Introduction

Effective degradation of plant biomass is a pre-requisite for biorefining of various plant feed stocks and underpins many food and feed production processes. However, plant biomass breakdown is challenging because of the inherent recalcitrance of plant cell walls. Many cell-wall-degrading enzymes are able to catalyze the selective cleavage of particular bonds in plant cell wall matrixes, and significant progress has been made within the last decade with respect to the discovery, characterization, cloning, and expression of plant cell-wall-modifying enzymes (Bauer et al. 2006). However, the main focus of the majority of the studies has been on carbohydrate-modifying enzymes and their industrial applications, e.g., cellulases, hemicellulases (Beg et al. 2001; Bhat 2000; Buckeridge et al. 2012; de Vries and Visser 2001), and pectinases (Mei et al. 2013; Silva et al. 2011), whereas proteases that may catalyze the degradation of cell wall structural proteins have been less studied. Nevertheless, it has been shown (a) that some of the most efficient microbial plant pathogens produce potent proteases (Fedatto et al. 2006) and (b) that specific proteases attack plant cell wall proteins (Dow et al. 1998).

Since the plant cell wall proteins are important structural components of the plant cell wall, their effective degradation plays a key role in plant biomass decomposition processes. The development of a better understanding of enzymes being able of efficiently catalyzing the decomposition of plant cell wall proteins is therefore an important, but surprisingly overlooked, research area in plant biomass processing.

Structural proteins in most plant cell walls are highly glycosylated, and they are classified into four categories: hydroxyproline-rich glycoproteins, arabinogalactan proteins, glycine-rich proteins, and proline-rich proteins (Showalter 1993). Proline and hydroxyproline rich glycoproteins are unique and abundant in plant cell walls. Hydroxyproline (Hyp) is generally believed to serve as an anchor point for glycosylation; for instance, the motif SerHyp4 of lectin and extensin is conserved throughout the plant kingdom (Showalter 1993; Kieliszewski and Lamport 1994). The bulky glycosylated side chains may provide a steric hindrance and thereby block the accessibility of proteases to the substrate. Microbial plant pathogens, however, have evolved sophisticated mechanisms to degrade plant cell wall components, including secreted enzymes that specifically target these glycoproteins.

Pectobacterium carotovorum (previously known as Erwinia carotovora) is a Gram-negative bacterium, which causes soft root disease in many plant species resulting in cell death through enzymatic plant cell wall destruction (Bell et al. 2004; Toth et al. 2003; Kyostio et al. 1991). The gene encoding the metalloprotease Prt1 from P. carotovorum subsp. carotovorum strain EC14 (MEROPS 001045, family M4 unassigned peptidase, Clan MA) (http://merops.sanger.ac.uk) was sequenced some 20 years ago, and it has been shown that this protease is an important virulence factor of P. carotovorum, as deletion of the gene hampers the pathogenesis of the strain (Kyostio et al. 1991). Nevertheless, the mechanisms of attack on plant cell wall proteins and substrate sub-site specificity of the enzyme have not been characterized in detail. Neither is it known how the N-terminal processing of this protease takes place.

With the perspective of developing new, targeted enzyme-based plant biomass processing strategies, the objective of the present work was to obtain further insight into the N-terminal processing, thermal stability, substrate targets, and substrate cleavage sites of Prt1. Hence, in this study, we cloned the prt1 gene and expressed it in Escherichia coli Rosetta™ 2 (DE3)plySs, assessed the reaction optimum, thermal stability, and metal ion dependency of the enzyme, and obtained insight into the N-terminal autoprocessing of Prt1 by sequencing the pre- and pro-peptides. In addition, by use of immuno-glycan microarrays, we report that the Prt1 enzyme can modify extensin present in extracts from different plant species.

Materials and methods

Strains and vectors

The prt1 gene was codon optimized and synthesized by DNA 2.0© (Palo Alto, CA, USA) based on the amino acid sequence available for the enzyme in the UniProt database (accession no. Q99132), but with one amino acid changed (T2A); this slightly changed, codon-optimized sequence has obtained GenBank accession no. KJ526380. The gene was cloned into the expression vector pETM-11 (EMBL©, Heidelberg, Germany), with a sequence for an N-terminal His6 tag and a TEV site. The E. coli strain DH5α was used as cloning host, and the E. coli strain Rosetta™ 2 (DE3)plySs was used as the expression host (the gene was codon optimized for Pichia pastoris expression, but since the Pichia expression was unsuccessful, E. coli was used). The chemically competent cells were purchased from Novagen® (EMD Millipore, Billerica, MA, USA)

Mutation of prt1

Primers for prt1 mutagenesis are listed in Table S1 (in the Supplementary Material). The mutations were generated by the QuickChange® site-directed mutagenesis kit (Agilent Techniques, Santa Clara, CA, USA). Four mutations were generated: insertion of aspartic acid after R37 (InsD), insertion of phenylalanine after P43 with the InsD: insD38/insF44 (InsF), mutation of arginine37 into alanine (MutA), and an alternative loop mimicking the Ca2+-binding loop of thermolysin (loop) (-R37/mutE41 → D/mutG42 → N/-P44).

Protease expression

The Rosetta™ 2 (DE3)plySs competent cells were transformed with the vector pETM-11-prt1 and cultivated in 1-L LB culture (in a 5-L flask) with kanamycin and chloramphenicol. The cell culture was incubated at 37 °C with shaking at 250 rpm until an OD600 of 0.6 was achieved. The culture was transferred to 20 °C and 150 rpm before 1 mM isopropyl β-d-1-thiogalactopyranoside (IPTG) was added, and induction of the expression was carried out for 16 h. The cells were harvested by centrifugation and disrupted by sonication in 20-mL lysis buffer (20 mM sodium phosphate, 500 mM sodium chloride, 10 mM imidazole, pH 7.0, with lysozyme and DNaseI). The cell debris was removed by centrifugation, and the extract was saved for purification.

His-tag affinity purification

A HisTrap HP Ni column (5 mL) and ÄKTApurifier (GE healthcare, Piscataway, NJ, USA) were used to purify the protease essentially as described by Silva et al. (2011), except that buffer A was 20 mM sodium phosphate containing 5 mM imidazole and buffer B was 20 mM sodium phosphate containing 200 mM imidazole (pH 7.0). To avoid imidazole affecting the thermal stability measurements on the enzyme (Silva et al. 2013), imidazole was removed by size exclusion chromatography (PD10) using 50 mM Bis-Tris containing 500 mM NaCl and 10 % glycerol (pH 6.0) for elution. The amounts of pure Prt1 protease were quantified with the bicinchoninic acid (BCA) assay (Pierce® BCA protein assay kit, Thermo Scientific, Waltham, MA, USA).

Enzyme activity assays

Two enzyme activity assays were applied: The first assay used the azurine cross-linked (AZCL) casein agarose plate for qualitative analysis. The second assay used azocasein for quantification. Both the AZCL casein and azocasein substrates were purchased from Megazyme (Megazyme, Bray, Ireland).

AZCL assay

A total of 10 g/L of melted agarose was suspended in a solution of 50 mM Bis-Tris buffer (pH 6). A total of 1 g/L AZCL casein or AZCL-collagen substrate was added, and after mixing, the suspension was poured into Petri dishes. After the agarose solidified, holes with a 7-mm diameter were cut and enzyme was added into each hole, and the plate was incubated at 37 °C overnight. Blue halos that formed around the holes indicated the endoprotease activity. The negative control was buffer, and the positive control was the commercial plant cell-wall-degrading multicomponent enzyme preparation Laminex C2K derived from Penicillium funiculosum (DuPont Nutrition Biosciences ApS, København K, Denmark).

Azocasein assay

The procedure was carried out as described by Megazyme. One unit of enzyme activity was defined as the amount of azocasein that was degraded to release an amount of soluble peptides producing an absorbance of 0.1 at 440 nm per min with 0.98 μM of enzyme at pH 6.0 and 50 °C.

Optimal conditions and kinetics

The optimal conditions for enzyme activity were studied with the following parameters: temperature, pH, and Zn ion concentration. The software MODDE® 7.0 (Umetrics, Umeå, Sweden) was used as an aid to design the experiments and to analyze the data by multiple linear regression. The pH buffers were 50 mM sodium acetate (pH 4.0), 50 mM Bis-Tris (pH 6.0), and 50 mM Tris base (pH 8.0), and reaction time was 10 min. The enzyme kinetic constants, K M, and k cat, were calculated from a Hanes-Woolf plot (Hanes 1932).

Thermal stability and effects of metal ions

Thermal stability was studied with the following temperatures: 40, 50, 55, 60, and 65 °C. The enzyme was preheated at different temperatures for 0, 5, 10, 15, and 20 min. After the pretreatment, the enzyme was cooled down to room temperature, and subsequently, the residual enzyme activity was measured at the optimal conditions.

First-order rate constants for inactivation (k D) of the enzyme at different temperatures were determined from the slopes of semi-logarithmic plots according to the equation: lnA = −k D t, while the half-life (t 1/2) at selected temperatures was determined according to t 1/2 = ln(2) / k D (Riffel et al. 2007).

In order to study the ion effects on thermal stability, 1 mM of ZnCl2 or CaCl2 or a mixture of ZnCl2 and CaCl2, respectively, was added to the Prt1 protease before incubation. General effects of other metal ions (added at a level of 1 mM each) were also carried out to compare the influence of MgCl2, CaCl2, ZnCl2, FeCl2, MnCl2, and CuCl2 on the Prt1 activity.

Protease inhibitors and metal chelators

The effects of protease inhibitors, phenylmethylsulfonyl fluoride (PMSF), 4-amidinophenylmethane sulfonyl fluoride (APMSF), aprotinin, bestatin, E-64 protease inhibitor, leupeptin, and pepstatin A (Calibiochem®, EMD Millipore, USA), and metal chelators ethylenediaminetetraacetic acid (EDTA), ethylene glycol tetraacetic acid (EGTA), and 1,10-phenanthroline were determined after incubation for 1 h at room temperature. The concentration of the inhibitors used was 1 mM. The residual enzyme activity was calculated after measurement of the initial rate.

MS sequencing

In-gel trypsin digestion and in silico trypsin digestion fingerprinting were carried out to enlighten the mechanisms of Prt1 processing. The protein bands were excised from the SDS-gel, and in-gel digestion with trypsin was performed. The extracted peptides were analyzed by LC-mass spectrometry(MS)/MS on an LTQ Orbitrap XL mass spectrometer (Thermo Fisher Scientific) equipped with a nano ESI ion source and an EasyLC nanoflow HPLC as described previously (Olsen et al. 2013). The data were searched against the enzyme sequence using the X! Tandem search engine (Craig and Beavis 2004) through GPMAW v. 9.21 (Lighthouse data) for interface and visualization. Search parameters were as follows: trypsin, precision 10 ppm, fragment error 0.50 Da, maximum missed cleavage 2, carbamidomethylation of cystein residues as fixed modification, oxidation of Met as variable modification, maximum value 0.01, and semi-cleavage as refinement. By comparing to the in silicotrypsin digestion, the sequences of all the peptides identified by MS were solved.

Analysis of Prt1 cleavage patterns

Reactions containing 200 μL of 50 mM Tris-HCl buffer, pH 6 or 8, with or without 1 mM ZnCl2, 20 μg/mL of the metalloprotease and 1 mg/mL of substrates, either bovine milk β-casein, bovine serum albumin (Sigma-Aldrich, St. Louis, MO, USA), or bovine pancreas ribonuclease A (Calbiochem, Darmstadt, Germany) were carried out at 40 °C under shaking at 600 rpm. Aliquots were taken every 30 min for 3 h, and the reaction was terminated by addition of formic acid to 2 % v/v. The supernatant was collected after centrifugation for 5 min at 13 000 g. Each substrate was also examined by a tryptic digestion (3 h at 37 °C), containing 50 mM ammonium bicarbonate, 20 μg/mL trypsin, and 1 mg/substrate. The samples were then analyzed using a 4800 Plus matrix-assisted laser desorption ionization tandem time-of-flight (MALDI TOF/TOF) Analyzer (Applied Biosystems, Foster City, CA, USA) in positive mode.

Cleavage of potato lectin

One percent (w/v) potato lectin (Sigma-Aldrich) was prepared in 50 mM Bis-Tris buffer (pH 6). One hundred microliter of lectin was mixed with 50 μL of Prt1 protease (1 mg/mL) and incubated at 50 °C for 1 h. Aliquots were taken at 1, 5, 10, and 60 min. Bovine chymosin in the form of the commercially marketed product CHY-MAX® M (Chr. Hansen A/S, Hørsholm, Denmark) and buffer were used as controls. After heat inactivation, the reaction mixture was analyzed by the use of 4–12 % gradient sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) (Bio-Rad, Hercules, CA, USA).

Immuno-glycan microarray analysis of extensin cleavage

The extraction of cell wall components and subsequent analysis was performed essentially as described before (Moller et al. 2007). Briefly, 50 mM diaminocyclohexanetetraacetic acid (CDTA) extracts from plant species were spotted using a microarray robot (Sprint, Arrayjet, Roslin, UK). Each extracted sample was printed as a twofold dilution followed by three fivefold dilutions. All sample dilutions were performed in printing buffer (55.2 % glycerol, 44 % water, 0.8 % Triton X-100). The samples were printed onto nitrocellulose membrane with a pore size of 0.45 μm (Whatman, Maidstone, UK). Once printed, arrays were blocked with phosphate-buffered saline (PBS) (140 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 1.7 mM KH2PO4, pH 7.5) containing 5 % w/v low fat milk powder (5 % modified phosphate-buffered saline (M-PBS)) for 1 h. Hereafter, they were washed in PBS and incubated with Prt1 at 0.5 mg/mL, Prt1 at 0.05 mg/mL, or xyloglucanase (E-XEGP, Megazyme International Ireland, Bray, Ireland) at 0.5 U/mL or 100 mM sodium acetate buffer pH 6 at 40 °C at 100 rpm overnight (protein solutions were prepared in 100 mM sodium acetate buffer, pH 6). After incubation, arrays were washed thoroughly with PBS, followed by 2-h incubation with extensin-directed monoclonal antibodies JIM11, JIM12, JIM20, and LM3 (PlantProbes, Leeds University, UK) diluted tenfold in 5 % M-PBS. Subsequently, arrays were washed in PBS and incubated with anti-rat secondary antibody conjugated to alkaline phosphatase (Sigma, Poole, UK) diluted 1/5000 in 5 % M-PBS. After washing with PBS and dH2O, the arrays were developed in a solution containing 5-bromo-4-chloro-3-indolylphosphate and nitro blue tetrazolium in alkaline phosphatase buffer (100 mM NaCl, 5 mM MgCl2, 100 mM diethanolamine, pH 9.5). Developed microarrays were scanned at 2,400 dpi (CanoScan 8800 F, Canon, Søborg, Denmark) and converted to TIFFs. Antibody signals were measured using appropriate software (Array-Pro Analyzer 6.3, Media Cybernetics, Rockville, USA). Data were presented as two data sets where maximal spot signal was set to 100 and all other values normalized accordingly. Color intensity is correlated to mean spot signal value, and a cutoff of 5 was applied.

CD measurements

The circular dichroism (CD) measurements of wild-type Prt1 and the mutants were carried out in Jasco J815 CD spectropolarimeter (Easton, USA) equipped with a temperature control unit. The enzyme was assayed in 20 mM potassium phosphate buffer (pH 7). Far-UV CD spectra were recorded at 4, 25, and 80 °C with protein concentrations between 2 and 15 μM in a 1-mm cell. Control baselines were measured with buffer in the absence of protein. The levels of secondary structure were estimated from the spectra using the software package CDSSTR, Sreerama and Woody 2000) through the Dichroweb analysis web server (http://dichroweb.cryst.bbk.ac.uk/html/home.shtml) (Whitmore and Wallace 2007). Thermal transitions were measured from 4 to 85 °C with a ramp rate of 1 °C/min and a protein concentration of 0.2–1.5 μM in a sealed 10-mm cell while stirring. After reaching the desired temperature, the enzyme was incubated for 60 s and data was collected for 32 s. Thermal denaturation was followed at 225 nm where the change in ellipticity as a function of temperature [Θ]T was most pronounced. Determination of the temperature at which 50 % of the native state CD signal remained was determined as described previously (Zeuner et al. 2014).

Results

Processing of the pre-pro-protease polypeptide

The expression of recombinant Prt1 produced a yield of 40 mg of pure Prt1 enzyme per liter of cell culture. The molecular weight (MW) of the full length Prt1 was 41 kDa (Fig. S1: Pellet in the Supplementary Material) whereas the MW for the purified protein was 37 kDa (Fig. S1: Prt1 in the Supplementary Material). Further cleavage gave rise to the mature Prt1 protease, which had a MW of 34 kDa (Fig. S1: Prt1* in the Supplementary Material). The MW of 41 kDa agrees with the theoretical value as determined by the ExPASy parameter (www.expasy.org), whereas the MW of 37 kDa is a result of cleavage of the pre-peptide, and the MW of 34 kDa is due to cleavage of the pro-peptide. Elucidation of the sequential cleavage points of Prt1 was accomplished by in-gel digestion followed by MALDI TOF/TOF analysis and MS sequencing. After fitting the MS data with an in silico digestion of Prt1, we were able to delineate the sequences of the pre-peptide, pro-peptide, and the mature Prt1, respectively (Table 1). It was found that the Prt1 pre-peptide comprised amino acid residues 2–34 (Table 1, italic, from residues KH to CS), and the pro-peptide comprised residues 35–67 (Table 1, bold, from residues VI to LM) (the full amino acid sequence of the pre-pro-Prt1 is shown in Fig. S2 in the Supplementary Material). The activation of Prt1 by N-terminal autoprocessing was corroborated by the findings that the protease activity increased 1.8 times upon incubation of the freshly purified enzyme at 37 °C for 45 min (Fig. S3 in the Supplementary Material). Longer incubation of the enzyme, however, appeared to cause autocatalytic hydrolytic cleavage that gradually eliminated the activity of the Prt1 enzyme (Fig. S3 in the Supplementary Material).

Table 1 Digestion of pre-pro-Prt1 by trypsin: peptide identification
Full size table

Proteolysis of casein, collagen, and lectin

The enzymatic activity of the purified mature Prt1 was validated by showing that Prt1 was active on both casein and collagen (AZCL casein and AZCL collagen) in plate screening. The plate screening assay thus showed that Prt1 was active on both proline- and hydroxyproline-rich substrates (casein and collagen, respectively, Brinckmann (2005)) (Fig. S4 in the Supplementary Material). Potato lectin is a hydroxyproline-rich glycoprotein representative of plant cell wall extensin. Pure potato lectin has a MW of 60–70 kDa, and incubation with Prt1 decreased the amount of the lectin as assessed by SDS-PAGE: After 60 min of incubation with Ptr1, only trace amounts were detected, whereas chymosin (MEROPS A01.006) had no effect on potato lectin (Fig. 1).

Fig. 1
figure 1

Separation by SDS-PAGE (4–20 %) of pure potato lectin subjected to hydrolysis by the Prt1 protease. Protein markers (lane 1); potato lectin control (buffer) (lane 2); potato lectin incubated with Prt1 for 1, 5, 10, and 60 min (lane 3–6); and potato lectin after incubation with chymosin (lane 7)

Full size image

Optimal conditions for enzymatic activity

Multivariate regression analysis showed that pH (4 to 8) was a significant factor on the protease activity (Table S2 in the Supplementary Material); in contrast, neither the temperature (30 to 70 °C) nor the Zn2+ ion concentration (0 to 1 mM) was significant reaction factor for the activity (Fig. 2 and Table S2 in the Supplementary Material). The predicted optimal conditions for activity were pH 6.0 and 50 °C (Fig. 2).

Fig. 2
figure 2

Surface response as a function of temperature and pH on the Prt1 activity. Ptr1 was incubated with azocasein, and the activity was measured by recording the release of soluble azocasein peptides by 440 nm after 10-min incubation: a at 0.5 mM ZnCl2 and b at 50 °C. The optimal conditions were determined to be as follows: pH 6.1, temperature 50 °C, and 0.5 mM ZnCl2. Further information on the surface response is given in Supplementary Material, Table S1

Full size image

At pH 6.0 and 50 °C, on azocasein as substrate, the K M was determined to be 1.81 mg/mL, and k cat was determined to be 1.8 × 107 U M−1 (calculation is shown in Fig. S5 in the Supplementary Material).

Thermal stability

At pH 6.0, Prt1 remained stable at 40 °C for at least 20 min (k D ~0.0), but the protease gradually became unstable above 50 °C (k D = 0.034) which was the optimal working condition. The protease lost half of the activity when incubated at 55 °C for 10 min and was rapidly inactivated when incubated at 65 °C (Fig. 3a). By addition of Zn2+, the protease became less heat stable and the activity started to decrease already at 40 °C (k D = 0.03). At 50 °C, the activity decreased much faster (k D = 0.12) than without Zn2+ added, i.e., the enzyme was 3.5 times less stable with Zn2+ added compared to the corresponding base case (Fig. 3b). In contrast, addition of Ca2+ ions did not affect the enzyme stability or the activity (Fig. 3c), while combination of Zn2+ and Ca2+ ions decreased the thermal stability to the same extent as Zn2+ alone (Fig. 3d).

Fig. 3
figure 3

Plot of residual activity of Prt1 after heat inactivation (LnU) versus heating time: a mature Prt1, b Prt1 with 1 mM of Zn2+, c Prt1 with 1 mM of Ca2+, and d Prt1 with 1 mM of Zn2+ and Ca2+

Full size image

Effects of protease inhibitors, chelators, and metal ions

The chelators EDTA, EGTA and 1,10-phenanthroline completely inhibited the Prt1 protease activity, whereas the enzyme was not sensitive to PMSF or to a cocktail of protease inhibitors (Fig. 4a). Zn2+ was the only divalent metal ion which stimulated the Prt1 protease activity (Fig. 4b), whereas Ca2+addition had no effect (Fig. 4b). Fe2+ and Cu2+showed ~30 and 15 % inhibition each, respectively, whereas no effect was observed with Mg2+ or Mn2+addition (Table S3 in the Supplementary Material). The effects of the EDTA treatment of Prt1 could thus be reversed by addition of Zn2+ with a recovery of >100 % of the initial activity (Fig. 4b); the increase in activity to 200 % relative activity with Zn2+ addition is presumed to be due to the base case not being completely saturated with Zn2+. Although Zn2+ did not seem to affect the yields of the reaction after 10 min (Fig. 2), it was observed that the initial rates of the reactions with Zn2+ added were higher (Fig. 4b). Taken together with the thermal stability data, this result implies that Zn2+ ions not only activated the proteolytic activity of Prt1 but also activated the autolytic activity, so that while Prt1 becomes more active, it also becomes less stable in the presence of Zn2+. Calcium was not able to reconstitute the activity of Prt1 pre-treated by EDTA (Fig. 4b).

Fig. 4
figure 4

a The effects of various protease inhibitors and chelators on relative Prt1 activity. b The effects of Zn2+ and Ca2+ (1 mM) on the recovery of Prt1 activity after EDTA treatment: E* is Prt1 after incubation and subsequent removal of EDTA. E* + Zn is the Prt1 reconstituted with ZnCl2, and E* + Ca is the Prt1 reconstituted with CaCl2. Relative activities are recorded in

Full size image

Prt1 cleavage patterns on β-casein, ribonuclease A, and BSA

To investigate the cleavage patterns of Prt1, three substrates (β-casein, ribonuclease A, and BSA) were used for digestion by the protease, and all three substrates were effectively cleaved by this enzyme. After mapping the sequence of the peptides detected by MALDI TOF/TOF analysis, 16, 23, and 17 cleavage sites were identified for β-casein, ribonuclease A and BSA respectively, i.e., a total of 56 cleavage sites for all three substrates. Based on these 56 cleavage sites, the frequency of each amino acid occurrence at the cleavage site (P4, P3, P2, P1, P1′, P2′, P3′, and P4’, nomenclature of Schechter and Berger (1967)) was determined (Table 2). A consensus sequence of amino acids around the cleavage site (P4-P3-P2-P1-P1′-P2′-P3′-P4′) as being (Ala)-(Pro/Glu/Gln)-(Lys/Glu/Gln)-(Pro)-(Ala/Val/Phe)-(Ala/Leu)-(Ser)-(Ser/Glu) could be proposed (Fig. 5). The results thus suggest that Prt1 has a preference for cleaving peptide bonds constituted by a proline followed by a hydrophobic residue. For comparison, in a previous study, the metalloprotease Bacillus thermolysin was found to be much less specific (Keil 1992). The relatively high frequency of preference for proline at or near the cleavage site corroborated that Prt1 may have high affinity for attacking proline-rich and hydroxy-proline-rich plant glycoproteins.

Table 2 Frequency of residues (%) around the substrate cleavage site when cleaved by Prt1
Full size table
Fig. 5
figure 5

Cleavage patterns of Prt1 indicating the proposed elevated preference for cleavage with P in position P1 and A, V, or F in position P1′ in the substrate (re. Table 2). P N and P N′ being the Nth residue, respectively, found before and after the cleavage site

Full size image

Prt1 activity on extensin extracted from plants

The immuno-glycan microarrays study showed that Prt1 could catalyze the degradation of extensin extracted from various plant species (Fig. 6). This type of assay is based on the principle that enzyme activities can be detected by a loss or reduction of antibody binding to the substrate after enzyme treatment. Reduction in antibody binding may result in modification or degradation of epitopes (Fangel et al. 2012). In this case, a set of well-established anti-extensin antibodies were used, and the substrates were array-immobilized crude extracts from a variety of plant species, all of which apart from Ginkgo biloba contained extensin, as evidenced by antibody binding to buffer-treated control arrays (Fig. 6a). In all the other species, the signals obtained from all the anti-extensin antibodies were reduced (and in some cases abolished, Fig. 6b) on arrays treated with Prt1 compared to buffer-treated arrays. The fact that xyloglucanase did not lead to a general decrease in antibody binding indicates that the observed reduction of signals was a Prt1-specific effect (Fig. 6a). Although the antibodies used have all been characterized as extensin-specific, their epitopes have not been fully resolved. However, all are thought to recognize carbohydrate moieties (Smallwood et al. 1995). The observed decreases in antibody binding therefore indicate an indirect effect, namely, that degradation of core proteins by Prt1 caused release of epitope-bearing glycan side chains. These data are significant because they indicate that Prt1 is capable of accessing protein substrates within the context of native extensin glycosylation. This may have important implications for the practical utilization of Prt1 during biomass processing.

Fig. 6
figure 6

a CDTA extracts from six different plant species were printed in a microarray format as described in Moller et al. (2007). Each extract was represented by two replicates and four dilutions resulting in a total of eight spots per sample. Identical arrays were treated with Prt1 at 0.5 mg/mL, Prt1 at 0.05 mg/mL, commercial xyloglucanase at 0.5 U/mL, and buffer, respectively. Afterward, arrays were probed with anti-extensin monoclonal antibodies (JIM11, JIM12, JIM20, and LM3), and detected binding for each antibody is presented as a sub-heat map where color intensities are proportional to mean spot signal values. The highest mean signal value was set to 100, and all other values were adjusted accordingly. Different array treatments are indicated at the top, while extracted plant species are listed on the right of the heat map. b JIM12 and LM3 antibody signals detected for Buxus sempervirens CDTA extracts are represented in two sub-heat maps. The normalized mean signal values for the four dilutions are shown individually. Below each heat map, a part of the corresponding arrays is shown in order to visualize the correlation between heat map values and spot intensities observed on the arrays

Full size image

Secondary structure elements by CD

CD spectra were measured for Prt1 and are shown in Fig. 7a. A comparison of the predicted secondary structure contents from the spectra to the structural content of thermolysin showed that 15 % of the residues in Prt1 were predicted to be in a helix conformation, 33 % in a sheet, 22 % in a turn, and 30 % in an unordered conformation (Table 3). The secondary structure contents deviated significantly from the structure of thermolysin, which exhibits higher helix content, but lower sheet content (Table 3). Thermal denaturation analysis of Prt1 by assessment of the residual native state mean residue molar ellipticity as a function of temperature revealed that the T m, i.e., the temperature at the transition midpoint (where 50 % of the native state signal remains) of the enzyme, was 52.6 °C (Fig. 7b).

Fig. 7
figure 7

a Far-UV circular dichroism spectra of Prt1 at 25 °C, recorded in 20 mM phosphate buffer, pH 7.0. b Temperature transitions of Prt1. The fraction of the native state mean residue molar ellipticity is shown as a function of temperature

Full size image
Table 3 Prt1 secondary structure determination from far-UV-CD measurements
Full size table

Mutations generated to improve thermal stability

To assess the options for generating a calcium binding site, in order to improve the thermal stability of Prt1, we introduced four site mutations based on the sequence alignment between the N-terminus of thermolysin and Prt1. Both proteases share a conserved motif with the sequence LPGX1LX2 where X 2 is a hydrophobic residue. This is the most N-terminal conserved motif in the consensus sequence of Prt1 and thermolysin and is followed in the latter by a calcium binding site. In Prt1, the amino acid sequence following the conserved motif seems shortened and lacks the calcium complexing aspartic acid residues. We therefore chose sites located in the immediate C-terminal vicinity of the conserved motif to mutate or insert amino acids (InsD, InsF, MutA, or loop). All four enzyme mutants were successfully expressed and purified, but their activity was lower than that of the wild type (Fig. S6a in the Supplementary Material). The CD spectra of the mutants InsD and InsF deviated strongly from the wild-type signal (Fig. 8a), exhibiting random coil spectra with little helix but significant sheet contents. Accordingly, the activity of these mutated enzymes was significantly decreased (Fig. S6a in the Supplementary Material). In contrast, the CD measurements showed that the MutA and loop mutant enzymes had secondary structure contents closer to those of thermolysin than the native wild-type Prt1, i.e., they contained more helix than the Prt1 wild type but less sheet structures, while the percentage of turns stayed the same (Table S4 in the Supplementary Material and Fig. 8a). Accordingly, these two mutants MutA and loop had slightly higher T m values than the wt (Fig. 8b). For MutA, a T m of 58.2 °C was determined, while it had only 75 % of the wild-type activity under optimal conditions (for the wild type) (Fig. 7a). The T m for the loop-mutant was determined to be 54.3 °C, but its activity was only 50 % of the wild-type activity when assessed at 50 °C (Fig. S6a in the Supplementary Material and Fig. 8b). Compared to the wild type, the two mutant enzymes MutA and loop also lost activity significantly faster than the wild-type Prt1 when incubated at 40 °C or above for 30 min (Fig. S6b in the Supplementary Material). Mutant enzymes with higher T m but lowered activity and thermal stability may immediately seem contradictory; however, taken together, the data essentially suggest that elevated temperatures did not affect the secondary structure but distorted the catalytic site. Our small set of mutations thus suggests that the N-terminus of Prt1 defines its thermal stability and helps maintain the enzymatic activity at elevated temperatures.

Fig. 8
figure 8

Far-UV circular dichroism spectra of wild-type Prt1 in comparison with four designed Prt1 variants at 25 °C (a). The data was recorded in 20 mM phosphate buffer, pH 7.0. Temperature transitions of wild-type Prt1 and the variants MutA and loop (b). The fraction of the native state mean residue molar ellipticity is shown as a function of temperature. The color code is the same as in (a), and the T m values are given in the figure

Full size image

Discussion

In this study, Prt1 was demonstrated to undergo a pre-pro-auto activation process, which appeared to be similar to that of other extracellular proteases (Wu and Chen 2011). Autoprocessing of protease N-termini, as observed for the pre-pro-Prt1 N-terminus, is very common in bacterial extracellular metalloproteases (Hase and Finkelstein 1993). The pre-peptide is automatically cleaved off during secretion from the host, while the pro-peptide is removed after secretion, leading to the active protease (Eijsink et al. 2011). Although the E. coli host did not secrete the pre-pro-Prt1, the pre-peptide was removed during the folding of the enzyme as proved by the SDS-PAGE study (Fig. S1: Extract in the Supplementary Material). Although the N-terminal His6 tag in the pre-peptide was lost early in the process, the seven adjacent histidines located in the N-terminus of the pro-Prt1 apparently provided sufficient binding to the Ni-NTA column, and the pro-Prt1 eluted at a relatively low concentration of imidazole (68 mM) (data not shown).

The predicted N-terminal signal pre-peptide in the UniProt database (www.uniprot.org) was about 20 amino acids long, which means that the cleavage site would be A46. However, the N-terminal study revealed that the pre-peptide was cleaved prior to V35, indicating that the UniProt prediction may need adjustment. In the present work, the positive effect of the N-terminal processing was evident from the increase in activity taking place during incubation at 37 °C (Fig. S3 in the Supplementary Material). However, extended incubation of the enzyme also caused degradation that resulted in the loss of activity. It has previously been reported that the N-terminal pre-pro peptide is functioning as a chaperone for correct folding of the protease, before it is excised automatically after expression (Inouye 1991). In order to verify that theory, the gene of the active protease without the pre-pro peptides was designed and ligated back to the pETM11 vector. When expressed, the majority of the truncated protease was insoluble (data not shown).

Based on sequence homology assessment, the Prt1 from P. carotovorum EC14 belongs to the thermolysin-like proteases (Kyostio et al. 1991). Prt1 has six active-site residues, three zinc-binding residues, and five out of seven substrate-binding residues as compared with that of thermolysin (Kyostio et al. 1991). Prt1 has 68.3 and 20 % sequence similarity with Bacillus thermolysin Fbk2 (UniProt G3GBT6 (www.uniprot.org)) and Bacillus thermolysin (Uniprot P00800 (www.uniprot.org)), respectively. The data expand the previous studies on the Prt1 enzyme published by Kyostio et al. (1991) mainly with respect to providing new knowledge about enzyme maturation, thermal stability, substrate specificity, and cleavage pattern. For thermolysin, it has been shown that during maturation, a stretch of amino acids comprising the pro-peptide/N-terminal boundary needs to insert productively into the catalytic center, a process that obviously requires conformational flexibility (Sauter et al. 1998). During maturation, the N-termini of the mature protease need to relocate from a position in the catalytic center to their final position in the mature enzyme (Sauter et al. 1998). Intensive studies have shown that the N-terminal β-sheets as well as the binding site of calcium ions are the most deleterious points of mutations. Mutations taking place in the N-terminal calcium binding site will cause dramatic changes of the thermal stability (Eijsink et al. 2011; Van den Burg et al. 1998). Thermolysin requires one zinc ion and four calcium ions to stabilize its structure. It is also known that thermolysin and thermolysin-like proteases (TLPs) all require calcium ions for proper folding and structure stability (Veltman et al. 1998; Alexander et al. 2001). However, Ca2+ was apparently not required for activation and did not induce activation (Fig. 4b) or stabilize (or destabilize) the thermal robustness of the Prt1 protease (Fig. 3c). The sequence alignment of N-terminus of Prt1 with thermolysin and TLPs showed that Prt1 lacks the N-terminal calcium binding site (D289, D291, and Q293 from thermolysin), which is a significant difference between Prt1 and thermolysin and some TLPs. This difference also explains the lower thermal stability of Prt1, since the N-terminus of Prt1 is not stabilized by calcium and is thus vulnerable to destabilization. Hence, when the Zn2+ ions activate the proteolytic as well as the autolytic activity of Prt1, the unstable N-terminus of Prt1 may very likely undergo autolytic digestion after translocation to the active pocket.

The sequence analysis of the Prt1 indicates that it belongs to the same family as thermolysin (classified as family M4 in the MEROPS peptidase database). According to the MEROPS database, this family of proteases usually has a cleavage preference with valine, leucine, isoleucine, and phenylalanine at P1′. However, our data on the Prt1 metalloprotease were not in accordance with this general rule. With β-casein as a substrate, proline showed the highest frequency in position P1, which may indicate a slightly increased preference for cleaving after proline residues. However, it must be noted that β-casein in itself is particularly rich in proline residues. Proline is also present in position P1 when using ribonuclease A and BSA as substrates, although to a much lesser extent than with β-casein. The cleavage of potato lectin and plant extensin further verified the ability of Prt1 to catalyze the hydrolysis of highly glycosylated proteins, signifying the prospects of using the Prt1 protease for degradation of plant cell wall structural proteins.