l-Arabinose Induces the Formation of Viable Nonproliferating Spheroplasts in Vibrio cholerae

Vibrios spp. among other bacteria form transient cell wall-deficient forms as a response to different stresses and revert to proliferating rods when permissive conditions have been restored. Such cellular forms have been associated with antimicrobial tolerance, chronic infections, and environmental dispersion.

ABSTRACT

Vibrio cholerae, the agent of the deadly human disease cholera, propagates as a curved rod-shaped bacterium in warm waters. It is sensitive to cold but persists in cold waters in the form of viable but nondividing coccoidal-shaped cells. Additionally, V. cholerae is able to form nonproliferating spherical cells in response to cell wall damage. It was recently reported that l-arabinose, a component of the hemicellulose and pectin of terrestrial plants, stops the growth of V. cholerae. Here, we show that l-arabinose induces the formation of spheroplasts that lose the ability to divide and stop growing in volume over time. However, they remain viable, and upon removal of l-arabinose, they start expanding in volume, form branched structures, and give rise to cells with a normal morphology after a few divisions. We further show that WigKR, a histidine kinase/response regulator pair implicated in the induction of high-level expression of cell wall synthetic genes, prevents the lysis of the spheroplasts during growth restart. Finally, we show that the physiological perturbations result from the import and catabolic processing of l-arabinose by the V. cholerae homolog of the Escherichia coli galactose transport and catabolic system. Taken together, our results suggest that the formation of nongrowing spherical cells is a common response of vibrios exposed to detrimental conditions. They also permit us to define conditions preventing any physiological perturbation of V. cholerae when using l-arabinose to induce gene expression from the tightly regulated promoter of the Escherichia coli araBAD operon.

IMPORTANCE Vibrio spp., among other bacteria, form transient cell wall-deficient forms as a response to different stresses and revert to proliferating rods when permissive conditions have been restored. Such cellular forms have been associated with antimicrobial tolerance, chronic infections, and environmental dispersion. The effect of l-Ara on V. cholerae could provide an easily tractable model to study the ability of vibrios to form viable reversible spheroplasts. Indeed, the quick transition to spheroplasts and reversion to proliferating rods by addition or removal of l-Ara are ideal to understand the genetic program governing this physiological state and the spatial rearrangements of the cellular machineries during cell shape transitions.

INTRODUCTION

Cholera is an acute diarrheal disease caused by ingestion of food or water contaminated with Vibrio cholerae, a curved rod-shaped bacterium that propagates in warm briny and salty waters. Cold stops V. cholerae proliferation. However, the bacterium has the ability to persist for months in cold water under the form of coccoid bodies and return to growth when the sea temperature rises (1–3). Similarly, V. cholerae is known to persist under a spherical form in biofilms (4) and to survive exposure to antibiotics inhibiting cell wall synthesis under the form of spheroplasts, i.e., a spherical cell in which the cell wall has been partially or completely removed (5).

The high propagation rate of V. cholerae and its capacity to survive unfavorable growth conditions have led to several pandemics, which have caused and are still causing major socioeconomic perturbations (6). Seven cholera pandemics have been recorded since the beginning of the 18th century. Isolates of the current on-going pandemic, which started over 50 years ago, are rapidly drifting (7–11). It is suspected that the constant appearance of new atypical pathogenic variants of V. cholerae will eventually lead to a more virulent strain that will start a new pandemic, which motivated extensive research on the physiology of the bacterium and its evolution toward pathogenicity (7–11). The use of a tightly regulated high-level expression-inducible system based on the regulation of the promoter of the Escherichia coli araBAD operon by the AraC regulator, known as the P_BAD system (12), played an instrumental role in many V. cholerae studies (13). The AraBAD enzymes allow E. coli to exploit l-arabinose (l-Ara), a component of the hemicellulose and pectin of terrestrial plants, as a carbon and energy source (14). AraC acts both as a positive and a negative regulator, repressing P_BAD in the absence of l-Ara and activating its transcription when bound to it (14). V. cholerae lacks a bona fide arabinose import and metabolization pathway. Nevertheless, the E. coli P_BAD system proved to be very effective in V. cholerae, which suggested that l-Ara was imported in the cytoplasm of the cells. However, we and others recently reported that l-Ara could interfere with the growth of V. cholerae (15, 16), calling for a better understanding of the impact of l-Ara on the physiology of this bacterium.

Here, we show that V. cholerae cells stop dividing or elongating and lose their characteristic curved rod cell shape in the presence of >1% (wt/vol) and >0.1% (wt/vol) l-Ara in rich and poor media, respectively. V. cholerae cells become spherical and morphologically similar to spheroplasts obtained by exposure to cell wall targeting antibiotics (5) or to coccoid bodies formed in cold temperatures (1, 2, 17). We further found that mutants with impaired physiology are more sensitive to the presence of l-Ara, morphologically transitioning to spheroplasts with as little as 0.01% (wt/vol) l-Ara in poor media. We show that the spheroplasts induced by l-Ara are able to revert to exponentially growing rods in only a few generations when l-Ara is removed, demonstrating that they remain viable. Finally, we demonstrate that l-Ara is imported and processed by the V. cholerae homologs of the E. coli galactose transport and catabolic enzymes and that one of its by-products perturbs the physiology of the cell by entering the glycolysis pathway. Taken together, these results suggest that formation of spherical cells might be a general physiological response of vibrios when they are faced with detrimental conditions. From a technical point of view, they permit us to define conditions that allow the use of the P_BAD expression system in V. cholerae while preventing any perturbation of the physiological state of the cells.

RESULTS

l-Arabinose induces the formation of nondividing spherical cells.

To study the effect of l-Ara on cell morphology and growth, we added increasing concentrations of l-Ara to wild-type V. cholerae strain N16961 in the early exponential phase in different liquid media. Cells grown in M9-minimal media (MM) appeared with a wild-type rod shape in the absence or up to 0.02% (wt/vol) l-Ara, but >90% of the cells became spherical in few hours in the presence of 0.1% (wt/vol) l-Ara (Fig. 1A). In cultures grown in M9-MM supplemented with Casamino Acids (CAAs), spherical cells started to appear at 0.2% (wt/vol) l-Ara, and the majority of the cells became spherical at 0.5% (wt/vol) l-Ara. In LB, the concentration of l-Ara had to be increased up to 1% (wt/vol) to induce morphological changes (Fig. 1A). The spherical shape of l-Ara-treated cells resembles that of nonproliferative cells obtained by treating V. cholerae cells with cell wall-targeting antibiotics (5) or by incubating them at 4°C (2, 4), even though at cold temperatures cells appear to be smaller (Fig. 1B).

FIG 1.

l-Ara induces loss of rod shape. (A) Phase contrast images of V. cholerae N16961 cells (strain EPV50) grown at 30°C in the indicated media with increasing concentrations of l-Ara. (B) Phase contrast images of V. cholerae N16961 cells (strain EPV50) incubated in M9-MM at 4°C for 6 weeks (left panel) and grown in LB with 100 μM ampicillin (Amp) at 30°C (right panel). (C) Phase contrast images of V. cholerae strain EGV299 (MCH1 P_BAD::YGFP-slmA) and EGV300 (N16961 P_BAD::ssb-YGFP) cells grown in M9-MM at 30°C in the presence of 0.01% (wt/vol) l-Ara. Scale bars = 2 μm.

In parallel to microscopic inspection, we followed the optical density of cell cultures over time. In all tested media, l-Ara had a detrimental effect on cell proliferation at the same concentrations at which it induced the formation of spherical cells (see Fig. S1 in the supplemental material). Cell growth was inhibited shortly after the addition of l-Ara, at an optical density at which cells would have otherwise kept growing exponentially, indicating that l-Ara induced a rapid metabolic arrest (Fig. S1).

V. cholerae is known to be sensitive to high concentrations of several carbon sources, including glucose (18). None of the 8 other commonly used carbon sources had similar effects on cell shape and growth, including d-arabinose (Table 1; see Fig. S2 in the supplemental material).

TABLE 1.

Carbon sources^a tested for N16961^b rod shape loss and growth inhibition

Carbon source	Attribute in the presence of indicated carbon source
	Growth	Spherical cellsc
d-Arabinose	+	−
l-Arabinose	−	+
l-Rhamnose	+	−
d-Glucose	+	−
d-Galactose	+	−
Glycerol	+	−
d-Sucrose	+	−
d-Xylose	+	−
Succinate	+	−

^aThey were added to M9-MM at a concentration of 0.2% (wt/vol), with the exception of glycerol at 10% (vol/vol).

^bStrain EPV50.

^c+, spherical shape; −, rod shape.

The l-Ara phenotype was not restricted to N16961, a strain isolated in Bangladesh in the 1970s, but it was shared among the most prominent pandemic V. cholerae serotype O1 El Tor strains, such as C6706, isolated in a cholera outbreak in Peru in the 1990s, and E7946, isolated in Bahrain in 1978. A V. cholerae non-O1 non-139 strain was similarly affected by l-Ara exposure (see Fig. S3 in the supplemental material).

Mutants with impaired physiology are more sensitive to l-Ara.

We had previously noticed that it was impossible to use the P_BAD system to produce proteins of interest in a few V. cholerae mutants because the presence of l-Ara inhibited their growth. The above results prompted us to revisit the phenotype of two of those mutants whose growth was completely arrested by the addition of as little as 0.01% (wt/vol) l-Ara in M9-MM, namely, a N16961 strain carrying two copies of the ssb gene, which encodes an essential single-strand DNA binding protein implicated in the regulation of replication, transcription, and homologous recombination repair (19); and a derivative of the MCH1 monochromosomal strain, in which the SlmA nucleoid occlusion protein was overproduced (20). In both cases, we observed that 0.01% (wt/vol) l-Ara was sufficient to induce the formation of spherical cells in the entire cell population (Fig. 1C).

Dynamics of the transition to spherical cells at the population level.

To visually inspect the morphological transition at the population level over time, we collected cell samples every hour for 10 hours after l-Ara addition and examined them under a microscope. Cells were divided in three categories based on their shape, as follows: cells with a rod shape; cells composed of a rod and a small or large irregular bulge protruding from the cell wall, which we refer to as bleb; and cells with a spherical shape (Fig. 2A). In cell cultures grown at 30°C, spherical cells started to appear after 5 hours and comprised 90% of the cell population after 9 hours. The sharp increase in the proportion of spherical cells in the population corresponded to an equally fast decline in rod-shaped cells, which dropped to less than 10% of the cell population at the end of the experiment. Blebbing cells appeared around 4 hours after l-Ara addition. Blebs were randomly located on the surface of the cells. In particular, there was no preference for midcell or cell pole locations (see Fig. S4 in the supplemental material). Cells with protruding blebs never represented more than 5% of the entire cell population and almost disappeared at the end of the time course, which suggested that they corresponded to a transient state between the rod and the spherical state. We refer to them as transitioning cells. Taken together, these results suggest that a few hours are required after growth arrest before morphological transition. However, once transition is initiated, the formation of spherical cells is very fast.

FIG 2.

Dynamics of the transition to spherical cells. Cells of V. cholerae strain EPV50 were grown in M9-MM at 30°C. l-Ara was added at a concentration of 0.2% (wt/vol) when indicated. (A) Kinetics of V. cholerae morphological change from rods to spherical cells. Cell shape was inspected at the microscope every hour after l-Ara addition. A representative image for each cell category (rod, transitioning, and spherical) is represented. The mean of three independent replicates and the standard deviation are represented. (B) Diameter distribution of V. cholerae spherical cells treated with l-Ara for 10 hours. (C) Average diameter of V. cholerae spherical cells over time. Mean of three independent replicates and the standard deviation are represented. (D) Transition from rods to spherical forms. EPV50 cells were mounted on a M9-MM agarose pad containing 0.2% (wt/vol) l-Ara. Bright-field still images from time-lapse microscopy experiments. Images were taken every 5 minutes for 12.5 hours. Scale bars = 2 μm.

In a similar time course experiment performed at 37°C, cells appeared to respond faster to l-Ara. The first blebs appeared 3 hours after l-Ara addition, and after 5 hours, about 40% of the cell population had completed transition to spherical bodies (see Fig. S5 in the supplemental material).

The size of the spherical cells was heterogeneous, and the diameter of the majority of cells was comprised between 1.5 and 1.8 μm (Fig. 2B). The average diameter of the spherical cells did not change between the 7-, 8-, 9-, and 10-hour time points after l-Ara addition, suggesting that once formed, spherical cells neither decreased nor increased in volume (Fig. 2C). Based on the measured dimensions of spherical and rod-shaped cells, we estimate that the cell volume of spherical cells is around 2.5 times bigger than that of exponentially growing rod cells.

Dynamics of the transition to spherical cells at the single-cell level.

We performed time-lapse video-microscopy experiments to inspect the transition process from rod to sphere at the single-cell level (Fig. 2D; see Movie S1 in the supplemental material). All the observed cells displayed the same transition pattern. After exposure to l-Ara, a single bleb appeared at the bacterial cell surface. As the bleb increased in size, the original cell was assimilated into the forming sphere until the original rod shape was completely lost. The time when a bleb became visible on the cell surface and its location varied from cell to cell. However, once started, completion of the morphological change was comparable in all cells, with the exception of a few rare cells that lysed during the process (see Movie S2 in the supplemental material). On an agarose M9-MM pad, blebbing cells transitioned to spheres in around 2 to 3 hours at 30°C (Fig. 2D and Movie S1).

l-Ara-induced spherical cells are cell wall deficient.

The spherical shape of l-Ara-treated cells is similar to cell wall-deficient forms that have completely or almost entirely lost the peptidoglycan (PG) layer, which suggested a process of cell wall degradation or a PG remodeling mechanism (21). To verify this point, we compared the PG content and composition of V. cholerae exponentially growing cells and l-Ara-induced nondividing spherical cells (Table 2). To limit contamination by the PG of cells that had not completely transformed into spheres in the presence of l-Ara, we used MCH1 cells expressing an additional copy of SlmA because they fully transitioned to spherical cells in the presence of as little as 0.01% (wt/vol) l-Ara (Fig. 1C).

TABLE 2.

Results for l-Ara-treated and nontreated V. cholerae cells^a

Parameter	Mean ± SD for V. cholerae cellsc
	Nontreated	l-Ara treated
Muropeptide groupb
Monomers (%)*	48.1 ± 0.6	40.2 ± 3.2
Dimers (%)	43.1 ± 1.9	47.2 ± 3.1
Trimers (%)	8.8 ± 2.5	12.6 ± 4.8
Anhydro-muropeptides (%)*	14.8 ± 2.3	21.9 ± 3.7
DAP-DAP cross-linked muropeptides (%)*	4.9 ± 0.7	7.5 ± 0.9
Peptidoglycan feature
Total cross-linkage (%)*	37.7 ± 1.4	47.2 ± 5.0
Glycan chain length*	11.7 ± 1.4	7.5 ± 1.1

^aStrain EGV217.

^b*, P < 0.05 (t test with a two-tailed distribution).

^cValues are the means and standard deviations of three independent experiments.

The PG of exponentially growing cells and cells that had completely transitioned to spheres after the addition of 0.1% (wt/vol) l-Ara was extracted and submitted to ultraperformance liquid chromatography (UPLC) analysis (see Fig. S6A in the supplemental material). The area of the UPLC profiles showed that the amount of PG per cell in l-Ara-treated bacteria was around 10 times lower than the amount present in rod-shaped cells (Fig. S6B). In addition, we found that glycan chains were twice as short in l-Ara-treated cells than that in untreated cells. The length of glycan chains is calculated based on the number of anhydro-muropeptides (22), which results from the activity of lytic transglycosylases (23, 24). Therefore, the observed reduction in the average glycan chain length suggests an increase of lytic transglycosylase activity in the presence of l-Ara. Finally, we observed that the amount of diaminopimelic acid (DAP)-DAP cross-linked muropeptides significantly increased in l-Ara-treated cells. Taken together, these results indicate that cell wall metabolism is perturbed after l-Ara exposure and that spherical cells are almost completely deprived of the PG layer, which corresponds to the definition of a spheroplast.

l-Ara-induced spheroplasts are viable.

To evaluate if l-Ara had a detrimental effect on cell viability, we estimated the number of viable bacteria in a time course experiment. A culture of N16961 cells was split in two after 2 hours of growth, and l-Ara was added to one of the two halves. The number of viable cells in each culture was determined by plating aliquots onto LB plates and counting the number of colonies. Viable cell count kinetics (represented as CFU) showed that l-Ara addition had an immediate inhibitory effect on cell proliferation but did not cause a corresponding decline in cell viability (Fig. 3A). Indeed, 75% of the number of cells before l-Ara addition gave rise to colonies after a 10-hour incubation with l-Ara (Fig. 3B).

FIG 3.

Recovery of growth and rod shape. Cells were grown in M9-MM at 30°C. l-Ara was added at a concentration of 0.2% (wt/vol) when indicated. Viable colony count (CFU) of V. cholerae cells (strain EPV50) grown with and without l-Ara over time (A) and after 10 hours (B). The mean of three independent replicates and the standard deviation are represented. In the time-lapse experiments, cells were grown in M9-MM + 0.2% (wt/vol) l-Ara until they became spherical and then were mounted on a M9-MM agarose pad in the absence of l-Ara. Bright-field still images were taken every 5 minutes for 14 hours. (C) N16961 cell (strain EPV50) rod shape recovery. (D) Spherical ΔwigKR cells (strain EGV515) are not able to recover rod shape. Scale bars = 2 μm.

Time-lapse video microscopy was performed to determine how nonproliferating spherical cells could return to proliferation and recover a curved rod shape after l-Ara removal. Single-cell analyses showed that reversion to proliferating rods started with the elongation of the spheroplasts, which was followed by the formation of multiple protrusions on their surface (Fig. 3C; see Movie S3 in the supplemental material). The protrusions elongated outward, giving rise to branched cells. Curved rod-shaped cells were recovered after a few division events. The time required to initiate elongation greatly differed from cell to cell; the recovery process started almost immediately after the removal of l-Ara in some cells but took a few hours to initiate in others. However, the time between the initiation of the recovery process and its completion was similar for all the cells. On an agarose M9-MM pad at 30°C, cells transitioned from elongating spheres to symmetrically dividing rods in around 4 to 5 hours. It was not necessary to add osmoprotectants in liquid or on agarose pads to avoid the lysis of spherical cells before and after the initiation of the proliferation recovery process.

Time-lapse video microscopy observations further suggested that the overall 25% loss of CFU after 10 hours of l-Ara treatment (Fig. 3B) was accounted for by the number of cells that lysed during the transition to spheroplasts after the addition of l-Ara (Movie S2) and by those that lysed during the recovery process after l-Ara removal (see Movie S4 in the supplemental material).

wigKR is essential for the recovery of cells after l-Ara treatment.

The histidine kinase/response regulator pair WigKR (also known as VxrAB) is thought to induce a higher expression of the full set of cell wall synthetic genes in response to cell wall damage (25). It was previously reported that WigKR was essential for the recovery of cell shape and the return to proliferation of V. cholerae cells treated with cell wall targeting antibiotics (25), which prompted us to inspect the effect of l-Ara on ΔwigKR cells. We observed no differences in the growth arrest and formation of spherical cells of the ΔwigKR strain in the presence of l-Ara (see Fig. S7A in the supplemental material). As observed before, spherical cell formation initiated from a single bleb randomly distributed on the surface of the cell. However, ΔwigKR spherical cells immediately started to grow in diameter when l-Ara was removed, expanding continuously in size until they exploded (Fig. 3D; see Movie S5 in the supplemental material). On the contrary, ΔwigKR cells that had not yet transitioned to spheroplasts were able to return to a proliferative state without any obvious defect (see Movie 6 in the supplemental material). After a 10-hour incubation with l-Ara, only 10% of the ΔwigKR cells could still form colonies, which corresponded to the proportion of cells that had not started transitioning to spheres (Fig. S7B). Taken together, these results suggest that the histidine kinase/response regulator pair WigKR plays an essential role in the recovery of the l-Ara-induced spheroplasts.

Identification of genes required for l-Ara sensitivity.

In contrast to cell wall-targeting antibiotics, growth arrest and spheroplast formation were unlikely to result from the action of l-Ara on the cell surface or in the periplasm of V. cholerae. V. cholerae lacks a bona fide arabinose import and metabolization pathway, but the effectiveness of the E. coli P_BAD promoter regulation by l-Ara suggested that it was at least passively imported in the cytoplasm of this bacterium. To identify putative factors involved in the response of V. cholerae cells to l-Ara, we performed two complementary genetic screens.

First, we used a screen based on the enrichment of l-Ara-insensitive mutants in a cell population grown in liquid in M9-MM supplemented with 0.2% (wt/vol) l-Ara. We constructed a library of N16961 mutants with a Mariner transposon (Tn), which randomly inserts at 5′-TA-3′ sites (26). We grew the library in M9-MM supplemented with 0.2% (wt/vol) l-Ara for 18 hours. Aliquots were collected before l-Ara addition and after 9 and 18 hours of incubation with l-Ara. Deep sequencing was used to determine all the positions at which the Tn was inserted in the collected libraries. We thus identified nine genetic loci in which Tn insertions were significantly overrepresented after growth in the presence of l-Ara (Fig. 4A). The transposition frequency and insertion profile differed in the overrepresented genes (Fig. 4B). Tn insertions covered both DNA strands of the entire gene length of vc1325, vc1327, vc1328, vc1595, vc1596, and vc0263, suggesting that the product of these genes was implicated in the physiological effect of l-Ara. Tn insertions covered the entire length of a specific DNA strand of vc0262, suggesting that they had a polar effect on the expression of vc0263. The overrepresentation of Tn insertions was restricted to only one of the multiple 5′-TA-3′ sites present in the entire gene length of vc0779 and vc2621, suggesting that the l-Ara-insensitive phenotype could be provided by an additional suppressor mutation located at another genetic locus. The suppressor capacity of the inactivation of vc1325, vc1327, vc1328, vc1595, vc1596, and vc0263 was confirmed by using the corresponding mutants in an ordered mapped Tn library of the wild-type C6706 strain (27). Inactivation of vc1325, vc1327, vc1328, vc1595, vc1596, and vc0263 fully restored growth of C6706 in the presence of l-Ara (see Fig. S8 in the supplemental material).

FIG 4.

Transposon insertion mutants insensitive to l-Ara-induced cell growth arrest. Tn-seq profile of a transposon insertion library in an EPV50 background before (top panel) and after incubation with 0.2% (wt/vol) l-Ara (middle panel for 9 hours and bottom panel for 18 hours). Transposon insertion profile of chromosome 1 is shown in panel A. Chromosomal positions are indicated below. Regions with overrepresented transposon insertions in the presence of l-Ara are shown at the bottom and zoomed in panel B. Transposon insertions in forward and reverse orientations are indicated by orange and blue vertical lines, respectively.

In addition, as l-Ara-insensitive mutants presenting a growth defect cannot be identified with a screen based on growth enrichment, we decided to directly plate a random Tn insertion library on M9-MM plates containing l-Ara. To limit the number of false positives, i.e., the formation of colonies by cells still sensitive to l-Ara, we used the MCH1 strain carrying an additional copy of the slmA gene, which fully transitions to spheres in the presence of as little as 0.01% (wt/vol) l-Ara (Fig. 1C). Plating of the Tn insertion library on M9-MM plates containing 0.1% (wt/vol) l-Ara resulted in the formation of 11 colonies, of which 9 were confirmed to be l-Ara insensitive after reisolation on fresh l-Ara plates. Sequencing showed that they corresponded to Tn insertions in six different genes (Table 3; see Fig. S9 in the supplemental material), of which five had already been detected in the growth enrichment screen (vc1325, vc1327, vc1328, vc1596, and vc0263). In addition, three independent Tn hits were obtained in vc2689. The strain corresponding to the inactivation of vc2689 was missing in the ordered mapped Tn library of the wild-type V. cholerae C6706 strain (27). Therefore, we directly engineered the mutation in the N16961 wild-type background to confirm that its inactivation suppressed the effect of l-Ara (Fig. S8). l-Ara reduced the growth rate of vc2689 mutants (Fig. S8). However, microscopic inspection revealed a majority of rod-shaped cells and a few isolated spherical cells, suggesting that l-Ara sensitivity was reduced even though it was not completely suppressed.

TABLE 3.

Suppressor mutants of l-Ara-induced growth arrest identified in a Tn-based screen^a on plate

Gene	Putative function	No. of hits
vc0263	Galactosyl-transferase	2
vc1325	Galactoside ABC transporter, periplasmic d-galactose/d-glucose binding protein	1
vc1327	Galactoside ABC transporter, ATP-binding protein	1
vc1328	Galactoside ABC transporter, permease protein	1
vc1595	Galactokinase	1
vc2689	6-Phosphofructokinase, isozyme I	3

^aThe screening was performed in V. cholerae strain EGV299.

DISCUSSION

It was recently reported that l-Ara inhibited the proliferation of V. cholerae (15, 16). Here, we show that it is associated with a change in the morphology of the cells from a curved rod shape to a spherical form (Fig. 1). The spherical cells lose the ability to divide and stop growing in volume over time, suggesting a major metabolic arrest (Fig. 2). We found that l-Ara-induced spherical cells are spheroplasts, i.e., they have almost completely lost their cell wall (Table 2 and Fig. S6). Nevertheless, they remain viable, and once l-Ara is removed from the environment, they resume proliferation and revert to the original cell shape after a few divisions (Fig. 3).

Metabolism arrest is linked to the processing of l-Ara by the galactose pathway.

We found that the addition of l-Ara almost immediately stopped V. cholerae proliferation under both fast- and slow-growing conditions, in contrast to the addition of 8 other commonly used carbon sources (Fig. 1 and Table 1). V. cholerae lacks a bona fide l-Ara import and degradation pathway. However, the effectiveness of the regulation of the E. coli P_BAD promoter in V. cholerae suggested that it was at least passively imported in the cytoplasm of the bacterium, where it interfered with the metabolism. We performed 2 genetic screens to determine which cellular processes might be involved in the action of l-Ara.

We identified 7 genes whose inactivation suppressed the sensitivity to l-Ara (Fig. 4 and Table 3). Of those 7 genes, 6 can be directly (vc1325, vc1327, vc1328, vc1596, and vc1596) or indirectly (vc0263) linked to the galactose Leloir catabolic pathway. vc1325, vc1327, and vc1328 code for homologues of the 3 components of the E. coli ABC galactose transport system, the periplasmic binding protein MglB, the ATP-binding protein MglA, and the integral membrane permease MglC, respectively (28, 29). vc1595 codes for a homologue of the E. coli galactokinase GalK, the first enzyme in the Leloir pathway of galactose metabolism (30). vc1596 codes for a homologue of E. coli GalT, the galactose 1-phosphate uridylyltransferase. vc0263 codes for a putative homologue of the enzyme initiating colanic acid synthesis in E. coli, WcaJ (31), which was also described to act as a galactose-1-phosphate transferase in vitro (32). These results suggest that l-Ara is imported in the cytoplasm of V. cholerae by the galactose transporter and processed by the galactose catabolic enzymes (Fig. 5). In Sinorhizobium meliloti, the arabinose transporter AraABC has been described to play a role in galactose uptake (33), suggesting a similarity in the activity of the arabinose and galactose transporter.

FIG 5.

l-Ara-insensitive mutants. Schematic representation of the galactose and glycolytic metabolic pathways. In red are the genes identified in the Tn screens. The number of Tn insertions (hits) obtained in the screen on plate are specified in between parentheses. Depletion or replacement of d-fructose 6-phosphate by a by-product of l-Ara and/or by the poisoning of GlmS could affect lipid II synthesis (in blue) and originate cell wall-deficient cells.

Finally, we identified the 7th suppressor gene vc2689 that codes for a homologue of the E. coli 6-phosphofructokinase PfkA, a key enzyme in the glycolysis pathway (34) (Fig. 5).

Taken together, these results suggest that l-Ara arrests the metabolism of V. cholerae because it is mistakenly recognized as a substrate of the Leloir metabolic pathway and that through a series of enzymatic reactions it is converted into a phosphorylated sugar by-product that cannot be further metabolized (Fig. 5). Likewise, several studies previously suggested that accumulation of a phosphate ester metabolite could perturb growth. l-Ara inhibits the growth of E. coli araD mutants because of the accumulation of l-ribulose 5-phosphate (35). Galactose inhibits the growth of E. coli galT mutants because of the accumulation of galactose 1-phosphate (36, 37). Rhamnose stops the growth of Salmonella enterica serotype Typhi strains defective in the rhamnose degradation pathway because of the accumulation of l-rhamnulose 1-phosphate (35, 38).

l-Ara-mediated metabolic perturbation does not prevent the use of P_BAD.

Importantly, the realization that l-Ara can perturb the metabolism of V. cholerae does not jeopardize previous results obtained with the E. coli P_BAD expression system in this bacterium since l-Ara concentrations lower than those that promote growth arrest and spheroplasts formation are almost always used (Fig. 1). However, our study indicates that special care should be taken in future work when using the E. coli P_BAD expression system in mutants of V. cholerae (Fig. 1). It also shows how metabolic artifacts linked to l-Ara can be avoided by performing experiments in cells that can import l-Ara but are insensitive to it by mutation of vc1595, which codes for the first enzyme probably processing l-Ara in the cytoplasm (Fig. 4 and 5).

Spheroplast formation results from an imbalance in cell wall degradation and synthesis.

The analysis of the muropeptide composition of l-Ara-treated V. cholerae cells showed that they still maintained a residual amount of PG whose structure was remarkably similar to that described for E. coli cefsulodin-induced L-forms (39). The dramatic decrease in the average chain length and corresponding increase in anhydro-muropeptides hint to a higher activity of lytic transglycosylases in cleaving the PG and producing shorter chains. The increase in DAP-DAP cross-linkage, an unusual kind of cross-linkage specifically generated by l,d-transpeptidases (40), further suggests that the metabolic arrest induced by l-Ara affects penicillin binding protein (PBP) activity and stimulates transpeptidation mediated by the l,d-transpeptidase LdtA (41). Taken together, these results suggest that l-Ara promotes the formation of spheroplasts because it induces a metabolic arrest that leads to an imbalance between PG synthesis and degradation. Interestingly, one of the suppressors we identified in the Tn screenings (vc2689) codes for a homologue of the glycolytic enzyme PfkA, whose normal metabolic substrate phosphor-sugar fructose 6-phosphate is an essential precursor for UDP-GlcNAc and subsequently lipid II and PG synthesis through the GlmS pathway (42) (Fig. 5). It suggests that the imbalance between PG synthesis and degradation might result from the depletion or replacement of phosphor-sugar fructose 6-phosphate by a by-product of l-Ara and/or by the poisoning of GlmS.

WigKR-dependent upregulation of cell wall synthesis genes is essential for recovery.

Like l-Ara, antibiotics inhibiting cell wall synthesis promote the formation of viable spheroplasts that can revert to the original cell shape after a few divisions when the antibiotics are removed (5). However, in contrast to l-Ara-induced spheroplasts, V. cholerae cells exposed to antibiotics inhibiting cell wall synthesis grow in volume over time, suggesting that they are not dormant (Fig. 3) (43, 44). In this regard, antibiotic-treated cells are more similar to bacterial L-forms, which can be obtained in osmotic stabilizing media in several microorganisms, including E. coli, by treating cells with lysozyme (45), by adding the β-lactam cefsulodin (a specific inhibitor of the penicillin binding proteins PBP1A and PBP1B) (39, 46) or by inhibiting synthesis of the lipid II cell wall precursor with fosfomycin (47).

The spheroplasts induced by the treatment of ΔwigKR cells by l-Ara increased in volume when l-Ara was removed from the growth media, demonstrating that cell metabolism was very rapidly restored (Fig. 3D). However, the expansion in volume of the cells led to lysis (Fig. 3D). These results indicate that the recovery of a constitutive level of PG synthesis was not sufficient to expand the residual amount of cell wall left in the spheroplasts to accommodate the increase in cellular material, as observed for the spheroplasts induced by cell wall-targeting antibiotics. These results fit with the idea that the histidine kinase/response regulator pair WigKR-dependent upregulation of the full set of cell wall synthetic genes is necessary for cell shape recovery of V. cholerae spheroplasts (25).

l-Ara sensitivity is a common feature of Vibrio.

l-Ara-induced spheroplast formation and inhibition of cell growth are observed in clinically relevant O1 and O139 strains as well as in environmental non-O1 non-O139 strains (Fig. S3) (15).

Interestingly, it was shown that l-Ara induced biofilm formation and had an inhibitory effect on cell growth in Vibrio fischeri (48). No remarks were made about cell morphology. However, mutations in GalK or the galactose transporter were found to suppress the phenomenon, as observed for the action of l-Ara in V. cholerae (48). Thus, l-Ara sensitivity could be a phenotype shared by different species of vibrios.

MATERIALS AND METHODS

Plasmids and strains.

Bacterial strains and plasmids used in this study are listed in Table 4. Strains were rendered competent by the insertion of hapR by specific transposition and constructed by natural transformation. Engineered strains were confirmed by PCR. Primers are listed in Table S1 in the supplemental material.

TABLE 4.

List of bacterial strains and plasmids used in this study

Strain or plasmid	Relevant genotype or feature	Reference
Strains
V. cholerae 104154	Wild-type strain, non-O1 non-O139 serogroup	Laboratory collection
V. cholerae C6706	Wild-type strain, O1 serogroup	Laboratory collection
V. cholerae E7946	Wild-type strain, O1 serogroup	Laboratory collection
V. cholerae EGV217	MCH1 hapR+ ΔlacZ::(PBAD::YGFP-slmA-Sh ble) zeoR, gmR	This study
V. cholerae EGV299	MCH1 hapR+ ΔlacZ::(PBAD::YGFP-slmA-lacZ-Sh ble) zeoR, gmR	This study
V. cholerae EGV300	N16961 hapR+ ΔlacZ::(PBAD::ssb-YGFP-lacZ-Sh ble) zeoR, gmR	This study
V. cholerae EGV515	N16961 hapR+ ΔlacZ wigKR::aadA specR, gmR	This study
V. cholerae EPV50	N16961 hapR+ ΔlacZ gmR	55
V. cholerae VC0263	C6706 with Tn-inactivated vc0263, kanR	27
V. cholerae VC1325	C6706 with Tn-inactivated vc1325, kanR	27
V. cholerae VC1327	C6706 with Tn-inactivated vc1327, kanR	27
V. cholerae VC1328	C6706 with Tn-inactivated vc1328, kanR	27
V. cholerae VC1595	C6706 with Tn-inactivated vc1595, kanR	27
V. cholerae VC1596	C6706 with Tn-inactivated vc1596, kanR	27
V. cholerae VC2689	N16961 with Tn-inactivated vc2689, kanR	This study
E. coli SM10 λpir	thi-1 thr leu tonA lacY supE recA::RP4-2-Tc::Mu, λpir, kanR	56
Plasmids
pEG258	PBAD::YGFP-slmA-Sh ble flanked by the upstream and downstream regions of lacZ; ori pUC; zeoR ampR	This study
pEG348	PBAD::YGFP-slmA-lacZ-Sh ble flanked by the upstream and downstream regions of lacZ; ori pUC; zeoR ampR	This study
pEG352	PBAD::ssb-YGFP-lacZ-Sh ble flanked by the upstream and downstream regions of lacZ; ori pUC; zeoR ampR	This study
pEG431	aadA flanked by the upstream and downstream regions of wigKR; ori p15a; cmR specR	This study

Growth curves.

If not otherwise indicated, cells were grown at 30°C in M9 minimal medium supplemented with 0.2% (wt/vol) fructose and 1 μg/ml thiamine (M9-MM), M9-MM + 0.1% Casamino Acids (M9-MM + CAA), and Luria-Bertani broth (LB) in a 96-well microtiter plate, and the optical density at 600 nm (OD₆₀₀) was followed over time in a Tecan plate reader. The growth curves plotted are the average of three replicates; the standard deviation is represented for each time point. For CFU and rod-to-sphere kinetics, cells were grown in flasks in M9-MM at 30°C and 37°C; 0.2% (wt/vol) l-Ara was added when indicated. Samples were taken every hour for plating and/or microscopic inspection. Three replicates were performed for each experiment. l-Ara was added to cell cultures with an OD₆₀₀ between 0.02 and 0.05.

l-Ara survival assay.

Overnight wild-type (EPV50) and ΔwigKR (EGV515) cultures were diluted 200 times in M9-MM, followed by 2 hours of growth at 30°C before 0.2% (wt/vol) l-Ara was added. Cells were checked for transition to spherical morphology under the microscope. Serial dilutions of T₀ (before l-Ara addition) and T₁₀ (after l-Ara treatment) samples were plated onto LB plates; afterward, the number of colonies was determined and used to calculate the CFU at T₀ and T₁₀. The ratio CFU T₁₀/CFU T₀ was used to calculate the percentage of cells able to survive l-Ara treatment and revert to proliferation.

Microscopy.

Cells were spread onto a 1% (wt/vol) agar pad (ultrapure agarose, Invitrogen) for analysis. For snapshots, images were acquired using a DM6000-B (Leica) microscope. For time-lapse analyses, the agarose pad was made using M9-MM with 0.2% (wt/vol) l-Ara if needed, and images were acquired using an Evolve 512 EMCCD camera (Roper Scientific) attached to an Axio Observe spinning disk (Zeiss). To observe rod-to-sphere transition on agarose pads, 0.2% (wt/vol) l-Ara was added in liquid M9-MM cultures 2 hours before transferring cells on agarose pads containing l-Ara and starting microscopic imaging.

Transposon insertion deep sequencing-based screen.

The Tn library was constructed in an EPV50 background, and libraries for Illumina sequencing were prepared as described by Espinosa et al. (49). Aliquots of the transposon library were thawed on ice, and ∼10⁹ cells were diluted into 100 ml of M9-MM and grown for 1 hour at 30°C before adding 0.2% (wt/vol) l-Ara. The culture was incubated at 30°C in a shaking incubator for an additional 18 hours. Samples for library construction and deep sequencing were collected before l-Ara addition and after 9 and 18 hours of incubation with l-Ara.

Transposon insertion sequencing analysis.

Cutadapt was used to remove adapters and transposon sequences. Genome sequences were mapped using BWA as described in references 50 and 51. Transposon insertions were visualized using the Artemis browser (52).

Transposon mutagenesis screen on plate.

The transposon mutagenesis was performed by conjugating the E. coli strain SM10 λpir/pSC189, which carries a mini-Himar transposon associated with kanamycin resistance, with the V. cholerae strain EGV299. In detail, 1 ml of culture of EGV299 grown to an OD₆₀₀ of 0.3 was mixed with 100 μl of donor strain SM10 λpir carrying the transposon donor plasmid pSC189 grown to an OD₆₀₀ of 0.5. Each mixture was pelleted, resuspended, and deposited onto a 0.45-μm filter (Millipore) on a LB agar plate supplemented with DAP. Conjugation was carried out for 6 hours at 37°C, and then cells were pooled and plated directly onto M9-MM plates containing kanamycin and 0.1% (wt/vol) l-Ara and incubated over night at 30°C. Eight conjugations were performed per each library. We constructed two Tn libraries of approximately 300,000 clones each. Mutants able to grow on a plate were isolated and inspected under the microscope for growth and morphology in the presence of l-Ara. An arbitrary PCR followed by DNA sequencing were performed to determine the Tn insertion site.

Peptidoglycan analysis.

EGV217 overnight cultures were diluted 200 times in M9-MM and grown at 30°C. A total of 100 ml was centrifuged after 7 hours of growth and 1 liter of culture, to which 0.2% (wt/vol) l-Ara was added after 2 hours, and was harvested after a further 7 hours of incubation. Cells were checked for complete transition to spherical morphology under the microscope before harvesting. Previously described methods were followed for muropeptide isolation and ultra-performance liquid chromatography (UPLC) analysis (53, 54). After cell pellets with sodium dodecyl sulfate (SDS) were boiled for 2 hours, the lysates were left stirring overnight at room temperature. Cell wall material was pelleted, washed with Milli-Q (MQ) water to remove the SDS, and digested with pronase E to remove Braun’s lipoprotein. Purified peptidoglycan was resuspended in MQ water and treated overnight with muramidase at 37°C. Soluble muropeptides were reduced with sodium borohydride, and the pH was then adjusted to 3.5 with phosphoric acid. Samples were injected in a UPLC system to obtain the muropeptide profiles. UPLC separation was performed on a Waters UPLC system equipped with an Acquity UPLC BEH C₁₈ column (130 Å; 1.7 μm; 2.1 mm by 150 mm; Waters) and a dual-wavelength absorbance detector using a linear gradient from buffer A (50 mM phosphate buffer [pH 4.35]) to buffer B (50 mM phosphate buffer [pH 4.95] and 15% [vol/vol] methanol) in a 28-min run with a 0.25 ml/min flow. Elution of muropeptides was detected at 204 nm. Identity of the peaks was assigned by comparison of the retention times and profiles to other chromatograms in which mass spectrometry data have been collected. The relative amounts of the muropeptides and the percentage of cross-linkage were calculated as described by Glauner et al. (22). To estimate the amount of peptidoglycan per cell, the total area of the chromatogram was normalized to the OD of the culture. All values are the means of three independent experiments.

Data availability.

Transposon insertion sequencing (Tn-seq) data are available in the ArrayExpress database (https://www.ebi.ac.uk/arrayexpress) under accession number E-MTAB-9747.