With Antibacterial Peptides Gram-Negative Bacteria and Are

Home , Antimicrobial peptides, Micrococcus luteus

Human Peptidoglycan Recognition Proteins Require Zinc to Kill Both Gram-Positive and Gram-Negative Bacteria and Are Synergistic with Antibacterial Peptides This information is current as of September 29, 2021. Minhui Wang, Li-Hui Liu, Shiyong Wang, Xinna Li, Xiaofeng Lu, Dipika Gupta and Roman Dziarski J Immunol 2007; 178:3116-3125; ; doi: 10.4049/jimmunol.178.5.3116 http://www.jimmunol.org/content/178/5/3116 Downloaded from

References This article cites 63 articles, 33 of which you can access for free at:

http://www.jimmunol.org/content/178/5/3116.full#ref-list-1 http://www.jimmunol.org/

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication by guest on September 29, 2021

*average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2007 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

Human Peptidoglycan Recognition Proteins Require Zinc to Kill Both Gram-Positive and Gram-Negative Bacteria and Are Synergistic with Antibacterial Peptides1

Minhui Wang, Li-Hui Liu, Shiyong Wang, Xinna Li, Xiaofeng Lu, Dipika Gupta, and Roman Dziarski2

Mammals have four peptidoglycan recognition proteins (PGRPs or PGLYRPs), which are secreted innate immunity pattern recognition molecules with effector functions. In this study, we demonstrate that human PGLYRP-1, PGLYRP-3, PGLYRP-4, and PGLYRP-3:4 have Zn2؉-dependent bactericidal activity against both Gram-positive and Gram-negative bacteria at physiologic Zn2؉ concentrations found in serum, sweat, saliva, and other body ﬂuids. The requirement for Zn2؉ can only be partially replaced 2؉ by Ca for killing of Gram-positive bacteria but not for killing of Gram-negative bacteria. The bactericidal activity of PGLYRPs Downloaded from is salt insensitive and requires N-glycosylation of PGLYRPs. The LD99 of PGLYRPs for Gram-positive and Gram-negative bacteria is 0.3–1.7 ␮M, and killing of bacteria by PGLYRPs, in contrast to killing by antibacterial peptides, does not involve ␣ ␤ permeabilization of cytoplasmic membrane. PGLYRPs and antibacterial peptides (phospholipase A2, - and -defensins, and bactericidal permeability-increasing protein), at subbactericidal concentrations, synergistically kill Gram-positive and Gram- negative bacteria. These results demonstrate that PGLYRPs are a novel class of recognition and effector molecules with broad 2؉ Zn -dependent bactericidal activity against both Gram-positive and Gram-negative bacteria that are synergistic with antibac- http://www.jimmunol.org/ terial peptides. The Journal of Immunology, 2007, 178: 3116–3125.

eptidoglycan recognition proteins (PGRPs)3 are innate im- zation Gene Nomenclature Committee changed their names to munity molecules present in insects, mollusks, echino- peptidoglycan recognition proteins 1, 2, 3, and 4 (PGLYRP-1, P derms, and vertebrates but not in nematodes and plants PGLYRP-2, PGLYRP-3, and PGLYRP-4), respectively, and (1–3). PGRPs have at least one PGRP domain, which is homolo- this terminology is used here. gous to bacteriophage and bacterial type 2 amidases (1–3). Insects Mammalian PGLYRP-2 is an N-acetylmuramoyl-L-alanine ami- have up to 19 PGRPs, classified into short (S) and long (L) forms dase that hydrolyzes bacterial peptidoglycan (5, 6). PGLYRP-2 is by guest on September 29, 2021 (3). The short forms are present in the hemolymph, cuticle, and fat secreted from liver into blood and is also induced by bacteria in body cells and sometimes in epidermal cells in the gut and hemo- epithelial cells (7–9). The remaining three mammalian PGLYRPs cytes, whereas the long forms are mainly expressed in hemocytes are bactericidal or bacteriostatic proteins that are secreted as (3). The expression of insect PGRPs is often up-regulated by ex- disulfide-linked homo- and heterodimers (10). Mammalian posure to bacteria. Insect PGRPs activate Toll or Imd signal trans- PGLYRP-1, PGLYRP-3, and PGLYRP-4 are a new class of bac- duction pathways or induce proteolytic cascades that generate an- tericidal proteins that have a different structure, mechanism of ac- timicrobial products, induce phagocytosis, and protect insects tion, and expression than currently known mammalian antimicro- against infections (3). Some insect PGRPs are enzymes that hy- bial peptides (10, 11). PGLYRPs are much larger than all currently drolyze peptidoglycan, a bacterial cell wall component (3). known vertebrate antibacterial peptides: they are disulfide-linked Mammals have four PGRPs, which were initially named glycosylated 44- to 115-kDa dimers, in contrast to vertebrate PGRP-S, PGRP-L, and PGRP-I␣ and PGRP-I␤ (for “short”, antimicrobial peptides, which are typically 3–15 kDa (10, 11). “long”, or “intermediate” transcripts, respectively), by analogy to PGLYRPs kill bacteria likely by interacting with cell wall pepti- insect PGRPs (3, 4). Subsequently, the Human Genome Organi- doglycan, whereas antibacterial peptides kill bacteria by permeabilizing their membranes (10). PGLYRPs require divalent cations Indiana University School of Medicine, Northwest Campus, Gary, IN 46408 and N-glycosylation for their bactericidal activity, and antibacterial Received for publication October 11, 2006. Accepted for publication December 14, peptides usually do not (10, 11). 2006. PGLYRP-1 is expressed primarily in PMN granules (12–15), The costs of publication of this article were defrayed in part by the payment of page and PGLYRP-3 and PGLYRP-4 are expressed in the skin, eyes, charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. salivary glands, throat, tongue, esophagus, stomach, and intestine 1 This work was supported by U.S. Public Health Service Grants AI28797 and (10). In addition to PGLYRPs, these tissues also express many AI56395 from the National Institutes of Health. antimicrobial peptides and proteins, such as defensins (16, 17), 2 Address correspondence and reprint requests to Dr. Roman Dziarski, Indiana Uni- phospholipase A2 (18–22), dermicidin (23), cathelicidin (24), versity School of Medicine, Northwest Campus, 3400 Broadway, Gary, IN 46408. E-mail address: [email protected] RNases (25), psoriasin (26), calprotectin (27–29), bactericidal permeability-increasing protein (22), and lysozyme (22). Secretions of 3 Abbreviations used in this paper: PGRP or PGLYRP, peptidoglycan recognition pro- 2ϩ teins; BPI, bactericidal permeability-increasing protein; HBD-3, human ␤-defensin-3; these tissues also contain several divalent cations, such as Ca , HNP-1, human neutrophil protein-1 (␣-defensin); LTA, lipoteichoic acid; MurNAc, Mg2ϩ,Zn2ϩ,Fe2ϩ,Cu2ϩ,Se2ϩ,Ni2ϩ, and Mn2ϩ (30), which influ- N-acetylmuramic acid; LB, Luria-Bertani; PLA , group IIA phospholipase A . 2 2 ence the activity of antimicrobial proteins. Because our previous Copyright © 2007 by The American Association of Immunologists, Inc. 0022-1767/07/$2.00 results indicated that bactericidal activity of human PGLYRPs for www.jimmunol.org The Journal of Immunology 3117

ϩ Gram-positive bacteria is Ca2 dependent (10) and because other Table I. Concentrations of divalent cations in human sweat and saliva divalent cations present in body secretions and tissue fluids influ- and in PGLYRP bactericidal assay buffer ence the activity of other antimicrobial proteins (16, 17, 27–40), in this study, we tested the hypothesis that other divalent cations may Cation Sweat Saliva Assay Buffersa also play an important role in bactericidal activity of human Ca2ϩ 0.5 mM 2 mM 1 mM PGLYRPs. Moreover, because PGLYRPs are secreted into body Mg2ϩ 100 ␮M40␮M50␮M fluids that also contain many antibacterial peptides with different Zn2ϩ 17 ␮M2␮M5␮M 2ϩ bactericidal mechanisms from PGLYRPs (10), we also tested the Fe 6 ␮M6␮M5␮M Cu2ϩ 5 ␮M5␮M5␮M hypothesis that PGLYRPs may kill bacteria synergistically with ϩ Se2 1 ␮M39nM these peptides. We discovered that PGLYRPs have higher bacte- Ni2ϩ 0.8 ␮M0 ricidal activity against Gram-positive bacteria and become highly Mn2ϩ 0.4 ␮M0 1␮M 2ϩ bactericidal for Gram-negative bacteria in the presence Zn and a Final concentration of ions in the bactericidal assays used individually or in also that PGLYRPs kill bacteria synergistically together with combinations as indicated in Figs. 1 and 2. Two times higher concentrations were membrane-damaging antibacterial peptides. used for protein purification and storage buffers, and other components were used as described in Materials and Methods. Materials and Methods Materials concentrations indicated in Results. The standard medium was 5 mM Tris ␮ Human PGLYRP-1, PGLYRP-3, PGLYRP-4, and PGLYRP-3:4 were ex- (pH 7.6), with 150 mM NaCl, 5 M ZnSO4, 5% glycerol, and 1% Luria- pressed in S2 cells and purified as previously described (10), except that Bertani (LB) for Gram-negative bacteria, or the same medium plus 1 mM Downloaded from 2ϩ 2ϩ ␮ instead of Ca ,40␮MZn (for testing on Gram-negative bacteria) or 2 CaCl2 and 5 M CuSO4 and either 1% LB or 1% BHI for Gram-positive mM Ca2ϩ,40␮MZn2ϩ, and 10 ␮MCu2ϩ (for testing on Gram-positive bacteria (10), unless otherwise indicated (i.e., in some experiments com- bacteria) were used in protein purification buffers, and 10 ␮MZn2ϩ (for binations of various divalent cations listed in Table I were used, as indi- Gram-negative bacteria) or 2 mM Ca2ϩ,10␮MZn2ϩ, and 10 ␮MCu2ϩ cated in Results). The numbers of bacteria were determined by colony (for Gram-positive bacteria) were used in the final dialysis buffer. Wher- counts on LB or BHI agar plates (10). In some experiments, as indicated in ever indicated, other combinations of divalent cations were also used in Results, 0.525 M sucrose was included in the assay and dilution medium, and bacteria were overlaid in 0.7% agar onto 1.5% agar LB plates, both some experiments in the purification and dialysis buffers as described in http://www.jimmunol.org/ Results. For Zn2ϩ removal, proteins were incubated with 100 ␮M EGTA containing 0.5 M sucrose (10). All other controls were performed as de- (10). De-N-glycosylation was performed as described previously (10). scribed previously (10). Bactericidal activity is defined as an at least 2 log10 decrease in the number of inoculated bacteria in 6 h; bacteriostatic activity Human group IIA phospholipase A2 (PLA2) (41) was amplified from the Image clone ID 152802 (obtained from Open Biosystems) by PCR using is defined as an inhibition of growth of bacteria or a decrease in the number Ͻ forward (GCAGATCTAATTTGGTGAATTTCCACAGAATG) and re- of inoculated bacteria of 2 log10 in 6 h. The results are means of three verse (ATGAATTCGCAACGAGGGGTGCTCCCTCTG) primers and (mostly) or two experiments; the average SEs were 18% of the mean subcloned into the BglII and EcoRI sites of the pMT/BiP/V5-His expression (ranged from 6 to 28%) and are not shown because the error bars were mostly smaller than the size of the symbols in the figures. The significance vector (Invitrogen Life Technologies). To generate human PLA2-PGLYRP-1 fusion protein, the PGLYRP-1 coding region (10) was amplified by PCR using of differences was calculated using the Student’s t test, and all differences Ͼ Ͻ forward (GTGAATTCGCTCAGGAGACAGAAGACCCGG) and reverse in bacterial numbers 1 log10 were statistically significant at p 0.05 (CATCTAGAGGGGGAGCGGTAGTGTGGCCAAT) primers and was (and most often at p Ͻ 0.01). by guest on September 29, 2021 Ј inserted 3 of the PLA2 into the EcoRI and XbaI sites. Human Fc IgG1 fragment was amplified from the clone ID 5480266 (obtained from Amer- Membrane permeabilization ican Type Culture Collection) by PCR using forward (CGAGATCTGA Membrane permeabilization was studied as previously described (10) using GCCCAAATCTTGTGACAAAACTC) and reverse (ATACCGGTTATT membrane-nonpermeable DNA-binding fluorescent dye SYTOX-Green (2 TCCATGCTGGGTGCCTGGG) primers (codons 239–506) and subcloned ␮M; Molecular Probes) and 3.5 ϫ 106/ml E. coli in the same medium as into the BglII and AgeI sites of the pMT/BiP/V5-His expression vector. To for the bactericidal assays without or with 0.525 M sucrose at 37°C. For generate human PGLYRP-1-Fc fusion protein, the Fc coding region was these experiments, bactericidal assays were performed as above with 3.5 ϫ amplified as above using forward (CGTCTAGAGAGCCCAAATCTTGTG 106/ml E. coli. ACAAAACTC) primer and the same reverse primer and was inserted 3Ј of the PGLYRP-1 (10) into the XbaI and AgeI sites. All constructs were confirmed by restriction digestion and by sequencing. The proteins were Results expressed and purified from S2 cell supernatants as described above for Bactericidal activity of PGLYRPs against Gram-positive PGLYRPs. Each of these purified proteins (10 ␮g/lane) yielded one bacteria is enhanced by Zn2ϩ band on Coomassie brilliant blue-stained gel of comparable purity to PGLYRPs (10). We previously demonstrated that human PGLYRPs were strongly Human ␣-defensin (human neutrophil protein-1 (HNP-1)) was obtained bactericidal for some pathogenic Gram-positive bacteria (S. aureus form Bachem Bioscience, human neutrophil bactericidal permeability-in- and L. monocytogenes) and nonpathogenic transient Gram-positive creasing protein (BPI) was from Athens Research and Technology, and bacteria (B. subtilis, Bacillus cereus, Bacillus licheniformis, and recombinant human ␤-defensin-3 (HBD-3) was from Cell Sciences. Ma- gainin (Ala8,13,18-magainin II amide), lysostaphin (recombinant), and all L. acidophilus) (10). This bactericidal activity required the ϩ ϩ other reagents were from Sigma-Aldrich, unless otherwise indicated. presence of Ca2 , which could not be replaced by Mg2 (10). However, under these conditions, these PGLYRPs were only bac- Bacteria teriostatic (but not bactericidal) for other pathogenic (S. pyogenes) The following bacteria were used: Bacillus subtilis ATCC 6633, Enterococcus or normal flora (Staphylococcus epidermidis, M. luteus, E. faeca- faecalis ATCC 19433, Escherichia coli K12 (used in most experiments), E. lis, and Streptococcus agalactiae) Gram-positive bacteria and for coli O18:K1:H7:Bort (42), Lactobacillus acidophilus ATCC 4356, Listeria monocytogenes ATCC 19115, Micrococcus luteus ATCC 4698, Proteus vul- all Gram-negative bacteria (E. coli, P. vulgaris, and Enterobacter garis ATCC 13315, Pseudomonas aeruginosa ATCC 27853, Salmonella en- cloacae) (10). However, these PGLYRPs were losing bactericidal terica ATCC 6539, Shigella sonnei ATCC 25931, Staphylococcus aureus Rb activity upon storage at 4°C in our buffer with Ca2ϩ, suggesting (10), and Streptococcus pyogenes ATCC 49399. that the ion or pH conditions of this buffer were not optimal for the Antibacterial assays stability and activity of these proteins. Because PGLYRPs are secreted (PGLYRP-1 into PMNЈs granules, PGLYRP-2 into the se- Bactericidal and bacteriostatic activity of human PGLYRPs and antibacterial peptides was assayed on logarithmically growing bacteria at 0.4– rum, PGLYRP-3, PGLYRP-4, and PGLYRP-3:4 into sweat, and 1.2 ϫ 106/ml as previously described (10) with recombinant human Fc PGLYRP-4 also into saliva) (10), we reasoned that they should be IgG1 fragment as a control protein or PGLYRPs or other test proteins at the more stable (and maybe more active) under the conditions that 3118 BACTERICIDAL PGRPs Downloaded from http://www.jimmunol.org/

FIGURE 2. PGLYRPs are bactericidal for E. coli in the presence of Zn2ϩ. E. coli was incubated with 100–150 ␮g/ml of the indicated PGLYRPs puriﬁed and assayed in the presence of the indicated ions (Ca2ϩ, Zn2ϩ, and Cu2ϩ) individually or in combination at the concentrations shown in Table I or Fc IgG fragment as a control. The numbers of bacteria were determined by colony counts. The results are means of two to three

experiments. by guest on September 29, 2021

PGLYRP-4 purified and tested as before (10) in the presence of Ca2ϩ only (Fig. 1A). To determine which of these cations were required for this higher bactericidal activity, we tested bactericidal activity of PGLYRP-4 purified and assayed in the presence of individual cations or various combinations of these cations. Equally high bactericidal activity of PGLYRP-4 for S. aureus was observed in the presence of three cations (Ca2ϩ,Zn2ϩ, and Cu2ϩ) and almost as high in the presence of Zn2ϩ only (Fig. 1A). These results demonstrate that Zn2ϩ is the single most essential cation needed for bactericidal activity of PGLYRP-4 with optimal effect at ϳ5 ␮M, which is the approximate concentration of Zn2ϩ found in body secretions, such as sweat and saliva (Table I), where FIGURE 1. Bactericidal activity of PGLYRPs for S. aureus and S. pyo- PGLYRP-4 is found. PGLYRP-4 at lower concentrations of Zn2ϩ 2ϩ genes is enhanced by Zn . S. aureus or S. pyogenes was incubated with was less active (data not shown). Zn2ϩ is probably required for the ␮ ␮ 50–100 g/ml (S. aureus) or 100–150 g/ml (S. pyogenes) of the indicated stability or proper conformation of PGLYRP proteins because it PGLYRPs purified and assayed in the presence of the indicated six ions ϩ ϩ ϩ ϩ ϩ ϩ had to be present during the entire purification of PGLYRPs and (Ca2 ,Mg2 ,Zn2 ,Fe2 ,Cu2 , and Mn2 ) individually or in combina- bactericidal assays. The high bactericidal activity of PGLYRPs tion (as indicated on the figure) at the concentrations shown in Table I or 2ϩ Fc IgG fragment as a control. The numbers of bacteria were determined by purified in the absence of Zn could not be restored by simply 2ϩ 2ϩ colony counts. The results are means of two to three experiments. adding Zn to the protein purified without Zn or to the bactericidal assay (data not shown). To determine whether other human bactericidal PGLYRPs closely mimic these body fluids. To test this hypothesis, we de- had the same ion requirements, we tested the effect of Zn2ϩ, termined the bactericidal activity of PGLYRPs that were puri- Ca2ϩ, and Cu2ϩ (alone and in combinations) on the bacteri- fied and assayed in the presence of six divalent cations that are cidal activity of PGLYRP-1, PGLYRP-3, and PGLYRP-3:4 found in sweat and saliva in at least micromole concentrations (PGLYRP-3: PGLYRP-4 heterodimer) for S. aureus.Zn2ϩ was (Table I) (30). the most essential cation for the bactericidal activity of all human PGLYRP-4, purified and tested in the presence of six divalent PGLYRPs (Fig. 1, A–D). Note that these assays were performed at cations (Table I), had higher bactericidal activity for S. aureus than lower concentrations of PGLYRPs than before (10). The Journal of Immunology 3119

FIGURE 4. PGLYRPs are bactericidal (kill 99% of bacteria) at ϳ45 ␮g/ml (0.4–1.0 ␮M) for S. aureus and at 30–200 ␮g/ml (0.3–1.7 ␮M) for E. coli. Bacteria were incubated for 4–6 h with increasing concentrations

of the indicated PGLYRPs or PLA2 puriﬁed and assayed in the presence of Ca2ϩ,Zn2ϩ, and Cu2ϩ (S. aureus)orZn2ϩ only (E. coli). The numbers of bacteria were determined by colony counts. The results are means of two to three experiments. Downloaded from

PGLYRP-1 had higher bactericidal activity for S. aureus at pH 6.4 than at neutral or alkaline pH (data not shown). These results are consistent with the location of PGLYRP-1 in PMNЈs granules (13–

15), which are usually acidic or become acidic following phago- http://www.jimmunol.org/ cytosis. PGLYRP-3, PGLYRP-4, and PGLYRP3:4 were most stable and most bactericidal for S. aureus and other bacteria at pH 7.6 (data not shown), which falls within the range of pH found in the intestine, saliva, and sweat (30). Because NaCl inhibits bactericidal activity of many mammalian antibacterial peptides (e.g., defensins) (16, 17), we next tested whether bactericidal activity of PGLYRPs was also salt sensitive. However, bactericidal activity of PGLYRPs for S. aureus and other bacteria was not inhibited by NaCl and was optimal at 100– by guest on September 29, 2021 150 mM NaCl (data not shown), which falls within the range of NaCl concentrations found in the intestine, saliva, sweat, and other FIGURE 3. PGLYRPs are bactericidal for several Gram-negative bac- body fluids (30). teria. The indicated bacteria were incubated with 100–150 ␮g/ml of the indicated PGLYRPs or Fc IgG fragment as a control purified and assayed PGLYRPs are bactericidal for Gram-negative bacteria in the ϩ in the presence of 5 ␮MZn2ϩ. The numbers of bacteria were determined presence of Zn2 by colony counts. The results are means of two to three experiments. In our previous study, human PGLYRPs purified and tested in the presence of Ca2ϩ (without Zn2ϩ) were only bacteriostatic but not bactericidal for Gram-negative bacteria (10). Therefore, we next In our previous study, PGLYRPs purified and tested in the pres- tested whether PGLYRPs purified and assayed in the presence of ence of Ca2ϩ (without Zn2ϩ) were bactericidal for some patho- Zn2ϩ and/or other cations present in body secretions had higher genic Gram-positive bacteria (S. aureus and L. monocytogenes) activity against Gram-negative bacteria. PGLYRP-1, PGLYRP-3, and for nonpathogenic transient Gram-positive bacteria (Bacillus and PGLYRP3:4 purified and tested in the presence of Zn2ϩ were sp. and L. acidophilus), but only bacteriostatic for other pathogenic highly bactericidal for E. coli (reduced the numbers of bacteria by (S. pyogenes) or normal flora Gram-positive bacteria (10). Because Ͼ5 logs in less than an hour), whereas at the same concentration, PGLYRPs purified and assayed in the presence of Zn2ϩ were more PGLYRP-4 was less active and required6htoreduce the numbers active against S. aureus, we next tested whether these PGLYRPs of bacteria by 3 logs (Fig. 2). PGLYRPs were less bactericidal for were also more active against other Gram-positive bacteria. In- E. coli in the presence of Zn2ϩ,Ca2ϩ, and Cu2ϩ or Zn2ϩ and Ca2ϩ deed, all four PGLYRPs (PGLYRP-1, PGLYRP-3, PGLYRP-4, and were not bactericidal and only bacteriostatic for E. coli (Fig. 2) and PGLYRP3:4) purified and assayed in the presence of Zn2ϩ and other Gram-negative bacteria (data not shown) in the absence instead of Ca2ϩ also had higher bactericidal activity against L. of Zn2ϩ.Ca2ϩ had an inhibitory effect on the bactericidal activity monocytogens, B. subtilis, and L. acidophilus (data not shown) and of PGLYRPs for Gram-negative bacteria, as all PGLYRPs purified were bactericidal for S. pyogenes at 100–150 ␮g/ml (Fig. 1E). At and assayed in the presence of both Ca2ϩ and Zn2ϩ were less 100–150 ␮g/ml, they were still only bacteriostatic for normal flora bactericidal than PGLYRPs purified and assayed in the presence of Gram-positive bacteria (M. luteus and E. faecalis), and at 300 ␮g/ Zn2ϩ alone (Fig. 2). ml, they were bactericidal for M. luteus but not for E. faecalis (data All other Gram-negative bacteria tested (S. enterica, S. sonnei, P. not shown). aeruginosa, and P. vulgaris) were also sensitive to the bactericidal PGLYRP-1 was most stable at pH 4.5, but its bactericidal ac- activity of PGLYRP-1, PGLYRP-3, and PGLYRP3:4 purified and tivity could not be assayed at such a low pH because pH Ͻ 6 assayed in the presence of Zn2ϩ and were less sensitive to PGLYRP-4 decreased the viability of S. aureus and other bacteria by itself. (Fig. 3), with S. enterica, S. sonnei, and P. aeruginosa being as 3120 BACTERICIDAL PGRPs

PGLYRP-3:4 (0.71 ␮M), and 200 ␮g/ml for PGLYRP-4 (1.7 ␮M), compared with ϳ45 ␮g/ml for S. aureus for PGLYRP-1 (1.0 ␮M), PGLYRP-3 (0.50 ␮M), PGLYRP-3:4 (0.46 ␮M), and PGLYRP-4 ␮ (0.39 M) (Fig. 4). The LD99 for PLA2, which is one of the most active antibacterial peptides (18–21), for E. coli and S. aureus were 30 ␮g/ml (18 ␮M) and 40 ␮g/ml (24 ␮M), respectively. The requirement for Zn2ϩ was further tested by removal of Zn2ϩ with EGTA, which has a high afﬁnity for Zn2ϩ (the log stability constant for Zn2ϩ is 12.9, compared with 11.0 for Ca2ϩ). Treatment of PGLYRPs with EGTA (in a buffer with 5 ␮MZn2ϩ only with no other divalent cations) abolished the bactericidal activities of PGLYRPs for Gram-negative bacteria (Fig. 5A), thus conﬁrming the requirement for Zn2ϩ for their bactericidal activity. N-Glycosylation of all PGLYRPs was also required for their bactericidal effect on Gram-negative bacteria (similar to Gram- positive bacteria; Ref. 10), as treatment of PGLYRPs with N-glycosidase abolished their bactericidal activity (Fig. 5B). Boiling of PGLYRPs for 30 min also abolished the bactericidal activity of all

PGLYRPs for Gram-negative bacteria (data not shown). Downloaded from

Bactericidal effect of PGLYRPs on Gram-negative bacteria does not involve permeabilization of cell membrane We have previously shown that PGLYRPs, in contrast to antibac- FIGURE 5. Bactericidal activity of PGLYRPs for E. coli requires Zn2ϩ terial peptides, do not permeabilize membranes in Gram-positive

and N-glycosylation. E. coli was incubated with the indicated PGLYRPs bacteria and kill them by targeting their cell wall (10, 11). There- http://www.jimmunol.org/ (100–150 ␮g/ml PGLYRP-1, PGLYRP-3, or PGLYRP-3:4 or 300 ␮g/ml fore, we next tested whether PGLYRPs kill Gram-negative bacte- PGLYRP-4) or Fc IgG fragment as control (all puriﬁed in the presence of ria by permeabilizing their cell membranes or by targeting their Zn2ϩ) treated or untreated as indicated with EGTA or N-glycosidase. The cell walls. We compared the effects of PGLYRPs on Gram-nega- numbers of bacteria were determined by colony counts. The results are means of two experiments. tive bacteria to the effects of membrane-permeabilizing antibacterial peptide (magainin) and an inhibitor of peptidoglycan biosyn- thesis (ampicillin). We measured membrane permeabilization in sensitive or more sensitive than E. coli. Similar high sensitivity real time using a membrane-nonpermeable dye (SYTOX-green, was also observed for the highly pathogenic encapsulated invasive which ﬂuoresces upon binding to DNA, after entering cells with by guest on September 29, 2021 strain E. coli O18:K1:H7:Bort (data not shown). The LD99 (the damaged membranes) and compared it with bacterial killing mea- concentration that kills 99% of bacteria) for E. coli was 30 ␮g/ml sured by colony counts. Vertebrate antibacterial peptides (e.g., ma- ␮ ␮ ␮ for PGLYRP-1 (0.68 M) and PGLYRP-3 (0.34 M), 70 g/ml for gainin, PLA2, and defensins) kill bacteria by traversing bacterial

FIGURE 6. Bactericidal activity of PGLYRPs for E. coli does not involve permeabilization of cytoplasmic membrane. Membrane permeabilization and killing of E. coli incubated with magainin (200 ␮g/ml), ampicillin (500 ␮g/ml), or PGLYRPs (200 ␮g/ml) were measured in medium without or with 0.5 M sucrose. The results are means of two to three experiments. The Journal of Immunology 3121 cell wall and permeabilizing bacterial cytoplasmic membrane (16, 17). Therefore, the kinetics of membrane permeabilization by antibacterial peptides correlates with the kinetics of bacterial killing and such an expected result was obtained in magainin-treated E. coli (Fig. 6). Similar results were obtained in the medium with 0.5 M sucrose, which protects bacteria from killing by antibiotics that inhibit cell wall synthesis (because bacteria can resynthesize their cell wall), but not from killing by membrane permeabilization, as conﬁrmed by our results with magainin- treated E. coli (Fig. 6). Antibiotics (such as ampicillin) target cell wall and kill bacteria by inhibiting peptidoglycan synthesis. However, ampicillin does not cause early permeabilization of bacterial cytoplasmic membranes because ampicillin-treated bacteria only die when they start to grow and are unable to synthesize peptidoglycan. Therefore, in ampicillin-treated bacteria, membrane permeabilization is substan- tially delayed until the bacteria have grown without synthesizing peptidoglycan, which we also observed in our experiments in ampicillin-treated E. coli (Fig. 6). Also as expected, killing of E. coli Downloaded from by ampicillin and delayed permeabilization of bacterial membranes were prevented in a medium with 0.5 M sucrose (Fig. 6). PGLYRPs rapidly killed E. coli (within 1 h, as measured by colony counts) but did not permeabilize their cytoplasmic membranes for the entire 6-h observation period (Fig. 6). Killing of E.

coli by PGLYRPs was not prevented by 0.5 M sucrose (Fig. 6). http://www.jimmunol.org/ These results contrast the ability of hyperosmotic medium with sucrose to prevent killing of Gram-positive bacteria by PGLYRPs, or by peptidoglycan-lytic enzymes, or by penicillin (Fig. 5 in Ref. 10) and killing of E. coli by ampicillin (Fig. 6). These results indicate that the effect of PGLYRPs on Gram-negative bacteria becomes irreversible in Ͻ1 h, i.e., from this point on the bacteria are unable to grow (hence, no formation of colonies when bacteria are plated). These results demonstrate that, in contrast to antibacterial peptides, PGLYRPs do not kill Gram-negative bacteria by permeabi- by guest on September 29, 2021 lizing their cytoplasmic membranes. These results also indicate that PGLYRPs do not kill bacteria by lysing their call wall because cell-wall lytic enzymes cause early permeabilization of cytoplas- FIGURE 7. Synergistic killing of Gram-positive bacteria by PGLYRPs mic membrane due to hypotonic shock and osmotic lysis, which and antibacterial peptides. S. aureus or L. monocytogenes was incubated (together with the killing) is preventable in hyperosmotic medium ␮ ␮ with subbactericidal concentrations of PLA2 (10 g/ml), HNP-1 (25 g/ with sucrose (10). By contrast, killing of Gram-negative bacteria ml), and PGLYRPs (25 ␮g/ml) alone or in combination (as indicated), and by PGLYRPs is not prevented in 0.5 M sucrose (Fig. 6), which the numbers of bacteria were determined by colony counts. The results are resembles the combined synergistic effect of peptidoglycan-lytic means of two to three experiments. enzymes and PGLYRPs on Gram-positive bacteria, whose killing under these conditions is also not prevented by sucrose (10). Fig. 9, p Ͻ 0.01). Similarly, incubation of E. coli with subbac- Synergistic killing of Gram-positive and Gram-negative bacteria tericidal concentrations of PGLYRP-1 or PGLYRP-3 and by PGLYRPs and antibacterial peptides ␤ PLA2, -defensin (HBD-3), or BPI resulted in 1,000–20,000 Several host defenses usually work together to eliminate bacteria times enhancement of bacterial killing compared with the kill- and they are often synergistic. PGLYRP-1 is present in PMNs’ ing by each compound alone, thus also demonstrating a strong granules (13–15), and PGLYRP-3 and PGLYRP-4 are present on synergistic killing of Gram-negative bacteria (Fig. 8, p Ͻ 0.001; the skin, mucous membranes, and in body secretions (10) together and Fig. 9, p Ͻ 0.01). with antibacterial peptides, such as PLA2, defensins, and BPI (16– In similar experiments, S. aureus was also synergistically killed 22). If two antibacterial agents have different bactericidal mecha- by subbactericidal concentrations of PGLYRP-3 or PGLYRP-4 nisms, they are likely to kill bacteria synergistically. Because an- and lysostaphin (a peptidoglycan-lytic enzyme) (100–1000 times tibacterial peptides kill bacteria by permeabilizing their enhanced killing, data not shown), which further conﬁrms different membranes (16–22) and PGLYRPs by targeting bacterial cell mechanisms of bacterial killing by PGLYRPs and lysostaphin. walls (10, 11), here we tested the hypothesis that PGLYRPs may Similarly, S. aureus was also synergistically killed by subbacteri- kill bacteria synergistically with antibacterial peptides. cidal concentrations of penicillin and lysostaphin (100 times en- When S. aureus or L. monocytogenes was incubated with hanced killing, data not shown), conﬁrming that another pair of subbactericidal concentrations of PGLYRP-1, PGLYRP-3, or bactericidal agents with different mechanisms of action also syn- ␣ PGLYRP-4 and PLA2 or -defensin (HNP-1), the killing of these ergistically kills bacteria. bacteria was enhanced 300–100,000 times, compared with the kill- To further test whether synergistic effect of PGLYRPs and an- ing by each compound alone, thus demonstrating a strong syner- tibacterial peptides was due to their action on different parts of the Ͻ gistic killing of Gram-positive bacteria (Fig. 7, p 0.001; and bacterial cell, we constructed a PLA2-PGLYRP-1 fusion protein 3122 BACTERICIDAL PGRPs

tein could only act on one target (cell wall or cytoplasmic membrane). However, if both proteins have the same target (e.g., cell membrane), the fusion protein could still show enhanced activity similar to the two unfused proteins. Our results showed that the

PLA2-PGLYRP-1 fusion protein did not have the enhanced bactericidal activity of the unfused PGLYRP-1 and PLA2 added together for S. aureus and E. coli (Fig. 9). As additional controls, we tested PGLYRP-1-Fc fusion protein alone, which had the same

bactericidal activity as PGLYRP-1, and PLA2-Fc fusion protein alone, which had the same bactericidal activity as PLA2 (data not shown), demonstrating that fusing PGLYRP-1 or PLA2 with another protein or increasing the size of PGLYRP-1 or PLA2 does not inhibit their bactericidal activities. These results further

conﬁrm that PGLYRPs and PLA2 target different parts of bacterial cell. In conclusion, our results demonstrate strong synergistic killing

of bacteria by PGLYRPs and PLA2, defensins, BPI, or a peptidoglycan-lytic enzyme, and further conﬁrm that PGLYRPs and

these antibacterial peptides kill bacteria by different mechanisms. Downloaded from

Discussion We have demonstrated here that bactericidal activity of human PGLYRP-1, PGLYRP-3, PGLYRP-4, and PGLYRP-3:4 for both Gram-positive and Gram-negative bacteria requires Zn2ϩ. For kill- 2ϩ ing of Gram-negative bacteria, Zn cannot be replaced by other http://www.jimmunol.org/ cations, but for killing of Gram-positive bacteria Zn2ϩ, can be partially replaced by Ca2ϩ. This Zn2ϩ dependence explains why in our previous experiments PGLYRPs purified in the presence of Ca2ϩ (without Zn2ϩ) were not bactericidal for Gram-negative bacteria and were only bactericidal for some Gram-positive FIGURE 8. Synergistic killing of E. coli by PGLYRPs and antibacterial bacteria (10). ϩ peptides. E. coli was incubated with subbactericidal concentrations of Physiologic concentrations Zn2 (ϳ2–5 ␮M) were sufficient for ␮ ␮ ␮ 2ϩ PLA2 (12.5 g/ml), HBD-3 (8 g/ml), BPI (15 g/ml), and PGLYRPs full bactericidal activity of human PGLYRPs. Zn is present in (12.5 ␮g/ml) alone or in combination (as indicated), and the numbers of sweat at ϳ17 ␮M, in saliva at ϳ2 ␮M, and in plasma at ϳ15 ␮M by guest on September 29, 2021 bacteria were determined by colony counts. The results are means of two (30). Zn2ϩ is an essential component of many proteins, such as to three experiments. enzymes and transcription factors, and is required for their structure and function (31). The concentration of extracellular and intracellular Zn2ϩ is tightly regulated by Zn2ϩ transporters and and compared its bactericidal activity to the activity of PGLYRP-1 Zn2ϩ-binding proteins (31–33). Proper function of host immunity ϩ ϩ or PLA2 alone or a mixture of PGLYRP-1 and PLA2.If requires Zn2 because Zn2 -deficient individuals have increased ϩ PGLYRP-1 and PLA2 indeed act on different parts of the bacterial susceptibility to infections and Zn2 supplementation restores or cell (cell wall and cytoplasmic membrane), the fusion protein enhances resistance to infections and the functions of innate and would not be expected to have a synergistic effect in contrast to the acquired immunity, although the exact mechanisms involved are two proteins not fused and added together because the fusion pro- mostly unknown (34–40). Several peptidoglycan- or polysaccha- ride-lytic enzymes contain Zn2ϩ (43–46), and some antimicrobial peptides require Zn2ϩ for their antimicrobial activity (47–52). However, the role of Zn2ϩ and other divalent cations in innate immunity is complex because, in addition to its enhancing effect, Zn2ϩ can also inhibit the activity of some antimicrobial proteins, especially Ca2ϩ-binding proteins psoriasin (26) and calprotectin (27–29). Although the exact mechanism of antimicrobial activity of psoriasin and calprotectin are unknown, they are thought to be antimicrobial through Zn2ϩ deprivation because they bind Zn2ϩ, and Zn2ϩ supplementation reduces their antimicrobial activity (26–29). However, Zn2ϩ may also inhibit antimicrobial effect of calprotectin by inducing a change in its conformation, rather than by reversing microbial Zn2ϩ deprivation (27). Therefore, proper compartmentalization of Zn2ϩ and antimicro- FIGURE 9. Synergistic killing of S. aureus or E. coli by PLA plus 2 bial proteins may be important. For example, PGLYRP-3 and PGLYRP-1 but not by PLA2-PGLYRP-1 fusion protein. S. aureus or E. 2ϩ ␮ PGLYRP-4 (which require Zn for activity) are present in sweat coli was incubated with subbactericidal concentrations of PLA2 (12.5 g/ ␮ glands, salivary glands, and the eye (10), and calprotectin and pso- ml) or PGLYRP-1 (12.5 g/ml) alone or with both PLA2 and PGLYRP-1 ϩ ␮ ␮ riasin (whose activity is inhibited by Zn2 ) are not (26). Moreover, (12.5 g/ml each) or with the PLA2-PGLYRP-1 fusion protein (25 g/ml), and the numbers of bacteria were determined by colony counts. The results although both calprotectin and PGLYRP-1 are expressed in neu- are means of two experiments. trophils, calprotectin is present in the cytosol and primary and The Journal of Immunology 3123 secondary granules (53), and PGLYRP-1 is present in the tertiary when both compounds are present at subbactericidal concentrations. granules (13). Also, concentrations of Zn2ϩ that enhance or inhibit This synergism is likely to greatly enhance antibacterial defenses in microbial killing are different: PGLYRPs require 2–5 ␮MZn2ϩ vivo because both PGLYRPs and antibacterial peptides are found in for killing bacteria, whereas inhibition of antimicrobial effect of the same locations in the body, such as PMN granules (PGLYRP-1, ␮ 2ϩ calprotectin and psoriasin requires 10–30 MZn , and therefore, PLA2, defensins, and BPI), skin (PGLYRP-3, PGLYRP-4, and 2ϩ at low physiological Zn concentrations, all these proteins are defensins), eyes (PGLYRP-3, PGLYRP-4, and PLA2), and oral 2ϩ antimicrobial. Thus, proper control of Zn concentration is es- cavity and intestinal tract (PGLYRP-3, PGLYRP-4, PLA2, and sential for efficient antimicrobial immunity. In this context, it is defensins) (3, 10–22). interesting to note that activation of innate immunity TLRs by The mechanism of bacterial killing by PGLYRPs seems to bacteria regulates Zn2ϩ homeostasis in dendritic cells and induces be unique and different from other known antibacterial agents. net export of Zn2ϩ from the cells (54). This process is essential for PGLYRPs do not permeabilize cytoplasmic membranes, which the activation of dendritic and other immune cells (54), and it is is the main mechanism of killing by antibacterial peptides, such 2ϩ tempting to speculate that it may also provide extracellular Zn as defensins or PLA2 (16–22). PGLYRPs also likely do not kill for antimicrobial peptides and PGLYRPs. This would be a feasible bacteria by hydrolyzing peptidoglycan because peptidoglycan- mechanism because intracellular Zn2ϩ concentration is ϳ150–300 hydrolytic activity of PGLYRP-1, PGLYRP-3, and PGLYRP-4 ␮M (33), and extracellular PGLYRPs require ϳ2–5 ␮MZn2ϩ for could not be demonstrated (6, 10), and because treatment of bactericidal activity. bacteria with peptidoglycan-lytic enzymes results in osmotic The requirement for Zn2ϩ for bactericidal activity of lysis of bacteria, which causes rapid membrane permeabiliza-

PGLYRP-1, PGLYRP-3, PGLYRP-4, and PGLYRP-3:4 is likely tion that is prevented in hyperosmotic medium (10). Such mem- Downloaded from through its effect on PGLYRPs and not on the bacteria (perhaps to brane permeabilization is not observed in bacteria treated with stabilize PGLYRPs) because Zn2ϩ presence is required both dur- PGLYRPs (Ref. 10 and this article). If PGLYRPs kill bacteria ing the purification of PGLYRPs and in the bactericidal assay. by hydrolysis of peptidoglycan, the rate of hydrolysis would PGLYRP-2 and other amidase-active PGRPs (e.g., Drosophila have to be very slow, still not noticeable after6hofexposure PGRP-SC1 and PGRP-LB) have a Zn2ϩ-binding site and require of bacteria to PGLYRPs. In our experiments, E. coli cell mem- 2ϩ Zn for their amidase activity (5, 6, 55, 56). These amidase-active brane was still not permeabilized after6hofincubation with http://www.jimmunol.org/ PGRPs have four Zn2ϩ-binding amino acids in their enzyme ac- PGLYRPs (Fig. 6), and if peptidoglycan was hydrolyzed, this tive site (5, 6, 55, 56). Other PGRPs, however, that do not have would have resulted in osmotic lysis of bacteria and membrane amidase activity (such as mammalian PGLYRP-1, PGLYRP-3, permeabilization. Therefore, to kill bacteria, such a slow pep- and PGLYRP-4), do not have all four Zn2ϩ-binding amino acids tidoglycan hydrolysis would have to be coupled with essentially conserved: they have Ser instead of Cys in the position corre- irreversible binding of PGLYRPs to bacterial cell wall, which sponding to the Zn2ϩ-binding Cys530 in human PGLYRP-2 (5, would result in delayed, rather than immediate killing of bac- 6). Thus, it is not known where the Zn2ϩ binding site in mam- teria. Long exposure of bacteria to bovine PGLYRP-1 did result malian PGLYRP-1, PGLYRP-3, and PGLYRP-4 may be. These in bacterial lysis, but it was concluded that bacterial lysis was PGLYRPs will have to be crystallized in the presence of Zn2ϩ to not required and was not responsible for bacterial killing be- by guest on September 29, 2021 unequivocally determine whether they bind Zn2ϩ and where the cause bacteriolytic activity of bovine PGLYRP-1 was heat la- Zn2ϩ binding site is located. bile, and bactericidal activity was heat stable (15). Our current and previous results demonstrate that human Measuring killing by colony counts on plates does not indicate PGLYRP-1, PGLYRP-3, PGLYRP-4, and PGLYRP-3:4 are bac- that at the time of bacterial dilution and plating the bacteria are tericidal proteins with broad bactericidal activity against most dead; it only indicates that at the time of plating the cidal effect is Gram-positive and Gram-negative bacteria, both pathogenic and irreversible—bacteria may die hours later or may be simply unable nonpathogenic. Only some normal flora Gram-positive bacteria to divide and form a colony. Many bactericidal antibiotics (␤- show partial resistance to bactericidal effects of these PGLYRPs, lactams, aminglycosides, etc.) have a similar effect—they irrevers- which is likely an adaptation of these bacteria needed for colo- ibly bind to their target, but bacteria actually die later, when they nizing the skin, mucous membranes, or the intestine, where are unable to grow or divide. For this reason, in ampicillin-treated PGLYRP-3 and PGLYRP-4 are produced, as discussed previously E. coli by5hofexposure, the cell membrane was beginning to be (10, 11). PGLYRPs are bacteriostatic for these normal flora bac- permeabilized indicating osmotic lysis of growing bacteria (Fig. teria, and this bacteriostatic effect may prevent the overgrowth of 6). The kinetics of killing and membrane permeabilization of normal flora bacteria on skin and mucous membranes. Gram-positive bacteria by penicillin and PGLYRPs were similar, Bactericidal activity of PGLYRP-1, PGLYRP-3, and PGLYRP-4 and the killing of Gram-positive bacteria by both compounds was for both Gram-positive and Gram-negative bacteria is likely to be a prevented in hyperosmotic medium (10). Killing of ampicillin- general characteristic of these PGLYRPs in all mammals because treated E. coli was prevented in hyperosmotic medium, as ex- their sequences are highly conserved, because they show similar pat- pected (Fig. 6), which reflects the ability of E. coli (similar to terns of expression, and because a bovine PGLYRP-1 ortholog has S. aureus in our previous experiments; Ref. 10) to resynthesize the been already shown to kill both Gram-positive and Gram-negative transpeptidases irreversibly inactivated by ampicillin (after dilut- bacteria (14, 15). Although Zn2ϩ was not used in the purification or ing and plating the bacteria in hyperosmotic medium). However, bactericidal assay of bovine PGLYRP-1, this was a native (not re- killing of E. coli by PGLYRPs was not prevented in hyperosmotic combinant) protein purified from the leukocyte granules, and it prob- medium (Fig. 6), similar to the combined killing effect of ably retained Zn2ϩ that was likely bound to it in vivo. Consistent with PGLYRPs and lysostaphin on S. aureus (Fig. 5 in Ref. 10). our results on human PGLYRP-1, PGLYRP-3, PGLYRP-4, and Peptidoglycan is a polymer of ␤(1-4)-linked N-acetylglu- PGLYRP3:4, bactericidal activity of bovine PGLYRP-1 for Gram- cosamine and N-acetylmuramic acid (MurNAc) cross-linked by negative bacteria was also inhibited by Ca2ϩ (15). short peptides containing alternating L-and D-amino acids (3). We have also demonstrated that human PGLYRPs synergisti- Each PGRP domain has a ligand-binding groove that binds pepti- cally kill both Gram-positive and Gram-negative bacteria together doglycan through its MurNAc-tripeptide or MurNAc-pentapeptide ␣ ␤ with antibacterial peptides, such as PLA2 and - and -defensins, (56–62). Binding of mammalian PGLYRPs to peptidoglycan is 3124 BACTERICIDAL PGRPs likely involved in bactericidal activity of these PGLYRPs because References exogenously added peptidoglycan inhibits bactericidal activity of 1. Kang, D., G. Liu, A. Lundstrom, E. Gelius, and H. Steiner. 1998. A peptidogly- PGLYRPs for Gram-positive bacteria (10). However, killing of can recognition protein in innate immunity conserved from insects to mammals. Proc. Natl. Acad. Sci. USA 95: 10078–10082. Gram-negative bacteria by PGLYRPs likely involves an additional 2. Werner, T., G. Liu, D. Kang, S. Ekengren, H. Steiner, and D. Hultmark. 2000. A step because, in Gram-negative bacteria, peptidoglycan is not ex- family of peptidoglycan recognition proteins in the fruit fly Drosophila melano- posed on the cell surface since it is located underneath the LPS- gaster. Proc. Natl. Acad. Sci. USA 97: 13772–13777. 3. Dziarski, R., and D. Gupta. 2006. The peptidoglycan recognition proteins containing outer membrane. Killing of Gram-negative bacteria (PGRPs). Genome Biol. 7: 1–13. 232. likely involves initial binding of PGLYRPs to the outer membrane 4. Liu, C., Z. Xu, D. Gupta, and R. Dziarski. 2001. Peptidoglycan recognition pro- because killing of Gram-negative bacteria by bovine PGLYRP-1 teins: a novel family of four human innate immunity pattern recognition molecules. J. Biol. Chem. 276: 34686–34694. (15) and by human PGLYRP-1, PGLYRP-3, and PGLYRP-4 (M. 5. Gelius, E., C. Persson, J. Karlsson, and H. Steiner. 2003. A mammalian pepti- Wang and R. Dziarski, unpublished results) is inhibited by LPS. In doglycan recognition protein with N-acetylmuramoyl-L-alanine amidase activity. Gram-positive bacteria, the initial binding of PGLYRPs to bacteria Biochem. Biophys. Res. Commun. 306: 988–994. 6. Wang, Z. M., X. Li, R. R. Cocklin, M. Wang, M. Wang, K. Fukase, S. Inamura, may involve lipoteichoic acid (LTA), in addition to peptidogly- S. Kusumoto, D. Gupta, and R. Dziarski. 2003. Human peptidoglycan recognition can, because LTA also inhibits killing of bacteria by bovine protein-L is an N-acetylmuramoyl-L-alanine amidase. J. Biol. Chem. 278: 49044–49052. PGLYRP-1 (15) and human PGLYRPs, although to a lesser extent 7. Zhang, Y., L. van der Fits, J. S. Voerman, M. J. Melief, J. D. Laman, M. Wang, than peptidoglycan (M. Wang and R. Dziarski, unpublished re- H. Wang, M. Wang, X. Li, C. D. Walls, et al. 2005. Identification of serum sults). LTA and LPS bind to PGLYRPs (in addition to peptidogly- N-acetylmuramoyl-L-alanine amidase as liver peptidoglycan recognition protein 2. Biochim. Biophys. Acta 1752: 34–46. can) (10, 12, 15), but the location of LPS- and LTA-binding sites 8. Wang, H., D. Gupta, X. Li, and R. Dziarski. 2005. Peptidoglycan recognition Downloaded from in PGLYRPs is unknown. These sites may be outside the pepti- protein 2 (N-acetylmuramoyl-L-ala amidase) is induced in keratinocytes by bac- doglycan-binding groove, as suggested by the recently shown teria through p38 kinase pathway. Infect. Immun. 73: 7216–7225. 9. Li, X., S. Wang, H. Wang, and D. Gupta. 2006. Differential expression of pep- binding of ligands outside the peptidoglycan-binding groove in tidoglycan recognition protein 2 in the skin and liver requires different transcrip- Drosophila PGRP-LE and PGRP-LC (62, 63). In addition, PGRPs tion factors. J. Biol. Chem. 281: 20738–20748. have a hydrophobic domain on the opposite side of the molecule 10. Lu, X., M. Wang, J. Qi, H. Wang, X. Li, D. Gupta, and R. Dziarski. 2006. Peptidoglycan recognition proteins are a new class of human bactericidal pro- from the peptidoglycan-binding groove, which was previously hy- teins. J. Biol. Chem. 281: 5895–5907. pothesized to interact with signal transduction molecules in insects 11. Dziarski, R., and D. Gupta. 2006. Mammalian PGRPs: novel antibacterial pro- http://www.jimmunol.org/ (56). In mammalian PGLYRPs, however, this hydrophobic domain teins. Cell. Microbiol. 8: 1056–1069. 12. Liu, C., E. Gelius, G. Liu, H. Steiner, and R. Dziarski. 2000. Mammalian pep- may play a role in interaction of PGLYRPs with bacteria. Analo- tidoglycan recognition protein binds peptidoglycan with high affinity, is ex- gous situation is found in bacteriophage amidases, which lyse pep- pressed in neutrophils, and inhibits bacterial growth. J. Biol. Chem. 275: 24490–24499. tidoglycan and allow bacteriophages to be released from bacteria. 13. Dziarski, R., K. A. Platt, E. Gelius, H. Steiner, and D. Gupta. 2003. Defect in One portion of a bacteriophage amidase is the binding domain neutrophil killing and increased susceptibility to infection with non-pathogenic specific for a cell wall component often different from peptidogly- Gram-positive bacteria in peptidoglycan recognition protein-S (PGRP-S)-deficient mice. Blood. 102: 689–697. can (e.g., choline in pneumococcal cell wall teichoic acid), 14. Tydell, C. C., N. Yount, D. Tran, J. Yuan, and M. Selsted. 2002. Isolation, characterization, and antimicrobial properties of bovine oligosaccharide-binding whereas another portion contains the peptidoglycan-lytic amidase by guest on September 29, 2021 domain (64, 65). protein. J. Biol. Chem. 277: 19658–19664. 15. Tydell, C. C., J. Yuan, P. Tran, and M. E. Selsted. 2006. Bovine peptidoglycan Moreover, binding of PGRPs to bacteria may involve multiple recognition protein-S: antimicrobial activity, localization, secretion, and binding interactions because both human and insect PGRPs use a dual properties. J. Immunol. 176: 1154–1162. 16. Ganz, T. 2003. Defensins: antimicrobial peptides of innate immunity. Nat. Rev. recognition strategy of binding to both MurNAc-pentapeptide and Immunol. 9: 710–720. to the peptide cross-bridge (60). Such binding requires interaction 17. Zasloff, M. 2002. Antimicrobial peptides of multicellular organisms. Nature 415: of at least two PGRP domains with peptidoglycan. This dual recog- 389–395. nition could be accomplished by PGRPs with two PGRP domains 18. Harwig, S. S. L., L. Tan, X. D. Qu, Y. Cho, P. B. Eisenhauer, and R. I. Lehrer. 1995. Bactericidal properties of murine intestinal phospholipase A2. J. Clin. In- and/or by dimeric PGRPs. PGLYRP-1 has one PGRP domain, but it vest. 95: 603–610. forms disulfide-linked dimers (10), whereas each molecule of 19. Weinrauch, Y., P. Elsbach, L. M. Madsen, A. Foreman, and J. Weiss. 1996. The Staphylococcus aureus PGLYRP-3 and PGLYRP-4 contains two PGRP domains, which are potent anti- activity of a sterile rabbit inflammatory fluid is due to a 14-kD phospholipase A2. J. Clin. Invest. 97: 250–257. not identical. Moreover, PGLYRP-3 and PGLYRP-4 form homo- or 20. Koduri, R. S., J. O. Gronroos, V. J. O. Laine, C. Le Calvez, G. Lambeau, heterodimers (10). Thus, PGLYRP-1 dimers have two identical T. J. Nevalainen, and M. H. Gelb. 2002. Bactericidal properties of human and murine groups I, II, V, X, and XII secreted phospholipases A2. J. Biol. Chem. PGRP domains, whereas PGLYRP-3 and PGLYRP-4 homodimers 277: 5849–5857. have two pairs of identical PGRP domains, and PGLYRP3:4 het- 21. Qu, X. D., and R. I. Lehrer. 1998. Secretory phospholipase A2 is the principal erodimers have four different PGRP domains. Therefore, two different bactericide for staphylococci and other Gram-positive bacteria in tears. Infect. Immun. 66: 2791–2797. PGRP domains and formation of homo- and heterodimers afford 22. Dziarski, R. 2006. Innate immunity. In Shaechter’s Mechanisms of Microbial PGLYRPs with considerable diversity of specificities and allow Disease 6, 4th Ed. R. Engelberg and V. DiRita, eds. Lippincott, Williams & highly avid binding to bacteria through interaction with multiple bac- Wilkins, Philadelphia. pp. 66–89. 23. Schittek, B., R. Hipfel, B. Sauer, J. Bauer, H. Kalbacher, S. Stevanovic, terial components. This avid binding may be responsible for the irre- M. Schirle, K. Schroeder, N. Blin, F. Meier, et al. 2001. Dermcidin: a novel versible bactericidal effect of mammalian PGLYRPs. human antibiotic peptide secreted by sweat glands. Nat. Immunol. 2: 1133–1137. 24. Nizet, V., T. Ohtake, X. Lauth, J. Trowbridge, J. Rudisill, R. A. Droschner, In conclusion, our results demonstrate that human PGLYRPs are V. Pestonjamasp, J. Piralno, K. Huttner, and R. L. Gallo. 2001. Innate antimi- a novel class of innate immunity pattern recognition and effector crobial peptide protects the skin from invasive bacterial infection. Nature 414: molecules with Zn2ϩ-dependent broad spectrum bactericidal ac- 454–457. 25. Harder, J., and J. M. Schroder. 2002. RNase 7, a novel innate immune defense tivity against both Gram-positive and Gram-negative bacteria. At antimicrobial protein of human healthy skin. J. Biol. Chem. 277: 46779–46784. low subbactericidal concentrations, PGLYRPs kill bacteria syner- 26. Glaser, R., J. Harder, H. Lange, J. Bartels, E. Christophers, and J. M. Schroder. gistically with antibacterial peptides, such as PLA and defensins. 2005. Antimicrobial psoriasin (S100A7) protects human skin from Escherichia 2 coli infection. Nat. Immunol. 6: 57–64. 27. Murthy, A. R., R. I. Lehrer, S. S. Harwig, and K. T. Miyasaki. 1993. In vitro candidastatic properties of the human neutrophil calprotectin complex. J. Immu- nol. 151: 6291–6301. Disclosures 28. Clohessy, P. A., and B. E. Golden. 1995. Calprotectin-mediated zinc chelation as The authors have no financial conflict of interest. a biostatic mechanism in host defence. Scand. J. Immunol. 42: 551–556. The Journal of Immunology 3125

29. Sohnle, P. G., B. L. Hahn, and V. Santhanagopalan. 1996. Inhibition of Candida 50. Grogan, J., C. J. McKnight, R. F. Troxler, and F. G. Oppenheim. 2001. Zinc and albicans growth by calprotectin in the absence of direct contact with the organ- copper bind to unique sites of histatin 5. FEBS Lett. 491: 76–80. isms. J. Infect. Dis. 174: 1369–1372. 51. Cole, A. M., Y. H. Kim, S. Tahk, T. Hong, P. Weis, A. J. Waring, and T. Ganz. 30. Lentner, C., ed. 1981. Geigy Scientific Tables, Vol. 1, Ciba-Geigy, Basel. 2001. Calcitermin, a novel antimicrobial peptide isolated from human airway 31. Prasad, A. S. 1995. Zinc: an overview. Nutrition 11: 93–99. secretions. FEBS Lett. 504: 5–10. 32. Palmiter, R. D. 2004. Protection against zinc toxicity by metallothionein and zinc 52. Dashper, S. G., N. M. O’Brien-Simpson, K. J. Cross, R. A. Paolini, B. Hoffmann, transporter 1. Proc. Natl. Acad. Sci. USA 101: 4918–4923. D. V. Catmull, M. Malkoski, and E. C. Reynolds. 2005. Divalent metal cations 33. Palmiter, R. D., and S. D. Findley. 1995. Cloning and functional characterization increase the activity of the antimicrobial peptide kappacin. Antimicrob. Agents of a mammalian zinc transporter that confers resistance to zinc. EMBO J. 14: Chemother. 49: 2322–2328. 639–649. 53. Stroncek, D. F., R. A. Shankar, and K. M. Skubitz. 2005. The subcellular dis- 34. Shankar, A. H., and A. S. Prasad. 1998. Zinc and immune function: the biological tribution of myeloid-related protein 8 (MRP8) and MRP14 in human neutrophils. basis of altered resistance to infection. Am. J. Clin. Nutr. 68: 447S–463S. J. Transl. Med. 3: 36. 35. Ibs, K. H., and L. Rink. 2003. Zinc-altered immune function. J. Nutr. 133: 54. Kitamura, H., H. Morikawa, H. Kamon, M. Iguchi, S. Hojyo, T. Fukada, 1452S–1456S. S. Yamashita, T. Kaisho, S. Akira, M. Murakami, and T. Hirano. 2006. Toll-like 36. Erickson, K. L., E. A. Medina, and N. E. Hubbard. 2000. Micronutrients and receptor-mediated regulation of zinc homeostasis influences dendritic cell func- innate immunity. J. Infect. Dis. 182: S5–S10. tion. Nat. Immunol. 7: 971–977. 37. Walker, C. F., and R. E. Black. 2004. Zinc and the risk for infectious disease. Annu. Rev. Nutr. 24: 255–275. 55. Mellroth, P., J. Karlsson, and H. Steiner. 2003. A scavenger function for a Dro- 38. Fraker, P. J., and L. E. King. 2004. Reprogramming of the immune system during sophila peptidoglycan recognition protein. J. Biol. Chem. 278: 7059–7064. zinc deficiency. Annu. Rev. Nutr. 24: 277–298. 56. Kim, M. S., M. Byun, and B. H. Oh. 2003. Crystal structure of peptidoglycan 39. Brooks, W. A., M. Yunus, M. Santosham, M. A. Wahed, K. Nahar, S. Yeasmin, recognition protein LB from Drosophila melanogaster. Nat. Immunol. 4: and R. E. Black. 2004. Zinc for severe pneumonia in very young children: dou- 787–793. ble-blind placebo-controlled trial. Lancet. 363: 1683–1688. 57. Guan, R., A. Roychowdhury, B. Ember, S. Kumar, G.-J. Boons, and R. A. 40. Rostan, E. F., H. V. DeBuys, D. L. Madey, and S. R. Pinnell. 2002. Evidence Mariuzza. 2004. Structural basis for peptidoglycan binding by peptidoglycan rec- supporting zinc as an important antioxidant for skin. Int. J. Dermatol. 41: ognition proteins. Proc. Natl. Acad. Sci. USA 101: 17168–17173. 606–611. 58. Chang, C. I., S. Pili-Floury, M. Herve, C. Parquet, Y. Chelliah, B. Lemaitre, Downloaded from 41. Seilhamer, J. J., W. Pruzanski, P. Vadas, S. Plant, J. A. Miller, J. Kloss, and D. Mengin-Lecreulx, and J. Deisenhofer. 2004. A Drosophila pattern recognition L. K. Johnson. 1989. Cloning and recombinant expression of phospholipase A2 receptor contains a peptidoglycan docking groove and unusual L,D-carboxypep- present in rheumatic arthritic synovial fluid. J. Biol. Chem. 264: 5335–5338. tidase activity. PLoS Biol. 2: 1293–1302. 42. Cross, A., L. Asher, M. Seguin, L. Yuan, N. Kelly, C. Hammack, J. Sadoff, and 59. Kumar, S., A. Roychowdhury, B. Ember, Q. Wang, R. Guan, R. A. Mariuzza, and P. Gemski. 1995. The importance of a lipopolysaccharide-initiated, cytokine- G. J. Boons. 2005. Selective recognition of synthetic lysine and meso-diamin- mediated host defense mechanism in mice against extraintestinally invasive Esch- opimelic acid-type peptidoglycan fragments by human peptidoglycan recognition erichia coli. J. Clin. Invest. 96: 676–686. proteins I␣ and S. J. Biol. Chem. 280: 37005–37012.

43. Lessard, I. A., and C. T. Walsh. 1999. VanX, a bacterial D-alanyl-D-alanine dipep- 60. Swaminathan, C. P., P. H. Brown, A. Roychowdhury, Q. Wang, R. Guan, http://www.jimmunol.org/ tidase: resistance, immunity, or survival function? Proc. Natl. Acad. Sci. USA 96: N. Silverman, W. E. Goldman, G. J. Boons, and R. A. Mariuzza. 2006. Dual 11028–11032. strategies for peptidoglycan discrimination by peptidoglycan recognition proteins 44. Labischinski, H., P. Giesbrecht, E. Fischer, G. Barnickel, H. Bradaczek, J. M. (PGRPs). Proc. Natl. Acad. Sci. USA 103: 684–689. Frere, C. Houssier, P. Charlier, O. Dideberg, and J. M. Ghuysen. 1984. Study of 61. Guan, R., P. H. Brown, C. P. Swaminathan, A. Roychowdhury, G. J. Boons, and the Zn-containing DD-carboxypeptidase of Streptomyces albus G by small-angle R. A. Mariuzza. 2006. Crystal structure of human peptidoglycan recognition pro- X-ray scattering in solution. Eur. J. Biochem. 138: 83–87. tein I␣ bound to a muramyl pentapeptide from Gram-positive bacteria. Protein 45. Trayer, H. R., and C. E. Buckley III. 1970. Molecular properties of lysostaphin, Sci. 15: 1199–1206. a bacteriolytic agent speciﬁc for Staphylococcus aureus. J. Biol. Chem. 245: 4842–4846. 62. Chang, C. I., Y. Chelliah, D. Borek, D. Mengin-Lecreulx, and J. Deisenhofer. 46. Blair, D. E., A. W. Schuttelkopf, J. I. MacRae, and D. M. F. van Aalten. 2005. 2006. Structure of tracheal cytotoxin in complex with a heterodimeric pattern- Structure and metal-dependent mechanism of peptidoglycan deacetylase, a strep- recognition receptor. Science 311: 1761–1764.

tococcal virulence factor. Proc. Natl. Acad. Sci. USA 102: 15429–15434. 63. Lim, J. H., M. S. Kim, H. E. Kim, T. Yano, Y. Oshima, K. Aggarwal, W. E. by guest on September 29, 2021 47. Schlievert, P., W. Johnson, and R. P. Galask. 1976. Isolation of a low-molecular- Goldman, N. Silverman, S. Kurata, and B.-H. Oh. 2006. Structural basis for weight antibacterial system from human amniotic ﬂuid. Infect. Immun. 14: preferential recognition of diaminopimelic acid-type peptidoglycan by a subset of 1156–1166. peptidoglycan-recognition proteins. J. Biol. Chem. 281: 8286–8295. 48. LaForce, F. M., and D. S. Boose. 1984. Effect of zinc and phosphate on an 64. Garcia, P., A. C. Martin, and R. Lopez. 1997. Bacteriophages of Streptococcus antibacterial peptide isolated from lung lavage. Infect. Immun. 45: 692–696. pneumoniae: a molecular approach. Microb. Drug Resist. 3: 165–176. 49. Brogden, K. A., A. J. De Lucca, J. Bland, and S. Elliott. 1996. Isolation of an 65. Loefﬂer, J. M., D. Nelson, and V. A. Fischetti. 2001. Rapid killing of Strepto- ovine pulmonary surfactant-associated anionic peptide bactericidal for Pasteu- coccus pneumoniae with a bacteriophage cell wall hydrolase. Science 294: rella haemolytica. Proc. Natl. Acad. Sci. USA 93: 412–416. 2170–2172.