Changing the net charge from negative to positive makes ribonuclease Sa cytotoxic

Abstract

Ribonuclease Sa (pI = 3.5) from Streptomyces aureofaciens and its 3K (D1K, D17K, E41K) (pI = 6.4) and 5K (3K + D25K, E74K) (pI = 10.2) mutants were tested for cytotoxicity. The 5K mutant was cytotoxic to normal and v-ras-transformed NIH3T3 mouse fibroblasts, but RNase Sa and 3K were not. The structure, stability, and activity of the three proteins are comparable, but the net charge at pH 7 increases from −7 for RNase Sa to −1 for 3K and to +3 for 5K. These results suggest that a net positive charge is a key determinant of ribonuclease cytotoxicity. The cytotoxic 5K mutant preferentially attacks v-ras-NIH3T3 fibroblasts, suggesting that mammalian cells expressing the ras-oncogene are potential targets for ribonuclease-based drugs.

The mammalian ribonucleases (RNases), of which RNase A is the best known, are the most rapidly evolving protein family. Some ribonucleases are cytotoxic, and they are being tested in clinical trials for use in cancer therapy (Leland and Raines 2001). Ribonucleases must enter the target cell and degrade the RNA to exert their cytotoxic effect (Wu et al. 1995). However, the molecular determinants of RNase-induced cell death are not well understood (Iordanov et al. 2000a). At least four members of the RNase A superfamily are endowed with antitumor activity and show cytotoxicity toward several tumor cell lines: BS-RNase from bull semen, onconase from oocytes of Rana pipiens, and two closely related sialic acid-binding lectins from oocytes of Rana catesbeiana and Rana japonica (Notomista et al. 2000). Ribonuclease activity is essential for cytotoxicity (Kim et al. 1995a,b). Members of the RNase A superfamily that are inhibited by ribonuclease inhibitor (RI) are not cytotoxic (Bretscher et al. 2000). For this reason, RNase A is not cytotoxic to cells containing RI. To circumvent this problem, site-directed mutagenesis is being used to alter the properties of RNase A so that it is not inhibited by RI (Bretscher et al. 2000).

The microbial RNases are another important family of RNases, and they are not inhibited by the RI from mammalian cells. A hybrid protein composed of Bacillus amyloliquefaciens ribonuclease (barnase) and Pseudomonas exotoxin A was toxic to different cell lines due to its ribonuclease activity (Prior et al. 1996). The ribonuclease from Bacillus intermedius (binase) exhibits antiproliferative action toward chicken embryo fibroblasts and mouse fibroblasts (Ilinskaya et al. 2001), and demonstrates genotoxic effects by induction of forward Ara^R-mutations and by histidine-reverse mutations as well as by prophage-induction activity (Ilinskaya et al. 1995). These results suggest that microbial ribonucleases might be useful agents for chemotherapy.

The smallest microbial ribonuclease is RNase Sa, a ribonuclease from Streptomyces aureofaciens. Wild-type RNase Sa is an acidic protein (pI = 3.5) that contains no Lys residues and has a net charge of −7 at pH 7 (Hebert et al. 1997). Replacement of Asp and Glu residues on the surface of RNase Sa with Lys residues was used to produce mutant RNases: 3K (D1K, D17K, E41K) with a net charge of −1 at pH 7, and 5K (3K + D25K, E74K) with a net charge of +3 at pH 7 (Shaw et al. 2001). These charge reversal mutants were stable and retained most of their enzyme activity (Shaw et al. 2001). These results and NMR studies (D. Laurents, unpubl.) suggested that the conformations of RNase Sa, 3K, and 5K are very similar. Futami et al. (2001) used chemical modification to increase the positive charge on bovine RNase A and human RNase 1. They found that this procedure made the enzymes less sensitive to inhibition by RI, and better able to enter cells. Finding that an extracellular human enzyme such as RNase 1 could be effectively internalized into cells by increasing the net positive charge suggested a simple strategy for improving the efficient delivery of other cytotoxic ribonucleases into cells (Futami et al. 2001).

In this paper, we report studies of the cytotoxicity of RNase Sa and its 3K and 5K mutants. We provide further evidence that the net charge of the molecule is an important determinant of ribonuclease cytotoxicity. The selectivity of toxic action to normal and tumor cells is another important parameter. For example, v-ras-transformed NIH3T3 cells were more sensitive to onconase than the parental NIH3T3 fibroblasts (Smith et al. 1999), and binase demonstrated greater growth inhibition of v-ras-transformed cells than normal fibroblasts (Ilinskaya et al. 2001). We report similar selectivity in the cytotoxic effect of the 5K mutant of RNase Sa to normal NIH3T3 fibroblasts and fibroblasts expressing v-ras-oncogene.

Results

The 5K mutant of RNase Sa is a far more potent toxic agent toward fibroblast cell lines than RNase Sa or the 3K mutant (Fig. 1A ▶, Table 1). At ∼50 μM concentration, this enzyme reduced cell viability to 24% for NIH3T3 and to 3% for v-ras-NIH3T3 fibroblasts, showing that the v-ras-transformed fibroblasts are more sensitive to the 5K mutant. The 3K mutant exhibited a weak toxic effect toward v-ras-transformed fibroblasts, and RNase Sa was nontoxic (Fig. 1A ▶). A more detailed investigation by direct cell counting and a determination of respiration rate did not detect toxic effects for RNase Sa or the 3K mutant even after 48 h (data not shown).

Fig. 1.

(A) Viability of normal NIH3T3 fibroblasts (blank columns) and v-ras-NIH3T3 fibroblasts (hatched columns) treated for 24 h with RNase Sa and its 3K and 5K mutants at a concentration of 500 μg/mL. Values are expressed as percent viability of control cells grown without RNase. (B,C) Effect of 5K RNase Sa on the cell number (B) and respiration rate (C) of normal NIH3T3 fibroblasts (blank columns) and v-ras-NIH3T3 fibroblasts (hatched columns) treated for 24 h with various concentrations of 5K RNase Sa: bars a and e, untreated; b and f, 10 μg/mL; c and g, 100 μg/mL; and d and h, 500 μg/mL.

Table 1.

Viability of v-ras-NIH3T3 fibroblasts treated for 24 h with RNase Sa and its 3K and 5K mutants at a concentration of 500 μg/ml compared to molecular characteristics of RNases

	Molecular characteristics of RNasesa
RNase	pI	Net charge at pH 7	Tmc (°C) at pH 7	kcat/KM (mM−1 s−1) at pH 7	Viability (%)b
Sa	3.5	−7	47.2	768	97
3K	6.4	−1	40.7	629	87
5K	10.2	+3	47.3	105	3

^a Shaw et al. (2001).

^b Viability of untreated cells was taken as 100%.

^c Midpoint of the thermal unfolding curve.

The 5K mutant reduced the number of viable cells and respiration rate in both cell lines in a concentration-dependent manner (Fig. 1B,C ▶). A more than two-fold reduction in the number of viable v-ras-NIH3T3 cells (Fig. 1B ▶) and a 16-fold decrease in respiration rate (Fig. 1C ▶) were observed. Normal fibroblasts retained a higher level of cell respiration than v-ras-NIH3T3 cells (Fig. 1C ▶). This finding indicates a higher resistance of normal fibroblasts toward the toxic action of the 5K mutant compared to v-ras-transformed cells. Additional evidence is the observation of morphological changes in both cell lines treated with 500 μg/mL 5K mutant: Normal fibroblasts underwent vacuolization but retained adhesive properties, whereas transformed cells lost their adhesion.

Discussion

Unlike the mammalian ribonuclease family, microbial RNases are not inhibited by the ribonuclease inhibitor from human cells. Our goal was to use RNase Sa and its charge reversed mutants to gain a better understanding of ribonuclease cytotoxicity for the possible development of microbial ribonucleases as chemotherapeutics. We replaced Asp and Glu residues on the surface of RNase Sa with Lys residues to produce mutants with a strikingly different net charge. By reversing five charges on RNase Sa, we have changed it from one of the most acidic to one of the most basic proteins, and our results show clearly that this is sufficient to generate a cytotoxic ribonuclease.

Normal NIH3T3 fibroblasts were less sensitive to the 5K mutant of RNase Sa compared to v-ras-transformed fibroblasts. Similar data were obtained for binase (Ilinskaya et al. 2001) and onconase (Smith et al. 1999), suggesting the key role of a positive mediator (Ras) of the proliferative signal transduction for a cytotoxic response in fibroblasts.

It is known that ribonucleolytic activity is essential for the cytotoxicity of RNases (Kim et al. 1995a,b). Onconase, an anticancer ribonuclease, damages cellular tRNA and causes caspase-dependent apoptosis in targeted cells that depends on its RNase activity (Iordanov et al. 2000b). The SOS response-inducing potency as a marker of genotoxicity of native and mutant binases with altered catalytic properties strongly correlated with ribonucleolytic activity (Ilinskaya et al. 1996). An alkylated derivative of onconase with 2% residual ribonuclease activity was not toxic (Smith et al. 1999), and a mutant barnase with <2% residual catalytic activity and photooxidized binase were not nephrotoxic (Ilinskaya and Vamvakas 1997). Cationic variants of RNase A and RNase 1 created using chemical modification of their carboxyl groups exhibited no cytotoxic effect if their catalytic activity was less than 0.01% of the nonmodified enzyme (Futami et al. 2001). So, as expected, ribonucleases must retain a definite level of ribonucleolytic activity to be toxic to cells.

All three ribonucleases used in our work are active enzymes (Shaw et al. 2001). In comparison to wild-type RNase Sa, the value of k_cat/K_M for cleavage of poly(I) is reduced by 18% for 3K, and by 86% for 5K RNase Sa (Table 1). Cytotoxicity was only observed for 5K RNase Sa, which has the lowest catalytic activity of the ribonucleases tested. Similarly, mammalian ribonucleases made more positively charged by chemical modification and retaining only about 1.5% catalytic activity were cytotoxic toward Swiss mouse albino 3T3 cells transformed by SV40 (Futami et al. 2001). Thus, a definite level of catalytic activity is essential but not sufficient to generate cytotoxicity.

Klink and Raines (2000) demonstrated that the conformational stability of RNase A mutants correlates directly with the cytotoxicity. The melting temperatures of RNase Sa, 3K and 5K in Table 1 show that a feature other than stability determines the cytotoxicity of 5K.

Our results show that the cytotoxic properties of RNase Sa correlate with the change in net charge from negative to positive (Table 1). This probably allows the 5K mutant to bind tighter to the negatively charged glycolipids in the outer part of the plasma membrane than do RNase Sa and 3K. This then promotes entry of the enzyme into the cell and may be the rate-limiting step for cytotoxicity. This is in agreement with the results of Futami et al. (2001) that show that the cytotoxicity and cell binding ability of chemically modified bovine RNase A and human RNase 1 correlate with their net positive charge. Site-directed mutagenesis allows the creation of positively charged ribonucleases with enhanced toxicity and avoids the unfavorable side effects of chemical modification. Thus, the development of mutant ribonucleases with increased positive charge may lead to more effective therapeutics.

Materials and methods

Wild-type RNase Sa and the 3K and 5K mutants were prepared as described in Shaw et al. (2001). Two cell culture lines were used: normal NIH3T3 mouse fibroblasts and NIH3T3 mouse fibroblasts with retroviral stock from v-Ha-ras-transfected psi-2-cells. Ras encodes G (guanine-nucleotide-binding and GTP-hydrolyzing)-protein that is located on the inner face of the plasma membrane and couples growth factor receptors to effector proteins in the cell. The cellular G-protein (Ras) is involved in signal transmission and plays an important role in mitosis and differentiation, and its viral counterpart (v-Ras) is a constitutively active protein which leads to cellular transformation (Bishop 1991). Animal fibroblasts were maintained at 37°C in a humidified, 6% CO₂/94% air atmosphere in Dulbecco's modified Eagle's medium (pH 7.2) containing 5% calf serum, 2 mM L-glutamine, and antibiotics. After confluence, cultures were treated with trypsin for 1 min and cells were plated at a density of 6 × 10⁵ cells per 35-mm dish. Cells were cultivated for 24 h before addition of a ribonuclease. The ribonuclease concentration varied from 10 to 500 μg/mL.

Viability was assessed with the cell proliferation reagent WST-1 (Roche Diagnostics, Mannheim, Germany). This test is based on cleavage of a water-soluble tetrazolium salt by mitochondrial dehydrogenases in viable cells. Cells were incubated with WST-1 for 20 min. After centrifugation of liquid phase (at 5000g for 5 min), absorbance of aliquots of supernatants was measured at 450 nm. The cell-free medium/reagent mixture was used as the background control. Cell viability was calculated as the difference in absorbance between each variant and background according to Ilinskaya et al. (2001). The viability of untreated cells was taken as 100%. The cell number of each dish was counted after trypsinization in a cell chamber counter using a light microscope. Respiration rate was determined as a change in the absorbance of WST solution per h per 10⁶ cells. Values reported are the mean of three measurements with the standard deviation. Two groups of data were regarded as different if they satisfied the t-criterion with P<0.01.