Per E.J. Saris, Department of Applied Chemistry and Microbiology, Division of Microbiology, PO Box 56, FIN-00014 University of Helsinki, Finland. E-mail: email@example.com
Aims: Immobilization of whole cells can be used to accumulate cells in a bioreactor and thus increase the cell density and potentially productivity, also. Cellulose is an excellent matrix for immobilization purposes because it does not require chemical modifications and is commercially available in many different forms at low price. The aim of this study was to construct a Lactococcus lactis strain capable of immobilizing to a cellulosic matrix.
Methods and Results: In this study, the Usp45 signal sequence fused with the cellulose-binding domain (CBD) (112 amino acids) of XylA enzyme from Cellvibrio japonicus was fused with PrtP or AcmA anchors derived from L. lactis. A successful surface display of L. lactis cells expressing these fusion proteins under the P45 promoter was achieved and detected by whole-cell ELISA. A rapid filter paper assay was developed to study the cellulose-binding capability of these recombinant strains. As a result, an efficient immobilization to filter paper was demonstrated for the L. lactis cells expressing the CBD-fusion protein. The highest immobilization (92%) was measured for the strain expressing the CBD in fusion with the 344 amino acid PrtP anchor.
Conclusions: The result from the binding tests indicated that a new phenotype for L. lactis with cellulose-binding capability was achieved with both PrtP (LPXTG type anchor) and AcmA (LysM type anchor) fusions with CBD.
Significance and Impact of the Study: We demonstrated that an efficient immobilization of recombinant L. lactis cells to cellulosic matrix is possible. This is a step forward in developing efficient immobilization systems for lactococcal strains for industrial-scale fermentations.
The efficient industrial usage of bacteria may require immobilization of the production strain. Immobilization often mimics what naturally occurs when cells grow on surfaces or within natural structures as many micro-organisms’ own ability to adhere to different kinds of surfaces in nature (Kourkoutas et al. 2004). For the industrial usage, the matrix for the immobilization should be cheap, durable, and it should be able to maintain high cell density cultures for long period of time. Cellulose matrix is a good candidate for immobilization matrix, because it is an inexpensive, chemically inert material, which is safe for use even in food or pharmaceutical applications (Linder et al. 1998), and the used carrier material has advantages in the waste treatment, when compared to other typical immobilization matrixes (Sakurai et al. 2000).
In nature, several micro-organisms produce enzymes capable of binding and hydrolysing cellulose. Cellulases are O-glycosyl-hydrolases (GHs) that hydrolyse β-1→4 glucosidic bonds in cellulose. Many cellulases and other GHs show a structural architecture based on one or more catalytic domain(s) and one or more carbohydrate-binding modules(s) (CBM, Hildén and Johansson 2004). The size of the CBMs varies from 4 to 20 kDA, and they are commonly located at one end of the protein sequence (Tomme et al. 1998). Traditionally, cellulose-specific CBMs have been called as cellulose-binding domains (CBD).
Gram-positive bacteria are suitable for whole-cell catalysts and whole-cell adsorbents because of the rigid structure of the cell walls in these micro-organisms; Bacillus and Staphylococcus strains have been used most often (Lee et al. 2003). However, L. lactis, a model lactic acid bacterium with a ‘generally recognized as safe’ status, is becoming an attractive alternative for heterologous protein secretion among Gram-positive bacteria (Morello et al. 2008). Heterologous expression of protein or peptide in the cell surface requires usage of an anchoring system that binds the surface expressed protein to the host cell. Several different mechanisms for anchoring of the proteins to the surface of L. lactis and other lactic acid bacteria (LAB) exist. Leenhouts et al. (1999) grouped these mechanisms under five different categories: transmembrane anchors, lipoprotein anchors, LPXTG type cell wall anchors (covalent binding), AcmA type anchors (later designated as LysM type anchors) and S-layer type anchoring (does not exist in L. lactis). In heterologous surface expression, the most common anchoring strategy is to use different kind of LPXTG type (Fischetti et al. 1990) cell wall anchors. In this case, the anchoring mechanism relies on the sortase activity as this membrane-anchored enzyme cleaves the sorting signal of the target protein at its pentapeptide motif (LPXTG) and promotes covalent anchoring of the target protein to the cell wall (Navarre and Schneewind 1994; Mazmanian et al. 1999). For the heterologous surface expression, the most commonly used LPXTG type cell wall anchors in lactic acid bacteria originate from M6 of Staphylococcus pyogenes, Protein A of Staphylococcus aureus and PrtP of L. lactis (Leenhouts et al. 1999; Lee et al. 2003; Detmer and Glenting 2006; Wells and Mercenier 2008).
Since Buist et al. (1995) identified and cloned the peptidoglycan hydrolase (AcmA) from L. lactis, the usage of the C-terminal repeat region(s) of this protein for noncovalent anchoring of heterologous proteins/peptides in L. lactis has been popular (Buist 1997; Åvall-Jääskeläinen et al. 2003; Steen et al. 2003; Lindholm et al. 2004; Raha et al. 2005; Bosma et al. 2006; van Roosmalen et al. 2006; Okano et al. 2008; Tarahomjoo et al. 2008). Earlier, similar repeated regions were discovered from the bacteriophage Φ29 of Bacillus subtilis by Garvey et al. (1986). Because these repeats were noted to be present in bacterial lysins, Ponting et al. (1999) used the term LysM to denote these repeats. In the AcmA of the L. lactis, there are three distinct LysM domains in the C-terminal part (Buist et al. 1995). So far, the number of the LysM domains used for binding the target protein in the cell surface has varied between 1 and 9 domains in different constructs (Raha et al. 2005; Steen et al. 2005; Okano et al. 2008). One major drawback when using the LysM repeats as anchor domains in hybrid proteins is the weak binding to the cell wall; usually, the most of the secreted fusion proteins have been detected from the culture supernatants (Bosma et al. 2006). Indeed, it has been found that the AcmA anchor of L. lactis is capable of binding to peptidoglycan cell walls of other cells when dosed extracellularly, and this attachment can be further enhanced if cells are washed with acid prior to binding (Steen et al. 2003). These observations indicated that there are other cell wall constituents that prevent binding of proteins containing LysM domains, and anchoring of these proteins is possible only in those locations on the cell surface where these components are absent. It seems that this hindering is attributed to the presence and specific composition of LTAs (lipoteichoic acids) in the bacterial cell wall (Steen et al. 2003, 2008; Buist et al. 2008). Bosma et al. (2006) further developed the strategy to use acid treatment to enhance the external binding of the LysM domain to the cell wall. In their work, Gram-positive enhancer matrix (GEM) particles were developed to be used as substrates for binding of externally added heterologous proteins containing LysM anchor. These GEM particles are nonliving and may thus provide a suitable non-GMO support for the heterologous proteins (Bosma et al. 2006).
The aim of this study was to construct a L. lactis strain capable of immobilizing to cellulose via the CBD of xylanase A (Hall et al. 1989) of Cellvibrio japonicus (previously known as Pseudomonas fluorescens subsp. cellulose, Humphry et al. 2003). The sequence coding the N-terminal CBD of C. japonicus was fused in frame with two different lengths of L. lactis PrtP (Kiwaki et al. 1989) cell wall anchor and L. lactis AcmA anchor. Fusion proteins were expressed under the P45 promoter (Sibakov et al. 1991) and the Usp45 secretion signal (van Asseldonk et al. 1990), and the surface display was confirmed by whole-cell ELISA. For the evaluation of the immobilization capability of these recombinant strains, a simple filter paper-based immobilization assay was developed.
Material and methods
Bacterial strains and growth conditions
All the bacterial strains are described in the Table 1. Escherichia coli strains were grown in Luria-Bertoni media (LB) supplemented with erythromycin 200 μg ml−1 and propagated at 37°C (with 160 rev min−1 shaking for the liquid media). If not otherwise stated, L. lactis strains were grown at 30°C in M17 media (Oxoid, Basingstoke, UK) supplemented with 0·5% w/v glucose (later stated as M17G), and when appropriate, erythromycin and chloramphenicol were added to media to a final concentration of 5·0 and 2·5 μg ml−1, respectively.
P45 promoter and ss45′ preceding the multiple cloning site (MCS)
Transcription termination loop downstream from the MCS
ssusp45 -CBD-spax– construct cloned into pLEB124 as HindIII-BamHI fragment, resulting in expression under P45 promoter
pLEB592 KpnI-BgllI digested and blund end ligated to enable further XbaI clonings
nisP-anchor fragment cloned into pLEB594 as XbaI-ApaI fragment (K. Kylä-Nikkilä unpublished data)
prtP153aa–sequence (anchor) cloned into pLEB594 as XbaI-ApaI fragment.
Constructed for expression of CBD-PrtP 153 aa fusion protein
prtP344aa–sequence (anchor) cloned into pLEB594 as XbaI-ApaI fragment
Constructed for expression of CBD-PrtP 344 aa fusion protein
acmA242aa–sequence (anchor) cloned into pLEB595 as BamHI-ApaI fragment
Constructed for expression of CBD-AcmA 242 aa fusion protein
Negative control for pLEB597 (lacks CBD)
Negative control for pLEB596 (lacks CBD)
Negative control for pLEB606 (lacks CBD)
Relevant new strains generated during this study
L. lactisMG1363 strain carrying pLEB597. Used for whole-cell ELISA. EmR
L. lactisMG1363 strain carrying pLEB596. Used for whole-cell ELISA. EmR
L. lactisMG1363 strain carrying pLEB606. Used for whole-cell ELISA. EmR
L. lactis MG1363acmAΔ1 strain carrying pLEB597. Used for immobilization test. EmR
L. lactis MG1363acmAΔ1 strain carrying pLEB596. Used for immobilization test. EmR
L. lactis MG1363acmAΔ1 strain carrying pLEB606. Used for immobilization test. EmR
L. lactis MG1363acmAΔ1 strain carrying pLEB685. Used for immobilization test. EmR
L. lactis MG1363acmAΔ1 strain carrying pLEB686. Used for immobilization test. EmR
L. lactis MG1363acmAΔ1 strain carrying pLEB687. Used for immobilization test. EmR
DNA isolation and enzymatic manipulations
Plasmid DNA from E. coli TG1 and L. lactis strains was isolated by a standard alkaline lysis method (Sambrook et al. 1989) and the method described by O’Sullivan and Klaenhammer (1993), respectively. Chromosomal DNA was essentially isolated by the method described by Marmur (1961). All enzymatic DNA manipulations were carried out by using established techniques (Sambrook et al. 1989). DNA fragments were isolated from 0·8% agarose gel by established techniques (Sambrook et al. 1989) or by E.Z.N.A.® Gel Extraction Kit (Omega Bio-tek, Norcross, CA). QIAquick PCR Purification Kit (Qiagen, Gaithersburg, MD) and E.Z.N.A.® Cycle-Pure Kit (Omega Bio-tek) were used to purify DNA after PCRs and prior/after certain enzymatic DNA manipulations. Polymerase chain reactions (PCR) were performed with the Mastercycler PCR (Eppendorf, Freiburg, Germany) apparatus using the standard procedures and recommendations given by the manufacturer (Finnzymes, Espoo, Finland) of the DNA polymerase enzyme (F-501L or F-530S).
Construction of the expression plasmid
The pN5 plasmid, containing the sequence coding the CBD of XynA (amino acids 26–137, Hall et al. 1989) of the C. japonicus fused with the sequences coding the Usp45 secretion signal of L. lactis (van Asseldonk et al. 1990) and Protein A anchor (SPA anchor) of Staph. aureus (Shuttleworth et al. 1987; Steidler et al. 1998), was kindly provided for us by Cultor Ltd. The nucleotide fragment coding the Usp45-CBD-SPA anchor fusion was digested out from the pN5 as HindIII and BamHI –fragment. For expression in L. lactis, this fragment was then cloned into pLEB124 (Ra et al. 1996) shuttle plasmid under the constitutive ss45 promoter (P45) and partial secretion signal, resulting in pLEB592. To enable further clonings, the other XbaI restriction site of pLEB592 was removed by double digestion (KpnI and BglII). After ligation, the resulting plasmid was designated as pLEB594.
Construction of the CBD-anchor fusions
Oligonucleotides used for the cloning and structure of the resulting CBD fusions are presented in Table 2 and Fig. 1, respectively.
Table 2. Oligonucleotides and DNA templates for PCR
The lactose-proteinase plasmid pLP763 (Kiwaki et al. 1989) carried by strain MQ421 (kindly provided by Kiwaki) was used as template for amplification of the PrtP anchors by PCR. The plasmid expressing the shorter PrtP-anchored CBD-fusion protein (carrying the 153 C-terminal amino acids of PrtP), designated as CBD-PrtP 153 aa, was constructed by cloning the PCR generated prtP fragment (Table 2) into pLEB594 as XbaI-ApaI fragment (SPA anchor was replaced by the PrtP anchor). The resulting plasmid was designated as pLEB596. Respectively, the plasmid expressing the CBD-PrtP 344 aa fusion protein was made with the same strategy and designated as pLEB597. The shorter CBD-PrtP construct covers AN and W domains and also the C-terminal part from the H domain (domains described as suggested by Siezen 1999). In addition to this, the longer construct covers the whole H domain and C-terminal part of the B domain. In this article, if not otherwise stated, the anchor term refers not only to the actual anchoring region but to the whole PrtP or AcmA-derived protein sequence.
The DNA fragment coding the AcmA anchor (242 C-terminal amino acids) was amplified using the chromosomal DNA of L. lactisMG1363 strain as template. The plasmid expressing the CBD-AcmA 242 aa fusion protein was constructed by replacing the DNA fragment (in pLEB595) encoding previously unpublished NisP anchor with the PCR generated acmA anchor (Table 2) as BamHI-ApaI fragment, resulting in pLEB606. The pLEB595 construct contains exactly the same promoter, signal and CBD region as pLEB594. The BamHI site used for the cloning is located immediately after the XbaI in pLEB595. As a result, all the CBD-anchor fusion proteins constructed in this work contain non-CBD and nonanchor-derived extra arginine, glycine and serine in the CBD-anchor seem.
Construction of the control plasmids
To evaluate the effect of the anchor for the adhesion to cellulosic material, control plasmids without the CBD domain were constructed. The P45 promoter and signal regions were amplified with oligonucleotides NIS265 and NIS266 using the pLEB597 plasmid as a template. The resulting DNA fragment was digested with ClaI (situated upstream of the P45 promoter) and BamHI (included in the NIS266 oligonucleotide) and cloned into pLEB597. In the resulting plasmid, designated as pLEB685, the original fusion construct of promoter-usp45 signals-CBD-prtP344 aa anchor was thus replaced by promoter-usp45 signals- prtP344 aa anchor fusion, also removing the codon coding the above-mentioned extra arginine from the PrtP seem, too. The two other control plasmids, designated as pLEB686 and pLEB687, were constructed by replacing the fragment coding the PrtP 344 aa anchor in pLEB685 with the anchor fragments digested out as BamHI-ApaI fragments from the pLEB596 and pLEB606, respectively.
Electroporation was used to transform E. coli TG1 with the plasmids constructed for expression of the fusion proteins and controls. The plasmids were first isolated and analysed by restriction enzyme digestion prior to electroporation into L. lactis strains by the method described by (Holo and Nes 1989). The CBD-anchor seems were also checked by sequencing. The transformation hosts and the resulting new strains generated in this work are listed in Table 1.
For the whole-cell ELISA test, one colony of bacteria grown on Petri dishes was used to inoculate 5 ml of M17G broth, followed by over-night propagation at 30°C. Then, 40 ml of M17G was inoculated with 800 μl of overnight culture (2% inocula). The bacteria were grown at 30°C until OD600 reached 0·9–1·0, followed by centrifugation of 1 ml of bacterial suspension for 10 min at 15 000 g in room temperature. The resulting pellet was washed with 1 vol of PBS (pH 7·4) and finally resuspended in 15 μl of 15% glycerol. As the growth rate of different strains was not equal, these glycerol-stabilized cell pellets were first freezed in liquid nitrogen and stored at −70°C until used for the actual assay. Later, thawed cells were used for the whole-cell ELISA assay, which was essentially performed as described by Åvall-Jääskeläinen et al. (2002). In our work, CBD antibody (from rabbit, kindly provided by Cultor Ltd.) was used together with anti-rabbit-IgG AP-conjugate (Promega), and detection was visualized by colour formation with Western Blue Stabilized Substrate (Promega).
Cellulose binding assay
To study the immobilization tendency of the recombinant strains to the cellulosic material, an assay utilizing cellulose filter paper was developed. One colony of bacteria picked from the Petri dish was grown over night at 30°C in 5 ml of M17G. Over-night incubation was used to inoculate 15 ml of M17G (3% inocula), and the bacteria were grown at 30°C until OD600 was between 0·35 and 0·55. Then, 10 ml from the culture was centrifuged at 2000 g for 10 min at RT. The resulting pellet was washed with 1 vol of PBST (Tween 0·5% v/v, pH 7·4) and then centrifuged again with the previously mentioned conditions. Washing was repeated, and after this second washing the resulting pellet was resuspended into PBST again. The optical density of this cell suspension was measured to further dilute it with PBST to meet OD600 of 0·35 ± 0·05.
For each test, two parallel Petri dishes (polystyrene, Ø 55 mm) were used: one carrying the filter paper and the other acting as a negative control (without filter paper). To minimize the unspecific binding, Petri dishes were saturated by pipetting 3 ml of cell suspension (OD600 ≈0·35) to both Petri dishes used for the test (using the same cell suspension for both the adhesion dish and for the control dish). After the saturation period of 15 min (100 rev min−1 shaking at 28°C), 1-ml samples were pulled out from both Petri dishes to measure the OD600 of the suspension prior starting the adhesion test with the filter papers. Then, Whatman No.1 filter paper (Ø 42·5 mm) was submerged on the other Petri dish, but the negative control Petri dish remained untouched. Adhesion of cells was carried out by shaking the Petri dishes for 1 h at 28°C with 100 rev min−1. After the adhesion, cell suspensions were completely collected for OD600 measurement by pipette from both of the Petri dishes, and 2 ml of washing solution (PBST) was added to both of the Petri dishes. After the washing period (10 min with 100 rev min−1 shaking at 28°C), liquid from both Petri dishes was completely collected by pipette, and OD600 of the resulting liquid was measured.
After the test, the amount of filter paper bound cells prior to washing was calculated by subtracting the OD600 of the adhesion test Petri dish solution from that of the negative control Petri dish solution. The resulting subtraction was then used to calculate the primary adhesion % by calculating the percentage of the bound cells compared to the average OD600 of both Petri dishes after the saturation phase. The effect of washing was taken into account when the final adhesion % was calculated. The OD600 of washing solution of the adhesion test Petri dish was subtracted from that of the negative control Petri dish, and it was calculated how much reduction had occurred compared to the average OD600 of both Petri dishes after the saturation phase. This resulting negative number (washing loss %) was then subtracted from the primary adhesion % to get the final adhesion %.
Construction of the CBD-anchor fusion plasmids and control plasmids
The aim of this work was to construct L. lactis cells capable of binding to cellulosic matrix. The DNA fragment encoding CBD of xylanase A of C. japonicus was fused with three different lactococcal anchors and expressed under the P45 promoter and Usp45 secretion signal (Fig. 1). Two of the fusion constructs were based on the usage C-terminal part (153 or 344 aa) of the PrtP of L. lactis, and the third construct contained the C-terminal part (242 aa) of the AcmA of L. lactis. The control plasmids contained exactly the same promoter and signal region when compared to those of the CBD-anchor fusions but lacked the nucleotide sequence coding the CBD.
Localization of the CBD-fusion proteins by using whole-cell ELISA
The binding of the L. lactis cells to the cellulose requires the presence of surface-exposed CBD. Therefore, we constructed fusion proteins in which CBD was fused with different C-terminal cell wall anchors. The surface exposure of the CBD in the resulting strains expressing these CBD-fusion proteins was then analysed by whole-cell ELISA with the CBD-specific antibody (Fig. 2). A clear positive reaction was seen with strains LAC242 (CBD-AcmA 242 aa) and LAC248 (CBD-PrtP 344 aa) when compared to that of the negative control (L. lactisMG1363 host strain) (Fig. 2). In the case of LAC247 (CBD-PrtP 153 aa), the colour intensity in the ELISA reaction was only marginally stronger when compared to that of the negative control. The CBD-PrtP 344 aa fusion protein clearly showed the strongest positive reaction among all the strains.
Immobilization of L. lactis strains to the cellulose filter paper
A simple assay utilizing cellulose filter paper discs was developed to study the cellulose immobilization of the recombinant strains expressing the CBD-anchor fusion proteins. First, preliminary tests with filter paper were performed to find out the optimal L. lactis host strain and conditions which gave the lowest background for the test (data not shown). Of the three host strains used in this study, the L. lactis strains MG1363 and L. lactisIL1403htrA proved unfavourable in some test conditions because they showed substantial nonspecific binding to cellulose. As expected by the observations made by Mercier et al. (2002), the acmA mutant strain showed the lowest background for nonspecific adhesion under various test conditions, and it was chosen as the host strain for the actual immobilization tests.
The results from the immobilization test confirmed the whole-cell ELISA observations indicating that the best surface expression was achieved with the CBD-PrtP 344 aa fusion protein. Indeed, this strain performed remarkably well under the test conditions applied, and almost all the cells were immobilized on the filter paper (Fig. 3). The immobilization efficiency of the shorter CBD-PrtP fusion protein (with 153 aa PrtP anchor) was only somewhat higher when compared to that of the negative plasmid control and host control, but the difference was statistically significant (P < 0·05, unpaired double-tailed t-test) when compared to strain expressing the control plasmid. In the case of the CBD-AcmA 242 aa fusion protein, statistically significant (P < 0·05, unpaired double-tailed t-test) immobilization was achieved, but the immobilization efficiency was about 36% lower when compared to the strain expressing the CBD-PrtP 344 aa fusion. The immobilization efficiency of the strains expressing the anchors without the CBD was statistically equal (P > 0·05, unpaired double-tailed t-test) with the host strain.
When the effect of the washing was closer examined, the result confirmed the previously mentioned findings (see Fig. 3). The lowest washing loss was measured for the CBD-PrtP 344 aa fusion strain, indicating a very strong binding of this strain to the filter paper. Also, the washing loss for the CBD-AcmA 242 aa fusion was clearly lower when compared to that of the negative control strain or the anchor control strains. However, the washing loss for CBD-AcmA 242 aa fusion strain was still about five times higher when compared to that measured for the CBD-PrtP 344 aa strain.
In this work, we have studied the usage of CBD of xylanase A from C. japonicus for immobilization of whole cells of L. lactis to cellulose. The cell surface display of CBD was based on usage of cell wall anchor (of LPXTG type) of L. lactis PrtP (last 153 or 344 aa) or the usage of the LysM domains of L. lactis AcmA (last 242 aa).
The estimated minimum length of the extended loop in the anchor (between the target protein and LPXTG box) to allow proper surface display of the target protein is usually at least 90–100 residues (Fischetti et al. 1990; Strauss and Götz 1996). Earlier, studies with the PrtP anchors have indicated that construct as short as 117 residues (total anchor length) has been enough for surface display of antibody fragment or antigen in LAB (Maassen et al. 1999; Krüger et al. 2002). However, when Krüger et al. (2002) increased the anchor length (PrtP of Lacto-bacillus zeae) for surface display of a single chain antibody fragment in Lact. zeae from 117 to 244 aa, the surface visibility was improved. Lindholm et al. (2004) compared PrtP and AcmA anchors (from L. lactis) for displaying the receptor-binding domain of the F18 fimbriae of E. coli on the surface of the L. lactis cells. In that study, the best functionality (adherence of whole cells to porcine intestinal epithelial cells) was achieved by using the combination of long PrtP spacer (516 aa) with the AcmA anchor.
Before the actual immobilization tests in this study, the surface visibility of the CBD in the generated recombinant strains was analysed with whole-cell ELISA. Briefly, a correlation between the anchor length and ELISA reaction intensity (with CBD antigen) was found. The signal for the shortest anchor (PrtP 153 aa) was only marginally stronger when compared to that of the wild-type strain. Only the longer PrtP anchor (344 aa) and the AcmA–anchor (242 aa) clearly indicated proper surface display for the CBD. Based on these results, it seems that the distance between the LPXTG box and CBD in the CBD-PrtP 153 aa construct (120 aa) is too short. Instead, the best surface visibility for the CBD was achieved with the long PrtP 344 aa anchor in which the spacer region is almost 200 aa longer. Similar comparison is not directly valid for the AcmA anchor, as the mechanism for the cell surface attachment for the LysM domains is different.
For the whole cell immobilization test, a method using filter paper as a solid cellulosic matrix was developed. As expected by the ELISA result, the PrtP 153 aa anchor was not long enough to enable efficient functionality for the CBD in the adhesion test. With the 344 aa PrtP anchor, the immobilization efficiency to filter paper was clearly improved. The nature of this binding was very strong as the washing loss measured for these cells was very low (about 1%). The immobilization efficiency of the strain expressing the CBD-PrtP 344 aa fusion protein was about 16 and 40 times higher when compared to that of the parental wild-type strain (MG1363acmAΔ1) or the anchor control (LAC351), respectively. Immobilization of the L. lactis cells expressing the CBD-AcmA 242 aa fusion protein was also efficient, and statistically significant (P < 0·05, unpaired double-tailed t-test) when compared to that of the host strain or the anchor control. However, the binding was not as strong as measured for the strain displaying the CBD-PrtP 344 aa fusion protein. As suggested by the ELISA result, the most probable reason for the weaker binding of the AcmA anchor strain is the lower number of the functionally displayed CBD units in the cell surface. Indeed, Steen et al. (2003) have noticed earlier that the binding of AcmA is not uniform but localized. Furthermore, during the washing step in our work, higher washing loss was measured for the CBD-AcmA strain. Putative explanation for this is the noncovalent nature of the binding of the LysM repeats to the cell wall which may result in increased detachment of the cells from the cellulosic matrix. Earlier Western results have indicated that for both of these fusion proteins, the most intense signal has been detected in the supernatant fraction. However, the proportional signal intensity in the cell wall fraction was much stronger for CBD-PrtP 344 aa fusion protein when compared to that of CBD-AcmA 242 aa fusion protein (results not shown).
Initially, Tween 20 was included into the immobilization test solution to prevent nonspecific binding of the lactococcal cells, but surprisingly, another benefit was gained with the Tween addition, too. After Tween 20 was included into the immobilization solution, the immobilization efficiency of the strain displaying CBD-AcmA 242 aa fusion increased significantly when compared to the control strains (results not shown). Steen et al. (2003) have proposed that LTAs (lipoteichoic acids) might be potential candidates for hindering of the AcmA binding in lactococci, and changes in the LTA composition affect the binding of the AcmA (Steen et al. 2008). Furthermore, Ohta et al. (2000) demonstrated that Triton X-100, another nonionic surfactant, enhanced the release of cell-bound LTA in Staph. aureus cells. It has been demonstrated that the LTAs are not uniformly distributed over the cell surface of L. lactis SK110 but are mainly present at sites where autolysin-anchored fusion protein is not able to bind (Steen et al. 2003). Thus, it can be speculated whether the improved binding with the CBD-AcmA 242 aa fusion protein with Tween 20 in our work was at least partly attributed to the reduced number of cell-bound LTA. Clearly, understanding the mechanism(s) behind the improved binding capability with Tween 20 needs further studies.
As reported by Poquet et al. (2000), HtrA is a unique extracellular housekeeping protease in L. lactis, and it is involved in propetide processing, maturation of native proteins and degradation of recombinant proteins. In our work, despite there was no need to use HtrA-negative host for achieving the proper immobilization, we compared Western detection of the CBD-Prtp 344 aa fusion protein between HtrA-deficient IL1403 host strain and HtrA-positive MG1363 strain (results not shown). Results from this study indicated that expression and/or stability of the CBD-Prtp 344 aa fusion protein was significantly better in the HtrA-deficient IL1403 strain, and the majority of the CBD-fusion protein was not detected in the supernatant but in the whole cell fraction.
In this work, PrtP- and AcmA-derived L. lactis anchors were successfully fused with CBD of XynA of C. japonicus and expressed on the surface of the lactococcal cells. As a result, a new cellulose-binding phenotype for L. lactis was achieved. The whole cell immobilization of these recombinant cells was indicated by a filter paper assay. To our knowledge, this is the first time when CBD-mediated immobilization of recombinant L. lactis cells to the cellulosic matrix has been described.
This work was supported by Tekes, the National Technology Agency (project number 40696), and the Education Fund.