Enhanced Synthesis of Internalin A in aro Mutants of Listeria monocytogenes Indicates Posttranscriptional Control of the inlAB mRNA

Abstract

Listeria monocytogenes mutants with deletions in aroA, aroB, or aroE exhibited strong posttranscriptional upregulation of internalin A (InlA) and InlB synthesis, which resulted in a more-than-10-fold increase in InlA-mediated internalization by epithelial Caco-2 cells and a 4-fold increase in InlB-mediated internalization by microvascular endothelial cells (human brain microvascular endothelial cells) compared to the wild-type strain. The increase in InlA and InlB production was not due to enhanced PrfA- and/or sigma factor B (SigB)-dependent inlAB transcription but was caused by enhanced translation of the inlAB transcripts in the aro mutants. All inlA(B) transcripts had a 396-nucleotide upstream 5′ untranslated region (UTR). Different deletions introduced into this UTR led to significant reductions in InlA and InlB synthesis; enhanced translation of all of the truncated transcripts in the aro mutants was, however, still observed. Thus, translation of the inlAB transcripts was subject to two modes of posttranscriptional control, one mediated by the UTR structure and the other mediated by the aro mutation. The latter mode of control seemed to be related to the predominantly anaerobic metabolism of the aro mutants.

Internalization of the facultative intracellular organism Listeria monocytogenes, a gram-positive bacterium responsible for severe food-borne infections in humans and animals, by nonphagocytic mammalian cells is mediated predominantly by two internalins, internalin A (InlA) and InlB (9). Both of these internalins belong to a large listerial protein family which in L. monocytogenes includes at least seven more internalins (12, 13, 43), all of which are typical members of the leucine-rich repeat superfamily that includes many other eukaryotic and prokaryotic proteins (26, 27). With the exception of InlC and InlB, the internalins have in their C-terminal portions a conserved region with an LPXTG motif, which anchors these proteins to the peptidoglycan of the bacterial cell surface (36). InlB, in contrast, is anchored via a GW repeat domain to the lipoteichoic acid of the bacterial cell wall (24), and InlC is a small secreted protein without any anchor sequences (13).

Besides InlA and InlB, only InlC, InlC2, InlD, InlE, InlF, InlG, and InlH have been studied so far to some extent (12, 13, 43). In contrast to InlA and InlB, none of these internalins seem to be able to induce phagocytosis by nonphagocytic mammalian cells (3). However, InlC may support InlA-mediated internalization into epithelial cells (3). As deletion of inlC leads to a significant reduction in L. monocytogenes virulence in the mouse model and, in particular, to significantly decreased bacterial loads in the livers of orally infected mice (13) and as an L. monocytogenes mutant with a deletion in the inlGHE gene cluster exhibits an increase in the 50% lethal dose after intravenous infection of mice and reduced replication in the livers of orally infected mice (43), it has been suggested that these internalins are involved in virulence, which has been clearly shown for InlA and InlB. The functions of all other internalins are basically unknown.

The inlA and inlB genes form an operon which is controlled by a complex regulatory region that is located upstream of inlA and contains several promoters and a binding site for the transcriptional activator PrfA (5, 12, 33, 48). Thus, expression of inlAB is either PrfA dependent or PrfA independent, possibly in response to the environmental conditions (5, 12, 33, 48). Transcription of inlC is more strictly dependent on PrfA than transcription of inlAB is and seems to occur under extracellular conditions (7, 13) and, in contrast to inlAB, especially under intracellular conditions (7, 13). The other inl genes do not seem to be regulated by PrfA.

Similar to inlAB and inlC, most of the other known virulence genes of L. monocytogenes are transcriptionally activated by PrfA, which appears to be the central regulator of virulence in L. monocytogenes.

However, recent investigations have shown that the synthesis of at least some listerial virulence factors, including PrfA itself and ActA, is controlled not only at the transcriptional level but also at the posttranscriptional level (23, 54). For the prfA and actA transcripts it was shown that the upstream 5′ untranslated regions (UTRs) can form secondary structures that affect the translation efficiency in response to external signals. For the prfA transcript elevated temperature was shown to enhance translation (23), while no signal for enhanced translation of the actA mRNA is known yet (54).

While studying expression of virulence genes in a virulence-attenuated aroA mutant of L. monocytogenes, we observed strongly increased production of InlA (and to a lesser extent InlB) which was mainly caused by enhanced translation of the inlAB transcripts.

Here we report this unexpected finding, which may be of major importance for in vivo infection by L. monocytogenes.

MATERIALS AND METHODS

Bacterial strains and plasmids.

The bacterial strains and plasmids used in this work are listed in Tables 1 and 2. Escherichia coli DH10b was used as the host for all DNA manipulations.

TABLE 1.

L. monocytogenes strains

Strain	Genetic properties	Reference
L. monocytogenes EGDe	Wild type	16
EGDe ΔaroA	Deletion of aroA	50
EGDe ΔaroB	Deletion of aroB	51
EGDe ΔaroA/B	Deletion of aroA and aroB	51
EGDe ΔaroE	Deletion of aroE	51
EGDe ΔaroA/E	Deletion of aroA and aroE	51
EGDe ΔpheA	Deletion of pheA	51
EGDe ΔtyrA	Deletion of tyrA	51
EGDe ΔtrpEG	Deletion of trpEG	51
EGDe ΔtrpEG tyrA	Deletion of trpEG and tyrA	51
EGDe ΔpheA trpEG tyrA	Deletion of pheA, trpEG, and tyrA	51
WL-112	Deletion of inlA and inlB	18
UTRdel15/45	Deletion in the UTR of inlA (15 to 45 bp in front of start codon)	This study
UTRdel15/45 ΔaroA	Deletion in aroA and deletion in the UTR of inlA (15 to 45 bp in front of start codon)	This study
UTRdel169/285	Deletion in the UTR of inlA (169 to 285 bp in front of start codon)	This study
UTRdel169/285 ΔaroA	Deletion in aroA and deletion in the UTR of inlA (169 to 285 bp in front of start codon)	This study
UTRdel50/360	Deletion in the UTR of inlA (50 to 360 bp in front of start codon)	This study
UTRdel50/360 ΔaroA	Deletion in aroA and deletion in the UTR of inlA (50 to 360 bp in front of start codon)	This study
UTRdel307/360	Deletion in the UTR of inlA (307 to 360 bp in front of start codon)	This study
UTRdel307/360 ΔaroA	Deletion in aroA and deletion in the UTR of inlA (307 to 360 bp in front of start codon)	This study
UTRdel255/306	Deletion in the UTR of inlA (255 to 306 bp in front of start codon)	This study
UTRdel255/306 ΔaroA	Deletion in aroA and deletion in the UTR of inlA (255 to 306 bp in front of start codon)	This study
UTRdel15/389	Deletion in the UTR of inlA (15 to 389 bp in front of start codon)	This study
UTRdel15/389 ΔaroA	Deletion in aroA and deletion in the UTR of inlA (15 to 389 bp in front of start codon)	This study

TABLE 2.

Plasmids

Plasmid	Description	Source or reference
pLSV101	Mutagenesis plasmid	T. Fuchs
pLSV1 ΔaroA	Used for deletion of aroA	50
pLSV101 ΔaroB	Used for deletion of aroB	51
pLSV101 ΔaroE	Used for deletion of aroE	51
pLSV101 ΔpheA	Used for deletion of pheA	51
pLSV101 ΔtrpEG	Used for deletion of trpEG	51
pLSV101 ΔtyrA	Used for deletion of tyrA	51
pLSV101 ΔUTRdel50/360	Used for deletion of bp 50 to 360 from UTR of inlA	This study
pLSV101 ΔUTRdel15/45	Used for deletion of bp 15 to 45 from UTR of inlA	This study
pLSV101 ΔUTRdel69/285	Used for deletion of bp 169 to 285 from UTR of inlA	This study
pLSV101 ΔUTRdel307/360	Used for deletion of bp 307 to 360 from UTR of inlA	This study
pLSV101 ΔUTRdel255/306	Used for deletion of bp 255 to 306 from UTR of inlA	This study
pLSV101 ΔUTRdel15/389	Used for deletion of bp 15 to 389 from UTR of inlA	This study

Bacteria were cultivated either in brain heart infusion (BHI) or in BHI containing 2% glucose. Vitamin K₂ (Sigma, Crailsheim, Germany) was dissolved in ethanol (concentration in stock solution, 25 mg/ml) and added to the medium to a final concentration of 60 mg/liter. For anoxic BHI cultures, medium containing 2% glucose was degassed at 60°C for 45 to 60 min under a vacuum and overlaid with mineral oil.

L. monocytogenes strains were grown to the late logarithmic growth phase (optical density at 600 nm [OD₆₀₀] for L. monocytogenes EGDe or supplemented aro mutants, 1.0; OD₆₀₀ for the aroA mutant without a shikimate or vitamin K₂ supplement, 0.8; OD₆₀₀ for the aroE mutant grown in the absence of vitamin K₂, 0.5) at 37°C. For infection bacteria were washed with phosphate-buffered saline (pH 7.4), resuspended in phosphate-buffered saline-20% (vol/vol) glycerol, and stored at −80°C.

Construction of Listeria mutant strains.

All deletion mutagenesis procedures were performed by using L. monocytogenes EGDe as the parental strain. For these procedures we used a homologous recombination technique (55) and constructs derived from the mutagenesis vector pLSV101 (derived from pLSV1; kindly provided by T. Fuchs).

For deletion of sigB, a 328-bp fragment (primers sigB5′-1 [5′-GTTCGTGGATCCGATAACGGCACAAGCTTCGA-3′] and sigB5′-2 [5′-CTGCAACGCCTCTCCCGGGTTTATCAGGTTGAGATACTTTTG-3′]) which was localized upstream of the deletion locus and a 162-bp fragment (primers sigB3′-3 [5′-GGGAATCCCGGGTGAGAGCGATTCGAG-3′] and sigB3′-4 [5′-ACATAGGAAGCGAATTCGGCA-3′]) downstream of the deletion locus were PCR amplified. Both fragments were used in a recombinant PCR with primers sigB5′-1 and sigB3′-4, which resulted in a 474-bp recombination fragment (ΔsigB) that was then cloned into pLSV101 by BamHI restriction, resulting in pLSV101 ΔsigB. The plasmid pLSV101 ΔsigB DNA was transformed into L. monocytogenes EGDe (for construction of a sigB mutant) or the prfA mutant (for construction of a prfA sigB mutant). After cultivation at 42°C with 5 μg of erythromycin per ml, clones with a chromosomally integrated recombination plasmid could be selected. The resulting clones were then cultivated at 30°C without erythromycin to obtain the sigB chromosomal deletion mutants.

Deletion mutagenesis of the UTR of inlA was performed by using the same technique. In all cases the 5′ and 3′ regions were amplified and ligated by a recombinant PCR (5′ and 3′ regions that overlapped by 32 bp were introduced by using the primers shown in Table 3) performed with primers inlAUTR5′-F (5′-GTAAACGGATCCACTGTCGGACCAACGAGC-3′) and inlAUTR3′-R (5′-CCTAATGGATCCGCCTGAAGCGTTGTAACT-3′), which resulted in recombination fragments that were cloned into pLSV101 by using BamHI restriction.

TABLE 3.

Primers used for construction of inlA UTR deletion mutants

Primer	Sequence (5′-3′)	Description
inlAUTR5′-F	GTAAACGGATCCACTGTCGGACCAACGAGC	External primer with BamHI site
inlAUTR3′-R	CCTAATGGATCCGCCTGAAGCGTTGTAACT	External primer with BamHI site
UTRdel15/45-F	CACTATATACACTCCTTTGTATAGGATTTTCTCCTGC	Internal primer to construct UTRdel15/45
UTRdel15/45-R	AGAAAATCCTATACAAAAGGAGTGTATATAGTGAGAAA	Internal primer to construct UTRdel15/45
UTRdel169/285-F	GATTTGCATGATTGAATTCCTATTAAGCTTCG	Internal primer to construct UTRdel169/285
UTRdel169/285-R	TTTTAAAAGGTGGAATGGATATCACTAAACGGCTCCG	Internal primer to construct UTRdel169/285
UTRdel150/360-F	TCAGGTTTCGTTGTACACTTCTTCTATATCATC	Internal primer to construct UTRdel50/360
UTRdel150/360-R	GATATAGAAGAAGTGTACAACGAAACCTGATAT	Internal primer to construct UTRdel50/360
UTRdel307/360-F	CCACCTTTTAAAATGCACTTCTTCTATATCATC	Internal primer to construct UTRdel307/360
UTRdel307/360-R	GATATAGAAGAAGTGGACCATTTTAAAAGGTGG	Internal primer to construct UTRdel307/360
UTRdel50/254-F	TCAGGTTTCGTTGTAGCAAATCATTTTATGTTG	Internal primer to construct UTRdel50/254
UTRdel50/254-R	CATAAAATGATTTGCTACAACGAAACCTGATAT	Internal primer to construct UTRdel50/254
UTRdel255/306-F	AATTCAATCATCTATTATTGACGAATTCAAGC	Internal primer to construct UTRdel255/306
UTRdel255/306-R	CGTCAATAATAGATGATTGAATTCCTATTAAGC	Internal primer to construct UTRdel255/306
UTRdel15/389-F	TTATTATTTAATGGGCTAAAGGAGTGTATATAGTGAGAAA	Internal primer to construct UTRdel15/389
UTRdel15/389-R	ACTATATACACTCCTTTAGCCCATTAAATAATAATAACT	Internal primer to construct UTRdel15/389

The aroA deletion in inlA UTR mutants was introduced as described previously (50) by using pLSV1 ΔaroA.

All deletion mutants were confirmed by PCR analysis and sequencing of the corresponding sections of the chromosome. The prfA and sigB mutants also were confirmed by Western blot analysis.

Cell culture and infection experiments.

Caco-2 (human colon adenocarcinoma) cells were cultured in RPMI 1640 medium supplemented with 2 mM l-glutamine (Gibco) and 10% fetal calf serum (Biochrom, Berlin, Germany); human brain microvascular endothelial cells (HBMEC) were grown in complete HBMEC medium (17). All cell lines were maintained at 37°C in a 5% CO₂ atmosphere.

For infection, cells were seeded into 24-well plates 1 day prior to infection. After the cells were washed with RPMI 1640 medium containing 2 mM l-glutamine, they were infected with 10 bacteria per cell for 1 h. Then the cells were washed with RPMI 1640 medium containing 2 mM l-glutamine and cultivated with medium containing gentamicin (100 μg/ml). Viable bacterial counts for intracellular bacteria were then determined by plating serial dilutions of mechanically lysed cell suspensions on BHI agar.

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis and immunoblotting.

Aliquots (10 ml) of logarithmic-phase Listeria cultures were centrifuged, and the pellets were resuspended in 100 μl of 4× Laemmli buffer per OD₆₀₀ unit. After incubation at 98°C for 20 min, sodium dodecyl sulfate—10% polyacrylamide electrophoresis was performed by the method of Laemmli (30). Immunoblotting was performed by a semidry method by using Hybond-ECL nitrocellulose membranes (Amersham Biosciences, Freiburg, Germany). After incubation with horseradish peroxidase-conjugated secondary antibodies (Dianova, Hamburg, Germany), a chemiluminescence-based immunoblot assay (ECL; Amersham Biosciences) was performed according to the instructions provided by the manufacturer.

The antibodies used in this work were a rabbit polyclonal antiserum raised against ActA (40), a rabbit polyclonal antiserum directed against listeriolysin O (provided by J. Kreft), a mouse polyclonal antiserum specific for listerial PlcB (41), a polyclonal guinea pig antibody specific for PrfA (4), a polyclonal antiserum directed against InlC (3), and affinity-purified rabbit polyclonal antibodies directed against InlA and InlB (38) (kindly provided by J. Wehland, Braunschweig, Germany). LpeA polyclonal antibody was provided by P. Kohlenbrander (44), and a LAP monoclonal antibody was provided by A. Bhunia (22).

Quantification of protein was performed by using a NightOWL LB981 low-light imager (Berthold Technologies, Bad Wildbad, Germany) equipped with WinLight³² software.

Determination of hemolytic and lecithinase activities.

The hemolytic and lecithinase activities of L. monocytogenes were determined by using supernatants of late-logarithmic-phase cultures as described previously (45).

Primer extension studies.

L. monocytogenes RNA was isolated from bacteria grown to the logarithmic phase in BHI by using an RNeasy mini kit (QIAGEN). Residual DNA was removed on a column with QIAGEN RNase-free DNase (QIAGEN). A primer extension analysis was performed essentially as described previously (2). Briefly, 0.5 pmol of γ-³²P-end-labeled oligonucleotide (5′-GTTGATATCACACGTGTCATTCCACC-3′; 270 bp upstream of the inlA coding sequence) was ethanol precipitated with 30 μg of RNA. The nucleic acids were resuspended in 10 μl of reverse transcriptase buffer (Roche Applied Sciences) containing deoxynucleoside triphosphates at a concentration of 0.4 mM, and the samples were heated to 90°C for 1 min. Twenty units of avian myeloblastosis virus reverse transcriptase (Roche Applied Sciences) was added, and the reaction mixtures were incubated for 45 min at 45°C. After RNase A treatment of the cDNAs, 5 μl of a stop solution (sequencing dye; Amersham Biosciences) was added, and the samples were subjected to gel electrophoresis on a 6% urea-polyacrylamide gel.

RT-PCR.

A real-time quantitative PCR (RT-PCR) was conducted with independently isolated total RNA from L. monocytogenes cells grown to the mid-log phase, which was extracted with an RNeasy mini kit (QIAGEN) used according to the manufacturer's protocol. Residual DNA on the column was removed with QIAGEN RNase-free DNase (QIAGEN). Before the RT-PCR was performed, the absence of DNA from RNA samples was verified by PCR amplification of the genes to be assayed with 1 μg of RNA as the template. cDNA synthesis was performed by using 5 μg of total RNA which was heat denatured together with specific (reverse) primers in an 18-μl (final volume) mixture for 5 min at 70°C. After incubation on ice for 5 min, a master mixture containing the following reagents was added: 20 mM dATP, 20 mM dCTP, 20 mM dGTP, 20 mM dTTP, dithiothreitol (40 pmol), RNase inhibitor (RNase out; 40 U; Invitrogen, Life Technologies), 1 μl of Superscript II reverse transcriptase (Invitrogen, Life Technologies), 5.8 μl of diethyl pyrocarbonate water, and reaction buffer (Invitrogen, Life Technologies) to a final concentration of 1×. The reaction mixture was incubated for 10 min at room temperature and then for 2 h at 42°C and for 15 min at 70°C.

Each RT-PCR was performed in a 25-μl (final volume) mixture. The protocol and the cycling conditions were those recommended in the manufacturer's protocol for a qPCRCore kit for SYBR Green-I (Eurogentec). The primers used for amplification were 5′-GCGGATGAAGAGGATAATTACG-3′ (forward) and 5′-GGAATCCATAGATGGACCGTT-3′ (reverse) for rpoB, 5′-AGCAGATGATGCTTCACCACA-3′ (forward) and 5′-CCCTGCACTTTTATCAACAATC-3′ (reverse) for actA, 5′-ATGCAATTTCGAGCCTAACCT-3′ (forward) and 5′-TTATTGTCTTGATTAGTCATAC-3′ (reverse) for hly, 5′-AAGCGCTAGGATGGAGCACA-3′ (forward) and 5′-CAACTGCAATAATCGAGCAAAG-3′ (reverse) for uhpt, 5′-GTATGGTTGAAAAGTATACTA-3′ (forward) and 5′-TGTTACATTCGTTTTTCCTAA-3′ (reverse) for inlA, 5′-TTTCTATCAGCCAGTCACTATTGGA-3′ (forward) and 5′-CGCGTCCCTGCTTCTACTTTTGT-3′ (reverse) for inlB, 5′-CAAATACAGGTGGACTAACTA-3′ (forward) and 5′-GATATCCATCTTCCATCTGGGT-3′ (reverse) for inlC, 5′-TTGCTCCAGAGGCCACTACAT-3′ (forward) and 5′-GATACCACTTTCCCAAACGAAGTG-3′ (reverse) for mpl, 5′-AATGCATCACTTTCAGGTGTATTAGA-3′ (forward) and 5′-GTTGATTAGTGGTTGGATCCGATAA-3′ (reverse) for plcA, 5′-TCAAGGAATATATGATGCGGATCAT-3′ (forward) and 5′-CTTTGCTCCTGTTATTTTCGCATTA-3′ (reverse) for plcB, and 5′-AGCCAAGCTTCCCGTTAATCGAAA-3′ (forward) and 5′-CAGGCTACCGCATACGTTATCAAA-3′ (reverse) for prfA. For quantification of RT-PCRs, a standard curve was established by using serial dilutions of an rpoB PCR fragment as a template in an RT-PCR, which served as an external standard. All results were normalized by establishing a normalization factor as follows. For each strain in all the growth media tested, rpoB expression was determined by using rpoB-specific primers. By setting the rpoB ratio equal to 1, a normalization factor for all strain combinations in all media tested was calculated and used to normalize all data sets. The specificity of all amplicons was confirmed by using melting curves. Final mean values and standard deviations were calculated based on the ratio for three independent RT-PCRs for each strain.

RNA structure analysis.

Potential secondary RNA structures of the 5′ UTR of inlA were obtained with the mfold algorithm described by Zuker (57). Secondary structure alignment was performed by using the RNA distance algorithm, which measured dissimilarity (14, 21, 46, 47).

RESULTS

Increased invasion rates of aroA, aroB, and aroE mutants in mammalian cells are due to increased production of InlA and InlB.

We recently constructed a set of aro mutants (aroA, aroB, aroE, aroA/B, and aroA/E mutants) as carrier strains for L. monocytogenes-based live vaccines (51). Using these mutants in infection assays with various cell cultures, we noticed 10- to 20-fold-greater rates of internalization of these mutants in Caco-2 cells, in which infection was mainly InlA dependent (31), and a 4-fold-greater rate in HBMEC, in which infection was strictly InlB dependent (17) (Fig. 1). In contrast, the efficiency of cytosolic replication and cell-to-cell spread was highly reduced in these aro mutants, and overall the virulence was therefore strongly attenuated (51).

FIG. 1.

Cells were infected as described in the text, and the relative uptake efficiencies of L. monocytogenes wild-type strain EGDe and aro mutants by epithelial Caco-2 cells (A) and endothelial HBMEC (B) were determined.

To examine the molecular basis of this unexpected observation, we determined the production of most proteins which reportedly affect invasion, including InlA (15), InlB (11, 17), InlC (3), ActA (52), p60 (29, 39), LAP (22), and LpeA (44) (Table 4).

TABLE 4.

Expression of virulence and putative adhesion genes in the aroA mutant compared to wild-type L. monocytogenes

Gene	Protein	Amt of transcripta	Amt of proteinb	Functional test
prfA	PrfA	1.0 ± 0.2	1.3 ± 0.1	NDc
plcA	PlcA	0.9 ± 0.1	ND	ND
hly	Listeriolysin O	0.8 ± 0.1	1.3 ± 0.0	1.9 ± 0.5d
mpl	Mpl	1.0 ± 0.5	ND	ND
actA	ActA	1.2 ± 0.3	1.8 ± 0.6	0.5 ± 0.1e
plcB	PlcB	1.5 ± 0.4	2.9 ± 0.3	2.1 ± 0.1f
uhpT	UhpT	1.5 ± 0.4	ND	ND
inlA	InlA	1.7 ± 0.3	10.2 ± 2.2	15.5 ± 6.4g
inlB	InlB	1.7 ± 0.3	3.7 ± 0.6	4.5 ± 1.0h
inlC	InlC	1.3 ± 0.2	2.3 ± 0.5	ND
iap	p60	ND	1.1 ± 0.3	ND
p104	LAP	ND	0.6 ± 0.1	ND
lpeA	LpeA	ND	1.7 ± 0.1	ND

^aThe amounts of transcripts of virulence genes in the aroA mutant and the wild-type strain L. monocytogenes EGDe were measured and compared by reverse transcription and real-time PCR. The data are means ± standard deviations for three independent experiments.

^bThe amounts of protein were compared after Western blotting by using a low-light imager for quantification. The data are averages ± standard deviations for two to five Western blots.

^cND, not determined.

^dListeriolysin O activity was measured by a hemolysin activity assay by using supernatants from late-logarithmic-phase cultures.

^ePlaque assays were performed as described previously (51). Smaller plaques after aroA mutant infection seemed to be due to the longer generation time for this mutant compared to wild-type L. monocytogenes.

^fPlcB activity was measured by a lecithinase assay by using supernatants of late-logarithmic-phase cultures.

^gCaco-2 invasion rates were used as a functional test for InlA expression.

^hHBMEC invason rates were used as a functional test for InlB expression.

Figure 2 shows that there was an almost 10-fold increase in the production of InlA and there was a 4-fold increase in the production of InlB, although transcription was increased less than 2.5-fold, while the production of all other invasion-enhancing proteins was not significantly increased (Table 4). If synthesis of all other PrfA-dependent virulence factors tested, including PrfA itself, was upregulated at all, it was only marginally upregulated (up to about 1.5-fold) at the transcriptional level, and at the protein level synthesis was upregulated 1.3- to 2.9-fold in the aroA mutant compared to the wild-type strain (Table 4). Thus, the increased production of InlA and InlB corresponded well to the observed enhanced internalization in Caco-2 and HBMEC cells, suggesting that there is a direct correlation between these two events. Furthermore, the data indicate that InlA production and, to a lesser extent, InlB production were enhanced at the posttranscriptional level.

FIG. 2.

Western blot analyses of InlA (A) and InlB (B) produced in the wild-type strain and aro mutants of L. monocytogenes EGDe. The upper band represents the intact InlA protein, while the faster-migrating proteins are degradation products of InlA. The arrows indicate the position of the 75-kDa marker (32).

Both PrfA- and SigB-dependent inlAB transcripts showed enhanced translation in the aroA mutant.

As reported previously, the inlAB operon is transcribed by several promoters (12, 25, 33); the activity of promoter PinlA2 depends on PrfA (33), while the activity of promoter PinlA3 depends on SigB (12, 25). Primer extension experiments carried out with RNA from BHI-grown L. monocytogenes wild-type strain EGDe, a prfA mutant, a sigB mutant, and a prfA sigB double mutant verified the dependence of PinlA2 on PrfA and the dependence of PinlA3 on SigB (Fig. 3A and B). The P1 transcript seems to be coupled to P3 transcription or may represent a degradation product of the SigB-dependent P3 transcript since no appropriate SigB-dependent −10 box which fits the identified start nucleotide (C) of P1 has been recognized (Fig. 3B). This is in line with the observation that the prfA sigB mutant produced very little inlAB transcript and was hardly invasive in Caco-2 cells (Fig. 3A and D). Interestingly, an aroA mutant showed preferential transcription from the PinlA2 promoter, while the aroA prfA mutant transcribed inlAB almost exclusively from PinlA3 (and PinlA1) and the aroA sigB mutant showed transcription of inlAB exclusively from PinlA2. The total amounts of transcripts in the wild-type strain and in the prfA, sigB, and prfA sigB mutants were not significantly altered, but the transcripts were upregulated about two- to threefold in the corresponding aroA mutant strains. The amount of InlA protein produced by the aroA prfA mutant was about 10-fold greater than the amount produced by the prfA mutant, and the amount of the InlA protein produced by the aroA sigB mutant was also about 10-fold greater than the amount produced by the sigB mutant; i.e., the relative increase in InlA production due to the aroA mutation in the prfA and sigB mutants was similar to the relative increase in the aroA mutant compared to the wild type (Fig. 3C and D). Much less InlA was detected in the prfA sigB mutant and the aroA prfA sigB mutant, but the aroA prfA sigB mutant still produced much more InlA than the prfA sigB mutant produced (Fig. 3C).

FIG. 3.

Primer extension experiments for determining the transcriptional start sites of the inlA mRNAs. (A) Primer extension experiments were performed with the transcripts obtained from aroA, prfA, and sigB deletion mutants, and the locations of the start nucleotides were mapped by comparison with the nucleotide sequence of the promoter region. (B) 5′ upstream sequence of inlA, including the promoters PinlA2 and PinlA3 and the start sites for the P1, P2, and P3 transcripts. The −10 box and the PrfA binding site of PinlA2 are underlined with solid lines, while the SigB-dependent −10 and −35 boxes of PinlA3 are underlined with dotted lines; the start codon of InlA is enclosed in a box. (C) Western blot analysis of InlA and its degradation products with anti-InlA antibodies. The amounts of InlA in the aroA prfA sigB and the prfA sigB mutants were determined in the same Western blot, but the blot was exposed for 15 min instead of 1 min to visualize the small amounts of InlA produced in these mutants. (D) Amounts of InlA determined by low-light imaging (upper panel) and relative efficiencies of uptake of the various strains by Caco-2 cells (lower panel). All quantitative data are averages and standard deviations for three measurements.

Translation efficiency of the inlAB transcripts depends on the 5′ upstream regulatory region.

Both major inlAB transcripts (P2 and P3), which start 49 bp apart, contain a 396 (445)-nucleotide UTR. To test whether this UTR might play a regulatory role in translation of the inlAB transcripts similar to the role shown recently for translation of the prfA and actA transcripts in L. monocytogenes (23, 54), we introduced different deletions into the UTR (Fig. 4A) and compared the effects of these deletions on InlA production in aroA⁺ and aroA backgrounds. The results are shown in Fig. 4C. Five of the seven deletions introduced into the UTR sequence reduced the amount of InlA and internalization in Caco-2 cells compared to the wild-type strain, suggesting that the UTR indeed plays an essential role in the expression of InlA (Fig. 4C, middle and right graphs). However, none of these deletions reduced the enhancement of InlA synthesis in the aroA background (Fig. 4C, middle graphs). These data suggest that the UTR and the aro mutations strongly affect the translation of both inlAB transcripts regulated by PrfA and SigB, respectively.

FIG. 4.

Analysis of inlA mutant strains containing truncated 5′ UTRs. (A) The deletions introduced into the UTR are indicated by dotted lines. (B) The mfold algorithm was used to predict putative secondary structures. The SD sequence is indicated by a red line, and the secondary structure harboring the SD sequence is scaled up. Similarities to the UTR structure of the wild-type strain were determined by using the RNA distance algorithm; higher values indicate higher levels of dissimilarity. (C) Effects of the UTR deletions (see panel A) on the amounts of inlA transcripts measured by RT-PCR (left graphs), the amounts of InlA protein produced in the strains carrying the various UTR deletions (middle graphs), and the relative efficiencies of uptake of these strains by Caco-2 cells (right graphs). All quantitative data are averages and standard deviations for three measurements.

To test whether the UTR may influence the stability of the inlAB transcript, we determined the amounts of the transcript in the different UTR deletion mutants by real-time PCR. As shown in Fig. 4C (left graphs), only one of the UTR deletion mutants (UTRdel15/389, in which the UTR was almost completely deleted) showed strongly altered levels of the inlAB transcripts. However, the changes observed in the amount of the inlAB transcript did not correlate with the amounts of InlA in the UTR deletion mutants with the aroA background.

Computational analysis of the RNA secondary structures of the wild type and the various truncated inlA UTRs with the mfold program (http://www.bioinfo.rpi.edu/applications/mfold) based on the mfold algorithm (57) (Fig. 4B) showed that the UTR RNA could indeed form a rather stable secondary structure which was significantly altered in the truncated UTRs. The UTR secondary structure may affect translation initiation of inlA mRNA since the Shine-Dalgarno (SD) sequence seems to be buried in a double-stranded stem. The altered structures of the truncated UTRs were compared to the structure of the original UTR by using the RNA distance algorithm (http://bioweb.pasteur.fr/seqanal/interfaces/rnadistance.html) (14, 21, 46, 47) (Fig. 4B). A high value indicated a high level of dissimilarity between the two structures compared (original and truncated), while similar structures resulted in lower values. The data obtained showed that there was no apparent correlation between the altered UTR secondary structures (compared to the wild-type UTR) and the reduced amounts of InlA, as was recently reported for translational control of the actA mRNA (54). However, it should be noted that in all truncated UTRs, including the most extensively deleted UTR, UTRdel15/389, the SD sequence remained in a double-stranded stem structure, but the stability was considerably less than that in the wild-type UTR. sigB was deleted in the UTRdel15/389 mutant strain to exclude new stem-loop formation due to transcription from the SigB-dependent promoter (P3), yet no significant change in the amount of the inlA transcript or the amount of InlA was detected.

The major contribution to enhanced InlA production in the aro mutants seemed to be caused by the switch to anaerobic physiology.

It seemed that it was rather unlikely that the increased InlA (and InlB) production in the aro mutants was caused by a possible shortage of the three aromatic amino acids since growth of these mutants occurred in rich culture medium (BHI) and even supplementation with additional phenylalanine, tryptophan, and tyrosine did not reduce the enhanced InlA production in the aro mutants (data not shown). To rule out the possibility that the aro mutants were impaired for uptake of these amino acids, we constructed deletion mutants which were still able to synthesize chorismate, the precursor of all aromatic amino acids, as well as folate and menaquinone, but were unable to synthesize phenylalanine, tryptophan, and tyrosine due to deletions in pheA, trpEG, and tyrA. These mutants grew like the wild-type strain in BHI but did not show increased InlA production (Fig. 5A).

FIG. 5.

(A) Amounts of the InlA protein in the wild-type strain, the aroA mutant, and different mutants with deletions in the specific aromatic amino acid pathways (determined by Western blot analysis with anti-InlA antibodies) under oxic conditions. (B) Amounts of InlA produced in the wild-type strain and the aroA mutant under anoxic conditions. (C) InlA production in the aroA and aroE mutants (solid bars) compared to the production in the wild-type strain after addition of shikimate (open bars) and additional vitamin K₂ (shaded bars). The quantitative data (which were determined by low-light imaging) are averages and standard deviations for three independent experiments.

As shown recently (51), all aro mutants of L. monocytogenes switch to predominantly anaerobic metabolism due to the lack of menaquinone synthesis. To test whether this altered growth physiology may cause the enhanced inlAB mRNA translation, we examined InlA production in the L. monocytogenes wild-type strain grown under anoxic conditions.

As shown in Fig. 5B, InlA production was indeed enhanced in the anaerobically grown L. monocytogenes wild-type strain, and the amount of InlA produced under these conditions was comparable to the amount produced by the aroA mutant cultured under oxic or anoxic conditions, although only 1.5- to 2-fold upregulation of inlA transcripts was detected for the bacteria grown in anoxic conditions compared to oxic conditions.

Oxidative respiration of aro mutants could be restored by addition of vitamin K₂ and shikimate to cultures of the aroA, aroB, and aroA/B mutants (51). Addition of shikimate to BHI also suppressed the enhanced InlA synthesis in the aroA mutant but not in the aroE mutant (which was unable to convert shikimate into the chorismate necessary for menaquinone production) (Fig. 5C). However, the enhanced InlA synthesis in the aroE mutant was suppressed by supplementation with vitamin K₂, which restored aerobic metabolism in this mutant (Fig. 5C).

DISCUSSION

Internalization of L. monocytogenes by nonprofessional phagocytic cells requires InlA and/or InlB (for a recent review see reference 9). Although this process may be supported by other internalins (3) or even noninternalin extracellular and surface proteins (20, 22, 29, 35, 39, 44), no (or only little) internalization by these host cells occurs in the absence of these two internalins (3). Previous reports have demonstrated that there is rather complex regulation of inlAB expression at the transcriptional level by PrfA-dependent and -independent promoters (12, 33).

The data described here show that transcription of the inlAB operon is basically achieved by a PrfA-dependent promoter (PinlA2) and a SigB-dependent promoter (PinlA3) which are 49 bp apart. All other transcripts may be degradation products of the P3 transcript since (i) a prfA sigB mutant produces undetectable levels of inlA transcripts and shows strongly decreased internalization by nonphagocytic cells comparable to that observed with a noninvasive Listeria innocua strain; (ii) transcript P1 is strongly enhanced in a prfA mutant, as is the SigB-dependent transcript P3; and (iii) transcript P1 is not produced in a sigB deletion mutant. However, it is unlikely that the P1 transcript is produced by a SigB-dependent promoter since there is no SigB-specific −10 box at the appropriate distance from the site where the P1 transcript should be initiated.

The enhanced InlA (and to a lesser extent InlB) production observed in the aro mutants does not seem to be caused by activation of PrfA and/or SigB, which in turn could lead to enhanced transcription of P2 and/or P3. This possibility can also be ruled out since the amount of the InlA protein is about 10-fold greater and the inlA transcripts are much less enhanced in an aroA background than in an aroA⁺ background even in the prfA sigB mutant strains.

Rather, the results argue for positive posttranscriptional control of the inlA(B) transcripts, which leads to increased InlA(B) synthesis. Both positive and negative modes of translational control have been repeatedly demonstrated in prokaryotes, and in most cases they involve the 5′ UTRs of the transcripts of the corresponding genes. These RNA sequences may form secondary structures, which often bury the ribosome binding site (SD sequence) of the mRNAs (10) or alter their stability (1, 8). Physical parameters, like temperature (23, 56), antisense RNA (6) and other small RNAs (34, 49, 53), and also riboswitches (37) may alter these secondary RNA structures, thereby modulating the translation efficiency.

Our data show that the posttranscriptional regulation of internalin A and the coregulated synthesis of internalin B involve two parameters. The first is the 396-nucleotide UTR which is common to the P3 and P2 transcripts. The UTR RNA can fold into an extended secondary structure which seems to be essential for translation efficiency of the P3 and P2 transcripts. In this structure the SD sequence of the inlAB mRNA that is essential for ribosome binding and hence translational initiation is buried in a double-stranded stem and therefore may not be easily accessible to the ribosome. Even in the truncated UTRs, including those with the most extended deletion (374 nucleotides), the SD sequence seems to be buried in putative double-stranded stem structures; however, the stabilities of the structures are widely different. The data that have been obtained also show that the UTR deletions do not significantly alter the stability of the P2 and/or P3 transcripts, with the exception of the most extended deletion, UTRdel15/389, which results in a strongly reduced amount of inlAB transcript that is probably due to destruction of the extended UTR structures of the inlAB transcript (8, 19).

Production of antisense RNA as a regulatory element for the observed translational control of the inlAB transcript can be ruled out, as demonstrated by Northern blot analysis (data not shown). Furthermore, the RNA profiles of the aro mutants and the wild-type strain grown under aerobic and anaerobic conditions, respectively, obtained by using whole genome microarrays, revealed high levels of similarity (data not shown). However, due to the complex RNA patterns at this time no conclusions can be drawn from these data regarding the trigger for the translational upregulation of InlA. The data therefore suggest that the UTR structure alone or in combination with an unknown factor (eventually necessary to facilitate access to the SD sequence for the ribosome) enhances translation of the inlAB transcripts.

Interestingly, translation activation of the inlA(B) transcripts in aro mutants apparently does not require an intact UTR structure since all transcripts with truncated UTRs show a similar increase in translational activation, although some of these transcripts yield a significantly reduced basal amount of InlA. This finding led us to the conclusion that a signal substance triggered by the aro mutation is a second parameter in the posttranscriptional control of InlA(B) synthesis, which seems to be unrelated to the translational control of the inlAB transcript by the UTR structure since a deletion that removed almost the entire UTR still led to enhancement of InlA production in the aro mutants.

A shortage of the aromatic amino acids can be ruled out as a trigger since a mutant unable to produce tryptophan, phenylalanine, and tyrosine due to mutations in the specific biosynthesis branches did not show increased InlA production and addition of all aromatic amino acids to the already rich BHI medium did not eliminate the increased InlA production in the aro mutants. A fortuitous additional mutation in the aro deletion mutants responsible for increased amounts of InlA is also unlikely since an aroE reversion mutant produced wild-type InlA levels (data not shown). These data indicate that other products branching off the basic aromatic amino acid biosynthesis pathway may be responsible for the observed enhancement of InlA(B) production by the aro mutants.

The predominantly anaerobic metabolism of all aro mutants even in the presence of oxygen, obviously due to the lack of menaquinone synthesis (51), strongly suggests that anoxic conditions is a trigger signal. Indeed, when wild-type L. monocytogenes is cultivated anaerobically in BHI, it produces increased amounts of InlA comparable to the amounts obtained with the aroA mutant. Furthermore, addition of shikimate and vitamin K₂, which allowed the aro mutants to perform aerobic respiration and restored growth (51), resulted in decreased InlA levels.

The absence of oxygen may also be an appropriate in vivo trigger for the upregulation of InlA in the intestinal tract of L. monocytogenes-infected humans or animals, where internalization of the listeriae by epithelial cells seems to play a decisive role (32, 42).

Taken together, the data clearly show that InlA synthesis is posttranscriptionally regulated and that this control is mediated by the intact UTR of the inlA mRNA and anaerobic metabolism; however, it is not clear whether these two posttranscriptional regulatory events are mechanistically linked.

The results reported here further support the view that in addition to the transcriptional control exerted by the central regulator PrfA on most virulence genes, posttranscriptional control mechanisms may represent a second line of virulence regulation in L. monocytogenes. In addition to the two internalins reported here, translation control has been demonstrated for the mRNAs of iap (28), prfA (23), and actA (54) and is likely for inlC, in which an extended UTR with a specific secondary structure(s) has been identified (13); Q. Luo and W. Goebel, unpublished data). None of the transcripts for the latter virulence factors exhibits significantly enhanced translation in the aro mutants, and the UTRs of these transcripts differ in their sequences and putative secondary structures, as well as in the conditions that activate translation. The UTR of the prfA transcript P2 appears to be a thermosensor, as an elevated temperature (above 30°C) leads to strongly enhanced translation of this prfA transcript (23). The inducing conditions for optimal translation of the iap and actA transcripts are not known, but they are unlikely to be similar to those for the inlA(B) transcript since no comparable increase in p60 or ActA production was observed in the aroA mutant.

Posttranscriptional control may thus provide a more general fine-tuning mechanism for enhancing synthesis of virulence factors in infection niches where the factors are specifically required and where specific signals for enhanced mRNA translation might be present.