Arginine-Specific Mono ADP-Ribosylation In Vitro of Antimicrobial Peptides by ADP-Ribosylating Toxins

Among the several toxins used by pathogenic bacteria to target eukaryotic host cells, proteins that exert ADP-ribosylation activity represent a large and studied family of dangerous and potentially lethal toxins. These proteins alter cell physiology catalyzing the transfer of the ADP-ribose unit from NAD to cellular proteins involved in key metabolic pathways. In the present study, we tested the capability of four of these toxins, to ADP-ribosylate α- and β- defensins. Cholera toxin (CT) from Vibrio cholerae and heat labile enterotoxin (LT) from Escherichia coli both modified the human α-defensin (HNP-1) and β- defensin-1 (HBD1), as efficiently as the mammalian mono-ADP-ribosyltransferase-1. Pseudomonas aeruginosa exoenzyme S was inactive on both HNP-1 and HBD1. Neisseria meningitidis NarE poorly recognized HNP-1 as a substrate but it was completely inactive on HBD1. On the other hand, HNP-1 strongly influenced NarE inhibiting its transferase activity while enhancing auto-ADP-ribosylation. We conclude that only some arginine-specific ADP-ribosylating toxins recognize defensins as substrates in vitro. Modifications that alter the biological activities of antimicrobial peptides may be relevant for the innate immune response. In particular, ADP-ribosylation of antimicrobial peptides may represent a novel escape mechanism adopted by pathogens to facilitate colonization of host tissues.


Introduction
Human defensins are cationic multifunctional arginine-rich peptides (molecular masses ranging from 3.5 to 6 kDa) characterized by three intramolecular disulfide bridges that stabilize their structure [1][2][3][4]. Defensins display microbicidal activity against a wide spectrum of Gram-negative and Gram-positive bacteria, fungi and viruses [5]. They are also cytotoxic for epithelial cells and chemotactic for T-cells. Based on the presence of six conserved cysteine residues and sequence homology, human defensins are grouped into aand bdefensins. The first group (a-defensins) includes human neutrophil peptides (HNP)-1 to 4, major components of the azurophilic granules of neutrophils, and two enteric human defensins, HD-5 and HD-6, isolated from the granules of Paneth cells in the small intestine, [6]. The second group (b-defensins), is mainly expressed in epithelial cells of various organs [7][8][9]. It has been shown that ADP-ribosylation of HNP-1 on arginine 14 reduces its antimicrobial and cytotoxic activities [10]. Mono ADP-ribosylation consists in the enzymatic transfer of the single ADP-ribose moiety of NAD to specific amino-acid residues of acceptor proteins coupled to the release of nicotinamide (nam) [11]. In mammals this reaction is catalyzed by a family of ADP-ribosyltransferases (ART1-5) [12,13], while the best studied ADP-ribosylation reactions are those catalyzed by bacterial ADP-ribosylating toxins. The ADP-ribosylation of a large panel of host proteins catalyzed by bacterial toxins leads to the interruption of cellular metabolic and regulatory pathways causing severe diseases [14]. Vibrio cholerae toxin (CT) [15], Escherichia coli heat labile enterotoxin (LT) [16], Pseudomonas aeruginosa exoenzyme S (ExoS) [17] and the recently discovered NarE, a toxin-like protein from Neisseria meningitidis [18], recognize arginine as an ADP-ribose acceptor in a similar fashion to ART1 and ART5 [13,19]. Arginine specificity is conferred to ARTs by the presence of the R-S-EXE triad signature in the active site [20]. Recent studies indicated that a-defensins display a novel biological function consisting in the ability to neutralize the activity of potent bacterial toxins like lethal factor, a metalloprotease produced by Bacillus anthracis [21], and toxin B produced by Clostridium difficile [22]. Moreover it has been shown that HNP1-3 neutralize the cytotoxic effects exerted by diphtheria toxin (DT) and Pseudomonas aeruginosa exotoxin A (ETA), while they were inactive on CT and pertussis toxin (PT) [23]. The neutralization of toxins with selected amino-acid specificity prompted us to hypothesize that mono ADP-ribosylation of specific amino-acids may block defensin ability to inhibit the activities of toxins. Therefore, we evaluated whether HNP-1 could be recognized by arginine-specific bacterial ARTs. In the present paper we provide evidence that CT and LT ADP-ribosylated aand bdefensins, which thus represent novel substrates for these bacterial ARTs. On the other hand, NarE and ExoS did not modify either aor bdefensins. Interestingly, unmodified HNP-1 exerted inhibition on NarE transferase activity suggesting a regulatory role. While the ADP-ribosyltransferase activity was inhibited by HNP-1, the NAD-glycohydrolase (NADase) activity remained unaltered. Furthermore, HNP-1 strongly enhanced the auto-ADP-ribosylation of NarE, a recently discovered catalytic activity of this toxin. Overall, our data highlight the interplay between ADP-ribosylating toxins and human defensins.

Results
To establish whether arginine-specific bacterial ARTs can ADP-ribosylate HNP-1, we incubated HNP-1 with the catalytic A subunit of CT (CTA), LT (LTA), ExoS or NarE individually. As shown in Fig 1A, CTA and LTA catalyzed the transfer of the biotin-ADP-ribose from biotin-NAD to HNP-1 with an efficiency that was comparable to that of ART1 ( Fig. 1 B). This incorporation was strongly reduced after heat-inactivation of the toxins ( Fig. 1 A). CTA and LTA have both transferase, and NADase activity [24,16]. The latter produces ADP-ribose that can react non-enzymatically with lysine residues in proteins [25]. However, since the incorporation of biotin-ADP-ribose on HNP-1 was strongly reduced in the presence of 2 mM unlabelled NAD (200-fold excess) but not with 2 mM ADP-ribose, we could rule out that the reaction was non-enzymatic. The enzymatic nature of the reaction was further confirmed in the dose dependent ( Fig. 1  D) and time-course experiments ( Fig. 1 E), showing that the increase of modified peptide is dependent on the level of free substrate and by the incubation time. In this respect the purification grade of the toxins (Fig. 1 C) is shown, to exclude the possibility of a blockage of the peptide by contaminating proteins. Under the same conditions, HNP-1 was a poor substrate for NarE ( Fig. 1 A) compared to ART1 (Fig. 1 B). ExoS was completely inactive towards HNP-1 (data not shown), in agreement with a previous report [26]. ADP-ribosylation of antimicrobials by CT and LT is not restricted to HNP-1. Also HBD1, which contains only one arginine at position 29 and is constitutively expressed by epithelial cells in the airway [27], was ADPribosylated ( Fig. 2 A). As for HNP-1, labelling did not occur in the presence of heat-inactivated toxins. The addition of an excess of unlabelled NAD to the reaction mixture decreased the incorporation of an ADP-ribose moiety on HBD1, while the incorporation of biotin-ADP-ribose on HBD1 was not reduced by the presence of 2 mM ADP-ribose. Dose-dependent reactions and time course experiments support the enzymatic nature of the modification also in the case of HBD1 (Fig. 2 C, D). NarE and ExoS did not modify HBD1 (data not shown). In contrast with a previous report [26], HBD1 was modified by ART1 to the same extent of HNP-1 (Fig. 2 B). To confirm that the observed modifications corresponded to the addition of the ADP-ribose unit, the products of the reaction of CTA with HNP-1 in the presence of NAD were identified by MALDI-TOF MS. As shown in Fig. 3, these included a peptide of 3442.12 Da, consistent with unmodified HNP-1 (theoretical mass: 3442.1 Da) and a peptide of 3983.15 Da. Although the amount of the modified peptide was low, we can conclude that the reaction is specific since we observed a mass increase consistent with mono ADP-ribosylated HNP-1 (theoretical mass: 3983.1 Da). Similar results were obtained with the LT catalyzed reaction (data not shown). To identify the preferred arginine residue of HNP-1 modified by CTA and LTA, we used two variants of HNP-1 in which a lysine replaced the arginines at positions 14 (HNP-1-R14K) or 15 (HNP-1-R15K).
We found that CTA and LTA selectively ADP-ribosylated HNP-1 at R14 (Fig. 4). Recent studies have shown that when HNP-1 is not recognized as a substrate, it is able to inhibit the ART activity of bacterial toxins such as ETA and DT [23] and also the eukaryotic ART5 multiple catalytic activities [26]. Therefore, since HNP-1 is only weakly modified by NarE, we investigated whether HNP-1 exerts a similar effect on NarE activities. The addition of HNP-1 to the reaction mixture seems to reduce the ADP-ribosyltransferase activity in a concentration dependent fashion (Fig. 5, grey bars) while the NADase activity was not greatly affected (Fig. 5, white bars). In contrast, HNP-1 enhanced the auto-ADP-ribosylation of NarE ( Fig. 6 upper panel), a recently discovered activity of this toxin (Picchianti et al. manuscript in preparation).

Discussion
ADP-ribosylating toxins are usually secreted by bacterial pathogens in the host environment. Some of them, which possess arginine-specificity, could recognize arginine-rich peptides such as aand bdefensins as substrates. Both aand bdefensins are released by neutrophils and epithelial cells respectively in high amounts at inflammatory sites. In this report we present evidence that synthetic HNP-1 and HBD1 are ADP-ribosylated in vitro by CTA and LTA. In contrast they are not recognized as substrates by ExoS and only poorly by NarE, suggesting specificity for both bacterial toxins and substrates. The artificial kemptide (PKA peptide substrate), which contains a di-arginine motif, was modified by CT on the first arginine of the motif while a mammalian ART recognized the second arginine within the R-R motif [28,29]. In contrast, our data indicate R14 as the preferred modification site, since the HNP-1-R14K was not ADP-ribosylated by the ADP-ribosylating toxins used in this study. Our findings are in contrast with studies performed by other groups, which failed to show toxin-catalyzed incorporation of the ADPribose unit on defensins [23,26]. Others evaluated the presence of the ADP-ribosylated-HNP-1 by monitoring the absorbance of the modified peptide in reversed-phase chromatography, but not being successful in identifying it [23,26]. Therefore we chose a chemiluminescence assay to detect ADP-ribosylation because of the higher sensitivity, allowing the detection of small amounts of modified HNP-1.
In agreement with previous findings we did not observe incorporation of ADP-ribose with PT [23] and ExoS [26]. Labelling of proteins can also result from the covalent noncatalyzed reaction of NAD [30] or free ADP-ribose with the eamino groups of lysines [25]. ADP-ribosylation of HNP-1, in which lysine residues are absent (Table 1), was not blocked by the addition of free ADP-ribose, while a reduction of incorporation was noticed when unlabelled NAD was added to the reaction mixture. Comparable results were obtained with HBD1 that contains four lysine residues. These data, further supported by mass spectrometry analysis, strongly indicate that an enzymatic ADP-ribosylation, and not a secondary reaction with NAD or ADP-ribose, was responsible for the modification. Defensins belonging to aand bgroup contain several conserved arginines (Table 1), which are recognized by CT and LT. However they are devoid of diphtamide and asparagine residues, which are the target amino-acid of DT, ETA and clostridial toxins. Furthermore, cysteines, present in aand bdefensins and recognized by PT as ADP-ribose acceptors, are engaged in disulphide bridges. It is well known that defensins have a variety of activities, but the antimicrobial function is by far the most important. Thus ADPribosylation of selected arginines might well correlate with recent discoveries, which show that antibacterial activity strictly depends on cationicity [31] and that only selective arginines support this activity [32]. Interestingly, toxic activities are not decreased in the case of CT [23], which ADP-ribosylates HNP-1 at an arginine residue. By contrast, HNP-1, which is devoid of the target aminoacids is not modified by ETA and DT, thus protecting cells from DT-or ETA-mediated cell death [23]. Besides its antimicrobial activity, several lines of evidence suggest an additional regulatory role for HNP-1 when it is not recognized as a substrate, as described for ART5 and ART1 [26]. Here we showed that HNP-1 is able to reduce NarE transferase activity. This reduction is more evident at high concentrations of HNP-1, likely to be present in the site of inflammation. Of note the NADase activity, a reaction not usually involved in the toxicity process, is not affected. On the other hand, auto-ADP-ribosylation, which could be an intramolecular mechanism regulating the two activities (Picchianti et al. manuscript in preparation) is enhanced. Although physiological substrates for CT and LT are well known and the extent of modification is limited, members of the antimicrobial peptide family may serve as novel substrates for these ADP-ribosylating toxins. At the onset of infection, bacterial pathogens have evolved different countermeasures to limit the effectiveness of antimicrobials [33] and to counteract the immune system. The modification of antimicrobial components of the innate immune system by bacterial ADP-ribosylating toxins may represent a mechanism that could facilitate bacterial colonization. In the context of the primary immune response, the ADP-ribosylation of HNP-1 may affect both the anti-microbial weapons released by neutrophils and the interplay between inflammatory cells, eventually facilitating the onset of an infectious disease. Hence, our study showing arginine-specific ADP-ribosylation of human defensins catalyzed by some bacterial toxins may be relevant in the onset of infectious diseases.

Preparation of ART1 and bacterial toxins
Chinese hamster fibroblast V79 cells transfected with pTet-ON-ART1 cDNA were kindly provided by Dr. J. Mac Dermot [34]. Human ART1 was synthesized as a GPI-linked protein on the surface of V79 cells after induction for 48 h with 2 mg/ml final concentration of doxycicline. The soluble form of ART1 was collected following treatment of intact V79 cells with 1 U/ml of phosphatidylinositol-specific phospholipase C for 1 hr at 37uC [34]. NarE was produced as previously described [18]. Briefly, the NarE gene cloned in the pET21b+ expression vector with a carboxyl-terminal 66 histidine tag was used to transform E. coli BL21 (DE3) competent cells (Invitrogen). Transformed cells were grown overnight in LB broth at 37uC with gentle shaking (180 rpm) and protein synthesis occurred after addition of 1 mM isopropyl-1-thio-b-D-galactopyranoside (IPTG) for 3 h at 25uC. Following induction, bacteria were harvested, centrifuged and lysed in B-PER (bacterial-protein extraction reagent, Pierce) in the presence of Mg ++ , DNase, lysozyme and phenyl-methyl sulfonyl fluoride (PMSF) as a protease inhibitor. After centrifugation to discard debris and membranes, the soluble fraction was loaded on a Ni-NTA (Pharmacia, Biotech, Stockholm, Sweden) affinity resin and the protein was eluted according to manufacturer's instructions. CTA was purchased from Sigma Aldrich, while LTA was cloned and purified by Drs. Paolo Ruggiero and Laura Pancotto (Novartis Vaccines & Diagnostics, Siena Italy). Since all of the above toxins need sulphydril agents like dithiothreitol (DTT) to exert full activity, we performed their activation as previously described [35,36]. Since DTT causes reduction of disulphide bridges causing linearization of HNP-1 and facilitating access of ADP-ribose to arginine acceptors, the activation mixture was extensively dialyzed in the same buffer lacking DTT using Microcon centrifugal filter devices (Amicon, Houston, TX). Pseudomonas aeruginosa exoenzyme S ADP-ribosyltransferase (transferase domain) and its activator FAS (Factor activating exoenzyme S) were kindly provided by Dr. Joseph Barbieri (Medical College of Wisconsin, Milwaukee WI).

ADP-ribosyltransferase enzymatic assay
Some of the ADP-ribosylating enzymes used in this study ART1, CTA, LTA, and NarE were tested by monitoring the transfer of ADP-ribose to agmatine using a standard assay [37]. The assay was carried out in a final volume of 0.3 ml containing 50 mM potassium phosphate, pH 7.5, 20 mM agmatine and 0.1 mM [adenine-U-14 C]NAD (0.05 mCi). After incubation at   30uC, duplicate samples (100 ml each) were applied to 1 ml columns of Dowex AG 1-X2. [adenine-14 C]ADP-ribosylagmatine was eluted for radio assay with 5 ml of H 2 O and the radioactivity counted in a Packard mod counter. The ART1 fraction exerted an activity of 6.8 nmol/h (U), while the activated CTA and LTA showed values of 2.5 nmol/h (U) and 1.8 nmol/h (U) respectively. NarE 0.9 nmol/h, Uwas used shortly after purification. Specific activities of the toxins used were 2.5 nmol/mg/h for CTA, 6.25 nmol/mg/h for LTA and 0.4 nmol/mg/h for NarE. ExoS activity was monitored using an auto-ADP-ribosylation assay in the presence of FAS ligand [38].

NAD-glycohydrolase assay
An analogous assay based on nam release was employed to measure the NADase activity of NarE [24]. This assay was carried out in the same conditions described above for the transferase assay with 0.1 mM [carbonyl-14 C]NAD (0.05 mCi) replacing [adenine-U- 14 C]NAD and without agmatine as ADP-ribose acceptor.
Detection of biotinylated ADP-ribose-HNP-1 and immunoblot analysis HNP-1 (3 mg, 43,56 mM) and HBD1 (3 mg, 38.18 mM) were ADP-ribosylated for 1 hour at 30uC in 50 mM potassium phosphate buffer (pH 7.4), with ART1, CTA, LTA, or NarE and 10 mM of biotinylated-NAD (biotin-NAD) in a final volume of 20 ml. In control experiments 2 mM NAD or 2 mM ADP-ribose were added individually to each assay mixture. In negative controls the ADP-ribosylating toxin was omitted from the reaction mixture or heat-inactivated. The ADP-ribosylated peptides were resolved by SDS-PAGE in a 10% NuPAGE gel, using MES as running buffer and transferred to nitrocellulose using a dry system apparatus (I-Blot, Invitrogen). The membrane was blocked for 1 h with 5% BSA in PBS containing 0.05% Tween-20 (PBS-T), extensively washed and then incubated in the same buffer containing streptavidin-HRP conjugated (1:10000 dilution) and mixed for 1 h on a nutator at room temperature (RT). After several washings with PBS-T, bound streptavidin was detected using an ECL immunoblotting detection system (Bio-Rad) according to the manufacturer's instructions. Molecular masses were estimated from the calibration standard included in each gel.

Mass-Spectrometry analysis of ADP-ribosylated-HNP-1
Peptide molecular masses were determined using a MALDI-TOF/TOF mass spectrometer UltraFlex (Bruker Daltonics, Bremen, GmbH). Ions generated by laser desorption at 337 nm (N 2 laser) were recorded at an acceleration voltage of 20 kV in linear mode. In general, about 200 single spectra were accumulated to improve the signal/noise ratio and analyzed by FlexAnalysis version 2.4 (Bruker Daltonics). Briefly, 1 ml of reaction solution (20-60 pmoles) was added to 1 ml of a saturated solution of sinapinic acid (3,5-dimethoxy-4-hydroxy-trans-cinnamic acid) in 30% (vol/vol) acetonitrile, 0.1% (vol/vol) trifluoroacetic acid (TFA). Then 2 ml of peptide/matrix mixture was spotted on a stainless steel sample target and air-dried at room temperature. Peptide mass spectra were calibrated using external peptide calibration standard (Bruker Daltonics).

Protein assay
Protein content was determined by the Bradford protein assay kit (Bio-Rad) using BSA for standardization.