The structural integrity exerted by N-terminal pyroglutamate is crucial for the cytotoxicity of frog ribonuclease from Rana pipiens

Abstract

Onconase, a cytotoxic ribonuclease from Rana pipiens, possesses pyroglutamate (Pyr) at the N-terminus and has a substrate preference for uridine–guanine (UG). To identify residues responsible for onconase’s cytotoxicity, we cloned the rpr gene from genomic DNA and expressed it in Escherichia coli BL21(DE3). The recombinant onconase with Met at the N-terminus had reduced thermostability, catalytic activity and antigenicity. Therefore, we developed two methods to produce onconase without Met. One relied on the endogeneous E.coli methionine aminopeptidase and the other relied on the cleavage of a pelB signal peptide. The Pyr1 substitutional variants maintained similar secondary structures to wild-type onconase, but with less thermostability and specific catalytic activity for the innate substrate UG. However, the non-specific catalytic activity for total RNAs varied depending on the relaxation of base specificity. Pyr1 promoted the structural integrity by forming a hydrogen bond network through Lys9 in α1 and Val96 in β6, and participated in catalytic activity by hydrogen bonds to Lys9 and P₁ catalytic phosphate. Residues Thr35 and Asp67 determined B₁ base specificity, and Glu91 determined B₂ base specificity. The cytotoxicity of onconase is largely determined by structural integrity and specific catalytic activity for UG through Pyr1, rather than non-specific activity for total RNAs.

INTRODUCTION

Ribonucleases are ubiquitous in nature and have been implicated in cellular RNA metabolism and the regulation of gene expression (1,2). However, several proteins in the ribonuclease family have other biological functions in addition to the ribonucleolytic activity. For example, human RNase 5 (angiogenin) induces blood vessel formation (3); human RNase 2 [human eosinophil-derived neurotoxin (EDN)] and RNase 3 [human eosinophil cationic protein (ECP)] possess neurotoxicity and antiviral activity (4,5) and bovine seminal ribonuclease exhibits both antitumor and immunosuppressive properties (6). In addition, a group of frog ribonucleases from Rana pipiens, Rana japonica and Rana catesbeiana exert cytotoxic activity towards tumor cells. These frog ribonucleases possess distinct characteristics which are not found in most mammalian ribonucleases: they have pyroglutamate (Pyr) at the N-terminus, prefer pyrimidine–guanine substrates, demonstrate high thermostability and are not neutralized by the ribonuclease inhibitor from human placenta (7–12). One of frog ribonucleases, onconase from R.pipiens, is currently being evaluated in combination with tamoxifen or doxorubicin in human phase III clinical trials for tumor therapy (7). To identify residues or domains responsible for the novel properties of frog ribonuclease, we cloned the onconase gene rpr from genomic DNA and mutated possible residues on the basis of amino acid sequences and structures of ribonucleases. We found that the Pyr1 is involved in thermostability and catalytic activity of onconase and that Asp67 and Glu91 residues participated in the B₁ and B₂ base specificity, respectively. Our results demonstrate that the cytotoxicity of onconase is primarily determined by structural integrity and specific catalytic activity for the innate substrate uridine–guanine (UG), but not by non-specific catalytic activity for total RNAs.

MATERIALS AND METHODS

Cloning of onconase gene from genomic DNA

Genomic DNA was extracted from the liver of R.pipiens (Nasco Biologicals Co., Fort Atkinson, WI) by DNA zol reagents (Molecular Research Center Inc., Cincinnati, OH) according to the manufacturer’s instructions. The onconase gene rpr was amplified by PCR using oligonucleotide primers rap-1 (5′-ATAAAGGCCTGATCACGACTTCCAG-3′, nucleotides 71–95 of 5′ UTR of rap LR1, AF165133) and α-rap-2 (5′-GCTGCTTATCACATCCCTGTTGTC-3′, nucleotides 514–491 of 3′ UTR of rapLR1, AF165133) as 5′ and 3′ primer, respectively (13,14).

Construction of expression vector

The amplified onconase gene containing an NdeI (CATATG) and a BamHI (GGATCC) recognition sequence at its 5′ and 3′ ends, respectively, was inserted into pET22b through the NdeI and BamHI sites (Novagen). Therefore, a Met is added to the N-terminus of onconase for translation initiation in Escherichia coli. Alternatively, the onconase gene is preceded by a NcoI-containing sequence CCATGGCT, which encodes for Met–Ala, and fused with a modified pelB signal peptide sequence through an NcoI site at its 3′ end, before inserting into pET11d through NdeI and BamHI sites (13). The signal sequence is presumably removed in vivo before Gln-ended mature onconase is secreted into the culture medium. Mutations were made by site-directed mutagenesis.

Expression and purification of recombinant ribonucleases

The recombinant onconase in pET22b or pET11d was expressed in E.coli BL21(DE3) overnight at 37°C. Inclusion bodies were collected, denatured in 6 M guanidine hydrochloride and then refolded in a mixture of reduced and oxidized glutathione (15). The refolded proteins were purified to electrophoretic homogeneity by carboxymethyl cellulose (CM52, Whatman) column chromatography. The soluble onconase secreted into the culture media was concentrated by polyethylene glycol and purified to electrophoretic homogeneity by phosphocellulose (P-11, Whatman) and carboxymethyl cellulose (CM52, Whatman) column chromatography (13). The native onconase was purified from the oocytes of R.pipiens as for RC-RNase (9).

Analysis of recombinant onconase and its mutants

Proteins were separated by 13.3% SDS–PAGE and stained by Coomassie brilliant blue R (16). For western blotting analysis, the immobilized proteins were probed with antibodies raised against onconase isolated from culture medium (17). The production of antibodies from rabbits was prepared by intra-spleen immunization (18). Onconase proteins with N-terminal block were incubated with pfu Pyr aminopeptidase (TaKaRa Biomedicals, Japan) at 70°C for 4 h before sequence analysis (13). Alternatively, proteins from inclusion bodies were separated by 13.3% SDS–PAGE and contact-transferred to PVDF membrane (Problot, Applied Biosystems) before sequence analysis (13). Automatic Edman degradation was performed on a Gas/Liquid phase Model 477A sequencer (Applied Biosystems) equipped with an on-line Model 492 phenyl thiohydantoin-amino acid analyzer (13).

Circular dichroism and mass spectrophotometric analysis

Circular dichroism (CD) experiments were carried out using an Aviv CD 202 spectrophotometer (Aviv, Lakewood, NJ) calibrated with (+)-10-camphorsulfonic acid at 25°C. In general, a 2 mm path-length cuvette with 10–20 µM ribonuclease in 20 mM sodium phosphate, pH 7.0, was used for CD experiments, and all protein solutions were made up to 1 ml. The spectra were recorded from 190 to 260 nm. After background subtraction and smoothing, all CD signals (millidegree) were converted into mean residue ellipticity (deg cm² dmol^–1). Equilibrium thermo-denaturing experiments were performed by measuring changes of molar ellipticity at 200 nm. Data were collected as a function of temperature at a scan rate of 2°C/min over the range 20–95°C in 20 mM sodium phosphate buffer, pH 7.0. The variation was monitored at 200 nm after 3 min equilibration at each point with a temperature controller. The mass spectrophotometric (MS) analyses of C4 column-desalted ribonucleases in 0.3% formic acid and 50% acetonitrile (v/v) were performed on a Micromass Q-TOF Ultima™ API spectrophotometer (Micromass, Wythenshawe, UK) equipped with an orthogonal electrospray source operated in the positive ion mode.

Ribonuclease activity assay

The ribonuclease activity of column eluates was determined by their abilities to cleave dinucleotide UG in 50 mM sodium acetate, pH 6.0, 50 mM sodium chloride and 0.1 µg/µl bovine serum albumin. The digestion products were separated by thin layer chromatography (PEI-cellulose, F Merck, Darmstadt, Germany) in 0.25 M lithium chloride–1.0 M acetic acid, and visualized under UV illumination at 254 nm (16). Ribo nuclease activities were also analyzed by zymogram assay on RNA-casting PAGE (19). Briefly, after electrophoresis the gel was washed twice with 25% isopropyl alcohol in 10 mM Tris–HCl, pH 7.5, to remove SDS for protein renaturation. The activity was visible after incubating the gel at room temperature for 30 min in 10 mM Tris–HCl, pH 7.5, then in 0.2% Toluidine blue O for 10 min. The non-specific catalytic activity of ribonuclease was determined by the release of acid-soluble ribonucleotides from the RNA substrate after ribonuclease digestion (20). Briefly, an excess amount of yeast total RNA (120 µg) was incubated with purified onconase or its variant in 50 µl of 50 mM sodium acetate, pH 6.0, 50 mM sodium chloride and 0.1 µg/µl bovine serum albumin, at 37°C for 15 min, and the reaction was terminated by the addition of ice-cold stop solution (7% perchloric acid, 0.1% uranyl acetate, 200 µl). The reaction mixtures remained on ice for 30 min before centrifugation (12 000 g, 4°C for 20 min). There was a linear relationship between enzyme activity and optical density at supernatent absorbance (at 260 nm) values between 0.1 and 0.3. One unit of ribonuclease activity is defined as the amount of enzyme producing one A₂₆₀ acid-soluble material at 37°C for 15 min.

Base specificity of ribonucleases

The specific cleavage sites of onconase on 5′ ³²P-labeled synthetic 18mer RNA, AAGGUUAUCCGCACUGAA, in 50 mM sodium acetate, pH 6.0, 50 mM sodium chloride and 0.1 µg/µl bovine serum albumin were determined by denaturing gel electrophoresis and autoradiography (21). The k_cat/K_m values of ribonuclease towards dinucleotides, e.g. CG and UG, were measured by HPLC separation and quantification (20,22). Briefly, the amount of onconase variant, which digests 1 µg of dinucleotide in a 20 µl reaction mixture containing 100 mM MES, pH 6.0, 50 mM NaCl and 0.1 µg/µl bovine serum albumin at 37°C in 10 min, was taken for analysis. Excess amounts of dinucleotides ranging from 2 to 16 µg were used as substrates for the reaction. Digested nucleotides were separated by reverse-phase HPLC using 1–5% acetonitrile (depending on products) in 0.1% trifluoroacetic acid on a Vydac C18 column with a Waters automated gradient controller. The amounts of substrate and product (guanosine) were determined using Waters 991 Integrator. The k_cat/K_m values were calculated by Lineweaver–Burk plot. Values were obtained from the average of at least two independent experiments.

Assay of cytotoxicity by ATP Lite-M measurement

HeLa cells (5 × 10³) were cultured in a 96-well plate, each well containing 100 µl of Dulbecco’s modified Eagle’s medium with 10% fetal calf serum, and treated with 2 µM onconase or variants for 72 h. Fifty microliters of lysis buffer was added to each well and incubated for 2 min before 50 µl of substrate solution was added. The luminescence was measured by a Top Count Microplate Scintillation Counter in a dark adapt plate (Packard A Canberra Company) according to the manufacturer’s instructions (ATP Lite-M assay system; Packard BioScience Company, the Netherlands) (13).

RESULTS

Cloning of onconase gene from genomic DNA

The onconase gene rpr was cloned from the genomic DNA of R.pipiens by PCR. Its sequence is deposited in the GenBank database under the accession number AF332139. The native rpr gene encodes a mature protein of 104 residues preceded by a putative signal peptide of 23 residues which is identical to the signal peptide of rapLR1 from R.pipiens (14). The deduced amino acid sequence of the rpr gene is identical to the corresponding sequence of native onconase isolated from oocytes, but is different from that of rapLR1 in four amino acid residues: I11L, D20N, K85T and S103H (14).

Expression and purification of onconase

The recombinant onconase, rMOnc, collected from inclusion bodies, was denatured, refolded and purified to electrophoretic homogeneity with a yield of 40–50 mg per liter of culture (Figs 1 and 2). Similar results were obtained for the Q1A and Q1S mutants, designated as rMOnc-Q1A and rMOnc-Q1S, respectively (Table 1). When a modified pelB signal peptide was introduced in front of the onconase, the recombinant protein was expressed predominantly in inclusion bodies with trace amounts in the soluble fractions (Fig. 1A and B). The proteins, rOnc and rOnc-Q1E, in inclusion bodies, were processed as described above with a yield of 40–50 mg per liter of culture (Table 1). In contrast, the ribonucleolytic activity of these secreted variants, sOnc, sOnc-Q1E and sMOnc-Q1S, was detected directly in the culture medium (∼70%) without refolding and purified to electrophoretic homogeneity with a yield of 2–5 mg per liter of culture (Figs 1 and 2).

Figure 1.

Analysis of onconase expression in E.coli. Escherichia coli cultures and equivalent amounts of supernatant and pellets of cell lysates were taken for analysis. (A) SDS–PAGE analysis of protein components. Twenty microliters of onconase and Met-tagged onconase-transformed E.coli culture were taken for 13.3% SDS–PAGE and Coomassie blue staining. (B) Western blotting analysis. Two microliters of the E.coli culture was taken for 13.3% SDS–PAGE and probed with antibodies raised against sOnc. (C) Zymogram analysis. Two microliters of E.coli culture was taken for RNA-casting 13.3% SDS–PAGE and stained by Toluidine blue O. C, purified onconase as a control [2 µg in (A), 0.1 µg in (B), 0.5 µg in (C)]; S and P, the supernatant and pellets of cell lysate, respectively; M, the E.coli culture medium.

Figure 2.

Analysis of onconase variants. (A) SDS–PAGE. Two micrograms of purified onconase variants were taken for 13.3% SDS–PAGE and Coomassie blue staining. (B) Western blotting. (C) Zymogram. Onconase variants (0.2 µg) were employed for (B) and (C). The abbreviation of each variant is explained in the text.

Table 1. Nomenclature of onconase and N-terminal variants expressed in E.coli BL21(DE3).

Ribonuclease	Expression vector	pelB signal peptide	Form	N-terminal residues
rMOnc	pET22ba	–	Insoluble	Met
rMOnc-Q1A	pET22b	–	Insoluble	Ala
rMOnc-Q1S	pET22b	–	Insoluble	Ser
rOnc	pET11db	+	Insoluble	Gln > Pyr
rOnc-Q1E	pET11d	+	Insoluble	Glu
sOnc	pET11d	+	Solublec	Pyr
sOnc-Q1E	pET11d	+	Soluble	Glu > Pyr
sMOnc-Q1S	pET11d	+	Soluble	Met

^aThe onconase gene was inserted into pET22b through NdeI and BamHI recognition sites.

^bA pelB signal peptide sequence was added, preceding the onconase gene, by a NcoI site, and was then inserted into pET11d through NdeI and BamHI recognition sites.

^cThe soluble onconase was collected from the culture medium.

Characterization of recombinant onconases

Determination of N-terminal residue. The N-terminal residue of rMOnc from inclusion bodies was Met, while that of rMOnc-Q1A and rMOnc-Q1S was Ala and Ser, respectively. The Met preceding Ala or Ser may be removed by the E.coli methionine aminopeptidase, which is unable to remove the initiator Met preceding the bulkier residue, i.e. Gln. However, the Met preceding the Ser of onconase secreted into the culture medium through the pelB signal peptide was not removed, as detected by Edman degradation and MS analyses (Fig. 3). The N-termini of pelB-derived rOnc and rOnc-Q1E in inclusion bodies were Gln and Glu, respectively, as detected by Edman degradation. However, the N-terminus of the above-mentioned rOnc was a mixture of Gln and Pyr because two forms were identified from MS analysis: one (11 819 Da) represents matured Pyr-initiated onconase and the other (11 836 Da) represents its precursor with Gln at the N-terminus (Fig. 3). The conversion of N-terminal Gln to Pyr was complete within 2 weeks at 4°C in 20 mM sodium phosphate, pH 7.0 (Fig. 3). The N-terminus of sOnc secreted the into culture medium was presumably Pyr, identical to native onconase, because no signal was obtained by Edman degradation and only one form (11 819 Da) was observed from MS analysis (Fig. 3). After pfu Pyr aminopeptidase treatment, however, the first five residues (Asp–Trp–Ala–Leu–Thr) are identical to those deduced from the DNA sequence. The N-terminal residue of pelB-derived rOnc-Q1E from inclusion bodies was Glu only (Table 1), but that of sOnc-Q1E from the culture medium was a mixture of Glu and Pyr based on MS analysis (two peaks, 11 819 and 11 837 Da) (Fig. 3). These results reveal that the N-terminal Gln of onconase from the E.coli culture media is totally converted into Pyr, whereas the N-terminal Glu of onconase from the culture medium is partially converted to Pyr, probably by some E.coli enzymes under the same culture conditions. The molecular masses of all the above-mentioned onconase variants are in good agreement with the values calculated from the DNA sequence by the ExPASY server (23) (Fig. 3).

Figure 3.

Mass spectrum analysis of onconase variants. The onconase variant was desalted by a C4 column in 0.3% formic acid and 50% acetonitrile (v/v), and analyzed in a Micromass Q-TOF Ultima™ API spectrometer equipped with an orthogonal electrospray source. The abbreviation of each variant is explained in the text. rOnc′, rOnc kept in 20 mM sodium phosphate, pH 7.0, at 4°C for 2 weeks.

Structural integrity and antigenicity. To see the effects of the N-terminus on the structural integrity, the CD spectra of these rOnc variants were recorded between 190 and 260 nm at pH 7.0 (Fig. 4A). The results indicate that at this resolution the secondary structures of onconase variants are not significantly perturbed by these N-terminal modifications. As far as thermostability is concerned, sOnc is the most stable (T_m = 92°C), followed by rOnc-Q1E (T_m = 89°C), rMOnc-Q1S (T_m = 89°C), rMOnc-Q1A (T_m = 89°C) and rMOnc (T_m = 85°C) (Fig. 4B). The antigenicity of onconase was markedly reduced by the addition of Met, but was slightly altered by Q1A or Q1E mutation at the N-terminus (Fig. 2B). These results suggest that the Pyr1 plays important roles in the structural integrity and the immunogenic epitope of onconase, and is disrupted by Met addition or residue substitution at the N-terminus.

Figure 4.

CD spectra and thermostability of onconase variants. (A) Spectrum analysis. CD spectra of 20 µM onconase variants in 20 mM sodium phosphate, pH 7.0, were recorded from 190 to 260 nm and presented as mean residue ellipticity and expressed in deg cm² dmol^–1. (B) Thermostability. The equilibrium CD titration experiment was measured at 200 nm as a function of temperature. The abbreviation of each variant is explained in the text.

Catalytic activity and base specificity. Although members of the RNase A superfamily possess a similar global structure that contains three α-helices and two anti-parallel β-strands, their substrate specificities are different. Onconase and RC-RNase 2 prefer UG, whereas RC-RNase 4 prefers CG. RC-RNase and RC-RNaseL1 prefer UG and CG, but RNase A prefers both UA and CA (7,10,24,25). To determine residues involved in the base specificity and their possible roles in cytotoxicity, mutations were made based on the base specificities and structures of the ribonucleases in the RNase A superfamily. There are two substrate-binding sites, the B₁ and B₂ subsites, and one catalytic center, the P₁ subsite_, in these ribonucleases (25). The K_m values of secretory onconase for CG and UG were similar, 47.3 and 50.1 µM, respectively, while their k_cat values differed substantially, 24.7 and 492.9 min^–1, respectively. These results show that onconase has similar binding affinity for CG and UG, but has a preferential catalytic activity for UG only. We mutated the B₁ base-responsible residues Lys33, Thr35 and Asp67, which are equivalent to Val37, Thr39 and Thr70 of RC-RNase, or to Val43, Thr45 and Asp83 of RNase A, respectively. The Val37 of RC-RNase has a hydrophobic interaction with the B₁ base (26). The Onc-K33A and Onc-K33E were less active than sOnc on zymogram (Fig. 5B), but their B₁ base specificities remained unchanged (Fig. 5C). The non-specific catalytic activity of sOnc-T35A for total RNAs and the k_cat/K_m values for both CG and UG markedly decreased, 12.2, 4.5 and 0.3%, respectively, compared with that of wild-type onconase (Table 2 and Fig. 5B). However, the absolute k_cat/K_m values for both CG and UG are similar, 2.2 × 10⁴ and 2.5 × 10⁴ M^–1 min^–1, respectively. The result shows that the B₁ base preference of sOnc-T35A was relaxed from uracil (U) to both U and cytidine (C), while that of sOnc-T35S was still U as seen in sOnc (Table 2 and Fig. 5C). These results indicate that the hydroxyl group of Thr35 is crucial for B₁ base preference and catalytic activity. In contrast, the k_cat/K_m values of D67A for UG markedly decreased, 3.1% left, but its k_cat/K_m values for CG increased 6-fold and its non-specific catalytic activity for total RNA increased 8-fold compared with that of wild-type onconase. However, its absolute k_cat/K_m value for CG (315 × 10⁴ M^–1 min^–1) was 10-fold higher than that for UG (30.7 × 104 M^–1 min^–1). The change of B₁ base preference of sOnc-D67A from U to C was also observed by RNA cleavage analysis on an 18mer RNA (Fig. 5C). The relaxation of B₁ base specificity from U to cytosine (C) and U was found in sOnc-D67T by similar analyses (Table 2 and Fig. 5C). These results indicate that residue 67 also participates in B₁ base recognition: Asp prefers U, Ala prefers C and Thr prefers both C and U.

Figure 5.

Analysis of residues involved in catalytic activity and base specificity. (A) SDS–PAGE analysis. One microgram of purified ribonucleases was separated by 13.3% SDS–PAGE and stained by Coomassie blue. (B) Zymogram. A purified onconase variant (0.2 µg) was subjected to RNA-casting SDS–PAGE and Toluidine blue O staining. (C) Base specificity of ribonuclease. 5′ End-labeled oligoribonucleotide was partially (approximately one fourth) digested with ribonuclease as indicated, in 50 mM sodium acetate, pH 6.0, 50 mM sodium chloride and 0.1 µg/µl bovine serum albumin, at 30°C for 10 min. These products were separated by 8 M urea–15% PAGE and visualized by autoradiography. The abbreviation of each ribonuclease is explained in the text. The cleavage bonds are shown in the right margin.

Table 2. Catalytic activities and cytotoxicities of onconase and its mutants.

Ribonuclease variants	Non-specific catalytic activity (Unit/µg; %)a	CG kcat (min–1)	Km (µM)	kcat/Km (M–1 min–1 × 104; %)	UG kcat (min–1)	Km (µM)	kcat/Km (M–1 min–1 × 104; %)	IC50 (µM)
Onc(native)	0.61 ± 0.04 (96.8)	24.3 ± 2.2	45.2 ± 1.5	53.7 ± 1.5 (102.9)	411.1 ± 19.2	46.0 ± 4.0	893.7 ± 60.3 (90.8)	0.2
N-terminus
sOnc(wt)	0.63 ± 0.09 (100)	24.7 ± 0.5	47.3 ± 1.7	52.2 ± 2.7 (100.0)	492.9 ± 17.7	50.1 ± 2.4	983.8 ± 14.3 (100.00)	0.2
rOnc	0.16 ± 0.009 (26.4)	12.4 ± 3.7	62.8 ± 20.2	19.7 ± 0.5 (37.8)	379.6 ± 22.3	64.9 ± 4.2	584.9 ± 3.4 (59.5)	1.5
rMOnc-Q1A	0.36 ± 0.007(57.1)	2.1 ± 0.2	11.3 ± 0.7	18.4 ± 1.9 (35.3)	7.4 ± 0.9	19.4 ± 1.9	38.1 ± 3.2 (3.9)	>8
rMOnc-Q1S	0.13 ± 0.007 (20.6)	4.7 ± 0.3	27.9 ± 2.5	16.8 ± 0.5 (32.2)	7.9 ± 0.4	17.1 ± 1.2	46.2 ± 1.1 (4.7)	4
sOnc-Q1E	0.09 ± 0.009 (14.6)	8.0 ± 1.1	100.2 ± 12.9	8.0 ± 0.7 (15.3)	171.4 ± 20.5	85.1 ± 11.0	201.4 ± 2.6 (20.5)	–
rOnc-Q1E	0.008 ± 0.0002 (0.9)	NDb	ND	ND	3.9 ± 0.6	38.4 ± 6.6	10.2 ± 0.2 (1.0)	4
RMOnc	0.003 ± 0.0003(0.5)	0.1 ± 0.01	22.1 ± 2.7	0.5 ± 0.02 (0.9)	4.6 ± 1.2	100.2 ± 29.4	4.6 ± 0.1 (0.5)	>>8
B1 site
sOnc-T35A	0.07 ± 0.016 (12.2)	0.6 ± 0.01	25.4 ± 1.4	2.2 ± 0.1 (4.5)	1.4 ± 0.2	56.8 ± 9.8	2.5 ± 0.1 (0.3)	>8
sOnc-T35S	0.71 ± 0.122 (112.3)	1.4 ± 0.1	90.1 ± 2.7	1.6 ± 0.01 (3.0)	509.2 ± 77.3	178.1 ± 13.6	285.9 ± 21.7 (29.1)	–
sOnc-D67A	5.10 ± 1.13(804.0)	321.1 ± 69.5	101.9 ± 21.6	315.1 ± 1.4 (603.7)	26.5 ± 0.3	86.2 ± 1.5	30.7 ± 0.8 (3.1)	4
sOnc-D67T	1.61 ± 0.235 (254.0)	87.8 ± 15.0	113.3 ± 21.0	77.5 ± 1.3 (148.5)	64.2 ± 2.7	77.6 ± 3.6	82.7 ± 7.6 (8.4)	–
sOnc-D67E	0.95 ± 0.244 (150.7)	152.9 ± 96.6	103.0 ± 68.4	148.4 ± 8.0 (284.4)	36.1 ± 4.0	30.3 ± 2.6	119.1 ± 8.5 (12.1)	–
B2 site
sOnc-T89A	0.45 ± 0.111 (71.5)	–c	–	–	–	–	–	–
sOnc-T89K	3.12 ± 0.675 (492.8)	3.0 ± 0.4	113.4 ± 16.5	2.6 ± 0.1 (5.1)	161.5 ± 40.5	81.1 ± 20.5	199.1 ± 0.5 (20.2)	–
sOnc-T89R	2.04 ± 0.164 (321.9)	3.9 ± 0.3	69.3 ± 6.1	5.6 ± 0.04 (10.8)	166.5 ± 41.9	86.9 ± 22.2	191.6 ± 12.9 (19.5)	–
sOnc-E91A	0.43 ± 0.056 (68.5)	0.8 ± 0.1	51.7 ± 14.1	1.5 ± 0.2 (3.0)	40.5 ± 0.1	139.8 ± 22.1	29.0 ± 0.6 (2.9)	7

^aThe value in parenthesis represents the relative activity compared with that of sOnc(wt) and is expressed as a percentage.

^bND, not detectable.

^cNot done.

With regard to the B₂ base specificity, it is known that Asn71 and Glu111 of RNase A are involved in B₂ base recognition. The RNase A Glu111-equivalent residue in onconase is Glu91, but there is no Asn71-equivalent residue. The k_cat/K_m values of sOnc-E91A for CG (3.0% left) and UG (2.9% left) markedly decreased, but its non-specific catalytic activity only decreased slightly (68.5% left) (Table 2). This suggests that some bases other than guanine (G) in the B₂ site may be recognized by sOnc-E91A. In fact, the B₂ base specificity of sOnc-E91A was relaxed from G to adenine (A), G and U because the CA, UA and UU bonds on the 18mer oligoribonucleotide are cleaved (Fig. 5C). The Lys95 and Glu97 residues are involved in the B₂ base specificity of RC-RNase. Their equivalents in onconase are Thr89 and Glu91, respectively (26). The k_cat/K_m values of sOnc-T89K for both CG and UG reduced (5.1 and 20.2% left, respectively), but its non-specific catalytic activity increased up to 5-fold. Thus, we suspect that the enhancement of catalytic activity in sOnc-T89K may be due to the relaxation of B₂ base specificity, similar to that of sOnc-E91A.

With regard to the N-terminal variants, the non-specific catalytic activity of sOnc was the highest (100%), then nOnc (96.8%), rMOnc-Q1A (57.1%), rOnc (26.4%), rMOnc-Q1S (20.6%), sOnc-Q1E (14.6%), rOnc-Q1E (0.9%), sMOnc-Q1S (0.8%) and rMOnc (0.5%) (Table 2). The catalytic activity of sQ1E-Onc isolated from culture media (14.6%) is higher than that of rOnc-Q1E refolded from inclusion bodies (0.9%) because the N-terminal Glu is partially converted into Pyr during or before secretion into the culture medium (Fig. 3). The difference in catalytic activity among these N-terminal variants may be due to the relaxation of base specificity in rMOnc-Q1A and rMOnc-Q1S, but not in rOnc-Q1E, similar to the N-terminal variants of RC-RNase (26). The K_m values of rMOnc-Q1A for CG (11.3 µM) and UG (19.4 µM) decreased as compared with that of wild-type onconase (47.3 and 50.1 µM, respectively). Similarly, the values of rMOnc-Q1S for CG (27.9 µM) and UG (17.1 µM) also decreased. These results show that the replacement of N-terminal Pyr by Ala or Ser may increase its affinity for CG and UG through B₂ G, but does not enhance its catalytic activity. The recombinant onconase, rOnc, refolded from inclusion bodies had nearly the same base specificity as that of sOnc, UG > CG, while Met-tagged variants, sMOnc-Q1S and rMOnc, cleaved not only CG and UG but also CA, UA and UU bonds on the cleavage sites on 18mer RNA (Fig. 5C). These results indicate that the B₂ base specificity of onconase is relaxed from G to A, G and U by Met addition or some residue substitutions at the N-terminus.

Cytotoxicities towards tumor cells. We have determined the cytotoxicity of onconase on five cell lines: human cervical carcinoma HeLa S-3 cells (IC₅₀ = 0.2 µM), human leukemia K562 cells (IC₅₀ = 1 µM), bovine aortic endothelial cells BAEC (IC₅₀ = 1 µM), human breast cancer MCF-7 cells (IC₅₀ > 8 µM) and normal human fibroblast cells HFW (IC₅₀ >> 8 µM). We then examined the cytotoxicity of N-terminal variants on HeLa S-3 cells. The cytotoxicities of rOnc-Q1E (IC₅₀ = 4 µM) and rMOnc-Q1S (IC₅₀ = 4 µM) were slightly less than that of rOnc (IC₅₀ = 1.5 µM), while that of rMOnc-Q1A (IC₅₀ > 8 µM) was much less than that of rOnc, although its non-specific catalytic activity (57.1%) is higher than that of rOnc (26.4%). The cytotoxicity was almost obliterated by the Met addition at the N-terminus, e.g. sMOnc-Q1S (IC₅₀ >> 8 µM) and rMOnc (IC₅₀ >> 8 µM) (Fig. 6A and Table 2). These results show that the Pyr1 of onconase plays a crucial role in the cytotoxicity as well as the catalytic activity; part of the cytotoxicity was retained if Pyr1 was substituted with Ser or Glu, although their specific catalytic activities for UG markedly decreased (4.7 and 1.0% left, respectively).

Figure 6.

Analysis of the cytotoxicity of onconase variants on HeLa S-3 cells. HeLa S-3 cells (3 × 10³) were seeded on a 96-well plate overnight and incubated with onconase variants for 72 h (six wells per treatment). The survival rates of onconase-treated cells were determined by ATP Lite-M and expressed as a percentage. (A) N-terminal variants of onconase. (B) B₁ and B₂ base variants of onconase.

The effects of B₁ and B₂ base specificity on cytotoxicity of onconases were also examined. The cytotoxicities of sOnc-T35A (IC₅₀ > 8 µM) and sOnc-E91A (IC₅₀ = 7 µM) decreased with the loss of non-specific catalytic activity for total RNAs (12.2 and 68.5% left, respectively) and specific activity for the innate substrate UG (0.3 and 2.9% left, respectively). However, the cytotoxicity of sOnc-D67A (IC₅₀ = 4 µM) decreased with the reduction of specific catalytic activity for UG (3.1% left), although it had an 8-fold increase in non-specific catalytic activity (Table 2 and Fig. 6B). These results reveal that the catalytic activity for the innate substrate UG is required for the cytotoxicity of onconase, but non-specific catalytic activity towards total RNAs is not sufficient to enhance its cytotoxicity.

DISCUSSION

Ribonucleases possessing cytotoxicity towards tumor cells are predominantly found in the oocytes of frog, which have Pyr at the N-terminus, prefer pyrimidine–guanine RNA substrates, demonstrate high thermostability and are not neutralized by the ribonuclease inhibitor from human placenta (7–12). We are interested in the roles of N-terminal Pyr on the frog’s cytotoxicity. Based on the computation of 27 000 curated (NP) and putative (XP) protein sequence entries from the NCBI RefSeq database using the SignalP program for the prediction of signal peptides and cleavage sites (27,28), it is known that 12.3% of proteins with signal peptides are initiated with Gln at the N-terminus after the signal peptide is removed. The resultant N-terminal Gln is converted to Pyr by deamination, which is usually essential for the biological functions of many proteins and hormones (29,30). With respect to the N-terminal Pyr in ribonucleases, the N-terminal Pyr of human RNase 4 or RNase 5 is structurally flexible and is not involved in catalytic activity (31,32). Pyr is also found at the N-termini of most mouse ribonucleases, but its function is still not clear (33). Pyr is found at the N-termini of all known frog ribonucleases and both catalytic activity and cytotoxicity were markedly reduced if Met was added to the N-terminus of frog ribonuclease, i.e. onconase from R.pipiens and RC-RNase 4 from R.catesbeiana (15,34).

The role of Pyr1 on the catalytic activity and cytotoxicity of frog ribonuclease could be further investigated by biophysical studies. From our previous study, three hydrogen bonds exerted by Pyr1 are observed in the RC-RNase·d(ACGA) complex (26). First, the Oε of Pyr1 binds to Lys9, which further hydrogen bonds to the catalytic phosphate, and the Lys9 is also bound to the side chain amide of Asn38 and the hydroxyl group of Tyr28 by hydrogen bonds. Secondly, the Nα of Pyr1 binds to Val102, which further hydrogen bonds to Lys95 which interacts with the B₂ substrate G. Thirdly, the Oα of Pyr1 directly binds to the N₂ amine of the substrate G. The hydrogen bond network contributed by these residues at different α-helices, β-strands and loops is able to strengthen the structural integrity of ribonuclease, e.g. Pyr1 (α1), Lys9 (α1), Val96 (β6), Tyr28 (loop 2), Asn38 (β1) and Lys95 (β5) in RC-RNase. In native onconase, two hydrogen bonds exerted by Pyr1 are bound to Lys9 and Val96, equivalent to Lys9 and Val102 of RC-RNase (26,35). In this study, we found that the thermostability (∼7°C reduction in T_m), catalytic activity (0.5–0.9% left) and cytotoxicity (IC₅₀ >> 8 µM) were severely reduced by Met addition at the N-terminus. The thermostabilities of all the Pyr1-substitutional variants slightly reduced (∼3°C reduction in T_m), but their specific catalytic activities for the innate substrate UG markedly reduced (1.0–4.7% left). However, their cytotoxicities towards HeLa S-3 cells varied, e.g. rMOnc-Q1A (IC₅₀ > 8 µM), rMOnc-Q1S (IC₅₀ = 4 µM), rMOnc-Q1E (IC₅₀ = 4 µM), even if the non-specific catalytic activity of rMOnc-Q1A (57.1%) is higher than that of rMOnc-Q1S (20.6%) and rMOnc-Q1E (0.9%).

It has been shown by DiDonato’s group that Met addition at the N-terminus markedly decreases the catalytic activity for total RNAs and the cytotoxicity towards tumor cells, but only slightly reduces its conformation stability and resistance against pepsin digestion. The depletion of the fourth disulfide bridge (Cys87–Cys104) markedly decreases the stability of the protein, but does not significantly reduce its catalytic activity and cytotoxicity (36). Rybak’s group has shown that rM^–1Onc-E1S and rM^–1Onc-E1Y restore significant catalytic activities and cytotoxicities from that of rM^–1Onc, and become susceptible to human ribonuclease inhibitor. They suggested that the hydroxyl group of Ser or Tyr behind Met is crucial for the catalytic activity and cytotoxicity of onconase (37). In this study, we prepared a sMOnc-Q1S with Met–Ser–Asp–Trp–Ala at the N-terminus and found that both catalytic activity and cytotoxicity are similar to that of rMOnc, which are much less than that of wild-type onconase. Therefore, we prepared the rMOnc-Q1S from inclusion bodies without the pelB signal peptide using the method as described by Rybak’s group for the production of rM^–1Onc-E1S and rM^–1Onc-E1Y (37), and found that the rM^–1Onc-E1S has a significant increase in catalytic activity and cytotoxicity (Table 2 and Fig. 6A). After the MS and Edman degradation analyses, we found the N-terminal residue of rM^–1Onc-E1S is Ser rather than Met, as expected (Fig. 3). Therefore, we suggest that the hydrogen bond network mainly contributed by N-terminal Pyr is able to maintain the structural integrity for specific catalytic activity for UG and cytotoxicity towards tumor cells, while the disulfide bridges linking β-strands distant from the catalytic center do not possess these functions. In addition, a hydrogen bond network similar to that exerted by Pyr may be formed by Ser or Glu, but not by Ala substitution or Met addition at the N-terminus.

The Met preceding the Ser of rMOnc-Q1S and the Ala of rMOnc-Q1A was removed, while the Met at the N-termini of rMOnc, rMOnc-Q1E and rMOnc-Q1Y in inclusion bodies and the sMOnc-Q1S in the culture medium was retained. The in vivo removal of Met from the small and uncharged penultimate residue, i.e. Ala and Ser, may be mediated through an E.coli methionine aminopeptidase inside the cell, which is unable to remove the Met from the bulky or charge penultimate residue, such as Gln, Glu and Tyr residues (38). The Met preceding the Gln of wild-type onconase was not removable by the enzyme except when the mutation occurs at the penultimate residue, e.g. Q1A or Q1S mutation; however, these mutations usually cause a significant reduction in thermostability and cytotoxicity compared with the wild type. In this study, we expressed the onconase gene from genomic DNA rather than from the synthetic or semi-synthetic onconase gene in E.coli and produced various N-terminal variants through the pelB signal peptide without treatment with cyanogen bromide or Aeromonas proteolytica aminopeptidase (15,31,37). Consequently, the method employed here, using the pelB signal peptide, is able to produce recombinant proteins with any amino acid residue at the N-terminus.

The base preferences for pyrimidine (C, U) and G in the B₁ and B₂ sites, respectively, are the unique properties of frog cytotoxic ribonuclease, whereas those for respective pyrimidine (C, U) and A are found in most mammalian ribonucleases. In this study, we found that the hydroxyl group of Thr35 of onconase is essential for pyrimidine recognition and Asp67 is crucial for U recognition at the B₁ site because sOnc-T35A and sOnc-D67T recognized both C and U at the B₁ site, while sOnc-D67A preferred C in contrast to U of wild-type onconase (Fig. 5C). These results are in good agreement with the structure of the RC-RNase·d(ACGA) complex. RC-RNase is a ribonuclease isolated from R.catesbeiana, which prefers both C and U at the B₁ site. The hydroxyl group of Thr39 serves as a hydrogen bond donor for the N₃ of cytosine (C) and simultaneously serves as a hydrogen bond acceptor from the hydroxyl group of Thr70. In contrast, the hydroxyl group of Thr39 may serve as a hydrogen bond acceptor from U and as a hydrogen donor for Thr70. The corresponding residues Thr39 and Thr70 of RC-RNase in onconase are Thr35 and Asp67 (26). Therefore, we suggest that the carboxyl group of Asp67 of onconase serves as a distal hydrogen bond acceptor from U through the hydroxyl group of Thr35, but not as a distal hydrogen bond donor for C. The B₁ base preference of onconase for U is similar to human RNase 4, which also possesses Thr44 and Asp80 and respective hydrogen bonds at the corresponding sites (32).

Despite having 10⁵-fold differences in relative ribonucleolytic activity, several members of the RNase A superfamily manifest other biological activities, e.g. human EDN and ECP possess neurotoxicity, and human angiogenin induces blood vessel formation (3–5). The relationship between catalytic activity and biological function remains unclear; much effort has been expended on this subject, but the results are still debatable. For example, alkylation at the ribonuclease active site simultaneously eliminates both ribonuclease and neurotoxic activity of EDN, ECP and onconase (39,40). The mutation of the catalytic residue, e.g. H103A, abolishes both catalytic activity and cytotoxicity of RC-RNase (13). On the other hand, the antiparasitic activity and the antibacterial activity of ECP appears to be unaffected by the presence of the ribonuclease inhibitor or the mutation of the catalytic residue (41). In this study, we found that the structural integrity and catalytic activity for innate UG exerted by Pyr1 are required for the cytotoxicity, but the non-specific catalytic activity for total RNAs is not sufficient to enhance the cytotoxicity. Therefore, we suggest that RNAs with nude UG sequence may be susceptible to internalized onconase in tumor cells, but resistant to mammalian ribonucleases which prefer a pyrimidine–adenine substrate.

DDBJ/EMBL/GenBank accession no. AF332139