Listeria monocytogenesl-forms respond to cell wall deficiency by modifying gene expression and the mode of division

Authors

  • Simone Dell'Era,

    1. Institute of Food Science and Nutrition, ETH Zurich, Schmelzbergstrasse 7, CH-8092 Zurich, Switzerland.
    Search for more papers by this author
  • Carmen Buchrieser,

    1. Institut Pasteur, Unité Biologie des Bactéries Intracellulaires and CNRS URA 2171, 28 Rue du Dr Roux, 75724 Paris, France.
    Search for more papers by this author
  • Elisabeth Couvé,

    1. Unité Génétique des Génomes Bactériens, 28 Rue du Dr Roux, 75724 Paris, France.
    Search for more papers by this author
  • Barbara Schnell,

    1. Institute of Food Science and Nutrition, ETH Zurich, Schmelzbergstrasse 7, CH-8092 Zurich, Switzerland.
    Search for more papers by this author
  • Yves Briers,

    1. Institute of Food Science and Nutrition, ETH Zurich, Schmelzbergstrasse 7, CH-8092 Zurich, Switzerland.
    Search for more papers by this author
  • Markus Schuppler,

    Corresponding author
    1. Institute of Food Science and Nutrition, ETH Zurich, Schmelzbergstrasse 7, CH-8092 Zurich, Switzerland.
      *E-mail: markus.schuppler@ilw.agrl.ethz.ch; Tel. (+41) 44 632 7862; Fax (+41) 44 632 1266;
    Search for more papers by this author
  • Martin J. Loessner

    Corresponding author
    1. Institute of Food Science and Nutrition, ETH Zurich, Schmelzbergstrasse 7, CH-8092 Zurich, Switzerland.
      **E-mail martin.loessner@ethz.ch; Tel. (+41) 44 632 3335; Fax (+41) 44 632 1266.
    Search for more papers by this author

*E-mail: markus.schuppler@ilw.agrl.ethz.ch; Tel. (+41) 44 632 7862; Fax (+41) 44 632 1266;

**E-mail martin.loessner@ethz.ch; Tel. (+41) 44 632 3335; Fax (+41) 44 632 1266.

Summary

Cell wall-deficient bacteria referred to as l-forms have lost the ability to maintain or build a rigid peptidoglycan envelope. We have generated stable, non-reverting l-form variants of the Gram-positive pathogen Listeria monocytogenes, and studied the cellular and molecular changes associated with this transition. Stable l-form cells can occur as small protoplast-like vesicles and as multinucleated, large bodies. They have lost the thick, multilayered murein sacculus and are surrounded by a cytoplasmic membrane only, although peptidoglycan precursors are still produced. While they lack murein-associated molecules including Internalin A, membrane-anchored proteins such as Internalin B are retained. Surprisingly, l-forms were found to be able to divide and propagate indefinitely without a wall. Time-lapse microscopy of fluorescently labelled l-forms indicated a switch to a novel form of cell division, where genome-containing membrane vesicles are first formed within enlarged l-forms, and subsequently released by collapse of the mother cell. Array-based transcriptomics of parent and l-form cells revealed manifold differences in expression of genes associated with morphological and physiological functions. The l-forms feature downregulated metabolic functions correlating with the dramatic shift in surface to volume ratio, whereas upregulation of stress genes reflects the difficulties in adapting to this unusual, cell wall-deficient lifestyle.

Introduction

Bacterial l-forms are cell wall-deficient derivatives of bacteria which have lost the ability to maintain a stable murein sacculus. Existence of this phenotype was first described in 1935, and l-forms were named in honour of the British surgeon Joseph Lister (1827–1912). Although l-forms may be considered similar to bacterial protoplasts, they show a variable morphology characterized by an extreme pleomorphism, sometimes featuring a significantly enlarged diameter of 10–20 μm (Edman et al., 1968; Allan et al., 1993), compared with 1–3 μm for cells with walls. Other characteristics which distinguish l-forms from normal bacteria include specific growth requirements, loss of peptidoglycan (PG)-associated molecules and their functions, and intrinsic resistance to antibiotics interfering with cell wall synthesis such as the β-lactams (Brem and Eveland, 1967; Edman et al., 1968; Mattman, 2001).

l-forms apparently occur in both Gram-positive and Gram-negative bacteria, although it is not clear whether these mostly older reports (reviewed by Mattman, 2001 and Madoff, 1986) actually refer to a stable or a transient state of cell wall deficiency. Spontaneous transformation to the l-form phenotype has been reported in only few species, but l-form generation can often be stimulated by exposure to inhibitors of cell wall synthesis such as antibiotics, cell wall-hydrolysing enzymes, or amino acids such as glycine or lysine which in high concentrations can also interfere with PG synthesis (Maisnier-Patin and Richard, 1996). Several (mostly older) studies describe morphological, serological and biochemical properties of l-form bacteria (Ghosh and Murray, 1966; Domingue and Woody, 1997). However, very little is known on the molecular changes accompanying l-form transition, the background of l-form transformation, and possible genetic determinants for l-form transition and/or persistence. Earlier studies (Landman and Halle, 1963; Landman, 1968; Wyrinck and Gooder, 1977) mostly focused on Bacillus subtilis and the expression of factors known or suspected to be involved in PG synthesis (Henriques et al., 1998), teichoic acid synthesis (Honeyman and Stewart, 1989), or cell shape determination (Jones et al., 2001), but did not yield evidence that l-form transition could be due to mutation. A recent report (Joseleau-Petit et al., 2007) suggested that unstable l-form-like variants of Escherichia coli retain low amounts of PG and require PG synthesis for growth, In contrast, Leaver et al. (2009) showed that PG synthesis is not required for B. subtilisl-forms. Interestingly, the latter authors also identified a single point mutation which predisposes B. subtilis to grow without a wall. Moreover, these l-form cells apparently use a novel membrane extrusion-resolution mechanism for propagation, which does not require FtsZ. Independently, we here discovered and described a conceptually similar, yet clearly distinct mechanism.

When bacteria shed their cell wall, they consequently also lose PG-associated proteins, carbohydrates and associated functions. With respect to pathogens, they might then also lose factors contributing to their specific ‘pathogen-associated molecular pattern’ important for recognition of the invader by the (innate) immune system of the hosts (Mogensen, 2009). Hence, absence of a cell wall in l-form bacteria may result in the inability of the immune system to recognize these cells as pathogens. The l-form transition can also have consequences for antibiotic therapy, because of the ineffectiveness of cell wall-active drugs (Domingue and Woody, 1997; Woo et al., 2001; Koch, 2003).

Listeria monocytogenes is a Gram-positive, facultative intracellular rod which causes invasive, often fatal infections in susceptible hosts such as newborns, pregnant women, elderly and immunocompromised individuals (Vázquez-Boland et al., 2001). Listeria is widely distributed in the environment and also occurs in the intestinal tract of healthy animals and humans. The L. monocytogenes strain Scott A used here is of serovar 4b, the most common cause of human epidemic listeriosis (Kathariou, 2002). It has been isolated during a listeriosis outbreak (Fleming et al., 1985), where the bacteria could be recovered from clinical cases but not from the pasteurized milk considered to be the source of the outbreak. Interestingly, strain Scott A was noted for its tendency to produce colonies of aberrant morphology on media containing cell wall-active antibiotics (Loessner et al., 1988), and also attracted interest because of the PSA prophage (Lauer et al., 2002; Zimmer et al., 2003).

Clearly, there is a need to understand l-form conversion and the molecular changes accompanying and possibly underlying this transition. Towards this aim, we have generated stable L. monocytogenesl-form lines and investigated the cells by a combination of imaging techniques and molecular approaches. Observations made using fluorescently stained l-form cells indicate a novel mode of growth and division for cell wall-deficient bacteria. Transcriptome analysis revealed manifold differences in gene expression, suggesting that Listerial-forms can adapt to their cell wall deficiency by adjusting expression levels of genes important for survival and adaptation to this unusual life style.

Results

l-form bacteria can grow and multiply

Putative L. monocytogenesl-form colonies showing a characteristic ‘fried egg’ like appearance on Listerial-form medium (LLM) agar media containing penicillin G (PenG) were detected after incubation for 16–24 days at 32°C. l-form colonies occurred at a frequency of approximately 10−5−10−6 per total cells plated. Neither different penicillin concentrations tested, nor incubation at different temperatures between 20°C and 35°C did significantly influence this ratio.

Stable l-form variants of parent Scott A and the green fluorescent protein (GFP)-labelled Scott A::pPL3-GFP strain were obtained after repeated passages in LLM soft agar, followed by step-wise decrease and eventual elimination of PenG. The procedure is outlined in Fig. 1. Over a period of more than 4 years, these l-form cultures have until now been passaged for approximately 200 generations in medium without antibiotic; we have never observed reversion to parental cell wall-containing bacteria. Further attempts such as supplementation of the LLM medium with trypsin (DeCastro-Costa and Landman, 1977) or cultivation at different temperatures did also not result in cell wall-containing revertants (data not shown).

Figure 1.

A. Generation of L. monocytogenesl-forms, monitored by phase-contrast microscopy. Parental L. monocytogenes Scott A cells (stage I) plated on LLM induction plates initially form intermediate, rounded l-form-like cells (stage II), which are able to revert to normal morphology (indicated by double-pointed black arrow). Stable l-form cells appear as round spheres (stage III) were obtained after at least five passages on LLM soft agar with and without penicillin. They are devoid of any cell wall material and are unable to revert to a wall carrying cell (see Experimental procedures for details).
B. Following incubation in soft agar medium, the bacteria grow as a ‘galaxy’ (right), consisting of many spherical colonies containing actively growing individual l-form cells (left).

The identical procedure was employed for generation of l-form cultures from the other Listeria strains tested (strains EGDe and ATCC 23074), with analogous results.

The growth of L. monocytogenesl-forms in media other than LLM, and in medium without addition of serum was poor. However, we found that the horse serum contained in the initial LLM formulation could be fully substituted by milk serum, without any negative effect on growth, viability and stability of the l-form cells.

Growth response and quantification of Listerial-forms

l-forms generally feature much slower growth compared with parental Listeria cells, which reach late log phase/early stationary growth phase after 16–20 h of incubation under standard culture conditions used here (data not shown). Regarding the Listerial-forms, approximately 6–8 days passed from inoculation in fresh LLM soft agar until the macroscopically visible, active growth phase came to an end. This log phase-like growth stage could be monitored by measuring concentration of total RNA extracted from growing l-form cultures at different time points, and quantified by employing real-time (quantitative) PCR for determining copy numbers of the single copy chromosomal actA gene (Fig. 2). The results confirmed that the maximum copy number, corresponding to the late log phase/early stationary growth phase, is reached approximately 6 days post inoculation, depending on the size of the inoculum.

Figure 2.

Growth and multiplication of L. monocytogenes Scott A stable l-forms in LLM soft agar. Total DNA extracted from the tubes was subjected to real-time quantitative PCR for determination of the total number of chromosomal actA copies, as a measure of cell number. The initial l-form inocula used for inoculation of the tubes (y-axis) are indicated in the legend.

l-forms lack a cell wall and show variable dimensions

The stable l-form cells are non-motile and completely devoid of the thick and multilayered Listeria murein sacculus. The absence of an organized cell wall in l-forms was demonstrated by electron microscopy (Fig. 3); freeze-substitution and high-resolution transmission electron microscopy of frozen-hydrated ultrathin sections did not yield any visible evidence for cross-linked PG structures.

Figure 3.

Freeze-substitution transmission electron microscopy of parent and l-form cells of L. monocytogenes Scott A shows integrity of the membranes and loss of the peptidoglycan cell wall sacculus in the l-forms.
A. Parental cells with the typical rod-shaped morphology.
B. Cell wall deficiency in l-form cells results in variable shape and dimension.
C. Magnification of the cell surface of a parental cell, showing the Gram-positive cell wall of Listeria, consisting of a double layered cytoplasmic membrane (white arrow), a thick and multilayered peptidoglycan layer (black arrow) and associated carbohydrates/teichoic acids (white bracket).
D. Detailed view of the l-form membrane outer surface with a cytoplasmic membrane only (white arrow), and no evidence for an organized cell wall.

Generally, l-forms showed a remarkable heterogeneity in size and shape, likely influenced by the age and propagation state of the individual cells (see below). The smaller cells appeared round and featured a thin and delicate surface, while older cells showed a more variable, sometimes inhomogeneous surface. The volume of an average parental rod shaped cell, assumed as a cylinder with an approximate diameter of 400 nm and a length of 1300 nm (Fig. 3A) was calculated to be 1.6 × 10−1 μm3, whereas the spherical l-form cells (diameter ranging between 1600 nm and 15 μm (Fig. 3B) have a volume of up to 104μm3. Thus, the volume of an l-form can be up to 105-fold larger than that of an average parental cell.

l-form vesicles contain DNA and are metabolically active

In an initial set of experiments, staining l-forms with the dsDNA-specific dye 4′-6-diamidino-2-phenylindole (DAPI) (Fig. 4A) indicated inhomogeneous distribution of DNA as genetic material within larger l-form cells. This observation could later be explained after employing confocal microscopy, which indicated an unusual mode of cell replication (see below).

Figure 4.

L. monocytogenesl-forms visualized using fluorescent dyes.
A. Phase-contrast microscopy (left), fluorescence microscopy (middle) and merged image (right) of a DAPI-stained, large l-form cell showing inhomogeneous distribution of genetic material.
B. Phase-contrast and fluorescence imaging of stable l-forms of L. monocytogenes Scott A::pPL3-GFP. The fluorescent protein allows the identification of metabolically active, viable l-form cells and their distinction from similar structures.
C. Phase-contrast (left), fluorescence (middle) and merged image (right) of a large fluorescent l-form vesicle, showing the presence of several subunits within the larger membrane (see Figs 8 and 9).

The development of green fluorescence in the gfp carrying l-form cells clearly demonstrated metabolic activity required for synthesis and maturation of the GFP protein (Fig. 4B). Fluorescence microscopy allowed identification of viable, living cells and their differentiation from empty membrane vesicles often observed in l-form cultures. Strong GFP signals were observed in large l-bodies consisting of up to 50 subunits, which appeared 3–4 days after transfer of first-stage l-form colonies from the drug-containing induction plates into the LLM soft agar (Fig. 4C). Note that the round vesicles within the larger units exhibit no fluorescence and appear dark; further experimentation indicated that these may represent the new, metabolically yet inactive daughter cells (see below, and Figs 8 and 9).

Figure 8.

Growth and multiplication of L. monocytogenesl-form cells. The l-forms were inoculated in LLM soft agar in flow-channel microscopy slides (see Experimental procedures), and incubated at 32°C on the stage of an inverted microscope equipped with a heating chamber.
A. Growth of an l-form cell monitored over time, by phase-contrast (upper row) and confocal laser scanning microscopy (lower row). Arrows indicate the development of intracellular vesicles.
B. Phase-contrast (upper row) and laser scanning confocal microscopy (lower row) of different phases of l-form growth and multiplication. Note that the images show different cells, representing the typical morphology, development and multiplication of l-form cells at different time points after inoculation into fresh medium.
C. Hypothetical model for the growth and multiplication of L. monocytogenesl-forms (based on the images above and many other similar or identical observations not shown here). The maternal l-form cell starts to develop intracellular vesicles (I). Approximately 6–12 h later, these internal vesicles grow in volume and number (II–IIIa) until the maternal cell is disrupted (IIIb), accompanied by an immediate disappearance of its GFP fluorescence and release of the daughter cells. The liberated l-forms develop fluorescence shortly after release, identifying them as metabolically active, new cells (IV).

Figure 9.

Visualization of L. monocytogenesl-form Scott A using compartment-specific dyes.
A. Phase contrast microscopy image of large L-form cell with internal vesicles.
B. Metabolic activity in the cytosol of the mother cell (but not in the vesicles) is indicated by GFP fluorescence.
C. The membrane-impermeable lipid dye FM4-64 visualizes only the surrounding cytoplasmic membrane, but does not decorate the internal vesicles.
D. Staining with DAPI demonstrates presence of dsDNA in each of the vesicles, suggesting previous genome transfer into the progeny cells.
E. Merged image generated by overlay of fluorescence signals from GFP, FM4-64, and DAPI.

l-forms produce PG precursors and carry membrane-anchored proteins

Vancomycin specifically binds to the terminal d-alanyl-d-alanine moieties of PG precursors which are exported through the cytoplasmic membrane prior to cross-linking to the network of existing PG. Staining with dye-conjugated Vancomycin-FL (Daniel and Errington, 2003) identified such precursors not only in the parental cells, but also on the l-form surface (Fig. 5). Parental rod-shaped cells exhibited a variable spatial distribution of the precursors, with most of the nascent PG located at the poles and in the septal region. In contrast, the l-form surface was decorated in an irregular, inhomogeneous pattern.

Figure 5.

Detection of peptidoglycan building blocks by Vancomycin-FL staining. Images are phase-contrast (left) and fluorescence (right) microscopy.
A. Rod-shaped cells of parental Scott A are stained at septal and polar regions.
B. l-form single cell showing an inhomogeneous distribution of the label on their surface.
C. Vancomycin-FL decoration of an l-form colony embedded in soft agar, consisting of many individual cells. As a control, incubation with fluorescein alone did not yield any signal.

We have also used antibodies raised against purified, protein-free Listeria cell walls (PG and teichoic acids) for immunostaining of parental and l-form cells (Fig. 6). The l-form cells showed inhomogeneous surface staining in particular on the surface of the larger (and older) l-forms, again suggesting presence of PG precursors or incompletely cross-linked murein components.

Figure 6.

Immunostaining of parental and l-forms using polyclonal antibodies directed against purified Listeria cell wall material. Images are phase-contrast (left) and fluorescence (right) microscopy.
A. Antibodies decorate the entire cell surface of parent Scott A cells.
B. l-forms showed an inhomogeneous decoration pattern, indicating the presence of cell wall precursors especially on older l-form cells with an irregular surface (arrows).

Further evidence for an absence of a mature cell wall consisting of peptide-linked PG strands and linked carbohydrates (i.e. wall teichoic acids) was obtained by using phage endolysin-derived cell wall binding protein GFP-CBD500 (Loessner et al., 2002), which failed to decorate the l-form cells (data not shown).

In agreement with our previous findings, protein-specific immunostaining further demonstrated that cell wall-linked Internalin A (InlA) is absent from the l-forms surface, whereas the membrane-anchored Internalin B (InlB) is present (Fig. 7). Loss of InlA is of course due to the missing PG to which the protein is normally attached via SrtA (Bierne et al., 2002). The InlB molecules attach to lipoteichoic acid polymers, which are membrane bound (Braun et al., 1997). Reverse transcriptase amplification of inlA and inlB mRNA transcripts demonstrated that both genes are transcribed, at similar levels (which is in agreement with the transcriptome analysis described below).

Figure 7.

Immunostaining of Listeria cells and l-forms with antibodies against InlA and InlB. Inserts show partial phase contrast images of the same fields.
A and D. L. monocytogenes EGDe cells produce InlA and InlB and are labelled by both antibodies (positive control).
B and E. Parental Scott A produces InlA and InlB and is labelled by both antibodies.
C. Scott A l-forms do not react with anti-InlA.
F. Immunostaining with anti-InlB demonstrates membrane-anchored Internalin B on the l-form surface.
G and H. Products of reverse-transcription (RT) PCR from l-form and parental RNA demonstrates transcription of InlA in parental (lane 1) and l-form cells (lane 2), and of InlB transcription for both parental (lane 6) and l-form cells (lane 7). As control for absence of contaminating genomic DNA, a standard PCR (without RT-step) from the same RNA preparations was performed (lanes 3, 4 and 8, 9). Lanes 5 and 10 are positive amplification controls using inlA and inlB as templates. Also shown is a DNA standard size marker (lanes M).

l-forms grow and multiply by formation of intracytoplasmic vesicles

Time-lapse confocal laser scanning microscopy (CLSM) of the growth and multiplication of GFP-labelled l-forms allowed us to identify several different phases reflecting the different steps of l-form propagation. Based on these observations, we propose a novel model for the growth and division of Listerial-forms. The generation of intracytoplasmic vesicles, clearly visible as smaller, round and non-fluorescent subunits within the larger maternal cells, is eventually followed by their release through collapse of the maternal cell (Fig. 8B). During this process, the membrane seems to first fold inwards (Fig. 8A), and then appears to pinch off to form the intracellular vesicles, which, at this point, still appear dark (Fig. 8B). The subunits become fluorescent after release from the mother cell, indicating activation of metabolic function followed by synthesis and/or subsequent oxidation of intracellular GFP. A schematic representation of this process is depicted in Fig. 8C.

In order to show that the small, round internal subunits actually contain DNA (assumed to represent a copy of the bacterial chromosome), we used compartment-specific dye labelling and CLSM to visualize the spatial arrangement of membranes (red), metabolically active cytosol (green) and dsDNA (blue). The result is shown in Fig. 9, and clearly shows that the non-fluorescent internal vesicles (proposed ‘daughter’ cells) contain DNA, and that they are fully contained within the cytoplasm of the larger ‘mother’ cells. These internal daughter cells (i.e. their membranes) are not yet accessible to the membrane-impermeable red dye and are therefore not labelled.

Comparative transcriptomics show differential expression of genes for metabolism, stress response, transport and regulation functions

Transcription profiles of parental L. monocytogenes Scott A and stable l-forms were recorded using DNA macroarrays featuring 2816 genes from L. monocytogenes EGDe (Milohanic et al., 2003), in addition to 153 L. monocytogenes serovar 4b-specific genes (Severino et al., 2007). Parent cell RNA was harvested from late log phase/early stationary phase cultures in LLM medium, whereas l-form RNA was extracted from 6 days' cultures in LLM soft agar, which represented the corresponding late log/early stationary growth phase of l-form cells (Fig. 2). This agrees well with findings from preliminary experiments in which we performed macroarray hybridizations using RNA isolated from l-forms 8 days post inoculation and found overexpression of groEL and groES, which indicated entry into the stationary phase (Lado et al., 2004). Younger l-form cultures (3 days' incubation) yielded only low amounts of mRNA. Taken together, the harvesting of 6 days' l-forms best represents the culture conditions used for the parent cells. However, it cannot be excluded that variations in growth media (soft agar versus liquid culture) might also influence gene expression.

For each cell type, two independent mRNA preparations were used for cDNA synthesis and hybridization, using two independent array sets. Only statistically significant differences in genes expression between parent and l-forms (ratio 2.5 or more) were considered and analysed further (results are listed in Tables S1 and S2; see Supporting information). Ratios were calculated for each gene represented on the array, using average values from duplicates of single experiments followed by determination of means from two independent experiments. Additional evidence indicating the reliability of this analysis came from the fact that genes clustered in co-transcribed operons generally showed very similar or identical values (up- or down-regulation).

Overall, 276 genes were found to be differentially expressed (minimum threshold value of 2.5-fold). In the l-forms, 78 genes were upregulated and 198 were downregulated (Fig. 10A). Most of the genes could be assigned to specific functional categories such as metabolism, transport, stress, regulators, ribosomal proteins, prophage genes and others. The majority of genes attenuated in the l-forms appear to be involved in metabolism and transport, whereas most of the induced genes encode stress response and regulatory functions (Fig. 10B). Figure 10C lists the 15 genes showing the strongest expression differences.

Figure 10.

Graphical representation of differential gene expression in parental and l-form L. monocytogenes, based on whole genome macroarray hybridization analysis.
A. Schematic representation of genes with significantly different expression ratio (> 2.5) in intact cells and l-forms (2969 genes total).
B. Overview of the 276 genes differentially expressed in l-forms, clustered into functional groups. Categories are represented as percentage of total differentially expressed genes (78 upregulated and 198 downregulated genes respectively).
C. Schematic representation of the expression ratios for the 15 genes that revealed strongest up- (top) or down-regulation (bottom).

l-form transcripts demonstrate induction of stress response and repression of metabolism

Many genes involved in stress-response were induced in the l-form state, such as ATP-dependent protease clpE and the related lmo1138, the two Clp genes clpP (ratio 2.15) and clpE (ratio 2.01), and low-temperature response factors cspD and ItrC.

In contrast, many metabolism and bioenergetics-associated genes were repressed in the l-forms, such as glycolysis genes pgi and ldh, citric acid cycle proteins encoded by the pdh operon, and several dehydrogenases, oxidoreductases and synthases [ldh, lmo0132, lmo0271, lmo0823, lmo1677, lmo2244, menB and others (see Table S2)]. The qox operon encoding the cytochrome d complex was also attenuated, and, as an indirect measure for metabolic activity, more than 20 ribosomal proteins were downregulated.

Reverse transcriptase real-time quantitative PCR confirms macroarray results

To confirm the overall validity of macroarray hybridization results, we selected a subset of the differentially expressed genes and independently determined their copy numbers (expression ratio) using real-time quantitative PCR. The amounts of RNA used as templates in each of the amplification reactions were identical as in the macroarray experiments. Primer pairs (see Supporting information Table S3) were chosen to amplify two upregulated and two downregulated genes each. As a control, two genes were selected that did not show any differences in expression. We found that the quantitative PCR results were in agreement with array hybridization data (see Supporting information Fig. S4).

No evidence for role of mutations in l-form transition

It was recently reported that a single point mutation in ispA, whose product is involved in the formation of essential lipids in the isoprenoid pathway (Fujisaki et al., 1990), predisposes B. subtilis for formation of l-forms (Leaver et al., 2009). To determine if this or other mutations in several genes involved in cell division could possibly play a role for Listerial-form transition, we analysed and compared the sequences of mreB, mreBH, mbl, ftsZ, ftsA and ispA from parent Scott A cells and l-forms derived thereof. However, we found no differences in any of the sequences obtained from the l-form compared with the parent strain, which indicates that l-form formation in Listeria cannot be attributed to mutations at least in these genes.

Discussion

l-forms have long been considered as unstable and short-lived pleomorphic bacterial variants, or even seen as artefacts. We show here that L. monocytogenes Scott A reproducibly transforms into the l-form phenotype during interference with cell wall synthesis by a β-lactam drug, albeit at a low frequency. Following step-by-step reduction and elimination of the drug, the l-forms became stable and could be cultivated without selection. Long-term l-form cultures maintained for more than 4 years clearly demonstrate the stability of this unusual phenotype, and suggests that cell wall deficiency represents an alternative mode of existence. Moreover, successful generation of stable l-forms of other L. monocytogenes strains such as EGDe (SV 1/2a) and ATCC 23074 (SV 4b) suggests the general ability of these bacteria to convert to the l-form lifestyle. In contrast to earlier reports regarding B. subtilisl-forms (DeCastro-Costa and Landman, 1977), we have yet no evidence that the stable l-forms described here may revert back to wall-containing cells, despite the presence of unassembled cell wall precursors. This suggests that true l-forms which have no residual cell wall may be unable to compensate loss of the organized cell wall structure by de novo synthesis just from PG precursors without an existing template, which would in some way be reminiscent of the synthesis of nucleic acids. It remains an intriguing task to determine if supplementation with PG fragments or building blocks could reinitiate assembly of a functional murein layer providing the functionality and shape of parental cells.

Electron microscopy of l-forms confirmed the absence of any recognizable cell wall-like outer structure. However, use of dye-conjugated Vancomycin suggested the presence of PG precursor molecules in l-form membrane regions corresponding to the spots appearing denser in phase contrast microscopy. A possible explanation might be accumulation of precursor molecules in these membrane regions, due to the lack of existing PG to which they would normally be cross-linked. This could suggest that PG precursors are exported from the cytoplasm but then float or stick in the membrane, and therefore remain unorganized on the l-forms surface. This interpretation is supported by the appearance of dense membrane regions especially in larger and older l-form bodies. Although PG synthesis appears to be needed for growth of unstable, l-form-like E. coli (Joseleau-Petit et al., 2007), it is not clear if PG precursors are required for the atypical cell division described here for stable Listerial-forms. However, it is possible that presence of these molecules triggers specific events which are essential for initiation of division and propagation.

Immunostaining demonstrated the absence of the cell wall-bound invasion protein InlA, whereas InlB, which associates to membrane-anchored lipoteichoic acids, is retained on the l-form surface. This correlates with the assumption that l-forms should generally lack all proteins and carbohydrates covalently attached to the PG of intact cells. However, they may retain proteins and other molecules anchored to or embedded within the membrane. Interestingly, the presence of InlB on l-forms and our finding that the expression of virulence genes does not change in l-forms, suggests that at least some specific pathogen factors of L. monocytogenes may be retained on l-forms, which opens the possibility that l-form Listeria may still feature some pathogenic potential.

Use of GFP as an intracellular marker in l-forms allowed, for the first time, the identification of viable, metabolically active true l-forms entities, and their distinction from membrane vesicles and other irregular structures frequently occurring in l-form cultures. The fluorescent tag greatly facilitated monitoring growth and multiplication of l-form cells. The absence of a GFP signal observed in smaller (internal) l-form vesicles is apparently not indicating non-viability of these cells, because fluorescence appeared only after release from the mother cells. This might be due to the requirement for oxidation of the GFP fluorophore by molecular oxygen (Yang et al., 1996), which may not be possible inside the metabolically attenuated (intracellular) daughter cells.

Labelling of bacterial DNA with DAPI showed that the genetic material contained in larger l-forms is not homogeneously distributed or specifically localized within the cells, and its amount seems to increase along with the increase in cell size. Together, this suggests the presence of multiple genomes within the larger l-forms. As genome replication is a prerequisite for cell division and multiplication we addressed the issue using time-lapse microscopy imaging to monitor growth and multiplication of l-form cells. Our findings point to an unusual cell division mechanism, which does not require an outer cell wall structure. Instead, new l-form vesicles are formed (by membrane constriction) and develop within a ‘mother’ cell, eventually followed by collapse of the extended mother cell and release of the new l-forms (see Fig. 8C). This model also explains the appearance of the (initially non-fluorescent) vesicular bodies within larger l-form cells after 3–4 days. Interestingly, the question whether bacteria lacking a rigid cell wall could possibly initiate division by membrane constriction was answered recently (Osawa et al., 2008). The authors show that FtsZ is necessary and sufficient to form a Z-ring and to generate the force required for membrane constriction inside lipid vesicles, even in the absence of other proteins required for normal cell division processes in rod-shaped bacteria. It might thus be possible that such a process, including formation of a Z-ring, membrane constriction and formation of new vesicles containing a copy of the genome can occur inside the l-forms. On the contrary, it was very recently reported that propagation and division of B. subtilisl-forms do not require FtsZ (Leaver et al., 2009). More research is needed in order to better understand this complex and novel process and to answer the many open questions such as spatial regulation of the division process and mechanism of genome transfer.

To address the question whether the stable l-form state is accompanied with a global change in gene expression, we analysed the l-form transcriptome and compared it with the parental cells (it should considered that the data had to be recorded using cells grown under different conditions, and are therefore correlative).

The genes showed the strongest induction in the l-form cells were stress response related. An example is the lmo2484–87 operon, homologous to the stress response regulator psp operon of B. subtilis, which is induced under osmotic stress conditions (Elderkin et al., 2002; Studholme and Dixon, 2003). In contrast, the downregulated qox operon results in decreased synthesis of the oxygen-induced cytochrome d complex (Chatterjee et al., 2006). These findings can be interpreted as a response of the l-forms to their markedly reduced surface-to-volume ratio, and agrees with the rule that enlargement of cells generally results in metabolic downregulation (Koch and Wang, 1982; Hess and Mikhailov, 1995). It also correlates with the downregulation of oxygen metabolism genes ctaA, ctaB and cydA, as a result of adaptation to a low partial oxygen pressure (Svensson et al., 1993).

Downregulation of the gbu operon is likely to represent an adaptation of the cell wall deficient state to growth in environment with high water activity (aw), because the respective genes encode synthesis and uptake of compatible solutes, which is required as an adaptation to a low aw environment (Mendum and Smith, 2002). Interestingly, some of the genes upregulated in l-forms reflect patterns typical for response to salt or cold shock (Duchéet al., 2002; Mendum and Smith, 2002; Wemekamp-Kamphuis et al., 2002; Sleator et al., 2003; Arous et al., 2004). Altogether, numerous regulatory proteins were found to be upregulated in l-forms, pointing to an active adaptation to cell wall deficiency.

An important question in l-form research is whether the cell wall-deficient state and all the dramatic changes accompanying it reflect an adaptation to loss of the PG sacculus and associated molecules, or could possibly be the result of one or more predisposing mutations affecting cell wall integrity or assembly. With respect to Listerial-forms, we found that important cell shape-associated genes and other genes such as ispA (Leaver et al., 2009) were not affected, at least in this strain. Still, it cannot be excluded that yet unrecognized mutations in other genes prevent correct synthesis or assembly of the wall or otherwise predispose to loss of the wall. Any such conditions are likely to be organism or strain specific; an answer may be provided by complete genome re-sequencing of parent and l-form lines.

In conclusion, we interpret the available data in the sense that the changes in gene expression occurring in l-form cells represent a response and active adaptation to the different lifestyle, although this appears to be a quite stressful one. Nevertheless, stable l-forms of L. monocytogenes Scott A are viable bacteria that are not only able to survive but also to replicate and propagate using a unique, previously unknown mechanism. Thus, l-forms are unlikely to be just artefacts, but seem to represent a pre-programmed, alternative phenotype of bacterial life.

Experimental procedures

Bacterial strains

Listeria monocytogenes strain Scott A (serovar 4b; Fleming et al., 1985), L. monocytogenes strains EGDe (serovar 1/2; Glaser et al., 2001) and ATCC 23074 (serovar 4b) were grown in brain–heart infusion (BHI) media (Oxoid, UK). The GFP-labelled derivative Scott A::pPL3-GFP was generated by single-copy chromosomal integration of pPL3-eGFP (Shen and Higgins, 2005) into a specific tRNAArg locus (Lauer et al., 2002).

Induction and stabilization of Listerial-forms

For the generation of cell wall-deficient L. monocytogenes, we modified previously published protocols (Brem and Eveland, 1967; Edman et al., 1968). Parental bacteria were cultivated for 16 h in BHI broth (Biolife, Italy) at 37°C on a rotary shaker (100 r.p.m.). Then, 10–100 μl overnight culture was surface plated onto LLM (3.7% BHI broth base (Biolife, Italy), 1.5% agar (Difco, USA), 0.25% magnesium sulphate and 15% sucrose] (J.M. Jay, pers. comm.). Initially, media were supplemented with 1% horse serum (Sigma, Germany). However, we later found that the blood serum could be substituted by addition of 0.5% lyophilized milk serum (soluble fraction of cows milk devoid of casein and most whey proteins; Emmi; Emmen, Switzerland). Different concentrations of PenG (5 μg ml−1, 50 μg ml−1 and 400 μg ml−1) were added to the media, and the LLM agar l-form induction plates were incubated at 32°C in sealed boxes for up to 4 weeks. Typical ‘fried-egg’ -like colonies appeared on the media surface after 16–24 days. They were picked and transferred to LLM soft agar tubes (0.2% agar), and further incubated at 30°C. Every 10–15 days, l-form cultures were transferred into fresh LLM soft agar medium, with gradually reduced PenG concentration (100 μg ml−1, 50 μg ml−1 and 10 μg ml−1). Finally, the stabilized l-form cells were cultivated in LLM soft agar medium without drug. The same procedure was used for generation of l-forms from the other Listeria strains listed above.

Stabilized Listerial-form cells were routinely cultured at 30°C in LLM soft agar, stored at 4°C, and serially passage using fresh medium every 2–4 weeks. The l-forms have since remained stable for more than 4 years and approximately 200 generations.

Molecular quantification of l-form cell numbers

l-form growth and multiplication cannot be monitored by convenient methods such as surface plating or optical density measurements. Therefore, quantification of l-forms following inoculation into LLM soft agar tubes (prepared with 0.4% low-melt agarose) was assayed as follows: at different time points (Fig. 2), total DNA was extracted after melting the soft agar by heating for 10 min at 90°C, shock freezing at −80°C for 15 min, followed by centrifugation (16 000 g) for 10 min. The DNA containing pellet was washed once with 250 μl purified water and the extraction repeated. Quantification of the total number of actA gene copies per tube was performed using real-time quantitative PCR (Rotor Gene 6000, Corbett Robotics, USA) with primers ActA-F (5′-AAGTGGCGAAAGAGTCAGTTGC-3′) and ActA-R (5′-ACTTTTAGGGAAAAATGGTTGTTGGT-3′). Amplification and melting curve analysis was performed using the following protocol: pre-incubation (95°C for 10 min); amplification (45 cycles): 10 s at 95°C; 15 s at primer-specific annealing temperature; 20 s at 72°C, with a single fluorescence measurement and melting curve analysis from 50°C to 95°C at 2°C s−1 with continuous measurement.

Listeria cell wall antibodies

Protein-free cell walls containing PG and teichoic acids only were prepared as described earlier (Wendlinger et al., 1996). Polyclonal rabbit antibodies against the cell wall fractions were raised using a standard 8 week immunization and harvesting protocol (nanoTools; Teningen, Germany).

Transmission electron microscopy

Thin slices of soft agar containing l-form colonies were placed on an aluminium specimen carrier with a central cavity of 3 × 0.5 mm (BAL-TEC, Balzers, Liechtenstein). Parental Scott A from a broth culture were soaked into cellulose capillary tubes (BAL-TEC), immediately submerged in 1-hexadecene (Sigma-Aldrich, Germany) at room temperature and placed on an aluminium specimen carrier (BAL-TEC). The specimen carriers were placed in the holder of a high-pressure freezer (HPM010, BAL-TEC) and shock frozen as described (Hohenberg et al., 1994). Freeze substitution with Osmium tetroxide, embedding in Epoxy resin, and thin sectioning with a cryo-ultramicrotom were performed as described elsewhere (Dahl and Staehelin, 1989; El-Kest and Marth, 1992; Hohenberg et al., 1994). Electron microscopy of the frozen-hydrated ultrathin sections was performed in a Zeiss EM 912 microscope at a temperature of 100 K (−173°C). Images were recorded using a high-resolution digital CCD camera.

Time-lapse CLSM

GFP-expressing L. monocytogenesl-forms were inoculated on the bottom of miniature glass microscopy dishes (IBIDI; Martinsried, Germany) pretreated with poly-l-lysine (Sigma, Germany), and covered with soft agar. l-forms were observed using an inverted stage Leica TCS SPE™ high-resolution confocal laser scanning microscope (Leica Microsystems; Wetzlar, Germany), equipped with a controlled temperature chamber set at 32°C. Images were recorded digitally, and processed using Leica software and Adobe Photoshop.

For compartment-specific dye labelling (Fig. 9), Scott A::pPL3-GFP l-forms were recovered from soft agar medium and carefully suspended in l-form washing buffer [PBS pH 7.4, 0.5% Tween20 (Sigma-Aldrich, Germany), 0.2 M sucrose] containing 1 μg ml−1 FM4-64 (Molecular Probes, Eugene, OR) and 1 μg ml−1 DAPI (Invitrogen, Carlsbad, CA), followed by CLSM.

Vancomycin-FL staining

Five hundred microlitres of the l-form cell cultures was removed from the LLM soft agar medium using a 1 ml pipette, carefully washed with 900 μl l-form washing buffer and centrifuged for 1 min at low speed (max. 1000 g). The upper phase of the supernatant (∼400 μl) was discarded and the procedure repeated four times. Then, cells were resuspended in 100 μl PBS-Tween20 (0.5%), followed by addition of DAPI (1 μl of a 10 μg ml−1 stock) (Sigma-Aldrich, Germany), or fluorescein-conjugated Vancomycin (2 μl of a 1 mg ml−1 stock) (Sigma-Aldrich, Germany). After 5 min, cells were washed three times with 900 μl l-form washing buffer, carefully centrifuged (1 min at 1000 g), and the supernatant discarded. Parental Scott A cells used as positive control and reference material were also stained, but centrifugation was performed at 8000 g and cells were exposed to staining dyes for 10 min. As negative control for background staining of both cell types, fluorescein was used. Samples were examined using CLSM and/or epifluorescence microscopy (Zeiss Axioplan; Carl Zeiss, Jena, Germany).

Immunolabelling of cell wall and membrane-anchored proteins

Detection of InlA and InlB on parental and l-form cells was performed using polyclonal antibodies raised against InlA and InlB (Parida et al., 1998; kindly provided by Sabine Pilgrim, University of Wuerzburg, Germany). On a piece of parafilm (Pechiney Plastic Packaging Company, USA), a drop of the respective cell suspensions was covered with a round cover slip to let the cells adhere. After 20 min at room temperature, the cover glass was transferred to formaldehyde solution (4%; Sigma-Aldrich) in PBS, for 30 min. After two washes in PBS, coverslips were incubated in 10 mM glycine for 10 min, washed again in PBS, and placed on a droplet of primary antibody (1:500 in PBST), followed by washing and application of the fluorescein-conjugated anti-rabbit secondary antibody (Sigma) in an identical manner. Finally, the cover slip was placed on a microscope glass slide using a 1:1 mixture of Vectashield™ embedding medium (Vector Laboratories, Burlingame, CA) and liquid low-melt agarose at 45°C. Samples were examined using epifluorescence microscopy (with appropriate filters), or high-resolution CLSM.

Cell wall decoration with GFP-CBD500

GFP-CBD500 (Loessner et al., 2002) is a cell wall binding domain polypeptide derived from Listeria phage endolysin Ply500 (Loessner et al., 1995), fused to the GFP. The CBD domain binds with very high affinity to cell wall-linked carbohydrates (teichoic acids) on Listeria cells. The assay was performed by incubation of 100 μl parental cells or l-forms for 10 min in GFP-CBD500 (20 μg ml−1 in PBS) in a total volume of 200 μl, followed by three washes with PBST (0.5%) and epifluorescence microscopy.

RNA extraction from parental and l -form bacteria

Extraction of total RNA was performed using a modified version of the Trizol™ method (Chomczynski and Sacchi, 1987). Parental L. monocytogenes was grown for 16 h in LLM broth at 32°C, diluted 1:50, and incubated for another 2 h at 32°C. Then, 10 ml culture was harvested by centrifugation (4000 g for 1 min at 4°C), the pellet resuspended in 400 μl resuspension buffer (glucose 10% (w/v), 12.5 mM TrisCl, pH 7.6), and transferred to a tube containing 500 μl phenolic acid and 0.4 g glass beads. Cell lysis was performed using a FastPrep™ instrument (Bio 101, USA) by five consecutive runs for 30 s each at a power setting of 6.5, interrupted by cooling on ice for 1 min.

For the l-form cells, the same protocol was applied with the following modifications: a total sample volume of 5 ml was collected from eight LLM soft agar tubes after 6 day incubation, and cell disruption was achieved by a single short run (3 s; power setting 6.5) in the FastPrep™ instrument. Further steps were identical for both RNA preparations.

Cell debris and glass beads were removed by centrifugation (5 min, 12 000 g, 4°C), and the supernatants gently mixed with 1 ml Trizol™ reagent (Invitrogen). After 5 min, RNA fractions were extracted twice using chloroform/isoamylalcohol (24:1 v/v), and precipitated after addition of 500 μl 2-propanol and incubation for 15 min on ice. Total RNA was pelleted by centrifugation (15 min 13 000 r.p.m. at 4°C), washed with 1 ml ethanol 70% for 5 min and dried at 37°C for 15 min. The pellets were resuspended in 50 μl RNase-free water, and incubated for 15 min at 37°C. The quality of the RNA was checked by agarose gel electrophoresis (0.8%), and the material was quantified using a Nanodrop™ spectrophotometer (Nanodrop Technologies, Wilmington, DE).

Macroarray hybridization and whole genome transcriptome analysis

The membranes employed for transcriptome analyses featured spot arrays of complementary sequences reflecting the entire 2816 predicted open reading frames of L. monocytogenes EGDe (Glaser et al., 2001), in addition to 153 additional PCR products specific for genes predicted in L. monocytogenes strain CLIP80459 (serovar 4b) not present in EGDe (Severino et al., 2007). Reverse transcription of bacterial mRNA, labelling and hybridization of cDNA fractions, analysis of macroarray signals, and identification of statistically significant differentially regulated genes in l-forms was performed as described previously (Milohanic et al., 2003). Only genes showing a greater than 2.5-fold up- or down-regulation were considered for data analysis and graphical representation. Functional classification was based upon shown or putative functions according to the ListiList database ( http://genolist.pasteur.fr/ListiList/) (Glaser et al., 2001).

The macroarray data have been deposited in the EMBL-EBI ArrayExpress database (http://www.ebi.ac.uk/arrayexpress) under the accession number A-MEXP-1602.

Reverse transcription real-time quantitative PCR

Reverse transcription of total extracted RNA was performed using TaqMan reverse transcription reagents and random hexamer primers according to the supplier's protocol (Applied Biosystems), using 50 ng μl−1 RNA in a total volume of 100 μl. Aliquots of 20 μl containing approximately 45 ng μl−1 cDNA were stored at −80°C. Real-time quantitative PCR was performed in a Rotor-Gene 6000 Instrument (Corbett Robotics, USA) using 90 ng cDNA as template and 5 μl SYBR™ Green PCR master mix (Applied Biosystems) in a total volume of 10 μl (primers are listed in supporting information Table S3). Amplification and melting curve analysis was performed using the following protocol: pre-incubation (95°C for 10 min); amplification (45 cycles): 10 s at 95°C; 15 s at primer-specific annealing temperature; 20 s at 72°C, with a single fluorescence measurement and melting curve analysis from 50°C to 95°C at 2°C s−1 with continuous measurement. We selected six genes from the total 2969 genes included in the macroarray hybridization: two genes each were chosen as representatives for upregulated and downregulated genes, and two control genes were selected that showed no differential expression. All reactions were run as triplicates, and calculation of initial copy numbers was performed by comparison with a standard curve based on high purity Listeria DNA dilution series with known numbers of templates.

Sequence analysis of genes involved in cell wall division and the bacterial cytoskeleton

Genomic DNA from the Scott A::pPL3-GFP parent was released using endolysin Ply511 (Loessner et al., 1995). This lysate was then treated as the l-form cells, by extraction using phenol/chloroform/isoamylalcohol (25:25:1); concentration and purity of precipitated DNA redissolved in TE Buffer was measured using a Nanodrop spectrophotometer. Five nanograms of DNA of each of the samples was used as template for PCR amplification the genes listed below. Primer pairs (see Table S5) were designed to amplify approximately 150 additional nucleotides upstream and downstream of the targeted sequences. The L. monocytogenes serovar 4b F2365 genome (NC_002973) served as a reference, and locus tags of the analysed genes were LMOf2356–1363 (ispA), LMOf2365–1567 (mreB), LMOf2365–2498 (mreBH), LMOf2356–1738 (mbl), LMOf2365–2065 (ftsA) and LMOf2365–2064 (ftsZ). PCR products were generated using a high fidelity, proofreading polymerase (Phusion Mastermix; Finnzymes, Espoo, Finland) according to the supplier's instructions. The DNA products were sequenced on both strands using the same primers, and data were analysed using CLC Main Workbench 5.0 software.

Acknowledgements

This work was funded by the Swiss National Science Foundation. We would also like to thank Darren Higgins (Harvard University) for providing pPL3eGFP, and Sabine Pilgrim (University of Würzburg) for providing antibodies. We gratefully acknowledge the support of Martin Müller (Electron Microscopy Center of ETH Zurich) with the state-of-the-art EM techniques, and the assistance of Christophe Rusniok (Institute Pasteur) in database submission of the transcriptome array results. This work is dedicated to the late James (Jim) Jay, a great teacher and visionary microbiologist, who has sparked the interest to study Listerial-forms when one of us (MJL) was a graduate student in his lab at Wayne State University more than 20 years ago, and provided valuable advice on how to culture and maintain this unusual form of bacterial life.

Ancillary