Effect of N-terminal deletions on the activity of pokeweed antiviral protein expressed in E. coli

Biochimie (1998) 80, 1069-1076 © Socirt6 franqaise de biochimie et biologie molrculaire / Elsevier, Paris

Effect of N-terminal deletions on the activity of pokeweed antiviral protein expressed in E. coli J i a n h u a X u a, A l i c e X. M e n g a, K a t h l e e n L. H e f f e r o n a, I v a n G . I v a n o v b, M o u n i r G. A b o u h a i d a r ~*

~Department of Botany, University of Toronto, 25 Willcocks St., Toronto M5S 3B2, Canada bDepartment of Gene Regulation, Institute of Molecular Biology, Bulgarian Academy of Sciences, 1113 Sofia, Bulgaria (Received 15 January 1998; revision received 18 March 1998; accepted 30 June 1998) A b s t r a c t - Pokeweed antiviral protein (PAP) from Phytolacca americana is a highly specific N-glycosidase removing adenine residues (A4324 in 28S rRNA and A 2660 in 23S rRNA) from intact ribosomes of both eukaryotes and prokaryotes. Due to the ribosome impairing activity the gene coding for mature PAP has not been expressed so far in bacteria whereas the full-length gene (coding for the mature 262 amino acids plus two signal peptides of 22 and 29 amino acids at both N- and C-termini, respectively) has been expressed in Escherichia coli. In order to determine: 1) the size of the N-terminal region of PAP which is required for toxicity to E. coli; and 2) the location of the putative enzymatic active site of PAP, 5"-terminal progressive deletion of the PAP full-length gene was carried out and the truncated forms of the gene were cloned in a vector containing a strong constitutive promoter and a consensus Shine-Dalgarno ribosome binding site. The ribosome inactivation or toxicity of the PAP is used as a phenotype characterized by the absence of E. coli colonies, while the mutation of PAP open reading frames in the small number of survived clones is used as an indicator of the toxicity to E. coli cells. Results showed that the native full-length PAP gene was highly expressed and was not toxic to E. coli cells although in vitro ribosome inactivating activity assay indicated it was active. However, all of the N-terminal truncated forms (removal of seven to 107 codons) of the PAP gene were toxic to E. coli cells and were mutated into either out of frame, early termination codon or inactive form of PAP (i.e., clone PAPAl 07). Deletion of more than 123 codons restored the correct gene sequence but resulted in the loss of the antiviral and ribosome inactivating activities and by the formation of a large number of clones. These results suggest that full-length PAP (with N- and C-terminal extensions) might be an inactive form of the enzyme in vivo presumably by inclusion body formation or other unknown mechanisms and is not toxic to E. coli cells. However, it is activated by at least seven codon deletions at the N-terminus. Deletions from seven through to 107 amino acids were lethal to the cells and only mutated forms (inactive) of the gene were obtained. But deletion of more than 123 amino acids resulted in the loss of enzymatic activity and made it possible to express the correct PAP gene in E. coli. Because deletion of Tyr94 and Va195, which are involved in the binding of the target adenine base, did not abolish the activity of PAP, it is concluded that the location previously proposed for PAP enzymatic active site should be reassessed. © Socirt6 franqaise de biochimie et biologie mol6culaire / Elsevier, Paris gene expression / N-terminal deletion / pokeweed antiviral protein / toxicity I. Introduction The pokeweed (Phytolacca americana) produces at least three ribosome-inactivating proteins (PAP, PAP-II and PAP-S) found in different tissues and at different stages of plant development (see for reviews [1, 2]). They belong to the type 1 ribosome-inactivating proteins (RIPs), consisting of a single polypeptide chain and possessing a rRNA sequence specific N-glycosidase activity. Two of the RIPs (PAP and PAP-II) are active on both eukaryotic and prokaryotic ribosomes, removing A 4324 from 285 rRNA [3] and m 2660 from 23S rRNA [411. In contrast, the type 2 RIPs (consisting of two polypeptide chains) are active on eukaryotic ribosomes only. The RIP target sites are part of a highly conserved region 5'AGUACGA43Z4/Z66°GAGGAAC-3' in the large ribosomal subunit rRNAs of both eukaryotes and prokaryotes which is responsible for binding elongation factors EF-G and * Correspondence and reprints

EF-Tu [5]. Naturally RIPs are found trapped within the cell wall matrix [6] and their antiviral activities are explained by host ribosome impairing after introduction into the cytoplasm by the pathogen [7]. Due to their antiviral as well as anticancer activities, the P. americana RIPs have attracted the interest of many researchers for their possible applications as therapeutic agents (see for reviews [l, 2]). The genes of the three P americana RIPs (PAP [8], PAP-II[9] and PAP-S [10]), have been isolated from tissue-specific cDNA libraries and sequenced. The PAP gene carries an open reading frame of 939 nt coding for the mature PAP protein (262 aa) plus an N-terminal signal peptide of 22 aa [8] and a C-terminal extra peptide of 29 aa [11]. This gene has been expressed in E. coli cells under an inducible (lac) promoter with an extremely low yield (0.13-0.16% of the total bacterial protein) [12]. It was found that even the low level of gene expression slowed down significantly bacterial growth, whereas higher expression levels (up to 8% of the total protein) of ricin A

1070 chain and Mirabilis antiviral protein genes did not affect cell growth [12-14]. Chen et al. [12] found that elimination of N-terminal signal peptide codons (22 amino acids) from the PAP gene led to an immediate cell death. The failure to express the PAP gene coding for mature PAP protein in E. coli, the high level of expression of active (A) chains of type 2 RIPs and direct in vitro experiments were the reason for formulating the generally accepted conclusion that PAP is highly toxic (in vivo) for both prokaryotic and eukaryotic cells [1, 2, 12, 15-17]. The high toxicity of PAP is a serious obstacle for applying the routine gene manipulation techniques for structure-function relationship studies. The structure of the PAP active site is predicted on the basis of high similarity between the three-dimensional structures of PAP and the ricin A-chain [1, 11, 17]. In order to check whether the structural differences between these two proteins are responsible for their different ribosome specificity, a series of swap PAP/ricin A-chain hybrids has been recently generated [17]. Biological studies showed that the major structural differences between the two proteins did not account for their different specificity. This study also shows that the C-terminal domains in PAP do not contribute significantly to the ribosome recognition. Aiming to find correlation between inhibition of bacterial cell growth and ribosome inactivating activity, two spontaneous mutants (F196Y and K211R) of PAP genes have been expressed in E. coli [16, 18]. The results and conclusions obtained by the two research groups are controversial. Here we present the results of cloning and expression of a series of N-terminal deletion mutants of the PAP gene in E. coli as well as biological properties of their products. Unlike in other published work, the PAP gene variants in this study were expressed under a strong constitutive promoter. This allowed us to use bacterial cell mortality as a phenotype indicating the presence of extremely toxic proteins. The full-length PAP is not toxic to E. coli cells and may be produced in large quantities. However, progressive amino acid deletion from the N-terminus unexpectedly renders the PAP toxic to E. coli cells. It is proposed that the region beyond the N-terminal 107 amino acids is responsible for the PAP protein toxicity to E. coli, in other words, the first 107 amino acids may not be essential for the protein's biological activity.

2. Materials and methods

2.1. Cloning of PAP cDNA in PBS + Total RNA was extracted from Phytolacca americana leaves as described by Ausubel et al. [119]. cDNA was prepared from total RNA using an oligo(dT12_]8 ) primer and M-MLV reverse transcriptase (Clontech Lab. Inc.). Oligonucleotides for PCR and sequencing were synthesized and the polymerase chain reaction (PCR) was

Xu et al. carried out in a 100pL total volume using Ventcx°+ polymerase (New England Biolabs). The reaction cycle was repeated 35 times under the following conditions: denaturation at 93 °C for 30 s, annealing at 62 °C for 30 s and extension at 72 °C for 2 min. The PCR cDNA product (1.1 kb) containing the PAP gene with BamHI sites at both ends was cloned into the unique BamHI site of pBS +. Clones were screened by restriction endonucleases and DNA structure was verified by sequence analysis. DNA sequencing was carried out using a commercial (US Biochemicals) Sequenase kit (version 2.0) following the manufacturer's manual. One of the positive clones designated pBS+-PAP was used in the next experiments. The initiation (ATG) codon in this clone was taken as a checkpoint for numbering nucleotides and codons in the PAP gene. 2.2. Construction of full-length PAP cDNA for constitutive expression in the vector pP1R9 PAP gene was placed into an introduced BamHI site in the expression vector pP~SD [20] containing a strong synthetic bacteriophage T5 early (T5P25) promoter and a synthetic Shine-Dalgarno (SD) ribosome-binding site (figure 1). Clones bearing the PAP gene in sense and antisense orientations were selected by restriction analysis. 2.3. Progressive N-terminal deletion mutagenesis of the PAP gene 2.3.1. Unidirectional deletion by exonuclease Ill Unidirectional deletion was carried out by an Exo-Size Deletion Kit (NEB) according to the manufacturer's instruction. Two restriction sites, ApaI and BgllI (one resistant and the other sensitive to exonuclease III), were introduced into the plasmid expression vector by PCR to form a new plasmid (figure 1). This new plasmid (3-5 ~tg) was digested with BgllI and ApaI to generate 3'recessive and 3'overhang ends, and then digested with five units of exonuclease III in 50 mM Tris-HC1, 5 mM MgC12, 10 mM [3-mercaptoethanol, pH 8.0 at 20 °C at different time intervals. Aliquots were taken out and the reaction was stopped by l FaL of 0.5 M EDTA, pH 7.5. DNA was precipitated and incubated with five units of mung bean nuclease in 50 pL of 50 mM NaAc, pH.5.0, 30 mM NaC! and 1 mM ZnS04 at 25 °C for 30 min. The reaction was stopped by 1 pL of 0.5 M EDTA, pH 7.5, and the DNA (1 lag/mL) was further treated with 10 units E. coli DNA polymerase (Klenow fragment) in a 50 pL (commercial buffer) at 37 °C for l h. Blunt ended DNA was finally self-ligated by T4 DNA ligase. 2.3.2. Site-directed deletions by PCR Five oligonucletides (four forward and one reverse) were used for the PCR mutagenesis. The four forward primers were designed to remove the first 22 (signal

Effect of N-terminal deletions

EcoRI

pBR322

I

Y0-rt

BamHI

/Apal

I(

BamHI

I

Bglll

Icl .... -............

-'~22........"-". . . . . . . .

262

NC I pBR322

29

+1

ataaatttgaacctactcgac~~a~ggatccgATG... SD

Figure I. Schematic structure of the vectors for constitutive expression of PAP gene constructs. Expression vectors are based on the cloning plasmid pBR322. Three successive elements are inserted after the EcoRI site: P1 promoter (a synthetic analog ~f the bacteriophage T5 early promoter) (19); SD, a consensus Shine-Dalgarno containing sequence (shown below) and PAP gene (full-length or N-terminal deletion mutants). BamHI site was used for the insertion of the full length PAP gene. For N-terminal exonuclease Ill deletions, two more restriction sites are introduced in the expression plasmid. N and C, N- and C-terminal signal peptides; NC, non-coding region in the PAP gene. Dashed area, the portion in the PAP gene (up to 190 codons) subjected to deletion mutagenesis; numbers under the PAP gene correspond to the mature protein (262) and the two terminal signal peptides (22 and 29). Boxed nucleotides represent Shine-Dalgarno consensus sequence.

peptide), 49, 71 and 89 codons respectively and the reverse primer was made matching the non-coding sequence of the PAP gene. PCR reaction was carried out as described above and the single band product was digested with BamHI, and cloned into the expression vector pP1SD (figure 1).

2.4. Preparation of PAP antibodies A male New Zealand rabbit was injected intramuscularly with 0.5 mg of pure plant PAP (CalbiochemNovabiochem Corporation) in 1 mL emulsion (Freund's incomplete adjuvant). Two booster injections were made at 3-week intervals and antibodies were prepared by routine procedures. The titre of PAP specific antiserum determined by serial dilution was 1:3000.

2.5. Western blot analysis E. coli LE392 cells were cultured overnight at 37 °C in TB broth (12 g/L bacto-tryptone, 24g/L yeast extract, 2.3 g/L K H z P O 4 and 12.5 g/L K2HPO4). Samples of 2.0 A59 o units of cultures were centrifuged and the cells were lysed in 200 pL of Laemmli [21] loading solution at 96 °C for 5 min. Samples of 10 pL were loaded on SDS polyacrylamide (15%) gels (SDS-PAGE) according to Laemmli [21 ]. Proteins were electro-transferred to a nitrocellulose membrane and treated consecutively with PAPspecific rabbit antiserum and goat anti-rabbit IgG conjugated with alkaline phosphatase (Gibco BRL). The PAP was quantitated on the basis of a serial four-fold successive dilutions. Pure plant PAP in cell lysates of E. coli transformed with the expression plasmid pPISD (without a PAP gene) was used as a standard.

2.6. Preparation of E. coIi cell lysates E. coli LE392 cells transformed with PAP expression plasmids were cultured overnight at 37 °C in LB (Luria Bertani medium) containing 0.2% glucose to a cell density of OD59 o = 2.0. Cells were harvested by centrifugation at 12000 g for 3 min, sonicated and pelleted again. The clear lysates were used for PAP analysis. Total protein concentration was determined by colorimetric measurements using bovine serum albumin as a standard. 2.7. In vitro translation inhibition assay In vitro translation experiments were carried out in wheat germ or rabbit reticulocyte lysates (Promega). The wheat germ/reticulocyte lysates (30/aL) were preincubated for 20 rain at room temperature with the E. coli lysates (10 ng). Then, brome mosaic virus (BMV) RNA (4 lag) or chloramphenicol acetyltransferase (CAT) mRNA (300 ng) were added as templates in the presence of L-[35S]-methionine (1099 Ci/mmole, ICN), incubated for 60 min at room temperature and applied on 12.5% SDSPAGE. The 35S-labelled protein bands were visualized by autoradiography for 24 h.

2.8. Local lesion assay Partially purified E. coli lysates were added to a suspension of tobacco mosaic virus (TMV) at a final concentration of 3 ktg/mL in 0.1 M phosphate buffer, pH 7.2. The solutions containing PAP were adjusted to a final PAP concentration of 1.0 mg/mL. The mixtures were inoculated on tobacco (Nicotiana glutinosa) leaves using 12 randomly chosen half-leaves for each treatment. Local

Xu et al.

1072 Table I, Analysis of N-terminal deletion clones of PAP gene. PAP gene 5' terminal Number of colonies containing PAP gene construct clones codons removed Total Sense Reverse PAP PAPA< 100 PAPA> 100 PAPA22 PAPA49 PAPA71 PAPA89

0 <100 > 100 22 (22+0) 49 (22+27) 71 (22+49) 89 (22+67)

44 45 110 52 16 22 16

21 45 b 110b 22 4 5 2

23 0 0 30 12 17 14

Immunoreactive clones Number

Percentage~

21 0 36 4 0 0 0

100 0 98. l 18.2 0 0 0

aThe N-terminal exonuclease III unidirectional deletion mutagenesis for the PAP gene (which is not released from the plasmid; as expected no reverse orientation clones were found). Values represent the quantities of N-terminal shortened immunoreative PAP in relation to the full-length PAP (100%) expressed. bThese clones were generated by exonuclease treatment. Theoretically only one-third of those clones should be in frame.

lesions were counted and the percentage of lesion inhibition was calculated by the formula: % inhibition= 1

No. of lesion (sample + TMV)] ~ o . ~ n ( - ~ a l ~ ) x 100

3. Results 3.1. Cloning of a full-length and 5' end truncated PAP gene in expression plasmids

Cloning strategy and structure of the plasmids for expression of PAP gene in E. coli are shown in figure 1. The full length PAP gene isolated from P. americana spring leaves contained the complete sequence coding for the mature PAP protein (262 amino acids (aa)) plus the two signal peptides at both N- and C-termini (22 aa and 29 aa respectively) and a portion of a non-coding sequence as has been already published [8]. This gene was subjected to both random and block deletion mutagenesis in order to study the significance of the N-terminal amino acids on the E. coli toxicity and antiviral and in vitro translation inhibition activities of PAP. All PAP gene variants (fulllength and mutants) were cloned in a pBR322 based expression vector containing a strong constitutive promoter (Pl) and a strong SD sequence which have already been used for high level expression of other eukaryotic genes in E. coli [20, 22, 23]. When the full-length PAP gene was cloned in this vector, 44 positive colonies were isolated and almost half of them contained the PAP gene in sense orientation (table 1). After the exonuclease Ill deletion mutagenesis, a total of 155 clones were analysed. In about one-third (45) of them the first 0-100 codons were deleted and in the rest (110 clones) more than 100 codons were deleted from the PAP gene. As expected, all the clones from the two groups were in the sense orientation but only one-third of them should be in the correct frame to translate into a protein. Clones (45) with less than

100 amino acids deleted from the N-terminus (without mutation) did not produce any truncated proteins as deduced from the lack of immunoreactivity by Western blot analysis (table I). However, about one-third (36) of a total of 110 clones containing N-deletion mutants of more than 100 amino acids produced immunoreactive and truncated forms of PAP protein (table I). When the PCR derived deletions of PAP gene were cloned, 106 positive colonies were isolated and sorted into four groups: (PAPA22, PAPA49, PAPA71, PAPA89) according to the number of N-terminal codons deleted. Table I shows that most clones which were derived from the second (PAPA49), the third (PAPA71) and fourth (PAPA89) groups carried the PAP gene in reverse orientation and that the number of clones which were immunoreactive was extremely low. The following clones (producing immunoreactive PAP protein or not) were chosen for further analysis: 1) PAP (carrying a full-length PAP gene); 2) three clones from the random deletion group PAP) > 100 (PAPA 107, PAPA 123 and PAPAl90); and 3) four clones from the PCR block deletion group PAPA22 (PAPA22a-c). DNA sequence analysis of these clones showed that the primary structures of the full-length gene (PAP) as well as of the three forms of PAP gene from the PAP) > 100 deletion group were correct (no mutation). However, deletion of the signal peptide 22 aa resulted in the production of a very small number of colonies. Some of them (four out of 22) produced PAP but at lower levels, others (18 out of 22) did not produce any detectable PAP. Analysis of these clones revealed that all contained mutations (table II). Two clones which produced an immunoreactive PAP protein contained a six-nucleotide T2°SACAAA insertion (PAPA22a) and a large insertion after A 652 (PAPA22b), while the other two clones which did not produce an immunoreactive PAP protein contained one nucleotide deletion at C 15 (PAPA22c) and a large insertion (more than 170 nucleotides) at T 225 (PAPA22d)(table II).

1073

Effect of N-terminal deletions Table II. Effect of N-terminal deletions on the antiviral activity of PAR

PAP gene construcU PAP PAPA7 PAPA22a PAP22b PAPA22c PAPA22d PAPA26 PAPA44 PAPA53 PAPA76 PAPA89a PAPA89b PAPA107 PAPAl 23 PAPA190

Yield of PAP (% of total E. coli proteins)

TMV antiviral activity (% inhibition)

PAP gene structure

1.24 0 0.12 0.08 0 0 0 0 0 0 0 0 0.91 0.83 0.90

100 0 59 53 N.D. N.D. 0 0 0 0 0 0 0 1.9 1.3

No mutation Frame shift T2°SACAAAinsertion Large insertion after A 652 Deletion of C ~5 Large insertion after T2e5 14 nt insertion Stop codon Stop codon Frame shift Frame shift Frame shift A2sv deleted and C:~2J added No mutation No mutation

'~Numbers after A represent the number of N-terminal codons deleted and the letters designate the clone number which was selected for study. N.D., not determined. Partial sequencing of other mutant clones positive for the PAP gene but not producing PAP protein was also carried out. Results obtained indicated all these mutants contained early stop codons, frame shifts or base substitutions (see table II).

3.2. Expression of the full-length and mutant PAP gene in E. coli LE392 Expression vectors containing the full-length and truncated forms of PAP gene were transformed in E. coli LE392 cells and the production of PAP was visualized by Western blotting and quantitated by dot-blot analysis. The Western blotting of the proteins from the constructs producing immunoreactive PAP showed an exact correlation between size of the insert and the protein molecular mass (see .figure 2). Data presented in table 11 indicate that the yield of PAP obtained from the mutant constructs PAPA22(a, b) was much lower than that of the full-length PAP gene. However, the yield of PAP increased (but still remaining less than that of the full-length gene) when a greater portion of the 5'terminal codons was deleted (PAPA 107, PAPA 123 and PAPA 190). In comparison with the result obtained by others, the yield of full-length PAP obtained with the constitutive (P~) promoter (1.24% of the total bacterial protein) was about ten times higher than that found by others with the inducible lac promoter [12]. The yield of truncated PAP protein is generally small for all constructs with deletions less than 100 aa from the N-terminus. However, it approached the yield of fulllength PAP when the deletions of more than 100 aa was carried out (table IlL Localization of the recombinant full-length PAP in E. coil was determined by immuno-gold electron micro-

scopy. The PAP protein is accumulated in cells mostly in the form of inclusion bodies although some gold grains were also found around the periplasmic area (data not shown).

3.3. Antiviral activit3, of recombinant PAP protein Antiviral activity of the protein coded by the PAP gene constructs described above was tested by a local lesion assay using tobacco plants and tobacco mosaic virus (TMV). When the TMV samples were pre-treated with lysates of E. coli cells transformed with full-length PAP construct, a 100% reduction in number of local lesions was observed (table II). The antiviral effect of the truncated PAP protein produced by the PAP- > 100 constructs (PAPA107, PAPAl23 and PAPAl90) varied between 0% and 1.9% as compared to lull-length PAP (table 11). Although two of the clones belonging to the PAPA22 deletion series had mutations in their nucleotide sequences, their antiviral activity ranged from 53% to 59% of that of full-length PAP (ruble 1I).

3.4. Ribosome-inactivating activity of recombinant PAP proteins Ribosome inactivating activity of PAP expressed in E.

coli was studied by an in vitro translation inhibition assay using either wheat germ or rabbit reticulocyte lysates. Our data showed that amounts of full-length PAP as low as 10pg arrested the protein synthesis completely (figure 3A). However, the products of the three PAP deletion constructs, PAPAl07, PAPAl23 and PAPA190, did not arrest the protein synthesis (figure 3B).

1074

Xu et al.

2 9 kDa

1

2

3

Figure 2. Western blot analysis of total proteins from E. coli cells expressing PAP gene. E. coli cells were lysed and subjected to Western immunoblot analysis. Lane 1, total protein (100 p,g) from E. coli cells transformed with plasmid PAP containing the fulllength PAP gene in sense orientation; lane 2, PAP (0.5 ~tg) purified from pokeweed plants (without the 22 aa N-terminal signal peptide and the C-terminal extension); lane 3, total protein (200 ~tg) from E. coli cells transformed with plasmid PAPA22a without the N-terminal signal codons. A 29 kDa molecular mass marker is depicted.

4. Discussion

ments [4] and its high toxicity to bacteria is predicted based on the failure to express the PAP gene devoid of N-terminal signal peptide codons in E. coIi ceils [12]. Full-length PAP (containing both N- and C-terminal

The capacity of the mature form of PAP to inactivate prokaryotic ribosomes is proven by in vitro experi-

b

24

kDa

--

1

1

2

2

3

4

5

6

3

Figure 3, Protein synthesis inhibition assay. A. Inhibition of protein synthesis by a recombinant full-length PAP protein. Wheat germ extracts (Promega) containing BMV-RNAs were supplemented with E. coli total cell proteins (10 ng) and the inhibition of protein synthesis was measured. Lane 1, proteins from E. coli cells transformed with plasmid pP~SD (no PAP gene); lane 2, protein sample as in lane 1 supplemented with 10 pg of purified PAP from pokeweed plants; lane 3, proteins from E. coli cells transformed with full-length PAP gene (10 pg recombinant PAP in the sample). B. Inhibition of protein synthesis by a full-length PAP recombinant and N-terminal deletion mutant proteins. Rabbit reticulocyte lysates containing chloramphenicol acetyl transferase (CAT) mRNA were supplemented with E. coli total cell protein ( 10 ng) and the inhibition of protein synthesis was measured. Lane 1, no CAT mRNA; lane 2, proteins from E. coli transformed with the plasmid pP~SD; lane 3, proteins from E. coli transformed with the plasmid containing full-length PAP gene (10 pg recombinant PAP in the sample); E. coli transformed with the plasmids: PAPA 107 (lane 4), PAPA 123 (lane 5) and PAPA190 (lane 6). The position of the 24 kDa molecular mass marker of chloramphenicol acetyl transferase is shown.

Effect of N-terminal deletions peptides) resembles the A-chain of type 2 RIPs (toxic for eukaryotic cells only) which allows their expression in E. coli cells. High toxicity of the mature PAP restricts the application of gene manipulation methodology to the isolation of non-toxic mutants only. The aim of this study was to investigate how massive is the N-terminal sequence in relation to its toxicity towards E. coli cells using a new expression system. Taking into consideration that the inducible gene expression could mask the toxicity of less toxic forms of PAP we chose to express the engineered PAP gene constructs under a strong constitutive promoter. Thus, a larger amount of (even of a moderate toxicity) protein is expected to kill the cells and cell mortality can be used as a phenotype for protein toxicity. A small number of survived colonies generated from N-terminal deletions of PAP gene were shown to contain only mutated forms of the PAP gene. Sequence analysis of these mutated PAP open reading frames may be used as toxicity indicators to E. coli cells. Our cloning experiments with the PAPA22 construct confirmed the results of Chen et al. [121]. The PAPA22 gene was found in the survived colonies in either reverse orientation or as a mutant. Similar results were obtained with the block-mutated genes PAPA49, PAPA71 and PAPA89 as well as with the 5' to 3'progressive deletion mutants PAPA < 107. As a result, we failed to find any clone containing a wild-type PAP gene in sense orientation until 123 or more codons were removed from the N-terminal coding region. All sequenced clones from the above constructs were only a mutant form of PAP gene. Wild-type (no mutation) genes were found only among the clones from the PAPA > 123 series. Local lesion assay using tobacco mosaic virus and tobacco cultivar is a common method to determine the antiviral activity of potential antiviral proteins. Since most of antiviral proteins like PAP are ribosome-inactivating proteins, when the antiviral protein and the virus mixture are rubbed onto tobacco leaves, the antiviral protein will enter the wounded cells along with the virus and inactivate the ribosomes and kills the cells before the viral replication can take place. Viral infection and protein synthesis inhibition assays with mutants of PAP gene showed that some of them were still active whereas others have lost their activity. There was a reverse correlation between the yield and the antiviral activity of the mutant PAP which indicates that they might have an increased toxicity to E. coli ceils. It is striking that all of the survived mutant clones belong to the series PAPA22 and PAPA > 100. This fact together with the extremely high mutation rate of the PAP gene (see for comments below) allows to assume that even mutated forms of PAP with more than 7 but less than 123 N-terminal amino acids missing are toxic for E. coli. Toxicity decreases drastically after deleting more than 123 amino acids from the PAP N-terminus but some residual antiviral activity remains even upon removing up to 190

1075 (63% of the full size PAP) of the amino acids. In a systematic deletion study with the ricin A-chain Morris and Wool [24] have found a residual ribosome impairing activity even when 222 (83%) amino acids were deleted from the protein. Sequence comparison revealed that Tyr94, Lys145, Glu198 and Arg201 (in cDNA) are invariant in the RIP family of toxins [ 11 ]. These conserved residues have been extensively studied in ricin and models for the active site were proposed mainly on the basis of crystallographic structure [25] and mutagenesis studies [26]. In this study, our results showed that deletion up to 107 codons of the N-terminus still produced a mutated form of PAP gene implying that the truncated PAP still toxic to E. coli cells. This result indicates that the deletion of the residue Tyr94 did not abolish the toxic effect of PAP to E. coli ceils. In other words, the Tyr94 residue which is presumably considered as an integrant part of the putative active site [27] may not be necessary for the toxic effect of PAP in E. coli. Our result is also supported by that of trichosanthin where a fragment (100 amino acid residues) was removed from the N-terminus without destroying the biological activity [28]. However, further deletion of 16 codons abolished the activity. It is therefore suggested that the residues between 107 and 123 play an important role in retaining the biological activity of PAP in E. coli. Furthermore, the S-S bonds between Cys56 and 281, and Cysl07 and 128 seem also to be not required for the toxic activity of PAP because the removal of both Cys 56 and 107 respectively did not seem to affect the toxicity of PAP to E. coli cells. In this study we also have observed an extremely high mutation rate of the PAP gene. High mutation rate was also described by Dore et al. [18] but the mutagenic source was not determined. It could be assumed that the PCR conditions themselves were mutagenic. However, the results from DNA sequence analysis of PAP constructs inserted in reverse orientation into the same expression vector (no clones with altered nucleotide sequence were found) disagree with such an assumption. It seems very likely that the mutagenesis of the PAP genes is an endogenous rather than exogenous process used as a protective mechanism against the toxicity of the gene products in E. coli. The constitutive expression of the full length PAP gene led to a ten-fold increase in the protein yield in comparison with the inducible expression of the same gene under a lac promoter 1112]. This implies that the E. coli cells can tolerate much higher intracellular concentrations of this protein. Ours and other studies [12, 16] have shown that the full-length PAP inactivates both eukaryotic and prokaryotic ribosomes in vitro with the same activity as the natural (mature) PAP. Chaddock et al. [16] also described a specific product (240 nt fragment) of PAP catalysed depurination of 23S rRNA extracted from E. coli host cells expressing full-length and mutant forms of PAP

1076 gene. It is difficult to b e l i e v e that P A P m R N A m i g h t be t r a n s l a t e d w i t h o u t the c o n v e n t i o n a l p a r t i c i p a t i o n o f the e l o n g a t i o n factors. T h e s e results m i g h t b e e x p l a i n e d a s s u m i n g that e i t h e r E. c o l i r i b o s o m e s m a y be resistant or the f u l l - l e n g t h P A P is i n a c t i v e in v i v o a n d it is a c t i v a t e d after d i s r u p t i o n o f r e c o m b i n a n t cells. T h e fastest w a y o f p r o t e i n i n a c t i v a t i o n is its c o a g u l a t i o n ( p r e c i p i t a t i o n ) or e x p o r t out o f the cell. In ours a n d o t h e r studies [ 12, 16] the r e c o m b i n a n t f u l l - l e n g t h P A P was f o u n d m a i n l y in the f o r m o f i n c l u s i o n b o d i e s a t t a c h e d to the i n n e r cell m e m b r a n e a n d partly in the p e r i p l a s m i c space. W e h a v e also f o u n d s o m e P A P activity in the culture m e d i u m (data not s h o w n ) w h i c h c o u l d b e d u e to e i t h e r p r o t e i n e x c r e t i o n o r spontan e o u s l y l y s e d b a c t e r i a l cells. W e a s s u m e that this is a quite efficient w a y o f i n a c t i v a t i n g P A P since w e did n o t find a n y s p o n t a n e o u s m u t a n t o f f u l l - l e n g t h P A P g e n e a m o n g the several s e q u e n c e d g e n e s i n s e r t e d in a s e n s e o r i e n t a t i o n in the p P 1 S D vector. It s e e m s likely that this m e c h a n i s m o f i n a c t i v a t i o n is i n a p p l i c a b l e for the p r o t e i n s d e v o i d o f N - t e r m i n a l signal p e p t i d e w h i c h are i n a c t i v a t e d b y g e n e alterations.

Acknowledgments This study is supported by an NSERC grant to M.G. Abouhaidar and K601/1996 to I. Ivanov.

References Ill Irvin J.D., Uckun EM., Pokeweed antiviral protein: ribosome inactivation and therapeutical applications, Pharmacol. Ther. 55 (1992) 279-302. [21 Barbieri L., Barrelli G.M., Stirpe E, Ribosome-inactivating proteins from plants, Biochim. Biophys. Acta 1154 11993) 237-282. [3] Endo Y., Tsurugi K., The RNA N-glycosidase activity of ricin-A chain: the characteristics of the enzymatic activity of ricin A-chain with ribosomes and with rRNA, J. Biol. Chem. 263 ll988) 8735-8739. 141 Hartley M.R., Legname G., Osborn R., Chen Z., Lord J.M., Single-chain ribosome-inactivating proteins from plant depurinate Escherichia coli 23S ribosomal RNA, FEBS Lett. 290 (1991) 65-68. 15] Moazed D., Noller H.E, Binding of rRNA to the ribosomal A and P sites protecls two distinct sets of nucleotides in 16S rRNA, J. Mol. Biol. 211 11990) 135-145. 161 Ready M.R, Brown D.J., Robertus J.D., Extracellular localization of pokeweed antiviral protein, Proc. Nail. Acad. Sci. USA 83 (1986) 5053-5056. [7[ Bonness M.S., Ready M.E, lrvin J.D., Mabry T.J., Pokeweed antiviral protein inactivates pokeweed ribosomes; implication for the antiviral mechanism, Plant J. 5 11994) 173-183. 181 Lin Q, Chen Z.C., Antoniw J.E, White R.E, Isolation and characterization of a cDNA clone encoding the antiviral protein fl'om Phytolacca americana. Planl Mol. Biol. 17 ( 1991 ) 609-614. [91 Poyet J.L., Radom J., Hoeveler A., Isolation and characterization of a cDNA clone encoding the pokeweed antiviral protein I1 from Phytolacca americana and its expression in E. coli. FEBS Lett. 347 I"1994) 268-272.

X u et al. [10] Kataoka J., Habuka N.. Masuta C., Miyano M., Koiwai A., Isolation and analysis of a genomic clone encoding a pokeweed antiviral protein, Plant Mol. Biol. 20 11992) 879-886. [ 11] Monzingo A.F., Collins E.J., Ernst S.R., Irvin J.D., Robertus J.D., The 2.5 J~ structure of pokeweed antiviral protein, J. Mol. Biol. 233 11993) 705-715. 112] Chen Z.C., Antoniw J.E, Lin Q., White R.E, Expression of lx~keweed (Phytola~x'a amer&ana) antiviral protein cDNA in Escherichia coli and its antivira] activity, Physiol. Mol. Plant Pathol. 42 11993) 237-247. [131 O'Hare M., Robertus L.M., Thorpe P.E., Watson G.J., Prior B., Lord J.M., Expression of ricin A chain in Escherichia coli, FEBS Lett. 216 (1987) 73-78. [ 14] Piatak M., Lane J.A., Laird W, Bjorn M.J., Wang A., Williams M., Expression of soluble and fully functional ricin A-chain in E. coli is temperature sensitive, J. Biol. Chem. 263 (1988) 4837--4843. [15J Prestle J., Hornung E., Schonfelder M., Mundry W.K., Mecbanism and site of action of a ribosome-inactivating protein type 1 from Dianthus barbatus which inactivates Escherichia coli ribosomes, FEBS Lett. 297 11992) 250-252. [ 16] Chaddock J.A., Lord J.M., Hanley M.R., Roberts L.M., Pokeweed antiviral protein (PAP) mutations which permit E. coli growth do not eliminate catalytic activity towards prokaryotic ribosomes, Nucleic Acids Res. 22 11994) 153f~1540. [17] Chaddock J.A., Monzingo A.E. Robertus J.D., Lord J.M., Roberts L.M., Major structural differences between pokeweed antiviral protein and ricin a-chain do not account for this differing ribosome specificity, Eur. J. Biochem. 235 (11996) 159-166. [18] Dore J.M., Gras E, Depien'e E, Wijdenes J., Mutations dissociating the inhibitory activity of the pokeweed antiviral protein on eukaryote translation and Escherichia coli growth, Nucleic Acids Res. 21 11993) 4200-4205. [19] Ausubel EM., Brent R., Kingston R.E., Moore D.D., Seidman J.G., Smith J.A., Struhl K., Current Protocols in Molecular Biology I 11989) 431-433. [201 Rommens J., Macknight D., Pomeroy-Cloney L., Jay E., Gene expression: chemical synthesis and molecular cloning of a bacteriophage T5 (P25) early promoter, Nucleic Acids Res. I1 (1983) 5921-5940. [21] Laemmli U.K., Cleavage o1! structural proteins during the assembly of the head of bacteriophage T4, Nature 227 (1970) 68111-685. [22] Jay E., Rommens J., Pomeroy-Cloney L., Macknight D., LutzeWallace C., Wishart E, Harrison D., Liu W.Y., Asundi V., Dawood M., Jay G., High-level expression of a chemic~dly synthesized gene for human interferon-gamma using a prokaryotic expression vector, Proc. Natl. Acad. Sci. USA 81 11984) 2291~2294. 123] Gigova L., Wishart E, Usbeva A., lvanova M., Bardarov S., Jay E., lvanov I., Expression of repetitive human calcitonin genes, in Escherichia coli, Biotechnol. Appl. Biochem. 11 (1989) I-I 2. [24] Morris K.N., Wool I.G., Determination by systematic deletion of the amino acids essential for catalysis by ricin A-chain, Proc. Nail. Acad. Sci. USA 89 11992) 4869-4873. 125] Robertus J,, The structure and action of ricin, a cytotoxic N-glycosidase, Semin. Cell Biol. 2 (1991) 23-30. 126] Frankel A., Welsh P., Richardson J., Robertus J.D., The role of arginine 180 and glutamic acid 177 of ricin toxin A chain in the enzymatic inactivation of ribosomes, Mol. Cell Biol. 10 tl990) 6257-6263. [271 Ago A., Kataoka J., Tsuge H., Habuka N., lnagaki E., Norma M., Miyano M., X-ray structure of a pokeweed antiviral protein, coded by a new genomic clone, at a 0.23 nm resolution. A model structure provides a suitable electrostatic field for substrate binding, Eur. J. Biochem. 225 (1994) 369-374. [28] Ke Y.B., Chen J.K., Nie H . L He X.H., Ke X.Y., Wang Y.H., Structure-function relationship of trichosanthin, Life Sci. 60 (1997) 465-472.