Báo cáo khoa học: Comparison of human RNase 3 and RNase 7 bactericidal action at the Gram-negative and Gram-positive bacterial cell wall

Comparison of human RNase 3 and RNase 7 bactericidal action at the Gram-negative and Gram-positive bacterial cell wall Marc Torrent, Marina Badia, Mohammed Moussaoui, Daniel Sanchez, M. Victo` ria Nogue´ s and Ester Boix

Departament de Bioquı´mica i Biologia Molecular, Facultat Biocie` ncies, Universitat Auto` noma de Barcelona, Cerdanyola del Valle` s, Spain

Keywords antimicrobial proteins; cell wall; ECP; immunity; RNase 7

Correspondence E. Boix, Departament de Bioquı´mica i Biologia Molecular, Facultat de Biocie` ncies, Universitat Auto` noma de Barcelona, 08193 Cerdanyola del Valle` s, Spain Fax: +34 93 5811264 Tel: +34 93 5814147 E-mail: ester.boix@uab.cat

(Received 19 November 2009, revised 25 January 2010, accepted 27 January 2010)

doi:10.1111/j.1742-4658.2010.07595.x

The eosinophil cationic protein ⁄ RNase 3 and the skin-derived RNase 7 are two human antimicrobial RNases involved in host innate immunity. Both belong to the RNase A superfamily and share a high cationicity and a common structural architecture. However, they present significant diver- gence at their primary structures, displaying either a high number of Arg or Lys residues, respectively. Previous comparative studies with a mem- brane model revealed two distinct mechanisms of action for lipid bilayer disruption. We have now compared their bactericidal activity, identifying some features that confer specificity at the bacterial cell wall level. RNase 3 displays a specific Escherichia coli cell agglutination activity, which is not shared by RNase 7. The RNase 3 agglutination process precedes the bacte- rial death and lysis event. In turn, RNase 7 can trigger the release of bacterial cell content without inducing any cell aggregation process. We hypothesize that the RNase 3 agglutination activity may depend on its high affinity for lipopolysaccharides and the presence of an N-terminal hydro- phobic patch, and thus could facilitate host clearance activity at the infec- tion focus by phagocytic cells. The present study suggests that the membrane disruption abilities do not solely explain the protein bacterial target preferences and highlights the key role of antimicrobial action at the bacterial cell wall level. An understanding of the interaction between anti- microbial proteins and their target at the bacterial envelope should aid in the design of alternative peptide-derived antibiotics.

Introduction

Abbreviations CFU, colony-forming unit; ECP, eosinophil cationic protein; MAC, minimal agglutination concentration; PGN, peptidoglycan; SEM, scanning electron microscopy.

FEBS Journal 277 (2010) 1713–1725 ª 2010 The Authors Journal compilation ª 2010 FEBS

1713

Human antimicrobial RNase 3 and RNase 7 are mem- bers of the RNase A superfamily that participate in the host immune response against pathogen infection. RNase 3 was first identified as an eosinophil secretion product and named as eosinophil cationic protein (ECP). ECP is secreted by activated eosinophils during inflammation and its levels in biological fluids are con- sidered to be a marker for the diagnosis and monitor- ing of allergy and eosinophilia disorders [1]. Recently, it was reported that eosinophils can mediate their anti- bacterial effect through the release of cationic granule proteins [2]. RNase 7 was first reported as a skin antimicrobial protein [3] and is considered to be one of the main components of the innate immunity first-line protection against infections at the epithelial level [4,5]. tissues, RNase 7 is expressed in several epithelial

M. Torrent et al.

RNase 3 and RNase 7 bactericidal activity

studies indicated the

[15] and some surface lysine

that both RNases show a high cationicity, they share approximately 40% amino acid identity; careful inspec- tion reveals a distinct evolutionary pressure that leads to the accumulation of an unusual number of either Arg (18 Arg out of 133 amino acids) or Lys (18 Lys out of 128 amino acids) at the mature protein (Fig. 1). involvement of Mutagenesis for positive and aromatic surface-exposed residues clusters RNase 3 for RNase 7 [9]. On the other hand, a binding domain for heparin in RNase 3 [16] may account for its high affinity for heterosaccharides at the bacterial cell wall. Indeed, recent studies using RNase 3-derived peptides revealed a key domain at the protein N-terminus, which retained most of the protein bactericidal activity and a considerable LPS binding capacity [17]. More- over, screening of the RNase 3 N-terminal sequence predicts a hydrophobic aggregation patch [9] and an antimicrobial prone sequence [18].

represent a distinctive

RNase 3

RNase 7

including skin, gut and the respiratory and genitouri- nary tracts, and its expression can be induced by inflammatory agents and bacterial infection [6]. Both RNases display a wide range anti-pathogen activity, with toxicity being reported against viruses, bacteria, fungi, protozoans and, in the case of RNase 3, even helminthic parasites [7]. Although both proteins belong to the RNase A superfamily and have conserved their catalytic RNase activity [3,8], studies indicate that their antimicrobial mechanism of action is strongly depen- dent on their membrane destabilizing mechanism of action [9,10]. The RNase A superfamily includes other members with antimicrobial properties [7] and recent evolution studies suggest that the family may have started with an ancestral antipathogen physiological function [11,12]. Previous experimental data on both RNases, using lipid vesicles as a membrane model, revealed that the lipid bilayer disruption event takes place with a distinct timing [10,13]. However, the data obtained also indicate that mechanic action at the cytoplasmic membrane does not solely explain the pro- tein bactericidal properties. Therefore, we also charac- terized RNase 3 activity at the surface of bacteria, identifying significant differences with respect to its action on both Gram-negative and Gram-positive strains. A key distinctive feature of RNase 3 is its high affinity for lipopolysaccharides (LPS) and Escherichia coli cell agglutination activity [14]. Despite the fact We have now compared the activity of both RNases at the bacterial cell wall level. Although RNase 7 dis- plays remarkable affinity for peptidoglycan (PGN) and LPS at the Gram-positive and Gram-negative outer surface, the very high LPS binding and cell agglutina- tion activities feature of RNase 3. By contrast, RNase 7 displays a high leakage activity and a high capacity for binding PGN. The comparison of both antimicrobial RNases conducted

A

B

C

Fig. 1. (A) Ribbon representation of the 3D structures of RNase 3 (1DYT.pdb) [43] and RNase 7 (2HKY.pdb) [9]. Molecules are coloured from the N- to C-terminus. The active site is marked by a circle. (B) Molecular surface representation of RNase 3 and RNase 7. Hydrophobic residues are labelled in grey, cationic residues in blue, anionic residues in red, cysteine residues in yellow, proline residues in orange and noncharged polar residues in cyan. (C) Sequence alignment of RNase 3 and RNase 7 primary sequences. Secondary structure elements of RNase 3 are depicted at the top. The sequence alignment was performed using ESPRIPT software (http://espript.ibcp.fr/ESPript/ESPript/) and molecular representations were drawn using PYMOL (DeLano Scientific, Palo Alto, CA, USA, http://www.pymol.org ⁄ ).

FEBS Journal 277 (2010) 1713–1725 ª 2010 The Authors Journal compilation ª 2010 FEBS

1714

M. Torrent et al.

RNase 3 and RNase 7 bactericidal activity

in the present study therefore contributes towards elu- cidating the main determinants of their distinct poten- tial in vivo anti-pathogen properties.

Results

Studies on the bacterial cell viability

Fig. 2. Remaining CFUs after exposure of bacterial cultures to (A) E. coli and (B) S. aureus. The response is registered as a function of the protein concentration. RNase 3 (triangles) and RNase 7 (squares) were dissolved in 10 mM sodium phosphate (Na2HPO4 ⁄ NaH2PO4) buffer, pH 7.5, and serially diluted from 10 lM to 0.2 lM. In each assay, protein solutions were added to each dilution of bacteria, incubated for 4 h, plated in Petri dishes and the colonies counted after overnight incubation.

the bactericidal process

We have compared the RNase 3 and RNase 7 antimi- crobial activities with respect to E. coli and Staphylococcus aureus cells, which are representative Gram-negative and Gram-positive strains. Both proteins display com- parable activity, as indicated by the reduction of col- ony-forming units (CFUs) as a function of protein concentration (Fig. 2). On the other hand, kinetic pro- files of bacterial viability show a similar overall pattern, although there were significant differences in the respec- tive activities for the two tested strains. The bactericidal activity profiles were monitored by staining of bacteria with a Live ⁄ Dead kit (BacLight(cid:2); Molecular Probes, Carlsbad, CA, USA), using syto 9 and propidium iodide to determine bacterial viability. Although syto 9 dye can cross the cytoplasmic membrane and label all bacterial cells, propidium iodide can only access the content of membrane damaged cells, competing and displacing the bound syto 9. Therefore, the integration of syto 9 and propidium iodide fluorescence provides an estimate of the percentage viability for monitoring the kinetics of (Fig. 3). Although RNase 7 shows a similar live ⁄ dead progres- sion for both studied bacterial species, RNase 3 is sig- nificantly more active on the E. coli population, as reflected by the ED50 values (Fig. 3 and Table 1). The relative percentage survival, as evaluated by the viability assay, also correlated with the reduction in the percentage of remaining CFUs (Table 1).

To determine the morphological changes in bacterial cell population upon incubation with both RNase 3 and 7, the process was also visualized using confocal microscopy, where live ⁄ dead cells are also labelled with the syto 9 and propidium iodide dyes, respectively.

structures formed (after 10–20 min of incubation) are only stained by syto 9, indicating that these filaments are formed by live bacteria. From 30 min onward, the bacterial population stained by propidium iodide is rapidly increased. Subsequently, the aggregates begin to bind propidium iodide and recruit new dead clusters of bacteria (Video S1). For S. aureus, this aggregation mechanism cannot be observed and only an increase in the propidium iodide-stained bacteria is detected. Although some small clusters of bacteria can be observed, they are not comparable to the aggregates obtained in the case of E. coli. For RNase 7, aggrega- tion is neither observed in E. coli, nor in S. aureus (Figs 4 and S2).

FEBS Journal 277 (2010) 1713–1725 ª 2010 The Authors Journal compilation ª 2010 FEBS

1715

A careful inspection on the culture population reveals how confocal microscopy behaviour by RNase 3 aggregates E. coli cells, and how bacterial cell death is a later event in relation to the aggregation process (Fig. 4). By comparison, RNase 3-treated S. aureus cells display a distinct behaviour, where bac- terial death takes place at only a slightly lower rate but without a significant aggregation pattern (Fig. S1). Therefore, we conclude that the results obtained for RNase 3 indicate that the key bactericidal events take place at different times. First, we observe an enlarge- ment on the filaments formed by E. coli cells. The To quantify the bacterial aggregation ability, the minimal agglutination concentration (MAC) was calcu- lated, with an estimated value of 1.5 lm for RNase 3 activity with E. coli cells, whereas no agglutination

M. Torrent et al.

RNase 3 and RNase 7 bactericidal activity

A

B

10 min

C

R N a s e 3

60 min

D

120 min

E

0 min

F

R N a s e 7

120 min

Fig. 4. Study of E. coli viability and population morphology visual- ized by confocal microscopy. E. coli cells (A) before protein addi- tion; (B–D) after 5 lM of RNase 3 at 10 min (B), 1 h (C) and 2 h (D); and (E, F) after adding 5 lM of RNase 7 at 0 and 2 h, respectively. Bacterial cells were stained using a 1 : 1 syto 9 ⁄ propidium iodide mixture. The left-hand panels correspond to the propidium iodide- stained cells (dead cells), excited using an orange diode. The cen- tral panels correspond to the syto 9-stained cells (live cells), excited using a 488 nm argon laser. The right-hand panels correspond to the overlay of both signals. Scale bar = 50 lm.

Fig. 3. Study of bacterial viability kinetics for (A) RNase 3 and (B) RNase 7. Cell viability for Gram-positive S. aureus (filled squares) and Gram-negative E. coli (filled circles) was analysed using syto 9 (for live bacteria) and propidium iodide (for dead bacteria). An aliquot of 1 mL of exponential phase cells was incubated with 5 lM of each protein. Duplicates were performed for each condition.

show that RNase 7 lacks the ability to agglutinate bacteria but retains bactericidal activity. To better understand the

Table 1. Kinetic analysis on the antimicrobial activity of RNases 3 and 7 using the Live ⁄ Dead bacterial viability kit as described in the Materi- als and methods. One millilitre of exponential phase cells was incubated with 5 lM of protein during a total period of 150 min. ED50 (mea- sured as the time needed to achieve a 50% decrease in live bacteria) and percentage survival were calculated by exponential fitting to the data presented in Fig. 3. The percentage of remaining CFUs is also indicated for each condition. Values are the average of three replicates.

S. aureus

Protein

E. coli

Remaining CFUs (%)

RNase 3 RNase 7

ED50 (min) 35 ± 1 56 ± 4

Survival (%) 12.0 ± 0.8 17 ± 2

7 ± 4* 13 ± 3*

Survival (%) 20.1 ± 0.8 13 ± 1

Remaining CFUs (%) 22 ± 3* 14 ± 2*

ED50 (min) 60 ± 2 56 ± 2

*P < 0.05 (Student’s t-test).

FEBS Journal 277 (2010) 1713–1725 ª 2010 The Authors Journal compilation ª 2010 FEBS

1716

activity was detected in the presence of S. aureus cells, nor for RNase 7 with the two tested strains, even with a 10 lm protein concentration. The results obtained correlation between aggregation and bacterial leakage, the release of cell content was monitored using activity staining gels (Fig. 5). With this technique, the endogenous bacterial

M. Torrent et al.

RNase 3 and RNase 7 bactericidal activity

A

B

Fig. 5. Record of bacterial lysis process by the detection of the release of endogeneous bacterial RNase by activity staining gel. (A) The clearance area corresponding to the bacterial RNase substrate degradation is indicated. The intensity of the areas showing substrate degradation was analysed by densitometry as described in the Materials and methods. The intensity values are referred to the 0 h incubation density area. The bacterial lysis activity of RNase 3 (filled symbols) and RNase 7 (empty symbols) on both E. coli (triangles) and S. aureus (squares) is shown. (B) Polycytidylic acid SDS-PAGE (15%) activity staining gel from the time course of E. coli cell incubation with RNase 3. Left lanes: control cells; right lanes: cells incubated with 5 lM of RNase 3 at 0, 1, 2, 3 and 4 h.

their action at pendent of EDTA chelation. This is not applicable to RNase 7, which has a lower membrane depolarization activity without EDTA treatment. On the other hand, RNase 7 appears to alter more easily the S. aureus cytoplasmic membrane than RNase 3. The distinct abilities of both RNases to access and alter the cyto- plasmic membrane may reflect the outer envelope level. results demonstrate that,

Studies at the bacterial cell wall

ribonuclease released upon membrane leakage can be detected and the leakage kinetics can be monitored. The bacterial cells were incubated with 5 lm of each RNase and aliquots were taken at 1-h intervals. For RNase 3, an important difference between E. coli and S. aureus is found. Whereas leakage in E. coli cells can be observed as soon as after 1 h of incubation, no release is detected for S. aureus, not even after 4 h of incubation. These even though RNase 3 is able to kill 80% of S. aureus cells after 4 h of incubation, the damage at the membrane level is insufficient to allow the release of a detectable amount of endogenous ribonucleases.

In the case of RNase 7, both E. coli and S. aureus endogenous RNases are released (Fig. 5). Nevertheless, RNase 7 leakage in S. aureus cells appears to be trig- gered later than in E. coli cells. The activity corre- sponding to the endogenous ribonucleases that are released by the bacteria is only registered after 2 h of incubation.

The bactericidal activity of both RNases is precluded by the protein binding to the cells. Proteins incubated with both E. coli and S. aureus cultures are recovered in the cell pellet fraction (Fig. S3). To gain insight on the bactericidal properties of both RNases, binding studies on different elements of the bacterial cell wall were carried out. Binding to PGN and LPS has already been studied in detail for RNase 3 [14]. The results obtained are now compared with RNase 7 binding affinities. The new data (Figs 6 and 7) indicate that RNase 7 can also interact with both Gram- and Gram-positive heteropolysaccharides. negative Affinity binding studies on LPS and PGN were com- plemented with scanning electron microscopy (SEM) microscopy to visualize the structural damage induced by the protein–cell wall interactions (Fig. 8).

FEBS Journal 277 (2010) 1713–1725 ª 2010 The Authors Journal compilation ª 2010 FEBS

1717

Finally, membrane depolarizing activity was also studied using the DiSC3(5) marker (Table S1). The results obtained show that RNase 3 is able to depo- larize E. coli cells more rapidly than S. aureus cells. When comparing membrane depolarization activities, we can observe that ECP easily accesses the Gram- negative cytoplasmic membrane, without any EDTA treatment being necessary to destabilize the cell outer membrane. RNase 3 activity on E. coli cells is inde- Binding to LPS was assessed using the Bodipy TR cadaverine marker (Invitrogen, Carlsbad, CA, USA).

M. Torrent et al.

RNase 3 and RNase 7 bactericidal activity

similar to that for polymyxin B, a powerful LPS binder, which was selected as a positive control (Fig. 6).

Fig. 6. Displacement of LPS-bound Bodipy TR cadaverine by RNase 7 (triangles), RNase 3 (circles) and polymyxin B (squares); [LPS]: 10 lgÆmL)1; [BODIPY TR Cadaverine]: 10 lM in 5 mM He- pes-KOH (pH 7.5).

We also assessed and compared RNase 7 binding to PGN, the main component of Gram-positive bacteria, with our previous results obtained for RNase 3 [14]. Microfluidic gel electrophoresis showed that, after RNase 7 incubation in the presence of S. aureus PGN, most of the protein sample is recovered together with the insoluble PGN fraction, as also observed for lyso- zyme, the positive control, and previously for RNase 3 [14]. A slight anomalous displacement in the virtual gel is observed for RNase 7, with a higher apparent molecular weight, as a result of its cationic nature. This behaviour is frequently observed for RNase A family members. By contrast, BSA, the negative con- trol, does not bind to the PGN fraction and is fully recovered in the supernatant fraction (Fig. 7A).

Moreover, a PGN binding assay using Alexa fluoro- phor-labelled RNase 7 also indicates a high binding affinity. A Kd value of 2 · 10)8 m was determined using the Scatchard plot as shown in Fig. 7B, which is a value considerably higher than that calculated for RNase 3 (2 · 10)7 m) [14].

SEM data were previously shown to be useful for surface damage upon RNase 3 assessing bacterial The results obtained show that RNase 3 is able to bind with higher affinity to LPS compared to RNase 7. In any case, RNase 7 still retains a high LPS binding affin- ity because it displays an effective displacement activity

A

100.0 75.0 50.0 37.0 25.0 20.0

B

Fig. 7. (A) Analysis by a microfluidic electrophoresis system of the binding of RNase 7 to PGN. Lysozyme and BSA were taken as positive and negative controls, respectively, for PGN binding. Molecular mass markers are indicated on the left. For each protein, the first lane corre- sponds to pellet (P) and the second lane to the supernatant fractions (S). PGN were incubated with each protein and the soluble and insolu- ble fractions were collected as described in the Materials and methods. Supernatant represents the soluble fraction, which contains the unbounded protein, whereas the pellet represents the insoluble fraction containing the PGN bound protein. (B) Scatchard plot and the corre- sponding binding curve of RNase 7 interaction with PGN. RNase 7 labelled with the fluorophor Alexa Fluor 488 at a concentration in the range 0.01–100 nM was incubated in the presence of 0.02 lg PGN in 200 lL of 5 mM Hepes-KOH (pH 7.5) and the free and bound fractions were quantified.

FEBS Journal 277 (2010) 1713–1725 ª 2010 The Authors Journal compilation ª 2010 FEBS

1718

M. Torrent et al.

RNase 3 and RNase 7 bactericidal activity

E. coli

S. aureus

Fig. 8. Scanning electron micrographs of E. coli and S. aureus incubated in the absence (top) and presence (bottom) of 4 lM RNase 7 for 4 h. The magnification scale is indicated at the bottom of each micrograph.

the mechanism of action of both RNases on model membranes [10,13]. RNase 7 has no significant mem- brane aggregation capacity compared to RNase 3, although it displays a much higher leakage capacity. On the other hand, initial studies on RNase 3 by site- directed mutagenesis indicated that the membrane dis- ruption ability could not solely explain the protein bactericidal properties [15]. Indeed, strain selectivity was reported for RNase 7 [3,9].

treatment [14], where severe damage on E. coli cells and the ability of protein to trigger cell population agglutination was reported. Accordingly, SEM was used to visualize changes in bacterial cell cultures upon incubation with RNase 7. The addition of RNase 7 at a final concentration of 4 lm is unable to induce either E. coli or S. aureus cell culture aggregation and all cells retain their characteristic morphology. Neverthe- less, several blebs can be observed on the bacterial cell surface in both E. coli and S. aureus, suggesting that local cell surface disturbance is taking place (Fig. 8).

Discussion

We have now analysed the time course profile of bacterial cell viability for both RNases (Fig. 3). The rapid decay during the first 30 min may reflect a rapid direct lytic process. We can differentiate between an initial active exponential growth phase, where the pro- tein may have easy access to the cell membrane during duplication, and a later stage, where protein action at the wall envelope may acquire a critical role. On the other hand, the viability assay, performed at a salt levels, rejects a concentration close to physiological mere unspecific electrostatic interaction and provides further corroboration for both proteins retaining their properties in vivo and being regarded as effective anti- microbial agents. As noted by Hancock and Sahl [23], many cationic peptides with few hydrophobic residues at crucial positions are prone to having some antimi- crobial activity at low ionic strength, although the term ‘antimicrobial’ should only be reserved for those that are able to kill microbes under physiological conditions.

the bacterial wall

FEBS Journal 277 (2010) 1713–1725 ª 2010 The Authors Journal compilation ª 2010 FEBS

1719

RNases 3 and 7 are the main representatives of the cytotoxic antimicrobial members of the RNase A superfamily. Both are cationic proteins with a high pI, and display a broad antimicrobial action against Gram-positive and Gram-negative strains [6,19–21]. The two RNases present, respectively, a high number of either Arg or Lys surface-exposed residues (Fig. 1) that may contribute to their distinct bactericidal mech- anisms of action. Previous work revealed that the RNase bactericidal mechanism was not dependent on its RNase enzymatic activity but on direct membrane disruptive action [9,10,15,22]. The contribution of bac- terial wall determinants was also suggested [15] and recent studies on RNase 3 indicated a high affinity for bacterial heterosaccharides [14]. Indeed, the present comparative characterization of both the action of RNase 3 and RNase 7 at level revealed some particular features that could explain their distinct abilities with respect to Gram-negative and Gram-positive strains. We previously compared The results obtained in the present study reveal dis- tinct behaviours not only on lipid bilayers, but also at the bacterial cell wall. In both strains, E. coli and S. aureus, RNase 7 displays a restricted disturbance causing local blebs, whereas no agglutination is

M. Torrent et al.

RNase 3 and RNase 7 bactericidal activity

observed (Fig. 8). These observations are much differ- ent from those observed in the case of RNase 3, where global cell damage has been observed in E. coli cells after complete bacterial agglutination [14].

Interestingly, the in vivo record of the RNase 3 trea- ted E. coli culture assessed by confocal microscopy illustrates how the cells first aggregate but still retain cytoplasmic membrane. Cell death, as an intact is then a observed by the propidium iodide uptake, later event (Fig. 4 and Video S1). We have further analysed RNase 3 bacterial agglutination activity and estimated a MAC of 1.5 lm on E. coli cell cultures. Cell agglutination comprises a characteristic feature that also is reported for other antimicrobial peptides [24] and proteins, as lectin RNases, which are amphi- bian members of the RNase A superfamily with a particular ability for binding heterosaccharides [25].

reported for other antimicrobial proteins and peptides [26,27]. If we compare the sequences and 3D structures available for both RNases (Fig. 1), we can identify some of the features that may account for the specific ability of RNase 3 to aggregate both lipid vesicles and bacterial cells. Scanning of both RNases with aggreg- scan software [28] reveals a distinct aggregation pro- file, in particular at the N-terminal zone. In the case of RNase 3, we can observe a hydrophobic patch in one side of the molecule, surrounded by polar residues. Indeed, a hydrophobic patch at the RNase 3 N-termi- nus that retains most of the protein antimicrobial activity, and may be responsible for the protein vesicle aggregation ability, was recently characterized by syn- thetic-derived peptides in our laboratory [17]. Bacteria agglutinating efficiency was also correlated with the presence of hydrophobic patches for de novo designed antimicrobial peptides. In the case of RNase 7, no hydrophobic patches on the protein surface can be observed. The protein cationicity, as a result of the high number of lysines present in the structure, is dis- tributed uniformly on the protein surface. The absence of hydrophobic patches may be responsible for the lack of agglutinating capacity of RNase 7.

the primary sequences

Although both RNases contain a high number of cationic residues, the bias on either Arg or Lys con- tent (18 Arg for RNase 3 and 18 Lys for RNase 7) suggests that the cationicity of both proteins has been acquired independently during their evolution. A comparison with other RNase A family members indicates that most Lys residues are retained in the includes RNases 6, 7 RNase A lineage group that and 8 [29]. Phylogenetic studies suggest the recent divergence of RNase 7 and RNase 8 as a result of a duplication event [29]. However, no homologues were identified in rodents [12] as described for the RNase2 ⁄ RNase 3 group, where members with antimi- crobial activity were reported in both rat and mouse. In turn, RNase 3 acquired many Arg residues during its divergence from RNase 2 [12,29]. However, a com- parison of antimicrobial RNases suggests that local positive clusters, rather than their overall pI, are key for protein bactericidal activities [30,31]. For example, a comparison of for fish, chicken and human antimicrobial RNases revealed a distinct Lys ⁄ Arg ratio but a similar total number of positive residues [30].

In turn, RNase 7 could follow another bacterial pro- cess. The ability to induce the bacterial cell content, as assayed by activity-staining gel analysis, has shown in S. aureus, RNase 7 presents an important that, leakage activity, whereas no significant activity is detected for RNase 3 at the assayed conditions (Fig. 5). This fact may be explained by the higher capacity of RNase 7 to cause leakage of membranes at low concentrations. These effects are in good agree- ment with the results observed in model membranes, where RNase 7 is able to trigger leakage at a lower protein : lipid ratio before any aggregation event takes place, suggesting a local membrane disturbance process [10]. Moreover, the higher binding affinity for PGN displayed by RNase 7 may also partially account for the higher membrane depolarization activity observed against the S. aureus strain (Table S1). RNase 7 was previously reported to display a particularly high bac- the Gram-positive Enterococ- tericidal activity for cus faecium [3]. Our membrane depolarizing assays confirm a distinct mechanism of action for both RNases on each of the two tested strains. Mainly for Gram-negative cells, RNase 3 does not require EDTA pretreatment. EDTA pretreatment would sequester the divalent cations that hold LPS together and secure the outer membrane structure. The higher affinity of RNase 3 for LPS (Fig. 6) could by itself facilitate outer membrane disturbance and access to the cyto- plasmic membrane. RNase 7 displays a similar capac- ity for depolarizing cell membranes, as observed in RNase 3, when E. coli cells are pretreated with EDTA, thus suggesting that the main differences may be restricted to the bacterial outer barrier.

FEBS Journal 277 (2010) 1713–1725 ª 2010 The Authors Journal compilation ª 2010 FEBS

1720

These results confirm that the capacity to bind bac- terial cell wall structures is of special importance for the antimicrobial properties of both RNases, as also On the other hand, arginine residues are implied in carbohydrate binding proteins because they display hydrogen bonding between the guanidinium group and sulphates or phosphates [32,33]. This fact may explain the higher binding affinity of RNase 3 for LPS (Fig. 6).

M. Torrent et al.

RNase 3 and RNase 7 bactericidal activity

WI, USA) and S. aureus 502 A (ATCC, Rockville, MD, USA) strains were used. PD-10 columns were purchased from GE Healthcare (Milwaukee, WI, USA).

Wild-type RNase 3 was expressed using a synthetic gene for human coding sequence. RNase 7 was expressed start- ing from a cDNA subcloned in the pET11c plasmid vector. Protein expression in E. coli BL21(DE3) strain, folding of the protein from inclusion bodies, and the purification steps, were carried out as described previously [8,10].

tissues, especially the Expression and purification of recombinant RNase 3 and RNase 7

Fluorescent labelling of proteins

RNases were labelled with the Alexa Fluor 488 fluorophor, in accordance with the manufacturer’s instructions. To 0.5 mL of a 2 mgÆmL)1 protein solution in NaCl ⁄ Pi, 50 lL of 1 m sodium bicarbonate (pH 8.3) was added. The pro- tein was incubated for 1 h at room temperature with the reactive dye, with stirring, in accordance with the manufac- turer’s instructions. The labelled protein was separated from the free dye by a PD-10 desalting column.

The tissue distribution of both RNases also suggests some functional differences. Whereas RNase 3 is mostly present in eosinophils and, to a less extent, in other cells of the immune system (e.g. neutrophils and basophils) [34,35], RNase 7 is expressed in multiple it somatic is skin, where described as a major antimicrobial agent [3,6]. Although both RNases are secreted, they may respond to distinct challenges. RNase 3 is stored in secretion granules and is depleted at the site of inflammation where these cells are recruited [36]. RNase 7 represents one of the major contributors to the antimicrobial activity involved in first-line host defence at the human skin barrier [37]. In the skin, basal RNase 7 secretion is detected but mRNA overexpression is observed as a result of bacterial challenge [37]. A correlation between a dysfunction in antimicrobial protein expression at the skin level during dermatitis and a predisposition to skin infections also highlights their contribution to a host defence role [38–40].

Antibacterial activity

In conclusion, in the present study, we have shown that RNase 3 and RNase 7 have particular antimicro- bial activities that are modulated by their action at the bacterial cell wall. We observed that RNase 7 displays a mechanism based on local membrane disturbance, in contrast to RNase 3 that demonstrated global action. Accordingly, we have shown that RNase 3 displays an E. coli agglutinating activity (not shared by RNase 7), which would probably be dependent on both the pres- ence of a hydrophobic patch and the capacity of the protein to bind LPS.

Antimicrobial activity was calculated by assessing the num- ber of CFUs as a function of protein concentration. Values were averaged from two independent experiments per- formed in triplicate for each protein concentration. Proteins were dissolved in 10 mm sodium phosphate (Na2HPO4 ⁄ - NaH2PO4) buffer (pH 7.5) and serially diluted from 10 lm to 0.2 lm. Bacteria were incubated at 37 (cid:3)C overnight in LB broth and diluted to give approximately 5 · 105 CFUÆmL)1. In each assay, protein solutions were added to each dilution of bacteria, incubated for 4 h, and samples were plated on Petri dishes and incubated at 37 (cid:3)C over- night. The number of CFUs in each Petri dish was counted and the average values were represented in a semi-logarith- mic plot.

An understanding of the molecular mechanism that is responsible for the high binding affinity of antimi- crobial protein for unique heterosaccharide structures at the bacterial envelope would also contribute to the development of new peptide-derived antibiotics, which would overcome the increasing emergence of antibiotic resistant strains.

Materials and methods

Kinetics of bacterial survival were carried out using the Live ⁄ Dead bacterial viability kit in accordance with the manufacturer’s instructions. Bacteria were stained using a syto 9 ⁄ propidium iodide 1 : 1 mix as provided with the kit. E. coli and S. aureus cells were grown at 37 (cid:3)C to the mid- exponential phase (D600 = 0.4), centrifuged at 5000 g for 5 min and resuspended in a 0.75% NaCl solution in accor- dance with the manufacturer’s instructions. One millilitre of stained E. coli or S. aureus bacteria (D600 = 0.2) was mixed with 5 lm of RNase 3 or 7 and the fluorescence intensity

Bodipy TR cadaverine, BC [5-(((4-(4,4-difluoro-5-(2-thienyl)- 4-bora-3a,4a-diaza-s-indacene-3-yl)phe-noxy)acetyl)amino) pentylamine, hydrochloride], 3,3-dipropylthiacarbocyanine [DiSC3(5)], Gramicidin D, Alexa Fluor 488 protein label- ling kit and the Live ⁄ Dead bacterial viability kit were all purchased from Molecular Probes (Eugene, OR, USA). LPSs from E. coli serotype 0111:B4, Polymyxin B sulfate, PGN from S. aureus, polycytidylic acid and lysozyme from chicken egg white were purchased from Sigma-Aldrich (St Louis, MO, USA). E. coli BL21DE3 (Novagen, Madison,

FEBS Journal 277 (2010) 1713–1725 ª 2010 The Authors Journal compilation ª 2010 FEBS

1721

Bacterial viability Materials

M. Torrent et al.

RNase 3 and RNase 7 bactericidal activity

was continuously measured using a Cary Eclipse Spectroflu- orimeter (Varian Inc., Palo Alto, CA, USA). RNase A was used in all cases as a negative control. The excitation wave- length was 470 nm and the emission was recorded in the range 490–700 nm. To calculate bacterial viability, the sig- nal in the range 510–540 nm was integrated to obtain the syto 9 signal (live bacteria) and from 620–650 nm to obtain the propidium iodide signal (dead bacteria). Then, the per- centage of live bacteria was represented as a function of time. ED50 was calculated by fitting the data to a simple exponential decay function.

Protein affinity to PGN was calculated using a fluores- cence-based method, employing a microtitre plate as described previously [14]. Protein labelled with the fluoro- phor Alexa Fluor 488 was incubated with insoluble PGN. Proteins at different concentrations, in the range 1–100 nm, were incubated in the presence of 0.02 lg of peptidoglycans in a 5 mm Hepes buffer at pH 7.5 in a final volume of 200 lL. The reaction mixture was kept at 4 (cid:3)C for 2 h with gentle shaking. Next, the remaining soluble protein was removed from the insoluble PGN fraction by a centrifuga- tion step at 13 000 g for 30 min and quantified with Victor 3 (Perkin-Elmer, Boston, MA, USA).

Agglutination activity was evaluated by calculating the MAC. An aliquot of 5 mL of E. coli cells was grown at 37 (cid:3)C to the mid-exponential phase (D600 = 0.6), centri- fuged at 5000 g for 2 min and resuspended in Tris-HCl buffer, 0.15 m NaCl (pH 7.5) until D600 of 10 was reached. An aliquot of 200 lL of the bacterial suspension was incu- bated in microtitre plates with an increasing protein con- centration at 0.1 and 0.5 lm intervals up to 10 lm and left overnight at room temperature. The aggregation behaviour was observed by visual inspection and checked with a bin- ocular microscope at ·50 magnification. The agglutinating activity is expressed as the minimum agglutinating concen- tration of the sample tested, corresponding to the first con- dition where bacterial aggregates are visible by the naked eye, as described previously [41].

LPS binding was assessed using the fluorescent probe Bodipy TR cadaverine as described previously [14]. Briefly, the displacement assay was performed by the addition of 1–2 lL aliquots of a solution of Polymyxin B, RNase 3, RNase 7 or RNase A to 1 mL of a continuously stirred mixture of LPS (10 lgÆmL)1) and Bodipy TR cadaverine (10 lm) in 5 mm Hepes buffer at pH 7.5. Fluorescence measurements were performed on a Cary Eclipse spec- trofluorimeter. The BC excitation wavelength was 580 nm and the emission wavelength was 620 nm. The excitation slit was set at 2.5 nm and the emission slit was set at 20 nm. Final values correspond to an average of four repli- cates and were the mean of a 0.3 s continuous measure- ment. Quantitative effective displacement values (ED50) were calculated.

Agglutination activity Affinity binding assay for LPS

Supernatant

loading

RNase 3 was incubated at 5 lm with E. coli bacterial cells grown to the exponential phase (D600 = 0.6) in 1 mL of NaCl ⁄ Pi buffer at 37 (cid:3)C for 1 h. After centrifugation at 13 000 g, proteins from the pellet were extracted with electrophoresis fractions buffer. were lyophilized and dissolved in loading buffer. Samples were analysed by SDS-PAGE (15%) and Coomassie blue staining.

Protein binding to bacterial cells SEM

E. coli and S. aureus cell cultures of 1 mL were grown at 37 (cid:3)C to the mid-exponential phase (D600 = 0.4) and incu- bated with 4 lm RNase 3 or RNase 7 in NaCl ⁄ Pi at room temperature. Aliquots were taken up to 4 h of incubation and were prepared for analysis by SEM, as described previ- ously [14]. The cell suspensions were fixed with 2.5% gluter- aldehyde in 100 mm Na-cacodylate buffer (pH 7.4) for 2 h at room temperature. Next, the cells were pelleted, a drop of each suspension was transferred to a nucleopore filter, which was kept in a hydrated chamber for 30 min allowing the cells to adhere, and then washed to remove the gluteral- dehyde, and resuspended in the same 100 mm Na-cacody- late buffer at pH 7.4. Attached cells were post-fixed by immersing the filters in 1% osmium tetroxide in cacodylate buffer for 30 min, rinsed in the same buffer, and dehy- in ascending percentage concentrations drated in ethanol [31, 70, 90 (·2) and 100 (·2)] for 15 min each. The filters were mounted on aluminum stubs and coated with gold- palladium in a sputter (K550; Emitech, East coater Grinsted, UK). The filters were viewed at 15 kV accelerat- ing voltage in a Hitachi S-570 scanning electron microscope

Protein binding to PGN was first analysed by electrophore- sis as described previously [14]. PGN at 0.4 mgÆmL)1 in 10 mm Tris-HCl (pH 7.5) was incubated with the protein at a protein ⁄ PGN ratio of 1 : 20 (w ⁄ w). Samples were kept at 4 (cid:3)C for 2 h with gentle mixing and centrifuged at 13 000 g for 15 min to separate the soluble and insoluble fractions. Lysozyme and BSA were chosen as positive and negative controls, respectively. Samples were resuspended directly in the electrophoresis loading buffer and evaluated using an system Experion automated microfluidic electrophoresis (Bio-Rad, Hercules, CA, USA).

FEBS Journal 277 (2010) 1713–1725 ª 2010 The Authors Journal compilation ª 2010 FEBS

1722

Affinity binding assay for PGN

M. Torrent et al.

RNase 3 and RNase 7 bactericidal activity

(Hitachi, Tokyo, Japan) and a secondary electron image of cells for topography contrast was collected at several mag- nifications. A total of ten micrographs were collected at random for each condition, and the number of isolated cells and aggregates was registered.

recorded for each condition, and the time required to achieve half of total membrane depolarization was esti- mated from the nonlinear regression curve. E. coli cells were also incubated in the presence of EDTA, allowing loss of the LPS outer membrane surface layer, as previously described [14].

Activity-staining gels (zymograms) were selected to analyse bacterial leakage upon incubation with ribonucleases. E. coli and S. aureus cells were grown at 37 (cid:3)C to the mid- exponential phase (D600 = 0.4) in LB medium, centrifuged at 5000 g for 5 min, and resuspended in a 10 mm Na2HPO3 buffer at pH 7.2. Cells were incubated with 5 lm of RNase 3 or 7 and 10 lL aliquots were collected at 1, 2, 3 and 4 h. The aliquots collected were mixed with nonreduc- ing loading buffer (60 mm Tris-HCl, 10% glycerol, 0.015% bromophenol blue, 3% SDS, pH 6.8) and analysed for RNase activity by zymogram on SDS-PAGE (15%) con- taining 0.6 mgÆmL)1 of polycytidylic acid as substrate. After elimination of SDS by incubation with a pH 7.5 solu- tion consisting of 10 mm Tris-HCl and 20% isopropanol, the gels were incubated at 25 (cid:3)C for 90 min in 100 mm Tris-HCl (pH 7.5). The relative intensity of the areas show- ing substrate degradation was analysed by densitometry. leakage was assessed by monitoring, as a Bacterial cell function of time, the increase of the clearance area corre- sponding to polynucleotide cleavage by the released bacte- rial RNase.

Experiments were carried out in a plate-coverslide system. Five hundred microlitres of E. coli or S. aureus bacteria (D600 = 0.4) were mixed with 40 lL of 60 lm to achieve a final concentration of 5 lm of RNase 3 or 7, and images were immediately recorded. RNase A was used in all cases as a negative control. Bacteria were pre-stained using the syto 9 ⁄ propidium iodide 1 : 1 mix provided in the Live ⁄ Dead staining kit. Syto 9 is a DNA green fluorescent dye that diffuses thorough intact cell membranes and propi- dium iodide comprises a DNA red fluorescent dye that can only access the nucleic acids of membrane damaged cells, displacing the DNA bound syto 9. The method allows the labelling of intact viable cells and membrane compromised cells, which are labelled in green and red respectively, referred to as live and dead cells [42]. Confocal images of the bacteria were captured using a laser scanning confocal microscope (Leica TCS SP2 AOBS equipped with a HCX PL APO 63, ·1.4 oil immersion objective; Leica Microsys- tems, Wetzlar, Germany). Syto 9 was excited using a 488 nm argon laser (515–540 nm emission collected) and propidium iodide was excited using an orange diode (588– 715 nm emission collected). To record the time-lapse experi- ment, Life Data Mode software (Leica) was used, obtaining an image every 1 min in a experiment lasting 180 min.

Confocal microscopy Bacteria leakage analysis by activity-staining gels

Acknowledgements

Bacteria cytoplasmic membrane depolarization assay

Confocal microscopy and scanning electron micros- copy were performed at the Servei de Microsco` pia of the Universitat Auto` noma de Barcelona (UAB). We thank Mo` nica Rolda´ n and Helena Monto´ n for their technical support with confocal microscopy, and Fran- cisca Cardoso, Francesc Bohils and Alejandro Sa´ nchez for their assistance with the electron microscopy samples. Spectrofluorescence and densitometry assays were performed at the Laboratori d’Ana` lisi i Fotodoc- umentacio´ , UAB. The work was supported by the Ministerio de Educacio´ n y Cultura (grant numbers BFU2006-15543-C02-01 and BFU2009-09371) and by the Fundacio´ La Marato´ de TV3 (grant number TV3- 031110). M.T. was the recipient of a predoctoral fellowship from the Generalitat de Catalunya.

References

1 Ahlstedt S (1995) Clinical application of eosinophilic

cationic protein in asthma. Allergy Proc 16, 59–62.

Membrane depolarization was assayed by monitoring the DiSC3(5) fluorescence intensity change in response to changes in transmembrane potential as described previously [14]. E. coli and S. aureus cells were grown at 37 (cid:3)C to the mid-exponential phase and resuspended in 5 mm Hepes- KOH, 20 mm glucose and 100 mm KCl at pH 7.2 until D600 of 0.05 was reached. DiSC3(5) was added to a final concentration of 0.4 lm. Changes in the fluorescence because of the alteration of the cytoplasmic membrane potential were continuously monitored at 20 (cid:3)C at an exci- tation wavelength of 620 nm and an emission wavelength of 670 nm. When the dye uptake was maximal, as indicated by a stable reduction in the fluorescence as a result of quenching of the accumulated dye in the membrane inte- rior, protein in 5 mm Hepes-KOH buffer at pH 7.2 was tested protein concentration of 4 lm. added at a final Gramicidin D was used as control reference protein. All conditions were assayed in duplicate. The time required to reach a stabilized maximum fluorescence reading was

FEBS Journal 277 (2010) 1713–1725 ª 2010 The Authors Journal compilation ª 2010 FEBS

1723

M. Torrent et al.

RNase 3 and RNase 7 bactericidal activity

2 Linch SN, Kelly AM, Danielson ET, Pero R, Lee JJ &

Gold JA (2009) Mouse eosinophils possess potent antibacterial properties in vivo. Infect Immun 77, 4976–4982.

16 Fan TC, Fang SL, Hwang CS, Hsu CY, Lu XA, Hung SC, Lin SC & Chang MD (2008) Characterization of molecular interactions between eosinophil cationic pro- tein and heparin. J Biol Chem 283, 25468–25474.

3 Harder J & Schroder JM (2002) RNase 7, a novel

innate immune defense antimicrobial protein of healthy human skin. J Biol Chem 277, 46779–46784.

4 Schroder JM & Harder J (2006) Antimicrobial skin pep-

17 Torrent M, de la Torre BG, Nogues VM, Andreu D & Boix E (2009) Bactericidal and membrane disruption activities of the eosinophil cationic protein are largely retained in an N-terminal fragment. Biochem J 421, 425–434.

tides and proteins. Cell Mol Life Sci 63, 469–486.

18 Torrent M, Nogues MV & Boix E (2009) A theoretical approach to spot active regions in antimicrobial pro- teins. BMC Bioinformatics 10, 373–382.

5 Koten B, Simanski M, Glaser R, Podschun R, Schroder JM & Harder J (2009) RNase 7 contributes to the cuta- neous defense against Enterococcus faecium. PLoS ONE 4, e6424.

6 Zhang J, Dyer KD & Rosenberg HF (2003) Human

19 Boix E, Torrent M, Sanchez D & Nogues MV (2008) The antipathogen activities of eosinophil cationic pro- tein. Curr Pharm Biotechnol 9, 141–152.

20 Domachowske JB, Dyer KD, Adams AG, Leto TL &

RNase 7: a new cationic ribonuclease of the RNase A superfamily. Nucleic Acids Res 31, 602–607.

Rosenberg HF (1998) Eosinophil cationic pro- tein ⁄ RNase 3 is another RNase A-family ribonuclease with direct antiviral activity. Nucleic Acids Res 26, 3358–3363.

7 Boix E & Nogues MV (2007) Mammalian antimicrobial proteins and peptides: overview on the RNase A super- family members involved in innate host defence. Mol Biosyst 3, 317–335.

8 Boix E, Nikolovski Z, Moiseyev GP, Rosenberg HF,

21 Rosenberg HF (2008) RNase A ribonucleases and host defense: an evolving story. J Leukoc Biol 83, 1079–1087.

Cuchillo CM & Nogues MV (1999) Kinetic and product distribution analysis of human eosinophil cationic pro- tein indicates a subsite arrangement that favors exonu- clease-type activity. J Biol Chem 274, 15605–15614. 9 Huang YC, Lin YM, Chang TW, Wu SJ, Lee YS,

22 Rosenberg HF (1995) Recombinant human eosinophil cationic protein. Ribonuclease activity is not essential for cytotoxicity. J Biol Chem 270, 7876–7881. 23 Hancock RE & Sahl HG (2006) Antimicrobial and

host-defense peptides as new anti-infective therapeutic strategies. Nat Biotechnol 24, 1551–1557.

Chang MD, Chen C, Wu SH & Liao YD (2007) The flexible and clustered lysine residues of human ribonu- clease 7 are critical for membrane permeability and antimicrobial activity. J Biol Chem 282, 4626–4633. 10 Torrent M, Sanchez D, Buzon V, Nogues MV, Cladera

24 Gorr SU, Sotsky JB, Shelar AP & Demuth DR (2008) Design of bacteria-agglutinating peptides derived from parotid secretory protein, a member of the bacterici- dal ⁄ permeability increasing-like protein family. Peptides 29, 2118–2127.

J & Boix E (2009) Comparison of the membrane interaction mechanism of two antimicrobial RNases: RNase 3 ⁄ ECP and RNase 7. Biochim Biophys Acta 1788, 1116–1125.

25 Irie M, Nitta K & Nonaka T (1998) Biochemistry of frog ribonucleases. Cell Mol Life Sci 54, 775–784. 26 Rosenfeld Y & Shai Y (2006) Lipopolysaccharide

11 Pizzo E & D’Alessio G (2007) The success of the RNase scaffold in the advance of biosciences and in evolution. Gene 406, 8–12.

(endotoxin)-host defense antibacterial peptides interac- tions: role in bacterial resistance and prevention of sep- sis. Biochim Biophys Acta 1758, 1513–1522. 27 Weidenmaier C, Kristian SA & Peschel A (2003)

12 Dyer KD & Rosenberg HF (2006) The RNase a super- family: generation of diversity and innate host defense. Mol Divers 10, 585–597.

Bacterial resistance to antimicrobial host defenses – an emerging target for novel antiinfective strategies? Curr Drug Targets 4, 643–649.

13 Torrent M, Cuyas E, Carreras E, Navarro S, Lopez O, de la Maza A, Nogues MV, Reshetnyak YK & Boix E (2007) Topography studies on the membrane interaction mechanism of the eosinophil cationic protein. Biochem- istry 46, 720–733.

28 Conchillo-Sole O, de Groot NS, Aviles FX, Vendrell J, Daura X & Ventura S (2007) AGGRESCAN: a server for the prediction and evaluation of ‘hot spots’ of aggregation in polypeptides. BMC Bioinformatics 8, 65.

14 Torrent M, Navarro S, Moussaoui M, Nogues MV & Boix E (2008) Eosinophil cationic protein high affinity binding to bacteria wall lipopolysaccharides and pepti- doglycans. Biochemistry 47, 3544–3555.

15 Carreras E, Boix E, Rosenberg HF, Cuchillo CM &

29 Cho S, Beintema JJ & Zhang J (2005) The ribonucle- ase A superfamily of mammals and birds: identifying new members and tracing evolutionary histories. Genomics 85, 208–220.

30 Cho S & Zhang J (2007) Zebrafish ribonucleases are bactericidal: implications for the origin of the

Nogues MV (2003) Both aromatic and cationic residues contribute to the membrane-lytic and bactericidal activity of eosinophil cationic protein. Biochemistry 42, 6636–6644.

FEBS Journal 277 (2010) 1713–1725 ª 2010 The Authors Journal compilation ª 2010 FEBS

1724

M. Torrent et al.

RNase 3 and RNase 7 bactericidal activity

41 Kawabata SI & Iwanaga S (1997) Big defensin and

vertebrate RNase A superfamily. Mol Biol Evol 24, 1259–1268.

tachylectins -1 and -2. In Methods in Molecular Biology: Antibacterial Peptide Protocols (Shafer WM ed). pp 51–61. Humana Press Inc, Totowa, NJ.

31 Nitto T, Dyer KD, Czapiga M & Rosenberg HF (2006) Evolution and function of leukocyte RNase A ribonuc- leases of the avian species, Gallus gallus. J Biol Chem 281, 25622–25634.

42 Roth BL, Poot M, Yue ST & Millard PJ (1997) Bacte- rial viability and antibiotic susceptibility testing with SYTOX green nucleic acid stain. Appl Environ Micro- biol 63, 2421–2431.

32 Hileman RE, Fromm JR, Weiler JM & Linhardt RJ (1998) Glycosaminoglycan–protein interactions: definition of consensus sites in glycosaminoglycan bind- ing proteins. Bioessays 20, 156–167.

33 Gandhi NS & Mancera RL (2008) The structure of gly- cosaminoglycans and their interactions with proteins. Chem Biol Drug Des 72, 455–482.

43 Mallorquı´ -Ferna´ ndez G, Pous J, Peracaula R, Aymamı´ J, Maeda T, Tada H, Yamada H, Seno M, de Llorens R, Gomis-Ru¨ th FX et al. (2000) Three-dimensional crystal structure of human eosinophil cationic protein (RNase 3) at 1.75 A resolution. J Mol Biol 300, 1297–1307.

Supporting information

34 Sur S, Glitz DG, Kita H, Kujawa SM, Peterson EA, Weiler DA, Kephart GM, Wagner JM, George TJ, Gleich GJ et al. (1998) Localization of eosinophil- derived neurotoxin and eosinophil cationic protein in neutrophilic leukocytes. J Leukoc Biol 63, 715–722. 35 Bystrom J, Garcia RC, Hakansson L, Karawajczyk M, Moberg L, Soukka J & Venge P (2002) Eosinophil cationic protein is stored in, but not produced by, peripheral blood neutrophils. Clin Exp Allergy 32, 1082–1091.

36 Venge P, Bystrom J, Carlson M, Hakansson L, Kara- wacjzyk M, Peterson C, Seveus L & Trulson A (1999) Eosinophil cationic protein (ECP): molecular and bio- logical properties and the use of ECP as a marker of eosinophil activation in disease. Clin Exp Allergy 29, 1172–1186.

37 Schroder JM & Harder J (2006) Innate antimicrobial peptides in the skin. Med Sci (Paris) 22, 153–157. 38 Nomura I, Goleva E, Howell MD, Hamid QA, Ong PY, Hall CF, Darst MA, Gao B, Boguniewicz M, Travers JB et al. (2003) Cytokine milieu of atopic dermatitis, as compared to psoriasis, skin prevents induction of innate immune response genes. J Immunol 171, 3262–3269.

39 Ong PY, Ohtake T, Brandt C, Strickland I,

Boguniewicz M, Ganz T, Gallo RL & Leung DY (2002) Endogenous antimicrobial peptides and skin infections in atopic dermatitis. N Engl J Med 347, 1151–1160.

40 Ong PY & Leung DY (2002) Atopic dermatitis. Clin

Allergy Immunol 16, 355–379.

FEBS Journal 277 (2010) 1713–1725 ª 2010 The Authors Journal compilation ª 2010 FEBS

1725

The following supplementary material is available: Fig. S1. Study of S. aureus viability and population morphology upon incubation with RNase 3 visualized by confocal microscopy. Fig. S2. Study of S. aureus viability and population morphology upon incubation with RNase 7 visualized by confocal microscopy. Fig. S3. Analysis by SDS-PAGE of RNase 3 cell bind- ing to E. coli and S. aureus cells. Table S1. RNase 3 and RNase 7 membrane depolar- ization activities. Video S1. Register of E. coli viability and population morphology upon incubation with RNase 3, stained with the Live ⁄ Dead kit and visualized by confocal microscopy. This supplementary material can be found in the online version of this article. Please note: As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer-reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.