|Home | About | Journals | Submit | Contact Us | Français|
The Mycobacterium avium complex (MAC) is an important group of opportunistic pathogens for birds, cattle, swine, and immunosuppressed humans. Although invasion of epithelial cells lining the intestine is the chief point of entry for these organisms, little is known about the mechanisms by which members of the MAC are taken up by these cells. Studies with M. avium have shown that cytoskeletal rearrangement via activation of the small G-protein Cdc42 is involved and that this activation is regulated in part by the M. avium fadD2 gene. The fadD2 gene indirectly regulates a number of genes upon exposure to HEp-2 cells, including transcriptional regulators, membrane proteins, and secreted proteins. Overexpression of two fadD2-associated regulators (MAV_5138 and MAV_3679) led to increased invasion of HEp-2 cells, as well as altered expression of other genes. The protein product of one of the regulated genes, named CipA, has domains that resemble the PXXP motif of human Piccolo proteins, which bind SH3 domains in proteins involved in the scaffold complex formed during cytoskeletal rearrangement. Although CipA was not detected in the cytoplasm of HEp-2 cells exposed to M. avium, the recombinant protein was shown to be potentially expressed on the surface of Mycobacterium smegmatis incubated with HEp-2 cells and, possibly, to interact with human Cdc42. The interaction was then confirmed by showing that CipA activates Cdc42. These results suggest that members of the M. avium complex have a novel mechanism for activating cytoskeletal rearrangement, prompting uptake by host epithelial cells, and that this mechanism is regulated in part by fadD2, MAV_5138, and MAV_3679.
Environmentally encountered organisms of the Mycobacterium avium complex (MAC) are known for being pathogens of birds and swine, and are a common cause of opportunistic infection. In non-AIDS patients, M. avium can be isolated as the etiologic agent of lung infections; while, in AIDS patients, M. avium enters the intestinal mucosa primarily through the epithelial lining of the small intestine (11). After translocation across the epithelium cell, the bacterium is taken up by submucosal macrophages and disseminates.
Understanding of the invasion of host mucosal epithelial cells by M. avium has been slow due to difficulties in genetic manipulation of these organisms. It is known, however, that other intracellular pathogens impact host signaling pathways triggering cytoskeletal rearrangement as a means to achieve invasion of nonphagocytic epithelial cells. For example, Cryptosporidium parvum has been shown to affect actin polymerization in host cells during invasion (15), while Bordetella (25) and Salmonella (35) can modify small GTPases such as Rac and Cdc42 through activities similar to eukaryotic guanine nucleotide exchange factors and GTPase-activating proteins.
A transposon mutant library of M. avium serovar 109 (MAC109) was screened in our laboratory for clones with impaired ability to enter human laryngeal epithelial (HEp-2) cells. A number of genes were found to be important for invasion of these cells, including the fadD2 gene (10). This gene encodes a fatty acyl coenzyme A synthetase involved in fatty acid degradation. In Salmonella, the fadD gene has been established as a regulator of invasion through hilA expression (22).
Further analysis of the ΔfadD2 mutant strain of MAC strain 109 (MAC109 ΔfadD2) showed that this strain did not activate the Cdc42 pathway, leading to cytoskeletal reorganization (10). Previous studies have shown that Cdc42 activates N-WASp indirectly through phosphorylation and that N-WASp subsequently binds and activates the Arp2/3 complex, leading to actin polymerization (30). A study indicated that invasion by the ΔfadD2 mutant was delayed by at least 15 min and did not result in N-WASp phosphorylation or binding to and activation of the Arp2/3 complex. The ΔfadD2 mutant invasion efficiency could be partially restored by the addition of supernatant from HEp-2 cells infected with the wild-type MAC109 strain (10), suggesting the presence of secreted proteins and secretory systems associated with this mechanism of invasion.
Very little is known about secretory systems and surface proteins of mycobacteria involved in epithelial cell invasion. In Mycobacterium tuberculosis and M. avium subsp. paratuberculosis, a number of secreted or surface proteins have been shown to be involved in macrophage or epithelial cell entry, including the mycobacterial cell entry (Mce) family of proteins (17), the ESAT-6 family of proteins (7), a tyrosine phosphatase (PtpA) (3), and the heparin-binding hemagglutinin protein (HbhA) (29, 33), but the mechanisms by which these proteins function in invasion are unknown. Kitaura et al. (20) found five M. avium proteins that bind fibronectin, including Ag85 and Mpb51. Fibronectin is expressed on the surface of M cells rather than enterocytes, while M. avium preferentially enters enterocytes (31), suggesting that these proteins are not primarily important for epithelial cell invasion. A recent study identifying secreted proteins of M. tuberculosis by proteomic methods indicated that a large portion of the secreted proteins were previously unknown and that almost 40% of the proteins were secreted by a mechanism other than the general secretory pathway (23), indicating there are also likely to be many surface and secreted proteins and systems by which these proteins are secreted by M. avium that are not yet identified.
In the present study, the role of the fadD2 gene in regulation of invasion was further examined, and putative surface or secreted proteins that could be responsible for the effect on the Cdc42 signaling pathway were identified. The results suggest that M. avium invasion of epithelial cells is regulated in part by fadD2 and other downstream transcriptional regulators and that the mechanism of invasion involves the activation of actin polymerization through interaction of a bacterial structure putatively expressed on the surface with the host cell membrane and Cdc42.
Laryngeal cells (HEp-2 cells) were obtained from the American Type Culture Collection (catalog no. CCL-23) and cultured in RPMI 1640 medium (Invitrogen, Carlsbad, CA) supplemented with 10% heat-inactivated fetal bovine serum (Gibco, Grand Island, NY). Cells were cultured in 25- or 75-cm2 flasks (Corning).
M. avium strain 109 (MAC109), serovar 4, and M. avium strain 104 (MAC104), serovar 1, are virulent clinical isolates obtained from the blood of AIDS patients. Mycobacterium smegmatis mc2 155 was a gift from William Jacobs, Jr. (Albert Einstein School of Medicine, New York, NY). All mycobacterial strains were cultured on 7H11 Middlebrook agar or in 7H9 Middlebrook broth (Difco Laboratories, Detroit, MI) with 10% oleic acid, albumin, dextrose, and catalase (OADC; Hardy Diagnostics). The fadD2 transposon mutant (MAC109:ΔfadD2) was generated as described by Dam et al. (10) and was grown on medium supplemented with 400 μg of kanamycin/ml. Other recombinant strains were generated as described below and cultured in broth or medium supplemented with either 400 or 50 μg of kanamycin/ml. For infection, bacteria were grown at 37°C to log phase in broth prior to inoculation. Luria-Bertani broth and agar (Difco) with the indicated antibiotics were used to culture all Escherichia coli strains.
Assays were performed as described previously by Sangari et al. (31). Briefly, MAC109, MAC109:MAV_5138, and MAC109:MAV_3579 (two clones overexpressing the transcription regulator [see Table Table2])2]) were adjusted to 108/ml by McFarland standards and verified by plating serial dilutions. For each strain, 100 μl of this inoculum was added to four wells of a 24-well culture plate containing HEp-2 cells. After a 30-min or 1-h infection period, the supernatant was removed, and the wells were washed three times with Hanks balanced salt solution (HBSS; Invitrogen) to remove extracellular bacteria. Sterile water containing a 1:5 dilution of 0.025% sodium dodecyl sulfate (SDS) was added to the wells to lyse the cells. The lysate was diluted serially and plated onto 7H11 agar to determine the CFU/ml. Assays were performed in replicate, and the resulting CFU from all assays were analyzed compared to the inoculum to determine the percent invasion after 30 min and 1 h for the three strains.
RNA was isolated and purified from mycobacterial strains as follows. Portions (30 ml) of mid-log-phase cultures of the wild type and mutant were each divided into two parts and then centrifuged at 3,000 × g at 4°C. The pellets were washed with HBSS and resuspended in HBSS at a concentration of 3 × 108/ml. HEp-2 cells that were first washed with HBSS were exposed to the wild-type and recombinant MAC109 strains for 15 min at 37°C. Extracellular bacteria were then recovered from the flasks and centrifuged at 3,600 rpm at 4°C, in addition to the bacteria resuspended in HBSS that were not exposed to cells. The pellets were suspended in 1 ml of TRIzol (Invitrogen) and transferred to 2- ml screw-cap tubes with 0.4 ml of glass beads. The samples were shaken three times for 30 s at maximum speed in a bead beater and periodically inverted. RNA was extracted from the aqueous TRIzol solution with chloroform-isoamyl alcohol (24:1) and isopropanol and washed with 75% ethanol. Resuspended RNA was treated with Turbo DNase (Ambion, Austin, TX) for 30 min at 37°C and precipitated with 100% ethanol and 3 M sodium acetate (pH 5.2). The purity and quantity were analyzed by using a spectrophotometer.
In order to gain some leads in the genes influenced by fadD2, we performed a DNA microarray using the M. tuberculosis template (the M. avium array was not available at the time of this study). Each oligonucleotide was 70 mers, and the array hybridization was performed using the Sigma-Genosys Panorama cDNA labeling and hybridization kit (Sigma-Genosys, St. Louis, MO) according to the manufacturer's protocol. The DNA array was repeated twice, and the significance of the fold difference was calculated. Because using the M. tuberculosis array can result in false hybridization or no hybridization if the oligonucleotide sequences are species specific, we used the results only to provide initial guidance.
To confirm the DNA microarray data, HEp-2 cells were exposed to wild-type MAC109 and the MAC109:ΔfadD2 strain. In additional studies, epithelial cells were also exposed to MAC109/pMH6 or MAC109/pMH7 transcription factor overexpression strains (see below). After no exposure or a 15-min exposure to epithelial cells, RNA was extracted from the bacteria as described above. By video microscopy, it was determined that this was the period of time necessary for bacterial internalization to begin (31). cDNA was generated from the RNA by using a SuperScript III First-Strand Synthesis for RT-PCR kit (Invitrogen) according to the manufacturer's instructions. Briefly, 3 μg of each RNA was combined with 50 ng of random hexamers/μl and a 10 mM deoxynucleoside triphosphate mix and then incubated at 65°C for 5 min. 10× RT buffer, 25 mM MgCl2, 0.1 M dithiothreitol, and 40 U of RNaseOUT were added to the primed RNAs, followed by incubation at room temperature for 2 min. Then, 200 U of SuperScript III reverse transcriptase was added to each sample, followed by incubation for 10 min at room temperature, followed in turn by incubation for 50 min at 50°C. The reactions were terminated by 15 min at 70°C and then treated with RNase H for 20 min at 37°C. Selected genes were amplified by real-time PCR in a Bio-Rad iQ iCycler, using Sybr green (Bio-Rad, Hercules, CA) and the primers listed in Table Table1.1. The fold change in gene expression after exposure to HEp-2 cells was determined by the following formulae:
where “exp” refers to RNA samples from bacteria after exposure to HEp-2 cells, “cont” refers RNA samples from bacteria in HBSS, and CT is the threshold cycle.
The transcription regulators MAV_5138 and MAV_3679 were chosen for additional studies because of the upregulation (2.5- and 3.6-fold, respectively) and the fact that homologous regulators have been associated with invasion of other bacteria. Strains of MAC109 overexpressing the MAV_5138 and MAV_3679 transcriptional regulator genes were constructed as follows. Using the primers listed in Table Table2,2, the gene sequences were amplified from MAC109 genomic DNA, and the amplicons were ligated into pLDG13, a Mycobacterium/E. coli shuttle vector containing the strong G13 constitutive promoter from Mycobacterium marinum (12), at the HindIII and EcoRI restriction sites. After screening for insertion in E. coli, the recombinant plasmids, pLDG13:MAV_5138 and pLDG13:MAV_3679 (named pMH7 and pMH6, respectively), were transformed to competent MAC109, resulting in MAC109/pMH6 and MAC109/pMH7.
Inactivation of the fadD2 gene resulted in the inability to activate Cdc42 when the bacterium interacted with epithelial cells. We then hypothesized that a M. avium protein might bind to Cdc42 and activate the host cell protein. In an attempt to identify the M. avium protein, we performed a two-hybrid system using the Cdc42 as bait and an M. avium genomic library. The primers indicated in Table Table22 were used to amplify human cdc42 placental isoform from the recombinant pcDNA3.1 plasmid provided by the Guthrie Research Institute (Sayre, PA). This fragment was then cloned into the pBT plasmid (Stratagene, La Jolla, CA) in frame with the λcI repressor gene at the EcoRI restriction site. Transformants were plated on LB agar containing 34.5 μg of chloramphenicol/ml and screened by digestion and PCR. MAC104 genomic DNA was partially digested with Sau3A and cloned into the pTRG plasmid (Stratagene) at the BamHI restriction site. Transformants were plated on LB agar containing 12.5 μg of tetracycline/ml. Seventy thousand colonies were selected from 10 ligations plated onto 100 plates and combined to create an M. avium library downstream of the RNAP-α gene fragment (pTRG:MAClib). The M. avium MAV_4671 gene (termed cipA for Cdc42 interacting protein) was amplified from MAC104 genomic DNA using the primers listed in Table Table2.2. The PCR product was cloned into the pBT plasmid at the EcoRI restriction site. The human cdc42 placental isoform gene product was amplified as indicated above and cloned into the pTRG plasmid at the EcoRI restriction site.
The recombinant pBT:cdc42 and pTRG:MAClib were cotransformed into the Bacteriomatch two-hybrid system (Stratagene) reporter strain E. coli. Transformants were plated onto LB agar containing 400 μg of carbenicillin/ml, 12.5 μg of tetracycline/ml, 34.5 μg of chloramphenicol/ml, and 50 μg of kanamycin/ml. Colonies that grew after 30 h at 30°C were picked from these plates and transferred to LB agar containing the same antibiotics except carbenicillin and 80 μg of X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside)/ml and 200 μM phenylethyl β-d-thiogalactoside (an X-Gal inhibitor), prepared in dimethyl formamide. Colonies that became blue in color in the presence of X-Gal and the inhibitor after 17 h at 30°C were further analyzed to determine the M. avium sequence of the insert in the pTRG vector. Cotransformation of the pBT:cipA vector with pTRG:cdc42 and of the pBT:cipA vector with pTRG:MAClib and screening of the resulting transformants was repeated as described above.
To determine whether CipA is delivered to the intracellular environment prior to bacterial invasion, we constructed a CipA:CyaA fusion protein, using primers listed in Table Table22 to amplify the DNA encoding amino acids 2 to 400 of the Bordetella pertussis cyaA gene from pACYA, kindly provided by Gregory V. Plano (University of Miami, Miami, FL). The cyaA PCR product was ligated into pLDG13 at the PstI restriction site and screened in E. coli, resulting in pMH4. The cipA PCR product was amplified from MAC104 using the primers listed in Table Table2.2. The pMH4 and the CipA PCR product were digested with EcoRI, ligated, and screened in E. coli, resulting in pMH5, a plasmid containing an in-frame fusion of cipA and cyaA behind the strong G13 promoter. The completed pMH5 construct was transformed by electroporation to MAC104 competent cells, and the resulting colonies were screened by PCR to confirm the presence of the plasmid. The resulting transformants were called MAC104/pMH5.
MAC104 and MAC104/pMH5 were grown to log phase in 7H9 medium without OADC. After centrifugation, the pellets were resuspended in HBSS, and the resulting inocula were used to infect HEp-2 cells in six-well tissue culture plates or 75-cm2 flasks at a multiplicity of infection of 100:1. All steps postinfection were completed at 4°C or on ice. At 30-min, 1-h, and 2-h time points, the extracellular bacteria were removed from the wells and pelleted by centrifugation. The infected HEp-2 cells were lysed in water containing a protease inhibitor cocktail (Sigma, St. Louis, MO) and then centrifuged to remove the cellular debris and intracellular bacteria. The contents of the HEp-2 cells after lysis and centrifugation were incubated with mouse monoclonal α-CyaA (Santa Cruz Biotech, Santa Cruz, CA) and agarose conjugate α-immunoglobulin G (α-IgG) beads (Santa Cruz Biotech) overnight at 4°C. The beads were washed four times with phosphate-buffered saline and resuspended in Laemmli sample loading buffer (Bio-Rad). After denaturation by boiling for 5 min, the protein samples were run on a 12% Tris-HCl protein gel (Bio-Rad) for 1 h at 150 V. The proteins were transferred to a nitrocellulose membrane by using a semidry transfer apparatus with a constant current of 15 V for 1 h.
Western blotting was performed by using an Odyssey Western blotting system (Li-Cor, Lincoln, NE) according to the manufacturer's instructions. Briefly, the membrane was blocked in a 1:1 dilution of Tris-buffered saline (TBS) and Odyssey blocking buffer overnight at 4°C. The membrane was then incubated with mouse α-cyaA antibody (1:300 in TBS containing 0.1% Tween 20) (Santa Cruz Biotech) for 3 h. After washes with TBS-0.1% Tween 20, the membrane was incubated with anti-mouse IgG linked to Alexa Fluor 680 (1:2,000 in TBS containing 0.1% Tween and 0.01% SDS) (Li-Cor). After additional washes the membrane was scanned by using an Odyssey infrared imager (Li-Cor).
Cyclic AMP (cAMP) was assayed in uninfected HEp-2 cells and at 30 min, 1 h, 2 h, and 4 h after contact between the bacteria and HEp-2 cells using a direct cAMP enzyme immunoassay kit (Sigma) according to the manufacturer's instructions. Briefly, MAC104 and MAC104/pMH5 were used to infect the cells at a multiplicity of infection of 100:1. At each time point, extracellular bacteria were removed from the wells, and the infected cells were washed with HBSS and then lysed in 0.1 M HCl. After centrifugation of the lysate at 600 × g, the supernatants were acetylated and neutralized in a 96-well plate. Each sample was then incubated with a cAMP conjugate and cAMP antibody for 2 h at room temperature and then washed. The plate was read at 405 nm after incubation with the p-NPP substrate for 1 h and addition of the stop solution. The levels of cAMP in the samples were calculated based on the standard curve generated from serial dilutions of known cAMP concentrations.
To investigate the possibility that CipA is a putative outer cell wall protein, we constructed an M. smegmatis strain expressing CipA (M. smegmatis genome does not contain a CipA homologue). A promoterless GFPmut2 gene (obtained from Rafael Valdivia and Stanley Falkow, Stanford University) was inserted in the HindIII and EcoRI sites of the pMV261 vector containing the Hsp60 promoter, resulting in pMV261:GFP (26). cipA was amplified, purified, and inserted between the promoter and the GFPmut2 gene at the BamHI and EcoRI sites, in frame with green fluorescent protein (GFP). Colonies were screened in E. coli, and the resulting pMV261:CipA:GFP was transformed to competent M. smegmatis mc2155. PCR was used to screen M. smegmatis expressing GFP for the presence of the CipA:GFP sequence.
HEp-2 cells were infected with wild-type M. smegmatis or bacterium containing pMV261:CipA:GFP plasmid overexpressing CipA protein. After 15 min of infection, cells were washed with TBS and lysed with lysis-binding-wash buffer containing protease inhibitor cocktail (Sigma) as previously described (1). For a positive control, uninfected cell lysates were treated with GTPγS to activate Cdc42 pathway. Activated Cdc42 were captured by using an EZ-Detect Cdc42 activation kit according to the manufacturer's instructions (Pierce, Rockford, IL). Eluted proteins were resolved on an SDS-12% polyacrylamide electrophoresis gel, transferred to a membrane, and blocked overnight with blocking buffer (Li-Cor) in TBS. The proteins were probed with anti-Cdc42 antibody (Pierce) and visualized with goat anti-mouse secondary antibody (Li-Cor). Membranes were scanned by using an Odyssey imager (Li-Cor).
M. smegmatis strains containing either the pMV261:GFP or pMV261:CipA:GFP vector, expressing GFP or GFP fused to the 3′ end of the CipA protein, and M. avium stained with fluorescein were resuspended in RPMI at 107/ml, estimated by McFarland's standards. A total of 100 μl of this inoculum was added to a monolayer of HEp-2 cells in each chamber of an eight-chamber slide with 100 μl of fresh RPMI or 100 μl of fresh RPMI containing 10 μl of a 10-mg/ml solution of cytochalasin D. After a 15-min, a 30-min, or a 1-h incubation period, the medium was removed, and the wells were incubated with a 4% paraformaldehyde solution for 1 h at room temperature. After being washed, the monolayers were examined by fluorescence microscopy or further prepared for immunohistochemistry as described below.
Cells were permeabilized with a 0.1% Triton X-100-0.1% sodium citrate solution for 2 min on ice. After being blocked overnight in 10% bovine serum albumin (BSA), the wells were incubated with rabbit α-Cdc42 (Santa Cruz Biotech), diluted 1:1,000 in 10% BSA for 1 h at room temperature, and then incubated with anti-rabbit IgG-Texas Red conjugate (Santa Cruz Biotech), diluted 1:200 in 10% BSA for 1 h. Monolayers were examined by fluorescence microscopy.
The M. avium 104 genome sequence is posted on The Institute for Genomic Research website (www.tigr.org). Sequences obtained from the microarray and positive pTRG vectors in the bacterial two-hybrid screen were analyzed by using the basic local alignment search tool (BLAST) to find similarity to the published MAC104 genome and for putative conserved domains. Once M. avium gene sequences were obtained, protein-specific iterated BLAST (PSI-BLAST) and the SignalP 3.0 server (5) were utilized to further characterize the gene sequences.
Statistical analysis was based on the mean of three experiments ± the standard deviation. The Student t test was used to compare values for each strain. P values of <0.05 were considered significant.
To explore whether FadD2 was playing a potential role in gene regulation upon epithelial cell invasion, a heterologous microarray was performed by hybridizing M. avium RNA from wild-type and the ΔfadD2 mutant strains against a Panaroma M. tuberculosis microarray obtained from Sigma-Genosys (data not shown). The microarray was strictly to gain leads and, among the genes upregulated in the array, several encoded for proteins of unknown function.
Using the preliminary results of the microarray and also from the preliminary bacterial two-hybrid screen described later, we performed real-time PCR to analyze the expression of a subset of genes appearing in one or both of the exploratory experiments. Five genes were selected with varied responses to HEp-2 exposure based on the results of the microarray: MAV_5138, MAV_3679, and cipA were all upregulated in the wild type (at least twofold increases over control) but not in the mutant upon exposure to HEp-2 cells; MAV_4139 and MAV_1190 showed no upregulation in either the wild type or the mutant. CipA, MAV_4139, and MAV_1190 also putatively interacted with human Cdc42, based on a two-hybrid screen. By real-time PCR analysis, MAV_5138, MAV_3679, and MAV_1190 were shown to be upregulated in the wild-type, but not the mutant, after exposure to HEp-2 cells (Fig. (Fig.1,1, P < 0.05). The cipA gene did not increase in expression upon exposure but had higher expression compared to the fadD2 mutant under the same conditions (although this difference was not significant) (data not shown). cipA and MAV_4139 were not upregulated in the wild-type or mutant strains after exposure to HEp-2 cells.
MAV_5138 and MAV_3679 are homologous to transcriptional regulator families found in other species that have been shown to play a role in invasion. Based on their upregulation in the wild-type M. avium, but not in the fadD2 mutant strain upon exposure to HEp-2 cells, MAV_5138 and MAV_3679 were overexpressed in M. avium behind the G13 promoter. Strains overexpressing either transcriptional regulator had an increased percent invasion of HEp-2 cells compared to MAC109 (Fig. (Fig.2).2). The MAC109:MAV_5138 strain had significantly higher invasion after a 30-min incubation, while the MAC109:MAV_3679 strain had significantly higher invasion after both a 30-min and a 1-h incubation (P < 0.05).
Because FadD2 is likely not a direct regulator of downstream gene expression, we hypothesized that the MAV_5138 and MAV_3679 gene products may play some role in regulating the genes involved in cytoskeletal rearrangement seen upon epithelial cell invasion by wild-type M. avium, but not in the ΔfadD2 strain. We selected five genes from the preliminary microarray and two-hybrid results, and analyzed their expression in strains of M. avium overexpressing either of the putative transcriptional regulators. Overexpression of the MAV_3679 repressor regulator led to upregulation of all 7 genes in broth-grown bacteria (Fig. (Fig.3A).3A). Similarly, all 7 genes were upregulated in broth-grown bacteria in the MAC109:MAV_5138 strain compared to the wild-type (Fig. (Fig.3B3B).
To connect the bacterial genes being expressed upon invasion of HEp-2 cells with the subsequent Cdc42 activation and cytoskeletal rearrangement in host cells, we wanted to identify bacterial proteins interacting with Cdc42. In an exploratory two-hybrid screen, we cotransformed pBT:cdc42 and pTRG:MAClib to the reporter strain E. coli and obtained three proteins that potentially interact with Cdc42: CipA, MAV_1190, and MAV_4139 (Table (Table3).3). As described previously, real-time PCR indicated that the MAV_1190 transcript is upregulated with exposure to HEp-2 cells in wild-type bacteria but not in the ΔfadD2 strain. The MAV_4139 and cipA genes showed no difference in regulation between the wild-type and ΔfadD2 strain. All three genes, however, were upregulated when MAV_5138 and MAV_3679 were overexpressed.
Although cipA transcripts were not affected by deletion in the fadD2 gene, the increase in expression when MAV_5138 and MAV_3679 were overexpressed and its amino acid sequence led us to further explore the role of this protein in epithelial cell invasion. The sequence of CipA has a number of interesting characteristics that lend support to this idea. Prior to the recent annotation of M. avium, an amino acid sequence corresponding to the cipA gene was listed as a putative multicopper oxidase in the NCBI database. The coding sequence listed in this entry is shorter than annotated homologues in other mycobacterial species, and there is no sequence or biochemical evidence to support this putative function. The current annotation lists this protein as a hypothetical protein. There are very few homologues to CipA, and their predicted functions are diverse, including the following: a putative membrane protein in M. tuberculosis (Rv0479c) and other sequenced mycobacteria; a hypothetical protein in M. avium subsp. paratuberculosis (MAP3972c); and putative secreted proteins in Mycobacterium leprae, Corynebacterium diphtheriae, Rhodococcus sp. strain RHA-1, and Burkholderia cepacia. According to the database, the first 100 amino acids are not found in any bacteria besides other mycobacterial species. A PSI-BLAST analysis of this sequence reveals that there are segments in the first 100 amino acids similar to the PXXP domains of the Piccolo domain proteins in human cells, which have been implicated to play a role in the scaffolding proteins involved in actin dynamics (16). These domains are present in only M. avium and M. avium subsp. paratuberculosis, and not other characterized mycobacteria, such as M. tuberculosis. The remaining 252 amino acids correspond to putative secreted proteins, and this region also contains a putative signal peptide, predicted by SignalP 3.0 (5). The region of this gene sequenced from the pTRG plasmid corresponds to the C-terminal 50 amino acids (Fig. (Fig.44).
If CipA interacts with Cdc42, it must be secreted into host cells or inserted into the membrane of these cells. To determine whether CipA is a secreted protein, it was fused to amino acids 2 to 400 of the B. pertussis adenylate cyclase protein. Using antibodies to the B. pertussis cyaA tag, expression of the CipA:CyaA fusion protein was detected in M. avium by Coomassie blue staining of whole-cell lysates (Fig. (Fig.5A).5A). After HEp-2 cells were incubated with the MAC104/pMH5 strain, the fusion protein could not be detected in the cytoplasm or insoluble fraction of the lysed HEp-2 cells by Western blotting (Fig. (Fig.5B).5B). When in the cytoplasm of eukaryotic cells, the fragment of the adenylate cyclase protein leads to an increase in cAMP that can be assayed by enzyme-linked immunosorbent assay. After incubation with MAC104/pMH5 for 1, 2, or 3 h, there was no significant increase in cAMP levels compared to the wild-type MAC104 (data not shown). There was an initial increase at 2 h, but it was not significant, and the levels of cAMP in the cells incubated with the wild-type increased to nearly the same levels after 3 h.
Because we could not identify CipA in the cytoplasm of host cells, we explored whether this protein is potentially expressed on the surface of the bacterium. A strain of M. smegmatis expressing CipA fused to GFP was used to infect HEp-2 cells. Construction of recombinant strains of M. avium expressing this fluorescent fusion protein was unsuccessful. After an incubation time of 1 h with the recombinant strains of M. smegmatis, a structure could be observed on the tip of bacteria near HEp-2 cells (Fig. (Fig.6A).6A). The HEp-2 cells had been treated with cytochalasin D to prevent the uptake of the bacteria. This structure was not present on wild-type M. smegmatis expressing GFP alone (Fig. (Fig.6B6B).
To determine whether Cdc42 was accumulating near the invading bacterium, immunohistochemistry was performed after infection of HEp-2 cells with the recombinant M. smegmatis strains, as well as fluorescein-labeled M. avium. After fixation, infected cells were incubated with antibodies against Cdc42 at 15 min, 30 min, and 1 h after infection. In cells infected with fluorescein-labeled M. avium, Cdc42 was observed to accumulate near mycobacteria at 15 and 30 min postinfection (Fig. 6C and D) but not in cells infected for 1 h with M. smegmatis control (Fig. (Fig.6E6E).
Pull-down assay for activated Cdc42 protein followed by a Western blot analysis with an anti-Cdc42 revealed that HEp-2 cells infected with wild-type M. smegmatis failed activation of Cdc42 at 15 min after infection. However, the bacterium expressing the CipA protein activates Cdc42 at the same time point of infection. Cdc42 is not activated in uninfected HEp-2 cells (Fig. (Fig.77).
If CipA is on the surface of the bacteria upon contact with epithelial cell, it may be part of a secretory complex, or interact a chaperone in the membrane. To identify putative M. avium proteins interacting with CipA, the M. avium protein library in the pTRG vector was screened with the CipA protein in the pBT vector, by cotransformation in the two-hybrid reporter strain. In five separate cotransformations, 87 colonies grew on plates containing the selective antibiotic. Of these colonies, 23 were further analyzed after becoming blue in the second screen. Sequences retrieved from the pTRG vector indicated that 7 proteins putatively interact with CipA (Table (Table3),3), including MAV_3034, an oxidoreductase, and MAV_1300, a hypothetical protein with some loose similarity to intracellular transport proteins (Fig. (Fig.88).
Similar to the initial two-hybrid screen performed, this screen was exploratory, and further biochemical analyses should be done to determine whether or not the interaction exists between Cdc42 and these proteins.
M. avium is similar to other intracellular pathogens, such as Salmonella sp., Yersinia sp., and Shigella sp., in its ability to enter the cells of the intestinal epithelium by an active process involving cytoskeletal reorganization. There is little known, however, about the molecular mechanism of this process in M. avium. Unlike these other intracellular pathogens, M. avium does not have genes encoding proteins similar to surface invasins or internalins, as in the case of Yersinia (18, 19) or Listeria (9), or those that comprise type III secretion systems of Salmonella (36) or Shigella (32). These secretion systems are commonly involved in the secretion of proteins affecting cytoskeletal rearrangement and subsequent epithelial cell invasion. It is plausible, however, to hypothesize that M. avium uses mechanisms for which the result is similar.
There is little doubt that, regardless of the pathogen, the genes involved in epithelial cell invasion are regulated in response to environmental cues such as pH, oxygen, and osmolarity. A study by Bermudez et al. (6) showed that exposure to low O2 tension and hyperosmolarity led to a significant increase of invasion of epithelial cells by M. avium. In Salmonella, Lucas et al. (22) identified a number of genes that work independently to activate hilA, which in turn activates genes in the pathogenicity island SP-1, encoding the type III secretion system (TTSS) involved in epithelial cell invasion. The regulators identified by this group included two-component response regulators similar to those involved in mycobacterial invasion of macrophages, as well as fadD. The FadD protein is involved in the breakdown of endogenous and long-chain fatty acids, although the mechanism of regulation by this protein is yet unknown (22).
In two separate screens of mutant libraries of M. avium, mutations in the fadD2 (10) and fadE20 (26) genes were associated with reduced invasion of epithelial cells. Further studies with the fadD2 mutant strain revealed that it was deficient in the ability to activate the host cell Cdc42 signaling pathway, leading to actin polymerization via N-Wasp and the Arp2/3 complex (10). This signaling pathway, from fadD in the bacterium to actin polymerization in the host, is involved in cell entry by other intracellular pathogens of both plants and animals (4, 22, 34). We hypothesized that the M. avium fadD2 gene may also be involved in the regulation of genes that affect epithelial cell invasion and conducted experiments to determine the role of fadD2 in this process.
In an exploratory heterologous microarray and real-time PCR comparing the upregulation of genes in a wild-type M. avium strain upon exposure to HEp-2 cells to the fadD2 mutant strain, we identified genes possibly regulated by fadD2. A number of homologues to transcriptional regulators were also identified, including MAV_3679 and MAV_5138. Real-time PCR confirmed that MAV_3679 and MAV_5138 are regulated by fadD2 upon invasion of epithelial cells, since they were not upregulated in the fadD2 mutant (Fig. (Fig.1).1). MAV_3679 encodes an ion-dependent regulator, similar to sirR from Staphylococcus epidermidis, dtxR from Corynebacterium diptheriae, mntR from Staphylococcus aureus, and both ideR and Rv2788 from M. tuberculosis. Many of these ion-dependent regulators, which bind the operators of and influence ABC transporters, have been shown to be involved in virulence (2, 24). Interestingly, constitutive expression of an ion-independent mutant of dtxR in both S. aureus and M. tuberculosis exhibited attenuation in mice (2). In the present study, constitutive overexpression of this gene behind a non-native strong promoter (G13) led to increased invasion of HEp-2 cells. Although it is possible that MAV_3679 is involved in the regulation of the mntH gene (regulated by mntR in S. aureus) also identified in the microarray (data not shown), the M. tuberculosis homolog of mntH is not involved in virulence (14), and the connection between the expression of MAV_3679 and M. avium mntH was not further explored.
For its genome size, M. avium has a very large number of tetR-like transcriptional regulators, and more than almost all other sequenced bacteria (28). The family of tetR regulators has many members, and the MAV_5138 gene is most similar to a homolog in M. tuberculosis of the AcrR family member. Proteins in the AcrR family have not been shown to play a role in virulence, but other TetR family regulators, such as HapR of Vibrio cholerae (21) and TvrR of Pseudomonas syringae (27), have been implicated in virulence. In general, the TetR family is important to the regulation of genes in response to the environment. M. avium has more than twice as many tetR-like genes than M. tuberculosis, so it is possible that there are as-yet-uncharacterized members in this large family involved in M. avium virulence. Similar to the sirR-like MAV_3679, overexpression of the MAV_5138 protein led to an increase in invasion of HEp-2 cells. TetR-like regulators are thought to influence their own expression (28).
Concurrently with the analysis of bacterial genes that may be regulated by MAV_3679 or MAV_5138, and in turn, by FadD2, we used a bacterial two-hybrid screen, with human Cdc42 as the bait, was used to explore this pathway from the host side. Fragments of three genes were retrieved from target plasmid after showing interaction with Cdc42: CipA, MAV_1190, and MAV_4139. These three genes were also upregulated in the strains of M. avium overexpressing both the MAV_3679 and MAV_5138 regulators.
The CipA protein sequence also does not contain any domains collected in the conserved domain database, but it does have similarity to domains by BLAST, as described in Results (Fig. (Fig.4).4). The PXXP piccolo domains in the first 100 amino acids of CipA, and the putative interaction with Cdc42, prompted us to analyze whether this protein is secreted or potentially expressed on the surface of M. avium in such a way that it is a part of a putative protein scaffold complex used to activate Cdc42 transmit signals to downstream proteins, such as N-WASp. Using a construct expressing a CipA:CyaA fusion protein, the presence of this protein could not be detected in the host cell cytoplasm by Western blot or increased levels of cAMP (Fig. (Fig.5).5). In addition, we looked for cleavage of the protein into two fragments using tags at the N and C terminus, using shorter tag sequences (13), but could detect neither fragment in the cytoplasm by Western blotting (data not shown). It is possible the protein is present in the host cells, but at levels below our ability to detect it. Together, these results suggest, however, that the protein is not secreted into the cytoplasm of the host cells. A study by Cain et al. (8) showed that the secreted effectors of Salmonella are also not secreted into the host cell cytoplasm but localize to the cell membrane instead, where they induce their effects on Cdc42 and downstream actin reorganization.
Because CipA was not detectable in the cytosol or the insoluble cell fractions, we hypothesized that this M. avium protein might be expressed on the surface of the bacterium upon invasion of epithelial cells, forming part of a structure which interacts with or inserts into the host cell membrane. Morphological changes in the recombinant strain of M. avium expressing the CipA:CyaA fusion protein lent support to the temporary localization of this protein to the bacterial membrane or cell wall (data not shown). Both CipA and MAP3985c (Cdc42 binding protein) are in small operons containing a hydrolase. MAP3985c was identified based on its interaction with an oxidoreductase shown to be involved in M. avium subsp. paratuberculosis invasion of epithelial cells (1). In the present study, CipA was also shown to putatively interact with an oxidoreductase (Table (Table33).
There are many hypothetical proteins in this region, corresponding to putative transmembrane proteins, a rho-kinase, and proteins with bacterial SH3 domains. Also present in this region are both characterized and uncharacterized transcriptional regulators, including regX3/senX3 and MAV_4676, a regulator that was shown by microarray to be upregulated upon exposure to HEp-2 cells, but not in the fadD2 mutant strain, and the hbhA gene, encoding the surface heparin-binding hemagglutinin protein shown to adhere to epithelial cells (29, 33). This region is currently being further analyzed for its role in the invasion of epithelial cells.
M. avium does not have a TTSS but may have an analogous mechanism for getting proteins into the host cell, where they can interact with host cell signaling pathways. Because the CipA protein has domains suggesting its binding to host cell proteins, but could not be shown to be secreted, we expressed this protein in M. smegmatis, fused to GFP, and were able to observe a structure at the end of M. smegmatis near host cells.
Our results lead us to hypothesize that the region of the genome including cipA is important to the invasion of epithelial cells by M. avium. Our model suggests that in the presence of host-specific environmental cues, various regulators, including fadD2, lead to the activation of additional transcriptional regulators, through as-yet-unknown mechanisms. These regulators, including MAV_3679 and MAV_5138, directly or indirectly activate the expression of proteins from this region, and likely other regions, that make up the components of a mechanism for altering host cell signaling. Observed putative protein-protein interactions and other data from our lab and others suggest that oxidoreductases may play a role in the regulation of this mechanism as well, perhaps acting as chaperones. As the bacterium comes into contact with the host cells, these proteins are inserted into the host cell membrane, where they form a complex with Cdc42 and other scaffolding proteins that are present leading to actin polymerization and subsequent uptake of the bacterium. Future work will address the identification and characterization of the proteins involved in this mechanism of epithelial cell invasion.
This study was supported by National Institutes of Health grant R01 AI43199.
We thank Denny Weber for editing the manuscript.
Published ahead of print on 5 December 2008.