The present invention relates to a process for the expression and secretion in the methylotrophic yeast Pichia pastoris of toxic chimaeric proteins having ribosome-inactivating activity, based on the plant ribosome- inactivating protein (RIP) saporin from soapworth Saponaria officinalis.

Background of the invention

Saporin belongs to the N-glycosidase family of plant toxins that includes the prototype RIP ricin: these are ribosome-inactivating proteins able to remove a specific adenine, found in an universally conserved "stem-loop" structure present in ribosomal RNAs from the large ribosome subunit (1, 2), leading to a block in protein synthesis and to apoptotic cell death in eukaryotic intoxicated cells.

The Pichia pastoris expression system makes use of a microbial eukaryotic host that can reach extremely high cell densities and has been widely used for high level expression of foreign secretory proteins which can be then purified from the extracellular culture media (3, 4). More recently, P. pastoris host cells have been successfully used for optimized expression of targeted toxins based on the diphtheria toxin (5) and also for the expression of the toxin PAP (Pokeweed Antiviral Protein) from the leaves of the plant Phytolacca americana (6).

Most of the targeting moieties such as antibody-fragments, growth factor domains or receptor ligands used to construct targeted toxins for anticancer therapy derive from secretory proteins. These normally fold in the oxidative environment of the endoplasmic reticulum (ER) and may require specific co-and post-translational modifications such as N-or O-glycosylation and disulphide bond formation to be produced in a proper conformation and in a biological active form. Therefore, their folding in bacterial cells can be quite inefficient, since bacterial host cells are devoid of the endomembrane system present in eukaryotic cells where newly synthesized polypeptides are assisted in their folding and undergo quality control mechanism(s) allowing the secretion of folded active proteins (4). In addition, the production of recombinant proteins in E.coli may have several disadvantages such as the formation of inclusion bodies containing heterologous protein needing to be refolded, and endotoxin contamination.

For instance, only low amounts of properly folded antimetastatic chimera made by the fusion of the Amino-Terminal Fragment of human urokinase (ATF) to the plant RJP saporin were recovered in bacterial host cells because of the presence in the ATF moiety of two disulfide-rich domains, a kringle and an EGF-like domain (7). ATF-saporin was instead secreted very efficiently when expressed in eukaryotic host cells protected from autointoxication by neutralizing antisaporin antibodies (8). This clearly shows that ER-targeting allows secretory ATF-saporin to reach micromolar concentrations in the extracellular medium, but also indicates that the production of these chimaeras in eukaryotic cells can be hampered by their intrinsic toxicity, due to the potential action of saporin towards the host ribosomes. In principle, because of the different sub-cellular localization of the newly synthesized recombinant polypeptide and the host ribosomes, this should not happen. On the other hand, efficiency of segregation of newly synthesized proteins in the endomembrane system can be far below 100% (9, 10) and mechanisms that can mediate the retrotranslocation to the cytosol are present in the endoplasmic reticulum (11). For all the above reasons, one may expect to observe some host toxicity during the expression of saporin polypeptides in an eukaryotic host cell. This may lead to a decrease in the biosynthetic ability of the host cells expressing saporin. Under these suboptimal conditions, factors controlling translation efficiency may become crucial for the expression of the recombinant polypeptide. Despite a plant RIP, PAP, has been previously expressed at good levels in the Pichia system (6), the differences in the aminoacid sequence and in specific enzymatic activity indicated that the task to express saporin in this eukaryotic system would not be as simple and would have required to explore several independent parameters in order to define the ones essential to achieve this goal. In fact we have tried to express saporin under the same conditions as in (6), but we failed in obtaining similar levels of expression (12). Thus here we describe a procedure which relies on codon optimization, on the selection of suitable strains and other conditions for the expression of saporin and saporin fusions in Pichia pastor is.

Description of the invention

The invention provides a process for the preparation of recombinant secretory saporin. Such a process comprises the steps of constructing a precursor saporin gene, inserting the precursor gene into Pichia pastoris GSl 15 (his4) host strain, followed by fermentation of the transformed clone to obtain secretory recombinant saporin. Preferably, said precursor is constructed using pPICZalphaA vector and the cloning is carried out downstream AOX-I promoter and Mating factor Alpha (MFalpha) signal sequence preserving the GIu- Ala repeats of Stel3 site and inserting a glutamic acid and a phenylalanine at the NH ₂ -terminus of mature saporin.

The invention also provides for optimized DNA gene sequences suitable for expression of recombinant secretory saporin (sequence ID#1) or saporin-containing fusions (from sequences ID#1; ID#2; ID#3). The optimized saporin sequence allows high-level expression of the seed saporin isoform in the selected Pichia pastoris strain. The invention also provides a process for the preparation of recombinant secretory chimeras containing the saporin sequence (sequence ID#1) fused to a targeting domain. In an embodiment of this invention, the secretory domain is constituted by the Amino Terminal Fragment of human urokinase (ATF). The production of secretory ATF-saporin comprises the steps of constructing an ATF-saporin precursor gene, cloning into an expression vector, stably inserting the precursor gene into a P. pastoris host strain, followed by fermentation of the transformed clone to obtain secretory recombinant fusion protein.

The invention, in still a further embodiment, provides a process for the preparation of recombinant proteins endowed with anti-CD22 specificity containing a secretory single-chain variable fragment fused to saporin, comprising the steps of constructing a codon-optimised scFv precursor optimised sequence as a V _H -218L-V _L gene, that is then inserted in either pPICSAPH vector for single step purification of the fusion chimaera or in the vector identified as pITalphawt, and stably transforming P. pastoris host strain, followed by fermentation of the transformed clones to obtain purified secretory recombinant αCD22-saporin chimaeras.

DNA sequences optimized for the expression in Pichia pastoris are also provided for recombinant secretory ATF-saporin (sequence ID# 1 and ID#2) and recombinant secretory αCD22 fusion chimaeras (sequence ID #1 and ID#3). The host is preferably Pichia pastoris GS l 15 (his4) which was found more tolerant to saporin expression (12).

Detailed description of the invention

It has now been found a gene sequence (Sequence ID #1) allowing an efficient and convenient expression and production of saporin-based fusion proteins. The invention refers therefore to said sequence as well as to a process for the expression of proteins coded by said sequence in Pichia pastoris yeast strains. The process of the invention is improved as a consequence of increased levels of accumulation of the expressed protein (FIGURE 1).

The sequence of the invention has been obtained by codon-optimization for Pichia species usage (13, 5). The invention also provides the selection of GS l 15 (his4) as a host strain particularly tolerant to saporin expression (FIGURE 2), making also available expression vectors for producing untagged or hexahistidine-tagged saporin fusions (FIGURES 3 AND 4) for easy purification of the fusion chimaeras.

The present invention discloses the procedures to select novel recombinant strains allowing the production of recombinant saporin as well as fusion toxins such as the ATF-saporin chimaera in P. pastoris host cells in much larger amounts than previously achieved in bacterial host cells (7). The invention also provides examples of production of two new scFv αCD22 fusion chimaeras, based on a single-chain variable fragment (scFv) obtained by RT-PCR of RNA extracted from a hybridoma producing a monoclonal antibody against the acute leukaemia marker antigen CD22 that was previously used to produce an anti-CD22 saporin immunotoxin (14). The sequence encoding the anti-CD22 scFv was first codon-optimized (Sequence ID #3) and included the flexible 218 linker between the light and heavy chains of the scFv (15) and was then fused either to the codon optimized saporin sequence in pITalphawt vector (FIGURE 3A) for secretion of 4KBopt218LSAPopt fusion or to a sequence encoding hexahistidine-tagged saporin in pPICSAPH vector for secretion of 4KBopt218L-SAPHis6. The process of the invention is advantageous allowing increased yields of secretory recombinant proteins.

The DNA sequences coding for the toxic protein/chimaera are under the control of an inducible and tightly regulated promoter, such as the well known alcohol oxydase- 1 promoter (AOXl) that must be repressed in the presence of dextrose/glucose during the biomass increase step. The addition of dextrose is necessary to avoid toxicity (of the synthesized saporin) during the cell growth phase, and in our hands glycerol feeding appears to be not optimal for the induction of this toxic protein. Linearized DNAs are used to transform electrocompetent GS l l 5(his4) cells for integration of the expression cassette(s) into the host genome, ensuring stability of the recombinant transformed strain after repeated sub-culturing. The domain(s) used in the production of saporin chimaeras should impart good secretory properties and would better undergo codon-optimisation to be best suited in the production of fusion toxins. Finally induction upon culturing the P. pastoris cells in the presence of methanol can be achieved.

The process of the invention preferably comprises the following steps:

1) Codon-optimization of the saporin cDNA sequence, based on the codon usage of P. pastoris (Sequence ID #1) and of the sequences for targeting domain(s) such as the Amino-terminal fragment of human urokinase (Sequence ID #2) and that coding for the single- chain variable fragment scFv αCD22 (Sequence ID #3).

2) Subcloning of said sequences in suitable recipient expression vectors for their integration into P. pastoris genome.

3) Transformation of electrocompetent GS l 15 (his4) host cells.

4) Selection of the best expressing clones in small-scale induction preparations by screening induced zeocine-resistant clones expressing secretory saporin or the fusion protein.

5) Medium-scale induction at 10 ODImL in shake flask conditions of the most promising clone(s) using dextrose feeding before the induction step to ensure a tight repression and low toxicity 6) Concentration of the media for Nichel-affinity (IMAC) purification of the histidine-tagged fusion proteins or for chromatographic purification of the secretory saporin and of chimeric proteins such as 4KBopt218L-SAPopt.

7) Assay for the cytotoxic activities of the purified recombinant protein(s) on leukaemia target cells.

In step 2, the toxic polypeptides should be directed to the extracellular space (16) by the signal and the propeptide region of the preproalpha mating factor domain of Saccharomyces cerevisiae (alpha- factor) which has been widely used for the expression of secretory proteins in P. pastoris (3,4) and is present in the pPICZalpha vectors commercially available from Invitrogen. It was found that the Stel3 aminodipeptidyl site(s) should be present, together with the Lys-Arg site recognized by the KEX-2 endoprotease, for proper processing by KEX-2 protease of the saporin precursor polypeptide.

The process of the invention is particularly advantageous for the expression of recombinant immunotoxins containing the saporin toxic domain fused to specific targeting domain(s) directing the fusion chimaera against selected tumor cells (17, 18). Examples of said targeting domains comprise antibodies or single-chain antibody- fragments (scFv) against tumor antigens as in the case of 4KBopt218LSAPopt and 4KBopt218L-SAPHis6 fusions, or ligand binding domains, as in the case of the ATF domain binding to the human urokinase receptor, or growth-factor domains.

According to a preferred embodiment of this invention, the chimaeric fusion toxin ATF-saporin, targeting potentially metastatic tumor cells overexpressing the human urokinase-receptor, is produced in P. pastoris cells. ATF-saporinH6 consists of the amino-terminal fragment of human prourokinase (ATF), including the ligand-binding domain to the urokinase receptor, fused through a trialanine linker peptide to a saporin variant having 6 histidines at the C-terminus. A different chimaeric fusion between ATF and saporin (ATF-SAP) has been previously expressed with some difficulties in the cytosol of E. coli because of folding problems in the ATF domain, due to the presence of 6 disulphides (7). The RIP activity of the ATF-saporin chimaera produced in Escherichia coli or Xenopus laevis oocytes has been documented previously in detail (7, 8). Xenopus laevis oocytes (8), are able to reproduce the co- and post-translational events during heterologous protein synthesis (19) but must be protected by co-injection of neutralizing anti- saporin antibodies, being sensitive to RIP activity (20, 21). Xenopus expression system has demonstrated to be a suitable model system to produce secretory ATF-saporin chimaera for analytical purposes but it is not a suitable expression system for the large scale expression needed for the clinical development of this class of molecules.

ATF-saporinH6 was secreted by GSl 15 (his4) cells at 3.5 mg/L under shake-flask conditions and was easily purified in one step, as detailed in the following examples. The recombinant secretory chimaera shows the expected RIP activity against target cells, as also did the recombinant secretory saporin purified from the extracellular medium of a clone secreting up to 30 mg/L saporin in flask conditions, as described below. We further characterized a fully optimized fusion of ATF and saporin (ATFopt-SAPopt) expressed at 15 mg/1 under flask growth conditions, that was successfully purified by a single step chromatographic procedure.

The scFvαCD22 fusion chimaera was also secreted at 2 mg/L in flask conditions whether having an hexahistidine-tag (4KBopt218L-SAPHis6) or not (4KBopt218LSAPopt) and the chimaeras were purified either by metal -affinity (4KBopt218L-SAPHis6) or by cation-exchange chromatography (4KBopt218LSAPopt). Importantly, both showed specific cytotoxic activity when assayed against target Daudi cells (FIGURE 5) in a protein synthesis inhibition (PSI) assay. Description of the figures:

Figure Legends:

Figure 1: Effects of codon-optimization on saporin expression levels

Pichia pastoris GS l 15 {his4) hosts cells were transformed with the native gene constructs or the codon-optimized constructs. (A-B) The histograms show the distribution of up to 20 independent GS \ \ 5(his4) clones (y-axis, number of clones) induced 48 h for expression of either native (white) or optimized (opt; black) constructs coding for active saporin (A) or active-site mutant SAPKQ (B). The amount of secreted polypeptide was quantified by densitometry, using seed-extracted saporin as a reference standard. Expression levels (x-axis) are reported in different ranges in mg/L from the lowest (0-4mg/L) to the highest expressors (20-25mg/L). In the case of catalytically active saporin (A) clones expressing high levels of the protein were present only when the codon-optimized sequence (SAP _OPT ) was used. Conversely, in the case of the active site mutant (B) clones accumulating high levels of the protein were present both when the native (nativeKQ) or the codon-optimized (KQ _OPT ) sequences were used. (C-D) Growth curves of at least two mock-induced clones (triangles) for each experiment were compared to the averaged growth curves of 20 clones induced for expression of either the optimized (diamonds) or native saporin genes (squares) having an intact (C) or mutated catalytic site (D). Please note that independently from gene codon-optimization or the relative expression levels, when GS 115(/zw4)cells synthesized active saporin polypeptides their growth rates decreased as compared to mock controls.

Figure 2: Selection of the best suited host strains for saporin expression

14-20 randomly picked colonies expressing SAPopt or KQopt in three different hosts (KM71H, SMDl 168, GS l 15 (his4)) were induced for 48h and the amount of saporin present in equivalent amounts of media was quantified by Western or slot-blots using seed-extracted saporin as a reference standard. The bars indicate the ranges of accumulation in the medium (mg/L) of either catalytically active (SAPopt, upper bars) or mutated (KQopt, lower bars) saporin in GSl 15 (his4) (black), KM71H (hatched) and SMDl 168 (white) strains. Please note that only in GS l 15 (his4) strain equivalent levels of secretion among active saporin and mutated KQ polypeptides could be obtained.

Figure 3: Schematic diagram of the Fusion Expression Vectors pITalphawt (A) containing the codon-optimised SAPopt gene or (B) pPICSAPH containing the SAPopt gene with an extra hexahistidine tag (6xHis)

The vectors derive from pPicZalphaB vector (Invitrogen) and contains the pUC origin of replication, shown together with Alcohol oxydase- 1 promoter (AOXl) and its transcription terminator, a zeocine resistance gene (ZeoR), the alpha mating factor signal peptide and prodomain, including the restriction enzyme sites present in the linker between EcoRI and Notl that is placed at 5' end of the SAPopt sequence: in this way a triple Alanine linker is inserted between any targeting domain and the plant toxin domain.

Figure 4: Expression, single step purification of ATF-saporinH6 and determination of its cytotoxic activity

(A) Western with anti-saporin serum of 48 h medium-scale induction of the best GSl 15 (his4) expressing clone producing ATF-saporinH6. Total cell lysate of cells expressing ATF-saporinH6 construct (lane 2) and with an equivalent amount of corresponding medium (lane 3) were analysed for comparison. Lane 1 : mock-induced lysate as a control.

(B) The 48 h medium from induced cultures containing ATF-saporinH6 was concentrated and ATF-saporinH6 nichel-purifϊed as described in Example 2 section. Total concentrated medium (Tot, lane 4), column flow-through (Flow-t., lane 5) and concentrated eluates (150 mM or 300 mM imidazole elution, lanes 6 and 7, respectively) were subjected to SDS-PAGE followed either by silver-staining (upper panel) or western-blotting with anti-saporin serum (lower panel). For quantifications, standard saporin (SAP-S) was loaded in lanes 1-3: 90, 225 and 450 ng (upper panel) or 30, 60 and 100 ng (lower panel). The position of ATF-saporinH6 is indicated by an arrowhead. The asterisk shows degradation products containing saporin-reactive polypeptides. Molecular weight markers positions are shown on the left. (C) ATF-saporinH6 activity was compared to that of saporin when assayed on acid-washed U937 human monocytes as described in (6, 7) and shows an IC ₅₀ of 6 x 10 ^"11 M. Squares: ATF-saporinH6. Dots: seed-extracted saporin.

Figure 5: Protein synthesis inhibition assay (PSI) on human leukaemia Daudi cells

4KBopt218L-SAPHis6 was produced and purified by Ni-affinity, essentially as described for ATF-saporinH6. Secretory saporin was purified by cation-exchange chromatography and human acute leukaemia Daudi cells were exposed as in (12) to increasing concentrations of purified histidine- tagged fusion (4KBop218L-SAPHis6, squares), recombinant secretory saporin purified from Pichia (Pichia Saporin, triangles), seed-extracted saporin (Saporin seed, diamonds) and to an anti CD22 saporin immunotoxin (4KB128-SAP IT, dots), as control, for 48h. At the end of the exposure the cells were washed and pulse-labelled with tritiated leucine essentially as described in (14), and results expressed as percent tritiated leucine incorporation relative to the untreated control Daudi cells. Please note that seed-extracted saporin and secretory saporin purified from P. pastoris medium have identical inhibitory activity showing an IC ₅₀ of 6 x 10 ^"7 M whereas the histidine-tagged fusion has the same cytotoxic activity as the untagged (not shown) being at least a hundred times more effective than non-targeted saporin against Daudi leukaemia cells, showing IC _5O of 1 x 10 ^~9 M. A conventional saporin immunotoxin against CD22 leukaemia marker antigen, used as a control, showed an IC ₅₀ of 1.7 x 10 ^{"1 1} M.

The invention is disclosed in more detail in the following examples.

EXAMPLE 1

Construction of a vector for expression of secretory saporin in P. pastoris

A synthetic gene (Sequence ID # 1) coding for a novel precursor of a seed-isoform of saporin whose DNA has been codon-optimized (12) based on the yeast codon-usage reported among available Pichia pastoris coding sequences (CDS) (choosing those most frequently represented in highly expressed P. pastoris proteins) was subcloned in pUC57 vector with EcoRI and EcoRV restriction sites (bold-face) and obtained as a pUC57SOLlGBXbaPst construct from GenScript Corporation, Piscataway,

NJ.

Sequence ID #1 :

IGAATTCACGATGAAAATCTATGTTGTTGCTACTATTGCTTGG ATTTTGTTGCAATTTTCCό 1

GCTTGGACCACTACTGATGCTGTTACCTCCATTACTTTGGATTT

GGTTAATCCTACTGCT 121

GGTCAATACTCCTCTTTCGTTGATAAGATTAGAAACAACGTTA AAGACCCTAATTTGAAG 181

TACGGTGGAACTGACATTGCTGTTATTGGACCACCATCTAAAG AGAAGTTTTTG AGAATT241

AACTTTCAATCTTCCAGAGGTACTGTTTCCTTGGGTTTGAAGAG AGACAACTTGTACGTT301

GTTGCTTACTTGGCTATGGATAACACTAATGTTAATAGAGCTT ACTACTTTAG ATCCG AA361

ATCACCTCTGCTGAGTTGACTGCTTTGTTCCCAGAAGCTACTAC AGCTAATCAAAAGGCT421

TTGGAATATACTGAGGATTACCAGTCCATTGAAAAGAATGCTC AAATTACTCAAGGAGAT481

AAGTCCAGAAAGGAGTTGGGTTTGGGAATCGACTTGTTGTTGA CTTCTATGGAAGCTGTT541

AATAAGAAGGCTAGAGTTGTTAAGAACGAAGCTAGATTCTTGT TGATTGCTATTCAAATG601

ACCGCTGAAGTTGCTAGATTCAGATACATTCAAAACTTGGTTA

CTAAAAATTTTCCTAAC661

AAGTTTGATTCTGACAACAAGGTTATTCAGTTCGAGGTTTCTTG GAGAAAGATTTCTACT721

GCTATCTACGGAGATGCTAAGAACGGAGTTTTTAACAAGGATT ACGATTTCGGTTTTGGA781

AAAGTTAGACAAGTTAAGGACTTGCAAATGGGTTTGTTGATGT ATTTGGGAAAGCCAAAG841

TCTTCCAACGAGGCTAATTCCACTGCTTACGCTACAACAGTTTT GTAACTCGAGGATATC

The sequence coding for the mature protein was amplified using

PFU-Taq DNA polymerase and the following oligonucleotides 5'CCGGAATTCGTTACCTCCATTACTTTGG-S ' (sense oligo) (Sequence ID #4).

5 '-CAGTAGCGGCCGCTTACTTTGGCTTTCCCA-S ' (anti-sense oligo) (Sequence ID #5). The digested and purified EcoRl-Notl fragment was subcloned into EcoRI-Notl cut pPICZalphaA commercial vector to obtain

SAPopt construct driving secretion of a mature saporin ending with LyS ₂₅₃ and having, following Stel3 cleavage sites, an extra GIu and Phe (encoded within EcoRl site) at the N-terminus just before VaI ₁ To accurately evaluate host toxicity an equivalent construct was prepared in which two key residues at the catalytic site of saporin were mutated to produce a SAPKQ variant in which GIu ₁₇₆ is changed into Lysine and Arg _π9 into Glutamine and that has greatly reduced RIP activity (22). Codon-optimization advantageously affected yields of active saporin in the GSl 15 {his4) strain (FIGURES 1 and 2). Three Pichia pastoris host strains have been further investigated and compared for the relative accumulation levels of the mutated SAPKQ and wild-type saporin (FIGURE 2) using the codon-optimized expression constructs. GS \ \5{his4) was the most tolerant P. pastoris strain for the expression of saporin, since essentially no difference in the range of expression levels was found among clones expressing the SAPopt or the KQopt construct (and this strain was selected as the most promising host strain).

EXAMPLE 2

Construction of ATF-saporinH6 for secretion and single step purification of ATF-saporin chimaera

The aminoacidic sequence 1-135 of human urokinase was amplified using as a template the plasmid pSP64TpAS (8) containing the human ATF DNA fragment, with a 5 '-CGCTCGAGAAAAGAAGCAATGAACTTCAT-S ' forward primer (Sequence ID #6) inserting the restriction enzyme site Xhol (underlined) and restoring the KEX-2 endoproteolytic Lys-Arg site just ahead the first aminoacid in ATF sequence (Serj) and a reverse primer 5 '-CAGTAGCGGCCGCTTTTCCATCTGC-S ' (Sequence ID #7) inserting a restriction enzyme site Notl (underlined) after LyS ₁₃₅ .

The fragment encoding human ATF was digested with Xhol and Notl, purified and ligated to pPICSAPH vector opened with the same enzymes. Electrocompetent GS l 15 (his4) cells were transformed and the best expressing clone selected following induction at 2 OD/mL with 0.5% Methanol. 10-25 μl of incubation medium after 48 h induction were assayed by slot-blot with an anti-saporin serum to identify the best expressing clones. Medium-scale preparation of the ATF-saporinH6 from this best expressing clone was achieved by growth 16h in liquid culture of a single freshly picked colony. For long term storage, cells (10-15 OD ₆ oo) were centrifuged 5 min at 1,560 g, resuspended in 1 ml of YPD [1% (w/v) yeast extract, 2% (w/v) peptone or tryptone, and 2% (w/v) dextrose] supplemented with 15% of sterile glycerol and frozen immediately. Stocks were stored at -80 ⁰ C until use. For medium scale inductions, a freshly plated single colony obtained from these stocks was inoculated in 10 ml of YPD with 50 μg/mL Zeocin and 5 ml of this overnight preculture inoculated in a 2 L-flask containing 250 ml of the same medium, incubated at 30 ⁰ C, 250 rpm overnigth. Cells were centrifuged at 159Og for 10 min at room temperature resuspended in 250 ml of BMMY [1% (w/v) yeast extract; 2% (w/v) peptone; 10OmM phosphate buffer, pH 6.0; 1,34% (w/v) yeast nitrogen base, 4xlO ^"5 % (w/v) biotin; 4xlO ^"3 % (w/v) histidine; 0,5% (v/v) methanol] (without antibiotics) at final OD ₆₀ o/ml of 10 and incubated at 30 ⁰ C, 250 rpm for 48h. Methanol 0, 5% (v/v) and histidine 0, 04% final were added at the beginning of the induction period and 24 h later. At the end of 48 h induction period, cells were removed by centrifugation at 5000 g for 10 min at 4°C. The medium was exchanged against PBS pH 7.6 [137mM NaCl; 2,7mM KCl; 1OmM Na ₂ HPO ₄ ; 2mM KH ₂ PO ₄ ] and concentrated 8-10 fold through a Hydrosart® Vivaflow 200 membrane, 10000 cut-off (Vivascience, Sartorius, Stedim Biotech, Firenze, Italy). The concentrated media containing the recombinant fusion proteins were supplemented with Complete™ or in the case of ATF-saporinH6 and 4kBopt218L-SAPHis6 with EDTA-free Complete™ (Roche, Monza, Italy), and then media were kept frozen before thawing at 30 ⁰ C for subsequent protein purifications. ATF-saporinH6 protein was purified from the concentrated medium with IMAC nichel affinity and eluted as a single peak at 150 mM imidazole (FIGURE 4). Eluates were exchanged against PBS pH 7.6 and concentrated to 1 ml with Vivaspin columns 10000 cut-off concentrators (Vivascience, Sartorius, Stedim Biotech, Firenze, Italy) by centrifuging at 5000 g. Samples were analysed by SDS- PAGE and subjected to silver staining and seed-extracted saporin (SAP-S) was used as a reference standard. ATF-saporinH6 cytotoxic activity was assayed as described (7, 8). SAP-S was purified from Saponaria officinalis seeds as previously described (23).

As a further improvement of the above described procedure, the sequence coding for the ATF domain was optimized for expression in Pichia pastoris. A codon-optimised ATF fragment (Sequence ID # 2) with codon- usage of Pichia pastoris species was designed following the same rules described for the saporin codon-optimised sequence. The resulting synthetic DNA (sequence ID#2) was obtained from GenScript Corporation, Piscataway, NJ subcloned in pUC57 with EcoRI and EcoRV restriction sites to obtain pUC57ATFopt construct.

Sequence ID #2: ATF optimised gene sequence (Xhol-Notl fragment, restriction sites in bold) is shown below:

CTCGAGAAGAGATCTAATGAATTGCATCAAGTTCCATCCAACT GTGACTGTTTGAACGGAGGTACTTGTGTTTCTAACAAGTACTTTTCTA ACATTCATTGGTGTAACTGTCCAAAGAAGTTTGGTGGTCAACATTGT GAAATTGACAAATCTAAGACTTGTTACGAGGGTAATGGTCATTTTTA CAGAGGAAAGGCTTCTACTGATACTATGGGTAGACCATGTTTGCCAT GGAACTCTGCTACTGTTTTGCAACAAACTTACCATGCTCATAGATCC GATGCTTTGCAATTGGGTTTGGGAAAGCATAACTACTGTAGAAATCC AGATAACAGAAGAAGACCATGGTGTTACGTTCAAGTTGGATTGAAA CCATTGGTTCAAGAATGTATGGTTCATGATTGTGCTGATGGTAAAGC GGCCGC An expression construct coding for ATF-Saporin was obtained by fusing the codon optimized ATF sequence to the optimized saporin sequence. The construct was inserted into GS l 15 (his4) cells and a clone was isolated that secreted ATF-Saporin in the culture medium at 15 mg/L in shake-flask conditions.

EXAMPLE 3

Construction of the scFv αCD22 fusion 4KBopt218L-SAPopt or the histidine-tagged 4KBopt218L-SAPHis6 for metal-affinity purification

The single-chain variable sequences were obtained with standard procedures by reverse transcription-PCR on RNA obtained from the hybridoma producing anti CD22 monoclonal antibody that was used for producing an anti-CD22 SAP immunotoxin (14). A synthetic codon-optimized DNA was obtained from GenScript Corporation, Piscataway, NJ and DNA subcloned in pUC57 with EcoRI-Hindlll to obtain pUC57αscFvCD22opt construct and including Sfil and Notl restriction sites for subcloning into pITalphawt or pPICSAPH. Sequence ID #3 is shown below and corresponds to the anti-CD22 optimised sequence including a flexible linker joining the V _H and V _L , known as 218 linker (15), in the V _H -218L-V _L configuration. Sfil and Notl restriction sites are shown in bold.

GGCCCAGCCGGCCATGGCTGAGGTTAAATTGATGGAATCCGG

TGGTGGTTTGGTTCAACCAGGTGGATCTATGAAGTTGTCCTGTGTTG CTTCTGGTTTTACTTTTTCCAACTACTGGATGAACTGGGTTAGACAAT CACCAGAAAAAGGATTGGAATGGGTTGCTGAGATTAGAGTTAAATC CAACAATTACGCTACTCACTACGCTGAATCTGTTAGAGGAAGATTCA CTACCTCCAGAGATGACTCCAAGTCTTCCGTTTACTTGCAAATGAAC AATTTGAGAGGTGAGGATACTGGAATCTACTACTGCAGTAGAGTTTA CTACTATGGTCACGACTACGCTATGGATTACTGGGGTCAAGGTACCT CCGTTACTGTCTCGAGTGGTTCCACTTCTGGTTCTGGAAAGCCAGGA TCAGGAGAGGGTTCTACCAAGGGATCCGCTGTTGACATTGTTTTGAC TCAATCTCCAGCTATTATGTCTACTTCATTGGGTGAAAGAGTTACTAT GACTTGTACTGCTTCATCTCCAGTCTCATCTACCTATTTGCACTGGTA CCAACAAAAGCCTGGTTCTTCTCCTAAGTTGTGGATCTACTCCACCTC TAAGTTGGCTTCCGGTGTTCCTGATAGATTTTCTGGATCTGGTTCCGG AACTTCATACTCATTGACTATTTCTTCCATGGAAGCTGAGGATGCTG CTACCTACTACTGTCACCAGTACCACAGATCACCAAGAACCTTCGGT GGTGGTACCAAATTGGAGATTAAAAGAGCGGCCGC.

This DNA was subcloned using the Sfil-NotI sites (in bold) into pPICSAPH opened with the same restriction enzymes to obtain 4KBopt218L- SAPHisό or into the vector pITalphawt containing SAPopt to obtain 4KBopt218L-SAPopt as an identical construct except for the absence of the C-terminal 6 histidines. The same procedure as described for ATF-saporinH6 was followed to obtain and analyze the induced clones expressing the scFv-saporin fusions and to select the best expressing clones to be induced at 10 OD/mL in flask conditions, except that induction was performed at 23 ⁰ C. At the end of the 48 h induction period the media were concentrated exactly as described for ATF-saporinH6 and subjected either to cation-exchange chromatography (4KBopt218L-SAPopt) or to nichel-affinity chromatography (4KBopt218L-SAPHis6), in the same conditions as reported above for ATF-saporinH6. The purified proteins were assayed against target leukaemia Daudi cells and their cytotoxicities compared, showing no significant differences when estimated as the concentration inhibiting 50% incorporation of tritiated leucine relative to control untreated Daudi cells (IC50). FIGURE 5 shows the dose-response curves obtained by exposing Daudi cells to 4KBopt218L-SAPHis6 or saporin secreted by P. pastoris, to seed-extracted saporin and to the anti-CD22 immunotoxin (14). References

1. Soria, M. R., et al. (1992) Targeted Diagn. Ther. 7, 193-21.

2. Peumans, W. J., et al. (2001) FASEB J. 15, 1493-1506.

3. Cereghino, J. L., et al. (2000) FEMS Microbiol. Rev. 24, 45-66.

4. Daly, R.., and Hearn, M. T. (2005) J. MoI. Recognit. 18, 119-38.

5. Woo, J. H., et al. (2002) Protein Expr. Purif. 25,270-282.

6. Rajamohan F., et al. (2000). Protein Expr. Purif. 18, 193-201.

7. Fabbrini, M. S., et al. (1997) FASEB J. 11, 1 169-1 176.

8. Fabbrini, M. S., et al. (2000) FASEB J. 14, 391-398.

9. Hegde, R. S., et al. (2006) Trends Biochem. Sci. 31,563-71.

10. Shaffer, K. L., et al. (2005) Dev. Cell 9,545-554.

11. Plemper, R. K., et al. (1999) Trends Biochem. Sci. 24, 266-70.

12. Lombardi A., et al. (2010) FASEB J. 24, 253-265.

13. Sreekrishna, K. (1993) In Industrial Microorganism: Basic and Applied Molecular Genetics, Am. Soc. Microbiol. (Baltz, R.H., Hegeman,G.D. and

Skatrud, P.L., eds) pp.119-126, Washington DC.

14. Flavell, D. J., et al. (1997) Cancer Res. 7, 4824-4829.

15. Withlow M. et al. (1993) Protein eng. 6, 989-995.

16. Rapoport, T. A. (1991) FASEB J. 5, 2792-8.

17. Fabbrini, M. S., et al. (2003) in "Bacterial, Plant&Animal Toxins" (Ascenzi, P., Polticelli, F. a., and Visca, P., Eds.), pp. 69-99, Research Signpost, Kerala, India.

18. Flavell, D. J. (1998) Curr. Top. Microbiol. Immunol. 234, 57-61.

19. Bossi, E., et al. Methods MoI. Biol. 375, 107-31.

20. Saxena, S. K., et al. (1989) J. Biol. Chem. 264, 596-601.

21. IT 1298437

22. Zarovni, N., et al. (2007) Cancer Gene Ther. 14, 165-73.

23. Flavell, DJ., et al. (2001) Br. J. Cancer. 84,571.