|Home | About | Journals | Submit | Contact Us | Français|
Examination of cell-free culture supernatants revealed that Legionella pneumophila strains secrete an endoglucanase activity. L. pneumophila lspF mutants were deficient for this activity, indicating that the endoglucanase is secreted by the bacterium’s type II protein secretion system. Inactivation of celA, encoding a member of the family-5 of glycosyl hydrolases, abolished the endoglucanase activity in L. pneumophila culture supernatants. The cloned celA gene conferred activity upon recombinant Escherichia coli. Thus, CelA is the major secreted endoglucanase of L. pneumophila. Mutants inactivated for celA grew normally in protozoa and macrophage, indicating that CelA is not required for the intracellular phase of L. pneumophila. The CelA endoglucanase is one of at least 25 proteins secreted by the type II system of L. pneumophila and the seventeenth type of enzyme effector associated with this pathway. Only a subset of the other Legionella species tested expressed secreted endoglucanase activity, suggesting that the type II secretion output differs among the different legionellae. Overall, this study represents the first documentation of an endoglucanase (EC 220.127.116.11) being produced by a strain of Legionella.
Legionella pneumophila is a gram-negative bacterium that is ubiquitous in natural and man-made water systems, surviving planktonically, in protozoa, and as a part of multi-organism biofilms (Fields, et al., 2002, Lau & Ashbolt, 2009). Human infection occurs following the inhalation of contaminated water droplets originating from various types of aerosol-generating devices. Within the lung, L. pneumophila infects and grows within macrophages, eventually resulting in a pneumonia known as Legionnaires’ disease (Fields, et al., 2002, Diederen, 2008). The bacterium also undergoes various forms of differentiation as it transitions between intracellular and extracellular niches (Dalebroux, et al., 2009, Morash, et al., 2009). Much of the ecology and pathogenesis of L. pneumophila is mediated by secreted factors (Bruggemann, et al., 2006, Chatfield & Cianciotto, 2007, De Buck, et al., 2007, Shin & Roy, 2008, Allard, et al., 2009, Cianciotto, 2009, Isberg, et al., 2009). For the secretion of enzymes and other proteins, L. pneumophila utilizes type II secretion, type IVA secretion, and type IVB secretion (Cianciotto, 2005, Vincent, et al., 2006, Bandyopadhyay, et al., 2007, Al-Khodor, et al., 2009, Rasis & Segal, 2009), and the genome database suggests the existence of type I and type V secretion (Jacobi & Heuner, 2003, Cazalet, et al., 2004).
Type II protein secretion (T2S) promotes the physiology of many environmental bacteria and the virulence of a variety of human, animal, and plant pathogens (Cianciotto, 2005). In T2S, proteins destined for secretion are first translocated across the inner membrane and into the periplasm by the Sec or Tat pathway and then a pseudopilus acts to push the proteins through an outer membrane (secretin) pore (Johnson, et al., 2006, Buddelmeijer, et al., 2009, Korotkov, et al., 2009). For L. pneumophila, T2S helps the bacterium to grow and survive at low temperature as well as within amoebae, macrophages, and the lung (Hales & Shuman, 1999, Liles, et al., 1999, Rossier, et al., 2004, Cianciotto, 2005, DebRoy, et al., 2006, Rossier, et al., 2008, Söderberg, et al., 2008, Rossier, et al., 2009). Recently, a two-dimensional polyacrylamide gel electrophoresis comparison of wild type and lsp mutant supernatants revealed that the T2S secretome of L. pneumophila includes at least twenty-five proteins (DebRoy, et al., 2006). Among the proteins that were identified was a protein that was predicted to encode endoglucanase activity. L. pneumophila is well known for elaborating many enzymes that degrade protein substrates (Rossier, et al., 2008) but several studies have indicated its ability to also utilize carbohydrates (Weiss, et al., 1980, Tesh, et al., 1983, Bruggemann, et al., 2006, Fonseca, et al., 2008), therefore we sought to investigate this new, predicted activity. Using mutational and cloning analysis, we demonstrate that L. pneumophila secretes, through its T2S system, an endoglucanase that belongs to the family-5 of glycosyl hydrolases.
L. pneumophila strain 130b (ATTC strain BAA-74, also known as AA100 or Wadsworth) served as our wild-type strain (Rossier, et al., 2004). A mutant of 130b containing a kanamycin-resistance (KmR) cassette inserted into lspF (NU275) was previously described (Rossier, et al., 2004). Other strains tested were L. pneumophila Philadelphia-1 (ATCC 33512), Dallas-1E (ATCC 33216), and Concord-3 (ATCC 35096), as well as L. erythra (ATCC 35303), L. feeleii (ATCC 35072), L. longbeachae (ATCC 33462), and L. parisiensis (= ATCC 35299). Legionellae were routinely cultured in buffered yeast extract (BYE) broth or on buffered charcoal yeast extract (BCYE) agar (Rossier, et al., 2004). Growth in broth was assessed by measuring the optical density of the culture at 660 nm. Escherichia coli DH5α and DH5α λ pir (Invitrogen, Carlsbad, CA), hosts for recombinant plasmids, were grown on LB agar (Rossier, et al., 2004). Antibiotics were added to media at the following concentrations (μg per ml): ampicillin, 100; chloramphenicol, 6 for L. pneumophila and 30 for E. coli; gentamicin, 2.5; and kanamycin, 25 for L. pneumophila and 50 for E. coli. Chemicals were purchased from Sigma-Aldrich (St. Louis, MO).
DNA and protein sequences were analyzed using Lasergene (DNASTAR). The Clustal method of Lasergene Megalign was used for protein alignments. Protein homologs were identified in the database using programs based on BLAST (Altschul, et al., 1990). Conserved domains were identified by searching the conserved domain database (Marchler-Bauer, et al., 2007). Signal sequences were identified using SignalP (Nielsen, et al., 1997), and further predictions about the cellular location of proteins were done using PSORTb (Gardy, et al., 2005). DNA was isolated as described previously (Rossier, et al., 2004). Primers for amplifying DNA from strain 130b were designed based on the Philadelphia-1 genome database (Chien, et al., 2004). A 2052-bp fragment containing celA as the only intact ORF was amplified using primers OR138 (5′-GGAATTCGTTAGCTGGCCAC) and OR139 (5′-CCTCTTAATGCTGCTACCTG) and ligated into pGEM-T Easy (Promega, Madison, WI) to yield pGCelA. A1825-bp fragment containing celA was amplified using primers MP22 (5′-TCACTCCATATGGTACCCACTCAA) and OR139 and cloned into pMMB2002 (Rossier, et al., 2004) to yield pMCelA. Allelic exchange was used to make celA mutants. To that end, a 1193-bp fragment containing the 5′ portion of celA was amplified using primers OR138 and MP2 (5′-ATTGACCTGATGCACTGCCT) and ligated into pGEM-T Easy, yielding pGCelA(N). The plasmid was digested with BstEII, which cuts 108-bp after the celA start, and then ligated to a KmR gene to yield pGCelA(N)::Km. Next, the SacI-SphI fragment of pGCelA(N)::Km containing the disrupted gene was cloned into pRE112 (Edwards, et al., 1998) yielding pRECelA(N)::Km. pRECelA(N)::Km was electroporated into 130b, and mutants were selected as before (Rossier, et al., 2004).
To assess secreted endoglucanase activity, we grew legionellae in BYE broth to late-log phase and then assayed cell-free culture supernatants for activity against carboxymethylcellulose (CMC) using the zymogram method (Schwarz, et al., 1987). 15 μl of supernatants were electrophoresed through non-denaturing 12% polyacrylamide gel containing 0.1% CMC. Next, the gel was incubated at 37° for 18 h and then stained with 1 mg/ml Congo Red and destained with 1M NaCl. As confirmed using purified cellulase of Aspergillus niger, yellow clearings against the red background were indicative of a CMC-hydrolysis. In order to assess cloned activity, E. coli lysates, which were obtained as before (DebRoy, et al., 2006, Rossier, et al., 2009), were placed into wells in 1% agarose containing 1% CMC. After 3 d of incubation at 37°C, the plates were stained with 0.5% Congo Red and then destained with 1M NaCl. In order to quantitate the amount of endoglucanase activity in L. pneumophila BYE cultures, 20 μl of 100-fold concentrated supernatants (DebRoy, et al., 2006) were incubated with 1% CMC in water at 37° for 20 hours with shaking, and then the amount of reducing sugars generated was determined using the dinitrosalicylic acid method (Miller, 1959, Percival Zhang, et al., 2006). The amount of reducing sugars in supernatant samples was expressed as glucose equivalents based on a glucose standard curve. To gauge substrate specificity, recombinant E. coli lysates were incubated with CMC, Avicel, or birchwood xylan in distilled water for 20 hours at 37° with shaking, and the release of reducing sugars was measured by the 2, 2′-bicinchroninate method modified (Zhang & Lynd, 2005). To assess cellobioside activity, L. pneumophila supernatants or E. coli lysates were incubated with 0.05 mM 4-methylumbelliferyl-β-D-cellobioside in 0.05 M sodium acetate pH 5.0 at 37° for 20 hours, and then the amount of released 4-methylumbelliferone was determined fluorometrically using excitation and emission wavelengths of 360 nm and 455 nm, respectively (Chernoglazov, et al., 1989, Barr & Holewinski, 2002). The presence of secreted lipase and phosphatase activities in L. pneumophila supernatants was determined as before (Aragon, et al., 2000, Aragon, et al., 2001, Aragon, et al., 2002).
To examine L. pneumophila growth in protozoa, H. vermiformis and A. castellanii were infected as previously described (Rossier, et al., 2008). Thus, 104 CFU were added to wells containing 105 amoebae and then, at various times post-inoculation, the numbers of bacteria per co-culture were determined by plating dilutions on BCYE agar. To assess L. pneumophila growth in macrophages, human U937 cells were infected as previously described (Rossier, et al., 2004). Briefly, monolayers containing 106 macrophages were inoculated with 105 CFU, and then, at various times, the numbers of bacteria in the monolayer were determined by plating on BCYE agar. For infection of A/J mice, 6-8-week-old females (Jackson Lab) were inoculated intratracheally with a 25 μl suspension containing 106 CFU of a 1:1 ratio of wild type and mutant strains (Rossier, et al., 2004, DebRoy, et al., 2006). One and three days later, infected lungs were homogenized, and the numbers of bacteria and the ratio of wild type to mutant were determined by plating. Animal experiments were approved by the Animal Care and Use Committee of Northwestern University.
Previously, we identified a protein that is present in wild-type strain 130b but not T2S mutant supernatants and is annotated in L. pneumophila databases (http://genolist.pasteur.fr/LegioList/) as being similar to endoglucanases (DebRoy, et al., 2006). In strains Philadelphia-1, Lens, Paris, and Corby, its corresponding gene is lpg1918, lpl1882, lpp1893, and lpc1372, respectively (Cazalet, et al., 2004, Chien, et al., 2004, Glockner, et al., 2007). The gene is monocistronic, with an upstream ORF encoding a hypothetical protein, and a downstream ORF encoding 3-deoxy-manno-octulosonate cytidylyltransferase. Compatible with its secretion, the 53-kDa protein had a Sec-dependent signal sequence (DebRoy, et al., 2006). The mature protein sequence contained an Asn-Glu-Pro domain, which, in known endoglucanases, is essential for activity (Baird, et al., 1990). For reasons below, we designated the protein and its gene as CelA and celA. Because Legionella had not been previously investigated for endoglucanases, we assayed supernatants from wild-type strain 130b for activity against the cellulose derivative CMC, as has been done in order to operationally define many endoglucanases (Molhoj, et al., 2002). The supernatant material readily cleaved CMC in a zymogram (Fig. 1A), indicating that L. pneumophila indeed secretes an endoglucanase activity. An lspF mutant defective for T2S expressed less activity in its supernatants than parental 130b did (Fig. 1A). To quantitate the amount of enzymatic activity in the supernatants, we measured the numbers of reducing sugars generated by hydrolysis of CMC using the dinitrosalicylic acid method (Fig. 2A). Once again, the T2S mutant had substantially less activity than its parental wild type. The residual activity seen for the mutant in the two different assays was likely due to cell lysis in the cultures, as we have observed before when monitoring other T2S-dependent activities (Rossier, et al., 2004, DebRoy, et al., 2006, Rossier, et al., 2008, Rossier, et al., 2009). Taken together, these data document that L. pneumophila secretes an endoglucanase that is dependent upon T2S. That both CelA and endoglucanase activity were lacking in lspF mutant supernatants suggested that CelA encodes the activity.
In order to determine if CelA is in fact the secreted endoglucanase, we used allelic exchange to construct a set of celA mutants from strain 130b. Three independent mutants, NU353, NU354, and NU355, were obtained. Similar to other lsp mutants (Rossier, et al., 2004), the celA mutants grew normally at 37° in BYE broth and on BCYE agar (data not shown), indicating that celA is not required for extracellular growth under standard conditions. When cultured in broth, mutant supernatants contained normal levels of acid phosphatase and lipase (data not shown), indicating the strains do not have general defects in T2S (Aragon, et al., 2001, Rossier & Cianciotto, 2001, Aragon, et al., 2002). The celA mutants did not display the altered colony morphology or reduced efficiency of plating at 25-17°C exhibited by lsp mutants (Rossier, et al., 2004, Söderberg, et al., 2004, Söderberg, et al., 2008). To examine secreted endoglucanase activity, we grew the legionellae in BYE broth to a similar stage in the late-log phase and then assayed cell-free supernatants for activity against CMC using both the zymogram and dinitrosalicylic acid methods. In contrast to wild type, the celA mutants completely lacked activity (Fig. 1B and and2B).2B). Since the two mutants tested behaved similarly and since celA is monocistronic, these data indicate that the loss of activity was due to inactivation of celA and not a second-site mutation or polar effect. That celA encodes an endoglucanase was confirmed when lysates from recombinant E. coli expressing celA, whether on a high-copy or low-copy number vector, cleaved CMC, while samples from E. coli containing only vector did not (Fig. 3). In addition, the E. coli expressing celA also cleaved microcrystalline cellulose (Avicel) but not xylan (Fig. 4) or cellobioside (data not shown). Because the celA mutants completely lacked reactivity, CelA is the only endoglucanase active against CMC that is secreted by L. pneumophila strain 130b under standard growth conditions.
On many occasions, we have observed that T2S mutants of strain 130b are impaired for infection of H. vermiformis and A. castellanii amoebae and human U937 cell macrophages (Liles, et al., 1999, Rossier & Cianciotto, 2001, Rossier, et al., 2004, Rossier & Cianciotto, 2005, DebRoy, et al., 2006, Rossier, et al., 2008, Söderberg, et al., 2008, Rossier, et al., 2009). Thus, to assess the role of type II-dependent endoglucanase activity in intracellular infection, we compared strain 130b and the celA mutants for their ability to infect these three, disparate hosts. Over the course of 72 hours, wild type and NU353 showed comparable growth in H. vermiformis (Fig. 5A) and A. castellanii (data not shown). The celA mutant also grew normally in the macrophage line (Fig. 5B). When the other celA mutants were tested, they grew like wild type did in amoebae and macrophages (data not shown). These data indicate that CelA is not required for intracellular infection. To ascertain the in vivo significance of the T2S endoglucanase, we analyzed NU353 for its relative ability to grow and survive in the lungs of A/J mice, using a competition assay that had detected the in vivo growth defects of other T2S mutants (Rossier, et al., 2004, DebRoy, et al., 2006). Because L. pneumophila chitinase mutants were unexpectedly defective in this assay (DebRoy, et al., 2006), we thought it possible that a cellulase, though traditionally viewed, like chitinase, as only being relevant in environmental niches, might be beneficial to L. pneumophila in the lung. However, the celA mutant persisted in the lungs as well as wild type did (data not shown), indicating that celA is not required for lung infection.
The detection of endoglucanase activity in the supernatants of several other wild-type strains of L. pneumophila (Fig. 6A) indicates that secreted endoglucanases are conserved in the L. pneumophila species. Compatible with this, a recent proteomic analysis of culture supernatants from the Philadelphia-1 strain detected a spot corresponding to CelA (Galka, et al., 2008). The Legionella genus includes 51 species besides L. pneumophila (Diederen, 2008). Therefore, we examined a sampling of other legionellae for endoglucanase activity (Fig. 6B). Whereas strains of L. erythra, L. feeleii, and L. parisiensis expressed activity, a strain of L. longbeachae lacked it, indicating that some but not all Legionella species secrete endoglucanase.
We have defined CelA as a type II-secreted endoglucanase of L. pneumophila, based upon its strong activity against CMC and more modest activity against Avicel (Percival Zhang, et al., 2006). BLASTP results indicate that CelA is a member of glycosyl hydrolase (GH) family-5 (Beguin, 1990, Wilson & Irwin, 1999). GH enzymes cleave the glycosidic bond in carbohydrates and are organized into 113 families based on amino acid similarity (Henrissat, et al., 2008). Family-5 is the largest group of GH and is comprised of mainly endoglucanases (EC 18.104.22.168) that cleave β-glucosidic bonds at any point along a cellulose molecule (Beguin, 1990, Wilson & Irwin, 1999). Although many environmental bacteria and fungi produce these endoglucanases (Lynd, et al., 2002), this is the first documentation of such an enzyme being made by a strain of Legionella. CelA showed sequence similarity, with E values ranging from 1 × 10−7 to 7 × 10−4, to endoglucanases of various bacteria, including CelC of Pseudomonas fluorescens, Xft818 of Xylella fastidiosa, and Cel5A of Acidothermus cellulolyticus (Ferreira, et al., 1991, Geelen, et al., 1995, Baker, et al., 2005, Wulff, et al., 2006). The Legionella enzyme displayed comparable levels of similarity to putative endoglucanases predicted from the genomes of Cellvibio japonicus, Clavibacter michiganensis, Reinekea blandensis, Ruminococcus albus, Saccharophagus degradans, and Xanthomonas axonopodis and oryzae (da Silva, et al., 2002, Xu, et al., 2004, Pinhassi, et al., 2007, DeBoy, et al., 2008, Holtsmark, et al., 2008, Salzberg, et al., 2008, Weiner, et al., 2008). Given the relatively modest level of sequence similarity between CelA and other known endoglucanases, it is possible that CelA is a novel Legionella enzyme that may have additional activities. CelA is the sixth example of an endoglucanase being secreted through a T2S system, with the other examples being from Erwinia carotovora, Dickeya dadantii (formerly E. chrysanthemi), Ralstonia solanacearum, X. campestris, and X. oryzae (Hu, et al., 1992, Chapon, et al., 2001, Corbett, et al., 2005, Liu, et al., 2005, Jha, et al., 2007).
As a group, endoglucanases perform a variety of functions, but, because it is secreted, CelA is most likely involved in the catabolism of exogenous polysaccharides. In degrading complex polysaccharides like cellulose, CelA would most probably act in concert with other enzymes, including other secreted glucosidases that are predicted from in silico analysis of the L. pneumophila genome database (DebRoy, et al., 2006, Percival Zhang, et al., 2006). One of the end-products of this sort of catabolism is glucose (Percival Zhang, et al., 2006), and early studies had indicated that L. pneumophila can utilize glucose (Weiss, et al., 1980, Tesh, et al., 1983). Therefore, we investigated the impact of CelA on L. pneumophila growth, but found that the celA mutants were not impaired for replication in broth and agar media, U937 cells, amoebae, or mice. Although these data suggest that CelA is not critical for L. pneumophila growth, it is possible that the protein promotes the growth of L. pneumophila in natural aquatic niches that contain higher levels and/or different types of polysaccharides, including cellulose produced by plants, amoebae, or other bacteria (Barrett & Alexander, 1977, Aksozek, et al., 2002, Linder, et al., 2002, Recouvreux, et al., 2008). Overall, the fact that L. pneumophila secretes an endoglucanase, in addition to a chitinase (DebRoy, et al., 2006), suggests that environmental legionellae survive part of the time as free-living saprophytes vs. being strict intracellular parasites of protozoa. Beyond its role in L. pneumophila ecology, CelA may have some utility in the industrial arena, as is the case for other endoglucanases and cellulases (Wilson & Irwin, 1999, Lynd, et al., 2002).
We can now add endoglucanase activity and CelA to the expanding list of functions that are linked to L. pneumophila T2S; that is, the secretion of more than 25 exoproteins and at least 17 different types of enzymatic activities, with the previously defined activities representing a metalloprotease, chitinase, ribonuclease, lysophospholipase A, cholesterol acyltransferase, phospholipase A, two aminopeptidases, two acid phosphatases, two phospholipases C, three lipases, and a surfactant (Hales & Shuman, 1999, Liles, et al., 1999, Aragon, et al., 2000, Aragon, et al., 2001, Flieger, et al., 2001, Rossier & Cianciotto, 2001, Aragon, et al., 2002, Flieger, et al., 2002, Rossier, et al., 2004, Banerji, et al., 2005, DebRoy, et al., 2006, DebRoy, et al., 2006, Rossier, et al., 2008, Stewart, et al., 2009). Thus, the L. pneumophila system is arguably providing us with the broadest appreciation for the impact of T2S on bacterial physiology, ecology, and pathogenesis (Cianciotto, 2009).
Although the connection between type II secretion, endoglucanase activity, and CelA is currently limited to L. pneumophila, we suspect that it holds for a number of other legionellae based upon the presence of endoglucanase activity in the culture supernatants of most of the other species tested as well as the conservation of lsp genes within the Legionella genus (Rossier, et al., 2004). However, the absence of activity in cultures of L. longbeachae indicates that some species lack expression of CelA or a related endoglucanase. Interestingly, previous examinations found that another Legionella species, L. micdadei, lacked other T2S-dependent activities (Stewart, et al., 2009). Thus, the picture is emerging in which the different species of Legionella, though all possessing the T2S machinery, are not all equivalent in terms of their secretion output. The further elucidation of these differences may over time help to explain differences in the ecology and pathogenicity of the different legionellae.
We thank past and present members of the Cianciotto lab, especially Jenny Dao and Marilyn Wells, for their technical assistance and helpful comments. This work was supported by NIH grant AI43987 awarded to N. P. C.