The Flexible and Clustered Lysine Residues of Human Ribonuclease 7 Are Critical for Membrane Permeability and Antimicrobial Activity*

The ubiquitous ribonucleases (RNases) play important roles in RNA metabolism, angiogenesis, neurotoxicity, and antitumor or antimicrobial activity. Only the antimicrobial RNases possess high positively charged residues, although their mechanisms of action remain unclear. Here, we report on the role of cationic residues of human RNase7 (hRNase7) in its antimicrobial activity. It exerted antimicrobial activity against bacteria and yeast, even at 4 °C. The bacterial membrane became permeable to the DNA-binding dye SYTOX® Green in only a few minutes after bactericidal RNase treatment. NMR studies showed that the 22 positively charged residues (Lys18 and Arg4) are distributed into three clusters on the surface of hRNase7. The first cluster, K1,K3,K111,K112, was located at the flexible coil near the N terminus, whereas the other two, K32,K35 and K96,R97,K100, were located on rigid secondary structures. Mutagenesis studies showed that the flexible cluster K1,K3,K111,K112, rather than the catalytic residues His15, Lys38, and His123 or other clusters such as K32,K35 and K96,R97,K100, is critical for the bactericidal activity. We suggest that the hRNase7 binds to bacterial membrane and renders the membrane permeable through the flexible and clustered Lys residues K1,K3,K111,K112. The conformation of hRNase7 can be adapted for pore formation or disruption of bacterial membrane even at 4 °C.


EXPERIMENTAL PROCEDURES
Preparation and Enzymatic Assay of RNases-The coding region of the human RNase7 (hRNase7) gene was obtained from the genomic DNA of HeLa cells by PCR. The NdeI and BamHI restriction sites were hung on the 5Ј and 3Ј end of the gene fragment, respectively, and subcloned into the T7 RNA polymerase-driven expression vector pET22b (Novagen) (11). Site-directed mutagenesis was made by PCR as previously described (12). The uniformly 15 N-or 13 C-labeled hRNase7 with an extra Met at the N terminus was produced in Escherichia coli BL21(DE3). The recombinant hRNase7 from inclusion bodies were refolded and purified as previously described (11). The ribonucleolytic activity of recombinant hRNase7 was analyzed by the zymogram assay on RNA-casting PAGE as previously described (13).
Binding of RNase to Bacteria-Small aliquots of RNases (2 g in 5 l) were incubated with 45 l (5 ϫ 10 6 cfu) of sodium phosphate-washed P. aeruginosa at 37°C for 30 min. The RNases bound onto bacteria were spun down, washed five times with 10 mM sodium phosphate, and analyzed by SDS-PAGE and silver staining.
Assays of the Permeability of Bacterial Membrane-The overnight culture of P. aeruginosa (10 7 cfu) was washed and resuspended in 100 l of water and incubated with 1 M SYTOX Green (Molecular Probes) in a dark 96-well plate for 15 min in the dark. After the addition of RNase, the increase of fluorescence due to the binding of the dye to the intracellular DNA was measured in the same microplate reader using 485and 520-nm filters for excitation and emission wavelengths, respectively (15).
Circular Dichroism (CD) Experiments-CD experiments were carried out using an Aviv CD 202 spectrometer (Lakewood, NJ) using a 1-mm path length cuvette with 20 M hRNase7 in 20 mM sodium phosphate. The CD spectra at different temperatures and pH values were recorded from 190 to 260 nm using a wavelength step of 0.5 nm. Equilibrium thermal denaturing experiments were performed using protein samples dissolved in 20 mM phosphate buffer, pH 3.5, by measuring the change of molar ellipticity at 201 nm. Data were collected as a function of temperature with a scan rate of 2°C/min and allowing 3 min to reach equilibrium over the range of 4 -95°C. Equilibrium unfolding induced by guanidine HCl was monitored by CD as described previously (11). The curves were fitted and analyzed using SigmaPlot version 8.02 (SPSS Inc.).
NMR Spectroscopy-All NMR experiments were performed on Bruker AVANCE 600 and 800 spectrometers equipped with triple ( 1 H, 13 C, and 15 N) resonance probes, including a shielded z-gradient. The RNase samples (0.6 mM in 0.35 ml) were prepared in 50 mM phosphate buffer in 90% H 2 O/10% D 2 O or 99.9% D 2 O at pH 3.5, 310 K and kept in a Shigemi NMR tube. All heteronuclear NMR experiments were carried out as described previously (16). Sequence-specific assignment of the backbone atoms of hRNase7 was achieved by the independent connectivity analysis of CBCA(CO)NH, HNCACB, HNCO, HN(CA)CO, and C(CO)NH. The 1 H resonances were assigned using TOCSY-HSQC, HAHB(CO)NH, and HCCH-TOCSY. Combined information from two-dimensional 1 H-15 N HSQC and three-dimensional NOESY-HSQC experiments yielded assignments for side chain amide resonances of the Asn and Gln residues. Aromatic resonances were assigned using twodimensional 1 H-13 C HSQC, NOESY, and TOCSY data. Linear prediction was used in the 13 C and 15 N dimensions to improve the digital resolution. All spectra were processed using the NMRPipe software package (17) and analyzed using NMRView version 5.0 (18).
NMR Restraints and Tertiary Structure Calculation of hRNase7-The dihedral angle information was predicted by the TALOS program (19). The hydrogen bonding information was obtained from D 2 O exchange monitored by two-dimen-  (9). c Liao, et al. (5).

RESULTS
Antimicrobial Activity of RNase Superfamily Proteins-The hRNase7 was more effective (0.1 M for 10 2 -fold reduction) in cfu compared with that of buffer only against bacteria (10 5 cfu P. aeruginosa) than bullfrog RC-RNase6 (5 M), whereas the active RNA-degrading RNases (bovine RNaseA and bullfrog RC-RNase) and all other bullfrog oocytic RNases were not bactericidal at 80 M. The 12-mer Arg-rich positively charged oligopeptide (R 3 S 1 ) 3 , however, possessed similar bactericidal activity (0.2 M) as that of hRNase7 (0.1 M) (Fig. 1, A and B).
The antimicrobial spectrum of hRNase7 was also examined, and the yeast P. pastoris X-33 (0.03 M for 10 2 -fold reduction in cfu compared with that of buffer only) was the most sensitive among all of the microbes tested and thereafter in order the Gram-negative bacterium P. aeruginosa (0.1 M), whereas the Gram-positive S. aureus (1 M) and Gram-negative E. coli bacteria (10 M) and yeast C. albicans (Ͼ 20 M) were not sensitive (Fig. 1, B and C).
Binding and Permeability to Bacterial Membrane-The cfu of susceptible bacteria was reduced 10 2 -fold in only a few min-  utes after RNase addition (data not shown). The bactericidal RNases (hRNase7 and RC-RNase6) were bound to the susceptible bacterium P. aeruginosa (Fig. 2, lanes 2 and 11). In contrast, the non-bactericidal bovine RNaseA and bullfrog RC-RNase did not bind to the bacteria (Fig. 2, lanes 5 and 8).
The membrane of P. aeruginosa became permeable to the DNA-binding dye SYTOX Green in only a few minutes after the addition of 2.5 M hRNase7 or bullfrog RC-RNase6 to the bacteria (10 7 cfu in 100 l). However, the non-bactericidal bovine RNaseA and bullfrog RC-RNase had no effect under the same conditions, which was in good agreement with the antimicrobial assay (Figs. 1A and 3). The bactericidal activity of hRNase7 was only slightly reduced at 4°C (0.3 M for 10 2fold reduction in cfu compared with that of 0.1 M at 37°C), whereas that of indolicidin, a 13-mer bactericidal oligopeptide from bovine neutroplils, was almost abolished (Fig. 4). These results suggest that the bactericidal activity of hRNase7 is not energy-dependent, whereas that of indolicidin is energy-dependent.
Conformational and Structural Stability of hRNase7-To elucidate the mechanism for the antimicrobial activity of hRNase7, the structure of the recombinant protein was analyzed by CD spectra and NMR studies. The important features are summarized and discussed below. First, the conformation of hRNase7 was pH-independent over the range of 3.5-9.5. The T m value of hRNase7 was 66.4°C. The chemical stability was determined with a C m value at 3.27 M guanidine-HCl. These T m and C m values indicate that hRNase7 is a stable protein in structure, similar to other RNaseA superfamily members.
Structure of hRNase7-The 800-MHz two-dimensional 1 H-15 N HSQC spectrum of the hRNase7 was obtained in which cross-peaks clearly dispersed (Fig. 5). For the determination of tertiary structures, a set of 1661 restraints were collected for simulated annealing calculations. Among these restraints, 1447 were interproton distances, 94 were hydrogen bonds, 116 were torsional angles, and 4 were disulfide bond restraints ( Table 2). The 15 structures of the lowest total energy were chosen to represent the ensemble of NMR structures (Fig. 6A). These structures were consistent with both experimental data and standard covalent geometry and displayed no violations Ͼ0.3 Å for distance restraints and no violations Ͼ3°for torsional angles. Superposition of each structure with the mean structure yielded an average root mean square deviation of 0.34 Ϯ 0.09 Å for the backbone atoms and 1.06 Ϯ 0.08 Å for the heavy atoms in residues 7-128. Analysis of the ensemble using PROCHECK-NMR revealed that 64.4% of the residues lay in the most favored regions and 33.9% of the residues lay in allowed regions in the Ramachandran , dihedral-angle plot ( Table 2). The solution structure of hRNase7 displayed an ␣ ϩ ␤ folding topology, which was composed of three ␣-helices and two antiparallel ␤-sheets, typical for the RNaseA superfamily. The structure was stabilized by four disulfide bridges (C 23 -C 81 , C 55 -C 106 , C 37 -C 91 and C 62 -C 69 ). Most of the hydrogen bonds were located in the ␣-helix and ␤-sheet regions. Most positively   FEBRUARY 16, 2007 • VOLUME 282 • NUMBER 7 charged residues were distributed into three clusters (Fig. 6B). The first was composed of Lys 1 , Lys 3 , Lys 111 , and Lys 112 ; the second, Lys 28 , Lys 32 , Lys 35 , Arg 36 , and Lys 38 , and the third, Lys 82 , Lys 94 , Lys 96 , Arg 97 , and Lys 100 . On the other hand, the negatively charged residues were randomly distributed over hRNase7 (Fig. 6C).

Residues Responsible for Antimicrobial Activity of hRNase7
Residues Responsible for Bactericidal Activity of hRNase7-Because cationic residues were abundant in bactericidal RNases as demonstrated in Table 1, we thus proceeded to determine the role of the RNA catalytic activity or cationic residue in the bactericidal activity. First of all, we found that the catalytic activitydeficient mutants of hRNase7 (H15A, K38A, and H123A) still conferred the same level of antimicrobial activity as the wildtype hRNase7 (Fig. 7, A and B; and Fig. 8A). The result shows that RNA catalytic activity is not essential for the antibacterial activity of RNases.
Deletion of the N-terminal Met and four residues (⌬ 1 KPKG 4 ) from recombinant hRNase7 markedly reduced the bactericidal activity but did not alter the RNA catalytic activities (Fig. 7, C and D; and Fig. 8B). Further analysis showed that the K3A mutant exerted less bactericidal activity than the K1A mutant (Fig. 7, C and D; and Fig. 8C). However, the substitution  of the K111Q/K112Q residues in the same cationic cluster reduced the bactericidal activity to a lesser extent than N-terminal deletion. In contrast, mutations in other positively charged clusters (K32N/K35Q and K96A/R97A/K100T) altered neither the catalytic nor the bactericidal activity (Fig. 7, C and D; and Fig. 8B). These results show that the flexible Lys 1 ,Lys 3 , as well as Lys 111 ,Lys 112 residues in the first cationic cluster are critical for the bactericidal activity. Structural Comparison of Wildtype and K3A Mutated hRNase7-The structure of K3A-hRNase7 was nearly identical to that of wild-type hRNase7, as there was no difference in the cross-peaks of two-dimensional N 15 -H 1 -HSQC and three-dimensional N 15 -NOESY and HSQC NMR spectra between them, except residues Gly 4 , Leu 124 , and Ala 3 , although K3A-hRNase7 had less antimicrobial activity than wildtype hRNase7 (Fig. 8C).

DISCUSSION
The structures of bactericidal RNases hRNase3 and hRNase7 are similar to those of RNaseA superfamily proteins with three ␣-helices and two triple-stranded antiparallel ␤-sheets. Among these RNases, only the bactericidal RNases contained abundant positively charged residues on the enzyme surface, but the non-bactericidal RNases did not ( Table 1). The abundance of the cationic residues Lys and Arg is also found in most antimicrobial peptides. These peptides possess amphipathic structures that are composed of clustered cationic and hydrophobic residues on each side of the structure, although they have diverse primary sequences and different secondary structures (23,24). The cationic residues may facilitate their interaction with the negatively charged components on the microbial surface, and the hydrophobic residues may permit their incorporation into microbial membrane (25).
In this study, we have found that three clusters of cationic residues are located on the surface of hRNase7. The first cluster consists of Lys 1 ,Lys 3 , Lys 111 ,Lys 112 residues, the second and third clusters contain Lys 32 ,Lys 35 and Lys 96 ,Arg 97 ,Lys 100 , respectively. Only cationic residues in the first cluster, K 1 ,K 3 ,K 111 ,K 112 , are Right, the ribbon representation of the best NMR structure that possesses the lowest total energy is shown. The hRNase7 is composed of three ␣-helices (red) and two triple-stranded antiparallel ␤-sheets (cyan). B, surface structure of hRNase7 is displayed as a 180°rotation with positively and negatively charged residues shown in blue and red, respectively. C, side chain conformations of all charged residues, Lys (K) in blue, Arg (R) in cyan, Asp (D) in red, and Glu (E) in pink, of 15 NMR structures of hRNase7 are displayed as a 180°rotation. The figure was generated based on the superposition of the backbone atoms in the full-length protein, and clarification of only one ribbon structure is shown. The charged residues that were mutated for antimicrobial activity studies are boxed with dotted lines. FEBRUARY 16, 2007 • VOLUME 282 • NUMBER 7 critical for the bactericidal activity, whereas the other two clusters, K 32 ,K 35 , and K 96 ,R 97 ,K 100 , are not (Figs. 6A and 8B). The result of Fig. 6C shows that the side chains of the Lys 1 , Lys 3 , Lys 111 , and Lys 112 residues in the first cluster are more flexible than those of cationic residues in other clusters. In addition, we also found that no anionic residue resides in the first cluster of hRNase7 near the K 1 ,K 3 ,K 111 ,K 112 residues by NMR studies (Fig. 6) and no anionic residue resides at each N-terminal region of bactericidal RNases by the comparison of amino acid sequences among the RNaseA superfamily members (Fig. 9). Thus, it is concluded that clustering of cationic residues without an adjacent anionic residue is critical for the bactericidal activity of hRNase.

Residues Responsible for Antimicrobial Activity of hRNase7
Several members in the RNase superfamily have antimicrobial activities, but the key residues/domains responsible for antimicrobial activity were different in human RNase3 and RNase7 (6 -10). The content of cationic residues of hRNase3 (Lys 1 and Arg 19 ) differs from that of hRNase7 (Lys 18 and Arg 4 ). Furthermore, the cationic residues Arg 101 and Arg 104 and aromatic residues Trp 10 and Trp 35 of hRNase3, which are suggested to be responsible for membrane binding and disruption, reside on the dispersed secondary structure (␣1, ␤4) (26), whereas the cationic residues K 1 ,K 3 ,K 111 ,K 112 , of hRNase7 reside on the flexible coil and loop at the N-terminal cluster. With regard to chicken RNaseA2, the Arg residues in domains II (residues 71-76) and III (residues 89 -104) are critical for the bactericidal activity, but the structures of the RNaseA2 remains unknown (10).
In addition to the flexibility of the side chains of cationic residues, we were also interested in the influence of similar tertiary structure folds for RNases on the antimicrobial activity.
Our results indicate that the bactericidal activity of RNase was not correlated with its backbone tertiary structure; for example, the root mean square deviation values between hRNase7 and bactericidal hRNase3 and hRNase5 are 2.57 and 3.37 Å, respectively, whereas those between bactericidal hRNase7 and nonbactericidal hRNase2, hRNase4, and bovine RNaseA are 2.34, 3.09, and 2.33 Å, respectively. This indicates that the tertiary structure fold of the whole RNase backbone is not critical for the antimicrobial activity.
The hRNase7 was effective in bactericidal activity at 4°C, whereas the indolicidin, a Trp-rich oligopeptide (ILPWKWW-PWWPWRR-NH 2 ) from bovine neutrophils, was not (27). The bactericidal RNases hRNase7 and RC-RNase6 from bullfrog  oocytes were bound to susceptible bacteria, whereas the nonbactericidal RNases were not. These results suggest that some component(s) on the bacteria is/are responsible for the binding of hRNase7 and the action mechanism of hRNase7 is energyindependent (28,29). The bacterial membrane became permeable to the DNA binding dye SYTOX Green just a few minutes after the addition of hRNase7 (Fig. 3). This indicates that the marked increase of membrane permeability is an important step for the bactericidal activity of hRNase7. Thus, we suggest that the hRNase7 may bind to the negatively charged components of the bacterial membrane through the flexible and cationic residues. The hRNase7 may change its own conformation, incorporates itself into the bacterial membrane through the hydrophobic scaffold, and triggers the disruption of membrane. Alternatively, we also propose that several pores, which are composed of hRNase7-binding proteins, may reside on the bacterial surface for the regulation of the transportation of ions and metabolites. The opening of pores may be triggered by hRNase7 binding through the flexible/clustered cationic residues and hydrophobic scaffold.
The efflux of ions and fatal depolarization of bacterial membrane may thus cause immediate cell death, even at low temperatures. The antibacterial mechanisms of these positively charged peptides/proteins by physical disruption of bacterial membrane is different from those for conventional antibiotics, which are inhibition of cell wall synthesis, DNA replication, RNA transcription, or protein synthesis. Due to the unique bactericidal activity of hRNase7, it has the potential to be a new therapeutic agent for bacterial infection, because it may not face the rapid emergence of drug resistance. The induction of endogenous hRNase7 gene expression or the administration of a synthetic oligopeptide designed from the hRNase7 structure would be possible in the clinical therapy for microbial infection.