Intraspecific diversity of Erwinia amylovora strains from northern Algeria

Abstract

Background

Fire blight, caused by Erwinia amylovora, is the most destructive bacterial disease affecting plants in the Rosaceae family, leading to significant economic losses. In Algeria, this disease has been reported since 2010. This study aimed to investigate the origin of fire blight in Algeria, in order to increase knowledge of the epidemiology of this serious disease and contribute to its management. A comprehensive characterization of 18 E. amylovora isolates recovered from northern Algeria between 2016 and 2021 to evaluate their phenotypical and genotypical diversity was conducted.

Results

Phenotypic differences, particularly in growth kinetics, virulence, and fatty acid profiles, allowed differentiation of strains into five groups, possibly indicating distinct introduction events. Genetic characterization revealed that only one strain lacked the ubiquitous plasmid pEA29, which is correlated with reduced virulence, while none harbored the pEI70 plasmid. Phylogenetic analysis using concatenated sequences of the recA, groEL, rpoS, ams, and hrpN genes grouped Algerian strains with those from a broadly prevalent clade. CRISPR genotyping identified a novel CR1 pattern and three genotypes, two of them previously unreported.

Conclusions

This study represents the first phenotypic, genetic, and phylogenetic investigation of E. amylovora strains in the region, and provides valuable information on the possible pathways of the introduction of this fire blight pathogen in northern Africa. The findings suggest one or more introduction events from a common ancestor, likely originating in northern Italy, followed by dispersal in various regions of Algeria.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12866-024-03555-3.

Background

Erwinia amylovora is a gram-negative bacterium that belongs to the Erwiniaceae family. It is the causative agent of fire blight, the most destructive disease affecting pome fruit trees such as apple, pear, quince or loquat and other ornamental and wild plants within the Rosaceae family [1–3]. It is the first bacterium known to cause phytopathogenic disease; it was first reported in the Hudson Valley of New York in 1793 [4] and has since spread to many regions around the world [5].

In Africa, the first country affected by fire blight was Egypt, where it was initially detected in 1964 [6], followed by Morocco in 2006 [7], Algeria in 2010 [8], and Tunisia in 2012 [9]. The incidence of fire blight was limited to North Africa and did not affect other parts of the continent.

E. amylovora preferentially enters the host plant through the nectarthodes present in the nectarial cup, although it can also enter through wounds or natural openings [10]. When environmental conditions are favorable, the bacterium moves rapidly from the flower to the pedicel and then to the twig, the main branch, and even down the trunk, rapidly invading the plant tissues and leading to blockage of water flow and ooze production, resulting in progressive necrosis from the upper parts of the tree to the rootstocks, which can result in the death of the tree [10]. The disease can spread by various means, including insects, wind, rain, or infected material, which can cause the destruction of an orchard in one season. As a consequence, significant economic losses can occur; in 2007, the Swiss government reported that the economic damage from fire blight cost $27.5 million [11], while in the United States, the costs are estimated annually to be over $100 million [12].

Determining E. amylovora diversity is critical for understanding the spread and evolution of the pathogen, as well as developing and implementing effective control measures. For this purpose, many approaches have been developed, such as rep-PCR fingerprinting [13], random amplified polymorphic DNA fragment (RAPD) [14], amplified fragment length polymorphism (AFLP) analysis [15], pulsed field gel electrophoresis (PFGE) [16], clustered regularly interspaced short palindromic repeats (CRISPR) [17, 18], variable number of tandem repeats (VNTR) analysis [19] and comparative genome analysis [20–22]. In recent years, numerous studies have adopted the CRISPR genotyping method to analyze the population structure of E. amylovora across various countries [3, 23–27], and this technique has been proven to be an efficient tool for evaluating the genetic diversity of this bacterium. Additionally, a combination of CRISPR and multilocus sequence typing (MLST) may provide further insights into the evolutionary and ecological pathways of these isolates [19], while phenotypic characterization may reveal more information about the evolution and adaptation of the pathogen to its environment [21, 28].

In Algeria, fire blight symptoms were first identified in many pear orchards across many provinces located in the north-central region (Algiers, Blida, Tipaza and Boumerdes), where the disease affected 284 commercial orchards in 2010 [8]. The presence of E. amylovora was confirmed in 2011 [29], although it is suspected that the disease may have been present earlier and that favorable weather conditions contributed to its occurrence [8]. Today, the disease has spread throughout the country and affects many apple- and pear-growing regions, which has become a major threat to the agricultural sector.

The aim of this research was to examine the genetic and phenotypic diversity of a collection of E. amylovora strains isolated between 2016 and 2021 from several provinces in the northwestern, northeastern, and northcentral regions of Algeria. In addition, the relationship between these strains and other strains has been investigated to determine the possible origin of their introduction.

Results

Isolates

Colonies with a morphology similar to that of E. amylovora were isolated from seven provinces (Chlef, Ain Defla, Tipaza, Blida, Sétif, Mila and Constantine) where pear and apple plants with symptoms compatible with the fire blight of Rosaceae have been reported (Fig. 1).

Fig. 1.

Locations of the orchards of pear and apple trees in different provinces of Algeria from which the 18 Erwinia amylovora strains used in this study were isolated. Dashed areas denote locations where fire blight was previously (prior to this study) reported. The map was edited using QGIS

The disease was reported for the first time in five of these seven locations (Chlef, Ain Defla, Sétif, Mila and Constantine). Then, 18 colonies were selected from the three culture media tested, which were representative of the different locations and host plants. All of them, listed in Table 1, were identified as E. amylovora by conventional [30] and real-time PCR [31, 32].

Table 1.

Erwinia amylovora strains isolated from apple and pear trees in different Algerian locations, and the reference strains used in this study

Strain	Host	Cultivar	Location	Plant organ	Date of sampling
EA16	Pear	Santa Maria	Mila	Fruit	2016
EA22	Pear	Santa Maria	Sétif	Fruit	2016
EA46	Apple	Unknown	Tipaza	Twig	2018
EA52	Apple	Golden delicious	Chlef	Twig	2018
EA53	Pear	Santa Maria	Chlef	Twig	2018
EA54	Pear	Santa Maria	Ain Defla	Twig	2019
EA55	Pear	Santa Maria	Ain Defla	Twig	2019
EA57	Pear	Santa Maria	Ain Defla	Exudate	2019
EA58	Pear	Santa Maria	Ain Defla	Twig	2019
EA59	Pear	Santa Maria	Chlef	Twig	2019
EA61	Pear	Santa Maria	Ain Defla	Twig	2019
EA68	Pear	Santa Maria	Ain Defla	Twig	2019
EA66	Apple	Anna	Ain Defla	Twig	2019
EA63	Pear	Santa Maria	Ain Defla	Fruit	2019
EA90	Pear	Santa Maria	Blida	Fruit	2019
EAF1	Pear	Santa Maria	Mila	Twig	2021
EAF2	Pear	Santa Maria	Mila	Twig	2021
EAPC	Pear	Santa Maria	Constantine	Twig	2021
CFBP1430	Crataegus sp.	-	France/Lille	-	1972
IVIA1614	Crataegus sp.	-	Spain/Segovia	-	1996
IVIA1892	Pear	-	Spain/Guadalajara	-	1998
PMV6014	Crataegus sp.	-	France/Lille	-	-

Characterization of phenotypic traits

The biochemical characterization revealed that all the strains were: positive for catalase and Voges-Proskauer activity, glucose, mannose, sorbitol, sucrose, and arabinose fermentation; and negative for arginine dihydrolase, lysine decarboxylase, ornithine decarboxylase, citrate, acetoin and hydrogen sulfide production, urease, tryptophan deaminase and gelatinase. None of them were able to ferment rhamnose or amygdalin, but they differed in the metabolism of inositol (33% positive), melibiose (6%) and β-galactosidase (44%) (Additional file 1).

Analysis of the cellular fatty acid composition revealed that all strains examined were identified as E. amylovora, with percentages ranging between 98.41 and 99.77%. These analyses revealed that ten different fatty acids were present in all the strains tested (lauric acid C12:0, myristic acid C14:0, palmitic acid C16:0, margaric acid C17:0, cis-9,10-methylenehexadecanoic acid C17:0 cyclo, stearic acid C18:0, and “summed feature 2”, consisting of β-hydroxymyristic acid/isopalmitic acid [14:0 3OH/16:1 iso I]; “summed feature 3”, consisting of palmitoleic acid [16:1 w7c/16:1 w6c]; “summed feature 8”, consisting of oleic acid [18:1 w7c/18:1 w6c]; and the unknown fatty acid 10.928). The most abundant fatty acid in all the tested strains was palmitic acid C16:0 (at an average percentage of 35.98%), followed by “Summed Feature 3”, which consists of 16:1 w7c/16:1 w6c palmitoleic acid (29.1%). On the other hand, traces of 0.17 and 0.09% tridecanoic acid C13:0 were found only in strains EA22 and EAF2, respectively, and 0.1% of capric acid C10:0 and the “Summed Feature 7”, consisting of 18.846/19:0 cyclo w10c/19w6c, were found only in strain EA53 (Table 2).

Table 2.

Fatty acid profiles of the Erwinia amylovora strains tested in this study, according to the percentage abundance of each fatty acid

Strain	Fatty acid
	Cluster	Capric acid (C10:0)	Summed in feature 7: 18.846/19:0 cyclo 10c/19w6c	Tridecanoic acid (C13:0)	Stearic acid (18:00)	Margaric acid (C17:0)	Unknown 10.928	Cis-9,10-methylenehexadecanoic acid (C17:0 cyclo)	Myristic acid (C14:0)	Lauric acid (C12:0 )	Summed in feature 2 : β-hydroxymyristic acid/14-methylpentadecenoic acid (14:03OH/16:1 iso I)	Summed in feature 8; Oleic acid (18:1 w7c/18:1 w6c)	Summed in feature 3: Palmitoleic acid(16:1 w7c/16:1 w6c)	Palmitic acid (C16:0)
EAPC	I	0	0	0	0.51	0.76	0.77	3.82	5.09	5.37	7.85	9.99	30	35.84
EAF1	I	0	0	0	0.47	0.57	0.75	4.1	4.25	5.28	8.26	10.68	29.74	35.9
EA55	I	0	0	0	0.39	0.49	1.13	4.69	5.32	5.49	7.69	8.97	30.96	34.88
EA22	I	0	0	0.17	0.68	1.54	1.06	2.33	4.67	5.19	8.34	8.6	31.17	36.24
EA66	I	0	0	0	0.5	0.8	1.21	2.62	5.42	5.57	7.88	9.09	31.22	35.67
EA16	I	0	0	0	0.72	0.86	1.15	2.63	4.89	5.65	8.41	9.03	30.14	36.52
IVIA 1614	I	0	0	0	0.35	0.37	0.99	3.65	4.05	5.54	8.63	10.07	31.33	35.03
EA53	II	0.1	0.2	0	0.44	0.72	0.84	4.82	4.17	5.15	8.7	9.3	29.59	35.97
EA58	II	0	0	0	0.31	0.73	0.87	5.88	4.42	5.12	8.39	9.17	29.49	35.62
EA59	II	0	0	0	0.44	0.54	0.9	4.56	5.14	5.32	8.22	9.02	29.4	36.48
EAF2	II	0	0	0.09	0.51	0.78	0.7	5.26	4.27	5.24	8.45	9.96	28.36	36.37
CFBP1430	II	0	0	0	0.45	0.89	0.98	4.98	4.51	5.53	7.24	10.23	28.59	36.6
EA46	III	0	0	0	0.65	0.65	1.04	4.53	5.27	5.54	7.82	10.22	28.01	36.29
EA61	III	0	0	0	0.61	0.57	1	4.56	5.12	5.48	8.12	10.55	27.87	36.11
EA63	III	0	0	0	0.58	0.64	0.82	4.9	5.14	5.5	7.85	10.03	28.36	36.19
EA52	III	0	0	0	0.57	0.68	1.42	4.83	5.49	5.81	7.58	9.99	27.97	35.67
EA57	III	0	0	0	0.66	0.7	1.1	4.62	5.31	5.38	8	10.23	27.4	36.59
EA68	III	0	0	0	0.75	0.58	1.14	4.52	4.45	5.62	8.44	10.4	27.52	36.58
EA54	III	0	0	0	0.57	0.75	0.89	5.72	5.24	5.41	8.08	9.87	27.1	36.36
EA90	III	0	0	0	0.54	0.71	1.22	5.46	5.52	6.75	8.39	8.95	27.78	34.68

The dendrogram of relatedness between the isolates determined by applying cluster analysis using Ward’s algorithm (Fig. 2) led to the identification of three distinct clusters based on quantitative fatty acid differences (Table 3). A significant difference in the amount of the fatty acid “summed feature 3” (P < 0.001) among the three groups was the main reason for separating the strains into clusters. For the 10 fatty acids as a whole, there was a greater difference (in 40% of the acids) between clusters II and III than between each of these clusters and cluster I (20%), with significant differences also in the content of lauric acid C12:0, myristic acid C14:0 and stearic acid C18:0. In fact, cluster I was similar to cluster II and cluster III by 80% in both cases, the difference between them being, in addition to the already mentioned acid that justifies the division into three groups, the lower content of cis-9,10-methylenehexadecanoic acid C17:0 cyclo (3.65).

Fig. 2.

Dendrogram of cluster analysis of fatty acid compositions of the Algerian Erwinia amylovora strains assayed in this study using Ward’s method

Table 3.

Differences in fatty acid composition between the clusters

Fatty acids	Fatty acids composition (%) a
	Cluster I (n = 7)	Cluster II (n = 5)	Cluster III (n = 8)	Significance
Unknown 10.928	1.06 (0.77; 1.115)	0.87 (0.77; 0.94)	1.07 (0.918; 1.2)	0.104
Lauric acid (C12:0)	5.49 (5.2800; 5.5700)AB	5.24 (5.1350; 5.4250) A	5.52 (5.428; 5.763) B	0.042
Myristic acid (C14:0)	4.89 (4.250; 5.320) AB	4.42 (4.220; 4.825) A	5.255 (5.125; 5.445) B	0.047
Summed in feature 2: β-hydroxymyristic acid/14-methylpentadecenoic acid (14:0 3OH/16:1 iso I)	8.26 (7.85; 8.41)	8.39 (7.73; 8.575)	8.04 (7.83; 8.323)	0.463
Summed in feature 3: Palmitoleic acid (16:1 w7c/16:1 w6c)	30.96 (30; 31.22) A	29.4 (28.475; 29.54) B	27.825 (27.43; 28) C	< 0.0001
Palmitic acid (C16:0)	35.84 (35.030; 36.240)	36.37 (35.795; 36.540)	36.24 (35.780; 36.525)	0.296
cis-9,10-methylenehexadecanoic acid (C17:0 cyclo)	3.65 (2.620; 4.100) A	4.98 (4.690; 5.570) B	4.725 (4.537; 5.320) B	0.004
Margaric acid (C17:0)	0.76 (0.49; 0.86)	0.73 (0.63; 0.835)	0.665 (0.595; 0.708)	0.485
Summed in feature 8: Oleic acid (18:1 w7c/18:1 w6c)	9.09 (8.97; 10.07)	9.3 (9.095; 10.095)	10.125 (9.9; 10.358)	0.301
Stearic acid (C18:00)	0.5 (0.39; 0.68) AB	0.44 (0.375; 0.48) A	0.595 (0.57; 0.658) B	0.014

^a The value of each fatty acid is the median (Q1; Q3). Values in the same row followed by different letters are significantly different (P ≤ 0.05). Significant differences are highlighted in bold

The growth curves were used to calculate the growth parameters for each strain (Additional file 2). Significantly different values were observed for the lag phase (ʎ), maximum growth rate (μ), and area under the curve (AUC) (P < 0.001). A thorough breakdown of these findings from the homogeneous subset analysis is available in Additional file 3.

For the length of the lag phase (ʎ), strain EA54 entered the exponential phase the earliest, with a median time of 13.83 h, while strain EA22 entered the exponential phase the latest, with a median time of 18.344 h (Fig. 3a). The maximum growth rates exhibited notable variations, ranging from 0.240 h⁻¹ for strain EA66 to 0.131 h⁻¹ for strain EA57 (Fig. 3b). The growth potentials (AUCs) of strains EA22 and EAF2 were the lowest at 32.769 and 33.311, respectively, whereas that of strain EA54 was the highest at 46.2 (Fig. 3c).

Fig. 3.

Kinetic parameters obtained from the growth curves of the Erwinia amylovora strains tested in this study: (a) latency phase (λ); (b) maximum growth rate (μ) estimated for the exponential period; and (c) growth potential obtained by calculating the area under each curve (AUC)

Pathogenicity assays

Pathogenicity tests revealed that all the Algerian E. amylovora strains were able to induce necrosis and ooze on inoculated immature pear fruits no later than 7 days after inoculation (Fig. 4). No symptoms were observed in the negative control fruits inoculated with sterile buffer. Interestingly, the isolates showed different degrees of severity [33]. Strains EA16, EA22, EA46, EA52, EA54, EA58, EA59, EA61, EA63, EA66, EA90, EAF1, and EAPC were the most virulent, with an SI of 3, showing necrosis extending from the point of inoculation through almost all the fruit tissue, while strains EA53, EA55, EA57, and EA68 showed medium virulence (SI = 2) with a lesser extent of necrosis. Strain EAF2 was the least aggressive (SI = 1), as necrosis did not extend beyond the inoculation site.

Fig. 4.

Symptoms induced by the Algerian tested Erwinia amylovora strains on immature pear fruits of cv. Blanquilla 7 days after inoculation. The severity index (SI) scale is shown in brackets [78]

A comprehensive multifactorial analysis (MFA) was conducted to explore the overall phenotype of strains based on fatty acids, growth kinetics, biochemical traits, and virulence (Fig. 5). In terms of strain clustering (Fig. 5a), five distinct groups emerged from hierarchical clustering analysis. The first phenotype cluster, indicated in black, consisted of strains EAF1, EAF2, EA22, EA16, and EA66, with a majority (80%) isolated from the eastern region of Algeria. Notably, strain EA22 appeared relatively distant from the other individuals of this cluster. The second cluster included only one strain, EA63. Cluster 3 comprised strains EA53, EA57, EA58, EA59, EA68, and EA90. Cluster 4, in blue, featured the reference strain CFBP1430 alongside EA46, EA52, EA54, and EA61. Clusters 3 and 4 contained only strains from the central-western regions of Algeria. The final cluster comprised two strains, EA55 and EAPC. The first dimension, explaining 25.32% of the total variability, demonstrated a notable positive correlation with growth kinetics, with a comparatively weaker association observed with fatty acid composition and biochemical traits. In contrast, the second dimension (20.09% variability) was primarily defined by virulence traits, while fatty acid and biochemical characteristics exhibited less explanatory power (Fig. 5b-c).

Fig. 5.

MFA analysis based on fatty acid content, virulence, growth kinetics, and biochemical characteristics of the studied Algerian strains of Erwinia amylovora. (a) Factor map illustrating the distribution of strains on the 1st and 2nd dimensions, with distinct colors representing different clusters. (b) Correlation circle plot showing the associations between the groups and the MFA dimensions. (c) Projection of the studied groups of parameters onto the first two dimensions of the MFA

Statistical analysis revealed a significant positive correlation (r = 0.336, P = 0.006) between virulence and growth rate and a significant negative correlation (r = -0.397, P = 0.001) between virulence and the length of the lag phase λ.

Genotypic characterization

Regarding the plasmid content, all the Algerian strains, except for strain EAF2, were carriers of the plasmid pEA29. In contrast, the plasmid pEI70 was not detected in any of the strains. With respect to the VNTR analysis the Algerian strains exhibited the same pattern as the reference strain CFBP1430.

For MLST, the 4311 bp fragments of each isolate resulting from the concatenation of the five housekeeping genes (recA, ams, groEL, hrpN and rpoS) showed identical sequences between all of them, with 100% similarity to CFBP1430 and ATCC49946 (Additional file 4).

The phylogenetic tree constructed from the concatenated sequences of the strains analyzed in this study and the sequences of other strains deposited in the NCBI database (Additional file 5, Fig. 6) indicated that the Spiraeoideae-infecting E. amylovora strains could be divided into four separate clusters. According to this tree, the Algerian strains are related to those of different origins: Belarus, China, Italy, France, Germany, South Korea, Spain, and the United Kingdom. Only one USA strain was grouped into cluster 1, while all other USA strains were grouped into the other clusters, with those from the eastern part of the country grouped on one side (cluster 2) and those from the western part grouped on the other (cluster 3). The CA3R strain differed from the others and formed its own cluster.

Fig. 6.

Phylogenetic tree of Algerian tested Erwinia amylovora strains (in bold) according to the concatenated sequences of the recA, ams, groEL, hrpN, and rpoS genes by using neighbor-joining reconstruction based on a Tamura 3-parameter model and bootstrap analysis (1000 replicates)

A total of 74 individual spacers were identified among the 18 strains analyzed in this work within the three CRISPR arrays present in E. amylovora (Additional file 6), with distinct spacers being considered when the changes were greater than 5 nucleotides. These distinctive CRISPR spacers were numbered consecutively in a 3’-to-5’ orientation with four digits, where the first digit denotes the affiliation to the equivalent CRR. Analysis of the three CRISPR regions of the E. amylovora strains isolated from Algeria revealed that the CRISPR 1 and CRISPR 2 arrays have 29 nt long direct repeat regions (DRs), similar to those present in CFBP1430. However, all Algerian strains had one single nucleotide in the 15th DR of the CR1 locus, with one inversion from C to T (GTG TTC TCC GCG TGA GCG GGG ATA AAC CG), which was not present in the reference strains used for comparison in the study (IVIA1614, PMV6014, IVIA1892).

With respect to the CRISPR results, the analysis of the CRISPR 1 sequences revealed that all the Algerian strains displayed 36 spacers corresponding to CRR1 A or pattern 4 [17, 18], except for strain EA54, which showed a new pattern (NP) due to the absence of the 22/1014 spacer (CTG ATG GCG TCA CGA GCC ATA CGG AAT GTG AC). For the CRISPR 2 sequences of the different Algerian strains, all but one (EA55) had 35 spacers matching the 24/a pattern. However, strain EA55 had a different pattern, known as pattern 29/e, resulting from the duplication of spacer 58/2011. CRISPR 3 sequencing revealed no distinction between the strains, as all of them exhibited this pattern (38/α). A total of three genotypes were found (Fig. 7). While most strains belonged to genotype 4, 24, 38/A a α (Fig. 7a), two new genotypes have been identified: (NP, 24, 38/n a α) for strain EA54 (Fig. 7b) and (4, 29, 38/A e α) for EA55 (Fig. 7c), although all strains belonged to group I [17].

Fig. 7.

Patterns/genotypes from CRISPR arrays CRISPR 1, CRISPR 2, and CRISPR 3 found in the Algerian Erwinia amylovora strains tested in this study (a–c), compared to the reference strain ATCC49946, Ea1/79Sm, and TS3128 (d–f)

(a) The most common genotype (4, 24, 38/A a α) found in this study, also present in strain CFBP1430; (b) The new genotype (NP, 24, 38/naα) identified in strain EA54; (c) The new genotype (4, 29, 38/Aeα) identified in strain EA55; (d) genotype (1, 21, 38/Bbα) found in strain ATCC49946; (e) genotype (5, 24, 38/Daα) previously found in strain Ea1/79Sm; (f) genotype (2, 22, 38) found in the Korean strain TS3128

CRISPR spacers are represented by boxes, with the spacer positions numbered at the top of the columns. Spacers that differed from others by more than 5 nucleotides were given a specific color, while white boxes indicate the absence of the spacer within the pattern/genotype. The spacers were numbered, and the patterns or genotypes were named according to the nomenclature presented by [17, 18]

Discussion

Algeria experienced a severe outbreak of fire blight in 2010 in Blida province, where it affected 400 ha of orchards throughout 2011 [8]. Despite the implementation of phytosanitary quarantine measures, which have led to the eradication of infected trees [29], the disease continues to spread and affect the main pome fruit-producing regions. In fact, numerous outbreaks have occurred, which have become a real concern for producers. Moreover, the epidemiology of the pathogen in these areas remains poorly understood, making outbreak management difficult.

Previous information available on the E. amylovora populations in Algeria is limited to the small region where the disease first occurred, and little is known about the extent and distribution of the pathogen in other areas of the country [8, 33]. To understand the epidemiology of fire blight in Algeria, in the present study, we performed a phenotypic and genotypic characterization of strains of E. amylovora isolated from apple and pear tree samples collected between 2016 and 2021 in the northern growing regions of the country within a radius of approximately 500 km.

Eighteen strains of E. amylovora were isolated from Chlef, Ain Defla, Tipaza, Blida, Sétif, Mila, and Constantine. Notably, Tipaza and Blida, which are located in the center of the country’s northern production area, serve as the primary epicenters for the dissemination of fire blight to other regions. However, fire blight has expanded to the western region, including Chlef. The areas located in the easternmost northern zone, Constantine, Mila and Sétif, are the areas where the disease has most recently been detected. This strain selection may provide more insight into the dissemination and evolution of the pathogen across a wide geographic region.

The biochemical characterization of the 18 selected strains was consistent with that described for E. amylovora [34]. Only some variability was observed in the metabolism of inositol, melibiose and β-galactosidase. This confirms the high degree of homogeneity of this species, which has already been revealed in previous studies [35].

Fatty acid analysis revealed that this method is a valuable tool for identifying E. amylovora strains and distinguishing them from other bacterial species, as indicated in previous work [36, 37]. The majority of fatty acids detected in Algerian strains were similar to those found in strains from other countries [37–39]. Nevertheless, significant variations in the fatty acid profiles allowed the strains to be classified into three clusters. The most abundant profile was that of cluster 3, which, however, was found only in strains from the central-western areas (Ain Defla, Blida, Chlef and Tipaza). Although cluster 3 was the most abundant cluster, strains from the eastern area (Constantine, Mila and Sétif), except for EAF2, were grouped in cluster I, even strains that were isolated 5 years apart. These results, comparable to those obtained with Serbian strains [39], suggest possible geographical differentiation, which could even be related to independent events of pathogen introduction in Algeria.

Differences in the growth kinetics of some of the strains may reflect a greater or lesser advantage in space and resource utilization, which may translate into differential aggressiveness in their interaction with the host plant. In fact, we found a correlation between growth kinetics and the severity index, in line with previous studies showing that bacterial strains with higher growth rates and shorter latency phases tend to be more virulent and adaptable [40–42].

As previously observed in different strains of E. amylovora belonging to the Spiraeoideae-infecting group [20, 27, 43], the isolates in this study also exhibited slight differences in virulence. Genetic variations, such as differences in plasmid or nucleotide content within a single plasmid [44, 45] or mutations in certain genes [21, 46], have been reported to lead to differences in virulence. In this sense, genotypic characterization of plasmid content in E. amylovora is important because plasmids are related to virulence as well as diversity and fitness [44]. The plasmid pEA29 plays a role in the pathogenesis and colonization of host tissues because it carries genes encoding thiamine metabolism, which is required for amylovoran production and biofilm formation and is considered an essential virulence factor [47]. The only Algerian strain lacking this plasmid is EAF2, which has the lowest virulence. The presence of natural strains lacking this plasmid is unusual and has been described in Egypt, Iran, Spain and Germany [48]. Therefore, this is the first time that an E. amylovora strain without the pEA29 plasmid has been described in Algeria, which could also be an indication related to independent introduction of the pathogen. This same strain, EAF2, has a fatty acid profile different from that of the other strains from the same geographical area. None of the Algerian strains have the pEI70 plasmid, which, although it seems to play a role in aggressiveness, is currently limited to a few European strains [49].

The MFA revealed differences between the phenotypes of the strains, a finding that has been confirmed in other studies with strains from other countries [28]. Indeed, phenotypic heterogeneity within genetically homogeneous populations is widely recognized [50–52]. The MFA showed a possible association between the phenotype and the region of isolation, which could be due to various introduction events and/or adaptations due to evolutionary pressure from the environment [53].

The high homogeneity of Algerian strains according to VNTR and MLST is not surprising, as it is a common feature in the genetic characterization of E. amylovora populations. Indeed, studies conducted in other countries such as Morocco [54], Tunisia [55], Iran [56], Serbia [57], Korea [58] and Portugal [28], using various techniques such as VNTR, multi locus sequence analysis, and whole-genome sequencing, also reveal strongly homogeneous populations, a fact attributed to the bottleneck event due to intensive cultivation of apple and pear trees in America in 1700s that led to the reduction of population size [18], and the relatively small genome comparing to other phytopathogenic bacteria with a highly conserved core [20, 59].

Nevertheless, the use of the MLST scheme with groEl, recA, ams, rpoS, and hrpN markers not only allows accurate taxonomic identification of E. amylovora strains but is also a good tool for differentiating strains according to their geographical origin. In this regard, the typology of the resulting phylogenetic tree was very similar to those produced by comparison of pEA29 sequences and the whole genome [21, 22], with four well-defined clusters: broad prevalence, eastern US, western US 1, and western US 2. All Algerian strains were part of the widely prevalent clade 1, which is composed of many closely related E. amylovora strains isolated from different parts of the world. However, within this clade, differences between strains are supported by only 4-259 SNPs [21], indicating a very low mutation rate, probably due to the limited time frame of less than 200 years of introduction of fire blight from the USA to other parts of the world and the low selective pressure exerted by the new environment [22].

CRISPR analysis was the most discriminatory, as it revealed some genetic diversity among the strains. Three genotypes were identified, with (4, 24, 38/Aaα) being the most abundant (Fig. 7a). This genotype was previously reported for strains isolated from other areas of the Mediterranean region, such as Egypt, France, Italy, Spain and Lebanon [17, 18]. Within the widely prevalent cluster 1, this genotype differs from those reported for other strains in the clade: (1, 21, 38) for ATCC49946 USA (Fig. 7d), (5, 24, 38) for E-2 from Belarus and Ea1/79Sm from Germany(Fig. 7e); (2, 22, 38) for TS3128 from South Korea(Fig. 7f); [60]; and others for strains from Portugal [27], Kyrgyzstan [3] and Georgia [61].

Notably, two new previously undescribed genotypes have been identified. On the one hand, that of the strain EA54 on the CR1 array was due to the deletion of the 22/1014 spacer [17, 18, 23, 27, 62, 63]. On the other hand, that of strain EA55 on the CR2 array (4, 29, 38/Aeα) differed from the previously identified genotype (4, 24, 38/Aaα). In this case, the difference was due to the duplication of the 58/2011 spacer. The UTRJ2, UT5P4, and UTFer3 strains isolated from Utah (USA) share a similar 29/e pattern with our strain named EA55 [17, 18], and, in fact, pattern 29 has not previously been reported outside the USA. The two strains EA54 and EA55 were isolated in the same area and the same year, together with others in the largest group of isolates. Within this group, there is some heterogeneity in the fatty acid profile. Differences between strains in this group could be the result of selective pressure on isolates competing for the same niche, since pressure can lead to misalignment during DNA replication [64]. Spontaneous mutations in the CRISPR system may be an adaptive mechanism for bacterial population survival under a wide variety of conditions, allowing a subpopulation to acquire beneficial plasmids or other genetic elements that increase their fitness [65].

The presence of the duplicate CRISPR spacer 1029 has been used as a marker to trace the origin of E. amylovora strains [24]. Since all Algerian strains are derived from the ancestral genotype due to the presence of the duplicated spacer 1029 within the CR1 array, it can be concluded that the strains present in Algeria could be linked back to the first outbreak that affected Europe in 1958 and spread to several countries, including England, Spain, Italy and France [19]. Interestingly, all Algerian strains examined in this study had a consistent mutation in the 15th DR of the CR1 array, regardless of year or place of isolation. Despite their different genotypes, strains EA54 and EA55 also exhibited this mutation. Given that the DRs are highly conserved [66], this could indicate a common ancestor to the Algerian strains, which, through one or more introduction events, would have since spread to different areas of the country. Comparison of the CR1 sequences with those deposited in the NCBI database revealed that strains from Trento (Italy) share the same DR mutation as Algerian strains, suggesting a close relationship between them, in line with previous studies showing a relationship between Italian and North African strains [67, 68]. A recent study using whole genome sequencing placed strains from Morocco and Italy in the same clade, separated from European strains [67]. Another study reported similar haplotypes between strains from Algeria and Slovenian strains originating from Italy [33, 69]. A shared ancestor between the Algerian and Italian strains has been suggested previously [68]. Our results also seem to indicate that fire blight could have been introduced into Algeria from southern Europe, most likely via Italy. It is noteworthy that Algeria has emerged as Italy’s largest export market for fruit from the Italian province of Trento [70], the site where strains with the same mutation as the strains in this study were isolated. The trade of nursery stock and other plant tissues, where E. amylovora may stay as an endophytic bacterium in internal tissues, remaining asymptomatic and thus undetectable during visual inspections, plays a key role in the dissemination of the disease over long distances. In Mediterranean countries, aerosols may help carry the bacteria over long distances [16, 71]. Birds also may contribute to the spread as well, by carrying the pathogen on their feet for up to 8 days, which can introduce it to new regions [72].

Conclusions

In conclusion, this study provides the first phenotypic, genetic and phylogenetic analysis of E. amylovora strains from Algeria. The results indicate that the E. amylovora strains from northern Algeria have a homogeneous structure despite being isolated over a 5-year time period and from distant geographical regions. Nevertheless, phenotypic and genotypic analysis revealed some heterogeneity between strains, which provides insight into the diversity, epidemiology and possible evolution of this pathogen. MLST and CRISPR analysis revealed the presence of two new genotypes derived from an ancestral genotype, suggesting that Italy could be the source of introduction of the fire blight pathogen in Algeria. This knowledge can help to predict and prevent future outbreaks and to improve quarantine measures by applying stricter border controls and inspections for the import of plants, especially asymptomatic ones, in a more targeted manner, resulting in more accurate disease management.

Methods

Sampling and strain isolation

From April 2016 to July 2021, samples of twigs, leaves, and fruits with incipient symptoms of fire blight were collected from pears and apple trees in orchards located in different geographical regions of Algeria. Small pieces were cut from the margin between healthy and necrotic tissues, crushed in sterile bags with phosphate-buffered saline (PBS: 0.36% NaCl, 0.018% NaH₂PO₄ 2 H₂O, 0.12% Na₂HPO₄ 12 H₂O; 10 mM, pH 7.2) at a ratio of 0.1 g in 4.5 ml, and macerated for 30 min at room temperature [34]. Aliquots of 50 μl of each macerate were plated on King’s B (KB), CCT and nutrient agar sucrose (NSA) media and incubated at 25 °C for 24–72 h [34]. Colonies on KB were observed under UV radiation, and nonfluorescent, creamy white, circular, and domed colonies were selected and purified. Smooth and mucous colonies on NSA and CCT media were also selected after 72 h and purified for further characterization. All the isolates selected were cryo-conserved at -80 °C and routinely grown on KB media for 48 h.

Identification of bacterial isolates

Colonies with the appearance of E. amylovora were identified by several real-time and conventional PCR protocols with different chromosomal targets [30–32]. To prepare the DNA template, a suspension adjusted to an OD₆₀₀ of 0.2 (approximately 1.5 × 10⁸ CFU/ml) in 2 ml of PBS was heated to 100 °C for 10 min. Conventional PCR, which amplifies a chromosomal region of 187 bp from a gene that encodes a hypothetical protein from E. amylovora [30], was carried out on a Veriti Dx Thermal Cycler (Thermo Fisher Scientific). The amplification products obtained were visualized through 1.5% (w/v) agarose gel electrophoresis in 0.5× TAE buffer (40 mM Tris, 20 mM NaOAc, 1 mM EDTA, pH 8/HOAc) with Good View™ staining (SBS Genetech). The real-time PCR protocols targeting a hypothetical protein-coding gene and the amsC gene [31, 32] were performed on a Lightcycler 480 II thermocycler (Roche). In all tests, the E. amylovora strain CFBP1430 was used as a positive control, and sterile water was used as a negative control.

Phenotypic characterization

Biochemical tests were performed by inoculating commercial strips of the API system 20E (bioMérieux) according to the manufacturer´s instructions, with cell suspensions adjusted to an OD₆₀₀ of 0.2. The incubation temperature was 25 °C, and the results were registered after 48 h. The catalase activity was also tested by the addition of hydrogen peroxide to the bacterial suspensions.

The composition of the cellular fatty acids of the strains was determined at the facility of the CECT, University of Valencia (Valencia, Spain), following the protocol recommended by the MIDI Microbial Identification System [73]. The cellular content of fatty acids was obtained using an Agilent 6850 gas chromatograph with the MIDI Microbial Identification System using the TSBA6 method. The strains were cultured on tryptic soy agar (TSA) at 28 °C for 24 h until the biomass was obtained [74]. A dendrogram based on differences in the fatty acid profiles was generated by hierarchical clustering using Ward’s method.

To analyze the growth kinetics of the strains, 20 μl aliquots of each fresh bacterial suspension adjusted to 1 × 10⁴ CFU/mL in PBS were transferred to 96-well plates containing 180 μL of 1:10 LB medium. The plates were incubated at 25 °C with shaking for 90 h, and the OD₆₀₀ was measured every 1 h on a Multiskan FC microplate photometer reader (Thermo Scientific). At least three replicates per strain were performed. Growth kinetics were determined using the AMiGA cross-platform Python package [75].

Pathogenicity assays

To confirm the pathogenicity of all the isolates, immature pear fruits, cv. Blanquilla, approximately 2–3 cm in diameter were used [76]. They were subsequently washed with tap water, superficially sterilized by immersing them in a 15% sodium hypochlorite solution for 20 min, rinsed 3 times with sterile distilled water and subsequently dried on sterile filter paper. Four 0.5 mm wounds were made on each disinfected immature fruit with sterile tips (three pears per strain). Then, 10 μl of bacterial suspensions in PBS adjusted to approximately 10⁷ CFU/ml were deposited in the wounds [77]. The inoculated fruits were placed in previously disinfected polystyrene trays, covered with plastic bags to maintain high humidity, and incubated at 25 °C. The development of disease symptoms was observed and evaluated until the seventh day [78]. Immature fruits inoculated with strain CFBP1430 and inoculated only with PBS were used as positive and negative controls, respectively.

Genotypic characterization

The presence of the ubiquitous plasmid pEA29 and plasmid pEI70 was screened by multiplex PCR with the primers AJ75-AJ76 and 1.7F-1.7R, respectively [49]. The amplification products obtained were visualized by 1.2% (w/v) agarose gel electrophoresis in 0.5× TAE buffer using a 1 kb DNA ladder as a weight marker (New England Biolabs) after staining with GoodView nucleic acid stain (SBS Genetech) and visualization with a UV transilluminator.

For further molecular characterization, the VNTR technique was performed using the primers VNTR 4 and VNTR 5 [54]. The PCR products were analyzed on a 1% agarose gel and visualized under the same conditions as those described above. Additionally, MLST analysis of all E. amylovora isolates was performed by the amplification of partial sequences from five housekeeping genes [56, 79], namely, groEl [80], recA [81], rpoS [82], hrpN [83] and ams [84], which encode the GroEl heat shock protein, RecA recombination protein, Sigma factor-38, T3-secreted proteins and proteins involved in the synthesis of amylovoran, respectively. PCR was performed by adjusting the amplification conditions to 50 μl of master mix containing 2 U of Taq DNA polymerase (Biotools), 0.2 mM dNTPs, 2 mM MgCl₂ and 0.4 μM of each primer. The mixed samples were subjected to an initial denaturation step at 94 °C for 10 min, followed by 35 cycles of denaturation at 94 °C for 1 min, annealing at 60 °C for 30 s (groEL), 47 °C for 1 min (recA), 58 °C for 30 s (rpoS), 55 °C for 30 s (hrpN), and 58 °C for 45 s (ams), extension at 72 °C for 2 min, and a final extension step consisting of 72 °C for 10 min. The amplification products obtained were visualized by 1% (w/v) agarose gel electrophoresis in 0.5× TAE buffer. Then, the PCR products were purified and sequenced using the Sanger sequencing procedure. The sequences generated were compared against those in GenBank using BLASTn. The phylogenetic analyses with all concatenated sequences were performed using the neighbor-joining method based on the Tamura 3-parameter model, and the support for the nodes was evaluated through 1000 bootstrap replicates using MEGA 11 [85].

Moreover, amplification of the CRISPR arrays was performed using the primers and PCR protocol described previously [18]. Briefly, the primer pairs CR1-F1/C1-R0, Cr2-F1/C2-R1, and CR3-F1/CR3-R1 were used to amplify the CRISPR 1, CRISPR 2, and CRISPR 3 arrays, respectively. All PCR products were purified using a mi-PCR Purification Kit (Metabion) according to the manufacturer’s instructions. Purified PCR products for CRISPR 1 and CRISPR 2 were sequenced using the additional primers CR1-F2, CR1-F3, and CR1RevRpt and Cr2-F2, Cr2-F3, and Cr2-R2, respectively, in combination with the amplification forward and reverse primers. The sequences generated were assembled and aligned using the Codon Code Aligner software V10.0.2, and CRISPR repeats and spacers were searched by the CRISPR Finder tool [86].

Statistical analysis

The Kruskal‒Wallis test was used to analyze the differences in growth kinetics, followed by post hoc analysis using stepwise step-down multiple comparison tests to construct homogeneous subsets. This test was also used to evaluate the differences between groups based on fatty acid analysis followed by pairwise comparison.

A Spearman correlation test was applied to assess the potential association between growth kinetics and virulence. Differences in all tests were considered statistically significant at P ≤ 0.05.

Multiple factor analysis (MFA) was employed to assess variations in the phenotype based on fatty acid composition, virulence, growth kinetics, and biochemical characteristics (as determined by API 20 analysis) using the R package FactoMineR [87].

The statistical analyses were performed using SPSS v 21 [88] and R software v 4.3.2 [89].

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

ʎ

Lag phase

μ

Maximum growth rate

AFLP

Amplified fragment length polymorphism

AUC

Area under the curve

CRISPR

Clustered regularly interspaced short palindromic repeats

DR

Direct repeat region

KB

King B

MFA

Multifactorial analysis

MLST

Multilocus sequence typing

NP

New pattern

NSA

Nutrient sucrose agar

PFGE

Pulsed field gel electrophoresis

RAPD

Random amplified polymorphic DNA

TSA

Tryptic soy agar

VNTR

Variable number of tandem repeats

Author contributions

L.T., S.B., M.S., and E.M-N. conceived and designed the experiments. L.T., S.B., and I.N-H. performed the experiments. L.T., S.B., and E.M-N. performed the data analysis. L.T., S.B., and E.M-N. wrote the manuscript. All the authors have read and approved the final manuscript.

Funding

This study was funded by the Algerian Ministry of Higher Education and Scientific Research (DGRSDT/MESRS) and by the project IVIA-GVA 52202D from Instituto Valenciano de Investigaciones Agrarias (project susceptible to being co-financed by the European Union through the ERDF Program 2021–2027 Comunitat Valenciana).

Data availability

The sequences generated in this study were deposited in the NCBI database https://www.ncbi.nlm.nih.gov/ under accession numbers PP841738-PP841905. Detailed information on the GenBank accession numbers can be found in additional File 4.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.