PAP (pancreatitis-associated protein) is a 16 kDa lectin-like protein, which becomes robustly up-regulated in the pancreatic juice during acute pancreatitis. Trypsin cleaves the N-terminus of PAP, which in turn forms insoluble fibrils. PAP and its paralogue, the pancreatic stone protein, induce bacterial aggregation and, more recently, PAP was shown to bind to the peptidoglycan of Gram-positive bacteria and exert a direct bactericidal effect. However, the role of N-terminal processing in the antibacterial function of PAP has remained unclear. In the present study, we demonstrate that N-terminal cleavage of PAP by trypsin at the Arg37–Ile38 peptide bond or by elastase at the Ser35–Ala36 peptide bond is a prerequisite for binding to the peptidoglycan of the Gram-positive bacterium Bacillus subtilis. The tryptic site in PAP was also efficiently cleaved by nprE (extracellular neutral metalloprotease) secreted from B. subtilis. Trypsin-mediated processing of PAP resulted in the formation of the characteristic insoluble PAP species, whereas elastase-processed PAP remained soluble. N-terminally processed PAP induced rapid aggregation of B. subtilis without significant bacterial killing. The bacteria-aggregating activities of trypsin-processed and elastase-processed PAP were comparable. In contrast with previous reports, the Gram-negative Escherichia coli bacterium was not aggregated. We conclude that N-terminal processing is necessary for the peptidoglycan binding and bacteria-aggregating activity of PAP and that trypsin-processed and elastase-processed forms are functionally equivalent. The observations also extend the complement of proteases capable of PAP processing, which now includes trypsins, pancreatic elastases and bacterial zinc metalloproteases of the thermolysin type.
PAP (pancreatitis-associated protein) was described by Keim et al. [1,2] as a 16 kDa secretory protein in rat pancreatic juice that appeared upon induction of pancreatic inflammation and represented up to 5% of total protein. The human orthologue was isolated in 1992 from the pancreatic juice of diabetic patients who underwent combined kidney and pancreas transplantation . In these patients, the donor pancreas developed acute pancreatitis and the levels of PAP in the juice reached up to 7.5% of the total secretory protein. Because PAP was hardly detectable in normal pancreatic juice, but became up-regulated in inflammation, it was considered a unique acute-phase protein, which was targeted to the exocrine secretions. Subsequent cDNA cloning revealed that PAP exhibited homology with the carbohydrate-binding region of Ca2+-dependent (C-type) lectins and thus PAP belongs to a larger secretory protein family which contain only a single C-type lectin domain [4,5]. Members of this protein family, which include the extensively studied PSP (pancreatic stone protein), have been described in excellent reviews [6–9]. Graf et al.  demonstrated that trypsin cleaved all three isoforms of rat PAP after Arg37 (Arg11 in the mature PAP protein) and the processed PAP became resistant to proteases and formed insoluble, fibrillar structures [10,11]. The authors proposed that insoluble PAP might be the biologically active form.
PAP expression was also detected in the intestines and other extrapancreatic tissues (reviewed in [6–9]), but the physiological function of PAP in the pancreas or elsewhere has remained contentious. Evidence from a large number of studies seemed to indicate that PAP had a variety of activities, which included anti-inflammatory, anti-apoptotic, proliferative and antibacterial effects (reviewed in [6–9]). With respect to the antimicrobial activity, PAP and PSP were shown to induce bacterial aggregation without inhibiting growth, and trypsin appeared to stimulate the effect of PSP [4,12]. More recently, Cash et al.  demonstrated that PAP can bind to the peptidoglycan layer of Gram-positive bacteria and exert a direct bactericidal effect. PAP also bound to chitin and the mannose polymer mannan, indicating that PAP recognizes carbohydrate patterns [13,14]. Cash et al.  proposed that PAP is part of the innate immune system and its intestinal expression plays an important role in the maintenance of the intestinal bacterial flora. By analogy, in the pancreas, PAP is probably up-regulated during acute pancreatitis to prevent bacterial infection. It remained unclear, however, whether or not the N-terminal processing of PAP by trypsin and the consequent insoluble fibril formation play a role in the antibacterial function. In the present study, we set out to answer this question.
MATERIALS AND METHODS
Bacillus subtilis subsp. subtilis (Ehrenberg) Cohn was obtained from the A.T.C.C. (ATCC no. 6051) and designated as wild-type B. subtilis in the present study. The protease-deficient B. subtilis BG2036 (Δapr-684, ΔnprE522) strain  was a gift from Evette S. Radisky (Mayo Clinic Cancer Center, Jacksonville, FL, U.S.A.). The Escherichia coli K-12 strain α-select [F−deoR endA1 recA1 relA1 gyrA96 hsdR17(rk−, mk+) supE44 thi-1 phoA Δ(lacZYA argF)U169 Φ80lacZΔM15 λ−] was purchased from Bioline and the E. coli B strain BL21(DE3) [F−ompT hsdSB (rB− mB−) gal dcm (DE3)] was from Novagen. All experiments using E. coli in the present study included both strains.
Numbering of amino acids of human PAP follows the genetic convention and starts with the initiator methionine (Met1) of the primary translation product. Because the secretory signal peptide comprising the first 26 amino acids is removed in the secretory pathway, the first amino acid of native, mature human PAP is Glu27 (Figure 1).
N-terminal sequences of proteolytically processed PAP forms
Plasmid construction and mutagenesis
The cDNA encoding the mature, secreted human PAP (amino acids Glu27–Asp175) was PCR amplified from IMAGE clone no. 5749980 (GenBank® Nucleotide Sequence Database accession no. BC036776) using the PAP NcoI sense primer [5′-AAATTTTCCATGGAACCACAAAGAGAACTGCCCTCT-3′ (where the NcoI site is underlined)] and the PAP SacI antisense primer [5′-AAATTTGAGCTCTAGTCAGTGAACTTGCAGACATAG-3′ (where the SacI site is underlined)]. The PCR product was digested with NcoI and SacI and ligated into the pTrapT7_Hu1 plasmid [16,17] in place of the human cationic trypsinogen gene. In the pTrapT7_PAP expression construct, the native secretory signal sequence of PAP is deleted and the N-terminal glutamic acid residue of mature PAP (Glu27) is replaced by a methionine residue (see Figure 1). In addition, codons for Pro29-Gln30-Arg31 have been optimized to increase expression levels in E. coli, as described by Cash et al. .
To express PAP with the first nine amino acids deleted [Δ9-PAP (elastase-processed PAP with the N-terminal nine amino acids removed); see Figure 1], we fused the N-terminus of Δ9-PAP to the C-terminus of a self-splicing mini-intein from the cyanobacterium Synechocystis sp. Previously, we have used this method successfully for the expression of human cationic trypsinogen with an authentic, native N-terminus . The intein fusion construct was generated by overlap extension PCR. First, the ∼460 nt DnaB mini-intein gene was amplified from the plasmid pTwin2 (New England Biolabs) using the Intein NcoI sense primer [5′-CGGGAGTCCATGGCTATCTCTGGCGATAGTCTGATCAGC-3′ (where the NcoI site is underlined)] and the Intein Δ9-PAP antisense primer (5′-GATCC-GTGCGTTGTGTACAATGATGTCATTCGCGAC-3′). Next, the PAP cDNA corresponding to amino acids Ala36-Asp175 was amplified using the Intein Δ9-PAP sense primer (5′-GTAC-ACAACGCACGGATCCGCTGTCCCAAAGGCTCC-3′) and the PAP SacI antisense primer described above. The underlined sequences in the Intein Δ9-PAP sense and antisense primers overlap and thus allow the PCR products obtained in the first round of amplifications to hybridize. PCR products from the first two reactions were gel-purified, mixed, extended and amplified using the Intein NcoI sense primer and the PAP SacI antisense primer. The resulting PCR product was digested with NcoI and SacI and ligated into the pTrapT7_PAP plasmid.
Expression and refolding of PAP
Human full-length PAP and Δ9-PAP were expressed in E. coli as cytoplasmic inclusion bodies and were subjected to in vitro refolding according to the protocol published by Cash et al. , with minor modifications. E. coli BL21 cells were transformed with PAP expression plasmids. Cells were grown in 600 ml of LB (Luria–Bertani) medium containing 100 μg/ml ampicillin at 37 °C until the D600 reached 0.6–0.9. PAP expression was then induced with 1 mM concentration of IPTG (isopropyl β-D-thiogalactoside) for 3 h. Cells were harvested by centrifugation at 12300 g for 10 min, and the pellet was resuspended in 15 ml of wash buffer (20 mM Tris/HCl, pH 8.0, 10 mM potassium EDTA and 1% Triton X-100, final concentrations). Cells were divided into 5 ml aliquots and sonicated for 3×1 min using a Heat Systems Ultrasonics cell disruptor, Model W-200R, with a microtip probe, at continuous mode on a power setting of 5. The bacterial cell lysate was centrifuged at 32800 g for 15 min, and the pellet containing the inclusion bodies was resuspended in 35 ml of wash buffer and centrifuged again. The washing procedure was repeated twice more. The final inclusion body pellet was resuspended in 10 ml of solubilization buffer [7 M guanidinium chloride, 30 mM DTT (dithiothreitol), 100 mM Tris/HCl (pH 8.0) and 2 mM potassium EDTA, final concentrations] and incubated at room temperature (22 °C) with continuous gentle mixing on an Adams Nutator mixer for 2 h followed by centrifugation at 32800 g for 15 min. The supernatant (∼10 ml) was then added to 125 ml of refolding buffer (50 mM Tris/HCl, pH 8.0, 10 mM KCl, 240 mM NaCl, 2 mM MgCl2, 2 mM CaCl2, 0.5 M guanidinium chloride, 0.4 M sucrose, 1 mM GSH, 0.1 mM GSSG and 0.5 M arginine/HCl, final concentrations) pre-equilibrated with argon. The refolding mixture was kept under argon at 4 °C for 24 h and then dialysed overnight against two changes of 10 litres of 20 mM sodium acetate (pH 5.0) buffer.
Purification of PAP
After dialysis, precipitated proteins were removed by ultracentrifugation at 40000 rev./min for 45 min at 4 °C using a Beckman Type 45 Ti rotor, and re-folded PAP (which typically increased in volume during dialysis to ∼200 ml) was loaded on to a MonoS HR 5/5 cation-exchange column (Amersham Biosciences) through the pump of an FPLC system (Amersham Biosciences) at a flow rate of 2 ml/min. The column was then developed with an NaCl gradient of 0–1 M [in 20 mM sodium acetate (pH 5.0)] at a flow rate of 1 ml/min. Fractions (1 ml) were collected and 15 μl of each peak fraction was analysed by SDS/15% PAGE and Coomassie Blue staining. Pure fractions were pooled, divided into aliquots and stored at −20 °C until use. Typical yields were 5–10 mg of purified PAP per 600 ml of initial culture.
Refolding and purification of the intein-Δ9-PAP fusion followed the same protocol. The self-splicing intein fusion tag is removed during expression, and the liberated Δ9-PAP can be re-folded and purified as described above. Yields were typically lower than those seen with full-length PAP, approx. 2–3 mg of protein per 600 ml of initial culture.
Expression and purification of human pancreatic protease zymogens
Human trypsinogens and pro-elastase 2A were expressed in E. coli as described previously [16,17,19–21]. Pro-elastases 3A and 3B and chymotrypsinogens B1, B2 and C were expressed in transiently transfected HEK-293T cells [human embryonic kidney 293 cells expressing the large T-antigen of SV40 (simian virus 40)] as previously reported . Trypsinogens, pro-elastases and chymotrypsinogen C were purified by ecotin-affinity chromatography , whereas chymotrypsinogens B1 and B2 were purified by MonoS cation-exchange chromatography followed by gel filtration on a Superose 6 HR 10/30 column (Amersham Biosciences) .
Concentrations of the purified protein solutions were calculated from their UV absorbance at 280 nm, using the following theoretical molar absorption coefficients (http://ca.expasy.org/tools/protparam.html). PAP and Δ9-PAP, 43805 M−1·cm−1; cationic trypsinogen, 37525 M−1·cm−1; anionic trypsinogen, 38890 M−1·cm−1; mesotrypsinogen, 41535 M−1·cm−1; pro-elastase 2A, 73505 M−1·cm−1; pro-elastase 3A, 76025 M−1·cm−1; pro-elastase 3B, 74535 M−1·cm−1; chymotrypsinogens B1 and B2, 47605 M−1·cm−1; chymotrypsinogen C, 64565 M−1·cm−1.
Activation of digestive pro-enzymes
Chymotrypsinogens and pro-elastases (1–2 μM concentration) were activated with immobilized bovine trypsin (Pierce) for 60 min at 37 °C. An aliquot (20 μl) of trypsin beads was pelleted by centrifugation, washed twice with 0.1 M Tris/HCl (pH 8.0), 1 mM CaCl2 and resuspended in 200 μl of the same buffer containing the pro-enzymes. After completion of the activation reaction, the trypsin beads were removed by centrifugation. Trypsinogens (2 μM) were activated to trypsin with human enteropeptidase (R&D Systems; 28 ng/ml) for 30 min at 37 °C, in 0.1 M Tris/HCl (pH 8.0) and 1 mM CaCl2.
Preparation of peptidoglycan from B. subtilis
Cell wall (crude peptidoglycan) was prepared from B. subtilis by the method of Leclerc et al.  with minor modifications. Cells were grown overnight in 3 litres of LB medium at 30 °C and harvested by centrifugation at 7900 g for 10 min at 4 °C. Cells were resuspended in 100 ml of 0.1 M Tris/HCl (pH 8.0), kept at 4 °C for 15 min and centrifuged again. Cells were resuspended in 100 ml of 4% (w/v) SDS solution. The suspension was incubated for 1 h at room temperature with continuous mixing on a rocking platform, followed by sonication on ice for 10 min. The extract was then heated to 100 °C for 20 min and centrifuged at 7900 g for 20 min at room temperature. To remove SDS and membranes, the pellet was washed in 100 ml of 0.2% (v/v) Triton X-100 with continuous mixing for 30 min at room temperature, followed by four washes with 80 ml of distilled water. Cell wall was centrifuged at 7900 g for 20 min at 4 °C and freeze-dried. The final yield was approx. 300 mg of cell wall powder per 3 litres of initial culture.
To remove peptidoglycan-associated proteins (including proteases), 2 mg of cell wall powder was resuspended in 1 ml of 100 mM Tris/HCl (pH 8.0), 1 mM CaCl2 and digested with 50 mg/ml proteinase K at 37 °C overnight. Next day, 100 mM DTT and 1% SDS (final concentrations) were added and the suspension was heated to 95 °C for 15 min. DTT and SDS were removed by washing the peptidoglycan preparation ten times with 2 ml of distilled water. Finally, protease-free peptidoglycan was suspended in water at a concentration of 1 mg/ml and stored at 4 °C.
Labelling of bacteria with FITC
B. subtilis and E. coli were grown to D600 of 0.6. A 1 ml portion of each culture was centrifuged and the bacterial pellets were washed with 50 mM carbonate–bicarbonate buffer (pH 9.0) containing 10 mM NaCl and resuspended in 1 ml of the same buffer. A 50 μl portion of FITC [Sigma 46950; dissolved in DMSO at 10 mg/ml (25.7 mM) concentration] was added (1.2 mM final concentration) and the reaction mixtures were incubated in the dark at room temperature for 1 h. Labelled cells were centrifuged at 16000 g for 1 min and the bacterial pellets were washed five times with 20 mM Tris/HCl (pH 8.0) and 150 mM NaCl to remove residual FITC. Finally, FITC-labelled cells were resuspended in 1 ml of 100 mM Tris/HCl (pH 8.0) and 1 mM CaCl2.
N-terminal processing of PAP by human trypsins
We expressed human PAP in E. coli according to a published protocol  and purified the lectin to homogeneity by cation-exchange chromatography. Incubation of PAP with 10 nM human cationic or anionic trypsin resulted in the rapid conversion of the PAP band into a lower-mobility species on SDS/PAGE (Figure 2A). N-terminal sequencing revealed that trypsin cleaved the Arg37-Ile38 peptide bond and generated an 11-amino-acid shorter PAP form [Δ11-PAP (trypsin-processed PAP with the N-terminal 11 amino acids removed); see Figure 1], in accordance with published reports [10,11]. In contrast with cationic and anionic trypsin, mesotrypsin did not process PAP at all. This observation is in agreement with previous results showing that mesotrypsin digests most protein substrates poorly and cannot activate pancreatic zymogens .
N-terminal processing of PAP by trypsin and pancreatic elastases
N-terminal processing of PAP by human pancreatic elastases
Human PAP was first discovered in the pancreatic juice of patients who had undergone pancreas transplantation and developed transient acute pancreatitis in the graft pancreas . We have obtained a sample of the pancreatic juice used in these studies (PS19, also see  for additional information) and fractionated it by MonoQ anion-exchange chromatography. The fractions were analysed by SDS/PAGE and Coomassie Blue staining and PAP was tentatively identified as a 17 kDa protein band present in fractions 16–18 (Supplementary Figure S1 at http://www.BiochemJ.org/bj/420/bj4200335add.htm). This band was transferred from fraction 17 to a PVDF membrane and subjected to Edman degradation, which confirmed the identity of PAP. Surprisingly, however, the N-terminal sequence of PAP present in this fraction lacked nine amino acids from its N-terminus (Δ9-PAP; see Figure 1), evidently as a result of cleavage at the Ser35-Ala36 peptide bond. To identify the protease responsible for this atypical processing, we digested recombinant PAP with a panel of human pancreatic proteases, which included chymotrypsin B1, chymotrypsin B2, chymotrypsin C, elastase 2A, elastase 3A and elastase 3B. Elastases 2A and 3B processed the N-terminus of PAP in vitro, and protein sequencing confirmed that the site of processing was the Ser35-Ala36 peptide bond, which conforms to the canonical elastase-sensitive peptide bonds (Figure 2B). Chymotrypsin B1, chymotrypsin B2 and elastase 3A did not cleave PAP, whereas chymotrypsin C slowly degraded the lectin (results not shown). The results indicate that, in the absence of trypsin, PAP can be processed N-terminally by an alternative, elastase-dependent mechanism. Although not shown, a commercial preparation of human sputum leucocyte elastase (SE563; Elastin Products Company) degraded PAP without appreciable N-terminal processing.
Addition of trypsin to elastase-processed Δ9-PAP converted this species into Δ11-PAP, as confirmed by N-terminal sequencing. The trypsin-sensitivity of Δ9-PAP posed a technical challenge since the potential existed for unnoticed trypsin contamination in the pancreatic elastase preparations used to generate Δ9-PAP. Unwanted tryptic processing of Δ9-PAP would not be easily detected, as Δ9-PAP and Δ11-PAP do not resolve into separate bands on SDS/PAGE. To circumvent this problem, we expressed Δ9-PAP in E. coli as a fusion protein with a self-splicing mini-intein as described in the Materials and methods section. Upon self-removal of the fusion tag, authentic Δ9-PAP could be refolded and purified. In the studies presented below, experiments were performed using both Δ9-PAP derived from intein fusions and Δ9-PAP generated through bona fide elastase processing, with identical results.
Solubility and protease sensitivity of trypsin-processed Δ11-PAP and elastase-processed Δ9-PAP
Upon cleavage of the Arg37-Ile38 peptide bond by trypsin, the newly generated Δ11-PAP presumably undergoes a conformational change, which results in two well-documented novel properties: (i) formation of insoluble aggregates that can be pelleted by centrifugation [10,11]; and (ii) resistance to proteolytic degradation by pancreatic proteases  or by proteinase K . To ascertain whether elastase-processed PAP acquires similar properties, first we compared the solubilities of Δ11-PAP and Δ9-PAP. As shown in Figure 3(A), intact unprocessed PAP was soluble, whereas, after processing by trypsin, aggregation of Δ11-PAP became detectable within 5 min and the greater part of Δ11-PAP was precipitated by 30 min at 22 °C. The tendency for aggregation of Δ11-PAP was comparable at 22, 30 and 37 °C, whereas aggregation proceeded significantly more slowly at 4 °C and only minimal precipitate was recovered after 30 min (Figure 3B). In contrast with trypsin-processed PAP, elastase-processed Δ9-PAP remained soluble over the same time course and only a small amount of precipitate appeared by 30 min (Figure 3C). As expected, addition of trypsin to elastase processed Δ9-PAP resulted in aggregation (results not shown).
Effect of N-terminal processing on the solubility of PAP
Next, we digested full-length PAP, Δ11-PAP and Δ9-PAP with proteinase K. Unprocessed PAP was degraded by the protease over the 60 min time course, whereas trypsin-processed Δ11-PAP was completely stable (Figure 4A). Similarly, Δ9-PAP showed increased resistance against proteinase K, although some degradation was noticeable by 60 min (Figure 4B). Taken together, these observations suggest that N-terminal processing by elastase generates a PAP isoform that is similar in conformation (as judged by increased protease resistance) to, yet not identical (as shown by lack of aggregation) with, trypsin-processed PAP.
Digestion of N-terminally processed PAP with proteinase K
N-terminal proteolytic processing of PAP is essential for peptidoglycan binding
Previously, Cash et al.  demonstrated that PAP binds to the peptidoglycan of Gram-positive bacteria and exerts a direct bactericidal effect. However, the significance of N-terminal processing of PAP in its peptidoglycan-binding activity was not addressed. We compared the ability of full-length PAP and its proteolytically processed Δ9-PAP and Δ11-PAP forms to bind to purified peptidoglycan of the Gram-positive bacterium B. subtilis. In the binding experiments, peptidoglycan was incubated with PAP at 22 °C for 30 min, the mixture was centrifuged at 4 °C and the distribution of PAP between the pellet (bound PAP) and the supernatant (unbound PAP) was analysed by SDS/PAGE and Coomassie Blue staining. Figure 5(A) demonstrates that full-length PAP did not bind to peptidoglycan, whereas both the trypsin processed Δ11-PAP and elastase processed Δ9-PAP were fully bound.
Binding of N-terminally processed PAP to peptidoglycan from B. subtilis
Since aggregation and consequent sedimentation of trypsin-processed Δ11-PAP at 22 °C might confound the results of the pull-down assay, we monitored the time course of peptidoglycan binding and compared it with the time course of Δ11-PAP precipitation. As shown in Figure 5(B), binding of Δ11-PAP to peptidoglycan was complete in 1 min at 22 °C, whereas significant precipitation of Δ11-PAP occurred only after 15 min (cf. Figure 3A). Furthermore, experiments performed at 4 °C demonstrated that binding of Δ11-PAP to peptidoglycan was complete in 30 min (Figure 5B), whereas precipitation of Δ11-PAP at this temperature was minimal after a 30 min incubation (cf. Figure 3B).
N-terminal processing of PAP by the B. subtilis nprE (extracellular neutral metalloprotease) protease
The results described above indicate that proteolytic processing of PAP by trypsin or elastase must precede peptidoglycan binding. These observations appear to contradict the results of Cash et al. , who showed peptidoglycan binding using intact, unprocessed PAP protein. The authors, however, observed that the commercial peptidoglycan preparation they used proteolytically cleaved PAP, which resulted in a mobility shift on SDS/PAGE gels. Therefore it seems reasonable to assume that secreted bacterial proteases might process PAP in a trypsin-like manner and thus promote peptidoglycan binding. To confirm this notion, we incubated PAP with the culture supernatant from wild-type B. subtilis and from strain BG2036, in which the genes for two major secreted proteases are deleted: the so-called extracellular neutral metalloprotease (nprE) and the serine alkaline protease (aprE, subtilisin E) . As shown in Figure 6(A), incubation of intact PAP with bacterial supernatant from the wild-type strain resulted in the characteristic mobility shift, whereas no processing was observed by culture supernatants of the protease-deficient BG2036 strain. N-terminal sequencing identified the site of processing as the trypsin-sensitive Arg37-Ile38 peptide bond. Addition of 20 mM EDTA (final concentration) to the bacterial supernatant abolished PAP processing (Figure 6B), suggesting that the bacterial protease responsible for PAP processing is the thermolysin-like neutral zinc metalloprotease (nprE). Consistent with this conclusion, a commercial preparation of thermolysin from Bacillus thermoproteolyticus Rokko (R&D Systems 3097-ZN) efficiently processed PAP, whereas subtilisin Carlsberg from Bacillus licheniformis (Calbiochem 572909) was ineffective (results not shown).
N-terminal processing of PAP by the nprE metalloprotease secreted from B. subtilis
N-terminally processed PAP induces aggregation of B. subtilis
To characterize the antibacterial effects of the various PAP forms, we incubated wild-type B. subtilis and the BG2036 strain with trypsin-processed Δ11-PAP, elastase-processed Δ9-PAP or intact, full-length PAP. We labelled B. subtilis with fluorescent FITC and monitored the PAP-induced aggregation of bacteria under the microscope. Figure 7 demonstrates that the protease-deficient BG2036 strain was aggregated by both trypsin-processed Δ11-PAP and elastase-processed Δ9-PAP, but not by unprocessed, full-length PAP. These results indicate that N-terminal processing is necessary for the bacterial aggregating activity of PAP and that trypsin-processed and elastase-processed forms are functionally equivalent. The wild-type B. subtilis strain was aggregated not only by Δ11-PAP and Δ9-PAP, but also by treatment with intact, full-length PAP, which was evidently processed by the nprE protease in the incubation medium.
Bacterial aggregation induced by N-terminally processed PAP
Neither PAP species promoted the aggregation of Gram-negative E. coli strains (results not shown). This observation stands in contrast with that in a previous report that described PAP-induced aggregation of E. coli .
PAP exerts no significant bactericidal activity against B. subtilis
Cash et al.  showed that PAP has a direct killing effect on Gram-positive bacteria, although B. subtilis was not studied . To test whether N-terminally processed PAP forms kill B. subtilis, first we utilized a dilution-plating assay. The bacteria were incubated with Δ11-PAP or Δ9-PAP at 22 °C for 60 min and 10-fold serial dilutions were plated on to LB agar plates. After overnight growth at 30 °C, colonies were counted. Surprisingly, the number of colony-forming units decreased upon PAP treatment only by ∼5–10%; however, this difference was within experimental error. Trypsin-processed Δ11-PAP and elastase-processed Δ9-PAP gave essentially identical results (Figure 8A).
Effect of N-terminally processed PAP on bacterial viability
To ascertain by an independent method whether PAP-treated bacteria are dead or alive, we stained B. subtilis with the Live/Dead® BacLight™ Bacterial Viability kit (Molecular Probes). This kit contains two DNA-binding fluorescent dyes, the green fluorescent dye SYTO 9, which binds to both dead and live bacteria and the red fluorescent DNA stain propidium iodide, which penetrates only dead bacteria. Importantly, binding of propidium iodide to the DNA of dead bacteria quenches the green fluorescence of SYTO 9. Figure 8(B) demonstrates fluorescence spectra of wild-type B. subtilis treated with trypsin-processed Δ11-PAP and then stained with both fluorescent dyes. Remarkably, the green peak of the spectrum exhibited only a 10–20% decrease upon PAP treatment, indicating that the large majority of the bacteria were alive. Identical results were obtained with elastase-processed Δ9-PAP (results not shown). Inspection of the PAP-treated bacterial suspension under a fluorescent microscope revealed green aggregates of live bacteria containing some red-stained dead bacteria (results not shown). Therefore we conclude that N-terminally processed forms of PAP do not kill B. subtilis to a significant extent.
The biological role of PAP has been in question ever since its discovery in 1984 . Recent studies by Cash et al.  provided evidence that one of the fundamental functions of human PAP and its mouse orthologue regIIIγ is an antibacterial activity. PAP was shown to bind to the peptidoglycan of Gram-positive bacteria and exert a direct bactericidal effect. E. coli and other Gram-negative bacteria, whose peptidoglycan layer is shielded by an outer membrane, were not affected by PAP. The authors proposed that PAP is a pattern-recognition molecule, responsible for an “evolutionarily primitive form of lectin-mediated innate immunity”. Previous studies from 1991 and 1993 also showed that rat PAP and the paralogue human PSP have antibacterial activity, as both molecules were shown to induce bacterial aggregation although without affecting growth [4,12]. Both PAP and PSP have been known to undergo tryptic cleavage and consequent insoluble fibril formation; however, the significance of the proteolytic processing in the antibacterial activity of PAP has not been studied. Iovanna et al.  noted that trypsin seemed to facilitate the bacteria-aggregating activity of human PSP, suggesting that PAP function might be similarly regulated by trypsin.
The present study convincingly demonstrates that binding of PAP to peptidoglycan and the consequent induction of bacterial aggregation are entirely dependent on the N-terminal proteolytic processing of PAP. Intact, full-length PAP does not bind to peptidoglycan and has no antibacterial activity whatsoever. We found that among the pancreatic proteases not only trypsins but also elastases can process PAP at the N-terminus and the trypsin-processed and elastase-processed PAP species exhibit comparable peptidoglycan binding and bacteria-aggregating activities. Furthermore, we demonstrated that bacterial proteases secreted extracellularly can cleave PAP at the trypsin-sensitive peptide bond. This phenomenon was also noted by Cash et al. . For B. subtilis, we tentatively identified the zinc metalloprotease nprE as the enzyme responsible for this activity. This latter observation might explain why intact, full-length PAP appeared to exhibit peptidoglycan binding and bacterial agglutinin activities in previous studies.
Arguably, pancreatic elastase-mediated PAP processing is one of the most interesting observations in the present study. However, the existence of a stable elastase-processed Δ9-PAP in the pancreatic juice might seem questionable, as this form should be rapidly processed further by trypsin. We isolated elastase-processed Δ9-PAP from the pancreatic juice of a patient with acute pancreatitis, which confirms that it is a naturally occurring variant. Premature activation of digestive protease zymogens inside the pancreas is a hallmark of acute pancreatitis, and we propose that, under certain conditions, elastase activation might precede trypsin activation, resulting in elastase-mediated PAP processing. In addition to pancreatic elastases, the elastase-sensitive Ser35–Ala36 peptide bond might be processed by certain bacterial proteases as well, although direct evidence for this is lacking. Nonetheless, it is intriguing to hypothesize that the N-terminus of PAP represents a protease bait region, which can be attacked by various proteases to generate biologically active, N-terminally processed PAP.
Tryptic cleavage of PAP and PSP results in the formation of insoluble fibrils, which were suggested to be responsible for the biological activity of these lectin-like proteins ([10,11]; PSP has been reviewed in [6–9]). The observation that elastase-processed PAP remains soluble but still exhibits peptidoglycan binding and bacterial aggregating activity argues that insoluble fibril formation by PAP is not required for these functions. Both trypsin and elastase processing alters the conformation of PAP in a similar manner, as judged by the increased resistance to proteinase K digestion. Despite this similarity, the unique solubility of elastase processed PAP indicates that this PAP form is in a somewhat different conformation than trypsin-processed PAP is. X-ray crystallographic and NMR structures have been reported for full-length, intact PAP ([26,27], PDB ID: 2GO0 and 1UV0), but not for N-terminally processed PAP, which could shed light on the structural changes that occur upon N-terminal processing. In this respect, the soluble elastase-processed PAP might be a more promising target for crystallographic analysis than the aggregation-prone trypsin-processed form.
In contrast with the results of Cash et al. , in our studies we did not find significant bactericidal activity of PAP and we concluded that the primary antibacterial activity of PAP was bacterial aggregation and not bacterial killing, at least with respect to B. subtilis. Cash et al.  demonstrated the killing effect of PAP on various Gram-positive strains, but B. subtilis was not tested. Thus strain-specific sensitivity to PAP might explain the differing results. Our conclusion that PAP is not bactericidal is also in agreement with previous reports indicating that PAP or PSP had no inhibitory effect on bacterial growth [4,12]. On the other hand, our observations contradict the same studies inasmuch as we could not demonstrate any effect of PAP on E. coli. Since the peptidoglycan layer is not exposed in E. coli, PAP should not bind to this bacterium and we cannot explain why aggregation was observed previously, although we note the long incubation times (3 h) used in these studies.
After the writing of this manuscript was completed, a paper reported that trypsin-mediated processing of PAP was required for its bactericidal activity against the Gram-positive bacterium Listeria monocytogenes . Using mutagenesis and NMR, the study also showed that PAP can undergo a conformational switch between biologically inactive and active states and the negative charges in the pro-peptide regulate this switch in an inhibitory manner. While our results agree with the major conclusion of this paper, there are notable discrepancies between the two studies on some important points. Thus Mukherjee et al.  claim that (i) N-terminal processing of PAP is not required for peptidoglycan binding; (ii) the antibacterial effect of PAP is independent of peptidoglycan binding; and (iii) the major antibacterial effect of PAP is killing. In contrast, our observations clearly establish a direct relationship between N-terminal processing of PAP, binding to peptidoglycan and induction of bacterial aggregation without significant killing. The inconsistent results between the two studies might be related to the use of different bacterial strains and peptidoglycan preparations, the possible interference of bacterial aggregation with the viability assays and the proteolytic instability of the PAP pro-peptide, which can result in unwanted processing of PAP preparations.
We gratefully acknowledge Niels Teich (University of Leipzig, Leipzig, Germany) for providing the pancreatic juice sample, Charles Bevins (University of California, Davis, CA, U.S.A.) for helpful suggestions and Evette Radisky (Mayo Clinic Cancer Center, Jacksonville, FL, U.S.A.) for providing the strain BG2036.
This work was supported by the National Institutes of Health [grant number DK058088 (to M. S.-T.)] and a scholarship from the Rosztoczy Foundation to P. M.