A Legionella pneumophila amylase is essential for intracellular replication in human macrophages and amoebae

Legionella pneumophila invades protozoa with an “accidental” ability to cause pneumonia upon transmission to humans. To support its nutrition during intracellular residence, L. pneumophila relies on host amino acids as the main source of carbon and energy to feed the TCA cycle. Despite the apparent lack of a requirement for glucose for L. pneumophila growth in vitro and intracellularly, the organism contains multiple amylases, which hydrolyze polysaccharides into glucose monomers. Here we describe one predicted putative amylase, LamB, which is uniquely present only in L. pneumophila and L. steigerwaltii among the ~60 species of Legionella. Our data show that LamB has a strong amylase activity, which is abolished upon substitutions of amino acids that are conserved in the catalytic pocket of amylases. Loss of LamB or expression of catalytically-inactive variants of LamB results in a severe growth defect of L. pneumophila in Acanthamoeba polyphaga and human monocytes-derived macrophages. Importantly, the lamB null mutant is severely attenuated in intra-pulmonary proliferation in the mouse model and is defective in dissemination to the liver and spleen. Our data show an essential role for LamB in intracellular replication of L. pneumophila in amoeba and human macrophages and in virulence in vivo.


Results
Identification of amylases in L. pneumophila. Based on domain sequence homology, three new putative amylases were identified in the L. pneumophila genome, in addition to the one described (GamA) amylase ( Fig. S1a-c) 56,60 . The Lpg2528 putative amylase is designated as LamB, which is encoded by a monocistronic gene (Fig. 1d). Considering LamB is only present in L. pneumophila and L. steigerwaltii, of the 60 Legionella species, it is more likely that LamB has been acquired after the speciation event of L. pneumophila and suggests that L. steigerwaltii may have arisen recently from L. pneumophila (Fig. 1a). The evolution of this gene in L. pneumophila mirrors that of the strain evolution (Fig. 1a). L. pneumophila strain Lens is most related in genome sequence homology to strain 130b/AA100, which is seen with lamB (Branch length, 99) 61 . Similarly, L. pneumophila strain Alcoy is most homologous to strain Corby, as also seen with lamB (Branch length, 94) 61 . LamB shares amino acids sequence homology only with other soil and freshwater organisms such as, Methylobacterium and Insolitispirillum (see Supplementary Fig. S2). Thus, it is likely that lamB may have been acquired by HGT from other intra-amoebal or planktonic, environmental organisms. Because L. pneumophila is responsible for 85% of Legionnaire's disease cases, we characterized the role of this enzyme in the intracellular infections of human monocyte-derived macrophages (hMDMs) and A. polyphaga 2,59 . Structure of LamB and its potential secretion. LamB has an α-amylase domain (residues 18-376) that is structurally similar to the crystalized glucosidase of Streptococcus mutants, SmDG (Fig. S1d) 62,63 . The putative catalytic site, which is located within the predicted catalytic pocket of the enzyme, is conversed amongst amylases and in LamB of L. pneumophila and L. steigerwaltii (Figs 1b,c and S1d). Iterative Threading Assembly Refinement (I-TASSER) is a bioinformatics method of predicting the three-dimensional structure of proteins based on fold recognition 64,65 . The server also predicts ligand binding sites, and gene ontology. Structural modeling of LamB using the I-TASSER database shows structural similarity with other crystalized glucosidases (Fig. 1b).
L. pneumophila proteins can access the host cytosol by two major routes, translocation via the T4SS or through secretion into the LCV lumen by the type II secretion system (T2SS) and into the cytosol through the semipermeable LCV membrane 22,66 . L. pneumophila secretes many proteins via the T2SS, which is critical for intracellular growth and pulmonary disease [67][68][69] . Twenty proteins were identified to be secreted by the T2SS of L. pneumophila while an additional 250+ proteins have been suggested to contain a putative T2SS signal 70 . LamB was not identified by any of these methods as potential type-II substrate 70 . In addition, LamB lacks the N-terminal secretion signal characteristic of T2SS substrates 69 .
In order to be translocated by the Dot/Icm translocation system, effectors are recognized by a translocation signal on the C-terminus; alterations at the C-terminus of effectors causes a failure in translocation 57 Machine learning techniques have identified conserved bi-and tri-residues at the C-terminal end of L. pneumophila effector proteins, 10% of known effectors do not contain at any of these motifs and some proteins harboring these motifs are not translocated effectors 57,73 . LamB contains seventeen bi-residues identified to be heavily enriched in the last 100 amino acids of the C-terminus of T4SS effector proteins (see Supplementary Fig. S3) 73 .
To determine whether LamB was Dot/Icm-translocated, the adenylate cyclase (CyaA) reporter function was used 74,75 . Transformation of plasmids expressing reporters CyaA-LamB or the positive control, CyaA-RalF, as fusion proteins was performed in WT L. pneumophila and the translocation-deficient mutant, dotA.  showed that the CyaA-LamB reporter was not translocated by the Dot/Icm translocation system, as there was no significant difference between the secretion of CyaA-LamB by WT L. pneumophila or the dotA mutant (Student t-test, p > 0.2); whereas, the control, CyaA-RalF, was readily translocated into the host cells during infection with WT L. pneumophila but not the dotA mutant (Student t-test, p < 0.01) (Fig. 2).
The amylase activity of LamB. To confirm the putative enzymatic activity of LamB as an amylase, in vitro biochemical activity was determined by using the standard amylase activity colorimetric assay, quantifying the cleavage of ethylidene-pNP-G7 to p-nitrophenol, which can be measured at 405 nm 76,77 . Three highly conserved residues, D193, E227, and D296 were identified in the catalytic pocket of LamB and confirmed with domain alignment to other amylases (Fig. 1b,c). Constructs harboring native LamB or three LamB variants of single amino acid substitutions in the catalytic pocket were expressed in E. coli as GST fusion proteins, controlled by an IPTG inducible promotor. Expression of these proteins was confirmed by western blot (Fig. S4). With IPTG induction, amylase activity was highest for the wild type protein compared to uninduced (Student t-test, p < 0.001) (Fig. 3). The three catalytic mutants showed no amylase activity after inducing with IPTG. These data confirmed that LamB is indeed an amylase and that the identified catalytic pocket is essential for enzymatic activity.  E227A, and D296A were also introduced into the lamB null mutant (Fig. 1c). Infections of Acanthamoeba polyphaga or hMDMs were performed, as we previously described 78 .

Requirement of
The lamB mutant was severely defective for intracellular growth in hMDMs and A. polyphaga (Figs 4 and 5). At 24 h post-infection, there was a significant difference in the replication of the null mutant and the three catalytically inactive mutants compared to WT L. pneumophila (Two-way ANOVA, p < 0.001). The growth defect was partially restored to the mutant by in trans-complementation of the gene, which is likely due to loss of the plasmid. However, complementation of the null mutant with any of the three catalytic variants did not restore any growth to the lamB mutant in A. polyphaga or hMDMs (Figs 4 and 5). This defect is not attributed to a growth defect in vitro, as the lamB mutant grows just as well as the WT strain in broth (see Supplementary Fig. S5). These data show that LamB is necessary for intracellular replication of L. pneumophila in both hMDMs and A. polyphaga. Indeed, it is the amylase activity of LamB that contributes to its essential role in intracellular growth, indicating the requirement for degradation of polysaccharides by L. pneumophila. This is surprising, considering the minimal role of glucose in metabolism of L. pneumophila, and that it is mainly utilized during late stages of growth to synthesize the PHB storage compound.
To determine if generation of host glucose by LamB was necessary for intracellular replication, A. polyphaga was supplemented with exogenous glucose during infection (Fig. S6). Glucose supplementation did not rescue the lamB mutant for its defect in intracellularly replication nor did it alter the growth of the WT strain or the complemented mutant (Fig. S6).

Role of LamB in vivo.
Given that A. polyphaga and hMDMs restrict the lamB mutant, we sought to determine the role of LamB in intrapulmonary growth in the mouse model, in vivo. Intra-trachael infection of A/J mice with WT L. pneumophila, the lamB mutant, or the complemented mutant (lamB/C) was performed, as we described previously 43 . Within 10 days, 50% of the mice infected with WT or the complemented mutant had died. However, 100% of mice infected with the lamB mutant survived for the 10 days of the study (Fig. 6a). Analysis of bacterial burden in the lungs of surviving mice, at 24, 48, and 72 hrs showed that the lamB mutant had significantly decreased numbers of bacteria within the lungs, compared to the WT strain (Student t-test, p < 0.05) (Fig. 6b). The defective phenotype was completely recovered by complementation, indicating minimal loss of plasmid in vivo compared to ex vivo infection (Figs 4, 5). Histopathology on pulmonary biopsies taken at 12 and 24 hrs post-infection, with wild type L. pneumophila showed severe inflammatory infiltrates of mononuclear cells (Fig. 6c,d). In contrast, following challenge with the lamB mutant, minimal inflammatory infiltration into the alveolar, bronchial, or peribronical spaces was observed (Fig. 6c,d).
The lamB mutant was less efficient in disseminating to the liver and spleen compared to the WT strain (Fig. 7a,b). At 48 hrs post-infection, there was a significant decrease in the amount of bacteria in the liver of mice infected with the lamB mutant compared to the WT strain (Student t-test, p < 0.01). Compared to the WT strain, fewer lamB mutants disseminated into the spleen at 48 hrs (Student t-test, p < 0.05) and 72 hrs (Student t-test, p < 0.01) post-infection compared to the WT strain. The reduced dissemination of the lamB mutant was completely restored upon complementation by lamB.

Discussion
A "bipartite" metabolism has been used to describe the nutritional needs and metabolic regulation of L. pneumophila 4,47,50,79 . During early intracellular replication within human macrophages or amoebae, L. pneumophila relies on amino acids to generate carbon and energy from the TCA cycle 37,80 . Once amino acid levels become low, the bacteria undergo growth phase transition, switching from the replicative phase to the transmissive phase [81][82][83][84] . At this point, L. pneumophila increases uptake and utilization of glucose and converts it into the storage compound PHB 46,47 . Experiments with 13 C-glucose demonstrated that glucose is used for de novo synthesis of amino acids and PHB during late stages of infection 39 . Additionally, labeling of glucose demonstrated a carbon flux from glucose to pyruvate via the Enter Doudoroff (ED) pathway but not the Pentose Phosphate Pathway (PPP) 39 . However, addition of excess exogenous glucose does not result in increased growth of the organism during any stage 46 . Therefore, generation of excess glucose in the host, as a source of carbon and energy, through degradation of polysaccharides by LamB is unlikely to support growth. Thus, it is surprising to identify a major role for LamB in intracellular growth, since L. pneumophila mainly utilizes amino acids for growth 85 . We speculate LamB is involved in processes other than nutritional virulence 85,86 . Uptake of glucose is increased by L. pneumophila during post-exponential growth, most notably for the generation of the storage molecule, PHB 39,46,49 , and nutrient importers are important for intracellular growth of L. pneumophila 6,87 . Having large stores of PHB allows the organism to persist outside of the host for extended periods of time 51 .Transcriptomic studies in the human macrophage cell line THP-1 have demonstrated that expression of lamB is highest early in infection (8 hrs) rather than later (14 hrs), opposite of when the organism starts increasing consumption of glucose 46,88 . Faucher et al. also classified this gene as "highly induced in cells" 88 . LamB may be involved in increasing availability of glucose in the host ahead of when the organism prepares to utilize it during the late stages of infection. However, our data excludes that possibility. Increasing the availability of glucose with an amylase could allow L. pneumophila to generate more PHB, promoting long-term survival. It is possible that failure to store sufficient amounts of PHB by the lamB mutant results in an early defect in intracellular growth, due to the lack of a rapid influx of acetyl-CoA from reduced levels of PHB. Alternatively, since amylases are known to act on the glycosylation of proteins, LamB may be acting on the post-translational modification of host proteins to control processes important for replication, independent of nutrition or PHB storage 89 . This could allow the bacterium to evade some aspect of the innate host immune response necessary for successful intracellular replication. Future studies are aimed at identification of target(s) of LamB and how they contribute to infection. Considering LamB is unique to L. pneumophila and its loss causes complete defect in intracellular growth in macrophages and amoebae, and attenuation in vivo, it may contribute to the enhanced virulence of L. pneumophila and its prevalence as a disease-causing species compared to other Legionella species.
Bioinformatical analysis indicates that LamB does not contain a T2SS secretion signal, but it does however contain putative T4SS translocation signals 57,58,70 . However, through the CyaA reporter assay our data show that LamB is not translocated. Previous reports have shown effectors that not translocated as CyaA reporter assay, were actually translocated T4SS effectors 25,58 . This reporter could interfere with the translocation of LamB, like seen with other effectors 25 . Predicted strength of the translocation signal is not a definitive answer to whether a protein is translocated, low-scoring predicted effectors have been shown to be translocated by the Dot/Icm System and high-scoring proteins have been shown to not be translocated 58 . Lifshitz et al. identified LamB to be a high-scoring putative effector and in the same study tested 10 new high-scoring putative effectors, of which three were confirmed to not be translocated by the Dot/Icm system using the CyaA reporter assay, but LamB was not tested 58 . Despite being a high-scoring putative effector, LamB may not be translocated, as observed in our CyaA reporter assay. Loss of lamB does not affect L. pneumophila's ability to grow in vitro, supporting the idea that LamB is likely secreted into the host cytosol or into the lumen of the LCV, but the mechanism remains to be determined.
In summary, we report an amylase essential for intracellular proliferation of L. pneumophila within the two evolutionarily distant hosts, human macrophages and amoebae. Given its uniqueness to L. pneumophila, LamB serves as an interesting enzyme that may contribute to the prevalence and virulence of L. pneumophila compared to other Legionella species.

Materials and Methods
Strains and cell lines. L. pneumophila strain AA100/130b (ATCC BAA-74) and the dotA T4SS-deficient mutant, were grown on Buffered Charcoal Yeast Extract (BCYE) agar, as we previously described 78 . To generate the isogenic mutant lamB (lpg2528), 2 kb flanking DNA on either side of lamB, was amplified using PCR with primers listed in Table S1. The resulting amplicon was cloned into the shuttle vector, pBCSk+, to generate pBCSK + lamBKO. To delete the entire gene of lamB, inverse PCR was employed using the primers listed in Table S1, resulting in pBCSK + lamBKO2. The kanamycin resistance cassette from the Ez-Tn5 transposon was amplified using primers listed in Table S1. The resulting PCR product was subcloned in pBSCK + lamBKO2 between the lamB flanking regions using standard molecular procedures, resulting in pBCSK + lamBKO3. This plasmid was introduced into L. pneumophila AA100/130b via natural transformation, as we previously described 90 . Natural transformants were recovered by plating on BCYE agar supplemented with 50 μg/ml kanamycin. To complement the lamB mutant, PCR was used to amplify the lamB gene and its upstream promotor region, using primers listed in Table S1, and subcloned into pBCSK + , generating pBCSK + lamB/C. Complement mutants of lamB with mutations in the catalytic pocket were made by substituting the amino acid for alanine, to generate pBCSK + lamBD193A, pBCSK + lamBD296A, and pBCSK + lamBE227A, using primers listed in Table S1. These plasmid was introduced into the lamA mutant, via electroporation, as previously described 91 . Complemented lamB mutants were selected on BCYE plates supplemented with 5 μg/ml chloramphenicol, resulting in the complemented strains, lamB/C, lamB/D193A, lamB/D296A, and lamB/E227A.

Intracellular replication.
For infection of cell monolayers, L. pneumophila strains were grown in BYE broth with appropriate antibiotic selection, at 37 °C with shaking, to post-exponential phase (OD 550nm 2.1-2.2).
A. polyphaga was cultured in PYG media at 22 °C, experiments were performed in PY media at 35 °C, as previously described 78 . Glucose supplementation experiments were done in presence of 100 mM glucose in the media. Human monocyte-derived macrophages (hMDMs) were isolated from healthy donors and cultured in RPMI 1640, supplemented with 10% fetal bovine serum, as previously described 78,92 . All methods were approved and carried out in accordance to the University of Louisville Institutional Review Board guidelines and blood donors gave informed consent as required by the University of Louisville Institutional Review Board (IRB # 04.0358). The wild type strain; the isogenic mutants, dotA and lamB; and complements lamB/C, lamB/D193A, lamB/ E227A, and lamB/D296A were grown to post-exponential phase in BYE broth at 37 °C with shaking, prior to infection and used to infect hMDMs and A. polyphaga, as previously described 78,92 . A total of 1 × 10 5 host cells were plated in 96 well plates and infected with L. pneumophila at an MOI of 10. Plates were centrifuged at 200 × g (5 mins), to synchronize infection. After 1 h, cells were treated for 1 h with gentamicin to kill extracellular bacteria, as previously described 78,92 . Over a 24 h time course, host cell were lysed with sterile water (hMDMs) or 0.02% v/v Triton X-100 (A. polyphaga). L. pneumophila CFUs were determined by plating serial dilutions onto BCYE agar.

Bioinformatics analysis of LamB. Protein domain analysis was performed using NCBI's Search for
Conserved Domains (https://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi). Phylogenetic analysis was determined using amino acid sequences of LamB, with the Phyologeny.fr platform. Branch length was determined by aLRT 93,94 . Predicted structures were generated via I-TASSER (https://zhanglab.ccmb.med.umich. edu/I-TASSER/) 60 . Structures generated from I-TASSER were aligned using TM-align to generate a TM-score of structural similarity (https://zhanglab.ccmb.med.umich.edu/TM-align/). Translocation Assay. To assess translocation of LamB by L. pneumophila T4SS, during infection of host cells, an adenylate cyclase fusion 74 was generated using standard biology techniques with primers listed in Table S1. A total of 1 × 10 6 hMDMs were infected with wild type or dotA mutant L. pneumophila harboring plasmids expressing various adenylate cyclase fusions at an MOI of 20 for 1 h, as previously described 74,92 . Following infection, the cell monolayers were lysed and processed to assess cAMP concentration by ELISA using the Direct cAMP ELISA kit (Enzo) according to the manufacturer's protocol and measure with a Synergy H1 microplate reader (BioTek).

Amylase activity.
To determine if LamB is a functional amylase, the lamB gene was cloned into the IPTG-inducible GST-fusion expression vector pGEX-6p-1 (Amersham) and expressed in E. coli BL21 using primers listed in Table S1. Additionally, residues within the predicted catalytic pocket were substituted to alanine using inverse PCR using primers listed in Table S1. E. coli cultures (5 ml) harboring either the empty vector, lamB, or the various catalytic inactive mutants were grown in LB broth at 37 °C with shaking until the OD 600nm reached 0.8. The cultures were spilt and one half was induced with 0.1 mM IPTG for 2.5 h at room temperature. One ml of each culture was pelleted by centrifugation and subjected to lysis with 0.5 ml buffer (0.1% v/v Triton X-100, 150 nM NaCl, 10 mM Tris pH7.5), containing protease inhibitors. Insoluble material was pelleted by centrifugation (16000 × g, 10 min, 4 °C) and the resulting supernatant was retained. Expression of fusion proteins was similar in all cultures (see Supplementary Fig. S3). To measure amylase activity, 25 μl of supernatant was analysed using an Amylase Assay Kit (Sigma), following the manufacturer's instructions. This kit utilizes an artificial substrate, ethylidene-pNP-G7, which when cleaved by an amylase generates a colorimetric product detectable at 405 nm.
Mouse model. For testing the virulence of the lamB mutant, specific pathogen-free, 6-8 weeks old A/J mice were used, as previously described 43,92 . Groups of 3 A/J mice, for each time point, were infected intratracheally with 1 × 10 6 CFUs. At 2, 12, 24, 48, and 72 h after infection mice were humanely sacrificed and lungs, liver, and spleen were harvested and homogenized in sterile saline (5 ml) followed by cell lysis in distilled water. To determine CFUs, serial 10-fold dilutions were plated on BCYE agar and incubated at 37 °C. For histopathology, lungs of infected mice were fixed in 10% neutral formalin and embedded in paraffin. Serial 5 μm sections were cut, SCIentIFIC REPORTS | (2018) 8:6340 | DOI:10.1038/s41598-018-24724-1 stained with haematoxylin and eosin (H&E), for light microscopy analysis. Twenty random high-powered fields (HPFs) were assessed to grade inflammation severity including alveolar and bronchial damage, as well as percentage of parenchyma involved. The histology assessment included the number of the mononuclear cells and percent of parenchyma involved by using modification of double-blind scoring method at a magnification of 40x, as we described previously 95 . The inflammation process was graded normal (score of 0), when there were 0-19 monocular cells infiltrates per HPF with no alveolar and bronchial involvement; mild (score of 1), for 20 to 49 cells per HPF, including mild damage of alveolar and bronchial regions; moderate (score of 2), for 50 to 99 cells per HPF with moderate alveolar and bronchial inflammation; or severe (score of 3), for 100 to 200 mononuclear cells per HPF with severe effacement of alveolar and bronchial regions. The murine lung section was examined in sagittal direction and percent of parenchyma involved was scored as 0 when no area was compromised. The involvement of the parenchyma was scored as 1 when up to 25% of the total area was occupied by inflammatory exudate, or scored as 2 when 26 to 50% of parenchyma area was occupied with inflammatory cells, and 3 if comprised of more than 51% of the total area. The total histology score was calculated as an average of individual criteria scores. Uninfected tissue was used as a baseline score. All the experimental procedures were in accordance with National guidelines and were approved by the Institutional Animal Care and Use committee (IACUC) at Faculty of Medicine, University of Rijeka.
Data availability. All data generated or analysed during this study are included in this published article (and its Supplementary Information files).