Expression, production and release of the Eis protein by Mycobacterium tuberculosis during infection of macrophages and its effect on cytokine secretion

Mycobacterium tuberculosis, the causative agent of tuberculosis (TB), is one of the oldest pathogens known to man and is responsible for the deaths of over 1.8 million people every year (Corbett et al., 2003). M. tuberculosis survives within the host macrophage by arresting the normal development of the phagosome (Chan et al., 1991; Fratti et al., 2003). Phagosomelysosome fusion is inhibited in part by effectors that are secreted by M. tuberculosis, such as mannose-capped lipoarabinomannan (ManLAM) (Chan et al., 1991; Fratti et al., 2003). In this manner, the bacterium is able to avoid the toxic effects of lysosomal enzymes and the normal acidified environment of a mature phagosome.

However, mycobacterial survival also hinges on its ability to modulate host cytokine secretion. M. tuberculosis infection or mycobacterial proteins alone induce the secretion of a large number of cytokines including interleukin (IL)-1, IL-2, IL-10, IL-12 and tumour necrosis factor alpha (TNF-α) by monocytes/macrophages (Barnes et al., 1992; Fulton et al., 1996; Lee et al., 2003b; Wallis et al., 1990). The important proinflammatory cytokine TNF-α plays a key role in the host immune defence against TB (Flynn et al., 1995; Keane et al., 2001). In contrast, IL-10 downregulates cytokine production and accessory cell function of mononuclear phagocytes (D'Andrea et al., 1993; de Waal Malefyt et al., 1991) and its expression is also induced by various mycobacterial products (Barnes et al., 1992).

A wide variety of mycobacterial proteins and lipids are secreted into the cytoplasm of M. tuberculosis-infected macrophages (Beatty & Russell, 2000; Beatty et al., 2001, 2000; Xu et al., 1994). The trafficking of these mycobacterial components within the endocytic network of the host cell may play a role in modulating macrophage function to the advantage of the bacterium. These mycobacterial components are also released from the macrophage into the extracellular milieu in vesicular compartments (Beatty et al., 2001; Rhoades et al., 2003). Furthermore, these vesicles are taken up by uninfected bystander cells (Beatty et al., 2001, 2000). The propensity of M. tuberculosis to extend its influence beyond the constraints of the infected host cell is in keeping with its ability to subvert the immune system of the host to enhance its survival.

The eis (enhanced intracellular survival) gene of M. tuberculosis is only found in members of the M. tuberculosis complex and has been shown to confer enhanced survival on Mycobacterium smegmatis in U-937 macrophages and human macrophages (Wei et al., 2000). The Eis protein is a 42 kDa protein that is secreted when M. tuberculosis is grown in liquid culture (Dahl et al., 2001). In the present study, we constructed and utilized an eis mutant (Δeis) and complemented strains of M. tuberculosis H37Ra and H37Rv to study the role the gene plays in the microbe's interactions with macrophages. We discovered that the Eis protein is produced and released by M. tuberculosis into the cytoplasm of the phagocyte during intracellular infection. We also found that Eis is able to modulate the secretion by human monocytes of the cytokines TNF-α and IL-10, which play a major role in the regulation of the inflammatory response to infection. Bioinformatic studies show that Eis is a putative acetyltransferase of the GCN5-related family of N-acetyltransferases (GNAT) that may play a role in deregulating macrophage cytokine production.

Strains and growth media.
M. tuberculosis H37Rv and H37Ra were grown in Middlebrook 7H9 (Difco) broth. For growth on solid culture media, Middlebrook 7H10 (Difco) with 10 % OADC was used. Mycobacterial inocula for infection of U-937 cells with MtbRa/gfp and MtbRaΔeis/gfp were prepared as previously described (Wei et al., 2000). For growing Escherichia coli transformants, LuriaBertani (LB) broth or agar with 200 µg hygromycin B ml¹ or 100 µg ampicillin ml¹ was used.

M. tuberculosis H37Ra wild-type and Δeis were transformed with the vector pBEN, which contains gfp driven by phsp60, a strong heat-shock promoter from Mycobacterium bovis BCG (Table 1) (Saviola et al., 2003) to generate the gfp-expressing strains MtbRa/gfp and MtbRaΔeis/gfp, respectively. The promoterless gfp vector pFPV27 was used as the negative control for these studies. Both vectors were generously supplied by Stanley Falkow, Stanford University School of Medicine. Bacteria were grown on Middlebrook 7H10 medium containing 25 µg kanamycin ml¹. Fluorescence of transformants was confirmed by fluorescence microscopy, as described below.

Table 1. Plasmids used in this study

Human macrophage-like U-937 cells (ATCC; 1593.2 CRL) were grown in RPMI 1640 medium (Sigma) containing 10 % fetal calf serum (FCS) (Atlanta Biologicals) under 5 % CO₂ at 37 °C (Nilsson & Sundstrom, 1974). U-937 macrophage monolayers were prepared by transformation of cells in T75 flasks (Corning) using phorbol myristic acetate (PMA), as previously described (Wei et al., 2000).

General DNA methods.
For DNA manipulations, standard protocols were followed (Sambrook et al., 1989). Agarose was purchased from Promega. Restriction enzymes, Klenow fragment of T4 DNA polymerase, T4 DNA ligase and calf intestinal alkaline phosphatase (CIAP) were from New England BioLabs or Gibco-BRL Life Technologies. All reagents were used according to the manufacturer's instructions. Plasmids were constructed in E. coli HB101 or DH5α cells and prepared using a Qiaprep spin miniprep kit (Qiagen).

Recombinant Eis.
In order to obtain purified Eis protein in sufficient quantities for use in our studies, we constructed a recombinant His-tagged Eis protein. The 1.3 kb eis gene was PCR amplified and cloned into the pET-15b vector (Novagen) so as to obtain a recombinant protein with an N-terminal His-tag. The plasmid bearing the insert (pET15b-eis) was transformed into the E. coli expression strain BL-21 DE-3 pLysS (Novagen) (Table 2). Recombinant protein was then induced, harvested and purified as per the manufacturer's protocol (Novagen). The one step Ni-NTA purification process resulted in a single protein band, of molecular mass similar to that predicted for Eis (42 kDa) (Wei et al., 2000). Purity of the recombinant Eis was greater than 99 %, as determined by scanning densitometry analysis of SDS-PAGE gels, and the His-tagged Eis protein also reacted with the anti-Eis antibody (data not shown). Limulus lysate amoebocyte assay was utilized to test for endotoxin contamination.

Table 2. Bacterial strains used in this study

Infection of mice.
Five- to eight-week-old female C57/BL6 mice were purchased from Taconic and aerosol infected as described previously (Boshoff et al., 2003). Each mouse was infected with approximately 50 c.f.u. M. tuberculosis and bacterial burdens were monitored by plating of lung and spleen homogenates from four or five mice at different times during the infection. In addition, mice lungs and spleens were fixed in 10 % neutral buffered formalin, embedded in paraffin, sectioned, and stained for histology with either haematoxylin/eosin or ZiehlNeelsen stain.

Immunofluorescence microscopy of infected U-937 macrophages.
One-millilitre aliquots of RPMI containing 2x10⁵ U-937 cells were plated into each well of a 24-well tissue culture plate (Falcon) containing sterile glass coverslips. Cells were infected the next day as described previously (Wei et al., 2000) with MtbRa/gfp and MtbRaΔeis/gfp at an m.o.i. of 20. At appropriate time points, coverslips were washed three times with PBS and cells fixed by addition of 1 ml 3 % formaldehyde in PBS to each well. After 15 min, coverslips were washed three times with PBS and cells were permeabilized by addition of 0.1 % Triton X-100 for 5 min. Coverslips were again washed, and incubated with a 1 : 500 dilution of anti-Eis antibody for 1 h at room temperature on an orbital shaker. Anti-Eis antibody is an affinity-purified antibody to Eis generated in rabbits as described previously (Dahl et al., 2001). Alexafluor 594 goat anti-rabbit IgG linked to a fluorescent red dye (Molecular Probes; 1 : 10 000) in PBS containing 10 % FCS was then added and incubated with the coverslips. Wells were washed and coverslips were removed to a paper towel to air dry in the dark. Two microlitres of p-phenylenediamine mounting medium was added to each coverslip, which was then inverted onto a glass slide, sealed with clear nail polish and viewed using a Nikon Eclipse 2000-S fluorescent/phase-contrast microscope. Images were obtained as TIFF files using Metacam software and adjusted for consistent contrast using Photoshop 7.0 (Adobe).

Infection, harvesting and fractionation of infected U-937 macrophage monolayers for Western blot analysis of macrophage cytoplasm.
Infection of U-937 monolayers was carried out as described above (Wei et al., 2000). An m.o.i. of 15 was used and the infection was allowed to proceed for 4 h. Monolayers were then washed and 15 ml RPMI+5 % human AB sera was added to each flask and incubated for 48 h. Monolayers were then washed with Hanks' balanced salts solution (HBSS) and harvested into 15 ml HBSS by agitation with 3 mm glass beads. Cells were washed and resuspended in cold homogenization buffer (250 mM sucrose, 0.5 mM EDTA, 0.5 mM EGTA, 20 mM HEPES) and then lysed by repeated passage through a 25 gauge needle. Lysate was centrifuged at 300 g for 10 min followed by two subsequent spins at 100 g for 5 min each. The supernatant was carefully layered over a 12 % sucrose cushion and centrifuged at 800 g for 45 min at 4 °C. The cytoplasmic fraction remained in a distinct upper layer, separate from mycobacteria and phagosomes containing mycobacteria in the lower layer and pellet, respectively.

Isolation of exocytic vesicles from culture supernatant of infected U-937 macrophages.
Exocytic vesicles were isolated from tissue culture supernatants as described previously (Beatty et al., 2001). Briefly, tissue culture supernatants from U-937 macrophages infected with M. tuberculosis H37Ra were harvested and subjected to a series of centrifugation steps to remove whole cells and bacteria. The supernatant was then centrifuged for 60 min at 100 000 g. The resulting pellet contained the vesicles released from the macrophages which were analysed by Western blotting for the presence of the Eis protein. Culture supernatant obtained after processing as described above was concentrated threefold using YM-10 Centriprep filters (Millipore) and analysed by Western blot for the presence of Eis protein.

Western blot analysis.
Sample protein concentrations were determined by BCA protein assay (Pierce). Samples were separated on 10 % SDS-PAGE gels along with prestained molecular mass standards (Gibco-BRL). Electrophoresis was performed at 200 V and 80 mA for 34 h. Proteins were transblotted for 1 h using MiniProtean 3 cells (Bio-Rad) onto PVDF membranes (Millipore) which were then blocked with 5 % skim milk in PBS containing 0.02 % sodium azide. Membranes were incubated with primary antibody in PBS containing 0.25 % powdered skim milk for 1 h, washed three times with PBS and then probed with secondary antibody for 1 h. Secondary antibody was goat anti-rabbit IgG conjugated to horseradish peroxidase (HRP) diluted 1 : 20 000 and used in blots. Membranes were then washed three times with PBS and incubated with chemiluminescent substrate for 5 min followed by exposure to X-ray film for appropriate periods of time. All reagents for development of the chemiluminescent Western blots were obtained from Pierce.

Isolation and cultivation of human monocytes for analysis of cytokine secretion in response to antigenic stimulation.
Venous blood was drawn from tuberculin-positive healthy subjects into sterile blood collection tubes, and peripheral blood mononuclear cells (PBMCs) were isolated by density sedimentation over Histopaque-1077 (Sigma). PBMCs were suspended at a density of 2x10⁶ cells ml¹ in complete RPMI medium (Gibco-BRL) with 10 % fetal bovine serum (Gibco-BRL), sodium pyruvate, nonessential amino acids, penicillin G (100 IU ml¹) and streptomycin (100 µg ml¹). Cells were incubated for 1 h at 37 °C and non-adherent cells were removed by pipetting off the supernatant. Adherent monocytes were collected as previously described (Song et al., 2003). The recovered cells were >95 % CD14⁺ cells, as determined by flow cytometry using an anti-CD14 antibody. The cells were then stimulated with either purified protein derivative (PPD) (Statens Serum Institut), Eis, or lipopolysaccharide (LPS, 10 ng ml¹; Sigma) and incubated at 37 °C in a 5 % CO₂ humidified air atmosphere. The supernatants from 18 h (for TNF-α) and 48 h (for IL-10) cultures were frozen at 80 °C until used in ELISA.

Preparation of inocula and infection of human monocytes with M. tuberculosis for analysis of cytokine secretion.
M. tuberculosis H37Rv was grown to late exponential phase in Middlebrook 7H9 as described previously, aliquoted and stored at 70 °C. Representative vials were thawed and c.f.u. enumerated on Middlebrook 7H10 agar. Single-cell suspensions of mycobacteria were obtained by a modification of standard methods. Briefly, aliquots of frozen M. tuberculosis were cultured in 7H9 broth with 0.5 % (v/v) glycerol at 37 °C in 5 % CO₂ for 710 days so that cultures reached mid-exponential growth phase. Bacterial cultures were pelleted at 3000 g for 10 min and resuspended in 7H9 broth. Clumped mycobacteria were dispersed with an ultrasonic cell disrupter (35 min, 35 kHz; Bandelin). Bacteria were then resuspended in 1 ml RPMI 1640 medium and clumps were disrupted by multiple passages through a 25 gauge needle. Mycobacterial viability, as assessed by the number of c.f.u., was 6070 %.

Adherent human monocytes were washed three times with Ca²⁺- and Mg²⁺-free PBS, and adherent monolayers were replenished with complete medium without antibiotics. The cells were incubated overnight without stimulation at 37 °C in 5 % CO₂. After overnight incubation, monocytes (2x10⁵ ml¹) were infected with mycobacteria using an m.o.i. of 1 for either 18 h or 48 h, and the supernatants were then recovered for ELISA determination of cytokine levels of TNF-α and IL-10, respectively.

ELISA.
ELISA was used to detect TNF-α and IL-10 levels (BD Biosciences) in monocyte culture supernatants, as previously described (Lee et al., 2003a). Assays were performed as recommended by the manufacturers. Cytokine concentrations in the samples were calculated using standard curves generated from recombinant cytokines, and the results were expressed in picograms per millilitre. The difference between triplicate wells was consistently less than 10 % of the mean.

Bioinformatic analysis.
Protein sequences for analysis were obtained in FASTA format and input into respective programs at the interface locations described below. Reverse position-specific (RPS)-BLAST is a variation of PSI-BLAST, which utilizes the Conserved Domain Database (CDD) to compare a single sequence against a database of PSSM (Position-Specific Score Matrix) models (Marchler-Bauer & Bryant, 2004). This is available for use at Pfam is a comprehensive database of protein domain families based on seed alignments using hidden Markov model profiles. The 3D-PSSM (three-dimensional position-specific scoring matrix) server is designed to take a protein sequence of interest and attempt to predict its 3D structure and its probable function. This program is able to confidently predict structural and functional relationships better than standard local alignment search tools such as PSI-BLAST (Kelley et al., 2000). 3D-PSSM is administered by Lawrence Kelley and is available for use at .

Statistical analysis.
For statistical analysis, data obtained from independent experiments were presented as the mean±SD and were compared by Student's t-test or for multiple comparisons by ANOVA. Differences were considered significant for P<0.05.

Survival of M. tuberculosis wild-type and eis mutant in vitro and in vivo
Having demonstrated previously that the eis gene enhanced intracellular survival of M. smegmatis (Wei et al., 2000), it was only natural to look for a similar phenotype in M. tuberculosis. The first step towards answering this question was to develop a Δeis and complemented strain in both M. tuberculosis H37Ra and H37Rv. Deletion and complementation of the eis gene was confirmed by both Southern and Western blot analysis (data not shown). Survival of the H37Rv wild-type, Δeis mutant and complemented strains in U-937 macrophage monolayers was then compared. There was no significant difference between the three strains in terms of intracellular survival in U-937 macrophages for the duration of the 7 day experiment (data not shown).

Survival of these three strains was then compared in an in vivo mouse aerosol model (Dahl et al., 2003). Six- to eight-week-old female C57BL/6 mice were infected via the aerosol route with M. tuberculosis H37Rv wild-type, Δeis mutant and complemented strains. At various time points over a 17 week period, the mice were sacrificed and lung and spleen homogenates were plated for c.f.u. determinations and tissue samples were also taken for histological evaluation. There appeared to be no significant difference between survival of the wild-type, Δeis mutant and complemented strains in the in vivo model as determined by viable plate counts (Fig. 1). Also, no histological differences were observed between infected lung or spleens from wild-type or Δeis mutants. This included severity of granuloma formation and inflammation, granuloma numbers and numbers of acid-fast bacteria per tissue section (data not shown).

(19K): Fig. 1. Growth of M. tuberculosis H37Rv, Δeis and complemented strains in the lungs (a) and spleens (b) of aerosol-infected mice. Female C57/BL6 mice were infected via the aerosol route and monitored by c.f.u. determinations of lung and spleen homogenates via viable plate counts. These data are representative of three independent experiments using four or five mice per time point after initial infection. , Wild-type; , Δeis; , complemented Δeis.

Bioinformatic analysis of Eis
Bioinformatic analysis of the Eis protein sequence using the proteinprotein BLAST (BLASTP) program (Altschul et al., 1990) yielded a large number of homologous proteins, many of which were members of the GNAT superfamily of proteins. RPS-BLAST (Marchler-Bauer & Bryant, 2004) analysis of Eis revealed the presence of an acetyltransferase domain that is characteristic of the pfam00583 family of GNAT acetyltransferases (Neuwald & Landsman, 1997). The acetyltransferase domain extends from residues 61 to 137 of the 402 aa Eis protein sequence and contains the conserved residues (V/I-x-x-x-x-Q/R-x-x-G-x-G/A) that are characteristic of this domain (Neuwald & Landsman, 1997). A substitution of glycine or alanine with lysine at the last conserved residue, as seen with Eis, is found in other members of this family (Neuwald & Landsman, 1997). Since members of the GNAT superfamily have poor sequence homology but share common 3D structural features (Dyda et al., 2000), the 3D-PSSM program was used to search for proteins showing significant fold compatibility with Eis. Fold analysis of a 154 aa section of the Eis sequence, which included the acetyltransferase domain using 3D-PSSM, indicates that Eis shares strong similarities with several proteins belonging to the GNAT family of acetyltransferases including the conserved residues (Table 3). These bioinformatics analyses suggest that Eis may function as an acetyltransferase.

Table 3. 3D-PSSM search for structural homologues to the Eis protein The predicted 3D structure of Eis shows significant homology to the known 3D structures of several members of the pfam00583 family of GNAT acetyltransferases as demonstrated by their E-values. The 3D structures of the proteins with the lowest E-values demonstrate the highest levels of similarity to the predicted 3D structure of Eis.

Presence of Eis in the cytoplasm of macrophages infected with M. tuberculosis
Because the Eis protein was shown to be secreted into the bacterial culture medium by M. tuberculosis in vitro (Dahl et al., 2001), studies were done to examine infected macrophages for the presence of the Eis protein in the host cell cytoplasm. M. tuberculosis H37Ra and Δeis strains were transformed with the pBEN plasmid, which contains the green fluorescent protein (gfp) gene driven by phsp60 from Mycobacterium bovis BCG (Saviola et al., 2003). M. tuberculosis H37Ra wild-type and Δeis strains expressing gfp were named MtbRa/gfp and MtbRaΔeis/gfp, respectively. The MtbRa/gfp- and MtbRaΔeis/gfp-infected cells were washed, permeabilized, and probed with rabbit anti-Eis antibody (Dahl et al., 2001) followed by goat anti-rabbit antibody labelled with a fluorescent red dye.

Macrophages infected with MtbRaΔeis/gfp, probed with both primary and secondary antibodies and visualized using immunofluorescence microscopy (IF) showed no red staining for the Eis protein at any time point due to the absence of Eis production in the MtbRaΔeis/gfp mutant (Fig. 2a, b). Starting at 4 h post-infection up to 96 h post-infection, U-937 cells infected with MtbRa/gfp demonstrated the presence of Eis protein in the cytoplasm as evidenced by red fluorescence when viewed by IF (Fig. 2ch). No Eis protein was visible when the cells were examined 30 min post-infection (data not shown). In many instances, macrophages appeared to contain Eis protein in the absence of visible infecting bacteria (Fig. 2f, g; arrowheads denote uninfected macrophages). In control experiments, no IF was observed in uninfected U-937 cells that were probed in a similar manner. This was also true for cells that were not permeabilized before the addition of the primary and secondary antibodies (data not shown), indicating that the Eis protein being visualized was not tethered to the surface of the macrophage. This experiment was repeated three times and similar results were obtained each time. Similar results were also observed using both J774 murine macrophages and primary human monocytes (data not shown).

(52K): Fig. 2. Immunofluorescence microscopy demonstrating the presence of Eis in the cytoplasm of U-937 macrophages infected with MtbRa/gfp. U-937 macrophages were infected at an m.o.i. of 20 and at various time points cells were fixed, permeabilized, probed and viewed by IF. Mycobacteria appear as fluorescent green clumps while the Eis protein appears as fluorescent red dispersed throughout the cytoplasm of MtbRa/gfp-infected macrophages. (a) IF showing fluorescent green MtbRaΔeis/gfp (white arrow) in U-937 macrophages. No Eis was detected in the cytoplasm of these macrophages as observed by the absence of red fluorescence. (b) Phase-contrast and fluorescent microscopy images of the field in (a) were merged, visualizing the macrophage and the fluorescent green bacteria (white arrow). (c) U-937 cells infected with MtbRa/gfp demonstrating red fluorescence throughout the cytoplasm due to the release of Eis protein from mycobacterial phagosomes at 4 h, (d) 12 h, (e) 24 h, (f) 48 h, (g) 72 h and (h) at 96 h after infection. In both (f) and (g) uninfected macrophages (indicated by white arrowheads) demonstrate the presence of the Eis protein in the cytoplasm in the absence of infecting bacteria. Bar, 20 µm.

Western blot analysis of the cytoplasm of infected macrophages for the presence of Eis
Western blot analysis using the anti-Eis antibody of lysates prepared from infected macrophages as previously described (Chakraborty et al., 1994) was used to confirm the IF observation of the presence of Eis in the cytoplasm of these macrophages. The Eis protein was not detected in the cytoplasmic fraction of macrophages infected with M. tuberculosis H37RaΔeis (Fig. 3, lane 1) but was found in the cytoplasmic fraction of U-937 cells infected with M. tuberculosis H37Ra wild-type at 48 h after infection (Fig. 3, lane 2). The infecting bacteria had been removed from these fractions by centrifugation of the lysate over a 12 % sucrose solution. To discount the possibility that the presence of Eis in the macrophage cytoplasm may be due to intracellular lysis of phagocytosed bacteria, the cytoplasmic fraction was probed by Western blot analysis for the presence of the 16 kDa alpha crystallin protein (Acr) at 48 h after infection. Acr is a mycobacterial protein encoded by hspX that is upregulated in M. tuberculosis during infection of macrophages and is known not to be secreted from the bacteria (Dubnau et al., 2002). Acr was not detected in the cytoplasmic fraction of macrophages infected with M. tuberculosis by Western blot analysis.

(114K): Fig. 3. Western blot detection of Eis in the cytoplasm and tissue culture supernatant of U-937 macrophages infected with M. tuberculosis H37Ra wild-type and Δeis mutant. Molecular mass markers and their sizes are represented at the left side of the blot. All lanes were probed with anti-Eis antibody. Lane 1, cytoplasmic fraction of U-937 macrophages infected with M. tuberculosis H37RaΔeis. No Eis protein is detected. Lane 2, presence of Eis in the cytoplasmic fraction of U-937 macrophages infected with M. tuberculosis H37Ra wild-type. Lane 3, extravesicular fraction from the tissue culture supernatant of M. tuberculosis H37Ra wild-type-infected U-937 macrophages. Eis was found to be associated with vesicles released from infected U-937 macrophages. Lane 4, presence of native Eis protein in the tissue culture supernatant of U-937 macrophages infected with M. tuberculosis H37Ra after removal of all vesicular material followed by concentration. In all cases, samples were loaded onto 10 % SDS-PAGE gels, separated by electrophoresis, and transblotted onto PVDF membranes for Western blot analysis. Two hundred micrograms of total protein was loaded per lane. The black arrowhead to the right of the blot denotes the position of the Eis protein.

We then attempted to explain the presence of Eis protein in macrophages that appeared to be visibly uninfected (Fig. 2f and g, arrowheads) by examining the vesicular fraction of the culture supernatant as described previously by Beatty et al. (2000). Tissue culture supernatant from U-937 cells infected with M. tuberculosis H37Ra was harvested at 48 h after infection and then subjected to a series of centrifugation steps to remove any contaminating bacteria and macrophage organelles and to separate out the microvesicular fraction. Analysis of this fraction containing the vesicles released by macrophages into the culture supernatant by Western blot detected the presence of the Eis protein (Fig. 3, lane 3). The 75 kDa band that is seen in this lane was not detected in either the cytoplasmic fraction (Fig. 3, lane 1) or the extravesicular fraction of U-937 cells infected with M. tuberculosis H37RaΔeis (data not shown). As the anti-Eis antibody used in these studies is affinity purified (Dahl et al., 2001), it is unlikely to cross-react with host cell proteins. Eis protein has a tendency to aggregate in solution and this band is probably the result of aggregate formation. Nevertheless, the possibility that the 75 kDa band may represent Eis covalently bound to a mammalian protein needs to be further investigated.

These findings are consistent with the reports of detection of other mycobacterial proteins and lipids in vesicles released into the tissue culture supernatant by M. tuberculosis-infected macrophages (Beatty et al., 2001, 2000; Rhoades et al., 2003). The steps utilized to obtain the vesicular pellet resulted in a tissue culture supernatant fraction devoid of any host cellular structural material. Western blot analysis of a concentrated aliquot of this fraction detected the presence of the Eis protein (Fig. 3, lane 4), the source of which may be the degradation of vesicles containing mycobacterial proteins in the culture supernatant or possibly due to exocytic release of protein from M. tuberculosis-infected macrophages.

Viability of infected macrophages was determined using an LDH release assay (Promega) to ensure that the release of Eis and vesicles containing Eis into the tissue culture supernatant was not due to lysis of infected macrophages. Based on LDH release, it was determined that macrophage viability was >90 % at time of harvest (48 h).

Cytokine release by human monocytes in response to stimulation with recombinant Eis
To evaluate cytokine production in primary human monocytes in response to recombinant Eis or PPD antigen (an M. tuberculosis antigen preparation commonly used in such immunological studies), assays were performed to detect TNF-α and IL-10 secretion by ELISA. These cytokines are known to play a significant role in the pro- and anti-inflammatory components respectively of the immune response to M. tuberculosis infection (Flynn et al., 1995; Gong et al., 1996). Mycobacterial proteins/lipids released from the phagosome of M. tuberculosis-infected macrophages have also been shown to modulate levels of cytokine secretion during infection, to the advantage of the pathogen (Gil et al., 2004; Placido et al., 1997; Rivera-Marrero et al., 2004).

Experiments using human monocytes from three healthy tuberculin reactors showed that TNF-α and IL-10 were not produced in freshly isolated cells but were detectable 3 h after and peaked from 1824 h (for TNF-α) and 48 h (for IL-10) after stimulation with recombinant Eis or PPD antigen (data not shown). Therefore, the 18 h and 48 h time points were used to study the levels of TNF-α and IL-10 production, respectively, after stimulation with the Eis or PPD antigens of M. tuberculosis. As shown in Fig. 4(a), Eis-induced TNF-α production was increased in human monocytes in a dose-dependent manner. However, Eis elicited significantly lower amounts of TNF-α at 5 and 10 µg ml¹, when compared with the same concentrations of the PPD antigen (P<0.05; Fig. 4a). Stimulation of human monocytes with 10 ng LPS (positive control) resulted in the secretion of TNF-α (10002500 pg ml¹).

(20K): Fig. 4. Cytokine secretion by human monocytes in response to Eis. Monocytes were isolated from tuberculin-positive healthy donors and either left untreated or stimulated with increasing amounts of PPD and Eis antigen. (a) TNF-α concentration was measured by ELISA in the cell culture supernatants harvested at 18 h following stimulation with each antigen and expressed as the mean±SE of three independent experiments. (b) IL-10 concentration was measured by ELISA in the cell culture supernatants harvested at 48 h following stimulation with each antigen and expressed as the mean±SD of three independent experiments. n=3 for each experiment. *P<0.05. (c) Monocytes from healthy tuberculin-positive donors were harvested and either not infected (1), or infected with M. tuberculosis H37Rv wild-type (2), Δeis (3) and complemented strains (4) at an m.o.i. of 1. Culture supernatants were harvested 18 h after infection to determine levels of cytokine production. TNF-α levels in the culture supernatants were monitored by ELISA. A representative experiment is shown, n=3. (d) IL-10 secretion by infected monocytes from healthy donors as described for (c). IL-10 levels in the supernatants were monitored by ELISA. A representative experiment is shown in (b), n=3. *P<0.05; **P<0.01.

As shown in Fig. 4(b), significant dose-dependent IL-10 production was observed by human monocytes after in vitro stimulation with the Eis protein. In contrast to the findings for TNF-α, the IL-10 production induced by Eis protein at 5 and 10 µg ml¹ was higher (P<0.05) than that induced by the PPD antigen at the same concentrations (Fig. 4b). Stimulation of human monocytes with 10 ng LPS (positive control) resulted in the increased secretion of IL-10 (20004000 pg ml¹). Recombinant protein preparations were tested to ensure absence of any contaminating endotoxins.

Cytokine release by human monocytes in response to infection with M. tuberculosis H37Rv wild-type, Δeis and complemented strains
Next, TNF-α and IL-10 secretion by human monocytes infected with wild-type M. tuberculosis H37Rv, Δeis and eis complemented strains at an m.o.i. of 1 were evaluated. As shown in Fig. 4(c), TNF-α production by human monocytes infected with the H37RvΔeis was significantly greater than those infected with wild-type H37Rv (P<0.01) or the complemented strain (P<0.05). Although the mean TNF-α production by M. tuberculosis-infected monocytes increased slightly with bacterial load, H37Rv wild-type infected monocytes still secreted significantly less TNF-α than those infected with the Δeis strain at an m.o.i. of 10 (data not shown).

In contrast to TNF-α, IL-10 secretion by H37RvΔeis-infected monocytes was significantly lower than induction by H37Rv wild-type (P<0.01) or eis-complemented strains (P<0.01) (Fig. 4d). With increased bacterial load (m.o.i. of 10), the H37RvΔeis-infected monocytes secreted significantly less IL-10 than those infected with wild-type or complemented strains (data not shown). These results show that strain H37RvΔeis induced more TNF-α but less IL-10 production in primary human monocytes than either H37Rv wild-type or complemented strains.

The eis gene of M. tuberculosis enhances the intracellular survival of M. smegmatis (Wei et al., 2000). However, its function in the virulence of M. tuberculosis has not yet been elucidated. No significant difference was noted in the survival of M. tuberculosis H37Rv wild-type, Δeis mutant and complemented strains in either in vitro or in vivo models (Fig. 1). However, Barczak et al. (2005) demonstrated that there was no significant difference in the growth of the hypovirulent CDC 1551 and the hypervirulent HN878 strains of M. tuberculosis in prestimulated macrophages. They similarly noted no difference in the mycobacterial loads of the lungs of mice infected with either strain. Instead the difference between the two strains lay in the growth kinetics and mouse survival (Barczak et al., 2005). Further studies are required to determine whether the same is true of strains deficient in eis. It is possible that the role of eis in the pathogenesis of M. tuberculosis may differ significantly from the phenotype observed in M. smegmatis (Wei et al., 2000), a surrogate host for this M. tuberculosis-derived gene, thus explaining the lack of an intracellular survival phenotype in the current study.

The discovery that eis is expressed during infection of activated human macrophages by clinical strains of M. tuberculosis has been confirmed by others (Cappelli et al., 2001). In these present studies, RT-PCR indicated that eis is constitutively expressed by the M. tuberculosis H37Ra strain whether in culture media or in macrophages (data not shown). This idea is corroborated by studies by Dahl et al. (2005) showing that eis is negatively regulated by the stringent response regulator Rel_Mtb so as to maintain constant levels of expression during starvation conditions. On the other hand, Wu et al. (2005) have demonstrated that the expression of eis was enhanced 12-fold in the clinical strain M. tuberculosis 210 as compared to the H37Rv strain when grown in a human monocyte cell line, whereas the expression was the same when the strains were grown in vitro. They also showed that SigA binds to the promoter region of eis and is responsible for the upregulation of eis expression in strain 210. They concluded that eis may contribute to the enhanced growth of the 210 strain in human macrophages. Morris et al. (2005) have also recently demonstrated that eis is a component of the whiB7 regulon, whose expression is upregulated by exposure to sublethal concentrations of antibiotics or fatty acids such as would be encountered within the host macrophage during infection. These reports in conjunction with the data presented here point to a significant but not yet clearly defined role for eis in the pathogenesis of M. tuberculosis.

Multiple sources have reported on the release of mycobacterial constituents from infected macrophages into the culture medium followed by their uptake into uninfected bystander cells (Beatty & Russell, 2000; Beatty et al., 2001; Rhoades et al., 2003). Our studies using IF of MtbRa/gfp-infected U-937 macrophages demonstrate the production and presence of Eis in the macrophage cytoplasm from 4 h to 96 h post-infection (Fig. 2ch). Furthermore, macrophage lysate fractions, purged of infecting bacteria, demonstrated the presence of the Eis protein by Western blot analysis (Fig. 3, lane 2). The presence of Eis in the extravesicular fraction and the tissue culture supernatant of infected macrophages (Fig. 3, lanes 3 and 4) places it in the company of other known mycobacterial effectors of virulence such as lipoarabinomannan (LAM) (Beatty et al., 2000). These mycobacterial protein/lipid-containing vesicles appear to be taken up by uninfected macrophages, which accounts for the presence of mycobacterial components within host cells in the absence of bacteria (Beatty et al., 2000). The release of Eis into the culture supernatant by infected macrophages, either in vesicles or in native form, followed by its uptake into uninfected macrophages (Fig. 2g, h), is the most likely explanation for its presence in these cells. Further work needs to be done to determine the localization of secreted Eis within the macrophage and examine its significance.

The critical importance of TNF-α in anti-mycobacterial defence is well established in mice and humans (Flynn et al., 1995; Keane et al., 2001). TNF-α helps prevent the reactivation of persistent TB and limits the pathological response of the host. The concept that M. tuberculosis plays a role in modulation of the TNF-α/IL-10 axis is supported by our data showing that Eis modulates TNF-α and IL-10 secretion by human monocytes in response to infection by M. tuberculosis (Fig. 4c, d). Because purified Eis elicits higher levels of IL-10 than TNF-α secretion, as compared to PPD (Fig. 4a, b), we examined the differences in cytokine secretion in response to infection with viable M. tuberculosis H37Rv wild-type, Δeis and eis-complemented strains. In contrast to H37Rv and the complemented strain, the Δeis mutant induced significantly higher levels of TNF-α and lower levels of IL-10 in human monocytes (Fig. 4c, d). This provides powerful evidence that M. tuberculosis utilizes Eis to manipulate the host TNF-α/IL-10 axis to its advantage by inducing a state of localized immunosuppression in the host.

IL-10 is thought to play an important role in the regulation of host antimicrobial immunity. The induction of down-regulatory cytokines, such as IL-10 and transforming growth factor-β, can inhibit interferon (IFN)-γ production by T cells and block or inhibit macrophage activation and apoptosis (Flesch et al., 1994; Othieno et al., 1999). It is widely known that in alveolar macrophages, apoptosis is a common defence mechanism against M. tuberculosis infection via a TNF-α-mediated pathway (Placido et al., 1997). Monocytes from patients with active TB when compared to those from healthy donors are not only likely to secrete more IL-10 than TNF-α in response to stimulation with PPD, but they are also more likely to undergo necrosis than apoptosis (Gil et al., 2004). Utilizing the slowly progressive primary tuberculosis (SPTB) mouse model, Abebe et al. (2006) have demonstrated a 20-fold increase in the expression levels of IL-10 in the lungs of mice with progressive tuberculosis as compared to mice with latent tuberculosis. Increased IL-10, but decreased TNF-α production driven by Eis, and possibly other M. tuberculosis antigens, may contribute to the intracellular survival of the pathogen by suppressing host cell apoptosis during mycobacterial infection. It is quite likely that the modulation of TNF-α and IL-10 secretion by Eis will have a further effect on the levels of other cytokines due to their roles as prominent components of the pro- and anti-inflammatory immune response. Immunohistochemical analysis of M. tuberculosis-infected tissues for secreted Eis as well as altered cytokine expression may shed light on the role of eis in the pathogenesis of M. tuberculosis.

Bioinformatic analysis suggests that Eis is an acetyltransferase of the family of GCN5-related N-acetyltransferases. 3D-PSSM fold analysis shows that the predicted structure of Eis bears significant similarity to the structures of known members of the GNAT superfamily (Table 3) and also shares the conserved residues that are characteristic of this family. This family of proteins is involved in a wide variety of activities ranging from transcriptional activation by histone acetyltransferases to antibiotic resistance by aminoglycoside acetyltransferases. The lack of sequence homology among members of this family makes classification of possible new members difficult unless 3D structural data are available (Neuwald & Landsman, 1997). In the absence of candidate substrates, it is extremely difficult to demonstrate whether the Eis protein possesses functional acetyltransferase activity. Current efforts are being directed towards identification of possible substrates and resolution of the crystal structure of the Eis protein, which may provide concrete evidence as to the putative acetyltransferase activity of Eis and also offer direction as to the identity of possible substrates.

The absence of an observable intracellular survival phenotype for eis, while confounding, should not detract from the significant nature of the other findings presented here. Recent reports regarding the role of eis in clinical strains of M. tuberculosis indicate that further studies should be conducted using these strains. The release of Eis into the cytoplasm of infected macrophages and its apparent role in the modulation of cytokine secretion in these important host immune cells, essential for defence against tuberculosis infection, raises many interesting questions as to the mechanism(s) whereby this modulation is achieved. Perhaps Eis acetylates histones as other acetyltransferases do (Brownell & Allis, 1995) in the macrophage nucleus, resulting in reduced binding to DNA and enhanced transcription and translation of IL-10. Alternatively, Eis may act directly on mammalian cell transcription factors and alter the response of the host cell to infection as described for other acetyltransferases (Sterner & Berger, 2000). Eis, by one of these mechanisms, may directly affect macrophage cytokine production via activity in the host cell nucleus. Whether Eis may actually enter and localize in the macrophage nucleus is unknown and is presently under investigation. The data presented here point toward a significant and novel role for eis as yet another virulence factor in the pathogenesis of M. tuberculosis that warrants further study.

We thank James Moulder, Janet Hatt, Amy Windley and Christopher Alteri for invaluable assistance and technical suggestions. This work was supported in part by National Institutes of Health grant AI45537-01A2 to R. L. F. and Grant R042 004 0001 00220 2004 from the Korea Research Foundation for the 2004 program year to E. K. J.

Edited by: G. S. Besra

Footnotes

^,†, Chang-Hwa Song², Jun Wei^{1 ,‡}, Esteban A. Roberts¹ †Present address: Department of Clinical Microbiology, The University of Rochester, 601 Elmwood Ave, Box 710, Rochester, NY 14642, USA. ‡Present address: Amgen Inc., Thousand Oaks, CA 91320, USA. §Present address: Department of Molecular Genetics and Microbiology, University of New Mexico, Albuquerque, NM 87131, USA.