Mutational analysis of the pro‐peptide of a marine intracellular subtilisin protease supports its role in inhibition

Abstract Intracellular subtilisin proteases (ISPs) have important roles in protein processing during the stationary phase in bacteria. Their unregulated protein degrading activity may have adverse effects inside a cell, but little is known about their regulatory mechanism. Until now, ISPs have mostly been described from Bacillus species, with structural data from a single homolog. Here, we study a marine ISP originating from a phylogenetically distinct genus, Planococcus sp. The enzyme was successfully overexpressed in E. coli, and is active in presence of calcium, which is thought to have a role in minor, but essential, structural rearrangements needed for catalytic activity. The ISP operates at alkaline pH and at moderate temperatures, and has a corresponding melting temperature around 60 °C. The high‐resolution 3‐dimensional structure reported here, represents an ISP with an intact catalytic triad albeit in a configuration with an inhibitory pro‐peptide bound. The pro‐peptide is removed in other homologs, but the removal of the pro‐peptide from the Planococcus sp. AW02J18 ISP appears to be different, and possibly involves several steps. A first processing step is described here as the removal of 2 immediate N‐terminal residues. Furthermore, the pro‐peptide contains a conserved LIPY/F‐motif, which was found to be involved in inhibition of the catalytic activity.


| INTRODUCTION
ISPs have key roles in cell cycle regulation, specifically in protein recycling by processing proteins during transition to the stationary phase. 1,2 To prevent proteolysis that may be lethal to the cell, the activity of an intracellular protease must be tightly controlled.
Although a potential ISP inhibitor protein has been identified, 3,4 the primary mechanism of regulation is likely intrinsic. 5,6 In the precursor protein, an N-terminal pro-peptide of typically 16-20 residues binds across the active site and inhibits activity. As shown for a few homologs, 6,7 the pro-peptide is released by intra-molecular maturation allowing the enzyme to act on exogenous substrates. ISPs are homodimeric, 6 which contributes to making ISPs a structurally distinct family of subtilases. The catalytic domain of ISPs are homologous to those of other members of the Subtilisin superfamily, such as the extracellular subtilisin proteases (ESPs), which is a "Peptidase S8" domain in the Pfam classification. 8 According to the MEROPS peptidase database, 9 both ISPs and ESPs belong to the S08 family in clan SB.
Within this domain a catalytic triad is arranged as an aspartate, a histidine and a serine (Asp32, His64, and Ser221, respectively, referring to the processed SubtilisinE from B. subtilis; Uniprot ID: CAB12870). In brief, the nitrogen-bonded protein (Nε2-H) of His64 is hydrogen bonded to the hydroxyl group proton of Ser221. This interaction causes a charge separation of the hydroxyl, deprotonating the serine oxygen and activating it for nucleophilic attack on the carbonyl of the peptide substrate, which ultimately leads to breakage of the peptide bond. 10 Aside from homology within the catalytic domain, significant architectural differences exist between ISPs and ESPs. The N-termini of ESPs contain short leader sequences of about 20-30 residues for protein secretion, 11 followed by a pro-domain of typically 60-80 residues, 12,13 which is not conserved in sequence, but vital to their folding and function. 14 In an analogous manner to the ISP propeptide, the ESP pro-domain is processed intra-molecularly during maturation of the enzyme into an active conformation. The prodomain has dual roles in acting as an inhibitor, 15,16 and as a molecular chaperone that guides folding of the active enzyme. [16][17][18] The first ESP structure was solved in 1969, 19 and has since been reported for several homologues 20,21 and a number of engineered mutants. 22 For ISPs, however, structural information is known from a single homologue, the Bacillus clausii ISP, 5 Notably, all B. clausii structures are from inactive mutants carrying catalytic Ser250 to Ala mutations.
In ISPs, the leader sequence and pro-domain of ESPs are replaced with a pro-peptide (also termed N-terminal extension). The propeptide binds across the active site, with residues Phe4-Leu6 forming a central β-strand of a 3-stranded antiparallel β-sheet. 6 The propeptide also contains a LIPY/F motif, not found in ESPs. In B. clausii ISP this motif is involved in inhibiting the active site. Residues within the motif contribute to disruption of the conformation of the catalytic triad by shifting the catalytic Ser and His residues apart. 5 According to a standardized residue nomenclature for peptide binding to the active site, 23 residues N-terminal to the scissile bond of the peptide substrate are termed P4, P3, P2, and P1, and those C-terminal to the bond are termed P1', P2', P3', and P4', where the scissile bond is between P1 and P1'. The corresponding sites in the enzyme are S4, S3, S2, S1, S1', S2', S3', and S4'. In B. clausii ISP, Leu6 and Ile7 correspond to P2 and P1 and are pointing inwards into the hydrophobic pocket at the S2 and S1 sites, respectively. Pro8 holds a unique position at the centre of a small curve, which displaces the peptide bond between Ile7 (P1 site) and Pro8 (P1' site) out of reach of the active site Ser, whereas Tyr9 is occupying the S1' site. The proline-centred curve is unique in B. clausii ISP, and contrasts the scissile bond in ESPs, which is positioned to allow autoproteolytic processing. Altogether, the structure suggests that the residues in the pro-peptide are involved in blocking the active site serine. 5,6 Both ESPs and B. clausii ISP harbor a conserved high affinity metal-binding site occupied by a metal ion that serves a structural role. 5,6,24,25 The high affinity metal-binding site in ESPs is occupied by calcium, 24,26 whereas in B. clausii ISP it is occupied by sodium. 5,6 In addition, B. clausii ISP has 2 unique binding sites for divalent metal ions, probably occupied by calcium ions, in each monomer: 1 close to the dimer interface and 1 in proximity to the active site. The latter is involved in ordering a loop that contributes to formation of 1 of the binding sites (S1) involved in catalysis. Due to the processing of the pro-peptide and the positioning of calcium, the catalytic triad and substrate binding cleft are significantly rearranged, especially at the S1 binding site. 5 In a proposed model for ISP regulation, 27 it was suggested that once a minor fraction of the pool of ISPs adopts an open conformation, calcium binding takes place and reshapes the S1 binding site, which ultimately releases the pro-peptide within this population and leads to a cascade of activation of other ISPs. The sequence of events and details of how the maturation precedes, in particular the role of calcium, are not known.
This study reports an ISP from a marine isolate, Planococcus sp. AW02J18, which is from a related, but phylogenetically distinct genus to B. clausii. Here, we present biochemical data for the recombinant enzyme, showing it is active in presence of calcium, at alkaline pH and moderate temperatures. We furthermore present a highresolution structure of an ISP with an intact catalytic triad and an inhibitory pro-peptide bound across the active site. The structure supports previous findings and unique features of ISPs, such as its dimeric nature, sodium binding in the high-affinity metal-binding site and active site blocking by the pro-peptide. The processing of the propeptide appears however to be different from reported ISPs, possibly involving multiple processing steps. We also present mutagenesis data supporting an inhibitory role of the LIPY/F motif of the pro-peptide.

| In silico identification of an intracellular subtilisin protease
The ISP sequence was identified from sequence-based mining of a marine bacterial isolate, Planococcus sp. AW02J18 (Supporting Information Table S1). This isolate was collected during expeditions in the coastal areas of Lofoten in 2009, and is stored in a bacterial collection at the University of Tromsø. The sampling procedure and collection has been presented elsewhere. 28 Genomic material was isolated for Illumina sequencing (MiSeq). Using a sequence-based approach, translated genomic sequences from a marine bacterial collection were mined for subtilisin-like proteases by searching for S08 family homologs against the MEROPS database. 9 The ISP candidate was identified in this data set, and the sequence has been deposited in GenBank with the accession code MG786190.

| The LIPY/F sequence conservation
Sequences homologous to Planococcus sp. AW02J18 ISP were identified using the UniProt blast search engine (default settings) against the UniRef90 database (UniProt release 2017_10). 29 Sequence hit number 156, UniRef90_A0A136C445, was the first sequence to contain 2 motif mutations (LVNE) making the motif unlikely to be functional and was used to define the distance cut-off (expect value 4e-107; 57% sequence identity to Planococcus sp. AW02J18 ISP). Hence, the top 155 sequence hits were used to make a multiple sequence alignment (MAFFT, default settings). 30 Three sequences were fragments that lacked the LIPY/F motif, and were manually removed (UniRef90: UPI00098840FB, UPI000590D2A7, UPI000689F3EC).
The alignment containing the remaining 152 sequences was used to construct a sequence logo (default parameters). 31

| Sub-cloning of the isp gene to expression vectors
To facilitate enzyme expression we used our previously developed screening procedure for subtilisin-like serine proteases. 32 The Planococcus sp. AW02J18 ISP protein sequence was used as template for gene synthesis (GenScript), and the synthetic isp gene was codonoptimized to improve its expression in E. coli. The isp gene was synthesized with flanking SapI sites, and delivered in a customized SapI-free pUC57 vector with kanamycin selection marker. The isp gene was sub-cloned from the delivery vector to a suite of expression vectors using a fragment exchange (FX) cloning method. 33 Construction of the expression vectors have been described previously. 32

| Gene truncation and mutagenesis
Truncation constructs and mutants were prepared from the pUC57 template. Primers were designed to contain a SapI-cloning site and a 15-20 bp gene-specific region targeting the desired truncation start. Table S2 were used to amplify the truncated ISP versions by PCR using Phusion polymerase. Gene fragments were purified, and cloned into the pINITIAL cloning vector by FX-cloning. 32 Plasmids were sequenced to confirm correct truncations. Point mutations were prepared by site-directed mutagenesis using primers in Supporting Information Table S2. Truncation constructs and mutants were sub-cloned into the p12 expression vector, using FX cloning.

| Small-scale expression and analysis of protein integrity
Small-scale recombinant expression was carried out according to the protocol described previously 32 in 4 mL culture volumes. Following expression, cells were collected and resuspended in 1 mL lysis buffer (50 mM Tris-HCl pH 8.5, 50 mM NaCl, 0.25 mg/mL lysozyme, 10% (v/v) glycerol). Lysis was completed by ultrasonication using two 5-s pulses at 40-60% amplitude with a CV-18 probe powered by an Ultrasonic Homogenizer 4710 (Cole Parmer). Lysates were cleared by centrifugation at 4600 × g for 20 min. Cleared lysate samples (representing soluble fraction) were analyzed by SDS-PAGE and immunoblot as described previously. 32 As background controls, lysates containing empty vector were used, herein termed GS due to the insertion of triple GS encoding sequence as a replacement of the ccdB gene in the expression vector. 32 Semi-quantitative analysis of recombinant protein in cleared extracts was performed in Image Lab 3.0 (BioRad). Target band intensities were extracted from image data of Coomassie-stained SDS-PAGE gels, and normalized to the total protein intensities in the lane excluding the target band intensities to adjust for variable growth rates and protein expression levels.

| Protease activity assays
The protease fluorescent detection kit (Sigma-Aldrich) was used for routine detection of proteolytic activity as previously described. 32,34 Briefly, 10 μL lysate or 5 μM enzyme was assessed for activity on FITC-casein in 50 mM TrisHCl pH 8.5 (at RT), 50 mM NaCl, in absence or presence of 1 mM CaCl 2 in a total volume of 50 μL at 37 C for 1 h unless otherwise stated. Temperature optimum was assayed using the FITC-casein assay. For the mutants, activity was assessed using EnzChek™ Protease Assay Kit (ThermoFischer).

| Determining the specific activity
Specific activity was determined using a protease colorimetric detection kit (Sigma-Aldrich). To avoid assay interference with amino groups from Tris, Asn3-ISP was dialyzed against 25 mM borate/NaOH pH 8.2, 50 mM NaCl before assaying. Casein was solubilized in water at pH 8.3. One unit is defined as the amount of enzyme that will hydrolyze casein to produce color (as determined by addition of Folin-Ciocalteu's Reagent) equivalent to 1.0 μmole tyrosine per minute at pH 8.3 at 37 C in presence of 10 mM CaCl 2 .

| Differential scanning calorimetry
Prior to Differential Scanning Calorimetry (DSC) measurements, aliquots of Asn3-ISP at approximately 1 mg/mL were dialyzed into the following conditions overnight at 4 C: 50 mM Hepes pH 8.0, 50 mM NaCl (DSC buffer); DSC buffer with 2 mM CaCl 2 ; DSC buffer with 1 mM ethylenediaminetetraacetic acid (EDTA). Thermal unfolding transitions were measured using a Nano-Differential scanning Calorimeter-III (Calorimetry Sciences Corporation) from 5 to 75 C with scan rates of 1 C/s. Buffer from the final dialysis step was used as a reference.
Data were analyzed using the NanoAnalyze software (TA Instruments).

| Crystallization
Crystallization experiments were performed with a stock solution of puri- Grenoble, France) beamline ID23EH1. The data were integrated by XDS/XSCALE, 35 scaled and analyzed by programs in the CCP4 program suite 36 through autoPROC. 37 A summary of the data collection statistics is found in Table 1.

| Structure determination
The crystal structure was solved by molecular replacement using Mol-Rep in the CCP4 program package 36 with 2XRM 5 as search model (a representative structure of the homologous ISP from B. clausii). The initial refinement was executed in Refmac 38 followed by automated model improvement in Buccaneer. 39 The manual building was done in Coot 40 interspersed by cycles of refinement in Phenix 41 Table S1). According to sequence analysis, this protease contained a catalytic domain (Peptidase_S8/PF00082) as annotated by Pfam (residues 40-311, Figure 1A). Sequence analysis also revealed that it shared 53% sequence identity to the previously described intracellular subtilisin protease (ISP) from B. clausii 6 (Supporting Information Figure S1). As expected from SignalP analysis, the ISP sequence does not contain a leader sequence to direct its export, 43 and is thus predicted to have  Figure S1). Although the LIPY sequence has been reported as a conserved motif, 6  3.2 | The first 2 residues of the calcium-dependent ISP is processed The full-length isp gene from Planococcus sp. AW02J18 was subcloned to a suite of expression vectors for heterologous expression.
From SDS-PAGE analysis, we found that all recombinant constructs yielded soluble enzyme, but that solubility was further improved by use of fusion tags ( Figure 1C). Since many serine proteases require calcium for proper folding and structural stability, activity was assessed on fluorescein isothiocyanate (FITC) conjugated casein in the absence or presence of calcium ions. Compared to extracts from strains carrying empty vectors, all recombinant enzymes were active, but required calcium for activity ( Figure 1D). The p1-construct encoding an N-terminal deca-histidine (His) tag was chosen for in-depth characterization due to its potential to yield a recombinant enzyme that would mimic the native processed ISP, and ease downstream purification ( Figure 1). In the absence of calcium, immobilized metal affinity chromatography (IMAC) was used for protein purification of His-ISP (approx. 38 kDa). In analogy to the ISP from B. clausii, the enzyme was incubated in presence of calcium to mature by autoproteolysis. From SDS-PAGE we obtained a "matured ISP", with an expected lower mass (35 kDa) than full-length, of 95% purity ( Figure 2A). With N-terminal sequencing we determined the starting residue on this protein entity to Asn3; we thus termed this protein Asn3-ISP. Using Asn3-ISP, we found that increasing concentrations of calcium had a positive effect on activity ( Figure 2B), whereas EDTA inactivated the enzyme ( Figure 2C). From SDS-PAGE analysis of the reaction products, we found that Asn3-ISP was further processed or degraded in presence of calcium ( Figure 2C). In absence of calcium or in calcium-depleted reactions, the enzyme was however persistent against proteolysis ( Figure 2C), and could be stored for 1 month without any effect on activity (Supporting Information Figure S2).
To further understand the processing, calcium chloride was added at various concentrations to the full-length recombinant enzyme  Figure S3 and Table S5). No obvious sequential pattern between protein entities was identified. Tag-removal was confirmed by immunoblot analysis and compared to a catalytic mutant designed by replacing the catalytic Ser251 with Ala (Supporting Information Figure S4).
The processing of the recombinant ISP from Planococcus sp. AW02J18 appears to occur in multiple steps. identified in the electron density and modeled. These occupy similar positions around the 2 protein molecules in the asymmetric unit.

| Planococcus sp. AW02J18 ISP operates at moderate temperatures and alkaline pH
To identify its optimal conditions for further activity assessments, Asn3-ISP was characterized with respect to the specific activity, temperature and pH optimum in casein assays ( Figure 3). It was found to operate optimally at pH 11.0, but was active across pH 7.0-11.0, whereas no activity was observed below pH 6.0 ( Figure 3A). Precipitation was observed at pH 4.0 in both citrate and acetate buffers, likely explained by an estimated pI around 4. The temperature optimum was found to be around 45 C ( Figure 3B). No activity was identified above 60 C, which indicates that the protein is destabilized at high temperatures. Using optimal temperature (45 C) in alkaline conditions (pH 8.3) and 10 mM CaCl 2 the specific activity of the ISP was determined on casein to be 13 ± 1 U/mg.
To determine the thermal unfolding temperature of Asn3-ISP, DSC measurements were carried out ( Figure 4). The enzyme unfolded as a single peak, which could be fitted to 2 two-state transitions with melting temperatures (Tm) separated by approximately 3.0 C (Table 3). In the DSC data, the apparent Tm in absence of calcium and EDTA was around 60 C, which is consistent with the data on temperature optimum and stability ( Figure 4A). Addition of CaCl 2 increased the directly measured T max by 1.7 C, and the apparent Tm by up to 3.0 C indicating that calcium has a stabilizing effect on the enzyme ( Figure 4B). The presence of EDTA slightly increased the apparent T m   Figure S1).

| Mutations in the LIPY/F motif of the propeptide relieve inhibition
Removal of the first 18 residues of B. clausii ISP by calcium treatment or by truncation released an ISP enzyme in an active conformation. 5 The proteolytic site for cleavage is however not conserved among ISPs (Supporting Information Figure S1). As calcium seemed to improve activity ( Figure 2B), but also further process the Asn3-ISP ( Figure 2D), we aimed at identifying the second processing site for  Figure S3 and Table S4). As an alternative approach, we designed various constructs where the N-terminal region of the Planococcus sp. AW02J18 ISP was truncated ( Figure 6A). To design a close mimic of the N-terminus of native and processed enzyme, a p12-based construct was chosen (ISP-His, 38 kDa). This mimicked the full-length ISP sequence and respective truncation mutants with Cterminal His-tags albeit with 2 artificial residues at the N-terminus of recombinant enzyme (MS, Figure 6A). A Leu6 truncation construct was designed to remove the first 5 residues, not affecting the LIPYsequence, to assay potential detrimental effects of removal of the β1-strand of the antiparallel β-sheet required for structural stability ( Figure 6B). An Arg10 truncation construct (that is, starting at Arg10) was designed to remove the LIPY-sequence from the native N-terminus, to release auto-inhibition induced by the motif. The Thr15-Arg20 truncations were designed to truncate the pro-peptide in search for an active enzyme that would mimic the processed B. clausii ISP. Truncations beyond Arg20 were considered to be destructive as these were anticipated to interfere with secondary structure elements in the core of the catalytic domain according to the B. clausii ISP structures. 5,6 Positions of ISP truncations are summarized in Figure 6. None of the truncations were expected to impair the high affinity metalbinding site or dimerization, as previous reports have identified the binding site and the dimer interface in other distant regions of the protein. 27 According to SDS-PAGE analysis recombinant enzymes were either not obtained or below our detection limits (Supporting Information Figure S6). Growth of E. coli was not affected by recombinant expression, suggesting that active enzymes, if present, were not lost due to cell death. In case the recombinant enzymes were present at undetectable levels, the truncated enzymes were assessed in an activity assay, but found not to present activity ( Figure 6C).
In B. clausii ISP the LIPY-sequence is involved in binding the hydrophobic pocket at the active site, wherein Pro holds a critical position in displacing the scissile bond between Ile and Pro out of reach of the active site serine. 6 According to structural data on Planococcus sp. AW02J18 ISP ( Figure 5B)  Leu6Ala, and both Ile single mutants were successfully expressed, but gave lower yields than wild-type ISP ( Figure 6D). Expression levels for the Leu6Lys single mutant and the double mutant were low, if any, and FIGURE 3 pH and temperature optimum of ISP. A, Using the N-succinyl-AAPF p-nitroanilide peptide, the activity of 1 μM Asn3-ISP at pH 3.0-11.0 was measured in the initial rate of the reaction at 25 C. Background from buffer was subtracted and data was made relative to measurement data at pH 11.0. Citrate buffer was used for pH 3.0-6.0 (diamonds), acetate buffer for pH 4.0-6.0 (square), sodium phosphate buffer from pH 6.0-8.0 (down-pointing triangles), Tris-HCl buffer for pH 7.0-9.0 (up-pointing triangles, dotted line between points) and glycine buffer (circles) for pH 9.0-11.0. Error bars represent deviation between 2 replicas in 1 representative experiment. The pI of the ISP is estimated to approximately 4.4 (vertical dotted line). B, Activity of 5 μM Asn3-ISP was monitored in the FITC-casein assay across a temperature range of 25-70 C. Background was subtracted and made relative to the measured data at 45 C. CaCl 2 was added immediately before assaying. The assay took place for 1 h at the respective temperatures. Error bars represent deviation between data points from 3 independent experiments. The horizontal dotted line represents the highest background measurement a ΔH cal (calorimetric enthalpy), ΔS (entropy of unfolding) and T max are calculated directly from the unfolding transition. ΔH vH and T m are derived from fitting 2 two-stated scaled models to each transition after subtraction of buffer scans and a sigmoidal baseline variation occurred in independent experiments. The ratio of soluble protein to expressed protein was generally higher for the mutants than for wild-type ISP (data not shown). Cleared lysates containing the wildtype ISP and mutants were assessed in an in vitro BODIPY-casein assay and compared to extracts from strains carrying the empty vector ( Figure 6E). As expected, the wild-type ISP was found to be active upon calcium treatment as determined from an increase in fluorescent signal.
Upon calcium addition, the Leu6Ala, and both Ile mutants showed a similar response, but mutants showed a higher than baseline level of activity even in the absence of calcium. No activity was detected for the Leu6Lys mutant, probably because it was not expressed. The double mutant was however found to be active, despite the low expression levels. The activity of the double mutant was similar both in absence and presence of calcium, albeit low. In all cases, EDTA prevented activity, likely by chelating calcium at 1 or several binding sites.

| DISCUSSION
An ISP from Planococcus sp. AW02J18 is herein characterized in terms of its catalytic activity, stability and structure. For recombinant expression, we explored the utility of N-terminal His, His-SUMO, or His-MBP fusion tags to promote soluble expression of ISP, as previous data have shown that N-terminal tags can be used for both intracellular 1 and extracellular serine proteases. 32 Expression trials showed that all fusion constructs were soluble (Figure 1). The ISP was active in the presence of calcium (Figure 2). The assumption that ISP requires propeptide processing for activation, for example, as in B. clausii ISP, allowed exploitation of its native protease activity for intrinsic tag  Figure S3).
The ISP operates at moderate temperatures, with optimal conditions at 45 C (Figure 3), and unfolds at about 60 C (Figure 4). The organism of which this ISP originates, Planococcus sp. AW02J18, was isolated from a marine habitat, and is known to thrive at cold to moderate temperatures (data not shown). Although some ISPs are active at neutral pH, 7 Planococcus sp. AW02J18 ISP, like the majority of ISPs, 2,[44][45][46] has optimal activity at alkaline pH ( Figure 3). So far, 1 ISP has been structurally characterized, namely the ISP from B. clausii. This study provides structural information on a second unique ISP that originates from a phylogenetically and physiologically distinct genus. 47 The ISP crystallized mostly at acidic pH (Supporting Information   Table S3), and calcium was not found in any of the crystals. The lack of activity and low processing below pH 7.0 (Figures 2 and 3) may partly explain why structures are in the inactive conformation.
Whether lack of crystals at conditions above pH 7.0 is caused by degradation or because the active state does not promote crystal growth is impossible to say. Processing is not induced by pH shift alone ( Figure 2D), but requires calcium. Both ISPs were found to crystallize in a dimeric state; thus, dimerization appears to be a generic feature of ISPs. According to size exclusion chromatography, the presence of calcium the Asn3-ISP lead to a mixed population of quaternary structures corresponding to approximately 2.5 and 3.7 monomers per oligomer (Supporting Information Figure S5). Whereas the dimeric form is From studies of B. clausii ISP, divalent metal ions, possibly calcium, bind close to the S1 pocket. 5,6 In the crystals of Planococcus  Figure S1). This could indicate a specific role of calcium in the transition from inactive to active enzyme, not only for the B. clausii ISP, but also for other ISPs. Asn3-ISP from Planococcus sp. AW02J18 was active in presence of calcium, but susceptible to self-degradation ( Figure 2). The fact that ISPs were not active without exogenous addition of calcium suggests that available metal binding sites were not occupied after production. Due to conservation of calcium-coordinating residues (Supporting Information Figure S1), and the need for high EDTA concentrations to inhibit activity ( Figure 2C), low affinity for calcium is likely not the case. DSC results suggest that additional calcium is only slightly stabilizing, and tightly bound calcium (removable with EDTA) is not essential for overall stability (Figure 4).
DSC showed however that calcium does have a minor stabilizing effect; thus suggesting that the added calcium in our assays contribute to minor structural rearrangements.
It is likely that there are structural rearrangements, such as propeptide flip-out or removal, in order for the 2 loops to order and coordinate calcium. The IP residues of the LIPY/F motif in the pro-peptide are spatially close to residues in 1 of the loops that need to be reoriented upon calcium binding. The 2 residues form hydrophobic interactions to the side-chain of Phe195 in our inactive structure and probably hinder this reorienting into the active conformation (this side-chain appears to be shifted almost 15 Å in the active state).
It is likely that the pro-peptide in the ISP from Planococcus sp. AW02J18 is removed, in analogy to several Bacillus ISPs. 5,7 From the available structures of ISPs with intact pro-peptides (PDB ID: 6F9M, 2X8J, 2WWT, 2WVT, whereof 2 are shown in Figure 5B) and the sequence alignment (Supporting Information Figure S1), we found that 2 short beta-strands in the pro-peptide are likely structurally conserved. The secondary structure elements are stabilized by main chain interactions, which are sequence independent. A unique feature of the Planococcus sp. AW02J18 ISP that is not found in homologous ISPs is the presence of the 2 consecutive proline residues in the transition between the pro-peptide and the catalytic domain (Supporting Information Figure S1). The removal of the ISP pro-peptide in Planococcus sp. AW02J18 appears to be different, and possibly involves several steps ( Figure 2D). In the first step the 2 first residues of the ISP (Met1, Lys2) are removed (protein band numbered 2, Figure 2D), as identified in the crystal and by N-terminal sequencing. Another product, which appears as the main product (around 30-35 kDa) at pH 8.5 in presence of 10 mM CaCl 2 (protein band numbered 3, Figure 2D), could possibly be functional. This product could in principle arise from processing of the C-terminal region of the protein, too, albeit not identified in the crystal or MS analyses (Supporting Information Figure S5 and Table S4). The N-terminal residues of this protein could not be identified. Unfortunately, MS analyses did not reveal obvious processing patterns at the N-terminal in the protein species from the SDS-PAGE analysis (Supporting Information Figure S5 and Table S4). This may partly be due to a lack of sequential degradation. Ultimately, we could not determine which ISP moiety that is responsible for or contribute to the activity identified in assays. A truncation experiment was conducted to trim the pro-peptide in the hunt for processing site(s). Two artificial residues (Met-Ser) are unavoidably added to the N-terminal end of these truncation constructs, which arise from fusion of the isp gene fragment to the start codon and the ligation seam added during sub-cloning ( Figure 6A), and their negative interference on protein stability cannot be ruled out. Sequence analysis of Planococcus sp. AW02J18 ISP, reveals that it contains 2 prolines in the transition from the pro-peptide to the catalytic domain (Supporting Information Figure S1). Whereas Pro at the P2 site is likely accepted, Pro at the P1 is highly unlikely due to the preference of hydrophobic residues at the S1 site. 27 Multiple prolines are normally not present in sites for autoproteolysis by serine proteases, 48 and the prolines may instead serve a structural role. 49 This does not however rule out that other proteases, for example proline-specific endopeptidases, could process and remove the pro-peptide in native conditions, or that processing site(s) are in other regions that were not included in this study.
Although it has been found that the pro-peptide of B. clausii ISP has a role in inhibition, the contribution of the conserved residues within the LIPY/F-motif has not been studied in detail. Due to the fact that Leu and Ile are conserved in the motif, and that the ISPs likely prefer hydrophobic amino acids at the S2 and S4 sites, 27 Figure 7). The substitution to lysine however seems to both reduce expression level ( Figure 6D). It furthermore does not respond on calcium addition in the activity assay ( Figure 6E). Assuming that the mutant is properly folded, inhibition could be explained by the possibility that lysine can form hydrogen bond and/or salt bridge interactions with the catalytic Asp49 and the Asn84 residues, respectively ( Figure 7). Structural explanations for the Ile7 mutants were not conclusive due to their proximity to the flexible region (183-193), but it is likely that both mutations cause reduced interactions with the propeptide. We thus conclude that the pro-peptide, with the LIPY/F motif in a central position, is involved in inhibition. Our data is in line with the proposed ISP model, 27 suggesting that calcium binding at the active site is prevented during pro-peptide inhibition.