Emergence of hypervirulent Pseudomonas aeruginosa pathotypically armed with co-expressed T3SS effectors ExoS and ExoU

Abstract

Pseudomonas aeruginosa is a significant pathogen mainly causing healthcare-associated infections (HAIs). Newly emerging high-risk clones of P. aeruginosa with elevated virulence profiles furtherly cause severe community-acquired infections (CAIs). Usually, it is not common for P. aeruginosa to co-carry exoU and exoS genes, encoding two type III secretion system (T3SS) effectors. The pathogenicity mechanism of exoS⁺/exoU⁺ strains of P. aeruginosa remains unclear. Here, we provide detailed evidence for a subset of hypervirulent P. aeruginosa strains, which abundantly co-express and secrete the T3SS effectors ExoS and ExoU. The exoS⁺/exoU⁺P. aeruginosa strains were available to cause both HAIs and CAIs. The CAI-associated strains could elicit severe inflammation and hemorrhage, leading to higher death rates in a murine acute pneumonia model, and had great virulence potential in establishing chronic infections, demonstrating hypervirulence when compared to PAO1 (exoS⁺/exoU^-) and PA14 (exoS^-/exoU⁺). Both ExoS and ExoU were co-expressed and co-secreted in abundance in exoS⁺/exoU⁺ strains. Their abundant protein secretion could boost exoS⁺/exoU⁺ strains’ potentials for cytotoxicity in vitro and pathogenicity in vivo. Genomic evidence indicates that exoU acquisition is likely mediated by horizontal gene transfer (HGT) of the pathogenicity island PAPI-2, while deletion of exoU was sufficient to mitigate virulence in the exoS⁺/exoU⁺ strains. Furthermore, bioinformatics analysis showed that such exoS⁺/exoU⁺P. aeruginosa strains turned out to be widely distributed across the globe. Overall, the research provide detailed evidence for the high virulence and epidemicity of exoS⁺/exoU⁺ strains of P. aeruginosa, highlighting an urgent need for surveillance against these high-risk hypervirulent strains.

Graphical abstract

Highlights

• •CAI-associated exoS⁺/exoU⁺ Pseudomonas aeruginosa strains have emerged as a new subset of hypervirulent P. aeruginosa.
• •The co-expression and co-secretion of T3SS effectors ExoS and ExoU enhanced exoS⁺/exoU⁺ strains’ potentials for cytotoxicity in vitro and pathogenicity in vivo.
• •ExoU acquisition of exoS⁺ P. aeruginosa is likely mediated by HGT of pathogenicity island PAPI-2.
• •ExoS⁺/exoU⁺ P. aeruginosa strains had widely distributed across the globe, highlighting an urgent need for surveillance against these high-risk strains.

INTRODUCTION

Pseudomonas aeruginosa is a gram-negative opportunistic pathogen notorious for causing a variety of human infections, resulting in significant morbidities and mortalities [1]. According to the latest research published in Lancet [2], we can estimate that about 1.03 million individuals were infected by P. aeruginosa worldwide in 2019, and nearly 40% of those cases turned fatal. P. aeruginosa instigates acute infections in the form of sepsis, ventilator-associated pneumonia, and infections of burns or postoperative wounds, predominantly in hospital settings. It also causes chronic airway infections in young patients with cystic fibrosis (CF), which is among the most recalcitrant in chronic infectious diseases [3]. Disquietingly, the risks for chemotherapy failure in P. aeruginosa infections can be profoundly raised by the pathogen’s intrinsic and acquired antimicrobial resistance (AMR). For its apparent capacity to become resistant to virtually all available antibiotics, P. aeruginosa is listed among the ESKAPE (acronym for Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, P. aeruginosa, and Enterobacteriaceae spp.) pathogens, and has been accorded with global concern by the World Health Organization and the Center for Disease Control and Prevention as a leading causative agent of healthcare-associated infections (HAIs) [4, 5].

To subvert host defense, P. aeruginosa employs a formidable armamentarium of virulence factors, with type III secretion system (T3SS) effectors being some of the most important [6]. T3SS, a needle-like structure, dictates the translocation of cytotoxic effector proteins into eukaryotic host cells [7]. Classical T3SS effectors include ExoS, ExoT, ExoU and ExoY, in which ExoS and ExoU are the most efficiently translocated effectors and contribute greatly to pathogenesis [1]. ExoS is a bifunctional enzyme with Rho GTPase-activating protein activity and ADP-ribosyltransferase activity [8]. It compromises cell-cell adhesion by disrupting the actin cytoskeleton and inhibits the DNA synthesis, vesicular trafficking and endocytosis of host cells, consequently leading to host cell rounding and cell death, which is reminiscent of apoptosis [[9], [10], [11], [12]]. ExoU is a potent phospholipase of the A2 family and is considered the most virulent T3SS effector capable of disrupting the plasma membrane and causing rapid cell lysis and necroptosis of mammalian cells [13]. The presence of ExoS and ExoU has been related to specific infections, patient prognosis and outcome [14].

Previous studies have extensively demonstrated that the majority of P. aeruginosa isolates encode either exoU or exoS gene [[15], [16], [17], [18]]. As the repository of sequenced isolates progressively expands thanks to advances in genome sequencing, new insights into the population structure of P. aeruginosa clone lineages and the distribution of their pathogenicity effectors become possible. In general, P. aeruginosa isolates are taxonomically classified into five groups based on their core genomes, including two major groups (designated group 1 and group 2) and three minor groups [17, 18]. Group 1 is characterized by the presence of exoS, and group 2 by the presence of exoU [18]. Thus, previous findings have favored the a priori assumption that the presence of exoS and exoU tends to be mutually exclusive in their occurrence among virulent strains. However, P. aeruginosa isolates co-carrying exoS and exoU, mostly of clinical origins, have been sporadically reported in at least 12 countries, suggestive of a wide distribution of such clones [19, 20] (Table S1). The term “hypervirulent P. aeruginosa” was applied to some cases, where the infection scope extended to the community or healthy individuals (Table S2). As an offending agent, ExoU-expressing P. aeruginosa strains are typically associated with relatively severe outcomes in clinical infections, in comparison with ExoS-expressing strains [21]. Indeed, PA14 (exoS^-/exoU⁺) as an exhaustively studied representative member of group 2 demonstrated higher virulence than PAO1 (exoS⁺/exoU^-), which is a group 1 representative strain [22]. However, the virulence profiles and public health risks of strains co-expressing exoS and exoU genes remain obscure.

To date, P. aeruginosa has been a major cause of HAIs. Prevalence of P. aeruginosa infections in the community is generally lower than that in the hospital, while severe cases of community-acquired infections (CAIs) in immunocompetent individuals are very rare [23]. In this study, we reported three cases of CAIs and two cases of HAIs, all confirmed to be caused by exoS and exoU co-carrying P. aeruginosa (exoS⁺/exoU⁺) clones. Four strains showed heightened potentials for cytotoxicity and pathogenicity vis-à-vis PAO1, and two CAI-associated strains (SY29 and SY351-1) showed a greater ability to cause cytotoxicity and pathogenicity than the case of PA14. Strikingly, SY29 was proved to be highly virulent and decimated mice in an experimental lung infection model. Moreover, in a murine model of subcutaneous infection, the same virulent strain was capable of efficiently establishing long-term cutaneous infections that gave rise to chronic abscess. Furthermore, it was determined that both ExoS and ExoU were abundantly expressed and secreted in these exoS⁺/exoU⁺ strains. Secretion levels of the effector molecules were directly proportional to measures of cytotoxicity and pathogenicity caused by these strains. Importantly, further bioinformatics analysis revealed that the five exoS⁺/exoU⁺ strains all belonged to group 1 P. aeruginosa, and that they likely acquired exoU through horizontal gene transfer. Deletion of exoU in the exoS⁺/exoU⁺ strains predictably attenuated virulence. Our research has provided compelling evidence for the emergence of clinically relevant and previously unrecognized hypervirulent P. aeruginosa clones co-carrying exoS and exoU, which fills a critical gap in our understanding of hypervirulent P. aeruginosa as a high-risk pathotype in CAIs.

RESULTS

exoS⁺/exoU⁺P. aeruginosa Strains are Capable of Causing Community-acquired Infections and Healthcare-associated Infections

A total of 214 P. aeruginosa clinical isolates were collected, of which 50 isolates came from a tertiary care hospital in south China, and 164 from a tertiary care hospital in north China (Table S3). Gene-specific PCR was performed for the detection of exoS and exoU genes in those isolates. While the majority of isolates carried either exoS or exoU, five (5/214 or 2%) isolates carried both exoS and exoU (exoS⁺/exoU⁺). Among these latter isolates, three strains (SY29, SY64, and SY351-1) stemmed from the hospital in south China, whereas the other two (PA563-1 and PA570-1) were from the hospital in north China (Table 1). Through analysis of the whole genome sequences of the five exoS⁺/exoU⁺ P. aeruginosa strains obtained by Illumina and Nanopore sequencing, we confirmed the coexistence of exoS and exoU in these strains. Multi-locus sequence typing (MLST) analysis reveals that the five exoS⁺/exoU⁺ strains belonged to fivesequence types (STs) (Table 1).

Table 1.

Clinical findings on the five exoS⁺/exoU⁺Pseudomonas aeruginosa strains in this study

Strain	Site	Patients’ age	Main complaint for hospitalization	Any underlying illness or not	Source	Isolation time	Infection type	ST
SY29	South China	30	Acute appendicitis	Nil	Periappendiceal pus	On the day of admission	CAI	ST768
SY64	South China	24	Right scrotal abscess	Nil	Purulent fluid secreted from abscess	1 day after admission	CAI	ST3089
SY351-1	South China	78	Instep wound ulceration and suppuration	Coronary atherosclerotic heart disease	Wound exudate	1 day after admission	CAI	ST3420
PA563-1	North China	57	Cerebral infarction	Hypertension	Sputum	14 days after admission	HAI	ST970
PA570-1	North China	69	Cerebral hemorrhage	Nil	Sputum	18 days after admission	HAI	ST3892

CAI, community-acquired infection; HAI, healthcare-associated infection; ST, sequence type.

P. aeruginosa strains SY29 and SY64 were isolated from two young patients, who were admitted to the hospital with acute appendicitis and scrotal abscess (Table 1). SY29 was isolated from periappendiceal pus during an operative procedure upon hospital admission, while SY64 was isolated from purulent fluid secreted from the abscess one day after admission. SY351-1 was isolated from an old patient who was admitted for treatment of instep wound ulceration and suppuration. The strain was recovered from the wound exudate one day after admission. Thus, the P. aeruginosa strains seemed to be an offending agent of community-acquired infections (CAIs) instead of healthcare-associated infections (HAIs) in the above three cases, as they were isolated within 48 h of hospital admission [36]. In comparison, the P. aeruginosa strains PA563-1 and PA570-1 were isolated from two patients (aged 57 and 69), who were admitted to the hospital with cerebral infarction and cerebral hemorrhage, respectively (Table 1). Their subsequent pulmonary infections developed during hospitalization. The strains PA563-1 and PA570-1 were recovered from sputum 14 and 18 days after initial admission, thus indicative of HAI.

exoS⁺/exoU⁺P. aeruginosa Strains have Different Potentials for Cytotoxicity and Pathogenicity in vitro and in vivo

To scrutinize the cytotoxicity potential of exoS⁺/exoU⁺ P. aeruginosa strains, we challenged human A549 alveolar epithelial cells with the five exoS⁺/exoU⁺ strains, using PAO1 (exoS⁺/exoU^-) and PA14 (exoS^-/exoU⁺) P. aeruginosa as controls. After incubation for 3 h, cell death in the host cells was measured based on the cytosolic release of lactate dehydrogenase (LDH). Remarkably, all of the exoS⁺/exoU⁺ P. aeruginosa strains (PA563-1, PA570-1, SY29, SY351-1, and SY64) were found to be significantly more cytotoxic than PAO1, while SY29 and SY351-1 induced greater cytotoxicity relative to the case of PA14 (Figure 1A). Next, we assessed the pathogenicity of five exoS⁺/exoU⁺ strains in a murine acute pneumonia model. Infections with PAO1 and PA570-1 were nonlethal with a challenge inoculum of 4×10⁶ colony-forming units (CFU), whereas the survival rates of mice infected with PA14, SY64, SY351-1, and PA563-1 were lowered at 25%, 60%, 40%, and 25%, respectively. Intriguingly, the strain SY29 demonstrated exceedingly high and acute pathogenicity, killing off all mice infected by SY29 within 17 h (Figure 1B). The physical appearance of lung lobes in PA563-1 and SY29 infected mice was dark red, with SY29 being a more prominent case. In contrast, the lung lobes of healthy or PAO1-infected mice appeared normal as pinkish white (Figure 1C). Histological examination of the lung lobes further suggests that intense inflammatory infiltration, hemorrhage, and tissue injury arose in both PA563-1 and SY29 infection groups (Figure 1C and 1D), corroborating their high capability to cause pneumonia in murine acute lung infections. In particular, lung infection with SY29 elicited the most severe inflammation and hemorrhage (Figure 1C and 1D).

Figure 1.

Cytotoxicity of exoS⁺/exoU⁺strains in A549 cells and their pathogenicity in murine infection models

(A) Lactate dehydrogenase release levels in A549 cells were measured at 3 hours post infection (hpi) by exoS⁺/exoU⁺ strains. (B) Survival rates of mice (n = 5/group) in the acute pneumonia model. (C) and (D) Results for anatomical examination and tissue sections of lungs in mice from the acute pneumonia group at 12 hpi. Inflammation-associated infiltration is highlighted by black arrows, and hemorrhagic lesions and tissue injuries are marked by red arrows. (E) Bacterial loads in the lungs of mice in the acute pneumonia model at 12 hpi. (F) Expression levels of pro-inflammatory cytokines in the murine lungs at 12 hpi. (G) Progression of infection as monitored on the fourth and seventh days in the murine cutaneous abscess model (n = 7/group). Black arrows indicate infiltration by neutrophils and macrophages; red arrows indicate necrotic tissues; blue arrow indicates deep-tissue inflammation and skeletal muscle injury. Scale bars for the tissue section images in D and G denote 200 μm. The infectious inoculum amount in acute pneumonia model and murine cutaneous abscess model were all 4×10⁶ colony-forming units. The images in C, D and G are representative of three independent replicates. Error bars represent SD. ∗: P < 0.05; ∗∗: P < 0.01; ∗∗∗: P < 0.001; ∗∗∗∗, P < 0.0001.

The colonization ability of PA563-1 and SY29 in the murine lungs was assessed in vivo by counting bacterial loads in the lungs after intranasal infection with 4×10⁶ CFU of each strain. At 12 hours post infection (hpi), bacterial loads of PA563-1 were significantly higher than those of PAO1, PA14and SY29 (Figure 1E). Interestingly, the SY29-infected group saw lower bacterial burdens than did PAO1-, PA14- or PA563-1-infected groups (Figure 1E). These observations suggest that strain PA563-1 likely has a greater ability for colonization at infection sites than does SY29. Next, we assessed the expression levels of proinflammatory cytokines, including interleukin (IL)-1β, tumor necrosis factor (TNF) -alpha, IL-6 and IL-8, in murine lungs post-infection with different bacterial strains. At 12 hpi, the expression levels of TNF-alpha and IL-1 Beta in SY29-infected murine lungs were significantly elevated in comparison to that of PAO1- or PA14-infected lungs (Figure 1F). These results furnish a possible explanation for the increased host fatality rates and poor colonization seen in the case of the strain SY29, wherein a hyperinflammation state provoked by SY29 infection led to rapid clearance of the pathogen, but at the expense of poorer survival of the host due to uncontrolled immune reactions (e.g., cytokine storm) and collateral tissue injury.

If SY29 is weak in colonization ability in acute pneumonia models, would it fail to establish long-term or chronic infections? To this end, a murine cutaneous abscess model was employed as a chronic infection model [37] for experimental investigation. A bacterial inoculum of 4×10⁶ CFU was subcutaneously injected via the back of the mouse. At day 4 postinfection, abscess/pus was predictably observed within local tissue swelling in the SY29-infected group, while visible scabs were formed on the skin surface in the PA14-infected group (Figure 1G). Evidence from histological sections reveals infiltration by immune cells, inflammation, and local necrosis in subcutaneous tissues infected with PA14 or SY29, whereas negligible damages were caused by PAO1 infection (Figure 1G). At day 7 postinfection, clear signs for healing were observable in mice infected with PA14, while there were still no qualifiable damages at injection sites in the PAO1-infected group (Figure 1G). In stark contrast, however, inflammation advanced well into the deep skeletal muscle area in the SY29-infected group, whereby cell debris accumulated extensively and caused underlying muscle tissues to extrude large abscesses to the skin surface (Figure 1G). Clearly, being an exoS⁺/exoU⁺ strain, SY29 has greater virulence potential and is superior to the PAO1 or PA14 strain in establishing chronic infections.

exoS⁺/exoU⁺P. aeruginosa Strains Abundantly Express and Secret both ExoS and ExoU

To elucidate the mechanisms underlying the hypervirulence of exoS⁺/exoU⁺ SY29 P. aeruginosa strain, we characterized the expression patterns of exoS and exoU in bacterial cells during growth under conditions of T3SS induction. Remarkably, qRT-PCR results show significant elevations in exoS and exoU transcript expression in the exoS⁺/exoU⁺ P. aeruginosa strains relative to exoS in PAO1 (Figure 2A) or exoU in PA14 (Figure 2B). Notably, the transcript level of exoS in SY29 surpassed that of PAO1 by 800 folds and that of other exoS⁺/exoU⁺ strains by 40 to 130 folds. Likewise, its transcript level of exoU was about 2000 folds higher than that of PA14, and 120-500 folds higher than that of other exoS⁺/exoU⁺ strains. Subsequently, we assessed T3SS activity of the exoS⁺/exoU⁺ strains by determining the expression of pcrV, which encodes the tip protein of the T3SS needle complex [8]. Likewise, elevated mRNA levels of pcrV in all 5 exoS⁺/exoU⁺ strains were detected compared to PAO1, with SY29 being the most prominent case (∼200-fold increase) (Figure 2C). These observations suggest that augmentation of the expression of both T3SS secretory machinery (pcrV) and major effectors (exoS and exoU) could contribute to the heightened virulence seen in exoS⁺/exoU⁺ strains, as exemplified by hypervirulence in the strain SY29. Concomitantly, we have also experimentally verified that exoS⁺ and exoU⁺ can co-occur abundantly at both the transcriptional and secretion levels within the same strain.

Figure 2.

Expression levels of exoS (A), exoU (B) and pcrV (C) in exoS⁺/exoU⁺ strains, as detected by quantitative real-time PCR, with data normalization with respect to PAO1 (pcrV and exoS) and PA14 (exoU). (D) Secretion levels of PcrV, ExoSand ExoU in exoS⁺/exoU⁺ strains, as detected by Western blot, with PAK and PA14 as controls. RpoA was selected as an internal control. Error bars represent SD. ∗, P < 0.05; ∗∗, P < 0.01; ∗∗∗, P < 0.001; ∗∗∗∗, P < 0.0001.

We next set out to determine whether the highly expressed ExoS and ExoU can be secreted. As the secretion of ExoS in PAO1 is generally low or elusive for detection by immunoblotting, we accordingly selected an ExoS high-secretion strain, PAK, as a reference strain. Secretion of ExoS in all 5 exoS⁺/exoU⁺ isolates was detected. SY29 was found to secrete ExoS much more prolifically into the supernatant than did other strains (Figure 2D). Consistent with the results on exoU expression above, higher levels of ExoU secretion in PA563-1, SY29, and SY351-1 were detected in the supernatant with respect to the case of PA14, the greatest secretion levels of ExoU being detected in SY29. These results support our hypothesis that the hypervirulence of exoS⁺/exoU⁺ P. aeruginosa strains is in part attributable to the robust secretion of T3SS effectors. PA563-1, SY29, and SY64 expressed evidently more PcrV than PA14 (Figure 2D), implying that these former strains are better armed with T3SS secretion injectisomes [38].

Quite importantly, our experiments demonstrated that PA570-1 and SY64 secreted ExoU to a lesser extent compared to PA14, even though the former’s ExoU expression levels were higher than that of PA14. In addition, pairwise alignment of the nucleotide sequences of exoU for each of the 5 exoS⁺/exoU⁺ isolates against that of PA14 indicated that the exoU gene in SY29 and SY351-1 was identical to PA14, even the upstream promoter region. We also detected a non-synonymous mutation (T504I) in the exoU of PA563-1 (Table S9). Remarkably, the ExoU protein of PA570-1 and SY64 occurred in a truncated form encoded by a stop codon-gain variant (Q512AfsX527) of the exoU gene due to a frameshift mutation (Table S9), which may eventually contribute to a loss of function. Such protein truncation of ExoU may also account for why secreted ExoU was not readily detectable in some strains.

The Newly Identified exoS⁺/exoU⁺ Strains Belong to the Group 1 P. aeruginosa

In order to elucidate the phylogenetic relationships between the exoS⁺/exoU⁺ strains in our collection, a phylogeny tree was constructed based on the core genes of the five exoS⁺/exoU⁺ P. aeruginosa strains described in this study, alongside 5172 P. aeruginosa strains with curated genomes available from NCBI database. To highlight, five distinct P. aeruginosa groups clearly emerged from the tree, in agreement with the results previously reported by Freschi et al. [17]. Group 1 and group 2 were predominant, being represented by 3728 (72%) strains and 1333 (26%) strains, respectively. Group 1 is characterized by the presence of exoS, and group 2 by the presence of exoU. The five exoS⁺/exoU⁺ strains isolated in this study along with 119 other exoS⁺/exoU⁺ P. aeruginosa strains belonged to group 1, and three strains of exoS⁺/exoU⁺ genotype belonged to group 2 (Figure 3A). Within group 1, the 124 exoS⁺/exoU⁺ strains were gathered into four clusters in the dendrogram (Figure 3A). There were 97 strains in cluster A, including PA563-1, SY29, and SY351-1 in our isolate set. Cluster B consisted of 21 strains, including PA570-1 and SY64. Additionally, four and two other strains formed cluster C and cluster D, respectively. It is noteworthy that 15 isolates available from the NCBI database contain the same exoU variants encoding truncated ExoU (as observed in SY570 and SY64). These potential high-risk clones of the exoS⁺/exoU⁺ genotype are thus quite widely distributed across the globe, suggesting hitherto underappreciated epidemicity and potential selective advantages associated with the exoU variants.

Figure 3.

Phylogenic tree of total 5177 Pseudomonas aeruginosa strains and population snapshot of the exoS⁺/exoU⁺P. aeruginosastrains

(A) Maximum likelihood tree based on core genes of 5177 P. aeruginosa strains. Colors of the branches and taxa represent five clonal groups of P. aeruginosa. Rings outside the tree represent the presence (light blue) or absence (blank) of type III secretion system effector genes. Corresponding to the inner through outer rings are exoT, exoY, exoS and exoU. Strains co-carrying the four effector genes, and the exoS⁺/exoU⁺ isolates, are indicated with a pentagram in red. Sequence types (STs) identified by the multi-locus sequence typing (MLST) scheme for each exoS⁺/exoU⁺ strain are shown alongside the corresponding red pentagram. The five strains collected in the present study are indicated in red. The four clusters encompassing the group 1 exoS⁺/exoU⁺ strains are also marked. (B) Population snapshot generated by go-eBURST based on MLST profiles of 127 exoS⁺/exoU⁺P. aeruginosa strains. Each circle represents an ST. Relative sizes stand for the number of strains belonging to an ST. Colors of the circles indicate clonal groups to which some corresponding strains belong. Green stands for group 1 and orange for group 2. A black line links STs, where strains only differ from each other by one allele.

In MLST analysis, these exoS⁺/exoU⁺ strains were classified into 29 STs (Figure 3B), of which 27 STs belonged to group 1 and 2 STs to group 2. This offers evidence for high genetic diversity among exoS⁺/exoU⁺ strains. We further examined the ratios of exoS⁺/exoU⁺ strains in each of the STs. The ratios were 100% in all STs, indicating some specificity between genotype and distribution of T3SS effectors. We further traced the isolation sources of the 122 exoS⁺/exoU⁺ strains in the NCBI dataset. These strains were isolated from clinical (80%), animal (3%), and environmental (4%) samples obtained across diverse geographical locales of the world including 19 countries (Figure S1). Notably, 74 exoS⁺/exoU⁺ ST463 P. aeruginosa strains were isolated from a single hospital in Zhejiang province, China (population: 65.4 million), which exhibited multidrug-resistance and carried KPC-2 carbapenemase, as documented during a hospital outbreak [39]. These findings reveal the unappreciated prevalence of exoS⁺/exoU⁺ P. aeruginosa in the world and the concomitant risks of their ability to cause outbreaks in hospitals.

Group 1 exoS⁺/exoU⁺ Strains Acquired exoU Through Horizontal Gene Transfer

In order to clarify the origin of exoU in exoS⁺/exoU⁺ strains in group 1, the complete genomes of PA563-1 and SY29 were analyzed comparatively with that of P. aeruginosa PA14, a representative strain of group 2. As shown in Figure 4, we found that PA563-1 and SY29 contain P. aeruginosa chromosomal pathogenicity island-2 (PAPI-2), which harbors exoU and predominates in group 2 [40]. Sequence-wise, the PAPI-2 of PA563-1 and SY29 shares 98.20% identity with that of PA14, except for the xerC gene, which encodes an integrase and is responsible for the transfer of the genomic island. The absence of xerC possibly leads to a loss of transmission and a gain of stability in PA563-1 and SY29. Furthermore, we analyzed the exoU surrounding sequences of all exoS⁺/exoU⁺ isolates and found that all of them acquired PAPI-2 (Figure 4). The majority of them carried PAPI-2 but not xerC. Our results suggest that group 1 strains could acquire exoU via the aid of PAPI-2.

Figure 4.

Schematic representation of sequence alignment of PAPI-2 in exoS⁺/exoU⁺strains

Detailed comparison among complete genomes of PA563-1, SY29 and PAPI-2 is as illustrated (upper panel). Genes of distinct series are shown in different colors: exoU and spcU in orange; tRNA-Lys in red; xerC in green; insertion sequence (IS) elements in olive green; hypothetical proteins in white; and genes located in conserved regions flanking PAPI-2 in wheat color. Regions sharing high DNA sequence identity are in grey shades, annotated with average identity values. Mapping results for all exoS⁺/exoU⁺ strains’ assemblies against PAPI-2 are as displayed (lower panel), with matching loci in the upper panel (grey dotted lines). Each row represents one exoS⁺/exoU⁺ strain, while each column represents a base position within the PAPI-2 sequence. Colors indicate mapping depths of each exoS⁺/exoU⁺ strain relative to PAPI-2 per base, ranging from 0 to 3 (grey to blue).

A phylogenetic tree of exoU genes from exoS⁺/exoU⁺ strains in group 1 and exoU⁺ strains in group 2 was constructed. exoU genes of exoS⁺/exoU⁺ P. aeruginosa strains were gathered with those of group 2 exoU⁺ strains and they shared 100% identities in the same branch (Figure S2). Such discrepancies between accessory and core gene phylogenies are generally indicative of horizontal gene transfer (HGT), and the mobile element PAPI-2 could mediate the transfer. There are several clusters of exoU genes of exoS⁺/exoU⁺ P. aeruginosa strains suggesting multiple independent HGT events.

ExoU Critically Determines the Virulence Strength of exoS⁺/exoU⁺ Strains in Acute Infections

To better understand how acquisition of T3SS effector ExoU contributes to enhance the virulence of group 1 strains, which is otherwise characterized primarily by the presence of exoS, we knocked out exoU and its chaperon gene spcU in the strains PA563-1 and SY29 to generate PA563-1△exoU and SY29△exoU mutants. As anticipated, their ability to trigger LDH release in A549 cells was blunted following exoU deletion in both strains (Figure 5A), indicating a diminished potential for cytotoxicity in △exoU mutants. In a murine acute pneumonia model, the deletion of exoU significantly decreased the death rates of mice infected with a bacterial inoculum of 4×10⁶ CFU (Figure 5B). Within 24 h after infection, all of the mice intranasally infected with wild-type (WT) strain of SY29 perished, and only 17% of the mice intranasally infected with WT strain of PA563-1 survived within 60 hpi. In comparison, within the △exoU mutant-infected group, mice started to die off only at 30 hpi. The survival rates of mice infected with PA563-1△exoU and SY29△exoU were both 67% within 60 hpi, which were even higher than that of PA14 (33%) (Figure 5B). Compared to the WT strains, the bacterial loads of PA563-1△exoU were decreased while those of SY29△exoU were increased slightly at 12 hpi (Figure 1, Figure 5C). Additionally, both of these mutant strains could cause immune inflammatory reactions, although weaker than the WT strains (Figure 5D). Inflammatory infiltration, hemorrhage, and tissue injury could also be observed in the lungs of the mutant strains’ infection groups (Figure 5E). Furthermore, there was no significant difference between the viability of the wild type and △exoU mutants of PA563 and SY29 (Figure S3). These findings thus implicate a role for exoU acquisition in enhancing the virulence of group 1 strains during acute infections. In addition, quantitative real-time PCR (qRT-PCR) results showed that the mRNA levels of pcrV and exoS rose slightly in SY29△exoU, compared with that of SY29 WT, whereas the mRNA levels of pcrV and exoS decreased in PA563-1 after knockout of exoU, suggesting a complex interplay between ExoU and other T3SS related proteins (Figure S4).

Figure 5.

Comparisonofpotentials for pathogenicity between the wild-type and mutants (exoU knockout strains) of the two exoS⁺/exoU⁺strains, PA563-1 and SY29

(A) Lactate dehydrogenase release levels in A549 cells as measured at 3 h after infection by the wild-type and mutants of PA563-1 and SY29. (B) Survival rates of mice after being intranasally infected by the wild-type and the mutants of PA563-1 and SY29, with PAO1 and PA14 as controls. (C) Bacterial loads in the lungs of mice in the acute pneumonia model at 12 hours post infection (hpi) by the mutant strains. (D) Expression levels of pro-inflammatory cytokines in the murine lungs at 12 hpi infected by the mutant strains. (E) Results for tissue sections of lungs in mice from the acute pneumonia group at 12 hpi infected by the mutant strains. Inflammation-associated infiltration is highlighted by black arrows, and hemorrhagic lesions and tissue injuries are marked by red arrows. Error bars represent SD. ∗∗, P < 0.01; ∗∗∗∗, P < 0.0001.

DISCUSSION

Pathogenic bacteria that are capable of instigating severe healthy individuals deserve special attention for surveillance, as they represent a distinct pathogen class with augmented virulence profiles. Classical Klebsiella pneumoniae, for example, causes infections primarily in hospital environments, whereas hypervirulent K. pneumoniae pathotypes evolve extra virulence factors to permit CAIs [41]. Thus far, the vast majority of reported cases of infections by strains of P. aeruginosa arise in hospital settings, where the use of antibiotics confers some selective advantages. CAIs in immunocompetent individuals by high-risk, hypervirulent clones of P. aeruginosa are rarely reported, whose infection machinery is at best poorly understood. Our findings here demonstrate that high-risk clones co-carrying the T3SS effector genes exoS and exoU can cause CAIs in non-immunocompromised young individuals, which was likely due to the strains’ augmented virulence profiles. P. aeruginosa strains with the same genotypic features could present an emergent threat to public health systems, thus urgently calling for attention.

Since the first report on P. aeruginosa co-carriers of exoU and exoS in 1999, strains of the exoS⁺/exoU⁺ genotype have been occasionally detected in clinical surveys, but they made up only a minor fraction of the total isolates (Table S1). Results from large-scale genomic analyses also corroborated these epidemiological patterns. For instance, Ozer et al. mentioned the presence of 1.3% (10/739) exoS⁺/exoU⁺ strains of P. aeruginosa in their dataset [18]. Taking advantage of the public genome database of NCBI, we expanded the dataset at hand to 5177 isolates (which represents an increase by an order of magnitude) and found that P. aeruginosa strains of exoS⁺/exoU⁺ genotype could well account for 2.5% (127/5177) of globally accessible sequenced isolates. Nevertheless, some studies noted the relatively high prevalence of exoS⁺/exoU⁺ P. aeruginosa strains among the total isolates in certain parts of the world, including locales in Mexico (66%) and Peru (23%). Based on further analysis of sequenced strains, we found that STs-specificity is another key characteristic of exoS⁺/exoU⁺ P. aeruginosa strains. Among the STs analyzed, ST463 is clearly identifiable as a multidrug-resistant high-risk clone that has an alarming track record of causing nosocomial outbreaks [39]. Collectively, these lines of evidence reveal that exoS⁺/exoU⁺ P. aeruginosa strains could indeed pose important risks for outbreaks. As most of the previous works on exoS⁺/exoU⁺ co-carriers covered only their identification and epidemiological typing, future investigations on potential hypervirulent strains of exoS⁺/exoU⁺ P. aeruginosa, particularly with respect to their virulence profiles and evolutionary trajectory, is warranted.

In terms of evolutionary relations, most P. aeruginosa isolates studied are phylogenetically clustered into group 1 characterized by exoS and group 2 by exoU. Majority of the exoS⁺/exoU⁺ strains (including our five strains reported herein) fall into group 1. Analysis of the genetic context of the exoU gene in exoS⁺/exoU⁺ isolates demonstrates that exoU is harbored within the pathogenicity island PAPI-2, while phylogenetic analysis on exoU genes suggests that the exoU gene of exoS⁺/exoU⁺ strains might have been acquired through several PAPI-2-mediated HGT events from exoU⁺ P. aeruginosa strains of group 2. ExoS-producing P. aeruginosa strains are typically associated with less severe outcomes in human infections, compared to ExoU-expressing counterparts [21]. Surprisingly, however, we noticed that these exoS⁺/exoU⁺ strains varied quite substantially in their expression and secretion patterns of T3SS effectors, which underscores the heterogeneity in virulence phenotypes. For instance, ExoU of PA570-1 and SY64 occurred in a truncated form due to frameshift mutations, which may ultimately result in a loss of function, as evidenced by the strains’ reduced potential for pathogenicity relative to SY29. It is worth noting that SY64 was a CAI-caused strain which was isolated from the purulent fluid secreted from the abscess of the right scrotal abscess of a 24-year-old male (Table 1). Although the exoU-truncated SY64 showed weaker virulence than the other exoS⁺/exoU⁺ P. aeruginosa strains, especially SY29, the cytotoxicity in vitro and pathogenicity in vivo of SY64 were both stronger than the exoS⁺ strain PAO1, as well as the expression level of pcrV and exoS, indicating the stronger invasiveness than PAO1. Therefore, SY64 might be invasive enough to cause CAIs. Moreover, within-host evolution usually enables the P. aeruginosa strains to convert acute infection to chronic infection by reducing virulence, and T3SS mutation is the most common strategy [42]. Thus, we proposed another explanation that during the establishment of CAI, the exoU of SY64 might be with the genotype of wild type. After the establishment of infection, the exoU-truncated mutants with lower virulence occurred to help the strain colonize better. And the strain with the frameshift mutation in ExoU happened to be isolated.

Strikingly, 15 exoS⁺/exoU⁺ isolates belonging to 8 STs available in the NCBI database contain exoU mutations identical to that of PA570-1 and SY64. The exoU variants encoding truncated ExoU proteins exist in exoS⁺/exoU⁺ strains within several independent clonal lineages across the globe, suggesting that such mutations of the exoU gene have taken place multiple times independently, with implications for selective advantages. PA563-1 and SY351 demonstrated similar potential for pathogenicity relative to PA14. Deletion of exoU in PA563-1 significantly diminished bacterial cytotoxicity and reduced host death rates in vivo, indicating that ExoU critically determines the acute virulence of P. aeruginosa. Our findings here are consistent with the previous report on the exoS⁺/exoU⁺ strain PA99, which showed that the secretion of ExoU had the greatest impact on modulating virulence [43, 44].

To illustrate, the strain SY29 in this study was found to have robust expression of T3SS components and prolifically secreted ExoS and ExoU as effector molecules, which culminated in excessive cytotoxicity and rapid host death in a murine model of acute pneumonia. Apparently, infection by SY29 elicited a hyperreaction of host cells to produce proinflammatory factors such as TNF-α and IL-1β, which did mediate rapid clearance of pathogens, but ironically led to runaway immune reactions and poor survival of the host [45]. In a murine cutaneous abscess model, SY29 showed a remarkable capacity for causing persistent infections and inducing the formation of chronic abscess. Predictably, deletion of exoU in SY29 significantly reduced bacterially induced cytotoxicity and improved host survival in vivo. However, the expression levels of exoS and pcrV remained high and even exceeded that of the wild-type control, implying some interplay between ExoU and ExoS during the infection processes. In addition, the acquisition of exoU by HGT in P. aeruginosa could be the key to unlocking an adaptive strategy for the pathogen to redouble its acute virulence and launch productive infections more efficiently. As a result, the exoS⁺/exoU⁺ P. aeruginosa strains may be considered as an advantage of adapting to diverse infection niches by modulating their expression and activity of T3SS effectors. Besides, we didn’t evaluate the contribution of other virulence factors to the virulence of these P. aeruginosa strains, since the significantly attenuated virulence in exoU-knockout mutants suggested the exoU as the main factor. Further in-depth scrutiny on the regulation of effector secretion in exoS⁺/exoU⁺ strains after excluding the influence of other virulence factors could promise exciting discoveries and new insights.

To summarize, our work has offered fresh insights into the evolutionary trajectory and virulence signatures of exoS⁺/exoU⁺ P. aeruginosa strains, highlighting the acquisition of exoU as a key event that enhances bacterial infection ability in both acute and chronic infections. One of these capabilities is manifested as strongly enhanced virulence arising from robust expression and secretion of T3SS effectors, particularly ExoS and ExoU. Despite that P. aeruginosa is a decade-old menace in HAIs, hypervirulent pathotypes or high-risk clones of exoS⁺/exoU⁺ P. aeruginosa present an emergent, although much underestimated, threat to public health systems with its expanded infection scope, which calls for urgent attention. Collectively, our findings here have helped advance a fuller mechanistic picture of hypervirulence in P. aeruginosa with direct relevance to CAIs and HAIs. It is hoped that incentives will be given to future studies seeking to elucidate the regulatory machinery of effector secretion in exoS⁺/exoU⁺ P. aeruginosa strains and develop rational surveillance and control strategies against their spread.

MATERIALS AND METHODS

Bacterial Strains and Culture Conditions

A collection of 214 clinical isolates of P. aeruginosa (Table S3) obtained from individual inpatients admitted to a tertiary care hospital (164 isolates) in Hebei province (population: 75.9 million), north China, and a tertiary care hospital (50 isolates) of Hainan province (population: 10 million), south China, was cultured in Luria-Bertani (LB) liquid medium at 37 °C with agitation at 200 rpm. Then, five exoS⁺/exoU⁺ P. aeruginosa isolates were screened out from the 214 P. aeruginosa isolates by polymerase chain reaction (PCR) for exoS and exoU genes. Primers used are as listed in Table S4. The parameters used for the thermal cycler were: 98 °C for 2 min; 30 cycles of 98 °C for 10 s, and 59-61°C (the optimal annealing temperatures for each locus are as listed in Supplementary Information, Table S4) for 15 s, and 72 °C for 45 s; and a final extension of 72 °C for 5 min.

DNA Extraction, Sequencing and Genome Assembly

The exoS⁺/exoU⁺ P. aeruginosa strains were inoculated into LB liquid medium at 37 °C and cultured overnight. Cells were harvested by centrifugation at 12,000 rpm for 1 min and used for genomic DNA extraction by using the Wizard® Genomic DNA Purification Kit (Promega, Madison, WI, USA). DNA was then sent to Novogene Co. Ltd. (Beijing, China) for sequencing, which generated larger than 1 G raw data of 151-bp paired-end reads with an average insert size of 350 bps. After filtering, high-quality sequencing reads (≥100×) were obtained for each of the samples. Reads were then assembled de novo by using Spades v3.13.1 [24].

Public Genome Data and Phylogenetic Analysis

Publicly available genomic sequences of 5172 P. aeruginosa strains with contig number less than 500 were retrieved from the National Center for Biotechnology Information (NCBI: http://ftp.ncbi.nlm.nih.gov) on December 30, 2020. The presence of exoS and exoU were detected by homologous alignment of the nucleotide sequences of exoU of UCBPP-PA14 (PA14_51530) and exoS of PAO1 (PA3841) against each genomic sequence. Background information on isolation sites, source of isolation, infection type, and relevant metadata was obtained from NCBI BioSample or BioProject records or manually reviewed from related publications. All of the 5177 genomes were further annotated by Prokka [25]; annotation results were used to identify the core/pan genomes through Roary [26] with -e -mafft -cd 100. Concatenated sequences of 124 core genes from the 5177 strains were used to construct a maximum likelihood (ML) tree by using RAxML v8.2.12 [27] with 100 bootstrap iterations and other variables set to default. Phylogenetic trees were visualized by using iTol [28]. Nucleotide sequences of exoU for exoS⁺/exoU⁺ P. aeruginosa strains and exoS^-/exoU⁺ group 2 strains were extracted with an in-house Perl script. The ML tree based on alignment of exoU genes was constructed by the same aforementioned method.

Multi-locus Sequence Typing Analysis

To determine the ST of exoS⁺/exoU⁺ P. aeruginosa isolates, MLST analysis was performed by using the P. aeruginosa database (https://pubmlst.org/paeruginosa). New alleles and STs were submitted to the database (Table S5). Burst analysis was performed with goeBurst [29] by using profiles of 28 STs found among 127 exoS⁺/exoU⁺ P. aeruginosa isolates.

Cytotoxicity Assay

Cytotoxicity was assessed by using a LDH release assay, as described previously [30]. Briefly, human A549 alveolar epithelial cells were cultured in Dulbecco’s modified Eagle medium with 2% heat-inactivated fetal bovine serum at 37 °C with 5% CO₂. A total of 2×10⁵ A549 cells were then seeded into wells of a 24-well plate in triplicate and cultured overnight. P. aeruginosa strains were cultured at 37 °C in LB medium to OD_600 nm of 1.0, and the bacterial cells were washed twice in phosphate-buffered saline (PBS). Then, the A549 cells were infected by each of the P. aeruginosa strains at a multiplicity of infection (MOI) of 10 for 3 h. LDH released in the medium was measured by using an LDH cytotoxicity assay kit (Beyotime, Shanghai, China) according to the manufacturer’s instructions.

Animal Infection Models

Animals used in these experiments were 6- to 8-week-old SPF female BALB/c mice (Charles River, Beijing, China). Infection of mice was performed as described previously [31]. Prior to infection, mice were anesthetized by trichloroethanol (1.25%) (Aibei Biotech, Nanjing, China) via intraperitoneal injection at a dose of 0.2 mL/10 g body weight. For the murine host survival model, overnight bacterial culture was diluted (1:100) in fresh LB and regrown at 37 °C up to a density of OD_600 nm = 1.0. Recovered bacterial cells were washed once with pre-chilled PBS. The density of resuspended bacteria was adjusted to 2×10⁸ CFU /mL in PBS. Then, 20 μL of the bacterial suspension containing 4×10⁶ CFU was intranasally inoculated into each mouse (5 or 6 mice per group). Survival of the mice was monitored and recorded every 12 h for a period of 3 days.

For the acute pneumonia model, bacterial density was first adjusted to 5×10⁷ CFU/mL in PBS. Then, 20 μL of the bacterial suspension was intranasally inoculated into each mouse, as described above. After 12 hpi, mice were sacrificed humanely with CO₂. Lungs were dissected, for homogenization. Recovered bacteria were serially diluted for agar plating, followed by enumeration of their CFUs. Lung tissues were also processed for hematoxylin and eosin (H&E) staining, or used in quantitative assays for the cytokines IL-8, IL-1β, TNF-α and IL-6 by using qRT-PCR. Gene-specific primers used are as listed in Table S6. For the mouse (chronic) cutaneous abscess model, each P. aeruginosa strain of the same infectious inoculum amount (4×10⁶ CFU resuspended in 50 μL PBS) as used in acute pneumonia model was injected subcutaneously into the dorsum of the mouse. Progression of infection was monitored, with randomly selected mice being sacrificed on the fourth and seventh day after the initial infection, whose skin abscesses were excised and processed for H&E staining. All inoculum used for the infection model did not induce the expression of T3SS in advance.

Total RNA Preparation and Quantitative Real-time PCR

P. aeruginosa strains were grown overnight in LB at 37 °C with vigorous shaking. These overnight cultures were inoculated into fresh LB, supplemented with 5 mM EGTA (Solarbio, Beijing, China) at 37 °C for 3 h. Subsequently, bacterial pellets were collected and their total RNA was extracted by a FastPure cell/tissue total RNA isolation kit V2 (Vazyme, Nanjing, China) and cDNA was synthesized with HiScript II Q Select RT SuperMix (Vazyme, Nanjing, China). qRT-PCR was performed by using a ChamQ Universal SYBR qPCR master mix (Vazyme, Nanjing, China), whose data were analyzed by using the 2^-ΔΔCt method. Gene-specific primers were used for these genes, with rpsL being used as a housekeeping control (Table S7). All experiments were performed in triplicate.

Western Blot

Rabbit polyclonal antibodies against RpoA, PcrV, ExoS, and ExoU proteins (histidine-tagged fusion proteins expressed by pET28b vector in Escherichia coli) were prepared by using standard protocols [32]. Overnight bacterial cultures were diluted in LB containing 5 mM EGTA and grown for 3 h at 37 °C. Then, 1 mL bacterial cultures (OD_600 nm = 1.0) were added to 1.5 mL Eppendorf tubes, and spun down by centrifugation at 12,000 rpm for 2 min. The supernatant was extracted, to which trichloroacetic acid (TCA) was added to precipitate proteins. The remaining bacterial cells (pellet) were washed 3 times with 1×PBS. Then, 80 μL of 1×SDS-PAGE loading buffer was added to the supernatant and pellet samples, followed by boiling at 99 °C for 10 min. After being fully dissolved by SDS-PAGE, the proteins were transferred onto a polyvinylidene fluoride membrane by using a semi-dry membrane transfer instrument (Bio-Rad, Hercules, CA, USA). Membranes were blocked with 5% non-fat dry milk for 2 h following incubation with primary antibodies (against RpoA, PcrV, ExoS, and ExoU) for 2 h. Thereafter, the membranes were probed with HRP-conjugated secondary antibodies (Abbkine, California, USA), followed by signal detection with an ECL-plus kit (Millipore, Billerica, MA, USA).

Construction of P. aeruginosa exoU Knockout Mutants

The △exoU mutants of PA563-1 and SY29 were constructed by methods described previously elsewhere [33]. First, fragments of about 1 kb in length (homologous arm) corresponding to the upstream and downstream regions of exoU and spcU genes were amplified by PCR with the primers exoU-up-F/exoU-up-R and exoU-dn-F/exoU-dn-R (Table S8), respectively. Digestion products were ligated into a pEX18Tc vector (which was previously linearized with Hind III and Sac I) to yield the recombinant plasmid pEX18Tc-exoU. The resultant plasmids were transformed into E. coli S17-1. Transformants with the correct recombinant plasmids were conjugated with P. aeruginosa PA563-1/SY29. Tetracycline-resistant P. aeruginosa colonies (representing the first crossover event) were picked, re-grown on LB containing 20% (w/v) sucrose, and subsequently plated on agar of the same medium. Presumptive ΔexoU colonies were verified for mutant phenotypes, by using gene-specific primers to confirm any second crossover events. Furthermore, the growth curve of the wild type and △exoU mutants of PA563 and SY29 was measured with a Multiskan FC Microplate Reader (Thermo Scientific, Waltham, MA, USA) at 600 nm for 48 h.

Detection of exoU-containing Pathogenicity Island PAPI-2

For the complete genomes of PA563-1 and SY29 and all genome assemblies of exoS⁺/exoU⁺ P. aeruginosa isolates, each genome was analyzed through homologous alignment against the sequence of the exoU-containing pathogenic island PAPI-2 by using cutoffs of 90% identity over 50% of the length. Results from comparative analysis among PA563-1, SY29, and PAPI-2 were presented by Easyfig [34]. Mapping depths for each genome assembly of exoS⁺/exoU⁺ P. aeruginosa isolates were calculated by means of soap.coverage (available online at https://github.com/gigascience/bgi-soap2/tree/master/tools/soap.coverage) and visualized as a heatmap in the R statistical package v3.5.1 [35].

Statistical Analysis

Statistical analyses in this study were performed with GraphPad Prism 7.0 (GraphPad Software, Inc., La Jolla, USA). P values were calculated by using one-way analysis of variance (ANOVA) and two-tailed unpaired Student’s t-test. Statistical significance was determined at P < 0.05, or as otherwise indicated.

ACKNOWLEDGEMENTS

This study has been graciously supported by grants from the National Key R&D Program of China (2021YFC2302005), the Joint Funds of the International Development Research Center of Canada (109282-001) and the National Key R&D Program of China (2021YFC2301004 and 2017YFE0125600).

Declaration of competing interests

The authors declare that they have no competing interests.

Ethics approval

All animal studies have conformed to the research ethics guidelines of the Institute of Microbiology, Chinese Academy of Sciences (IMCAS), China, on the use of animals in research. Mice were housed and bred in specific pathogen-free (SPF) mouse facilities in compliance with research ethics guidelines. All experimental works with animals have been approved by the Research Ethics Committee of IMCAS.

Data availability

Genomic sequence data used in this study have been deposited into the National Microbiology Data Center (NMDC; https://nmdc.cn/en) under the BioProject accession no. NMDC10018006.

Author contributions

J.F. and F.B. conceived the study and designed the experimental procedures. Y.S., Y.M., Z.Y, J.L., M.Y., X.Z., J.H., G.Z. and C.W. performed the experiments. Y.S., Y.M., D.W., F.B. and J.F. analyzed the data. Y.S., N.K.W., J. F., F.B. and W.W. wrote the manuscript. All authors had full access to all study data and had final responsibility for the decision to submit for publication.

Appendix A. Supplementary data

The following is the Supplementary data to this article: