High-level soluble production and characterization of porcine ribonuclease inhibitor.

Ribonucleases can be cytotoxic if they retain their ribonucleolytic activity in the cytosol. The cytosolic ribonucleolytic activity of ribonuclease A (RNase A) and other pancreatic-type ribonucleases is limited by the presence of excess ribonuclease inhibitor (RI). RI is a 50-kDa cytosolic scavenger of pancreatic-type ribonucleases that competitively inhibits their ribonucleolytic activity. RI had been overproduced as inclusion bodies, but its folding in vitro is inefficient. Here, porcine RI (pRI) was overproduced in Escherichia coli using the trp promoter and minimal medium. This expression system maintains pRI in the soluble fraction of the cytosol. pRI was purified by affinity chromatography using immobilized RNase A and by anion-exchange chromatography. The resulting yield of 15 mg of purified RI per liter of culture represents a 60-fold increase relative to previously reported recombinant DNA systems. Differential scanning calorimetry was used to study the thermal denaturation of pRI, RNase A, and the pRI-RNase A complex. The conformational stability of the complex is greater than that of the individual components.

A single amino acid substitution changes ribonuclease 4 from a uridine-specific to a cytidine-specific enzyme.

The structural features underlying the strong uridine specificity of ribonuclease 4 (RNase 4) are largely unknown. It has been hypothesized that the negatively charged alpha-carboxylate is close to the pyrimidine binding pocket, due to a unique C-terminal deletion. This would suppress the cleavage of cytidine-containing substrates [Zhou, H.-M., and Strydom, D. J. (1993) Eur. J. Biochem. 217, 401-410]. Replacement of the alpha-carboxylate by an alpha-carboxamide in a fragment complementation system decreased both (kcat/Km)CpA and (kcat/Km)UpA , thus refuting the hypothesis. However, model building showed that the deletion allowed the side chain of Arg-101 to reach the pyrimidine binding pocket. From the 386-fold reduction in (kcat/Km)UpA in RNase 4;R101N, it is concluded that this residue functions in uridine binding, analogous to Ser-123 in RNase A. In addition, it may have an effect on Asp-80. The 2-fold increase in (kcat/Km)CpA in the mutant R101N and the close proximity of the side chains of Arg-101 and Asp-80 suggested that the latter could be involved in suppressing CpA catalysis. The substrate specificity of RNase 4;D80A was completely reversed: (kcat/Km)UpA decreased 159-fold, whereas (kcat/Km)CpA increased 233-fold. The effect on CpA was unexpected, because the corresponding residue in bovine pancreatic RNase A (Asp-83) hardly affects cytidine-containing substrates. Furthermore, the residue is conserved in nearly all sequences of mammalian RNase 1. Thus, an evolutionary highly conserved residue does not necessarily function in the same way in homologous enzymes. A model, which proposes that the structure of RNase 4 has been optimized to permanently fix the position of Asp-80 and impede its movement away from the pyrimidine binding pocket, is presented.

Structural determinants of the uridine-preferring specificity of RNase PL3.

RNase PL3 is a structurally highly conserved, pyrimidine-specific RNase, which strongly prefers to cleave at the 3'-side of uridine. Here, question of which residues are involved in determining substrate specificity is addressed. The difference in the rate of cleavage of UpA and CpA was found to result from a 375-fold larger kcat for the former substrate, whereas the values of Km were essentially the same. The pyrimidine specificity of this class of RNases is thought to result from hydrogen bonds between the base and a threonine residue in the B1 subsite. Mutation of this residue (Thr-44) in RNase PL3 resulted in strongly reduced activity with UpA and poly(U). However, the activity with CpA and poly(C) had increased. Comparison with the effect of the same mutation in RNase A [delCardayre, S. B., & Raines, R. T. (1994) Biochemistry 33, 6031-6037] and angiogenin [Curran et al. (1993) Biochemistry 32, 2307-2313] showed that the function of this threonine in substrate recognition is different in three RNase subfamilies. Previous studies have shown that the 36-42 region contains one or more residues that are involved in substrate recognition [Vicentini et al. (1994) Protein Sci. 3, 459-466]. Site-directed mutagenesis of amino acids in this region identified Phe-42 as the only single residue that affected the cytidine/uridine specificity ratio. The mutation F42V resulted in a 10-fold increase in kcat and a 1.9-fold decrease in Km for CpA. The properties of the double mutant F42V/T44A suggested that a suboptimal binding of cytidine is caused by Phe-42, partially through an effect on Thr-44.

Residues 36-42 of liver RNase PL3 contribute to its uridine-preferring substrate specificity. Cloning of the cDNA and site-directed mutagenesis studies.

Within the superfamily of homologous mammalian ribonucleases (RNases) 4 distinct families can be recognized. Previously, representative members of three of these have been cloned and studied in detail. Here we report on the cloning of a cDNA encoding a member of the fourth family, RNase PL3 from porcine liver. The deduced amino acid sequence showed the presence of a signal peptide, confirming the notion that RNase PL3 is a secreted RNase. Expression of the cDNA in Escherichia coli yielded 1.5 mg of purified protein/liter of culture. The recombinant enzyme was indistinguishable from the enzyme isolated from porcine liver based on the following criteria: amino acid analysis, N-terminal amino acid sequence, molecular weight, specific activity toward yeast RNA, and kinetic parameters for the hydrolysis of uridylyl(3',5')adenosine and cytidylyl(3',5')adenosine. Interestingly, the kinetic data showed that RNase PL3 has a very low activity toward yeast RNA, i.e., 2.5% compared to pancreatic RNase A. Moreover, using the dinucleotide substrates and homopolymers it was found that RNase PL3, in contrast to most members of the RNase superfamily, strongly prefers uridine over cytidine on the 5' side of the scissile bond. Replacement, by site-directed mutagenesis, of residues 36-42 of RNase PL3 by the corresponding ones from bovine pancreatic RNase A resulted in a large preferential increase in the catalytic efficiency for cytidine-containing substrates. This suggests that this region of the molecule contains some of the elements that determine substrate specificity.

Protein chemical and kinetic characterization of recombinant porcine ribonuclease inhibitor expressed in Saccharomyces cerevisiae.

A cDNA encoding porcine ribonuclease inhibitor was used to express this protein in yeast under control of the PHO5 promoter. The recombinant protein was purified to homogeneity with a yield of 0.2 mg/g of yeast cells (wet weight) and was found to be indistinguishable from the inhibitor isolated from porcine liver on the basis of the following criteria: the amino acid composition, the number of free sulfhydryl groups, the molecular weight of the native and the denatured protein, peptide mapping, and amino acid sequence analysis of the N- and C-terminal regions of the protein. A simple method was developed for measuring accurately the slow, tight-biding kinetics of the inhibition of ribonuclease by ribonuclease inhibitor. From the dependence of the observed inhibition constant on the substrate concentration, it could be concluded that RI was competitive with the substrate UpA. The dependence of the observed association rate constant on the substrate concentration was consistent with a two-step mechanism in which the substrate only competed in the second (isomerization) step. The values for the inhibition constant for the inhibition of RNase by the recombinant inhibitor, 67 fM, the association rate constant, 1.5 x 10(8) M-1.s-1, and the dissociation rate constant, 8.3 x 10(-6) s-1, were in good agreement with those obtained for the porcine liver RNase inhibitor.

Rapid construction of large synthetic genes: total chemical synthesis of two different versions of the bovine prochymosin gene.

We have tested several different synthesis designs and assembly methodologies to develop an improved gene synthesis strategy which enables significantly longer nucleotide sequences to be easily constructed. This strategy, based in part upon our ability to synthesize high-quality extended-length oligodeoxynucleotides (over 100-mer in length), together with the use of chemical 5'-phosphorylation, and simplified low-melting-temperature agarose gel purification methods, combines ease, speed and high overall efficiency. We show that it is now feasible to synthesize routinely even long genes (at least 1-2 kb). To demonstrate this capability we have chemically synthesized and assembled two different versions of the gene encoding the bovine enzyme prochymosin (prorennin). One gene is essentially the natural bovine prochymosin gene sequence. In the second gene the codons have been optimized with regard to the codon bias of highly expressed yeast genes. Each synthetic gene was in excess of 1100 bp, yet they were assembled from only 13 or 14 pairs of complementary oligodeoxynucleotides (oligos), the average lengths of which were 87 and 82 bp, respectively. The 'mutation' rate was low enough to assess that more than 75% of all such oligo pairs (160-170 total nt) were error-free.