Lactobacillus fermentum CGMCC 1.2133 strain has high homology with L. fermentum IFO 3956, of which the genome has already been completely sequenced. To confirm whether any putative NTD had already been reported in this strain, we used NCBI blast Protein and found two putative N-deoxyribosyltransferase homologs in L. fermentum IFO 3956: LAF 0141 (NCBI gi|184154617), which encodes a 158-amino acid hypothetical protein, and LAF 0655 (NCBI gi|184155131), which encodes a 148-amino acid hypothetical protein. Their functions have not yet been revealed experimentally. Alignment of five amino acid sequences including LAF 0141, LAF 0655 and other reported NTDs (Fig. 1) showed that, in addition to the three critical catalytic sites for 2′-deoxyribosyl transfer activity (Armstrong et al., 1996; Anand et al., 2004; Miyamoto et al., 2007), the LAF 0141 gene encodes a substrate binding site that interacts with both purine and pyrimidine bases of 2′-deoxyribonucleosides (Miyamoto et al., 2007). This made LAF 0141 a perfect candidate as an NTD despite the fact that protein sequence identity between LAF 0141 and known NTDs (Kaminski et al., 2008) was only 34%.
Figure 1. Multiple sequence alignment of NTDs from Lactobacillus helveticus (AY064167), Lactobacillus leichmanni (Q9R5V5.3), L. fermentum (Q6YNI5.1), hypothetical protein LAF 0141 (YP_001842957.1), and LAF 0655 (YP_001843471.1). Identical amino acid residues are marked black and gray. Bullet indicate the catalytic sites of the three known enzymes. Arrow show the binding sites to the hydroxyl group binding of C5′ of 2′-deoxyribose. Two asterisks indicate the substrate binding sites: the first is only present in NTDs, rendering LAF 0141 a better candidate than LAF 0655 to test for NTD activity.
Download figure to PowerPoint
Cloning and expression of NTD from L. fermentum
The LAF 0141 homolog from L. fermentum CGMCC 1.2133 was amplified using PCR, cloned, and overexpressed in E. coli BL21. The recombinant plasmid was sequenced and there were no differences at the nucleotide level between LAF 0141 and the homolog. To identify the function of the LAF 0141 homolog gene product, the recombinant protein was purified by a combination of two ion-exchange chromatography steps and further via a gel filtration column (Fig. 2a). Purified recombinant LAF 0141 homolog gene product migrated as an 18-kDa protein on 12.5% SDS-PAGE, which was identical with the theoretic molecular mass of 18.28 kDa (a total of 160 amino acids, with two additional amino acids present at the N-terminus). The concentration of the purified protein was 2.9 mg mL−1.
Figure 2. Purification and characterization of NTD. (a) SDS-PAGE of different purification steps of LAF 0141 homolog. Lane M, molecular weight standards; lane 1, total protein after Resource Q column; lane 2, total protein after Mono Q column; lane 3, 58 μg purified protein after gel filtration on a Superdex 75 column. Arrows on the right sides of the gels mark the positions of the purified protein. (b) HPLC chromatograms showing the enzymatic activity of NTD. Thymidine served as the deoxyribosyl donor and adenine the deoxyribosyl acceptor; deoxyadenosine and thymine are the products of the reaction (bottom). A control consisting of the standard reaction mixture, but without enzyme was performed (top).
Download figure to PowerPoint
The N-deoxyribosyltransferase activity of the purified recombinant protein was determined by reactions between adenine and thymidine under standard conditions. The amount of deoxyribose transferred after 30 min in citrate buffer was 73.3%. The control reaction, which did not contain the enzyme, showed no conversion of the substrate to a product (Fig. 2b). As PTDs can only catalyze deoxyribosyl transfer to and from purines, and the nucleoside phosphorylases require inorganic phosphates for their enzyme reactions, the LAF 0141 homolog gene product should be classified as an NTD.
Subcellular localization of N-deoxyribosyltransferase
Subcellular localization of the NTD was determined using the polyclonal antibodies raised against recombinant NTD. The specificity of the purified antibodies was confirmed using whole cell extract of L. fermentum in Western blotting (Fig. 3a). The bacterial cells were separated into their different compartments, and NTD was detected both in the cytoplasmic fraction and the cell wall/plasma membrane fractions (Fig. 3b). Washing the debris with buffer could exclude possible contamination with cytoplasmic proteins. However, after two washes, NTD signal remains detectable in the washing supernatant indicating that the cell wall/plasma associated NTD might be washed off by the buffer.
Figure 3. Western blot analysis. (a) The specificity of the purified antibodies. Lane 1, SDS-PAGE of the whole cell extract; lane 2, Western blotting of the whole cell extract with anti-NTD. (b) Detection of NTD in the cytoplasmic fraction and the cell wall/plasma membrane fractions. Lane 1, Western blotting of the purified NTD with anti-NTD, positive control; lane 2, cytoplasmic fraction; lane 3, cell wall/plasma membrane fractions after one wash with washing buffer; lane 4, cell wall/plasma membrane fractions after two washes with washing buffer; lane 5, first washing buffer; lane 6, second washing buffer. Lane M, molecular weight standards.
Download figure to PowerPoint
Immunogold labeling of NTD on ultrathin sections of lactobacilli cells was clearly visualized under the electron microscope, whereas background labeling was relatively low (Fig. 4). The electron-transparent granules can be inferred to be PHB (polyhydroxybutyrate) granules (data not shown). The positive signal, visualized as black dots, was localized within or in close proximity to the cell membrane. Unexpectedly, there were a number of gold particles spread over the surface of the cell wall (Fig. 4). According to PSORTb 3.0 analysis of the amino acid sequence of NTD, we found that NTD contains neither established cell wall-anchoring motifs nor signal sequences that could target it into secretory pathways. The immunofluorescence (Fig. 5a) and Western blotting results (Fig. 5b) support the surface association of N-deoxyribosyltransferase. This phenomenon is reminiscent of recent studies of the surface association of anchorless proteins in probiotics. These ‘anchorless’ proteins, including GroEL (Bergonzelli et al., 2006), EF-TU (Granato et al., 2004), glyceraldehyde-3-phosphate dehydrogenase (GAPDH), and enolase (Antikainen et al., 2007b), have been identified on the surface of lactobacilli. These housekeeping proteins do not possess any exporting motifs or surface-anchoring domains. The mechanism by which they cross the cytoplasmic membrane is still unknown. Enolase and GAPDH are essential intracellular glycolytic enzymes. However, the major function of surface GAPDH and enolase is the immobilization of human plasminogen onto the bacterial surface, subsequently enhancing its activation (Hurmalainen et al., 2007). In addition, enolase was found to bind to the extracellular matrix proteins, such as laminin and Collagen I (Antikainen et al., 2007a). They are considered to be anchorless multifunctional proteins or moonlighting proteins (Sanchez et al., 2008).
Figure 4. Immunogold staining of NTD on Lactobacillus fermentum thin sections. NTD was detected both in the cytoplasm and in contact with the cell membranes and cell surface by anti-NTD antibodies and 5-nm colloidal gold-labeled secondary antibody (a–f ). Sections processed without primary antibody did not label (g). The arrows indicate the location of NTD in the cytoplasm, and the arrowheads indicate NTD in contact with cell membranes or around the cell surface. The electron-transparent intracellular PHB granules were indicated by asterisks (a). Scale bar: 100 nm.
Download figure to PowerPoint
Figure 5. Association of NTD with the cell wall of Lactobacillus fermentum. (a) Phase-contrast (A, C, E) and fluorescence (B, D, F) images are shown for indirect immunofluorescence of the cells. Lactobacilli cells were directly fixed (A, B, C, D) or fixed after incubation in Tris–HCl buffer, pH 8.0 (E, F), incubated with anti-NTD immunoglobulin and subsequently with DyLight 594 Goat Anti-Rabbit IgG. Control slide was incubated only with the labeled secondary antibody (C, D). The exposure time was the same (10 s) for all immunofluorescence photographs. (b) Western blotting analysis of the release of NTD from the cell surface. Lane 1, supernatant after incubation of the cells in 100 mM Tris buffer (pH 8.0); lane 2, culture supernatant after 20 h growth. (c) Release of NTD into buffer at pH values from 3.5 to 8.0. Cells from 80 mL of culture medium were incubated for 40 min in 100 mM PBS-citrate buffer at the indicated pH. The resulting supernatant was filtered through a 0.45-μm membrane filter. All the samples were precipitated with trichloroacetic acid and analyzed using Western blotting. Lane M, molecular weight standards.
Download figure to PowerPoint
A few reports have shown that incubation in neutral or alkaline buffer can release enolase and GADPH from the surface of Lactobacilli, so that these extracellular proteins can be detected in the culture medium (Hurmalainen et al., 2007). Our results demonstrated that the NTD could also be released from the L. fermentum surface in Tris–HCl buffer at pH 8.0. Surface-exposed NTD was verified using indirect immunofluorescence (Fig. 5a), showing that the NTD was bound to the cell surface under normal culture conditions, whereas it was released after incubation in 100 mM Tris–HCl buffer at pH 8.0. This result was supported by Western blotting analysis of the supernatant (Fig. 5b). Microscopic examination of the cell suspension did not reveal any obvious cell lysis after 1 h of incubation, neither did we detect DNA in the cell-free supernatant (data not shown). Previous studies have also demonstrated that incubation would not result in the autolysis of Lactobacillus cells (Antikainen et al., 2007b; Hurmalainen et al., 2007). We have also detected NTD in the culture medium (pH value is 5.6 after 20 h culture) of L. fermentum (Fig. 5b). The release of NTD from the cell surface remained detectable after the incubation buffer was changed to 100 mM PBS-citrate buffer with pH values from 3.5 to 8.0 (Fig. 5c). This means that NTD could dissociate from the cell wall at pH values below or above its isoelectric point (the theoretic PI of NTD is 4.6), and indicates that the release of NTD is not just a response to the Tris–HCl buffer environment. This is not consistent with the ‘anchorless’ proteins thus far identified, including enolase and GAPDH, whose dissociation from the outer surface of Lactobacillus crispatus was favored when the pH was above the isoelectric point of these enzymes (Antikainen et al., 2007b). It has been reported that treatment with buffers normally used for cell washing (Tris–HCl or PBS) at pH 7.3 allowed the extraction of 12-fold higher protein concentrations compared with buffers adjusted to pH 4 (Sanchez et al., 2009), suggesting that most of the surface-associated proteins that interact with the cell envelop may depend on electrostatic interactions and thus are sensitive to pH. However, this does not apply to NTD. Further research is necessary to explore in detail the mechanism involved.
To determine whether the NTD activity also presents on the lactobacillus surface, enzymatic assay was carried out using whole lactobacillus cells. However, it is conceivable that lactobacillus cells may take up the highly concentrated substrates efficiently as other bacteria, as there have been many reports concerning nucleoside synthesis using bacteria whole cells (Fernandez-Lucas et al., 2007; Zheng et al., 2008), which implies that a set of membrane transportation system exists to facilitate the substrate import and product export. This uptake occurs very fast due to Nucleoside-specific membrane transporters in lactic acid bacteria (Kilstrup et al., 2005; Martinussen et al., 2010), namely, the conversion of nucleoside may be attributed to cytoplasmic enzyme when a whole cell assay was performed. Thus, considering that the incubation in conventional buffer will strip most of the NTD from the cell surface (Fig. 5a), whole cells after incubation in PBS-citrate buffer for 40 min were used in the same assay as a control. If surface-located NTD retain its biologic activity, washed cells were supposed to exhibit lower catalysis rate at the start point, at which time the activity of intracellular enzymes is yet limited by transportation kinetics. Data presented in Fig. 6 reveal a significantly reduced catalysis activity of washed whole cells after 1 min reaction. However, as the reaction time increases, the activity difference between washed cells and original cells gradually diminishes. This is consistent with our assumption that the uptake kinetics of nucleoside by lactobacillus is fairly fast. In a short period of reaction time, the conversion was mainly catalyzed by surface enzyme, as time elapsed, intracellular enzyme activity became dominant due to the function of membrane-located substrate and product transporters.
Figure 6. NTD activity of whole Lactobacillus fermentum cells. The NTD activity was observed at 254 nm of the lactobacillus whole cells by determining the amount of deoxyribose transferred from thymidine to adenine in the buffer system as described in 'Materials and methods'. After washing with buffer, transformation ratio of the whole cells is lower at first, but progressively increased to a level similar to the non-washed cells. Each column represents the average of three parallel experiments. Error bars indicate standard deviation.
Download figure to PowerPoint
From a physiologic point of view, the presence of NTD at the outside is puzzling, as the role of the enzyme is to balance the deoxynucleotide pools inside the cell (Kilstrup et al., 2005). Although NTD retains its deoxyribose transferase activity when localized on the cell surface, whether NTD do ‘moonlighting’, i.e. acquire another function when surface-associated (Jeffery, 2009), remains unclear. Current research is ongoing in our lab to determine its precise role on the surface of lactobacilli.
In conclusion, the data presented here show that NTD from L. fermentum can be added to a growing list of enzymes that one would expect to see only in the cytoplasm, but which have been detected on the cell surface (Granato et al., 2004). It is not known how these anchorless proteins cross the cytoplasmic membrane. They are thought to bind to the cell surface through non-covalent interactions and, thus, can be extracted by buffers or released into the culture medium. To our knowledge, we are the first to confirm experimentally the localization of an essential deoxynucleoside catabolic enzyme that has dual location both in the cytoplasm and on the surface in L. fermentum. The results reported here may serve as the basis for further work to characterize the surface-associated NTD, identify the specific roles of surface-associated NTDs in nucleoside metabolism or the extracellular environment, and also determine the surface-association mechanisms of the anchorless proteins.