Selective Recognition of Synthetic Lysine and meso-Diaminopimelic Acid-type Peptidoglycan Fragments by Human Peptidoglycan Recognition Proteins Iα and S*

The interactions of a range of synthetic peptidoglycan derivatives with PGRP-Iα and PGRP-S have been studied in real-time using surface plasmon resonance. A dissociation constant of KD = 62 μm was obtained for the interaction of peptidoglycan recognition protein (PGRP)-Iα with the lysine-containing muramyl pentapeptide (compound 6). The normalized data for the lysine-containing muramyl tetra- (compound 5) and pentapeptide (compound 6) showed that these compounds have similar affinities, whereas a much lower affinity for muramyl tripeptide (compound 3) was measured. Similar affinities were obtained when the lysine moiety of the muramyl peptides was replaced by meso-diaminopimelic acid (DAP). Furthermore, the compounds that contained only a stem peptide (pentapeptide, compound 1) and (DAP-PP, compound 2) as well as muramyldipeptide (compound 3) exhibited no binding indicating that the muramyltripeptide (compound 4) is the smallest peptidoglycan fragment that can be recognized by PGRP-Iα. Surprisingly, PGRP-S derived significantly higher affinities for the DAP-containing fragments to similar lysine-containing derivatives, and the following dissociation constants were measured: muramylpentapeptide-DAP, KD = 104 nm; muramyltetrapeptide-DAP, 92.4 nm; and muramyltripeptide-DAP, 326 nm. The binding profiles were rationalized by using a recently reported x-ray crystal structure of PGRP-Iα with the lysine-containing muramyltripeptide (4).

sition of position 3 of the peptide chain, PGNs are classified as either L-lysine-type (Lys-type) or meso-diaminopimelic acid-type (DAP-type). The lysine-type, typical for Gram-positive bacteria, is normally connected to the D-Ala of another peptide chain by a short bridge varying in length and amino acid composition, depending on the bacteria. In the case of Gram-negative bacteria and Gram-positive bacilli, DAP-type is normally found as the third amino acid and is directly connected to D-Ala of another peptide chain.
Drosophila PGRP-SA has been shown to interact with lysine-type PGN, activating the Toll receptor pathway (23). On the other hand, PGRP-LC and PGRP-LE recognize DAP-type PGN activating the Imd/ Relish pathway (24 -28). PGRPs have high homology with the T7 lysozyme, a type 2 N-acetylmuramoyl-L-alanine amidase that hydrolyzes the bond between MurNAc and L-Ala of PGN (29). In this respect, Drosophila PGRP-SC1b and PGRP-LB have been shown to possess amidase activity. (30) Each of the four mammalian PGRPs (PGRP-L, PGRP-I␣, PGRP-I␤, and PGRP-S) is able to bind peptidoglycan; however, possible selectivities for lysine or DAP-type PGN have either not been determined or remain controversial. In addition, the mode of cellular activation and bactericidal activity of these PGRPs is largely unknown. The limited data for PGRP-L indicates that this protein exhibits lytic activity (31,32). The function of PGRP-I␣ and PGRP-I␤ is unknown, and most research has thus far focused on PGRP-S. Mouse PGRP-S found in neutrophil tertiary granules participates in the intracellular neutralization of bacteria. Mice deficient in this PGRP are much more susceptible to intraperitoneal infections with low pathogenic Gram-positive bacteria. (33) However, bovine PGRP-S, located in neutrophil and eosinophil granules, has been shown to inhibit the growth of both Gram-positive and -negative bacteria (34).
To determine in the ligand requirements for various PGRPs, we have synthesized a range of partial structures of PGN (see Fig. 1) that contain lysine or DAP as the third amino acid. The interactions of these compounds with human PGRP-S and the C-terminal domain of human PGRP-I␣ containing two tandem domains have been studied in realtime using surface plasmon resonance (SPR).

Materials
Procedures for the expressing recombinant PGRP-I␣C (residues 177-341) by in vitro folding from Escherichia coli inclusion bodies have been described previously (35).
MTrP_Lys ( 13  The DAP containing muramyl tripeptides 7-9 were synthesized similarly to procedures described above whereby the Lys(Mtt)-OH was replaced by a suitable protected DAP derivative (38) (38.7 mg, 42 mol) to afford the desired protected DAP derivative. Once cleaved from the resin, the DAP-PGNs were treated with 20% trifluoroacetic acid to deprotect the tert Butoxycarbonyl and tert butyl protecting groups on the side chain of the DAP. The deprotected derivative was precipitated from cold diethyl ether to afford an off-white compound. To a solution of this compound in EtOH:H 2 O:1 (N) HCl (4:2:0.01, 0.6 ml), 10% Pd/C (5 mg) was added and stirred at room temperature for 16 h. The solution was filtered and purified by Sephadex G10 size exclusion column chromatography to afford the target compound as a mixture of ␣/␤ anomers (9.3 mg, 30% overall). MTP_DAP (7)

SPR Analysis of PGRP-Ligand Interactions
The biospecific interaction analysis was performed using BIAcore 3000 biosensor system (Biacore Inc., Uppsala, Sweden). The CM-5 research grade sensor chip, HBS-EP buffer, and immobilization reagents (1-ethyl-3-(3-N,N-dimethylaminopropyl)carbodiimide (EDC), N-hydroxy succinimide (NHS), and ethanolamine) were obtained from BIAcore Inc. phosphate-buffered saline buffer was purchased from Sigma, and MDP was purchased from Calbiochem. All solutions were filtered using a 0.22-m PES membrane syringe filter and degassed prior to use. PGRP-I␣C was covalently immobilized by a standard amine coupling procedure using the amine coupling kit supplied by the manufacturer. A fixed flow rate of 10 l/min was used throughout the immobilization procedure with HBS-EP (pH 7.4, 0.01 M HEPES, 150 mM NaCl, 3 mM EDTA, 0.005% P20) as the running buffer. The surface was activated using 70 l of freshly mixed 1:1 100 mM NHS and 391 mM EDC for 7 min. Upon activation, a 60 g/ml solution of PGRP-I␣ in 10 mM NaOAc (pH 4.5) was injected for 8 min. The remaining active esters on the surface were quenched using 70 l of 1.0 M ethanolamine (pH 8.5) for 7 min. A ligand density of ϳ10,000 RU was achieved. Blocking of a control flow cell was accomplished by activation followed by an immediate quenching of the flow cell as described above. PGRP-S immobilization was accomplished using the same protocol as above with a change to 5 mM maleate buffer (pH 6.0) as the immobilization buffer. A ligand density of ϳ5,000 RU was achieved after 8 min of protein injection. Although a higher initial immobilization density was targeted, longer PGRP-S injections resulted in no change of protein density. Initial binding studies of the DAP containing compounds with PGRP-S suggested high affinity prompting the formation of a low density immobilization surface for kinetic analysis. A similar protocol was used with an injection of For PGRP-S, HBS-EP buffer was selected as both the running and dissociation buffers. Due to a greater variation in binding affinity, opti-mum concentration ranges for each analyte were established. For the lysine-containing compounds, MTP was examined from 500 to 1000 M, MTrP from 200 to 500 M, and MPP from 50 to 300 M. Initial binding studies were conducted on a flow channel with 5000 RU density. The DAP-containing muramyl peptides showed high affinity biding with on and off rates in the measurable range for kinetic analysis. These studies were conducted on a flow channel containing 2700 RU of PGRP-S to minimize rebinding and mass transport effect. A concentration range of 10 -1000 nM was selected for MPP-DAP and MTrP-DAP and for MTP-DAP, a concentration range of 100 -1000 nM was selected. The kinetic studies were performed using wizard kinetic software with a flow rate of 10 l/min and association and dissociation times of 5 and 10 min, respectively. The surface was regenerated by a 60-s injection of 10 mM NaOH, pH 11.4, at a flow rate of 30 l/min.

Data Analysis
The responses near equilibrium (R eq ) for the comparative affinity and rate constants k 1 and k Ϫ1 for kinetic analysis were obtained by fitting the primary sensogram data using the BIAevaluation 3.1 software. The dissociation rate constant is derived using,

Selective Ligand Recognition by Human PGRPs
where R t 0 is the amplitude of the initial response, and k Ϫ1 is the dissociation rate constant. The association rate constant k 1 can be derived from the measured k Ϫ1 values, using, where R t is response at time t, R max is the maximum response, C is concentration of the analyte in the solution, and k 1 and k Ϫ1 are association and dissociation rate constants, respectively. The ratio of k 1 and k Ϫ1 yields the value of association constant K A (k 1 /k Ϫ1 ).

RESULTS
The interactions of human PGRP-I␣C and PGRP-S with a range of synthetic lysine-and DAP-containing PGN fragments were probed using surface plasmon resonance (SPR). SPR is a rapid and sensitive method for the evaluation of affinities of bimolecular interactions (39). A benefit of this method is that it relies exclusively on mass changes, thus allowing interactions to be studied in real-time without the need for external labels such as fluorophores, which in some cases can alter the nature of the interaction. In this study, recombinant human PGRP-I␣ and PGRP-S were immobilized on research-grade CM5 sensor chips, and the synthetic compounds 1-9 were employed as analytes. Collecting SPR data for low molecular weight analytes, such as compounds 1-9, is challenging, because the refractive index monitored during a binding event is relatively small, thus resulting in responses with much lower magnitudes than those observed in typical protein-protein interactions. Despite these challenges, the high sensitivity and reproducibility of modern instruments combined with proper experimental design permits the direct monitoring of the binding of low molecular weight analytes to immobilized proteins. (40) An alternative approach whereby the synthetic peptidoglycan part structures would be immobilized and the PGRPs employed as analytes is expected to lead to artifacts, because the immobilization may destruct vital functional groups of the small synthetic compounds.
The compounds 1-9 ( Fig. 1) were synthesized by polymeric support synthesis using a Sieber amide resin, Fmoc-protected amino acids, and a properly protected muramic acid derivative. The synthetic compounds were designed in such a manner that the significance of each amino acid and the muramic acid moiety could be addressed for binding. Furthermore, by employing two series of compounds that contain    NOVEMBER 4, 2005 • VOLUME 280 • NUMBER 44 either lysine or DAP as the third amino acid, the importance of these residues could be studied. A relatively high immobilization of ϳ10,000 RU of PGRP-I␣C was accomplished on the N-hydroxy succinimide-activated groups of a CM-5 research grade sensor chip surface to achieve good signal-tonoise ratios. For all SPR studies, bulk refraction caused by the difference in refractive index of the running buffer and sample injection was negated by using a control cell that was functionalized by ethanolamine.

Selective Ligand Recognition by Human PGRPs
A representative sensogram for the interaction of the lysine-containing muramyl pentapeptide 6 with the immobilized PGRPI-␣C is shown in Fig. 2. Fitting using steady-state conditions gave a dissociation constant K D ϭ 62 M (TABLE ONE). The compounds 4 and 5 showed binding at concentrations Ͼ50 M, whereas peptide 1 (Lys-PP) and MDP (3) exhibited no binding at concentrations up to 1 mM. The latter results indicate that the muramyl moiety is required for complexation and that MTP 4 is the smallest fragment of PGN recognized by PGRP-I␣. Equilibrium-based kinetic analysis was not possible for MTP 4 and MTrP 5 due to an apparent biphasic nature of the real-time sensogram slope shifts at higher concentrations. Given that SPR detection is based on changes in mass at the surface of a sensor chip, normalization of respective RUs by the molecular weight of each analyte allows a qualitative comparison of analyte sets. A prerequisite of such a comparison is that all variables are kept constant, most importantly the protein surface employed. Normalization of the binding data of 4-6 ( Fig. 3) demon-strated that MTrP 5 and MPP 6 are significantly more potent ligands than MTP 3. Furthermore, the data suggest that MPP 6 has a slightly higher affinity than MTrP 5.
Normalized SPR data for the DAP-containing compounds 7-9 showed a trend similar to that of the lysine-containing derivatives 4-6, whereby PGRP-⍜␣ had the weakest affinity for the MTP-DAP, whereas MTrP-DAP and MPP-DAP had similar and significantly higher affinities (Fig. 4). A comparison of the normalized data for the DAP-containing compounds 7-9 and lysine-containing derivatives 4-6 indicated that the PGRP-I␣ has a low selectivity for the lysine-or DAP-containing compounds.
Next, the interactions of compounds 1-9 with a CM-5 research grade sensor chip, containing 5000 RU of PGRP-S, were studied. The normalized data for the binding of immobilized PGRP-S with lysinecontaining muramyl peptides 4-6 showed that MPP 6 has a slightly higher affinity than MtrP 5, which are both significantly better ligands than MTP 4 (Fig. 5). Steady-state equilibrium analysis of MPP 6 gave a K D of 189 M (TABLE ONE). The normalized data for the DAP-containing compounds 7-9 showed a similar trend to that observed of the lysine-containing derivatives 4-6 whereby MTrP-DAP 8 and MPP-DAP 9 were significantly better ligands than MTP-DAP 7. However, unlike the binding with PGRP-I␣C, which displayed similar binding affinities for the lysine-and DAP-type part structures, PGRP-S demonstrated significantly higher affinities for the DAP-con- taining fragments. Normalized data for MPP 6 and MPP-DAP 9 with both proteins at concentrations from 0.50 to 200 M illustrate the highly selective nature of PGRP-S for PGN part structures, whereas PGRP-I␣ does not discriminate between the lysine-and DAP-containing fragments (Fig. 6). Due to the high affinity for this set of analytes, a lower density surface containing 2700 RU of PGRP-S was prepared for kinetic analysis. The kinetic wizard method and simultaneous kinetic fitting gave the following dissociation constants: MPP-DAP K D ϭ 104 nM, MTrP-DAP ϭ 92 nM, and MTP-DAP ϭ 326 nM ( Fig. 7 and TABLE ONE).

DISCUSSION
Although the structure of the carbohydrate backbone of peptidoglycan is preserved among all bacteria, considerable structural variability exists in the peptide moiety (22). The archetypal stem peptide of Grampositive bacteria is L-alanine-␥-D-glutamate-L-lysine-D-alanine. In the case of Gram-negative bacteria and Gram-positive bacilli, m-diaminopimelic acid (DAP-type) is normally found as the third amino acid. Drosophila PGRP-SA has been shown to interact with lysine-type PGN activating the Toll receptor pathway, (23) whereas PGRP-LC and PGRP-LE recognize DAP-type PGN activating the Imd/Relish pathway (24 -28). Previous studies have shown that each of the four mammalian PGRPs is able to bind peptidoglycan. However, possible selectivities for lysine-or DAP-type PGN have either not been determined or remain controversial.
In this study, the complexation of human PGRP-I␣C and PGRP-S with a range of synthetic PGN fragments, containing either lysine or DAP as the third amino acid, have been studied by SPR. A dissociation constant of K D ϭ 62 M for MPP-Lys 6 was determined by a fitting of steady-state conditions. Furthermore, the normalized data of the lysinecontaining compounds 4-6 indicate that MTrP 5 and MPP 6 have similar affinities, whereas a much lower affinity for MTP 4 was measured. In addition, peptides 1 (Lys-PP) and 2 (DAP-PP) and MDP (3) exhibited no binding at 1 mM concentration.
Recently, the crystal structure of PGRP-I␣C ligated with MTP 4 was reported at 2.3 Å (Fig. 8) (41). In this complex, the tripeptide stem of MTP was held in an extended conformation at the deep end of the binding groove, whereas the MurNAc moiety lies in a pocket in the middle of the groove, with the pyranose ring oriented perpendicular to the base of the pocket. The structure indicates that the protein can accommodate a fourth D-Ala residue making contacts with Gln-261, Tyr-266, and Asn-269. The SPR data reported here demonstrate that these proposed interactions contribute significantly to binding. By contrast, D-Ala at position 5 is expected to extend beyond the binding groove. Therefore this amino acid should contribute little to the binding, an observation consistent with the SPR data. The observation that the pentapeptide 1 does not bind with PGRPI␣C indicates that the interactions with the muramic acid moiety are critical for binding. In this respect, the lactyl moiety forms a hydrogen bond with Tyr-242, whereas the acetamido of the saccharide moiety interacts with His-2231 and Arg-235. SPR experiments with DAP-containing fragments 7-9 revealed that they complex with similar affinities as compared with the Lys-containing compounds 4-6.
The complexation of PGRP-S with DAP-containing fragments 7-9 exhibited high affinities with dissociation constants in the nanomolar range. The affinity trend was similar to that observed for the binding of PGRP-I␣C with DAP-containing compounds 7-9. Thus, MTP-DAP (7) was the smallest fragment to be recognized, and significantly higher affinities were observed for MTrP-DAP (8) and MPP-DAP (9). Interestingly, much lower affinities were measured for the interactions of PGRP-S with the lysine-type compounds 4-6. Thus, although only a minimal selectivity was observed for binding of PGRP-I␣C with lysineand DAP-containing compounds, PGRP-S displayed a significant preference for DAP-containing PGN fragments.
The PGRP-I␣C-MTP complex shows that Asn-236 and Phe-237 form a number of van der Waals contacts with the side chain of L-lysine. Sequence alteration at these two positions may account for the discriminatory ability of different PGRPs toward Lys-type and DAP-type PGNs. For example, the corresponding sequence in Drosophila PGRP-LCx and PGRP-LE, which recognizes DAP-type PGNs, (1,42) is Gly-Trp. PGRP-S also has Gly-89 and Trp-90 at these positions.
The structural difference between lysine and DAP is an extra carboxylic acid at the ⑀C of the side chain. This chemical difference does not allow the different PGRPs to exhibit absolute discriminatory ability toward one type of PGN. For example, a certain degree of cross-reactivity is also exhibited by Drosophila PGRP-SA and mouse PGRP-L, which hydrolyzes PGNs from both Gram-positive and negative sources (31). Similarly, in the case of PGRP-I␣C, despite having Asn-236 and Phe-237, the protein can recognize both Lys-and DAP-type PGNs with similar affinities. Selectivity for lysine-or DAP-containing peptidoglycan fragments has been observed for other pattern recognition receptors. For example, transfection studies have shown that NOD2 recognized MDP and muramyltripeptide containing lysine as the third amino acid. (43) On the other hand, NOD1 senses DAP-containing peptidoglycan (44,45). The structure activity relationships determined for the NOD proteins shows a different profile from what has been determined for PGRP-I␣ and PGRP-S. In this respect, the NOD proteins do not respond to muramyl tetra-and muramyl pentapeptide structures. It appears that the different pattern recognition receptors for peptidoglycan have evolved in such a way that they can recognize different part structures.