A New Bacterial Disease on Mandevilla sanderi, Caused by Pseudomonas savastanoi: Lessons Learned for Bacterial Diversity Studies

Abstract

Leaf lesions of Mandevilla sanderi were shown to be caused by Pseudomonas savastanoi. While BOX fingerprints were similar for P. savastanoi isolates from different host plants, plasmid restriction patterns and sequencing of plasmid-located pathogenicity determinants revealed that Mandevilla isolates contained similar plasmids distinct from those of other isolates. A repA-based detection method was established.

TEXT

The ornamental plant Mandevilla sanderi (Dipladenia sanderi [family Apocynaceae]) originating from Middle and South America has become increasingly popular over the last decade, mainly because of its copiously formed red flowers. In 2008, breeders of Mandevilla sanderi observed for the first time large necrotic lesions with chlorotic rings on leaves and tumor formation on stems (see Fig. 1). The potential causal agents isolated from the lesions of leaves of diseased plant material were identified initially by metabolic profiling (Biolog) as Pseudomonas savastanoi pv. glycinea or pv. nerii (data not shown), pathogens of soybean (Glycine max) or oleander (Nerium oleander), respectively. Stem and leaf inoculation of healthy Mandevilla plants with these isolates indeed caused identical symptoms (Fig. 1) fulfilling Koch's postulates.

Fig 1.

Symptoms caused by Pseudomonas savastanoi on leaves (A and C) and stems (B) of Mandevilla sanderi.

P. savastanoi strains are Gammaproteobacteria which belong to genomospecies 2 of the Pseudomonas syringae complex (6). Although the species P. savastanoi was established only in 1992 (5), sequencing of the genomes and plasmids of several P. syringae and P. savastanoi isolates reinforced the discussion about the taxonomic affiliation of P. savastanoi, as the core genome is clearly shared by P. syringae and P. savastanoi strains. Several pathovars of P. savastanoi infect woody plants, e.g., P. savastanoi pv. savastanoi is known as an important pathogen of olive trees (Olea europaea) in the Mediterranean area (11). The typical symptoms of olive knot disease, the formation of tumors in the stem and branches, were already described by Theobrast Eresos in the year 300 BC. The disease causes massive yield losses, and breeding for tolerant varieties is the main strategy to overcome the commercial losses due to infections with P. savastanoi pv. savastanoi (16).

Taking into account that other known hosts of P. savastanoi are oleander (Nerium oleander) and privet (Liguster vulgare), leaf and stem inoculations of Mandevilla isolates were done also on oleander, olive, and privet plants. Nine weeks after inoculation, the response patterns of oleander were similar to those observed on Mandevilla plants; however, while no symptoms were observed for olive plants, privet plants displayed only leaf lesions (unpublished results). Thus, it was hypothesized that the most likely causal agent of the novel bacterial disease of Mandevilla sanderi might have originated from infested oleander plantations in the vicinity of the Mandevilla sanderi-producing companies in the South of France. To shed light on this hypothesis, P. savastanoi isolates from Mandevilla sanderi were characterized and compared to isolates originating from olive trees, oleander, jasmine, and privet (information on the isolates, their hosts, and their geographical origins is given in Table 1). Furthermore, we aimed to use this information as the basis for the development of a sensitive and specific method for detection and differentiation of the pathogen from total community DNA.

Table 1.

Bacterial strains and isolates used in this study

Isolate	Species (strain name)	Host plant	Origin	Reference or source
Ph1	Pseudomonas savastanoi pv. glycinea (DSMZ 19341)	Olea europaea	Yugoslavia	5
Ph2	Pseudomonas savastanoi (B202)	Mandevilla sanderi	France	This study
Ph3	Pseudomonas savastanoi (B203)	Mandevilla sanderi	France	This study
Ph4	Pseudomonas savastanoi (B204)	Mandevilla sanderi	France	This study
Ph5	Pseudomonas savastanoi (B205;H16931.1)	Mandevilla sanderi	Germany	This study
Ph6	Pseudomonas savastanoi (B205;H16931.2)	Mandevilla sanderi	Germany	This study
Ph7	Pseudomonas savastanoi (16973)	Mandevilla sanderi	Germany	This study
Ph8	Pseudomonas sp. (B209)	Mandevilla sanderi	France	This study
Ph9	Pseudomonas sp. (B211)	Nerium oleander	Germany	This study
Ph10	Pseudomonas sp. (B213)	Nerium oleander	Germany	This study
Ph11	Pseudomonas sp. (B215)	Nerium oleander	Germany	This study
Ph12	Pseudomonas savastanoi (B217)	Olea europaea	Amendolare, Italy	This study
Ph13	Pseudomonas savastanoi (B218)	Olea europaea	Bari, Italy	This study
Ph14	Pseudomonas savastanoi (B219)	Nerium oleander	Italy	This study
Ph15	Pseudomonas savastanoi (B220)	Liguster vulgare	Bari, Italy	This study
Ph16	Pseudomonas savastanoi (B221)	Jasminum sp.	Greece	This study
Ph37	Pseudomonas savastanoi (IVIA 1628–3/Psv29)	Olea europaea	Spain	17
Ph38	Pseudomonas savastanoi (NCPPB 2327–3/Psv31)	Olea europaea	Italy	17
Ph39	Pseudomonas savastanoi (IVIA 1657-b8/Psv32)	Olea europaea	Spain	17
Ph40	Pseudomonas savastanoi (CFBP 2074/Psv35)	Olea europaea	Algeria	17
Ph41	Pseudomonas savastanoi (NCPPB/Psv37)	Olea europaea	Portugal	17
Ph42	Pseudomonas savastanoi (NCPPB 1344/Psv47)	Olea europaea	United States	17
Ph43	Pseudomonas savastanoi (NCPPB 3335–3/Psv48)	Olea europaea	France	17
Ph44	Pseudomonas savastanoi (CFBP 1670/Psv62)	Olea europaea	Italy	17
Ph45	Pseudomonas savastanoi (ITM 317/Psv416)	Olea europaea	Serbia	17

The P. savastanoi strains used in this study (Table 1) were grown on King's B agar medium (12) and incubated for 2 days at 28°C. A loop full of freshly grown bacterial cell material was resuspended in 1 ml 0.85% NaCl and harvested by centrifugation for 5 min at 13,000 × g. This step was repeated once or twice to reduce slime due to copiously produced exopolysaccharides. Crude cell lysates were obtained using a Qiagen genomic DNA extraction kit (Qiagen, Hilden, Germany). The DNA was extracted using a silica-based kit (silica bead DNA extraction kit; Thermo Scientific, St. Leon-Rot, Germany). Checking the DNA yields under conditions of UV transillumination after agarose gel electrophoresis and ethidium bromide staining revealed the presence of both plasmid and genomic DNA in all strains. 16S rRNA gene fragments amplified from genomic DNA (Table 2) and digested with AluI-MspI or Hin6I-Bsh1236I displayed identical restriction patterns for the subset of strains (Ph1 to Ph16) tested, suggesting that these strains most likely belong to the same species (data not shown). The 16S rRNA gene sequence was determined for all seven isolates from Mandevilla sanderi (Ph2 to Ph8). All sequences were 100% identical. Phylogenetic analysis of partial nucleotide sequences (1,413 bp) of the 16S rRNA gene showed that the isolates from Mandevilla sanderi clustered together with those of P. savastanoi pv. nerii (ITM313) and P. savastanoi pv. savastanoi (NCPPB 3335)(Fig. 2).

Table 2.

Primers, probes, and PCR conditions used in this study

Target	Primer	Sequence (5′–3′)	Annealing temp (°C)	Product size (bp)	Probe(s) generated from strain	Reference
BOX	BOX_A1R	CTACGGCAAGGCGACGCTGACG	53			14
repA	repA-F1	AGCTTCAAGAYCAGGGMAA	55	1,100	Ph4	13
	repA-R2	ARRTCCATCARYCGGTCRAA
16S rRNA gene	U8-27	AGAGTTTGATC(AC)TGGCTCAG	56	1,506		9
	R1494	CTACGG(T/C)TACCTTGTTACGAC
iaaM	iaaM-F	CATATGTATGACCATTTTAATTCACCC	57	1,674	Ph3, Ph38	17
	iaaM-R	GGTACCTTAATAGCGATAGGAGGC
iaaL	iaaL-F	GGCACCAGCGGCAACATCAA	66	456	Ph3	17
	iaaL-R	CGCCCTCGGAACTGCCATAC
hopAB1	hopAB1-F	GCCCGCCTCGCAGACTCAT	63	700	Ph3, Ph38	17
	hopAB1-R	CTGCGCGGATATCATTCACAACTT
hopAF1	hopAF1-F	CTTATCAAGCAGAAAGACGG	55	339	Ph3, Ph43	17
	hopAF1-R	AAGGGAGCAGATGGAATACG
hopAO1	hopAO1-F	TCTCAGTCACAGCATTCC	60	305	Ph43	17
	hopAO1-R	GCTTACGATGTCGTACTC

Fig 2.

Phylogenetic analysis of partial nucleotide sequences of the 16S rRNA gene from strains of the P. syringae complex. All strains included in the tree are identified by their pathovar and strain names. Neighbor-joining trees were constructed using 14 nucleotide sequences (1,413 bp). See Fig. 4 for methodology and Table S3 in the supplemental material for locus tags.

Subsequently, BOX-PCR fingerprints were generated for all isolates as previously described (18). Interestingly, the BOX-PCR fingerprints generated showed high similarity and were almost identical, independently of the strain's host and geographical origin (see Fig. S1 in the supplemental material). Highly similar BOX-PCR patterns were also obtained for olive isolates Ph12, Ph13, and Ph37 to Ph45 (data not shown). While the resolution of ARDRA (amplified ribosomal DNA restriction analysis) is at the genus or species level, BOX-PCR fingerprints have a much finer level of resolution. BOX-PCR fingerprinting is a powerful tool for strain differentiation in medical microbiology, epidemiology, and microbial ecology (10). The PCR products resolved by gel electrophoresis represent a genomic DNA fingerprint pattern that is assumed to be unique for each bacterial strain and isolate (10, 19). While BOX-PCR fingerprint patterns are stable over many generations, they are affected by polymorphism, rearrangements, recombination, or acquisition of foreign DNA (10). On the basis of the BOX-PCR fingerprints, it was concluded that the P. savastanoi strains were highly similar with respect to genomic diversity, indicating that the strains infecting Mandevilla sanderi might have originated from diseased or latently infested olive or oleander trees.

However, another picture emerged when plasmid DNA extracted from all isolates by means of a Qiagen plasmid minikit (Qiagen, Hilden, Germany) was analyzed. Plasmid DNA left undigested or digested with Bst1107I and PstI enzymes (Fermentas) was analyzed in 0.8% or 1% agarose gels, respectively. All strains contained plasmids and displayed multiple plasmid restriction patterns. The Bst1107I-plus-PstI restriction patterns of isolates from Mandevilla sanderi were distinct from those of all other strains. Moreover, plasmid restriction patterns of isolates from olive trees and oleander displayed high diversity, most likely due to the presence of two or more plasmids (Fig. 3A). In fact, olive isolates Ph37 to Ph45 have been reported to contain at least two to six different native plasmids (17). The Southern-blotted plasmid restriction digests were subsequently hybridized with different digoxigenin (DIG)-labeled probes, which were obtained from plasmid-borne genes in P. syringae strains (25) or P. savastanoi strains (17). The probes were generated by PCR under the primer system, plasmid template DNA, and PCR conditions detailed in Table 2 and subsequent DIG labeling according to the manufacturer's instructions (Roche, Mannheim, Germany). Hybridization of Southern-blotted plasmid restriction digests was performed with the repA probe generated with Mandevilla isolate Ph4 according to the method of Götz et al. (7). The repA hybridization patterns of all Mandevilla isolates were identical and clearly distinct from the hybridization patterns of all other P. savastanoi isolates (Fig. 3B). Three strongly hybridizing fragments and two smaller fragments with less hybridization intensity were observed for all isolates from Mandevilla sanderi. While the repA probe was generated from Mandevilla sanderi isolate Ph4, all other probes used in this study were generated with DNA of olive tree isolates Ph38 and Ph43 and of Mandevilla sanderi isolate Ph3. The DIG-labeled probe for iaaM, coding for tryptophan-2-monooxygenase, an enzyme involved in the biosynthesis of indole-3-acetic acid (IAA), was generated from Ph38 and Ph3, whereas the probe for iaaL (IAA-lysine-synthase) was generated from Ph3. The iaaM and iaaL probes hybridized with the same restriction fragment size for all PstI/Bst1107I-digested plasmids of Mandevilla isolates (see Fig. S2A and D in the supplemental material). In agreement with previously reported data (17, 25), these probes hybridized with plasmid DNA from all oleander strains; however, only 2 of the 12 olive isolates tested hybridized with the iaaM probe. The sizes of the hybridization fragments obtained for Mandevilla isolates were similar to the sizes of the fragments detected for all samples giving a hybridization signal (see Fig. S2A and D in the supplemental material). These results clearly show that both the iaaM and iaaL genes are carried on plasmids in all Mandevilla isolates tested.

Fig 3.

Restriction profiles of plasmid DNA determined with enzymes Bst1107I and PstI (A); hybridization of the Southern blot with a repA probe derived from Mandevilla isolate Ph4 (B). 1 kb, DNA molecular weight marker Generuler Plus DNA Ladder (Fermentas, St. Leon-Rot, Germany). DIG, DNA molecular weight marker VI (DIG labeled) (Roche, Mannheim, Germany).

Several genes encoding type III secretion system (T3SS) effectors have been reported to be carried on plasmids in P. syringae and P. savastanoi strains. However, different strains isolated from the same host largely differ in the number and type of T3SS effectors carried on their plasmids (17, 25). This was not the case for Mandevilla isolates, as identical fragment sizes were also observed for the Southern blot hybridizations performed with the DIG-labeled hopAF1 gene probes (see Fig. S2C in the supplemental material) obtained from strain Ph43 or strain Ph3 as the template DNA and tested against plasmids from Mandevilla isolates. In addition, the sizes of the two hybridizing fragments observed for the isolates from Mandevilla were different from those of the isolates from all other strains analyzed. Only 4 of 12 isolates from olive and all 3 oleander isolates hybridized with the hopAF1 probe. No hybridization signal was observed for isolates from privet and jasmine (see Fig. S2C in the supplemental material). The similarity in the gene content of the plasmids of all Mandevilla isolates was further confirmed using hopAO1 and hopAB1 probes. The hopAO1 and the hopAB1 probes did not hybridize with any of them. In contrast, the numbers and sizes of the plasmid fragments hybridizing with these probes largely differed among the remaining P. savastanoi isolates (see Fig. S2B in the supplemental material; data not shown for the hopAB1 probe), as expected from a variable distribution of plasmid-carried T3SS effector genes in these strains. The restriction patterns and hybridizations showed that all P. savastanoi isolates from Mandevilla isolates carried indigenous plasmids which belong to the pPT23A family and share the replication gene repA. However, these plasmids were clearly distinct from the plasmids carried by the other P. savastanoi isolates from all other host plants (Fig. 3). Plasmids belonging to the pPT23A family are assumed to contribute to host specificity and pathogenicity. Recently, the sequence and role in virulence of plasmids carried by the tumor-inducing Pseudomonas savastanoi pv. savastanoi NCPPB 3335 bacterium were determined (1). At least some of the genes involved in the biosynthesis of virulence factors such as iaaM, iaaL, and hopAF1 identified on the plasmid complement of the olive tree NCPPB 3335 isolate could be also amplified on the plasmid DNA from the P. savastanoi isolates from Mandevilla sanderi. Amplicons obtained were cloned using pGEM-T Easy vector (Promega Corporation, Madison, WI), and two clones per gene fragment were sent for sequencing.

In a phylogenetic analysis of repA, the three plasmids of P. savastanoi pv. savastanoi NCPPB 3335 clustered with diverse plasmids from other P. savastanoi olive isolates (group C of repA sequences), although they were separated from plasmids isolated from other pathovars of the genomospecies 2 included in group A of repA sequences (1). In fact, the repA sequence from the Mandevilla Ph3 isolate clustered closely to the repA sequences of a plasmid from P. savastanoi pv. glycinea race 4, which belongs to group A (Fig. 4A). However, a phylogenetic analysis of iaaL (Fig. 4B) showed that the sequence of this gene from Ph3 clustered together with that from P. savastanoi pv. nerii PLVM2 and with one of the alleles (iaaL-1) of pv. savastanoi NCPPB 3335 (15, 20), suggesting that they share a recent common origin. Phylogenetic analyses were also performed for hopAF1 and iaaM sequences (see Fig. S3A and B in the supplemental material, respectively). In both cases, the sequences from Mandevilla isolate Ph3 clustered together with those of P. savastanoi pv. savastanoi NCPPB 3335. Furthermore, the iaaM sequence from Ph3 also clustered in the same branch as that from P. savastanoi pv. nerii PLVM2 (see Fig. S3B in the supplemental material), providing further support for the idea of a recent common origin for all these plasmid-carried sequences. The sequence data have been submitted to the DDBJ/EMBL/GenBank databases (see below).

Fig 4.

Phylogenetic analysis of partial nucleotide sequences of the repA (A) and iaaL (B) genes from strains of the P. syringae complex. The evolutionary history was inferred by the neighbor-joining method (21) using MEGA5 (23); evolutionary distances were computed in numbers of nucleotide substitutions per site. Strains included in the repA and iaaL trees are identified by their plasmid, pathovar, and strain names and by pathovar and strain names, respectively. Sequences from Mandevilla isolate Ph3 (in bold) (Table 1) were amplified using the primers indicated in Table 2. The topologies were identical for trees produced by the minimum evolution and maximum parsimony methods. The percentages of replicate trees in which the associated taxa clustered together in the bootstrap test (10,000 replicates) are shown next to the branches (4). The trees were constructed with published genome sequences plus those from Ph3. Neighbor-joining trees were constructed using (A) 16 nucleotide sequences from the repA gene (330 bp) and (B) 11 nucleotide sequences from the iaaL gene (356 bp); all positions containing gaps and missing data were eliminated using the option of complete deletion. Nucleotide sequences corresponding to the P. syringae complex were downloaded from NCBI, and their locus tags are included in Table S1 (repA) and Table S2 (iaaL) in the supplemental material.

PCR-based system for detection of P. savastanoi in isolates from Mandevilla sanderi.

PCR amplicons with the primers repA-F1 targeting the repA gene of pPT23A-like plasmids and repA-R2 targeting a downstream region of repA (13) were obtained from most P. savastanoi isolates (Fig. 5A). Only the amplicons generated from Mandevilla isolates had a size of approximately 1,100 bp and hybridized with the repA probe derived from the Mandevilla isolate Ph4 (strain B204) (see Fig. 5B). The 5′ sequence of the amplicon was 98.7% identical to the repA 3′ fragment of the pPT23A-like pREP601 plasmid (13). The system was successfully used to detect the pathogen in total community DNA extracted from leaf and tumor material of plants inoculated with strain Ph4 (data not shown). The hybridization method is an important means to increase not only the sensitivity but also the specificity of detection. While the repA PCR provides fast and specific detection of P. savastanoi isolates from Mandevilla sanderi, the application of the detection system in combination with hybridization can be also used to detect the pathogen in DNA directly extracted from plant material or soil.

Fig 5.

PCR amplicons of Pseudomonas savastanoi and Pseudomonas sp. from different host plants (see Table 1) with the primers repA-F1 targeting the repA gene of pPT23A-like plasmids and repA-R2 targeting a downstream region of repA (A) and hybridization with a repA probe derived from Mandevilla isolate Ph4 (+) (see Table 2) (B). NTC, no-target control. DIG VI, DNA molecular weight marker VI (DIG labeled) (Roche, Mannheim, Germany).

Lessons learned for bacterial diversity studies.

Although the ARDRA and BOX-PCR fingerprints indicated that the isolates originating from various diseased plant species and different geographic origins showed high genomic similarity, comparison of plasmids of these isolates showed clear differences and allowed us to withdraw our research hypothesis that the new bacterial disease observed for Mandevilla sanderi is caused by strains from olive or oleander. The plasmids present in the Mandevilla sanderi isolates seemed to be unique, and we hypothesize that properties contributing to the interaction with the host are carried on the mobilome. However, only the plasmid sequence can provide more insights. This study not only provides insights into the diversity of P. savastanoi isolates from woody host plants but also is an example illustrating the resolution level of 16S rRNA gene-based bacterial diversity studies. The analysis of 16S rRNA gene fragments amplified from total community DNA by cloning and sequencing or fingerprinting methods such as terminal restriction fragment analysis, denaturing gradient gel electrophoresis, phylochip analysis, or pyrosequencing analysis has provided fascinating insights into the diversity of bacterial communities in rhizosphere and bulk soils over the last 2 decades. The effects of the soil type, the plant species, or the cultivar on the composition of bacterial communities were unraveled (2, 3, 22, 24). In particular, ultradeep amplicon sequencing techniques promise dramatically improved resolution. However, microbial ecologists need to be aware of the caveats and limitations with respect to bacterial diversity studies based on 16S rRNA genes or other genes belonging to the core gene pool. Despite the challenges encountered in studying the mobilome, we need to realize the important contribution of the mobilome for bacterial diversification, adaptation to changing environments, and the ability to colonize new ecological niches or interact with plants (8).

Nucleotide sequence accession numbers.

The strain Ph3 sequence data have been submitted to the DDBJ/EMBL/GenBank databases under accession numbers JX227983 to JX227986 and JX678983.