WO1993000367A1

WO1993000367A1 - Abrin variants and immunnotoxins

Info

Publication number: WO1993000367A1
Application number: PCT/EP1992/001419
Authority: WO
Inventors: Aslak Godal
Original assignee: Hafslund Nycomed As
Priority date: 1991-06-21
Filing date: 1992-06-20
Publication date: 1993-01-07
Also published as: GB9113475D0; NO934700D0; EP0590011A1; JPH06510525A; NO934700L

Abstract

Certain naturally occurring holotoxin variants of the plant toxin abrin are disclosed. The holotoxin variants have lost cell-binding activity and may be coupled to ligands such as antibodies or lectins to produce immunotoxins. The immunotoxins may be formulated into pharmaceutical compositions.

Description

ABRIN VARIANTS AND IMMUNOTOXINS

This invention relates to novel variants of the toxin abrin and to immunotoxins based thereon.

Abrin consists of two subunits (called A- and B-chains) linked together with a single disulfide bridge and having different functions. The B-chain binds to galactose-containing receptors on the cell surface and, following internalization of the molecule, the A-chain is liberated into cytosol where it inhibits cellular protein synthesis by enzymatic inactivation of the 60S ribosomal subunit (Olsnes & Pihl, 1982).

Immunotoxins (IT's) are conjugates of highly potent toxins or active fragments thereof with monoclonal antibodies (MoAbs) , and they represent a relatively new class of compounds designed to have cell type-specific activity. The rationale for their use is that the IT will bind specifically to cells expressing the appropriate antigen, followed by cellular uptake of the IT and intracellular release of the toxic moiety. If the antigen expression is restricted to a certain cell type (e.g. cancer cells) , specific cytotoxicity may be achieved (Rev. in Frankel, 1988).

In addition to antibody-mediated binding, the IT may bind via the toxin B-chain (where present) to toxin receptors which are present on most mammalian cells, causing unspecific activity (Thorpe & Ross, 1982) . To avoid this problem many investigators have omitted the B-chain by conjugating the toxin A-chain directly to the MoAb. However, such A-chain conjugates often display a much lower activity than their corresponding holotoxin counterparts (Vallera, et al. , 1984; Leonard, et al. , 1985) . Furthermore, when the ability of free (Youle & Neville, 1982; Mclntosh, et al. , 1983; Eccles, et al. , 1987) or conjugated ricin B-chain (Vitetta, et al. , 1983; Vitetta, et al.. 1984; Vitetta, 1986) to enhance the activity of prebound ricin A-chain conjugates was studied, it was concluded that the B-chain also has a function beyond binding which can not be exerted by the MoAb. The fact that the A-chain is far more susceptible to proteolytic enzymes than the intact toxin (Olsnes, et al. , 1975) suggests that the B-chain may increase the cytotoxic potential of the A-chain by protecting it from lysosomal degradation. This is also indicated in experiments where ineffective ricin A-chain conjugates displayed high potency after addition of compounds known to increase lysosomal pH or alter intracellular routing (Casellas, et al. , 1984; Raso & Lawrence, 1984). There is also reason to believe that the B-chain assists in transporting the A-chain into the cytosol, but detailed knowledge of this process exists only in the case of diphtheria toxin (Olsnes, et al. , 1985).

Holotoxin IT's are in general far more potent than the corresponding A-chain conjugates. However, unspecific activity resulting from toxin B-chain mediated binding of the IT constitutes a problem to which several solutions have been proposed, including selection of IT's with sterically hindered toxin binding sites by affinity chromatography (Thorpe, et al.. 1984), and IT's constructed from toxins devoid of B-chain activity after chemical substitution (Leonard, et al.. 1985; Pietersz, et al. , 1988; Brusa, et_al. , 1989).

Recently, by genetic engineering, toxins with aberrations in their B-chain lacking the ability to bind • to their native receptors have been produced. Such variants of diphtheria toxin (Greenfield, et al. , 1987) and Pseudo onas Exotoxin A (Kondo, et al. , 1988) are now being currently studied with a special reference to their possible use in the construction of IT's (Johnson, et al. , 1989; Batra, et al. , 1990; Kreitman, et al.. 1990; Loberbaum-Galski, et al.. 1990; Siegall, et al. , 1990) .

The present invention is based on the discovery and isolation from natural sources of holotoxin variants of the plant toxin abrin which are devoid of B-chain binding activity and which display only a fraction of the activity as compared to the fully active holotoxin. The inactive variants have been coupled to different ligands, and the activity and specificity of the IT's have been assessed and compared to IT's constructed from Pseudomonas exotoxin (PE) , abrin or abrin A-chain, respectively. It had not previously been appreciated that such inactive variants existed.

According to the present invention we provide a naturally occurring variant holotoxin of abrin, the B chain of which is substantially devoid of cell-binding activity, when substantially free from active cell- binding variants of abrin holotoxin. By the term •naturally occurring' is meant that the abrin holotoxin occurs in natural sources and may be isolated therefrom by fractionation whereby active forms or variants of the holotoxin are separated from the desired inactive variant(s) .

Two inactive variant holotoxins of abrin have been isolated and termed abrin I and abrin II. The physical and biochemical characteristics of these variants are given hereinafter.

Abrin I appears from gel filtration and electrophoretic studies (SDS-PAGE) to consist of a mixture of holotoxins characterised by a double-band at about 67 kD and a single band at about 60 kD. Reduction to separate the A- and B-chains shows the double band to comprise A-chains of 32 kD and 34 kD conjugated to a B-chain of 35 kD; the single band appears to comprise an A-chain of 29 kD and a B-chain of 35 kD. While active abrin also has a B-chain of 35 kD, differences in binding to cells show that abrin I B-chain is different from that present in active abrin and has the advantage of much lower non¬ specific binding when conjugated with the A-chain and an immuno-carrier. The A-chains of abrin I are also different from the A-chain of active abrin (30 kD) but retain toxic activity. The A- and B-chains of abrin I, separately or in the native conjugated AB-forms, constitute a further feature of the invention.

Abrin II appears to consist of a single band on SDS-PAGE (65 kD) which on reduction runs as a single A-chain (30 kD) and a B-chain (37 kD) which is clearly different from the 35 kD B-chain of active abrin. The A- and B-chains of abrin II constitute a still further feature of the invention.

The isoelectric points of abrin, abrin I and abrin II have been measured in a pH-gradient polyacrylamide gel (Bio-Rad Model 111 Mini IEF Cell) . Bio-Lyte 3-10 (Bio- Rad) was used as an a pholyte and the pi values were measured in accordance with the manufacturer's instructions. The results are as follows:

pi

Abrin 6.5

Abrin I 8.2

Abrin II 6.5

All the separated A-chains were found to have a pi value of about 4.5. This value is, however, slightly uncertain because this band (a single band) is located near the electrode. The invention further provides an immunotoxin comprising a specific cell-binding moiety conjugated to a variant holotoxin according to the invention.

The invention also includes pharmaceutical compositions containing the above immunotoxins together with a pharmaceutical carrier or excipient, for example water for injection or physiological saline. The compositions may be useful for in vitro purification of bone marrow and intratecal injection into the Canalis vertebralis. For injections into rats lyophilized immunotoxin has been solubilized in phosphate buffered saline, pH 7.2-7.6 in doses ranging from 0.4-2.5 μg per animal (approx. 150g) . The immunotoxins can be stored lyophilised or frozen.

The invention also includes depot forms of the immunotoxin which may release the latter over a long period of time.

As indicated above, the target cells for treatment with immunotoxins according to the invention include tumour cells such as leukaemia cells, lung cancer cellε, cancer mammae, lymphomas and medulloblastomas.

The specific cell-binding moiety to which the inactive abrin variant is attached may, for example, be an antibody specific to a desired target cell, e.g. a tumour cell, or a lectin or protein binding specifically to such cells. Monoclonal antibodies are particularly useful in view of their specificity for particular cell- antigens, but proteins such as transferrins are specific to certain tumours as are certain cytokines such as IL- 2.

The inactive abrin holotoxin variant may be conjugated to the targeting cell-binding moiety by conventional techniques as described below, provided these do not block any of the functions of the toxin.

The immunotoxins according to the invention may be potentiated by an ionophore such as monensin or a monensin conjugate, eg. a conjugate with a protein such as human serum albumin (HSA) (Colombatti et al, Cancer Research, 5_Q, 1385-91, 1990) . We have found that the potentiating effect of monensin may be enhanced to an even greater extent by conjugation to certain long-chain ligands other than HSA. Such ligands should be highly water soluble, since monensin itself is hydrophobic. Ligands of interest include polysaccharides such as dextrans, (preferably with molecular weight less than 10 kD) aminodextrans, basic proteins and aminoacid copolymers, for example copolymers of Glu, Lys and Tyr (GLT) or Lys, Ala, Glu and Tyr (AGLT) which are available from Sigma. The enhancement of abrin I and, more notably abrin II, by monensin and conjugates thereof is many times greater than the enhancement observed with active abrin. The invention thus extends to the combined therapeutic use of the above immunotoxins with monensin and conjugates thereof and to pharmaceutical compositions comprising these.

In general, the most preferred cytotoxic agent according to the invention is abrin I conjugated to transferrin and potentiated by monensin or a conjugate thereof.

The inactive abrin holotoxins of the invention may be obtained by subjecting a natural source of the native •active' abrin holotoxins to fractionation whereby the active and inactive abrin holotoxins are separated. The activity of the AB-holotoxin variants can readily be assessed by standard cytotoxicity tests as described hereinafter.

Such isolation is facilitated by the surprising finding that inactive variants of abrin holotoxins may be more abundant than the active forms. Thus, for example, seeds from Abrus Precatorius contain about 3-4 times as much inactive abrin as active abrin.

Thus, seeds of Abrus precatorius may be crushed and extracted into an aqueous medium. We have found that extraction at pH 7.7 using Tris/HCl gave particularly good extraction of the desired material, as compared with 5% acetic acid as used for extraction of active abrin (Olsnes and Phil 1973) .

After extraction, the aqueous extract may be fractionated, for example by anion-exchange chromatography, eg. using DEAE-Sephacel.

Further purification can be effected by passage through a column of acid treated Sepharose 4B to remove any contaminating cell-binding abrin.

The A and B-chains may be separated by reduction of SS- bonds, eg. with ercaptoethanol, whereupon the B-chains precipitate. The supernatant may be treated with acid- treated Sepharose to remove any residual B-chain material and purified on a mono-Q column with elution by a linear salt gradient.

Affinity chromatography of the abrin variant immunotoxin is not necessary and thus avoids loss of the immunotoxin due to exposed binding sites.

The invention is illustrated in the following examples and drawings: PMSF is phenyl methyl sulphonyl fluoride:-

FIGURE 1 - Panel A. Purification of inactive forms of abrin by anion chromatography on DEAE-Sephacel. Elution was performed at a speed of 0.5 ml/min, and four ml fractions were collected and pooled, as indicated by bars. Panel B. Affinity chromatography on acid treated Sepharose of Peak I obtained after chromatography on DEAE-Sephacel. The Sepharose was equilibrated with PBS and chromatography was performed at a speed of 0.5 ml/ in. Four ml fractions were collected, and pooled as indicated by the bar. Panel C. Affinity chromatography of Peak II obtained after chromatography on DEAE- Sephacyl performed as described for Panel B. Bound material was eluted by 100 mM Lactose in PBS as indicated by the arrow. ( ) absorbance at 280 nm;

(- - -) concentration of NaCl.

FIGURE 2 - Purification of abrin on DEAE-Sephacel. The chromatography was performed as described hereinafter, and in the legend to Fig. 1. As revealed by SDS-PAGE, Peaks I and II contained small amounts of abrin I in addition to low molecular weight material, Peak III contained only coloured, low molecular weight material, whereas Peak V contained Abrus agglutinin (not shown) . The fractions containing abrin (Peak IV) were pooled (as indicated by the bar) and submitted to affinity chromatography on acid-treated Sepharose as described in the legend to Fig. IC.

FIGURE 3 - SDS-PAGE analysis (10% acrylamide) of inactive abrins in the absence ( - 2-me) or presence (+ 2-me) of 2-mercaptoethanol. Abrin and its constituent polypeptide chains were used as molecular weight standards, as indicated. Lanes 1,5 and 8: Abrin; Lanes 2 & 6: Abrin I; Lanes 3 & 7: Abrin II; Lane 4: Abrin A- chain. DF: dye front.

FIGURE 4 - SDS-PAGE analysis (10% acrylamide) of the constituent chains of abrin I. The two components of abrin I (denoted 67 kD and 60 kD) were cut out from a separate gel, eluted and re-electrophoresed in the absence (- 2-me) or presence (+ 2-me) of 2- mercaptoethanol. Lanes 1 & 5: abrin I; lanes 2 & 6: 67 kD; lanes 3 & 7: 60 kD. FIGURE 5 - SDS-PAGE analysis (10% acrylamide) of abrin (lane 5) , abrin I (lane 6) and abrin II (lane 7) after treatment with 2-mercaptoethanol and removal of precipitated material and reducing agent. Abrin (lane 1) , abrin I (lane 2) and abrin II (lane 3) were run under reducing conditions as controls. DF: Dye front.

FIGURE 6 - Separation of the A-chains of abrin I. After reduction, centrifugation and removal of reducing agent, the remaining components of abrin I were separated on a Mono Q (HR 5/5) column (FPLC system, Pharmacia) . The bound material was eluted (1 ml/min) with a linear salt gradient from 0 to 0.5M NaCl in 10 mM tris-HCl (pH 8.5). Left panel: A-chains from abrin I - Right panel: Abrin A-chain.

Example 1

Extraction of inactive-abrin

Four grams of decorticated Semen Jeσuiriti (the seeds from Abrus Precatorius) were allowed to swell overnight at 4°C in 50 mM Tris-HCl (pH 7.7) containing 1 mM PMSF. The material was then ground in a mortar, mixed with 40 ml of 50 mM Tris-HCl (pH 7.7) containing 1 mM PMSF and further extracted for 2 hrs. at 4°C. The extract was centrifuged at 10,000 x g for 10 min. followed by dialysis against 4500 ml of 10 mM Tris-HCl (pH 7.7) overnight at 4°C. After centrifugation, the dialyzed material was stored frozen until use.

Extraction of abrin

Abrin was extracted as described by Olsnes & Phil (1973) . Four grams of decorticated seeds from Abrus Precatorius were allowed to swell overnight at 4°C in 5% (v/v) acetic acid. The material was then ground as described above, mixed with 40 ml of 5% acetic acid, and further extracted for 2 hrs. at 4°C. The extract was centrifuged at 10,000 x g for 10 min. followed by dialysis against 4500 ml of water for four hrs., then against 4500 ml of 10 mM tris-HCl (pH 7.7) overnight at 4°C. The dialyzed material was clarified by centrifugation, and stored frozen until use.

Purification of inactive abrin

The crude extract was applied onto a column (1.5 x 11 cm) packed with DEAE-Sephacel equilibrated in 10 mM Tris-HCl (pH 7.7), washed with the same buffer and eluted with a linear salt gradient. Peak fractions were collected, pooled and applied onto a column (1.5 x 10 cm) packed with acid-treated Sepharose 4B (Godal, et al. , 1986) to remove any contaminating abrin. The "fall-through fraction" was dialyzed against PBS and concentrated by ultrafiltration to obtain a protein concentration of approx. l mg/ml.

Purification of abrin

The crude extract was applied onto a column packed with DEAE-Sephacel equilibrated in 10 mM Tris-HCl (pH 7.7), and run as described above. Fractions containing abrin were pooled and applied onto a column (1.5 x 10 cm) packed with acid-treated Sepharose 4B. Unbound material was removed by washing, and abrin was eluted with 100 mM lactose in PBS.

Purification of abrin A-chains

Abrin or inactive abrin were treated with 5% (v/v) 2- mercaptoethanol (2-me) for 20 hrs. at room temperature. The mixture was then centrifuged to remove the precipitated B-chain, and passed through a small column of acid-treated Sepharose to remove any remaining B- chain. The eluate was then purified on a Mono Q (HR 5/5) column (FPLC System, Pharmacia) equilibrated with 10 mM tris-HCl (pH 8.5). The different A-chains were eluted with a linear salt gradient.

Characterisation of inactive abrin

Polyacrylamide electrophoresis in the presence of SDS (SDS-PAGE) under reducing and non-reducing conditions was performed according to Laemmli (1970) . Antisera against purified, intact abrin were raised in rabbits (Godal, _____________, 1981) , and used in Western blot analysis performed with an Immun-Blot Assay Kit (Bio- Rad, Richmond, CA) with HRP-conjugated goat anti-rabbit, according to the manuf cturer's description.

Cell lines

The melanoma cell lines FEMX and LOX, the sarcoma lines OHS, SAOS 2 and KPDX and the lung cancer cell line SELS were all established in our laboratory from patient biopsy specimens. The glioblastoma cell lines SNB 19 and SF 295 were obtained from NCI (Frederick Cancer Research Facility) through the courtesy of Dr. R. Shoemaker. The mouse melanoma B-16 was obtained from NIH (Bethesda, MD) , and the B-cell lymphoma Raji was a gift from Dr. G. Klein, Stockholm, Sweden. All cell lines were grown in monolayer cultures, with the exception of Raji line which was kept as a suspension culture.

Measurement of biological activity

The cytotoxic effects of toxins and immunotoxins were assessed by measuring their ability to inhibit protein synthesis. Briefly, 5 x 10⁴ cells in 1 ml of RPMI medium containing 20 mM HEPES (pH 7.2) and 10% FCS were added to each well of a tissue culture plate (Falcon, Oxnard, CA) and incubated 3 hrs. at 37°C to allow the cells to adhere. Then appropriate amounts of toxin or immunotoxin were added and the cells were incubated for 20 hrs. at 37°C. After washing with PBS, the cells were incubated in 0.5 ml of leucine-free medium containing 2 μCi ³H-Leucine/ml (Amersha , Little Chalfont, UK) for 30 min. at 37°C, followed by processing as described by Sandvig and Olsnes (1982) . The TCA-precipitable radioactivity was measured and compared to that of untreated controls. The Raji cells were treated as described by Godal, et al. (1986) .

The dose required to inhibit protein synthesis by 50% (ID50) was calculated from dose-response curves. To obtain a more meaningful parameter of cytotoxic potential, the sensitivity (=1/ID50) was calculated in each experiment.

Conjugation of abrin, abrin A-chain and inactive abrin to monoclonal antibodies and to transferrin

The monoclonal antibody 9.2.27 against malignant melanoma (Morgan, et al. , 1981) was kindly supplied by Dr. A.C. Morgan (NeoRx Corporation, Seattle, WA) . The anti-sarcoma antibody TP-3 was developed in our laboratory (Bruland, et al. 1986) and the anti lymphoma antibody HD 39 was kindly supplied by Dr. Bernd Dόrken (Heidelberg, Germany) . Human Transferrin (Sigma Chemical Company, St Louis, MO) was saturated with iron according to Shindleman, et al.. (1981) before conjugation.

Abrin, inactive abrin and Pseudomonas exotoxin (PE, Swiss serum and Vaccine Institute, Bern, Switzerland) were conjugated to the antibody by a thioether bond formed with SMCC (Pierce, Rockford, IL) according to (Morgan, et al.. 1990) . Abrin A-chain was disulfide- linked to the 9.2.27 antibody using the heterobifunctional reagent SPDP (Pharmacia, Sweden) as described by Wawrzynczak, et al.. 1990. Inactive abrin and PE were conjugated with transferrin as described by Johnson, et al. (1989) . All conjugates were purified by gelfiltration, and fractions containing purified conjugate, as judged by SDS-PAGE, were pooled and used in the experiments. Purification of inactive abrin

Typical chromatographic patterns are shown in Fig. 1. Pilot experiments (not shown) indicated that Peak I contained abrin-derived material of low toxicity where Peak II contained highly toxic material, mainly abrin (Fig. 1A) . The pooled fractions of Peak I and Peak II from the DEAE column were run separately on acid treated sepharose to remove molecules with an ability to bind to the matrix, and the peaks indicated by bars (Figs. 1 B and C) were collected and used in further experiments. Peak a (Fig. IB) was designated abrin I, and peak b (Fig. IC) was designated abrin II. The large peak (Fig. IB) consisted mainly of low molecular weight material (not shown) . As expected, the material eluted with lactose (peak c) proved to be abrin (Fig. IC) .

Purification of abrin

The material extracted with acetic acid was purified by ion-exchange chromatography as described for inactive abrin (Fig. 2) . The pooled fractions from Peak IV were applied onto acid-treated Sepharose as described above, and abrin was eluted with 100 mM lactose (not shown) .

Structure of inactive abrin

SDS-PAGE analysis (Fig. 3) of abrin I revealed a double band migrating at an apparent molecular weight (Mw) of 67 kD, and one band migrating slightly faster (60 kD) , as compared with abrin (65 kD) .

In contrast, abrin II seemed to consist of one component which migrated identically with abrin (65 kD) (Fig. 3) . However, upon reduction, abrin II dissociated into one component with a Mw identical to that of the normal abrin A-chain (30 kD) , and one with a slightly higher Mw (37 kD) than the normal abrin B-chain (35 kD) .

SDS-PAGE analysis of reduced abrin I revealed four bands: one migrated as "normal" abrin B-chain (35 kD) and three components which migrated at Mw's 34, 32 and 29 kD, respectively (Fig. 3) . To identify the origin of these components, electrophoresis of abrin I was performed under non-reducing conditions, and gel-pieces containing either the double band (67 kD) or the 60 (kD) component where cut out, eluted and re-electrophoresed. It was found (Fig. 4) that, after reduction, the double band dissociated into one seemingly normal B-chain (35 kD) and two "A-chains" (34 and 32 kD) . The 60 kD component also contained the 35 kD component in addition to a component of 29 kD) .

In order to investigate whether abrin I and abrin II contain a B-chain, or a B-chain-like component, advantage was taken of the finding (Olsnes, et al.. 1973) that reductive dissociation of abrin in the absence of lactose leads to a complete precipitation of its B-chain. Abrin I and abrin II were treated with 2- mercaptoethanol over night and centrifuged to remove precipitated material. The reducing agent was removed by gelfiltration and the protein fraction analysed by SDS-PAGE. It was found (Fig. 5) that after reduction of abrin I only the 34, 32 and 29 kD components remained in solution, i.e. that the 35 kD component was completely precipitated. In the case of Abrin II, the 37 kD component precipitated. Thus, the largest moieties of abrin, abrin I and abrin II, respectively, all precipitated after reduction, indicating similarities in structure and, possibly, also in biological function.

To investigate the possibility that the different components of abrin I might be associated under native conditions, i.e. that its native molecular weight is larger than 67 kD, gel-filtration experiments were performed. These revealed (not shown) that the elution volume of abrin I was slightly higher than that of human serum albumin (Mw = 67 kD) , strongly indicating that the components of abrin I exists as monomers under native conditions.

Western blot analysis revealed that all components present in abrin I and abrin II crossreacted with an antibody raised against abrin (not shown) , strongly indicating a close relationship between active and inactive forms of abrin.

Purification of A-chains from inactive abrin

After treatment of Abrin I with 2-me, and removal of precipitated material, attempts were made to separate the remaining components (29 kD, 32 kD and 34 kD) by ion exchange chromatography using the FPLC (Pharmacia) system. From typical elution diagrams (Fig. 6) , it can be seen that the abrin A-chain eluted as a single peak with a retention time of 6.3 min., whereas the A-chains of abrin I separated into two peaks, hereafter referred to as and β, with retention times of 3.9 and 4.6 min., respectively. SDS-PAGE (not shown) revealed that the a peak consisted of the 32 and 34 kD components, whereas the β peak contained the 29 kD component. The 29 kD component is referred to as Abrin I β-A-chain.

Example 2

Biological activity of inactive abrin

The cytotoxic activity of abrin I and abrin II was measured after addition to three different cell lines, and compared to that of abrin. The results, which are outlined in Table I, showed only negligible differences between abrin I and abrin II. However, abrin I and abrin II displayed a dramatically lower activity than did abrin. Thus, in the case of the abrin-sensitive cell lines OHS and SELS, the cytotoxic potential of inactive abrin was approximately 1000 times lower than that of abrin, while in the case of the less abrin-sensitive cell line FEMX, the cytotoxic activities differed with an average factor of 45. Interestingly, the sensitivity of the cells to the inactive forms of abrin correlated, although weakly, with their inherent sensitivities to native abrin.

It was found (Table I) that the toxic action of both abrin I and abrin II was only modestly reduced in the presence of lactose, a competitive inhibitor of abrin- binding to cells (Olsnes, et al. 1974) , whereas the toxicity of abrin was strongly reduced. Altogether, the data in Table I indicate that the low cytotoxic activity of abrin I and abrin II either results from a poor ability to bind to cells, or that their A-chains somehow are unable to penetrate into the cytosol after cellular uptake. In attempts to elucidate the underlying mechanism, we first measured the capacity of abrin I and abrin II to bind to red blood cells. In control experiments it was found (Table II) that abrin showed a strong ability to bind to cells, and that this ability was almost abolished by the addition of lactose, as expected. However, abrin I and abrin II showed a low, but definite binding, which was not affected by the presence of lactose, indicating that the nature of this binding is different from that of abrin.

Biological activity of immunotoxins containing inactive abrin

The poor ability of abrin I and abrin II to bind to cells (Table II) is an interesting feature with regard to their use of toxic moieties of IT's. Provided that abrin I and abrin II A-chains are capable of penetrating efficiently into the cytosol after cellular uptake, IT's constructed from these abrin variants would have been expected to display high activity due to the presence of the B-chain, and high specificity due to the lack of B- chain mediated binding.

The two inactive forms of abrin were conjugated to the antimelanoma antibody 9.2.27 and the cytotoxic activity of the resulting I 's was measured in three antigen- positive and one antigen-negative cell lines and compared to that of the correspondent abrin, abrin A- chain and abrin I-beta-A-chain conjugates.

The results are outlined in Table III. It is seen that the abrin-IT showed virtually no specificity, i.e. there were no difference in cytotoxic activity towards antigen positive and antigen negative cell lines. In the presence of the ionophore Monensin (Mo) , a somewhat higher specificity was achieved. Interestingly, almost identical results were obtained with the abrin II-IT (not shown) . Thus, the differences between the B-chains of abrin and abrin II, respectively, seem to be of little importance from the point of view of specificity. Therefore, the similar behaviour of abrin and abrin II as toxic moieties must rely on their A-chains, in concordance with the finding (Fig. 3) that these seem to be identical.

As is often seen with A-chain IT's, the abrin A-chain conjugate displayed a low activity against all cell lines. However, in the presence of Mo, the activity against two target cell lines was dramatically improved, whereas the activity against the non-target cell lines SELS was unaltered, thereby increasing the specificity by a factor of approximately 10,000. Unfortunately, the cytotoxic activity against the target cell line LOX was not improved by the presence of Mo, for reasons which are unknown.

Also abrin I conjugated to the 9.2.27 antibody displayed a low activity and specificity in the absence of Mo. However, the presence of the ionophore had a considerable impact on conjugate activity against all target cell lines, but not on the non-target cell line SELS. Thus, in the presence of Mo the abrin I conjugate was highly specific; in fact, this conjugate displayed the highest overall specificity of the four conjugates tested. Possibly the observation (Table III) that Mo had a stronger enhancing effect on the abrin I-IT than on the abrin IT, is most likely due to the structural differences between abrin A-chain and abrin I A- chain(s), respectively.

To investigate whether the differences between the conjugates containing abrin or abrin I was due to differences between their A-chains, we constructed 9-2- 27-abrin A chain and 9.2.27-abrin I β-A-chain conjugates and measured their cytotoxic activities in the absence or presence of Mo.

It was found (Table III) that the abrin A-chain conjugate displayed a low activity against all cell lines. However, in the presence of Mo, the activity against two target cell lines was dramatically improved, whereas the activity against the non-target cell line SELS was unaltered. However, the activity of 9.2.27- abrin A chain against the target cell line LOX was not improved in the presence of Monensin. In contrast, Mo improved the cytotoxic activity of the abrin I-β-A-chain conjugate by a factor of at least 10. It is interesting in this regard that Mo enhanced the activity of the 9.2.27-abrin against the cell line LOX only by a factor of 2 , whereas it enhanced the activity of the 9.2.27- abrin I with a factor of 133.

To further investigate the possible use of abrin I as a toxic moiety, abrin I was conjugated to three different ligands, and the cytotoxic activities of the resulting conjugates were measured in both target and non-target cells. For comparison, the commonly used Pseudomonas exotoxin (PE) was linked to the same ligands, and activities of the conjugates were measured as for the abrin I conjugates.

In Table IV the results obtained with the antisarcoma antibody TP-3 are listed. It is seen that the PE conjugate displayed a good selectivity, except a failure to kill the target cell line SA OS 2. In separate experiments it was found (not shown) that this cell line was resistant to PE. The abrin I conjugate displayed a poor selectivity in the absence of Mo, but in the presence of the ionophore the conjugate displayed good selectivity for all target cell lines tested.

More encouraging results were obtained with the anti- lymphoma antibody HD 39 (Table V) . Here, the abrin I conjugate displayed a tenfold higher selectivity than the PE conjugate also in the absence of Mo. In the present of the ionophore, the selectivity was approx. 500 times higher than that for the PE conjugate.

Several investigators have used transferrin (Tfn) as a ligand for specific cellular uptake of toxins or other effector molecules, based on the fact that most tumor cell lines express the transferrin receptor. This allows measurement of the activity of such Tfn- conjugates in a large number of cell lines derived from histologically different tumors. We constructed Tfn-Abrin I and Tfn-PE and measured their cytotoxic activities in a number of cell lines. It is evident from Table VI that, with the exception of the Raji cells, the activity of both conjugates correlated well with the level of receptor expression. In the absence of Mo, Tfn-abrin I and Tfn-PE displayed almost identical activities, but Mo enhanced the cytotoxic activity of Tfn-abrin I by an average factor of approximately 35. The reason for the low activity observed in the Raji cells, which expressed the highest number of Tfn-receptor in this study, is obscure but is possibly due to inherent metabolic properties of this particular cell line.

Example 3

Comparison of cytotoxic activity of sterically hindered abrin with abrin I and abrin II

In order to determine whether the cytotoxic activity of abrin conjugates prepared with a sterically hindered binding site (Thorpe) was different from that of conjugates prepared from inactive/binding deficient abrin variants, we coupled Transferrin (Tfn) to abrin, abrin I and abrin II, respectively. Tfn-abrin was passed through a column of acid-treated Sepharose before use to remove molecules with exposed binding sites. Their biological activities were measured in the medulloblastoma cell line SNB 19 in the absence or presence of 10^"7 M Monensin (Mo) , using the standard 20 h assay as shown in Table VII hereinafter.

It is seen that Mo potentiated the cytotoxic effect of Tfn-abrin I and, in particular, Tfn-abrin II more strongly than the effect of Tfn-abrin. Example 4

Potentiation of cvtotoxicity by Monensin conjugate

It was demonstrated above that the presence of Mo strongly enhanced the in vitro activity of abrin I and -II conjugates. In vivo. Mo must be bound to a high molecular ligand, for example HSA to obtain suitable pharmacological properties (Colombati 1990) .

Two random copolymers of amino acids were tested because they contain:

- a high fraction of amino groups for coupling

- a high fraction of negatively charged Glu, possibly assuring good solubility after conjugation with Mo

- Tyr which absorbs light at 280 nm allowing detection during chromatography. Tyr can also be labelled with ¹²⁵I for binding- and pharmacological studies.

Materials:

Poly(Glu,Lys,Tyr) = 6:3:1 (Sigma P4409 - Mw 20-50 kD) =

GLT

Poly(Lys,Ala,Gly,Tyr) = 5:6:2:1 (Sigma P 1278 - Mw 30-70 kD) = AGLT

Method:

10 mg Mo was mixed with 6 mg S-acetylmercaptosuccinic anhydride (SAMSA) , dissolved in 2 ml of methanol and incubated overnight at 20°C in a small glass test tube. The methanol was evaporated at 30°C under a stream of N_ rendering a clear gel at the wall of the tube. Immediately before use, the gel was washed twice (5 min. each) with water to remove uncongated SAMSA.

5 mg of polymer was dissolved in 1 ml of 20 mM borate buffer (pH 8.5) and applied onto a column (0.7 x 27 cm) packed with Sephadex G-50 equilibrated in 0.15 M NaCl in borate buffer. The high molecular fraction eluting in the void volume was collected - typically 2.5 ml (2 mg/ml) . This fraction was treated with a 30 fold molar excess of Trauts reagent (2-Iminothiolane) for 30 min at ambient temperature to introduce SH-groups into the polymer. Then, a few grains of Ellmans reagent was added and the mixture was incubated for another 10 min. to allow reaction with the SH-groups in the polymer, which generates a strong yellow color. The mixture was then passed through a PD-10 column to remove low Mw material, and the eluate was mixed with solid NH-OH to a final concentration of 1M, and added to the water- extracted Mo-SAMSA gel (see above) . Incubation was carried out for 1 hr at ambient temperature with end over end rotation. The development of a yellow colour due to the reaction:

Polymer-S-S-Ellmans + Mo-..SH >. Polymer S-S-Mo +

Ellmans-SH (Yellow)

can be used to monitor the process. After completion, usually within 1 hr, the mixture was passed through PD- 10 (PBS containing 10% EtOH) , and the eluate containing Mo-polymer was stored at 4°C.

Mo was conjugated with HSA exactly as described above, except that the last PD-10 step was performed with PBS.

Results:

To test the potentiating activity of conjugated Mo, OHS cells were incubated with increasing concentrations of TP-3-abrin I in the absence or presence of free or conjugated Mo, for 20 hrs at 37°C, followed by measurement of protein synthesis. Results are shown in Table VIII hereinafter. 6-7 μg/ml of Polymer-Mo results in a similar in vitro enhancing effect as Mo alone. To obtain the same effect by HSA-Mo, more than 100 μg/ml must be added.

^a Sensitivity = 1/ID50 where ID50 is the dose (expressed in ng/ml) required to inhibit protein synthesis by 50 per cent. ^b Assayed in the presence of 30 mM lactose.

Table II

Binding of inactive abrin to red blood cells³

Treatment Bound/Total added (%)^b

No lactose 100 mM lactose

Abrin 24.5 0.4

Abrin I 0.8 0.9

Abrin II 1.3 0.7

^a Human erythrocytes (Type O) from 1 ml blood were washed twice in PBS and resuspended in 5 ml PBS containing 1 mg/ml HSA. Hundred μl of this suspension was mixed with 1 ml PBS/HSA with or without lactose. Radiolabelled toxin (50 ng - 3 x 10⁶ cpm) was added, and was incubated at 37"C with occasional shaking. After 30 min. the blood cells were sedimented by centrifugation, washed twice in PBS before measuring bound radioactivity. ^b Average from two experiments.

Table III

Sensitivity³ of antigen-positive and antigen negative tumor cell lines to different forms of abrin conjugated to the antimelanoma antibody 9.2.27

Cell Ag Abrin Abrin- Abrin I Abrin I- line A-chain β-A-chain

Mo^c + Mo Mo + Mo - Mo + Mo Mo + Mo

10^"4 0.25 <10~ 10^"3 ND^d ND

ND ND

^a See footnote to Table I . ^b Antigen-expression was determined by indirect immuno fluorescence . ^c Cytotoxic effect was assayed in the absence (- Mo) or presence (+ Mo) of 10^"7 M Monensin. ^d Not done. Table IV

Sensitivity of different cell lines to inactive abrin and PE conjugated to the antisarcoma antibody TP-3

^a Antigen expression was assayed by indirect immunofluorescence. ^b See footnote to table III. ^c Not done.

Table V

Sensitivity³ of Raji and OHS cells to PE and abrin I conjugated to the antilvmphoma antibody HP 39

Conjugate Cell line

Raji (Ag +) OHS (Ag -)

HD 39 - PE 0.17 0.03

HD 39 - Abrin I 0.4 0.003

HD 39 - Abrin I + Mo^b 50 0.01

³ See footnote to Table I. ^b In the presence of 10^"7 M Monensin.

³ Binding was assessed by exposing 10⁶ cells in 0.5 ml PBS/HSA to 10^s cpm (approx. 10 ng) ¹²⁵I-Transferrin (Amersham, Little Chalfont, Buckinghamshire, England) for 4 hrs. on ice with occasional shaking. After washing, the bound radioactivity was measured, and calculated relative to the total amount added. ^b See footnote to table I.

Table VII

Cytotoxic activity measured in the medulla blasto a cell line SNB 19

Tfn-abrin II

0.05 11

220X

* Cytotoxic activity = 1/IC50, where IC50 is the concentration (in ng/ml) required to inhibit cellular protein synthesis by 50%.

Table VIII

Effect of free and conjugated Monensin on the cytotoxic activity of TP-3-Abrin I against a sarcoma cell line OHS

Enhancer Cytotoxicity* Cytotoxicity** Enhancing of enhancer of TP-3 abrin I effect

None 70

Mo (10^"7 M) 20 0.8 88

* Cytotoxicity of enhancer is presented as percentage inhibition of protein synthesis in the presence of enhancer relative to an untreated control. ** Given as IC50 (ng/ml) .

REFERENCES

Batra, J.K., FitzGerald, D. . , Gately, M. , Chaudhary, V.K., and Pastan, I. (1990) J. Biol. Chem. 265, 15198- 15202

Bruland, 0. , Fodstad, 0. , Funderud, S., and Pihl, A. (1986) Int. J. Cancer 38, 27-31

Brusa, P., Pietribiasi, F. , Bussolati, G. , Dosio, F. , Arione, F. , Comoglio, P.M., Prat, M. , and Cattel, L. (1989) Cancer Immunol. Immunother. 29, 185-192

Casellas, P., Bourrie, B.J.P., Gros, P., and Jansen, F.K. (1984) J. Biol. Chem. 259, 9359-9364

Eccles, S.A., Mclntosh, D.P., Purvies, H.P., Cumber,

A.J. , Parnell, G.D., Forrester, J.A., Styles, J.M. , and

Dean, C.J. (1987) Cancer Immunol. Immunother. 24, 37-41

Frankel, A.E., (ed.), (1988) Immunotoxins. Kluwer Academic Publishers, Boston/Dordrecht/Lancaster

Godal, A., Fodstad, 0., Morgan, A.C., and Pihl, A. (1986) J. Natl. Cancer Inst. 77, 1247-1253

Godal, A., Olsnes, S., and Pihl, A. (1981) J. Toxicol. Environm. Health 8, 409-417

Greenfield, L. , Johnson, V.G., and Youle, R.J. (1987) Science 238, 536-539

Johnson, V.G., Wrobel, C. , Wilson, D. , Zovickian, J. , Greenfield, L. , Oldfield, E.H., Youle, R. (1989) J. Neurosurg. 70, 240-248 Kondo, T., FitzGerald, D. , Chaudhary, V.K., Adhya, S., and Pastan, I. (1988) J. Biol. Chem. 263, 9470-9475

Kreitman, R.J. , Chaudhary, V.K. , Waldmann, T. , Willingham, M.C., FitzGerald, D.J. , and Pastan, I. (1990) Proc. Natl. Acad. Sci. 87, 8291-8295

Laemmli, U.K. (1970) Nature 227, 680-685

Leonard, J.E., Wang, Q.-c, Kaplan, N.O., and Royston, I. (1985) Cancer Res. 45, 5263-5269

Lorberboum-Galski, H., Garsia, R.J., Gately, M. , Brown, P.S., Clark, R.E. , Waldmann, T. , Chaudhary, V.K. , FitzGerald, D.J. , and Pastan, I. (1990) J. Biol. Chem. 265, 16311-16317

Mclntosh, D.P., Edwareds, D.C. , Cumber, A. . , Parnell, G.D., Dean, C.J., Ross, W.C.J., and Forrester, J.A. (1983) FEBS Lett. 164, 17-20.

Morgan, A.C., Galloway, D.A. , and Reisfeld, R.A. (1981) Hybridoma 1, 27-38

Morgan, A.C., Sivam, G. , Beaumier, P., Mclntyre, R. , Bjorn, M. , and Abrams, P.G. (1990) Mol. Immunol. 27, 273-282

Olsnes, S., and Pihl, A. (1973) Eur. J. Biochem. 35, 179-185

Olsnes, S., and Pihl, A. (1982) in: Molecular action of toxins and viruses. (Cohen, P., and van Heyningen S., eds.) pp. 51-105, Elsevier, Amsterdam

Olsnes, S., Refsnes, K. , Christensen, T.B., and Pihl, A. (1975) Biophvs. Acta. 405, 1-10 Olsnes, S., Saltvedt, E. and Phil, A. (1974) J. Biol. Chem. 249, 803-810

Olsnes, S., and Sandvig, K. (1988) in: Immunotoxins (Frankel, A.E., ed) pp. 39-73, Kluwer Academic Publishers, Boston/Dordrecht/Lancaster

Pieterz, G.A. , Kanellos, J., and McKenzie, I.F.C. (1988) Cancer Res. 48, 4469-4476

Raso, V., and Lawrence, J. (1984) J. Exp. Med. 160, 1243-1240

Sandvig, K. , and Olsnes, S. (1982) J. Biol. Chem. 257, 7495-7503

Shindelman, J.E., Ortmeyer, A.E., and Sussman, H.H. (1981) Int. J. Cancer 27, 329-334

Siegall, C.B., FitzGerald, D.J., and Pastan, I. (1990) J. Biol. Chem. 265, 16318-16323

Thorpe, P.E., Ross, W.C.J., Brown, A.N.F., Myers, CD., Cumber, A.J., Foxwell, M.J. , and Forrester, J.T. (1984) Eur. J. Biochem. 140, 63-71

Vallera, D.A. , Quinones, R.R., Azemove, S.M., and Soderling, C.C.B. (1984) Transplantation 37, 387-392

Vitetta, E.S. (1986) J. Immunol. 136, 1880-1887

Vitetta, E.S., Cushley, W. , and Uhr, J.W. (1983) Proc. Natl. Acad. Sci. 80, 6332-6335

Vitetta, E.S., Fulton, R.J. and Uhr, J.W. (1984) J. Ex . Med. 160, 341-346 Wawrzynzcak, E.J., Cumber, A.J. , Henry, R.V. May, J., Parnell, G.D., Worrell, N.R. , and Forrester, J.A. (1990) Cancer Res. 50, 7519-7526

Claims

1. A naturally occurring variant holotoxin of abrin, the B chain of which is substantially devoid of cell- binding activity, when substantially free from active cell-binding variants of abrin holotoxin.

2. An abrin variant as claimed in claim 1, being abrin

I having a molecular weight by SDS-PAGE of 67 kD and consisting of an A-chain having a molecular weight of 29-34 kD together with a B-chain having a molecular weight by SDS-PAGE of 35 kD.

3. An abrin variant as claimed in claim 2 having a pi value of 8.2 measured in a pH gradient polyacrylamide gel.

4. An abrin variant as claimed in claim 1, being abrin

II having a molecular weight by SDS-PAGE of 65, and consisting of an A-chain having a molecular weight by SDS-PAGE of 30 kD together with a B-chain having a molecular weight by SDS-PAGE of 37 kD.

5. An abrin variant as claimed in claim 4 having a pi value of 6.5 measured in a pH gradient polyacrylamide gel.

6. The A-chains and B-chains of the variants as claimed in claims 2 and 4.

7. An immunotoxin comprising a specific cell-binding moiety conjugated to an abrin variant as claimed in claim 1.

8. An immunotoxin as claimed in claim 7 in which the specific cell-binding moiety is an antibody, a lectin or a protein. 9. An immunotoxin as claimed in claim 8 in which the cell-binding moiety is transferrin.

10. Pharmaceutical compositions comprising an immunotoxin as claimed in claim 7 together with a pharmaceutical carrier or excipient.

11. Compositions as claimed in claim 10 additionally containing an ionophore which enhances the cytotoxic activity of the immunotoxin.

12. Compositions as claimed in claim 11 in which the ionophor is monensin or a conjugate thereof.

13. A process for the isolation of an abrin variant as claimed in claim 1 wherein a natural source of mixed active and inactive variants of abrin is subjected to fractionation whereby the active and inactive abrin variants are separated.

14. A process as claimed in claim 13 in which said source is seeds of Abrus precatorius.

AME.NDED CLAIMS

[received by the International Bureau on 11 December 1992 (11.12.92); original claims 1-14 replaced by amended claims 1-15 (3 pages)]

1. An immunotoxin comprising a specific cell-binding moiety conjugated to a naturally occurring variant holotoxin of abrin, the B chain of which is substantially devoid of cell-binding activity, when substantially free from active cell-binding variants of abrin holotoxin.

2. An immunotoxin as claimed in claim 1 in which the specific cell-binding moiety is an antibody, a lectin or a protein.

3. An immunotoxin as claimed in claim 2 in which the cell-binding moiety is transferrin.

4. A naturally occurring variant holotoxin of abrin, the B chain of which is substantially devoid of cell- binding activity, when substantially free from active cell-binding variants of abrin holotoxin, being either abrin I having a molecular weight by SDS-PAGE of 67 kD and consisting of an A-chain having a molecular weight of 29-34 kD together with a B-chain having a molecular weight by SDS-PAGE of 35 kD or abrin II having a molecular weight by SDS-PAGE of 65, and consisting of an A-chain having a molecular weight by SDS-PAGE of 30 kD together with a B-chain having a molecular weight by SDS-PAGE of 37 kD

5. An abrin variant as claimed in claim 4 having a pi value of 8.2 measured in a pH gradient polyacrylamide gel.

6. An abrin variant as claimed in claim 4 having a pi value of 6.5 measured in a pH gradient polyacrylamide gel. 7. The 29, 30 and 34 kD A-chains and 35 and 37 B- chains of the variants as claimed in claim 4.

8. A naturally occurring variant holotoxin of abrin, the B chain of which is substantially devoid of cell- binding activity, when substantially free from active cell-binding variants of abrin holotoxin, for use in an immunotoxin.

9. An abrin variant as claimed in claim 1, being abrin

I having a molecular weight by SDS-PAGE of 67 kD and consisting of an A-chain having a molecular weight of 29-34 kD together with a B-chain having a molecular weight by SDS-PAGE of 35 kD, for use in an immunotoxin.

10. An abrin variant as claimed in claim 1, being abrin

II having a molecular weight by SDS-PAGE of 65, and consisting of an A-chain having a molecular weight by SDS-PAGE of 30 kD together with a B-chain having a molecular weight by SDS-PAGE of 37 kD, for use in an immunotoxin.

11. Pharmaceutical compositions comprising an immunotoxin as claimed in claim 1 together with a pharmaceutical carrier or excipient.

12. Compositions as claimed in claim 11 additionally containing an ionophore which enhances the cytotoxic activity of the immunotoxin.

13. Compositions as claimed in claim 12 in which the ionophor is monensin or a conjugate thereof.

14. A process for the isolation of an abrin variant as claimed in claim 4 wherein a natural source of mixed active and inactive variants of abrin is subjected to fractionation whereby the active and inactive abrin variants are separated.

15. A process as claimed in claim 14 in which said source is seeds of Abrus precatorius.