Whole Genome Sequencing Highlights the Pathogenic Profile in Nocardia Keratitis

Abstract

Purpose: Nocardia keratitis is a serious and sight-threatening condition. This study aims to reveal the virulence and antimicrobial resistance gene profile of Nocardia strains using whole genome sequencing.

Methods: Whole-genome sequencing was performed on 23 cornea-derived Nocardia strains. Together with genomic data from the respiratory tract and the environment, 141 genomes were then utilized for phylogenetic and pan-genome analyses, followed by virulence and antibiotic resistance analysis. The correlations between virulence genes and pathogenicity were experimentally validated, including the characteristics of Nocardia colonies and clinical and histopathological evaluations of Nocardia keratitis mice models.

Results: Whole-genome sequencing of 141 Nocardia strains revealed a mean of 220 virulence genes contributed to bacterial pathogenesis. The mce gene family analysis led to the categorization of strains from the cornea into groups A, B, and C. The colonies of group C had the largest diameter, height, and fastest growth rate. The size of corneal ulcers and the clinical scores showed a significant increase in mouse models induced by group C. The relative expression levels of pro-inflammatory cytokines (CD4, IFN-γ, IL-6Rα, and TNF-α) in the lesion area exhibited an increasing trend from group A to group C. Antibiotic resistance genes (ARGs) spanned nine distinct drug classes, four resistance mechanisms, and seven primary antimicrobial resistance gene families.

Conclusions: Whole genome sequencing highlights the pathogenic role of mce gene family in Nocardia keratitis. Its distribution pattern may contribute to the distinct characteristics of the growth of Nocardia colonies and the clinical severity of the mice models.

Nocardia, a genus classified within the family Nocardiaceae, under the order Mycobacteriales within the class Actinomycetes,¹ is commonly found in soil, dust, and decaying organic material.² Nocardia can infect various organs in the human body, with the lungs, skin, and cornea being the most common sites.³ To date, more than 100 Nocardia species have been identified through 16S rRNA gene sequencing analysis, and 9 common species or complexes are closely associated with human diseases.⁴ As the cornea is exposed to environmental stress, Nocardia keratitis often arises after corneal trauma or surgery.⁵ 

Nocardia strains are obligate aerobes with limited nutritional requirements, exhibiting resistance to destruction by neutrophils and macrophages.⁶ Moreover, Nocardia has the capability to survive and replicate slowly within inflammatory cells, leading to the frequent development of refractory infections with prolonged course.⁴ Nocardia keratitis, as a severe and sight-threatening condition, presents as a chronic, persistent infection that is unresponsive to commonly used antibiotics.⁵ In clinical settings, the manifestations of Nocardia keratitis are notably variable, sharing close similarities with fungal and other bacterial keratitis. Consequently, this similarity poses intricate challenges for timely and precise early-stage differential diagnosis.⁷ Additionally, consistent variations in the severity of Nocardia keratitis were observed among patients with similar presentation times, along with diverse clinical manifestations and outcomes among patients infected with the same Nocardia species, as identified through mass spectrometry and sequencing analysis.^8,9 These observations revealed the difficulties in determining the severity and prognosis of Nocardia keratitis based on the species of the strains, which emphasized the need for further investigation into the clinical features and pathogenesis underlying Nocardia keratitis. 

Whole genome sequencing (WGS), as a high-throughput sequencing technology, has significantly promoted clinical microbial research at the gene level and generated diverse applications.¹⁰ Studies have exhibited that WGS can be exploited to detect the comprehensive presence of virulence genes and antimicrobial resistance (AMR) genes among clinical bacterial strains.^10,11 Thus, by analyzing the pathogenicity, antimicrobial susceptibility, the ability to form biofilms, and survival in various environments of bacteria, the bacterial genetic profiles and functional phenotypes can be predicted, thereby guiding appropriate clinical diagnosis and treatment strategies. 

In this study, we aim to integrate clinical features and conduct a comprehensive genomic analysis to identify the pathogenic traits of Nocardia in a clinical cohort of 23 patients. Additionally, we seek to identify virulence genes that correlate with the severity of clinical lesions. This integrated approach will support the development of a more precise and tailored diagnostic classification for Nocardia keratitis. 

Materials and Methods

Establishment of Nocardia Keratitis Cohort

Twenty-three consecutive patients (23 eyes) with Nocardia keratitis were included in the study, from the Department of Ophthalmology at Beijing Tongren Hospital, Capital Medical University between 2003 and 2022. The inclusion criteria were patients with typical clinical manifestations of Nocardia keratitis and at least one positive etiological examination (corneal scraping and microbial culture). Informed consent was signed after explicating the study purposes and examination content to the patients. Basic information (age, sex, and first visit time), complete medical history (risk factor and presentation time), and clinical and etiological examination results (visual acuity before treatment, clinical manifestation, corneal scraping, culture, in vivo confocal microscopy, visual acuity after treatment, and resolution time) were collected. Based on the infectious corneal ulcer scoring system,¹² the severity of patients with Nocardia keratitis was evaluated as mild, moderate, and severe. This study adhered to the tenets of the Declaration of Helsinki and obtained the approval of Beijing Tongren Hospital, Capital Medical University Ethics Committee (TRECKY2021-024). 

Identification and Antimicrobial Susceptibility Testing of Nocardia Strains From the Cornea

A total of 23 strains of Nocardia were isolated from corneal scrapings between 2003 and 2022 and stored at −80°C in the Ocular Microbial Resource Bank (OMRB) at the Beijing Institute of Ophthalmology, China. To revive the strains, they were inoculated on blood agar plates (BAPs; Thermo Fisher Scientific, Waltham, MA, USA) and cultured in a bacterial incubator (Thermo Fisher Scientific, Waltham, MA, USA) at 35°C for 3 days. After purification using the streak plate method, the strains were identified to the species level using matrix-assisted laser desorption/ionization-time of flight mass spectrometry (Smart MS; DL Biotech, Zhuhai, China). 

To perform the antimicrobial susceptibility testing, 23 strains of freshly grown Nocardia isolated from the cornea were added to deionized water and adjusted to 0.5 McFarland using a turbidimeter (Thermo Fisher Scientific, Waltham, MA, USA).^13–15 Next, 100 µL of the suspension was transferred into 11 mL of Mueller Hinton broth medium (Thermo Fisher Scientific, Waltham, MA, USA). After mixing the suspension to no visible precipitate, 100 µL of the mixture was introduced into every well of a 96-well plate, which was then combined with lyophilized antibiotics (ceftriaxone, clarithromycin, imipenem, linezolid, rifampin, tobramycin, vancomycin, moxifloxacin, amikacin, and trimethoprim-sulfamethaxole) in each respective well (Thermo Fisher Scientific, Waltham, MA, USA). The plates were sealed and incubated at 35°C for 24 to 72 hours. Final results were based on the 72-hour data, following cutoffs specified in the Clinical and Laboratory Standards Institute (CLSI) M62 document. 

Whole Genome Sequencing and Annotation of Cornea-Derived Nocardia Strains

Subsequently, genomic DNA (gDNA) was extracted from each strain using the DNA isolation mini kit-box2 (DC103-01; Vazyme, Nanjing, China). The concentration and purity of the gDNA samples were evaluated using Nanodrop2000 (Thermo Fisher Scientific, Waltham, MA, USA). The gDNA samples were then submitted for whole genome sequencing on the Illumina Novaseq platform in the PE150 mode (Novogene, Tianjin, China), ensuring that all sequencing depths exceeded 100 folds. The raw data obtained from the WGS were filtered to remove low-quality data, including reads containing low-quality bases (>40%), reads containing >10% N, and reads with an overlap between adapters exceeding 15 bp. This filtering process was performed using readfq 10 (https://github.com/cjfields/readfq). The remaining clean data were de novo assembled by SPAdes 3.15.5,¹⁶ and the coding sequences (CDSs) were predicted using Prokka 1.14.5.¹⁷ 

Phylogenetic Analysis and Pan-Genome Analysis

To gain a more comprehensive understanding of the biological functions and pathogenicity of Nocardia strains isolated from corneal lesions, it is crucial to conduct a thorough analysis of the phylogenetic relationship and virulence genes from various sources of Nocardia. As a result, we downloaded the whole genome sequences of 81 Nocardia strains isolated from the respiratory tract and 41 Nocardia strains isolated from the environment from the National Center for Biotechnology Information (https://www.ncbi.nlm.nih.gov/datasets/genome/). The quality control of 145 genome assemblies was carried out using QUAST 5.2 (CAB, St. Petersburg, Russia).¹⁸ The inclusion criteria were N₇₅ value > 10,000 bp, undetermined bases < 500 per 100,000 bases, and completeness > 98%. As a result, 141 Nocardia genomes that met the quality control standards were identified at the species (132 strains) or genus (9 strains) level and subjected to phylogenetic analysis using GTDB-Tk 2.3.2.¹⁹ The phylogenetic tree was visualized using iTOL 6.8 (https://itol.embl.de/).²⁰ Pan-genome analysis, including core genome, accessory genome, and unique genome, was performed using Bactopia 2.2.0²¹ and Bacterial Pan-Genome Analysis 4.6.6 (BPGA).²² Furthermore, based on clusters of orthologous groups (COGs),²³ the functional genes were categorized into 20 categories (from category C to category V). 

Analysis of Virulence Genes and Antibiotic Resistance Genes

The protein sequences of 141 genomes were aligned using BLASTp version 2.13.0 with specific criteria: a cutoff of 40% identity, 70% coverage, and an E value of 1e-10 against the Virulence Factor Database (VFDB).²⁴ The adherence, invasion, and other virulence genes were identified. Based on the specific distribution and presence of certain virulence genes, the 23 strains of Nocardia isolated from the cornea were grouped into 3 clusters (group A, group B, and group C) using R software version 4.2.2 (The R Foundation, Vienna, Austria). This clustering analysis will help us identify and validate the distinct virulence genes, which in turn will aid in assessing the severity and prognosis of Nocardia keratitis. At last, the antibiotic resistance genes (ARGs) were identified using Resistance Gene Identifier (RGI) version 6.0.2 with a cutoff of 60% identity and 70% coverage against the Comprehensive Antibiotic Resistance Database (CARD).²⁵ 

Experimental Validation of Virulence Genes as Criteria for Grouping Corneal Pathogenicity

Characteristics Analysis of Nocardia Colonies

Three strains from each group (group A, group B, and group C) of Nocardia were cultured on BAPs and incubated at 35°C for 3 days. The growth of Nocardia colonies was observed and photographed every 12 hours using a slit lamp microscope (Topcon SL-D7, Tokyo, Japan). At the same time, the cross-sections of the colonies were captured using a telecentric optical coherence tomography (OCT) cornea lens by Optovue iVue 80 (Fremont, CA, USA). The diameter, height, and angle (elevation angle) of the colonies were measured using the software in Optovue iVue 80, and each parameter was calculated as the mean value of 3 colonies. 

Establishment of a Mouse Model of Nocardia Keratitis

Twelve 8-week-old female C57BL/6N mice were purchased from Vital River Laboratory Animal Technology Co., Ltd. (Beijing, China) and kept in a specific pathogen-free animal room for 1 week. They were then divided into four groups with the random number table: a control group (corneal stroma injection with 2 µL phosphate-buffered solution) and three Nocardia keratitis groups (group A, group B, and group C). Only the left eyes of each mouse were used for the experiments. The right eyes of each mouse model were set as absolute control and were intact throughout the animal experiments. Three strains from each group of freshly grown Nocardia were diluted with Mueller Hinton broth medium (Thermo Fisher Scientific, Waltham, MA, USA). The selected dosage for the corneal stroma injection was 2 µL of the diluent at a concentration of 1.5 × 10⁸ colony forming unit (CFU)/mL, based on the bacterial keratitis mouse model by Chojnacki et al.²⁶ and our preliminary experiments.^27–29 Corneal scrapings of the mice were performed 1 day after the injection, and positive for weak acid-fast staining indicated the mouse model was established successfully. All animal studies followed the guidelines of the Association for Research in Vision and Ophthalmology and were approved by the Institutional Animal Care and Use Committee. 

Clinical and Histopathological Evaluations of the Models

The models with Nocardia keratitis were examined daily to monitor the disease progression and photographed with a slit lamp microscope (Topcon SL-D7, Tokyo, Japan) on day 1, day 3, and day 7. Additionally, 1 mL of 1% sodium fluorescein was instilled into the inferior conjunctival sac of the mice, and the corneal staining patterns were photographed and fitted with a cobalt blue filter. The diameter of the corneal ulcer was measured manually with the photographs of sodium fluorescein staining, and the clinical scores (ranging from 0 to 4) were determined based on the observable severity of ocular disease in the infected mice.³⁰ The mice were euthanized on day 7, and their cornea tissues were preserved in 4% paraformaldehyde solution and conducted hematoxylin and eosin (H&E) staining for histological analysis under a light microscope (Olympus BX51; Olympus, Tokyo, Japan). 

Colony Forming Unit Assessment of the Models

The left corneas were enucleated and homogenized into 1.0 mL sterile phosphate-buffered solution (PBS) using tissue grinder (Kimble Kontes Dounce, Rockwood, TN, USA) on day 1, day 3, and day 7 post-infection. To assess the CFU of Nocardia isolated from the models, 0.1 mL of the homogenate was serially diluted 1:10 in PBS. Serial dilutions (0.1 mL/plate) were plated onto BAP (Thermo Fisher Scientific, Waltham, MA, USA) in triplicate and incubated at 35 °C for 3 days. After the colonies were counted, CFU/mL were calculated. 

Immunohistochemistry and Immunofluorescence Staining

For CD4 and IFN-γ staining, slides with 5-µm-thick sections were deparaffinized in xylene and rehydrated using a series of alcohol concentrations (from 100% to 70%). Antigen retrieval for CD4 and IFN-γ staining was performed in a microwave oven using specific buffer solutions. Endogenous peroxidase activity was blocked using 3% hydrogen peroxide for 10 minutes. The samples were then blocked with 5% goat serum in PBS for 1 hour. Primary antibodies (CD4 and IFN-γ, 1:1000; Abcam, Cambridge, UK) were applied to the slides and left to incubate overnight at 4°C. After washing with PBS, the slides were treated with HRP-conjugated goat anti-rabbit IgG (1:1000; Abcam, Cambridge, UK) for 1 hour at room temperature. Subsequently, the slides were visualized using the 3, 3-diaminobenzidine (DAB; Beyotime Biotech, Guangzhou, China) kit and counterstained with hematoxylin. Finally, the stained sections were observed under a light microscope at 400-fold magnification (Olympus BX51; Olympus, Tokyo, Japan), and three fields of view per section were selected for further analysis. For immunofluorescence staining of IL-6Rα and TNF-α, following the above procedures, the slides were incubated at 4°C overnight with the primary antibodies (IL-6Rα: 1:200; Santa Cruz, Dallas, TX, USA; and TNF-α: 1:100; Abcam, Cambridge, UK). After washing with PBS, slides were incubated with Alexa Fluor 488-conjugated goat anti-mouse IgG (1:500; Abcam, Cambridge, UK) for 1 hour in a dark room. Cell nuclei were counterstained with 4, 6-diamidino-2-phenylindole dihydrochloride (DAPI; Sigma-Aldrich, Darmstadt, Germany). Slides were washed and examined using a fluorescent microscope (Olympus DP72, Tokyo, Japan) with 400-fold magnification. 

Statistical Analysis

All bioinformatics data were analyzed and visualized using R software 4.2.2 (The R Foundation, Vienna, Austria). Wilcoxon rank-sum tests were used to compare corneal genomes with respiratory or environmental genomes within core, accessory, and unique genomes. Parameters such as Nocardia colony size, clinical scores of mice models with Nocardia keratitis, CFU of Nocardia, and staining results were analyzed using GraphPad Prism 9.0 software (GraphPad, La Jolla, CA, USA). Statistical significance was determined using 1-way analysis of variance (ANOVA) followed by the Tukey post hoc test. All P values were 2-tailed, and P < 0.05 was considered statistically significant. 

Results

Nocardia Keratitis Cohort

In this study, a total of 23 patients (23 eyes) diagnosed with Nocardia keratitis were included, and their average age was 40.34 ± 3.49 years. All of them had unilateral infections, and 16 of them (69.6%) were men. Among these patients, seven cases had a history of trauma (3 cases caused by vegetable materials), six cases had a history of contact lens wear, and two had a history of surgery (keratoplasty and meningioma surgery). The mean duration from symptoms onset to hospital admission was 20.50 ± 2.86 days. Based on the clinical manifestations, patients were categorized into three severity levels: mild (8 of 23), moderate (6 of 23), and severe (9 of 23). The mean resolution time was 33.00 ± 7.07 days. After medical treatment, 18 patients showed improvement in clinical manifestations and visual acuity. However, 5 out of 23 (all categorized as severe) required procedures, such as penetrating keratoplasty (PKP) or deep anterior lamellar keratoplasty (DALK; Table 1). 

Table 1.

 

Clinical and Genomic Profiles of 23 Patients With Nocardia Keratitis

 

Among the Nocardia strains isolated from the cornea and identified at the species level, N. puris were the most numerous (7 strains, 30.4 %), followed by N. farcinica (5 strains, 21.7 %), and N. abscessus (4 strains, 17.4 %). To confirm the antimicrobial susceptibility of Nocardia isolated from the cornea, the minimum inhibitory concentration (MIC) was examined using the broth microdilution method by CLSI M62, calculated MIC₅₀ and MIC₉₀, and finally listed with different species in Table 2. All 23 strains were sensitive to amikacin, imipenem, and linezolid, whereas 13 strains were resistant to rifampin, and 12 strains were resistant to vancomycin. Interestingly, the values of MIC₅₀ and MIC₉₀ of 2 powerful antibiotics, vancomycin (MIC₅₀ = 16 mg × L⁻¹ and MIC₉₀ = 64 mg × L⁻¹) and rifampin (MIC₅₀ = 4 mg × L⁻¹ and MIC₉₀ = 8 mg × L⁻¹), were notably higher than that of the others. Similarly, except for N. puris, over half of the strains in each species were resistant to vancomycin. For the other five antibiotics (ceftriaxone, clarithromycin, tobramycin, trimethoprim-sulfamethaxole, and moxifloxacin), one to eight strains exhibited resistance. 

Table 2.

 

Antimicrobial Susceptibility and MICs of 23 Nocardia Isolated From the Cornea (n = 3)

 

Genomic Characteristics and Whole-Genome Phylogenetic Analysis

After passing quality control filters, the whole genomes of 141 multiple Nocardia strains were successfully investigated. The average nucleotide identity (ANI) values for 141 Nocardia strains surpassed 95%, underscoring the accuracy of our identification against reference sequences. The mean genome sizes of corneal, respiratory, and environmental genomes were 6.95 Mb, 6.44 Mb, and 6.67 Mb, with mean GC content values of 69.1%, 68.5%, and 66.9%, respectively. 

Among the Nocardia identified at the species level, N. farcinica was the most numerous (34 strains, 24.1%), followed by N. cyriacigeorgica (28 strains, 19.9%), N. puris (9 strains, 6.4%), and N. abscessus (8 strains, 5.7%). To further evaluate the relationship between the strains, a circular whole-genome phylogenetic tree was constructed using the whole genome sequence of 141 multiple Nocardia strains (Fig. 1). The label of each strain in the phylogenetic tree consists of the identification results of Nocardia and the strain name. Remarkably, most of the same species of Nocardia were harbored in the same clade suggesting the accuracy and precision of our whole-genome phylogenetic analysis. The general background profiles including isolated source, collection year, and geographic origin, of 141 multiple Nocardia strains were added around the circular whole-genome phylogenetic tree with 3 rings, respectively. The isolated sources from which most strains were obtained were the respiratory tract (79 strains, 56.0%), environment (39 strains, 27.7%), and cornea (23 strains, 16.3%). Nocardia isolated from the cornea included seven different species, most of which were N. puris (7 of 23). As to the collection year and geographic origin, 65.3% of Nocardia were collected in the second decade of the 21st century and China contributed 72.3% (102 strains) of the Nocardia whole genome sequence, followed by Venezuela (8 strains, 5.7%), and the United States (5 strains, 3.6%). 

Figure 1.

 

Whole-genome phylogenetic analysis of 141 multiple Nocardia strains. The innermost colored bands represent 44 different species of Nocardia and Nocardia spp. (unable to identify species). (A) A colored ring showing three different isolated sources of Nocardia. (B) A colored ring showing five different periods of collection years and the unknown collection years. (C) A colored ring showing 18 different geographic origins of Nocardia.

Functional Differences of Three Isolated Sources With Pan-Genome Analysis

To further investigate the functional differences among 3 isolated sources, pan-genome analysis was performed among the 141 genomes by BPGA with 50% sequence identity cutoff by default. The corneal genome size comprised 31,959 genes, whereas the number of respiratory and environmental genomes was 45,835 and 60,187 genes. Four hundred twenty-one genes were identified as core genes, shared among all corneal strains, whereas 1458 core genes were present in respiratory duct strains, and 224 core genes in environmental strains. Consequently, the accessory genes numbered 31,538 genes, 44,377 genes, and 59,963 genes for corneal, respiratory, and environmental genomes, respectively (Fig. 2A). Moreover, the corneal, respiratory, and environmental genomes were open, with the Bpan of these genomes less than 1 (0.46, 0.35, and 0.47). Based on the openness of the genomes, the gene function profiles, and the functional differences were further assessed among three isolated sources. The COG categories and annotations of predicted genes within the core, accessory, and unique genomes were evaluated and graphed as stacked bar charts (Fig. 2B). Twenty functional COG categories in 3 isolated sources were observed with significant differences in 2 categories (J and R) of the core genome, 5 categories (H, K, M, P, and Q) of the accessory genome, and 10 categories (E, F, G, H, I, J, K, P, Q, and S) of the unique genome. These data indicated the characteristic living patterns of Nocardia from the different sources which were assigned to nucleotide transport and metabolism (category F), cell wall/membrane/envelope biogenesis (category M), and secondary metabolites biosynthesis, transport and catabolism (category Q). Additionally, there is the possibility that other categories hold associations with Nocardial virulence, potentially playing a pivotal role in the onset and progression of Nocardia keratitis. 

Figure 2.

 

Pan-genome analysis of 141 multiple Nocardia strains grouped by three different isolated sources. (A) Genome size evolution curves of the pan-genome (blue) and core genome (orange) grouped by three different isolated sources. (B) The stacked bar chart shows the relative abundance (%) of COG categories and annotations of predicted genes within the core, accessory, and unique genomes of three isolated sources of Nocardia. All the categories are graphed as the percentages of the total number of genes in the core, accessory, and unique genomes.

Analysis of Virulence Genes and the mce Gene Family

Virulence factors play a crucial role in regulating the initiation and progression of Nocardia infection. Based on the virulence genes within the whole genomes of 141 multiple Nocardia strains, a dot chart was generated (Fig. 3), and 541 virulence genes were identified using BLASTp against the VFDB database. Each genome had a mean of 220 virulence genes, which contributed to various aspects of bacterial pathogenesis, such as adherence (31 genes on average, 14.10%), invasion (7 genes on average, 3.25%), motility (20 genes on average, 9.01%), exotoxin production (8 genes on average, 3.82%), biofilm formation (5 genes on average, 2.37%), etc. Among them, hemB, hemE, hemH, and hemN were present in almost all strains, and the flp genes and hem operons mainly encoded the adherence function of Nocardia. The hly operons and mls genes, which encoded hemolysin and mycolactone almost took up all the exotoxin-related genes. The presence of motility-related genes in Nocardia was limited, with only a small number of flm operons contributing to this function, and minimal variation observed among the 141 strains. The mce gene family, a class of genes encoding proteins involved in Nocardia cell invasion, had nine clusters (mce1-9). Among these, mce4, mce5, and mce7 were found in the majority of Nocardia strains, whereas mce1, mce3, mce6, mce8, and mce9 showed distinct distributions among the 141 strains. Interestingly, mce2 was detected in only one strain (NBRC108239). Furthermore, six virulence genes or operons that occur with high frequency in cornea-derived Nocardia (flp genes, glc operon, hly operon, mls genes, flm operon, and ibe genes), but to our knowledge, these genes or operons have not been reported or comprehensively researched in other studies of Nocardia. 

Figure 3.

 

Distribution of four typical virulence factor categories within the whole genome of 141 multiple Nocardia strains. Each dot represents the presence of a virulence factor. Four colors (violet, green, blue, and red) correspond to four virulence factor categories (adherence, exotoxin, motility, and invasion), respectively.

Categorization of Nocardia Isolated From the Cornea Based on the mce Gene Family

The invasion function in Nocardia bacteria primarily relies on the mce gene family. Cluster analysis of the mce gene family distribution categorized 23 corneal Nocardia strains into 3 discernible groups: group A, group B, and group C. Within group A, the mce5A, mce5C, and mce5E genes were consistently present in almost all strains. Group B exhibited the presence of mce4C and mce4F genes across all strains. In group C, mce3C, mce4F, and mce5D were detected in all strains, whereas mce3C was only observed within group C (Fig. 4A). Within these groups, group A included eight isolates from patients (34.8%) with mild symptoms, including N. wallacei, N. asiatica, and N. abscessus. Group B had seven isolates from patients (30.4%) with moderate symptoms covered the species of N. asteroids, N. cyriacigeorgica, N. puris, and N. farcinica. In addition, group C comprised eight isolates from patients (34.8 %) with severe symptoms, which only belonged to N. farcinica and N. puris. Figure 4B depicted representative images of the anterior segment for each group. This distribution pattern provided insights into the specific gene variations that might contribute to the distinct characteristics of these Nocardia strain groups, emphasizing a potential correlation between the severity of clinical manifestations and the categorization of these groups. 

Figure 4.

 

Twenty-three strains of Nocardia isolated from the cornea can be categorized into three groups. (A) Twenty-three strains of Nocardia were grouped based on the distribution of the mce gene family. Each dot represents the presence of a mce gene. (B) The representative anterior segment photographs of patients with Nocardia keratitis correspond to three different groups. (C) The representative slit lamp microscopic and AS-OCT images of Nocardia colonies on BAP media correspond to three different groups (n = 3). The morphological changes were observed and photographed at 0.5, 1, 1.5, 2, 2.5, and 3 days post-infection. (D) The increasing trends of diameter, height, and angle in three groups of Nocardia (n = 3). One-way ANOVA followed by Tukey post hoc tests were performed to assess differences among three groups of Nocardia at the same incubation time (***P < 0.001, **P < 0.01, *P < 0.05, group A versus group B; ###P < 0.001, ##P < 0.01, #P < 0.05, group A versus group C; …, P < 0.001, …, P < 0.01, …, P < 0.05, group B versus group C).

Analysis of Corneal Pathogenicity Characteristics Based on Grouping With Virulence Genes

Validation of Colony Growth Characteristics of Nocardia

After culture on BAPs and incubated at 35°C, Nocardia colonies among the 3 groups were identified and their different growth patterns were photographed over a 12-hour interval (from day 0 to day 3) using slit lamp microscope and anterior segment OCT (AS-OCT; Fig. 4C). The diameter, height, and angle of the colonies were quantified and displayed in the line charts (Fig. 4D). Colonies of group A exhibited a distinctive ring-like growth pattern with a notably slower growth rate. The high-reflective signal beneath the center of the bottom in the cross-section images correlated with the thinner center of the ring seen in the slit lamp microscopy images. Group B colonies showed a mound-like shape with a smaller diameter, height, and moderate growth rate. The colonies of group C had mound-like morphology with the largest diameter (1850 ± 82.87 µm; P = 0.0006, versus group A; and P < 0.0001, versus group B), height (520 ± 13.59 µm; P < 0.0001, versus group A; and P < 0.0001, versus group B), angle (33.57 ± 0.76 degrees, P < 0.0001, versus group A; and P < 0.0001, versus group B), and fastest growth rate. 

Validation of Clinical Manifestation of Mice Nocardia Keratitis

To assess the differences in pathogenicity of Nocardia keratitis after grouping based on mce genes, a Nocardia keratitis mouse model was established. Clinical scores and photographs of corneal ulcers with and without 1% sodium fluorescein were conducted at specified time points (day 1, day 3, and day 7; Fig. 5A). For the control group, mice had intact corneas with minor changes due to PBS injection on day 1. Conversely, all intervention groups exhibited evident corneal infiltrates and ulcers. In group A, on day 1, the cornea responded to epithelial ulcer and moderate inflammation with dense stromal infiltrate. However, the lesion would be improved without treatment, showing a self-limited trend in the following days. In group B, the clinical manifestations and score on day 3 (P = 0.0021) and day 7 (P = 0.0015) were significantly more severe or higher than those in group A (Fig. 5B). Furthermore, the cornea response in group C was significantly more severe than those in group A and group B. On the first day, mice in group C exhibited clinical indications of corneal thinning. The inflammation steadily worsened, leading to more intense infiltration and opacity, consistent with the increasing trend of clinical scores (see Fig. 5B). For the outcome of Nocardia keratitis, both the size of corneal ulcers and the clinical scores increased in group C compared with the group B, with statistically significant differences (for clinical score, P = 0.0167; and for corneal ulcer, P = 0.0156; see Figs. 5B, 5C). The CFU assessment exhibited that group A had the lowest viable Nocardia count (3.16 × 10³ CFU/mL) on day 1 post-infection, followed by group B (8.19 × 10³ CFU/mL), and group C had the highest quantity (1.83 × 10⁴ CFU/mL). The differences among the three groups are statistically significant (F = 84.61, P < 0.0001). As the observation period progressed, the bacterial count in the corneas of the three groups steadily decreased. Interestingly, by the seventh day, the mean CFU of group A had reduced to 0.56 × 10³ CFU/mL, exhibiting no statistically significant difference compared to the control group (Supplementary Fig. S1). 

Figure 5.

 

The anterior segment photographs and clinical scores of Nocardia keratitis model mice in four groups (control group, group A, group B, and group C). (A) The representative anterior segment photographs with and without fluorescein staining under slit lamp microscopy on days 1, 3, and 7 (n = 3). (B) The clinical scores and (C) the diameters of the corneal ulcers were assessed in each group, and three mice were assessed in each group for the indicated time (n= 3). One-way ANOVA followed by Tukey post hoc tests were performed to assess differences between four groups of Nocardia at the same time post-infection (****P < 0.0001; ***P < 0.001; **P < 0.01; *P < 0.05; ns, no significance).

Validation of Inflammation, Immune Reaction, and CFU of Viable Pathogen of Mice Nocardia Keratitis

Compared with the control group, the inflammation and immune reaction on day 7 were present with an exacerbating tendency from group A to group C. Compared to the control group, the mean number of cells staining with H&E were 242.00 cells per field (P = 0.0378), 449.33 cells per field (P < 0.0001), and 621 cells per field (P < 0.0001), the mean relative optical density of IFN-γ was 2.74 folds (P = 0.0019), 3.53 folds (P = 0.0001), and 4.17 folds (P < 0.0001), and the mean count of CD4+ cells were 2.00 cells per field (P = 0.2503), 4.67 cells per field (P = 0.0032), and 8.33 cells per field (P < 0.0001), which both exhibited ascending trends from group A to group C. This effect was particularly pronounced in the corneal stroma, aligning with the outcomes of histological examination and clinical presentation. The relative expression levels of IL-6Rα and TNF-α demonstrated analogous patterns to the aforementioned biomarkers. For example, the mean relative expression levels of IL-6Rα in groups A, B, and C were 2.28 folds (P = 0.0086), 2.90 folds (P = 0.0006), and 3.89 folds (P < 0.0001), respectively, and the mean relative expression levels of TNF-α were 1.69 folds (P = 0.0116), 2.44 folds (P < 0.0001), and 3.72 folds (P < 0.0001), respectively (Fig. 6). 

Figure 6.

 

Histological analysis and detection of CD4+ cells, IFN-γ, IL-6Rα, and TNF-α expression in the cornea of Nocardia keratitis mouse model. Representative images of histological analysis by H&E staining revealed the presence of epithelial, stromal, endothelial, and inflammatory cells in the control group and three intervention groups. Representative images of immunohistochemistry (IHC) staining of IFN-γ and CD4+ cells, immunofluorescence (IF) staining of IL-6Rα and TNF-α reveal the inflammation in the cornea of Nocardia keratitis mouse model (n = 3). The results were plotted as the number of cells stained with H&E, positive cells per field (CD4), or fold changes compared with the control group (IFN-γ, IL-6Rα, and TNF-α). Three fields of view per section were selected for mean value calculation and statistical analysis (****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05). Scale bar, 50 µm.

Analysis of Antibiotic Resistance Genes Among the Nocardia Isolated From the Cornea

To detect antibiotic resistance properties of Nocardia at the genomic level, ARGs were identified by BLASTp against CARD. These ARGs spanned nine distinct drug classes, four resistance mechanisms, and seven primary AMR gene families. Table 3 presents the average gene length and distribution of these ARGs. Particularly noteworthy, the vancomycin resistance gene family covered every strain isolated from the cornea. Among them, vanB, vanI, and vanG were the most widely distributed. Moreover, the rifampin resistance gene family (rox-sv and rox-nf) covered 87.0% of the corneal isolated strains (20 strains). 

Table 3.

 

The Antibiotic Resistance Genes Within the Whole Genome of 23 Strains of Nocardia Isolated From the Cornea

 

Discussion

In this study, whole genome sequencing was performed on 23 Nocardia strains isolated from the cornea in the current cohort. The objective was to gain a thorough understanding of the phylogenetic relationship among Nocardia strains from different sources, including the cornea, respiratory tracts, and environment, as well as to explore the similarities and differences in their biological functions and virulence properties. As a vital virulence factor in Nocardia strains isolated from the cornea, the mce gene family was also further investigated. The distinct distribution of mce genes led to the categorization of cornea-derived Nocardia strains into three groups. The accuracy of this categorization was verified by in vitro analysis of strain growth patterns and in vivo testing using a Nocardia keratitis mouse model. The model was assessed using clinical scores and various histopathologic analyses, including H&E staining, immunochemistry, and immunofluorescence staining with specific biomarkers (CD4, IFN-γ, IL-6Rα, and TNF-α). To date, it was the largest series with Nocardia keratitis on clinical profiles and pathogenetic grouping based on WGS. Additionally, the study involved an analysis of antimicrobial susceptibility testing and ARGs to detect the unique antimicrobial resistance properties of the Nocardia strains in this particular cohort. 

In this cohort, Nocardia keratitis had a longer mean presentation time (20.50 ± 2.86 days) and resolution time (33.00 ± 7.07 days) compared to other forms of bacterial or infectious keratitis (4.0–14.0 days^31–33 and 5.8–14.0 days^33–35), and consistent with the previous cohorts of Nocardia keratitis.^36,37 Among the various strains, N. puris was the most common isolate, followed by N. abscessus, N. cyriacigeorgica, and N. farcinica. Despite the genetic similarities observed among the corneal Nocardia isolates, there were variations in terms of antimicrobial resistance properties of the strains, the clinical characteristics, severity, and outcomes of Nocardia keratitis. Notably, all strains showed sensitivity to amikacin, imipenem, and linezolid. However, apart from N. puris, over half of the strains within each species were resistant to vancomycin. Furthermore, 13 strains displayed resistance to rifampin, showcasing peculiar antimicrobial resistance profiles that differed from previous studies.^36–38 

Combining with the whole genome sequences online, most of the same species of 141 strains of Nocardia from different sources were harbored in the same clade suggesting the accuracy and precision of our whole-genome phylogenetic analysis. Among the strains studied, 32 of 34 strains of N. farcinica and all strains of N. puris were isolated from China, suggesting certain species of Nocardia were restricted in specific regions with the influence of climate and environmental factors.³⁹ The pan-genome analysis focused on the collective set of genes found across all strains and revealed that the pan-genomes of Nocardia from all three sources were “open,” which means that these Nocardia strains have the capacity to colonize and exploit various environmental niches. This adaptability is likely achieved through horizontal gene transfer,⁴⁰ allowing them to acquire new genetic material and the capacity to survive in a complicated host environment.⁴¹ The distinctive functional distribution among COG indicated the characteristic living patterns of Nocardia from different isolated sources, which were assigned to multiple categories. Especially for core and accessory genomes of Nocardia, the gene cluster functioning as cell wall/membrane/envelope biogenesis (category M), displayed significant differences between corneal and environmental isolates. 

Unlike environmentally isolated Nocardia, the cornea-derived Nocardia needed to interact with and invade corneal tissue, therefore, had the possibility that categories M-related genes hold associations with the virulence of Nocardia, potentially playing a pivotal role in the onset and progression of Nocardia keratitis. In this study, a mean of 220 virulence factors and corresponding virulence genes were detected, which played crucial roles in regulating the initiation and progression of Nocardia infection. For Nocardia isolated from the cornea, virulence genes predominantly contributed to the pathogenesis of the strains and interactions between host and pathogen, such as adherence, invasion, motility, exotoxin production, biofilm formation, etc. After adherence to host cells, invasion is the key pathogenetic process of Nocardia keratitis.⁴² For genes among them, the mce gene family is a cell wall-related protein, which has an association with category M in COG and is proven to have crucial effects on the invasive process of Nocardia,⁴³ especially into the cornea and their survival in phagocytes, epithelial cells, under hypoxic condition.^44–48 Because mce4, mce5, and mce7 were found in the majority of Nocardia strains, mce1, mce3, mce6, mce8, and mce9 showed distinct distributions. In addition, every cluster of mce gene had multiple genes involved (normally from A to F), these provided a basis for grouping based on the distribution of the mce gene family. 

For the peculiar distribution of specific mce genes, within group A, the mce5A, mce5C, and mce5E genes were consistently present in almost all strains and potentially played a critical role in the invasion of host cells under hypoxic conditions however further research is needed.⁴⁹ All strains across group B displayed the presence of mce4C and mce4F genes with multiple pathogenic effects which were well-studied. For encoding a cholesterol transporter,^50,51 mce4 contributes to the entrance of Nocardia into a tryptophan aspartate-containing coat protein (TACO) – coated phagosomes to prevent the degradation of lysosomes,⁵² meanwhile promoting the localization and invasion of Nocardia to host cells through cholesterol-rich domains of the plasma membrane.^53,54 Furthermore, mce4C and mce4F genes allow Nocardia to grow under hypoxic conditions, by taking in cholesterol as a carbon source, therefore, providing energy for Nocardia survival inside the cornea and deeper invasion.⁴⁸ All strains in group C contained the mce4F gene, thus obtaining similar basic pathogenic properties to group B. Mce4F, an essential component of the mce4 operon, is found in the genomes of mycobacteria and Nocardia.⁵⁵ Mce4F encodes external membrane proteins, crucial for bacterial pathogenesis.⁵⁶ This study indicates the highly conservation of mce4F across various Nocardia strains, making it a valuable and reliable diagnostic marker, especially for early detection, saving time compared to the culture method. Additionally, Kendall et al.⁵⁷ demonstrated that the KstR|Rv3574 transcriptional regulon controls mce4F expression through lipid catabolism, vital for bacterial survival. Therefore, mce4F shows promise as a potential therapeutic target for combating these infections. Remarkably, the mce3C gene was only detected in the strains of Nocardia in group C, which triggered the immune escape by promoting Nocardia to invade and survive within macrophages and other inflammatory cells.^47,58 The ubiquitous presence of mce3C within group C might be the cause of poor clinical manifestations and outcomes of the patients and also can be the biomarker to evaluate the prognosis of Nocardia keratitis, especially in the early stage of the disease. 

In previous studies, Mce proteins were suggested as diagnosis markers for various mycobacterial infections.^54,59,60 However, specific mce gene (such as mce3C, mce4C, mce4F, etc.) that can be easily detected using polymerase chain reaction (PCR) has not been proposed yet. Interestingly, distinct growth patterns were observed among the three groups of Nocardia strains, and these were confirmed using slit lamp microscopy and AS-OCT. The appearance of the colonies also differed: group A exhibited a distinctive ring-like colony, whereas groups B and C were mound-like. The growth rate increased significantly from group A to group C, as evident from trends in colony diameter, height, and angle. Furthermore, the growth pattern of Nocardia colonies might correlate with the clinical manifestations, especially the ring-like colony of group A, whose patients’ lesions showed wreath-like infiltration. Additionally, the rate of Nocardia colony growth was linked to patient outcomes. Group C, with the fastest growth rate, had a higher proportion of patients in the cohort requiring therapeutic keratoplasty, indicating worse outcomes. 

The phenotypes of Nocardia keratitis corresponded to genotype-based groupings of Nocardia strains isolated from the patients. In group A, Nocardia keratitis is milder and self-limiting, with clinical scores and corneal ulcer tapering off and ending up with corneal scarring formation on day 7. Group B presented moderate course and outcomes with relatively slow progression, whereas group C initiated with acute onset in one day and exacerbated in the following week, resulting in central corneal thinning and perforation in some of the mouse models. The results of CFU assessment were consistent with clinical outcomes, with mice in group A having the lowest CFU, group B the next highest, and group C the most. The total amount of Nocardia isolated from the cornea was reduced in all three groups, which was mainly a result of the immune response to the infection and thus killing the pathogen. Compared to the control group, H&E staining exhibited an increasing trend of inflammatory cell activation in the lesion area. IFN-γ, a pro-inflammatory cytokine produced by various inflammatory cells (neutrophils, macrophages, B lymphocytes, and T lymphocytes, etc.).⁶¹ The relative OD value of IFN-γ also reflects the exacerbation of inflammation condition, especially in the corneal stroma which corresponded to the extent of corneal infiltration and opacity in the animal model. Previous studies demonstrated that IL-6R was essential to the immune response,⁶² furthermore, IL-6Rα as the main functional subunit of IL-6R,⁶³ acted as the key protein in the signal pathway for driving naive CD4+ T cells to the Th17 lineage T cells (mature CD4+ T cells) by dendritic cells.⁶⁴ Thus, Nocardia infection triggered the immune response in the cornea and upregulated the expression level of IL-6R, as well as IL-6Rα, further promoting the maturation of CD4+ cells to Th17 lineage T cells with an increasing trend from group A to group C, which was validated by immunofluorescence staining in the current study. In addition, the mononuclear phagocyte system (MPS) was also activated by Nocardia infection in identical trends, as the positive staining of TNF-α, which was primarily released by MPS.⁶⁵ 

The ARGs among 23 cornea-derived Nocardia were eventually detected. Similar to the antimicrobial susceptibility testing, the ARGs of two powerful antibiotics, vancomycin and rifampin widely distributed, differing from previous clinical strains⁶⁶ and the isolates from respiratory tracts and environment (Supplementary Fig. S2), which probably was the result of horizontal gene transfer and topical antibiotic therapy. Moreover, clusters vanB and vanI contributed to the mainstay of the vancomycin resistance gene, which varied from other vancomycin-resistant bacteria.^67–70 Of the 122 downloaded Nocardia strains, 27 had antibiotic susceptibility test results (Supplementary Table S1). Among the 10 antibiotics included in this study, there was no significant difference in the susceptibility of cornea-derived Nocardia to the four antibiotics (i.e., amikacin, ceftriaxone, moxifloxacin, and tobramycin) when compared with non-cornea-derived Nocardia; significantly higher susceptibility to clarithromycin (P < 0.001), imipenem (P < 0.001), linezolid (P < 0.05), and methotrexate-sulfamethoxazole (P < 0.01); and significantly lower susceptibility and resistance to rifampin (P < 0.05) and vancomycin (P < 0.05), which was generally consistent with the results of resistance gene analysis. 

Our study still had several limitations. First, the bioinformatics methods were mainly used to detect the presence of specific genes in Nocardia strains. Although the algorithms of those were proven accurate and reliable, it would be better to validate the mce gene family distribution in vivo on nucleic acid levels with molecular biological methods. Second, due to a rather rare ocular infection and its relatively low incidence, the cohort of patients with Nocardia keratitis was relatively limited in size. It would be advantageous to construct a prospective cohort study involving a larger number of patients. This expanded cohort would provide more comprehensive data for further research and a better understanding of Nocardia keratitis. Third, although based on the practical considerations, only female mice were involved in these animal experiments. Further studies with both male and female mice would reduce the gender bias and enhance the understanding of Nocardia keratitis classification. At last, this study only proposed a categorization of cornea-derived Nocardia based on the distribution of the mce gene family, however, more research focusing on the pathogenesis of Nocardia keratitis is needed. 

In summary, our study has provided valuable insights into the profile of virulence and antimicrobial resistance genes among various Nocardia strains. This has enhanced our comprehension of Nocardia keratitis and offered guidance for improved diagnosis and treatment strategies. By using whole genome sequencing, we have underscored the significance of the mce gene family in genotype-based classification to Nocardia keratitis. 

Acknowledgments

Supported by grants from the National Key Research and Development Program of China (2021YFC2301000). 

Disclosure: X. Guo, None; Z. Zhang, None; Q. Chen, None; L. Wang, None; X. Xu, None; Z. Wei, None; Y. Zhang, None; K. Chen, None; Z. Wang, None; X. Lu, None; Q. Liang, None