Tropical Strains of Ralstonia solanacearum Outcompete Race 3 Biovar 2 Strains at Lowland Tropical Temperatures

Abstract

Bacterial wilt, caused by members of the heterogenous Ralstonia solanacearum species complex, is an economically important vascular disease affecting many crops. Human activity has widely disseminated R. solanacearum strains, increasing their global agricultural impact. However, tropical highland race 3 biovar 2 (R3bv2) strains do not cause disease in tropical lowlands, even though they are virulent at warm temperatures. We tested the hypothesis that differences in temperature adaptation and competitive fitness explain the uneven geographic distribution of R. solanacearum strains. Using three phylogenetically and ecologically distinct strains, we measured competitive fitness at two temperatures following paired-strain inoculations of their shared host, tomato. Lowland tropical strain GMI1000 was only weakly virulent on tomato under temperate conditions (24°C for day and 19°C for night [24/19°C]), but highland tropical R3bv2 strain UW551 and U.S. warm temperate strain K60 were highly virulent at both 24/19°C and 28°C. Strain K60 was significantly more competitive than both GMI1000 and UW551 in tomato rhizospheres and stems at 28°C, and GMI1000 also outcompeted UW551 at 28°C. The results were reversed at cooler temperatures, at which highland strain UW551 generally outcompeted GMI1000 and K60 in planta. The superior competitive index of UW551 at 24/19°C suggests that adaptation to cool temperatures could explain why only R3bv2 strains threaten highland agriculture. Strains K60 and GMI1000 each produced different bacteriocins that inhibited growth of UW551 in culture. Such interstrain inhibition could explain why R3bv2 strains do not cause disease in tropical lowlands.

INTRODUCTION

Ralstonia solanacearum, which causes bacterial wilt disease of many crops, is found on six continents, and its hosts include plants in over 50 dicot and monocot families. The pathogen forms a heterogenous species complex that contains thousands of distinct strains (1, 2). No single strain can attack all known hosts, but most members of the species complex can wilt tomato. The best way to manage bacterial wilt is by planting resistant crop varieties, but the high variability of the pathogen complicates breeding efforts because wilt resistance is often strain specific (3, 4). Sequencing over 20 diverse strains revealed that R. solanacearum has a fluid genome with evidence of extensive horizontal gene transfer (5–7). However, the specific mechanisms by which this pathogen has adapted to diverse environments and hosts are unknown. For example, most strains of R. solanacearum are tropical or subtropical, but one highly adapted group, known historically and for regulatory purposes as race 3 biovar 2 (R3bv2), causes economically important wilt of both potato and tomato in cool highland tropics (8).

R. solanacearum strains fall into four genetically distinct phylotypes that correspond to their geographic origin: phylotypes I (Asia), II (the Americas), III (Africa), and IV (Indonesia) (1) However, because of human activity, multiple phylotypes are now commonly present in a single region. Recent surveys found strains from two or three phylotypes in Guatemala, Florida, Reunion Island, and Cameroon (9–12). Despite this homogenization, highland tropical R3bv2 strains and warm temperate or tropical strains are not known to cause disease in the same plants or even in the same fields. Field surveys have not found tropical lowland or warm temperate strains in the cool temperate tropical highlands where R3bv2 is endemic (10, 11). Conversely, R3bv2 has not been isolated from wilting plants in tropical lowland regions, even though these strains frequently enter tropical zones via infected plant material and contaminated surface water (13).

An obvious explanation for the failure of R3bv2 strains to proliferate in the lowland tropics is that they cannot grow or cause disease effectively at high temperatures. However, we previously found no difference in either growth rate or virulence between R3bv2 strain UW551 and a tropical or a subtropical strain at the tropical lowland temperature of 28°C (14, 15). In contrast, the tropical strains were significantly less virulent than R3bv2 at cool temperatures. It appears that UW551 has adapted to growth and virulence at cool temperatures without losing its ability to thrive under warmer conditions.

These findings raise several questions. If R3bv2 strains are highly virulent at tropical temperatures, why do they not cause disease in the lowland tropics? This is especially puzzling since R3bv2 strains can overcome the bacterial wilt resistance in most commercial tomato varieties, which should give them a competitive advantage over the lowland strains. Why are strains from multiple phylotypes not isolated from the same host plant in regions where they co-occur? How do temperature and intraspecific competition affect the ability of co-occurring R. solanacearum strains to cause disease?

We hypothesized that R3bv2 is not found in warm lowland sites because of differences in competitive fitness among R. solanacearum strains in those environments. We predicted that lowland tropical R. solanacearum strains have a competitive advantage over tropical highland strains in a warm environment, while highland strains are more competitive at cooler temperatures. Such differences would lead to niche exclusion of nonadapted strains and could explain the observed differences in geographic distribution of strains. We tested this hypothesis with paired-strain competition experiments at two temperatures in planta and in culture. GMI1000 was isolated from tomato in the lowland tropics (16). K60, the R. solanacearum type strain, was isolated from tomato in warm temperate North Carolina (17, 18). UW551 is a typical R3bv2 strain originally isolated from geraniums grown in the cool highlands of Kenya (19). It is highly virulent on both potato and tomato plants at temperatures ranging from 18°C to 28°C (15, 20).

Under cool temperate conditions, highland tropical strain UW551 was most competitive in rhizospheres of susceptible tomato plants. At a tropical lowland temperature, however, the warm temperate and tropical lowland strains outcompeted UW551 in both rhizospheres and stems. These more competitive strains secreted proteins that specifically inhibited growth of UW551.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

The bacterial strains used in this work are listed in Table 1, along with their relevant characteristics. R. solanacearum strains were cultured on Casamino Acid-peptone-glucose broth medium (CPG) (21) or on CPG plates containing TZC (2,3,5-triphenyl tetrazolium chloride) (22) at 28°C, unless otherwise stated. Boucher's minimal medium (BMM) was used where defined medium was required (23). Pectobacterium carotovorum, Xanthomonas campestris pv. vesicatoria, X. campestris pv. campestris, Agrobacterium tumefaciens, Escherichia coli, and Bacillus subtilis were grown on Luria-Bertani (LB) (24) medium at 37°C or 28°C. Pseudomonas fluorescens and Pseudomonas aureofaciens were grown at 28°C on King's B medium (25). When necessary, antibiotics were added to the medium at the following concentrations: rifampin, 25 mg/liter, and tetracycline, 15 mg/liter. Stock inoculum for all strains was prepared from 24-h cultures grown with shaking at 225 rpm at 28°C or 37°C according to the optimal growth temperature of the strain. Bacteria were pelleted and washed with sterile Milli-Q (SMQ) water, repelleted, and resuspended in SMQ water to an optical density at 600 nm (OD₆₀₀) of 0.2 and further diluted to the desired final cell density.

TABLE 1.

Bacterial strains used in this study, with relevant characteristics

Strain	Phylotype/sequevara	Race/biovarb	Geographic origin	Isolation host	Reference
Ralstonia solanacearum (alternate name)
K60 (UW25)	IIA/7	1/1	North Carolina	Tomato	22
K60 Rifr	IIA/7	1/1	North Carolina	Tomato	30
K60 Tetr	IIA/7	1/1	North Carolina	Tomato	30
UW551	IIB/I	3/2	Kenya	Geranium	20
UW551 Tetr	IIB/I	3/2	Kenya	Geranium	58
GMI1000 (UW643)	I/18	1/3	French Guyana	Tomato	23
GMI1000 Rifr	I/18	1/3	French Guyana	Tomato	14
Pectobacterium carotovorum subsp. carotovorum WPP14	NAc	NA	Wisconsin	Potato	59
Xanthomonas campestris pv. vesicatoria	NA	NA			Allen laboratory collection
Xanthomonas campestris pv. campestris	NA	NA			Allen laboratory collection
Agrobacterium tumefaciens	NA	NA			Allen laboratory collection
Escherichia coli (DH5α)	NA	NA			60
Bacillus subtilis	NA	NA			Allen laboratory collection
Pseudomonas fluorescens	NA	NA			Allen laboratory collection
Pseudomonas aureofaciens	NA	NA			Allen laboratory collection

^aPhylotype corresponds to strain phylogeny and geographic origin; sequevar clusters strains by phylogenetic analysis of partial endoglucanase gene sequence.

^bRace is a historic designation with little predictive value, based loosely on host range. Strains are divided into biovars based on ability to oxidize a panel of disaccharides and hexose alcohols.

^cNA, not applicable.

Growth curves.

Growth rates of the four antibiotic-resistant R. solanaceraum strains used in this study, GMI1000 Rif^r, K60 Tet^r, K60 Rif^r, and UW551 Tet^r, were measured and compared to those of their wild-type (WT) parents, GMI1000, K60, and UW551. Bacterial growth was measured by OD₆₀₀ using a Synergy/HT microplate reader (BioTek Instruments, Winooski, VT). Two hundred microliters of a bacterial culture at an OD₆₀₀ of 0.01 was aliquoted into a 96-well sterile microplate in triplicate and with rapid shaking. Optical density was measured every hour for 94 h. Cultures were grown at a constant 28°C, simulating a tropical lowland environment, or at cycling temperatures of 24°C for 14 h and 19°C for 10 h (24/19°C), simulating a tropical highland environment.

CPG broth competition assay.

The competitive fitness of strains GMI1000, K60, and UW551 was measured in liquid medium by inoculating 50 ml of CPG in 125-ml Erlenmeyer flasks with a 1:1 bacterial suspension of either GMI1000 Rif^r and UW551 Tet^r, GMI1000 Rif^r and K60 Tet^r, or UW551 Tet^r and K60 Rif^r at a final concentration of ∼2.0 × 10⁵ CFU/ml of medium (1 × 10⁵ CFU/ml of each competing strain). Cultures were shaken at 225 rpm, and bacterial populations were measured after 48 h by serial dilution and plating onto selective media. Plates were incubated at 28°C, and colonies were counted after 48 h. All three paired strain combinations were grown in triplicate at temperatures simulating tropical lowland and tropical highland conditions as indicated above.

Sampling bacterial populations in rhizospheres and stems.

Unwounded tomato plants (bacterial wilt-susceptible cv. Bonny Best) were inoculated via a naturalistic soil soak as previously described (26). Briefly, 50 ml of a bacterial suspension containing a 1:1 mixture of two competing R. solanacearum strains, each at a concentration of 2 × 10⁷ CFU/ml, was poured over the soil of 16-day-old unwounded tomato plants. Competing strain combinations were as described above, and each pairwise combination was performed on at least 45 plants under both temperature conditions. At the first sign of wilt symptoms (1 to 25% of the leaves displaying wilt symptoms), bacterial population sizes were quantified in the stem and rhizosphere of each tomato plant. Rhizospheres were sampled by excising the entire root system from each plant. These were shaken to remove loosely adhering soil, weighed, placed in a 50-ml conical test tube, and vortexed in 10 volumes of sterile water for 1 min. The supernatant was serially diluted and plated onto TZC plates containing the relevant selective antibiotic. Bacterial population sizes in tomato stems were measured by grinding an ∼100-mg transverse slice of tomato stem harvested at the height of the first leaf petiole in 1 ml of sterile water. The supernatant was serially diluted and plated onto selective media as described above. Colonies were counted after 48 h of incubation at 28°C. R. solanacearum populations in each sample were normalized to log CFU per gram of plant tissue + 1.

In planta fitness of individual strains.

We determined the fitness of each individual strain in the absence of competition in tomato plant rhizospheres and stems at both tropical and cool temperatures by inoculating tomato plants with one of the three competing strains as described above. Bacterial population sizes in the rhizospheres and stems were measured as described above. Each treatment included 10 plants.

CI values.

Competitive index (CI) values were calculated as previously described (27). Briefly, the proportion of the two bacterial populations recovered in a paired-strain competition assay was divided by the proportion of the same two bacterial populations in the initial inoculum for that assay. The CI value was calculated individually for each rhizosphere and stem sample and for all three paired-strain combinations. A CI value of 1 indicates no competitive advantage of one strain over another. A CI value greater than one indicates an advantage of the strain in the numerator compared to that in the denominator, and a CI value less than one indicates a competitive disadvantage for the strain in the numerator compared to the strain in the denominator.

Inhibition activity assays.

A qualitative assay for antimicrobial activity was adapted from a previously described overlay plate growth inhibition assay (28). R. solanacearum strains were tested for the ability to inhibit growth of other strains from the same species and selected non-R. solanaceraum strains. The overlay consisted of molten medium inoculated with a target bacterial strain to a final concentration of 2 × 10⁵ CFU/ml. This inoculated overlay was vigorously vortexed for 30 s, and 20-ml aliquots were quickly poured onto petri plates containing 20 ml of the solid medium on which the target strain was inoculated. After the agar solidified, 3-mm-diameter wells were created in each plate with a sterile no. 1 cork borer. Each well was then filled with 25 μl of culture supernatant from a test strain. Supernatants were prepared by centrifuging an overnight broth culture of the test strain at 6,000 × g for 10 min and passing the supernatant through a sterile 0.2-μm-pore-size cellulose acetate syringe filter. The overlay inhibition plates were incubated at either 24/19°C or 28°C. Plates were photographed at 48 h postinoculation. Zones of inhibition were measured using ImageJ (http://imagej.nih.gov). Each assay was replicated at least three times, with two technical replications for each bioreplicate.

Characterization of inhibitory activity in culture supernatants.

Supernatants from R. solanacearum strains GMI1000, UW551, and K60 were obtained from 50-ml overnight cultures grown in CPG medium as described above. Supernatants were passed through a 10-kDa Amicon centrifugal filter (Millipore Corporation, Billerica, MA) following the manufacturer's instructions. Filtrate (<10 kDa) and concentrate (>10 kDa) were each collected and used immediately. Fifty-microliter quantities of both filtrate and concentrate were heat treated for 10 min at 65°C. Proteinase K was added to supernatant filtrate or concentrate to a final concentration of 50 μg/ml and incubated at 37°C for 1 h. Untreated supernatant from each of the three strains was used as a control. Overlay plate inhibition activity assays were then performed as described above.

Statistical analysis.

One-way analysis of variance (ANOVA) was used to analyze data from in planta fitness assays of individual strains, inhibition activity assays, and growth curves. Data describing bacterial population sizes recovered from paired-strain combinations in plants were analyzed using a Wilcoxon signed-rank test (29). Statistical analyses were carried out using JMP, version 9 (SAS Institute Inc., Cary, NC).

RESULTS

We assessed the competitive interactions among three phylogenetically and ecologically distinct R. solanacearum strains in the contexts of varying temperature and a shared plant host, tomato. We measured the ability of each strain to multiply in culture and in planta first in the absence of competition and then in paired-strain competition assays in culture and in planta. Because temperature can affect competitive interactions, each experiment was conducted at both a typical lowland tropical temperature of 28°C and a highland tropical regimen of 24°C day/19°C night.

Antibiotic resistance markers did not decrease strain fitness.

Each strain was marked with antibiotic resistance genes to allow quantification of the strains in planta and in paired-strain competition studies. There were no significant differences between growth rates of the antibiotic-resistant strain and its wild-type parent over a 144-h period in minimal broth at either 24/19°C or 28°C, indicating that the antibiotic resistance genes did not cause a measurable fitness loss (data not shown). R. solanacearum strains carrying the spontaneous rifampin resistance used in this study retain wild-type bacterial wilt virulence on tomato (30, 31).

In culture, UW551 grew better than GMI1000 and K60 at cool temperatures but lost its advantage under tropical conditions.

To assess the effect of temperature in the absence of competition, we compared growth curves of all three strains in rich CPG broth at either 24/19°C or 28°C. As expected, at cool temperatures R3bv2 strain UW551 grew faster than tropical lowland strain GMI1000 or warm temperate strain K60 (P < 0.0001, repeated-measures ANOVA) (Fig. 1). K60 and GMI1000 had similar growth rates (P = 0.0604, t test). Under tropical conditions UW551 lost its advantage, with a growth curve indistinguishable from that of K60 at 28°C (P = 0.0667) (Fig. 1, bottom). Interestingly, at 28°C GMI1000 grew more slowly than UW551 and K60 (P < 0.0001, repeated-measures ANOVA).

FIG 1.

Growth of ecologically distinct R. solanacearum strains in culture without competition. Rich CPG broth was inoculated with strain GMI1000, K60, or UW551 to a final OD₆₀₀ of 0.01 and cultured with shaking for 36 h in a 96-well plate at either 24/19°C (top) or 28°C (bottom). The OD₆₀₀ of the samples was determined every 30 min by a BioTek plate reader. At 28°C tropical lowland strain GMI1000 grew significantly more slowly than UW551 and K60 (P < 0.0008, one-way ANOVA). At the cooler temperatures strain UW551 grew more quickly and reached a higher final optical density than either K60 or GMI1000 (P < 0.0001, one-way ANOVA). Data represent the means of the results from three technical replicates in a representative experiment.

Without competition, all three strains grew similarly in planta at both temperatures.

R. solanacearum strains generally invade host plant roots from the soil and then colonize and spread through the xylem elements of the plant's vascular system (7, 32). We therefore focused on two plant-associated environments where competing pathogen strains interact: the rhizosphere, known to be a microbially diverse and highly competitive niche (33, 34), and the mid-stem, where R. solanacearum strains can reach very high population sizes and interact with a relatively small number of other xylem-inhabiting microbes (35–37). Any growth differences between strains in planta could result either from competitive interactions between strains or from innate differences in ability to colonize tomato rhizospheres and stems. To separate these traits, we first determined the reproductive success of each strain growing alone in plants following a naturalistic soil-soak inoculation and incubation under either warm or cool conditions. Bacteria were isolated from rhizospheres and stems of tomato plants when wilt symptoms first appeared. The three strains reached indistinguishable population sizes of around 10⁷ CFU/g in rhizospheres of tomato plants incubated at 24/19°C (Fig. 2) (P = 0.8775). The stems of these plants contained over 10⁸ CFU/g of each strain, with UW551 colonizing stems slightly better than the other two strains under the cool condition (Fig. 2). At 28°C, the three strains reached comparable cell densities of around 2.5 × 10⁸ CFU/g in tomato rhizospheres (P = 0.8739, one-way ANOVA) and 2.6 × 10⁹ CFU/g in stems (P = 0.4836, one-way ANOVA).

FIG 2.

Growth of ecologically distinct R. solanacearum strains in tomato stems and rhizospheres under cool and warm conditions without competition. Sixteen-day-old unwounded tomato plants were inoculated via a naturalistic soil soak with a bacterial suspension of R. solanacearum strain GMI1000, K60, or UW551 at a concentration of 6.0 × 10⁶ CFU/g of soil. At the first sign of wilt symptoms, bacterial population sizes in rhizospheres and stems were determined by serial dilution plating. There were no significant differences between group mean population sizes in tomato rhizospheres at 24/19°C (top) and 28°(bottom) and in tomato stems at 28°C, as determined by one-way ANOVA. In tomato stems at 24/19°C, strain UW551 populations were significantly higher than those of GMI1000 and K60, as indicated by an asterisk (F = 4.34; P = 0.02; n = 10).

Strain competitive fitness differed in culture, with warm temperatures favoring K60 and cooler temperatures favoring R3bv2 strain UW551.

To determine the effect of temperature on bacterial strain interactions in the absence of a plant host, we conducted paired-strain competition assays in vitro under both temperature regimens. Bacterial populations were enumerated after 48 h of growth in CPG rich broth inoculated with equal numbers of each competing strain. At the cooler temperatures in culture, GMI1000 was outcompeted by both K60 (competitive index value [CI] = 2.65) and UW551 (CI = 6.17) (Fig. 3; Table 2). There was no significant difference in competitiveness between UW551 and K60 at cool temperatures, although the mean population size of K60 was slightly larger than that of UW551 (1.08 × 10⁶ CFU/ml and 7.06 × 10⁵ CFU/ml, respectively) (Table 2). At 28°C, warm temperate strain K60 easily outcompeted both tropical strain GMI1000 (CI = 6.37 × 10⁴) and R3bv2 strain UW551 (CI = 8.57 × 10⁴). UW551 was outcompeted to a lesser degree at 28°C by tropical strain GMI1000 (CI = 1.74 × 10⁻¹) (Fig. 3; Table 2). Overall, warm temperate strain K60 dominated in culture, outcompeting both other strains at 28°C and GMI1000 at 24/19°C, while matching UW551 at the cooler temperatures.

FIG 3.

Competitive fitness of ecologically distinct R. solanacearum strains growing in coinoculated culture at cool and warm temperatures. Fifty milliliters of CPG broth was inoculated with a 1:1 bacterial suspension containing 3 × 10⁵ cells each of strains GMI1000 and K60, GMI1000 and UW551, or K60 and UW551. The population size of each strain was determined by serial dilution plating after 48 h of shaking incubation at either 24/19°C (top) or 28°C (bottom). Results are the means of three biological replicates. Asterisks indicate different mean population sizes (P < 0.01); for detailed statistical analysis, see Table 2.

TABLE 2.

Results of paired-strain competitions in culture with statistical analysis

Competing strains	Result with CPG at 24/19°C				Result with CPG at 28°C
	Mean population size (CFU/g)	Competitive indexc	Fd	P valuee	Mean population size (CFU/g)	Competitive indexc	Fd	P valuee
GMI1000a	8.44 × 104	2.65	11.847	0.0034	1.94 × 104	6.37 × 104	3,340	<0.0001
K60b	2.24 × 105				1.24 × 109
GMI1000a	1.62 × 105	6.17	19.0799	0.0005	2.37 × 107	1.74 × 10−1	5.355	0.0343
UW551b	1.00 × 106				4.15 × 106
K60b	1.08 × 106	1.53	0.0676	0.7982	2.44 × 104	8.57 × 104	2,997	<0.0001
UW551a	7.06 × 105				2.09 × 109

^aThis strain is the reference strain for competitive index calculations (denominator).

^bThis strain is the competing strain for competitive index calculations (numerator).

^cCompetitive index values represent the proportional competitive advantage of the competing strain over the reference strain. A competitive index of 1 indicates no competitive advantage.

^dF is the critical value associated with one-way ANOVA statistical analysis.

^eP values indicate the level of difference in bacterial population sizes between competing strains in each paired-strain combination in liquid broth medium as determined by one-way ANOVA.

In tomato plants, tropical highland strain UW551 outcompeted other strains under cool conditions.

Having established the effects of temperature on competitive interactions between R. solanacearum strains, we added a biologically relevant context for these interactions in the form of the plant environment. Unwounded tomato plants were inoculated with a 1:1 mixture of each pathogen strain pair, and the population size of each strain was measured in rhizospheres and stems when symptoms first appeared (approximately 4 to 8 days postinoculation). At a cool 24/19°C, UW551 significantly outcompeted both GMI1000 and K60 in tomato rhizospheres, but K60 and GMI1000 reached similar population sizes when they competed under these conditions (Fig. 4; Table 3). Once the bacteria invaded plant stems, UW551 outcompeted K60 to an even greater degree, with a CI of 4.66 × 10⁶. Results of the other competitions in stems at 24/19°C were more complex. GMI1000 slightly outcompeted UW551 (CI = 12) but was strongly outcompeted by warm temperate strain K60 (CI = 6.84 × 10⁹) (Fig. 5; Table 3).

FIG 4.

Competitive fitness of R. solanacearum strains in the rhizosphere at two temperatures. Box-whisker plots show the distribution of bacterial population sizes recovered from the rhizospheres of 16-day-old tomato plants soil-soak inoculated with a 1:1 mixture of two different R. solanacearum strains as indicated and grown at either 24/19°C (top) or 28°C (bottom). At first wilt symptoms, the population size of each strain was determined by serial dilution plating of rhizosphere soil. The black line in the middle of each box represents the median, and dashed lines highlight the relative trends in population size for each paired-strain competition. The boxes represent the interquartile range (from 25 to 75%) of the results, and each dot represents the bacterial population size in an individual plant (n = 50). See Table 3 for detailed statistical analyses.

TABLE 3.

Results of paired-strain competitions in planta with statistical analysis

Competing strains	Rhizosphere at 24/19°C				Rhizosphere at 28°C				Stem at 24/19°C				Stem at 28°C
	Median population size (CFU/g)	Competitive indexc	Zd	P valuee	Median population size (CFU/g)	Competitive indexc	Zd	P valuee	Median population size (CFU/g)	Competitive indexc	Zd	P valuee	Median population size (CFU/g)	Competitive indexc	Zd	P valuee
GMI1000a	4.66 × 106	7.81 × 103	0.703	0.482	6.42 × 107	−243	5.93	<0.0001	4.66 × 106	6.84 × 10−9	4.68	<0.0001	1	1.39 × 109	6.73	<0.0001
K60b	3.70 × 107				2.59 × 105				5.33 × 109				1.90 × 109
GMI1000a	1.27 ×106	10	3.39	<0.0007	1.20 × 107	−5 × 10−5	3.24	0.0012	1.98 × 108	−12	0.114	0.909	2.31 × 108	−5.0 × 10−5	4.67	<0.0001
UW551b	1.20 × 107				5.25 × 105				1.40 × 106				1
K60a	1.08 × 105	6.5 × 10−3	5.85	<0.0001	4.59 × 107	−108	7.96	<0.0001	1	4.66 × 106	4.18	<0.0001	6.41 × 108	−4.89 × 108	8.23	<0.0001
UW551b	8.97 × 107				2.58 × 105				2.99 × 108				1

^aThis strain is the reference strain for competitive index calculations (denominator).

^bThis strain is the competing strain for competitive index calculations (numerator).

^cCompetitive index values represent the proportional competitive advantage of the competing strain over the reference strain. A competitive index of 1 indicates no competitive advantage.

^dZ is the critical value associated with the Wilcoxon signed-rank test for each paired-strain competition.

^eP values indicate the level of difference in bacterial population sizes between competing strains in each paired-strain combination in plants as determined by Wilcoxon signed-rank test.

FIG 5.

Competitive fitness of R. solanacearum strains in plant stems at two temperatures. Box-whisker plots show the distribution of bacterial population sizes recovered from the stems of 16-day-old tomato plants soil-soak inoculated with a 1:1 mixture of two different R. solanacearum strains as indicated and grown at either 24/19°C (top) or 28°C (bottom). At first wilt symptoms, the population size of each strain was determined by serial dilution plating of ground mid-stem tissue. The black line in the middle of each box represents the median, and dashed lines highlight the relative trends in population size for each paired-strain competition. The boxes represent the interquartile range (from 25 to 75%) of the results, and each dot represents the bacterial population size in an individual plant (n = 50). See Table 3 for detailed statistical analyses.

In planta, warm temperate strain K60 outcompeted the other strains at 28°C.

When strains were coinoculated onto tomato plants growing in a lowland tropical environment, warm temperate strain K60 outcompeted tropical highland strain UW551 and tropical lowland strain GMI1000 in both rhizospheres and stems (Fig. 4 and 5). With CI values greater than 10⁸ in the stems of infected tomato plants, K60 was a much stronger competitor when strains shared this specialized niche (Table 3). Tropical lowland strain GMI1000 also outcompeted UW551 in both tomato rhizospheres and stems at 28°C (Fig. 4 and 5; Table 3). Together, these data suggest that tropical highland strain UW551 is not well adapted to compete with warm temperate and tropical lowland strains in plants growing at warm temperatures (summarized in Fig. 6).

FIG 6.

Summary of the relative competitive fitness of three ecologically distinct R. solanacearum strains in tomato plants following paired-strain inoculations. The competition outcomes are shown for each strain pair in rhizospheres (A and C) and in tomato mid-stems (B and D) at temperate (A and B, in blue) and tropical (C and D, in red) conditions. The colored shading and connecting shapes represent the relative competitive success of each strain pair in the given environment. For example, in tomato rhizospheres at 24/19°C, tropical highland R3bv2 strain UW551 outcompeted both tropical lowland strain GMI1000 and warm temperate strain K60, as indicated by the spear points, while GMI1000 and K60, connected by a rectangle, had similar levels of fitness.

Strains K60 and GMI1000 secrete intraspecies growth inhibitors.

The most direct explanation for the observed differences in competitive fitness between R. solanacearum strains is interstrain chemical inhibition. To test this hypothesis, we measured the ability of each strain to inhibit growth of the others in vitro using an assay wherein cell-free supernatants of one strain were placed in a well on an agar plate overlaid with a lawn of the putative target strain. These experiments revealed that both warm temperate strain K60 and tropical lowland strain GMI1000 produced a diffusible factor(s) that limited growth of highland tropical strain UW551, resulting in a clear zone of inhibition around the well (Fig. 7). GMI1000 supernatants slightly inhibited growth of K60, while K60 supernatants had a noticeably greater effect on GMI1000, indicating that K60 and GMI1000 produce different inhibitors. In contrast, UW551 did not inhibit either K60 or GMI1000 (Fig. 7). No strain was inhibited by its own supernatant. Temperature was not a factor in inhibitor production or response, since similar zones of inhibition were observed when cultures were grown and assayed at 24/19°C (data not shown).

FIG 7.

Cell-free culture supernatants from R. solanacearum strains inhibited growth of other R. solanacearum strains in culture. Agar plates were overlaid with a suspension of the target strain in molten agar. Wells in the plate were filled with 25 μl of cell-free supernatant from an overnight culture of the producing strain. Plates were photographed after 48 h and the diameter of the growth inhibition zone was measured; representative images of inhibition zones are shown under the corresponding bar. Each bar represents the mean inhibition zone diameter from 6 replicates. Results shown are from cultures grown and plates incubated at 28°C; similar results were observed when cultures were grown and plates were incubated at 24/19°C. Asterisks indicate that inhibition zones were different from empty well control (P < 0.0001), as determined by one-way ANOVA. UW551 was inhibited by producing strains GMI000 and K60 (F = 435); K60 was inhibited by producing strain GMI1000 (F = 471) and K60 was inhibited by producing strain GMI1000 (F = 397).

To determine if these inhibitors affected strains outside the R. solanacearum species complex, we used the plate overlay assay to measure the effects of supernatants from K60, UW551, and GMI1000 on growth of six plant-associated bacterial species, as well as E. coli. None of the R. solanacearum culture supernatants noticeably affected growth of Pectobacterium carotovorum subsp. carotovorum, Xanthomonas campestris pv. vesicatoria, X. campestris pv. campestris, Agrobacterium tumefaciens, Bacillus subtilis, Pseudomonas fluorescens, P. aureofaciens, or E. coli (data not shown). This suggests that R. solanacearum strains produce narrowly targeted intraspecific inhibitors.

Preliminary characterization of inhibitors.

We further characterized the compound(s) responsible for the zones of growth inhibition. Cell-free culture supernatants from K60 and GMI1000 completely lost the ability to inhibit growth of other R. solanacearum strains after 10 min at 65°C or 1 h of incubation with 50 μg/ml of proteinase K at 37°C. When cell-free culture supernatants were passed through a 10-kDa centrifugal concentrator, the eluates had no detectable activity, but the concentrated liquid on the membrane retained inhibitory activity (Table 4). Thus, the inhibitory activity is proteinaceous and larger than 10 kDa. These results suggest that K60 and GMI1000 inhibit growth of UW551 with bacteriocins, commonly defined as antibacterial proteins that affect only closely related strains (38–40).

TABLE 4.

Characterization of the extracellular inhibitory activities produced by R. solanacearum strains K60 and GMI1000

Strain or supernatant	Producing strain (well)a
	UW551		K60
Target strain (overlay)	K60	UW551	GMI1000	UW551
Supernatant untreated	+	−	+	+
Proteinase Kb	−	−	−	−
Supernatant, 65°C for 10 minc	−	−	−	−
10-kDa retentated	+	+	+	+
10-kDa eluatee	−	−	−	−

^aOvernight bacterial cultures were pelleted by centrifugation and 25 μl of supernatant was placed in a 3-mm-diameter well in a CPG agar plate previously overlaid with a suspension of 2 × 10⁶ CFU/ml of the target strain in molten agar. A plus sign indicates that a clear zone of growth inhibition was visible around the well after 48 h of incubation at 28°C; a minus sign indicates absence of a clear zone.

^bCulture supernatants were treated with proteinase K to a final concentration of 50 μg/ml for 1 h at 37°C.

^cCulture supernatants were incubated at 65°C for 10 min.

^dCulture supernatants were collected after passage through a 10-kDa-cutoff centrifugal concentrator.

^eCulture supernatants were retained on the membrane of a 10-kDa-cutoff centrifugal concentrator.

DISCUSSION

To our knowledge, only R3bv2 strains of R. solanacearum cause crop losses in cool temperate tropical highlands (41). Based on extensive field observations, Thurston speculated that R3bv2 strains are specifically adapted to lower temperatures (42). Some laboratory studies show that R3bv2 strains grow better at cool temperatures in culture than tropical strains, but growth in culture at a given temperature does not always correlate with ability to cause disease (14, 41, 43). Under conditions tested in this study, R3bv2 strain UW551 reached modestly higher cell densities in culture than K60 or GMI1000 at the cool temperatures of 24 and 19°C.

The cold adaptation phenotype of R3bv2 strains is more than a physiological capacity to grow at cooler temperatures, because it is specifically expressed during interaction with host plants. For example, at the typical potato storage temperature of 4°C, UW551 survives in potato tubers 2 to 4 months longer than K60 or GMI1000 even though in 4°C water K60 survives better than UW551 (14). Similarly, R3bv2 strains are dramatically more virulent on tomato than lowland tropical strains at cool temperatures (14, 15). Other controlled inoculation studies have reported that R3bv2 strains can wilt tomato and potato plants at temperatures as low as 16°C, while warm temperate and tropical strains like K60 and GMI1000 cause little or no disease below 20°C (22, 43, 44). We found that at 24/19°C, UW551 colonized tomato stems better than the other two strains in the absence of competition. Thus, a substantial body of research supports the idea that R3bv2 strains are better adapted to survive in planta and cause disease at cool temperatures than other R. solanacearum strains. It should be noted, however, that cold adaptation is not entirely unique to R3bv2 strains, since under laboratory conditions phylotype II sequevar 4 strains isolated from Pothos and Anthurium could wilt tomato and potato plants at 18°C (45).

Although differential temperature adaptation appears to explain the absence of lowland tropical R. solanacearum strains from cool regions, it does not explain why R3bv2 strains are not a problem in warmer cropping systems. In the laboratory at 28 to 30°C, there is little difference between R3bv2 and tropical strains with respect to either growth in culture or virulence on plants (14, 15). Consistent with these previous studies, R3bv2 strain UW551 grew as well as warm temperate strain K60 and tropical strain GMI1000 at 28°C under our conditions. Accumulated experimental data thus demonstrate that R3bv2 strains are capable of infecting, colonizing, and wilting plants at warm temperatures in the absence of competition.

No obvious physical barrier confines R3bv2 strains to the tropical highlands. These widespread bacteria contaminate surface waterways in many tropical highlands, including rivers used to irrigate downstream lowlands (11). Such rivers transport R3bv2 cells from highlands to lowland fields, where they likely encounter potential hosts, such as tomato. However, R3bv2 has not been identified from tropical lowland sites. Our goal was to understand why R3bv2 strains are not found in the warm lowland tropics despite their demonstrated biological ability to thrive and cause disease under hot conditions. One explanation for this absence is that R3bv2 strains are directly outcompeted by closely related lowland R. solanacearum strains that occupy the same niche in host plant xylem tissue. We explored this hypothesis by coinoculating tomato plants with pairs of ecologically distinct R. solanacearum strains under controlled conditions.

We first measured the independent growth of the three bacterial strains on and in tomato plants at warm and cool temperatures. Although highland tropical strain UW551 reached a slightly higher population size in tomato stems at 24/19°C, all strains colonized the plant habitats equally well under the other three conditions. These results indicated that any observed differences in population size following coinoculation resulted from competitive interactions between bacterial strains.

Following paired-strain inoculation of tomato plants, we determined the relative competitive indices (CIs) of each strain in both the rhizospheres and the mid-stems. Although there was little or no difference between the three strains in their ability to colonize tomato without competition at either temperature, highland tropical strain UW551 generally outcompeted K60 and GMI1000 in planta at cool temperatures, while warm temperate strain K60 was the best competitor under tropical lowland conditions. These results demonstrate that in vitro experiments cannot accurately predict competitive interactions among strains in plants. Whereas K60 and UW551 grew to equal population sizes in paired-strain competition in culture at 24/19°C, at that temperature UW551 outcompeted K60 by 3 and 6 orders of magnitude in tomato rhizospheres and stems, respectively. The host plant environment is thus a critically important factor mediating interactions between R. solanacearum strains.

Although strain CI values varied among plants within assays (see Fig. S2 and S3 in the supplemental material), the mean CI for a given strain pair was generally much higher in stems than in rhizospheres. For example, K60 outcompeted UW551 108-fold in tomato rhizospheres at 28°C, but it had a 4.89 × 10⁸-fold advantage in tomato mid-stems. These large differences may reflect the fact that the inoculum containing equal numbers of each competing strain was applied directly to the soil, so at the beginning of the assay the rhizosphere contained comparable populations of both strains. However, bacteria in the plant mid-stem have survived intense successive selective pressures that would magnify any minor advantage of a strain. These pressures include persisting in the pathogen's microbially diverse infection court on the root surface, invading the plant by traversing the root cortex and entering the developing protoxylem, overcoming plant defenses, multiplying in the relatively nutrient-poor xylem fluid, and finally spreading up into aboveground stems. During these processes the coinoculated strains are in direct competition for optimal colonization sites and limited nutrients. Moreover, they are confined together in xylem vessels; this close physical proximity would strongly favor bacteria that could chemically inhibit other strains.

We explored the possibility of such a direct inhibitor using paired-strain growth assays in culture. Indeed, both GMI1000 and K60 outcompeted R3bv2 strain UW551 in rich medium. These in vitro CI differences suggested that the tropical lowland strains' higher competitive fitness could result from chemical inhibition that acts even in the absence of a plant host. This would not be surprising because microbes often use antibiosis against competitors (46). As described above, the biology of R. solanacearum would make chemical inhibition an effective competitive strategy, especially within xylem vessels. Niche theory, increasingly applied to microbial ecology, predicts that antibiosis would be particularly important in competition with a closely related strain (47, 48). Different species that occupy the same habitat can functionally partition resources to avoid competition, but members of a single species that exploit the same niche must compete directly for resources (49).

We found that strains K60 and GMI1000 each produced distinct proteins that inhibited growth of UW551 on plates. These proteins had no effect on seven other bacterial species, indicating that they are putative bacteriocins, proteins that specifically target conspecific or closely related strains (28, 50). R. solanacearum has long been known to produce bacteriocins, but their biological role(s) was unknown (38, 51–53). Many plant-associated bacteria make bacteriocins, such as agrocin, produced by Agrobacterium radiobacter strain K84. This bacteriocin excludes the crown gall pathogen A. tumefaciens from the infection court so effectively that A. radiobacter is widely used for biocontrol (54).

Interestingly, both of the lowland strains studied in this investigation could inhibit highland tropical strain UW551. This observation is consistent with the latitudinal diversity gradient hypothesis, which postulates that adaptation to biotic competition drives increasing diversity in tropical zones, while adaptation to environmental stresses is a more important limiting factor for organisms living in colder habitats closer to the poles or at higher elevation (55). This geographic diversity gradient has been observed within species (56) and in microbial populations (57). It is possible that by adapting to cooler temperatures in the tropical highlands, R3bv2 strains escaped the interstrain competition experienced by lowland tropical R. solanacearum strains, and as a result, they were no longer under selection pressure to maintain bacteriocin production and immunity.

We identified a consistent correlation between the ability of strains K60 and GMI1000 to outcompete and exclude UW551 in planta and production by these strains of bacteriocin(s) that specifically inhibited UW551 growth. Bacteriocin production could explain why GMI1000 outcompeted UW551 in tomato stems at 24/19°C even though UW551 outcompeted GMI1000 in the rhizospheres of the same plants and GMI1000 is not as virulent as UW551 at this temperature. To directly determine if these bacteriocins contribute to in planta competitive fitness of R. solanacearum strains, it will be necessary to identify and mutate the gene(s) encoding bacteriocin production and see if the resulting mutant strain has reduced competitive fitness.

Our results suggest that interstrain competition mediated by bacteriocins can explain the geographic distribution of R3bv2 strains. However, in the field other factors may also contribute to this phenomenon. For example, R3bv2 may be a generally poor competitor in lowland soils relative to tropical R. solanacearum strains. It would be interesting to compare survival of R3bv2 and tropical strains in the natural microbiome of plant roots in live tropical soils. Alternatively, tomato plants growing at 24/19°C and 28°C may vary in antimicrobial defenses or other physiological attributes that differentially affect the bacterial strains studied here.

The bacteriocins described here may be practically useful for management of potato brown rot and southern wilt of geranium. Because R. solanacearum R3bv2 causes both diseases, it inflicts both direct yield losses and indirect losses associated with the fact that R3bv2 is a quarantine pathogen in Europe and North America and a highly regulated select agent pathogen in the United States. We identified at least two different bacteriocins that inhibit UW551, which is typical of the nearly clonal group of R3bv2 strains that have been globally distributed along with the potato (12). Transgenic potato or geranium plants that express these bacteriocins might have increased resistance to infection by R3bv2 strains of R. solanacearum.