Fluorescence-Based Isolation of Bacterial Genes Expressed Within Host Cells

See allHide authors and affiliations

Science  26 Sep 1997:
Vol. 277, Issue 5334, pp. 2007-2011
DOI: 10.1126/science.277.5334.2007


A selection strategy was devised to identify bacterial genes preferentially expressed when a bacterium associates with its host cell. Fourteen Salmonella typhimurium genes, which were under the control of at least four independent regulatory circuits, were identified to be selectively induced in host macrophages. Four genes encode virulence factors, including a component of a type III secretory apparatus. This selection methodology should be generally applicable to the identification of genes from pathogenic organisms that are induced upon association with host cells or tissues.

Many bacterial pathogens survive in professional phagocytes by coordinately regulating the expression of a wide spectrum of genes (1). Because a microbe's ability to survive killing by phagocytes correlates with its ability to cause disease (2, 3), the identification of genes that are preferentially transcribed in the intracellular environment of the host is central to our understanding of how pathogenic organisms mount a successful infection. So far, selections for in vivo–expressed genes (4-6) have been limited to bacterial pathogens with tractable genetic systems, because of the requirement for high frequencies of homologous recombination and extensive strain manipulation before gene selection (4,6).

We have developed a selection methodology, on the basis of differential fluorescence induction (DFI) (7), for the rapid identification of bacterial genes induced upon association with host cells that would work independently of drug susceptibility and nutritional requirements. Green fluorescent protein (GFP) (8) was used as a selectable marker in conjuction with fluorescence-activated cell sorting. Host cells infected with a bacterium bearing a transcriptionally active gfp gene fusion were separated by a fluorescence-activated cell sorter (FACS) and lysed, and the bacteria recovered were then grown under ex vivo conditions. Bacteria were then isolated by FACS on the basis of low to no fluorescence in the absence of host cells. Because all screening and selection steps are performed by a FACS, the biases inherent to manual screening are avoided.

Salmonella typhimurium survives in phagocytes (2) and causes human gastroenteritis and murine typhoid fever (9). We applied DFI to identify S. typhimurium genes that are induced in phagocytic cells (Fig.1). To analyze bacterial clones that were isolated in the selection scheme, we compared the fluorescence intensity of individual bacteria grown in tissue culture media with the fluorescence intensity of the same bacteria clone after release from infected cells (Fig. 1G). Intracellular-dependent induction of thegfp gene fusions isolated was further confirmed by fluorescence microscopy (Fig. 2). Approximately a third to a half of all S. typhimurium that were recovered after one enrichment cycle contained agfp gene fusion with host cell–dependent activity.

Figure 1

Fluorescence-based selection for intracellularly induced genes. (A) A library of S. typhimurium–bearing plasmids with random DNA fragments inserted upstream of a promoterless gfp gene (39) was used to infect a monolayer of RAW 264.7 macrophage-like cells at a multiplicity of infection (MOI) of 5:1. Fluorescent bacteria are shaded. (B) After a 6-hour infection, RAW 264.7 cells were gently scraped from the tissue culture wells, resuspended in cold Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (FCS), and analyzed in a FACStar cytometer. Cells infected with a fluorescent bacterium were readily distinguished from uninfected cells or cells infected with S. typhimurium–bearing unproductivegfp gene fusions. (C) Fluorescent RAW 264.7 cells were sorted, lysed with 1% Triton X, and the recovered bacteria were plated on L-agar. Bacterial colonies were pooled and a sample grown in DMEM (supplemented with 10% FCS) at 37°C in 5% CO2 for 6 hours. (D) FACS analysis of this population yielded bacteria with a wide range of fluorescence intensities. Bacteria with the lowest fluorescence (lower 15%) were sorted as described (7). (E) The bacteria with low fluorescence in the absenceof host cells were amplified on L agar and used to reinfect RAW 264.7 cells at an MOI of 0.5 to 1 for 6 hours. (F) Flow cytometric analysis showed enrichment for fluorescent infected cells as compared to infection with the initial library. The fluorescent cells were collected and the bacteria within recovered. (G) Independent clones exhibited varied levels of intracellular-dependent gene induction. Histograms show three different strains in which the relative fluorescence intensity of bacteria grown under ex vivo conditions (DMEM + 10% FCS) (solid line) and fluorescence of bacteria released from infected cells after 6 hours (shaded) were compared (7). Analysis and quantitation of fluorescence were performed with CellQuest software (Beckton Dickinson).

Figure 2

Intracellular-dependent bacterial gene expression. Salmonella typhimurium bearing anssaH::gfp fusion (mig-10) were used to infect RAW 264.7 cells for 4 hours. (A andC) Differential interference contrast (DIC) image showing the relative topology of bacterial association with the host cell. External bacteria were immunostained with a polyclonal antisera toSalmonella followed by a second stage phycoerythrin conjugate (red). (B) Only internalized bacteria show active synthesis of the GFP tag (green). Fluorescence images were acquired and processed on an Applied Precision Deltavision Deconvolution System. (C) The fluorescence and DIC images were superimposed with Adobe Photoshop software.

Fourteen promoters with intracellular-dependent activity were identified (Table 1). The genes downstream of these promoters were isolated by recombinational cloning (10), sequenced, and characterized. Eight of these macrophage-inducible genes (migs) have either been described or have close homologs with known functions in other bacterial species. Two migs encode known virulence factors. The himAgene (mig-23) encodes subunit A of integration host factor (IHF), which is involved in DNA replication, recombination, and gene regulation (11), and is preferentially expressed in vivo (4). The mig-10 gene product is homologous to a family of bacterial virulence proteins (YscF, MxiH, and PrgI) that are essential components of virulence-associated type III secretion systems present in several bacterial pathogens (12-14). Sequence analysis of the loci associated with mig-10 indicated that this open reading frame (ORF) is present within theSalmonella pathogenicity island 2 (SPI-2), which is necessary for S. typhimurium survival in BALB/c mice (15). For consistency, mig-10 will be referred to as ssaH (16). The mig-1 gene encompasses the intergenic region between the gene encoding the galactose repressor (galR) and a close homolog of theEscherichia coli aas gene that encodes acyl-acylglycerolphosphoethanolamine acyltransferase, which is involved in phospholipid recycling and possibly membrane repair (7,17). mig-2 (pagA/ugd) encodes Ugd, which is necessary for growth in a low-magnesium environment, is homologous to a Streptococcus pneumoneae gene product involved in capsule biosynthesis (18), and is transcriptionally active inside macrophages (7,19). The mig-4 gene product is a homolog of theE. coli periplasmic phosphate transport protein PhoS (20), the induction of which may represent S. typhimurium's response to phosphate deprivation within the host cell. The mig-26 gene encodes the exclusion protein (TraT) of the S. typhimurium virulence plasmid, which is partially responsible for the serum resistance phenotype of S. typhimurium (21). The mig-29 gene product is a homolog of the E. coli heat shock–inducible proteasome component Hslu (22) and represents an example of a stress response protein that is synthesized during cell infection.

Table 1

Characterization of macrophage-inducible genes (mig). DNA fragments (0.2 to 1.2 kb) with intracellular-dependent activity trapped by differential fluorescence enrichments (Fig. 1) were sequenced and compared with the available DNA sequence databases at the National Center for Biotechnology Information (NCBI). Putative functions were assigned either when a particular gene had been previously described in S. typhimurium or when the function of a close homolog (>50% amino acid identity) had already been determined. For promoter regions with no significant homology to previously described genes or with homology to genes with unknown function, flanking DNA was isolated by recombinational cloning (10) and sequenced. Five ORFs downstream of macrophage-inducible promoters were inactivated either by insertion of a kanamycin resistance gene flanked by transcriptional terminators (ΩKn) or by insertion of the suicide vector pGP704 (Ampr) into coding regions (27). ND, not determined.

View this table:

Six of the mig gene products showed either no significant homology to previously identified bacterial genes or homology to gene products of unknown function. The mig-3gene is present within an ORF with homology to phage tail-fiber assembly proteins and likely represents a promoter within an integrated phage. Although virulence determinants may be acquired by horizontal gene transfer (14), we have no evidence that any virulence genes were transduced within these phage sequences. Themig-30::gfp fusion is present at the 3′ end of a gene encoding a homolog of an E. coli putative transmembrane protein (YjsH) (23). However, the polarity of transcription is in the opposite orientation to that of yjsH and would therefore generate an antisense transcript that could down-regulate this protein's synthesis during intracellular infection (24). The mig-5, mig-7,mig-13, and mig-14 genes encode either small, highly hydrophobic proteins or contain motifs characteristic of bacterial lipoprotein-specific signal sequences. The mig-5gene maps to the S. typhimurium virulence plasmid and is located ∼2 kb upstream of the spv virulence operon, which is required by Salmonella sp. to cause systemic disease (25, 26). The mig-14 gene product has an internal stretch of amino acids that is significantly homologous (30% amino acid identity) to members of the LysR family of transcriptional activators. This observation suggests a possible role for mig-14 in the regulation of bacterial gene expression in the intracellular environment.

To establish the role in virulence of these macrophage-inducible proteins, we disrupted five of the unknown mig genes by insertion of either an ΩKn (Kanr) element or plasmid pGP704 (Ampr) into each ORF (27). These mutants were tested in competition assays against the wild-type S. typhimurium for the ability to colonize the spleens of BALB/c mice and for survival in the macrophage-like cell line RAW 264.7. Insertions in mig-5 and mig-14 resulted in a decrease in spleen colonization for each mutant strain (Table 1) (28). However, none of the mutant strains tested showed any growth defect in RAW 264.7 cells. These results, coupled with the roles of the previously reported virulence determinants himA andssaH, suggest that genes induced in the intracellular environment, although not essential for survival in tissue culture models of infection, are important for bacterial survival within the host.

A key regulator of S. typhimurium gene expression inside host cells is the two-component regulatory system PhoP/PhoQ (29). To determine how many migs were under the control of PhoP/PhoQ, we conjugated plasmids bearingmig::gfp fusions into a S. typhimuriumstrain bearing a phoP::Tn10 insertion (30). Eight of the 14 migs required PhoP/PhoQ for induction in macrophages (Table 1), including two migs (mig-5 and mig-14) important for S. typhimurium virulence in BALB/c mice. Thus, the PhoP/PhoQ two-component system is a major regulator of macrophage-inducible genes important in Salmonella sp. virulence as previously noted (19, 29, 31, 32). BecauseS. typhimurium can invade and replicate in a wide variety of cell types (9), we tested mig induction in RAW 264.7 macrophages, activated RAW 264.7 macrophages (10 U of interferon-γ and 1.5 ng/ml lipopolysaccharide), the epithelial cell line Hep-2 , HeLa fibroblasts, and FSDC dendritic cells (33). The level of intracellularmig::gfp expression was indistinguishable among these different cell types. From these results, we conclude thatmigs respond to common intracellular cues rather than those unique to macrophages. Our finding that PhoP/PhoQ-dependent genes are not macrophage specific is consistent with a previous report of a PhoP/PhoQ-dependent gene that is induced in polarized epithelial cells (34). The macrophage-inducible virulence factor (ssaH) present within SPI-2 did not require PhoP/PhoQ for intracellular activity. ssaH is part of an operon that includes other essential components of the type III secretion apparatus such as ssaJ (16). This suggests that the regulation of the synthesis of components of this secretory apparatus may occur through a different mechanism, possibly the putative two-component regulatory system (SsrA/SsrB) that is also present within SPI-2 (15, 35). Therefore, we tested anssaH::gfp fusion in S. typhimuriumbackgrounds bearing a miniTn5Km insertion in either ssrA orssrB (15). Indeed, disruption of either the sensor or regulator components resulted in a loss ofssaH::gfp induction within RAW 264.7 cells.

Type III secretion systems in pathogenic bacteria are also called contact-dependent secretory systems because of their role in the translocation of bacterial proteins across the eukaryotic plasma membrane upon cell contact (36). Although the virulence defects associated with mutations in components of the secretion apparatus in SPI-2 have been described, little is known about the molecular function and targets of SPI-2 gene products (15,16, 35). On the basis of our results, it seems plausible that a secretory apparatus assembled within the host cells may function in translocating bacterial proteins across the vacuolar membrane.

To establish whether mig::gfp fusions were expressed within infected animal cells, we injected BALB/c mice intraperitoneally with 105 S. typhimuriumbearing different mig::gfp fusions. Three days after infection, the spleens and livers were homogenized and analyzed by fluorescence microscopy and flow cytometry (Fig3). All mig::gfpfusions were active within infected spleenocytes and hepatocytes. In heavily colonized animals, as much as 5% of all spleenocytes were infected with one or more fluorescent organisms. The ability to track infected cells in an animal host should extend the utility of this selection technology to enable the identification of host cell subsets that are preferentially targeted by a pathogen.

Figure 3

mig gene expression within infected spleenocytes. FACS analysis of spleen homogenates from (A)S. typhimurium– or (B) S. typhimurium (mig-5::gfp)–infected BALB/c mice. (C) A transmitted light and fluorescence image of sorted fluorescent cells from (B) as imaged through a long-pass fluorescein-rhodamine filter (Chroma Technologies). BALB/c mice were injected intraperitoneally with 105S. typhimurium bearing various mig-gfp fusions. After 3 days the spleens were removed and gently homogenized in phosphate-buffered saline supplemented with 10% FCS. The cell suspension was analyzed in a FACScalibur cytometer (Beckton Dickinson) and in a Nikon epifluorescence microscope.

Although many intracellular-induced bacterial genes (for example,spv and those contained in SPI-2) do not seem to have a clear role in survival in tissue culture models of infection, they are essential for virulence in animals (15, 25). This suggests that intracellular induction of key bacterial virulence determinants is required for systemic infection. By tracking the expression of these genes in infected animals, we might be able to further define when and where virulence factors are expressed. New generations of spectrally distinct fluorescent protein reporters (37) may soon permit the simultaneous monitoring of multiple bacterial genes in response to the host environment.

The selection methodology described here represents a tool with which to dissect the genetic basis of the interactions between two or more organisms. The only genetic requirements are that the test organism be able to maintain an episomal element and express a functional gfp. The technology should be widely applicable to the study of gene induction by bacterial, fungal, viral, and protozoan pathogens or to the study of symbiosis and gene regulation in complex microbial communities.

  • * To whom correspondence should be addressed. E-mail: valdivia{at}


View Abstract

Navigate This Article