Report

A Bacterial Protein Targets the BAHD1 Chromatin Complex to Stimulate Type III Interferon Response

See allHide authors and affiliations

Science  11 Mar 2011:
Vol. 331, Issue 6022, pp. 1319-1321
DOI: 10.1126/science.1200120

Abstract

Intracellular pathogens such as Listeria monocytogenes subvert cellular functions through the interaction of bacterial effectors with host components. Here we found that a secreted listerial virulence factor, LntA, could target the chromatin repressor BAHD1 in the host cell nucleus to activate interferon (IFN)–stimulated genes (ISGs). IFN-λ expression was induced in response to infection of epithelial cells with bacteria lacking LntA; however, the BAHD1-chromatin associated complex repressed downstream ISGs. In contrast, in cells infected with lntA-expressing bacteria, LntA prevented BAHD1 recruitment to ISGs and stimulated their expression. Murine listeriosis decreased in BAHD1+/– mice or when lntA was constitutively expressed. Thus, the LntA-BAHD1 interplay may modulate IFN-λ−mediated immune response to control bacterial colonization of the host.

Listeria monocytogenes is a food-borne pathogen that can cause serious illness in pregnant women and immunocompromised individuals (1). This intracellular bacterium uses an arsenal of effectors to exploit cellular functions in various ways (2). Host cells respond to this invasion by turning on appropriate defense transcriptional programs (3). Listeria and other pathogens can manipulate chromatin to reprogram host transcription (4, 5). However, very few bacterial molecules have been shown to enter eukaryotic cell nuclei, and knowledge about microbial factors that may act directly on the chromatin-regulatory machinery is limited (6).

To identify factors involved in bacterial pathogenicity, we screened the L. monocytogenes strain EGDe genome for genes encoding secreted proteins absent in nonpathogenic Listeria species. lmo0438/lntA (listeria nuclear targeted protein A) was one such gene (fig. S1A). lntA was expressed at very low levels by the EGDe strain grown in brain-heart infusion (BHI) medium (fig. S1B) (7). Two major regulators of virulence genes, PrfA and σB, were required for basal lntA transcription (Fig. 1A). lntA expression was significantly higher in bacteria harvested from spleens of infected mice, 48 hours after intravenous inoculation, compared with that of bacteria grown in BHI (Fig. 1B). In addition, deletion of lntA led to a decrease in bacterial colonization of spleens and livers, as well as blood bacteraemia (Fig. 1B). lntA thus contributes to L. monocytogenes virulence. It encodes a 205–amino acid basic protein with a N-terminal signal peptide but no sequence similarity with any known polypeptide. The 2.3 Å resolution structure of LntA reveals a compact α-helical fold (fig. S2) [Protein Data Bank (PDB) ID no. 2xl4]. Consistent with low lntA transcription levels in vitro, LntA was undetectable in either total extracts or supernatants of wild-type (WT) bacteria grown in BHI (Fig. 1C).

Fig. 1

The secreted virulence factor LntA targets the nuclear protein BAHD1. (A) lntA is regulated by PrfA and σB. Real-time reverse transcription quantitative polymerase chain reaction (RT-QPCR) analysis of lntA levels in WT, ∆lntA, ∆prfA, or ∆sigB strains. (B) lntA is up-regulated and contributes to virulence 48 hours post infection in an intravenous mouse model. (Left) RT-QPCR analysis of lntA and control lmo2845 levels in WT Listeria extracted from spleens. (Right) Bacteria were numerated in organ or mL blood from mice infected with WT or ∆lntA strains. (A and B) **P < 0.01; ***P < 0.001 (two-tailed t tests). (C) LntA is a secreted protein. Bacterial total extracts (TE) and supernatants (Sn) of WT or lntAV5+ strains were analyzed by immunoblot, with ActA and InlC used as controls. WT bacteria do not produce LntA in BHI. P, precursor; S, secreted. (D) LntA localizes to the nucleus of C3SV40 fibroblasts. V5 immunolabeling and 4′,6′-diamidino-2-phenylindole (DAPI) staining in cells infected for 22 hours with lntA or lntAV5+ bacteria. (E) Purified LntA binds BAHD1. GST or GST-LntA were incubated with nuclear extracts from CFP-V5– or BAHD1-V5–expressing HEK293 cells. Immunoblots of inputs and eluted fractions were probed with antibodies against V5 (α-V5) or GST. (F) LntA localizes to BAHD1-induced heterochromatin foci. BAHD1-YFP and either LntA-V5 or CFP-V5 were cotransfected into C3SV40 cells and detected by immunofluorescence. (D and F) Scale bars, 5 μm.

To address the role of LntA during L. monocytogenes cellular infection, we generated strains that constitutively expressed lntA under the control of a heterologous promoter, either on the chromosome (lntAc+) or on a plasmid in fusion with the V5 tag (lntAV5+) (tables S1 and S2). Both strains produced and secreted LntA (Fig. 1C and fig. S1C) and showed no noticeable difference in entry or multiplication in cultured cells compared with the WT or lntA-deficient strains (ΔlntA, lntA, or lntAc–) (table S3A and fig. S3). Secreted LntA accumulated in the nucleus of fibroblasts after 22 hours of infection with lntAV5+ bacteria (Fig. 1D and fig. S4A). We thus assessed whether LntA interacted with nuclear proteins in a large-scale yeast two-hybrid screen of a human cDNA library. One of the strongest LntA interactors was BAHD1, a silencing factor that orchestrates heterochromatin assembly at specific genes such as that for insulin-like growth factor 2 IGF2 (8). A fusion of LntA with glutathione S-transferase (GST) pulled down V5-tagged BAHD1 from nuclear extracts, which confirmed the capacity of LntA to specifically interact with BAHD1 (Fig. 1E). When produced ectopically in human fibroblasts, LntA colocalized with heterochromatin nuclear foci that were induced by overexpression of BAHD1 tagged with yellow fluorescent protein (BAHD1-YFP) (8), both in fixed (LntA-V5) (Fig. 1F) and in living cells (LntA tagged with cyan fluorescent protein, LntA-CFP) (fig. S4B).

Because BAHD1 is involved in gene silencing, LntA might control host gene expression. To assess this hypothesis, we performed a transcriptome analysis of colon carcinoma epithelial LoVo cells infected for 24 hours with either lntAV5+ or lntA bacteria (GEO database, GSE26414). The lntAV5+ bacteria specifically up-regulated the expression of a subset of genes, out of which 39 displayed a more than twofold induction (table S4). Of these genes, 83% belonged to the interferon-inducible genes regulon: 28 are known interferon-stimulated genes (ISGs), including three genes (IL29, IL28A, and IL28B) that encode type III interferons (IFN-λ1, -λ2, and -λ3), and four are predicted ISGs. LntA may thus play a role in the IFN-III–mediated immune response. This pathway controls various viral infections, especially in epithelial tissues (913).

We confirmed that WT L. monocytogenes triggered the expression of IFN-λ2 in intestinal LoVo and placental JEG-3 epithelial cells (Fig. 2A and fig. S5A), while type I IFN-β1 was induced a little and type II IFN-γ was undetectable. However, the induction of downstream ISGs was modest (Fig. 2A), except for CCL5, which, like IFN-λ genes, is controlled both by nuclear factor κB (NF-κB) and interferon regulatory factors (IRFs) (14). These data suggested that interferon signaling was down-regulated in infected cells. We wondered whether the host factor BAHD1 could act as a repressor of ISGs, as it does for IGF2 (fig. S5B) (8). Knockdown (depletion) of BAHD1 had no or minor effect on ISG expression in noninfected LoVo cells (Fig. 2B). However, infection of these BAHD1-depleted cells with L. monocytogenes induced the expression of several ISGs, which highlighted that BAHD1 could act as a negative regulator of this pathway after bacteria-triggered signaling (Fig. 2B and fig. S5B).

Fig. 2

The BAHD1 complex represses ISGs in Listeria-infected epithelial cells. (A) RT-QPCR analysis of IFN and ISG expression in response to Listeria infection in LoVo cells infected for 16 hours with WT L. monocytogenes, compared with noninfected cells. (Left) Type I (IFN-β1), II (IFN-γ), and III (IFN-λ2) interferon genes. (Right) Various ISGs. †, below detection limits. (B) Quantification of ISGs mRNA in LoVo cells treated for 72 hours with control small interfering RNA (siRNA), siRNA against BAHD1 or KAP1 and infected for 16 hours (WT) or not (NI). (C) Tandem affinity purification of the BAHD1-associated complex. Solubilized chromatin extracts from HEK293 cells expressing the HPT-BAHD1 fusion or control cells were first purified on anti–protein C affinity matrix, followed by polishing on nickel-Sepharose. Eluted fractions from the first (E1) and second (E2) affinity columns were analyzed by colloidal Coomassie staining (left) or immunoblot (right). I, input; FT, flow-through from first column. Histone H4 was a control for nonspecific binding of chromatin components.

ISG expression is governed by IRF-STAT (signal transducer and activator of transcription) transcriptional activators and by chromatin structure regulators. Except for HP1 proteins, found as BAHD1 partners (8) and repressing ISGs (15), there is no reported link between BAHD1 and STAT signaling. To address whether BAHD1 was associated with other proteins involved in ISG regulation, we purified the BAHD1-associated complex from the chromatin fraction of human embryonic kidney (HEK293) cells expressing His6protein Ctagged BAHD1 (HPT-BAHD1) by tandem affinity chromatography (Fig. 2C and fig. S6). Mass spectrometry analysis of the complex revealed several polypeptides involved in chromatin and transcriptional regulation, including KAP1, HP1γ, and histone deacetylases HDAC1 and 2, as confirmed by immunoblots (Fig. 2C). HDAC1 and 2 directly bind STAT (16), as does the scaffolding protein KAP1, which represses both basal and IFN-I–mediated STAT-driven transcription (17, 18). We investigated whether KAP1 also repressed ISGs during infection with Listeria. KAP1 knockdown induced ISG expression in noninfected cells, and bacterial infection greatly enhanced this induction (Fig. 2B and fig. S5C). The BAHD1-KAP1 corepressor complex thus inhibits ISGs downstream of IFN-III stimulation during L. monocytogenes infection (fig. S7).

Because WT L. monocytogenes does not express lntA in vitro, we further explored the role played by LntA in the IFN-III signaling pathway using lntA constitutive strains. In agreement with the transcriptome data (table S4), ISG expression was higher in LoVo cells infected with lntAV5+ (Fig. 3A) or lntAc+ (fig. S5D) bacteria, compared with noninfected cells or lntA-infected cells. This effect was observed only in epithelial cell lines (fig. S5E). The expression of CCL5 (Fig. 3A) and IFN-λ2 (fig. S5F), which are ISGs themselves, was also increased upon infection with lntA constitutive strains. Thus, LntA can activate ISGs specifically in Listeria-infected epithelial cells, phenocopying BAHD1 depletion (Fig. 2B and fig. S5B). As LntA interacted with BAHD1, we addressed whether it inhibited BAHD1-mediated silencing. Chromatin immunoprecipitation (ChIP) revealed that the recruitment of BAHD1 at the promoter of representative ISGs (IFIT3 and IFITM1) was impaired in lntAV5+-infected cells, compared with lntA-infected cells (Fig. 3B). This correlated with an enrichment of acetylated histone H3 at Lys9 (H3K9Ac) at these genes, consistent with increased transcriptional activity. Thus, by displacing the BAHD1-HDAC complex from ISGs, LntA derepresses these genes in infected cells (for a model, see fig. S7).

Fig. 3

LntA impairs BAHD1-mediated repression of ISGs. (A) LntA induces ISGs. mRNA levels were estimated by RT-QPCR on total RNA from LoVo cells infected for 16 hours with lntA or lntAV5+, compared with noninfected cells. (B) LntA impairs BAHD1 recruitment and increases acetyl-H3 levels at ISGs. ChIP analysis was performed on LoVo cells infected as in (A), with antibodies against BAHD1 or H3K9Ac. Promoter DNA levels were assessed by QPCR and normalized to cognate levels in histone H3-ChIP. (C) Constitutive expression of lntA in Listeria decreases bacterial burden during murine systemic listeriosis and triggers overproduction of IFN-III. Mice were infected intravenously with either ∆lntA or lntAc+ strains. (Left) Colony-forming units (CFUs) per organ were numerated at 72 hours postinfection. (Right) Mouse IFN-λ3 concentration was quantified by enzyme-linked immunosorbent assay (ELISA) in clarified total extracts of infected spleens and livers. (D) BAHD1+/– mice are less sensitive to systemic listeriosis. BAHD1+/+ or isogenic BAHD1+/– mice were infected intravenously with L. monocytogenes EGDe strain. CFUs per organ were numerated at 72 hours postinfection. (C and D) *P < 0.05; **P < 0.01; ***P < 0.001.

Neither lntA expression (fig. S3 and table S3) nor cellular stimulation with recombinant IFN-λ2 (fig. S8) altered bacterial infection in tissue-cultured LoVo cells. We thus assessed the consequences of LntA-BAHD1 interactions on the outcome of infection in vivo. To this end, (i) we infected BALB/c mice with lntA constitutive or lntA-deficient bacteria, and (ii) we generated C57BL/6 BAHD1+/– mice (figs. S9 and S10) and infected them with WT bacteria. We observed a strong decrease in bacterial burden in spleens and livers of BALB/c mice infected with lntAc+ or lntAV5+ relative to ∆lntA or WT bacteria, whereas the IFN-λ3 concentration increased in infected organs (Fig. 3C and fig. S11). Thus, constitutive expression of lntA promotes the IFN-III response and decreases bacterial colonization in vivo. Moreover, in BAHD1+/– mice infected with WT L. monocytogenes, the bacterial burden in organs was reduced compared with that of BAHD1+/+ mice (Fig. 3D). Thus, increasing lntA expression in Listeria had effects similar to impairing BAHD1 expression in the host, i.e., decreasing infection. Furthermore, although controlled secretion of LntA by WT bacteria is beneficial to the pathogen, either its constitutive secretion or its absence is detrimental. We propose that a tight control of lntA expression during infection allows Listeria to fine-tune localized immune responses and to escape antibacterial response (19). Given the tropism of Listeria (20) and IFN-III (913) for epithelia, unraveling the role of LntA in these tissues is now a key issue. Our work identifies the BAHD1 complex as a negative regulator of ISGs in the context of listeriosis and highlights the importance of chromatin remodeling in bacterial infections.

Supporting Online Material

www.sciencemag.org/cgi/content/full/science.1200120/DC1

Materials and Methods

SOM Text

Figs. S1 to S11

Tables S1 to S7

References

References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. Supporting discussion is available on Science Online.
  3. We are very grateful to O. Dussurget for help in virulence assays. We thank E. Gouin for antibodies against LntA; J.-Y. Coppée for microarrays facilities at the IP Genopole; S. Jacquot and M. C. Birling at the Targeted Mutagenesis and Transgenesis department of the Mouse Clinical Institute (MCI/ICS), where the BAHD1+/– mouse line was generated. Work in the Cossart laboratory received financial support from the Pasteur Institute, French National Institute for Agricultural Research (INRA), INSERM, The French National Research Agency (ANR)–European Research Area (ERA)–NET-PathoGenomics (grant SPATELIS), French Ligue Nationale Contre le Cancer (LNCC RS10/75-76 Bierne), and European Research Council (Advanced Grant 233348). P.-J.M. received a Ph.D. fellowship from Région Rhône Alpes. Work in the Cabanes laboratory was supported by Fundação para a Ciência e a Tecnologia (FCT) (PTDC-SAU/MII/65406/2006; Ph.D. fellowship to A.C. SFRH/BD/29314/2006) and ERANET-PathoGenomics (grant SPATELIS). P.C. is an international research scholar of the Howard Hughes Medical Institute. The LntA molecular structure data are deposited at the Worldwide Protein Data Bank (http://www.wwpdb.org), ID no. 2xl4, structure factor file no. r2xl4sf.
View Abstract

Stay Connected to Science

Navigate This Article