless otherwise stated, all chemicals are GIBCO tissue culture grade.

Each individual assay measurement requires 2ml at 2-3 x 10⁶ cells/ml. At the start of the procedure, the log phase cells were counted, and then the required number of cells (21 x (2-3 x 10⁶) ) were spun at 1,000 r.p.m. for 5min. The cells were resuspended in fresh THP-1 growth medium at 2-3 x 10^e cells/ml and FURA-2AM

(Sigma) was added to a final concentration of lμM . the cells were incubated at 37°C in 5% C0₂ atmosphere for

45min, then spin for 5min at 1,000 r.p.m., resuspended and washed in 50ml of Tyrodes buffer pre-incubated at 37°C. the cells were spun again for 5min at 1,000 r.p.m. and resuspended in Tyrodes buffer at room temperature to give 2-3 x 10⁶ cells/ml . Wrap in foil to exclude stray light and maintain at room temperature.

Fluorescence emission intensity is measured using a Perkin-Elmer LS-50 fluorimeter, with a cell holder thermostatically controlled at 37°C (Grant LTD-6 thermocirculator) and with a built in magnetic stirrer

(normally set to the high speed setting) . The sample is excited at 340nm with a lOn bandwidth and emission is measured at 500nm with a 5nm bandwidth using the time- drive function to record over a 5min period. Hellma UV- grade 5 window quartz glass cuvettes with a 10mm pathlength and 3.5ml volume are used together with 7 x 2mm (len x dia) magnetic stirrer bars. The assay is set up in the following way: Mix the cell suspension, remove 2ml into an Eppendorf tube and spin for 5s in an Eppendorf centrifuge. Remove and discard the supernatant with a pipette into a suitable disposal container, replace Eppendorf lid and flick to resuspend pellet. Add 2ml of Tyrodes buffer (pre-incubated at 37°C) , pipette up and down twice to mix and then transfer to the cuvette in the fluorimeter cell holder. Add 20μl of the lOOmM stock CaCl₂ solution to give a final concentration of ImM in the cuvette and leave for 2min^"to equilibrate. Set the fluorimeter time-drive function to run for 300s and at time 0 start run. After 60s, add 20μl of the test sample (made up at lOOx required concentration) . After approximately 200s add 20μl of the digitonin stock solution and once the level has stabilised (F^^) , add 20μl of stock EDTA solution (F • )

The experimental protocol described by Mantel et al . (1993) {loc . ci t . ) for assay of SCI monomer requires a pre-dilution of the monomeric material (in aceto- nitrile/TFA) into PBS prior to assay (the time courses of which are over a number of hours to days) . This pre- dilution is claimed not to affect the monomeric state of the molecule significantly. Direct addition of sample in 20μl of acetonitrile/TFA does not affect the cells in the calcium mobilisation assay (described here) which has a response time of <2 seconds. It was, therefore, possible for us to add monomeric wild-type huMIP-lα directly into this functional assay with no pre-dilution. The results of a comparative study of the potency of wild-type huMIP-lα. in PBS and 30% acetonitrile, 0.1% TFA are shown in Figure 2. The results in Figure 2 show that on average, wild-type SCI prepared in acetonitrile/TFA is around 3-4 fold more active than the polymeric form in PBS. These data does not agree with the claim (Mantel et al . (1993) (loc . ci t . ) that a 1,000-fold increase in potency is obtained under these conditions. The present experimental protocol is actually more certain in terms of direct addition of monomer into the assay and the response is measured immediately.

EXAMPLE 3 - Size-exclusion chromatocrraphv of wild-tvpe human MIP-lo? in 12.5mM Tris, 1M Glycine PH8.3

lOOμg of wild-type human MIP-lo; was reconstituted in 0.2ml of 12.5mM Tris, 1M glycine pH8.3 and injected at a flow rate of 1ml/min onto a Superdex 75 (HR10/30) column equilibrated in the same buffer on the FPLC. Eluting fractions were detected by absorbance at 280nm.

The SEC elution profile (Figure 3) shows a major asymmetric peak of high molecular weight ( 70kDa) protein partially excluded from the column, a small dimer peak at approx. 15kDa and a major symmetrical peak corresponding to the monomer mass around 8kDa.

Sedimentation equilibrium studies (as described in PCT/GB92/02390) of huMIP-lo; at a protein concentration of 0.5mg/ml under these buffer conditions reveal the presence of a polydisperse population of mass species ranging from 8,000Da (M_w(T=0)) to >300,000Da (M_w(f=l)) .

The presence of a large population of monomeric huMIP- lo; in equilibrium with high molecular weight multimers suggests that the huMIP-lo; quaternary structure is extremely sensitive to these buffer conditions. The high concentration of amino acid glycine does not give a high ionic strength so dissociation must be related directly to an effect of the buffer components. huMIP- lQ? has a high affinity for the divalent metal Cu²⁺. Metal ions are classically used to chelate to free amino and carboxyl groups of amino acids to enable the modification of side-chain groups in reaction mixtures. Only a small population of di eric huMIP-lo; is evident in t e SEC profile and no tetrameric species are present. It appears, therefore, that metal ions play a key role in stabilization of both the huMIP-lo? tetramer and dimer units.

EXAMPLE 4 - Size-exclusion chromatocrraphy of wild-type human MIP-lo? in the presence and absence of EDTA in 150mM PBS PH7.4

Wild-type human MIP-lo; was reconstituted at 0.5mg/ml in 150mM PBS pH7.4 ± lOmM EDTA and pre-equilibrated at 4'C overnight before chromatography at 1ml/min on Superdex 75 (HR10/30) on the FPLC pre-equilibrated in the same buffer as the sample was prepared in. The elution profiles (Figure 4) show that the equilibrium of association only partially shifts towards the dimer and monomer species in the presence of the metal chelator EDTA at this concentration. It is clear, however, that the equilibrium predominantly favours the high molecular weight complexes. EXAMPLE 5 - Size-exclusion chromatoσraphv of human MIP-lo;(Asp26>Ala) in the presence and absence of EDTA in 150mM PBS PH7.4

The SEC elution profile of huMIP-lo;(Asp26>Ala) reconstituted and pre-equilibrated 0.5mg/ml in 150mM PBS ± lOmM EDTA as described in Example 4, is shown in Figure 5. (M10 is the designation of huMIP- lα;(Asp26>Ala) ) . In the absence of EDTA the protein elutes with the expected retention time of a tetramer. Following overnight incubation at 4°C in the presence of lOmM EDTA, profile B shows that >85% of the protein elutes as a monomer. This represents a major shift in the association equilibrium, and is very marked compared to the wild type (shown in Figure 4) . The change in mass of the sample after incubation with EDTA confirms that multivalent metal ions are involved in stabilising the self-association of MIP-lo;.

EXAMPLE 6 - Formulation of human MIP-lo;(Asp26>Ala) containing a chelator of multivalent metal ions intended for human administration

Current recommendations for the use of the chelator EDTA as an anticoagulant in vivo state concentration limits of 0.1% - 0.3% (w/v) depending on the volume of administration and intended endpoint. The upper limit of 0.3% corresponds approximately to a lOmM concentration of EDTA in a small volume bolus dose. An ideal formulation for administration of an SCI would be a vehicle close in nature to physiological ionic strength/composition. Such a vehicle would be 150mM phosphate buffered saline pH7.4, well known to those skilled in the art as equivalent to physiological ionic strength and pH.

Specifically for the huMIP-lα variant containing a substitution of Aspartic acid-26 to Alanine, a protein concentration of 0.5mg/ml in lOmM EDTA (=0.3%(w/v) ) , 137mM NaCl, 3mM KCl, lOmM phosphate pH 7.2 - 7.6 is recommended as the formulation.

The protein can be formulated in this buffer during production/ purification from a recombinant expression system (typically by desalting on a gel-filtration column or a diafiltration system) and subsequently stored prior to administration. Alternatively if a lyophilized preparation is available, the protein should be reconstituted in the above formulation and allowed to equilibrate at 4°C for a minimum of 36hr prior to administration.

The bolus administration of huMIP-lo;(Asp26>Ala) at 0.5mg/ml under these conditions will ideally not exceed a total volume of 10ml per dose.

Claims

1. A pharmaceutical formulation comprising a stem cell inhibitor, the concentration of metal ions in the formulation being sufficiently low to avoid unwanted multimerisation of the stem cell inhibitor.

2. A pharmaceutical formulation as claimed in claim

1, wherein the stem cell inhibitor is wild-type murine or human MIP-lo; or a variant thereof.

3. A pharmaceutical formulation as claimed in claim

2, wherein the variant contains an amino acid substitution at one or more aspartic acid or glutamic acid residues.

4. A pharmaceutical formulation as claimed in claim

3, wherein the stem cell inhibitor is huMIP-lo?(Asp26>Ala) , huMIP-lo?(Glu56>Ser) , huMIP- lα (Phel2>Gln) , huMIP-lo? (Argl7>Ser) , huMIP- lo?(Glu66>Ser) , huMIP-lo;(Asp26>Ser) or huMIP- lα(Phe23>Ala) .

5. A pharmaceutical formulation as claimed in any one of claims 1 to 4 containing means for removing or reducing the concentration of available metal ions.

6. A pharmaceutical formulation as claimed in claim 5, wherein the means for removing or reducing the concentration of available metal ions comprises a chelating agent.

7. A pharmaceutical formulation as claimed in claim

6, wherein the chelating agent is ethylenediaminetetra- acetic acid.

8. A pharmaceutical formulation as claimed in claim

7, wherein the chelating agent is present at an amount of from 3 mM to 10 mM.

9. A pharmaceutical formulation as claimed in claim 6, wherein the chelating agent is an amino acid.

10. A pharmaceutical formulation as claimed in any one of claims 1 to 9 comprising phosphate buffered saline.

11. A process for preparing a stem cell inhibitor, the process comprising harvesting the stem cell inhibitor from a stem cell inhibitor-producing cell, removing metal ions from the stem cell inhibitor or from a medium in which it is contained and purifying the stem cell inhibitor, for example by chromatography.

12. A citrate-free composition comprising a stem cell inhibitor, the concentration of metal ions in the composition being sufficiently low to avoid unwanted multimerisation of the stem cell inhibitor.