Background

Staphylococcus aureus is an important human pathogen causing a variety of infections in both healthcare facilities and community settings [1]. S. aureus is among the top three causes of nosocomial infection in Ethiopia. It was found to be the leading cause of nosocomial infection, accounting for 26.2% and 35.6% of infections in two hospitals in Ethiopia [2, 3]. In another hospital, S. aureus was the third leading cause, accounting for 20.6% of hospital-acquired infection [4]. The majority of studies were focused on antimicrobial susceptibility testing, with limited information on molecular epidemiology in Ethiopia [5, 6]. There was a single study showing the epidemiology of S. aureus using strain typing [7]. S. aureus are constantly changing, with novel strains appearing in different geographical regions [8,9,10]. Molecular characteristics of MRSA can be diverse in different hospitals within the same country [11].

Pulsed-field gel electrophoresis (PFGE), Multi Locus Sequence Typing (MLST) and staphylococcal protein A (spa) typing have been used extensively to identify different S. aureus strain types. For methicillin-resistant S. aureus (MRSA), molecular typing of the staphylococcal cassette chromosome (SCC)mec, which harbors the gene encoding methicillin resistance, provides additional strain discrimination [12].

The most frequently reported global MRSA clonal complexes (CCs) include the following: CC1, CC5, CC8, CC22, CC30, CC45, CC59 CC80, and CC239. Many of these are distribute globally, while others are restricted to particular region [13]. The most widely distributed healthcare associated (HA)-MRSA clone include ST239-MRSA-III, ST22- MRSA-IV while the most common community associated (CA)- MRSA clones include ST8-MRSA-IV (USA300), ST80-IV-MRSA and ST30-IV-MRSA which has been reported in many countries around the world. On the other hand, clones such as ST59-MRSA-IV and ST93-MRSA-IV have displayed comparatively restricted geographical spread. The pandemic HA-MRSA clone, ST239/ST241-III-MRSA has been reported since 1980 and 1990 s from most parts of the world including Africa, Australia, Europe, Asia, North and South America [14,15,16].

Although, genotypic reports of S. aureus and MRSA in Africa are limited, some have highlighted the major clones circulating within the continent. Commonly reported CC circulating in Africa include CC5, CC7, CC21, CC30, CC121 and CC152 [7, 17, 18], as well as MRSA strains with different SCCmec types belonging to CC1, CC8, CC22, and CC88. Notably, the globally-distributed HA-MRSA strain ST239/ST241-MRSA-III has only been identified in Egypt, Ghana, Kenya, and South Africa [19].

Molecular genotyping of S. aureus prospectively in healthcare setting can determine prevalent strains, identify outbreaks and transmission routes of newer strains, and implement control and prevention of S. aureus spread within healthcare settings. The aim of the current retrospective study was to determine the molecular epidemiology of MSSA and MRSA from multiple antimicrobial resistance (AMR) surveillance sites in Ethiopia to evaluate for any potential clusters outbreak transmissions prospectively.

Methods

Study site description

Ethiopia has been implementing a laboratory-based AMR surveillance program since 2016 [20]. Currently, more than 10 sentinel sites are networked within the national AMR surveillance system. Among these sites, four have been actively participating in the surveillance program since the program was initiated. The four sites included in this study were: Tikur Anbessa Specialized Hospital (TASH), Addis Ababa; Amhara Public Health Institute - Dessie Branch (APHI), Dessie; Ayder University Hospital (AUH), Mekelle; and the Clinical Bacteriology and Mycology National Reference Laboratory at the Ethiopian Public Health Institute (EPHI), Addis Ababa.

Sampling strategy

A total of 190 stored S. aureus isolates from the aforementioned AMR Surveillance sites were characterized in this study. The clinical specimens included wound/pus (n = 167), blood (n = 8), ear swabs (n = 6), and other body fluids including eye swabs (n = 9). The isolates were collected from 2016 to 2019 from their respective sites and transported to EPHI. The isolates used in the study were stocked using 20% glycerol and tryptic soya broth in a cryotube and stored at -80 freezer for further analysis. The isolates were tested at the respective AMR surveillance sites using classical microbiological methods. Specimens were cultured on sheep blood agar plate (BAP) and beta-hematolytic colonies with characteristics indicative of S. aureus were further sub-cultured on mannitol salt agar (MSA). Yellow colonies were then sub-cultured onto nutrient agar and isolates were identified as S. aureus based on catalase and coagulase positivity. The isolates were then shipped to The Ohio State University (OSU) for molecular characterization.

Diagnostic testing

All of the isolates were tested for antimicrobial susceptibility, spa type, Panton-Valentine leucocidin (lukF-PV), toxic shock syndrome toxin (tst), and 5 staphylococcal enterotoxin genes (sea, seb, sec, seh, sej) as previously described [21]. In addition, the isolates were characterized using PFGE, spa typing and the MRSA isolates were tested by using SCCmec typing. A subset of isolates with novel spa type patterns were also subjected to MLST analysis. The study protocol was approved by the EPHI Institutional Review Board (IRB) (Unique identifier: “EPHI-IRB-029-2017”). Data analyses were anonymous, and all phases of the study did not identify patients in any way.

Nucleic acid isolation

Genomic DNA was extracted using a commercially available kit (QIAamp DNA mini Kits, Germany), following the manufacturer’s protocol [22]. DNA extraction and PCR tests were done at the Infectious Disease Epidemiology Molecular Laboratory (IDEML) at OSU. The extracts were stored at -20 °C until further analysis. MLST and spa typing was performed at the Center for Discovery and Innovation, Hackensack Meridian Health, Nutley, New Jersey, USA.

Staphylococcal Cassette chromosome (SCC)mec typing and SCCmec-IV sub-typing

Nine MRSA isolates that were confirmed with mecA detection were tested for (SCCmec typing using PCR. Identification of cassette chromosome recombinase (ccr) alleles and mec class was used to determine SCCmec type as previously described [23]. For PCR master mix, illustra™ PuReTaq™ Ready-To-Go™ PCR Beads (GE Healthcare Bio-Sciences, USA) were used in 25 µl reactions, and reference strains of known SCCmec types were used as controls throughout the PCR test.

Staphylococcal protein A (spa) typing

Spa typing was performed using PCR followed by Sanger sequencing, as previously described [24]. Assignment of spa type Based Upon Repeat Pattern (BURP) analysis for determination of spa clonal complex (spa-CC) was performed using Ridom StaphType. The Ridom SpaServer was used to predict the multi-locus sequence types (STs) as described previously [25]. Phylogenetic trees were constructed using RAxML Tree, Geneious Version 2022.2, (https://www.geneious.com).

Multi-locus sequence typing (MLST)

MLST was performed using the method described by Enright et al. [26] for 12 isolates with novel/unknown spa type patterns (11 MSSA and 1 MRSA). Allelic sequences for each gene were analyzed using Geneious software, and used to query the S. aureus database (https://pubmlst.org/saureus/) [27]. Clonal complexes (CC) were inferred for 6 of the strains using the BURST analysis software available on the PubMLST server; the remainder were singletons or novel sequence types unrelated to any other in the database.

Pulsed Field Gel Electrophoresis (PFGE)

PFGE was performed for further characterization of both MSSA and MRSA isolates. The isolates were selected randomly, while considering the site and specimen type to identify clonal relatedness. DNA fingerprinting was performed by macro-restriction of chromosomal DNA using SmaI (New England Biolabs, Ipswich, MA, USA) and pulsed field gel electrophoresis (PFGE) as described previously [28]. The PulseNet “universal” standard strain Salmonella enterica serovar Braenderup H9812 was used as a reference marker. The chromosomal fragments were separated using a CHEF-DR®III Pulsed-Field Electrophoresis System (Bio-Rad Laboratories, Hercules, CA, USA). Gel images were analyzed using Bionumerics Gelcompar II version 6. 6. software (Applied Math inc., Belgium). Cluster analysis was performed using the unweighted pair group method with arithmetic averages (UPGMA). Similarity coefficients were determined using Bionumerics by calculating the Dice coefficient similarity index. A similarity coefficient of 80% was selected to define individual pulso-types.

Results

Detection of virulence factor genes

Among the 190 total isolates, 172 were confirmed S. aureus. The remaining 172 S. aureus isolates tested for the presence of spa gene yielded positive results, while the mecA gene was detected in 9 of the isolates. None of the tested isolates were positive for mecC. Among the S. aureus isolates, 102 (59.3%) possessed the lukF-PV gene. A total of 66 (38.4%) isolates harbored at least one staphylococcal enterotoxin gene, while 31 (47.0%) isolates had more than one.

Methicillin-susceptible Staphylococcus aureus (MSSA)

Spa typing

The spa typing analysis revealed 56 distinct spa types among the 170 S. aureus isolates, the most common being: t355 (56/170, 32.9%), t085 (13/170, 7.6%) and t314 (11/170, 6.5%). The other spa types had less than 5% frequency each. Eleven novel spa types were also identified (Table 1) and registered in the Ridom SpaServer database (https://spaserver.ridom.de/). The spa types identified were clustered into 15 spa-clonal complexes (spa-CCs) by BURP analysis; however, 18 of the isolates could not be identified by this method (Fig. 1). The majority of the isolates belonged to spa-CC 152 (62/170, 36.5%), and consisted of four spa types: t1172 [1], t1299 [3], t355(56) and t454[2]; followed by spa-CC 121 (19/170, 11.18%), spa-CC 5 (18 /170, 10.59%), spa-CC 15 (15/170, 8.82%), and spa-CC 22 (10 /170, 5.88%). MLST was performed (Table 2) for several spa-types which could not be assigned to spa-CC, including t777, t1916, t1991, and t5338.

Table 1 Novel spa types identified
Full size table
Fig. 1
figure 1

Clonal complexes of Staphylococcus aureus isolates identified by Based on Repeat Pattern BURP analysis. spa-CC, clonal complex

Full size image

CC: Clonal complex; unk: unknown.

Multi-locus sequence typing (MLST)

Table 2 shows the MLST housekeeping genes, spa and clonal complex (CC) information for the 11 MSSA and one MRSA isolate. Among the 11 MSSA strains, 5 were assigned to CCs, including CC5, CC15, CC25, CC121 and CC398; these strains corresponded to spa-type t777, t605, t1916, t1991, and t5338, respectively. The remaining 6 MSSA strains displayed considerable variety, with multiple novel/unknown alleles in all 7 of the MLST housekeeping genes, which did not correlate to any existing sequence types (ST), nor to any known CCs. The isolate with spa-type t20037 is a double-locus variant of ST-5695, while isolate with spa-type t20038 is a single-locus variant of ST-2294; isolates with spa-types t20040 and t20041 are both singletons most closely related to ST-1852 and ST-5696, respectively; while isolate with spa-type t20036 is distantly related to ST-6984 (only matching 3 out of 7 loci). the closest match to the MRSA strain was ST 140 and was assigned to CC398; this corresponded to spa-types t3487.

Table 2 Multi-locus sequence typing housekeeping genes, spa type, and clonal complexes of methicillin-susceptible Staphylococcus aureus and methicillin-resistant Staphylococcus aureus isolates
Full size table

PFGE

Based on the PFGE analysis, MSSA isolates (n = 41) exhibiting more than 80% similarity were grouped into eight pulso-types (A, B, C, D, E, F, G, H and I) (Table 2). PFGE pulso-type D comprised 8 isolates (8/51 15.7%). Except for one isolate (body fluid), the pulso-type D clusters were all from pus specimens.

Pulso-types E and G consisted of 3 isolates each, whereas the other pulso-types comprised 2 isolates each. All the tested isolates have relatedness. The three isolates contained in a cluster and sub-clusters E, F and I were identified from sample all collected from EPHI. Cluster B strains were identified from sample collected from TASH and EPHI. Cluster A and C strains were identified from sample collected from TASH and APHI, and Dessie respectively. Cluster G strains were identified from sample collected from EPHI and AUH. Cluster H strains were identified from sample collected from APHI and AUH, Dessie and Mekelle. The PFGE pulso-types and spa CC showed close correlation of strains (Table 3).

Table 3 Methicillin Susceptible Staphylococcus aureus isolates pulsed field gel electrophoresis results using 80% similarity cut off
Full size table

Methicillin Resistance Staphylococcus aureus (MRSA)

Spa typing

The 9 MRSA isolates were assigned to four spa clonal complexes (CC 8, CC 55, CC 88, and CC 239) and seven spa types (t30, t86, t306, t311, t688, t1476, t3487) as shown in (Table 4). One MRSA isolate (t3487) could not be characterized by BURP analysis; MLST analysis determined it as a single-locus variant of ST-140, previously associated with CC398.

Table 4 Staphylococcal Cassette Chromosome (SCC)mec and spa type of methicillin-resistant Staphylococcus aureus (MRSA) isolates
Full size table

PFGE

The PFGE analysis for the MRSA isolates (n = 9) yielded three pulso-types (A, B and C) with 93.7, 88.0, and 90.3% similarity, respectively. Each pulso-type consisted of two MRSA isolates (Fig. 2). Two isolates had 58.3% relatedness and the distant MRSA were 50.03% similar with other strains. All MRSA strains showed over 46% relatedness. The two isolates in pulso-type A had identical antimicrobial resistance patterns. Isolates in pulso-types B and C had similar AMR patterns Table 4.

Fig. 2
figure 2

Pulsed-field gel electrophoresis dendrogram showing relatedness of methicillin resistant Staphylococcus aureus isolates

Full size image

SCCmec typing and SCCmec IV sub-typing

Four of the nine (44.4%) MRSA isolates were found to be SCCmec type IV, 2/9 (22.2%) isolates were SCCmec type III, 2/9 (22.2%) isolates were SCCmec type V and 1/9 (11.1%) isolate was SCCmec type VI (Table 4).

Discussion

The molecular epidemiology of S. aureus isolates circulating in Ethiopia has not been well-described previously. In this study, 56 different spa types and 15 spa clonal complexes were identified. Eleven of the spa types were novel, while six could not be assigned to spa or MLST clonal complexes, highlighting the diversity of S. aureus strains in Ethiopia. Among the S. aureus spa types identified, the most common included t085, t314 and t355. Interestingly, spa  type t355, clonal complex cluster CC152 is the most prevalent spa type reported in other east African countries [29]. This cluster was also the most prevalent in another study conducted in Ethiopia [7]. Spa-CC 152, spa-CC 5, spa-CC 8 and spa-CC 30 are among the most prevalent lineages identified in other African countries [17, 29, 30].

The strains characterized by MLST, were all associated with novel sequence types not found in the PubMLST database, with multiple unique alleles identified for each of the seven genes. Similar results were obtained in another study from Ethiopia, where more than half of the strains were shown to comprise novel STs with unique allelic combinations not found within the database [7]. The MRSA strain analyzed by MLST (spa type t3847) was shown to be closely related to ST-140, previously associated with CC398, a livestock-associated lineage [31]. This strain was reported on inanimate objects and from patient infection associated with hospital transmission elsewhere in Africa [18, 32].

Among the MSSA isolates, PFGE analysis identified eight distinct pulso-types, as well as several distantly-related strains. The PFGE results indicated that strains from different geographical areas were genotypically related. Another PFGE-based study conducted on S. aureus isolates from distinct regions of Ethiopia reported similar patterns [33]. The presence of similar strains in widespread areas is possibly related to the large-scale movement of people within Ethiopia, especially to the central region where the capital city Addis Ababa is located. In this study, the spa type assignments were closely correlated with PFGE pulso-types, similar to what has been described in other studies [34].

The three clusters of MRSA strains displayed relatedness regardless of geographic separation. One strain was found to exhibit > 80% relatedness with strains from AUH, Mekelle and EPHI. In addition, all MRSA isolates showed more than 46% similarity, suggesting relatedness between strains from different geographic location. Moreover, strains clustered within a single pulso-type also showed similar antimicrobial resistance properties. Other studies have also highlighted correlations between strain relatedness and antimicrobial resistance patterns [35, 36].

In this study, the most common spa-CC for MRSA strains were spa-CC 5, comprising SCCmec types, IV, V, and VI, and spa types t306, t311 and t688. Another study from Africa also reported spa-CC 5 as the most common lineage among MRSA strains [30, 37]. Previously, SCCmec types I–III were considered to be HA-MRSA, whereas SCCmec types IV and V were considered CA-MRSA [38]. In recent years, however, the distinction between HA-MRSA and CA-MRSA has blurred increasingly in recent years, as a growing number of reports have demonstrated that CA-MRSA lineages are now prevalent in hospitals [39, 40].

Two MRSA strains displayed 97.3% similarity using PFGE, and had identical SCCmec (type IVa) and spa (t086) types, as well as identical antimicrobial resistance patterns. Another two of MRSA strains were spa-CC 239 with SCCmec type III, and displayed multidrug resistance, suggestive of HA-MRSA. This result was consistent with other studies describing CC 239 SCCmec type III as being commonly associated with multidrug resistance and treatment failure [41, 42]. Moreover, CC 239 SCCmec type III was known to cause MRSA pandemic, circulating in many countries and also associated with serious illness such as admittance to ICU and high rate of death [16, 43].

Despite identification of some cluster, the study is limited by the retrospective nature of the isolates tested and lack of clinical and epidemiological information for linking the cases. In addition, the use of PFGE and spa typing are primarily used for local epidemiological investigation and cannot be applied to multiple sites without knowing if these are the endemic strains versus emerging outbreak strains. Unfortunately, more robust genotyping with MLST for all the isolates and/or whole genome sequence (WGS) could not be performed due to budgetary constraints.

Conclusion

The most predominant spa type in Ethiopia was found to be t355, belonging to spa-CC 152. The spa types identified in this study were closely associated with the PFGE pulso-types. Eleven new spa types were identified among the MSSA isolates, while among the MRSA isolates, strains with high antimicrobial resistance and global epidemic potential were identified. These findings highlight the diversity of S. aureus strains in Ethiopia, as well as the presence of potentially epidemic strains circulating in the country necessitating further characterization of S. aureus for antimicrobial resistance detection and infection prevention purposes in prospective study with MLST and/or WGS for country wide analysis.