Introduction

S. aureus is a versatile opportunistic pathogen of humans that has been implicated in various infections1,2. It has an outstanding ability to cause diverse skin to systemic infections in both community and hospital settings3. Its ability to elaborate important enzymes and proteins including ARGs implicated in its pathogenicity makes it an outstanding pathogen3,4,5. The development of the ARGs has led to several studies aimed at searching for newer antibiotics from diverse sources6,7. However, the search for newer antibiotics may not solve the resistance menace by S. aureus taking a cue from the introduction of methicillin, whose resistance was acquired barely two years after its introduction into clinical use8. Antimicrobial resistance (AMR) driven by ARGs is a significant global public health issue that has a global spread, and comes with huge financial and health burdens4,5.

As a region, Africa bears the highest and most significant brunt of the AMR burden compared to other regions of the world4,5. A cross-country study by Sartorius et al. on the burden of bacterial antimicrobial resistance in the World Health Organisation (WHO) African region in 2019 stressed the importance of addressing the burden of AMR in Africa5. First, addressing the resistance menace is only feasible if there is a comprehensive understanding of the various resistance genes elaborated by S. aureus. Second, there is also a need to understand the mechanisms utilised by S. aureus to resist antibiotics used in clinical practice. Sadly, most of the studies on ARGs that have emanated from Africa are usually focused on a particular gene or a few genes using a combination of culture and molecular-based methods9. Among the molecular-based studies, the method of choice remains the use of polymerase chain reaction, where they utilise one or a few primers to detect resistance gene(s) of interest9. Thus, these studies have been unable to completely capture the entire set of resistance genes that may be present in S. aureus8,9. Other non-molecular-based studies have reported the prevalence of resistance to various antibiotics, which again does not reveal all the ARGs in S. aureus10,11. This is further compounded by the gross absence of AMR surveillance systems in some African countries11. This trend indicates the need to utilise bioinformatics as an approach to infer the diversity of ARGs.

By applying bioinformatics to whole genome sequences of S. aureus from diverse hosts in a study conducted in the United States of America (USA) between 2017 and 2019, it was shown that each genome carried at least eight resistance genes12. Some of these genes included the norA (fluoroquinolone), arlRS (fluoroquinolone), lmrS (multidrug), tet(38) (tetracycline), and mepAR (multidrug and tigecycline resistance)12. In the same study, using whole genomes, it was further revealed that the bacterium and the host share resistance genes12. Such a diverse number of genes cannot be revealed by molecular-based studies alone as previously reported8,9,12. The ARGs and their spread among pathogens like S. aureus pose a significant global health risk, especially so in Africa. Even though there are reports of S. aureus from diverse samples, the diversity of AMR genes in S. aureus whole genomes isolated in Africa, as well as their geographical dissemination or evolution on the continent, are also unknown. A comprehensive understanding of the diversity will aid the national and global efforts to reverse the spread of the AMR menace by revealing hotspots for surveillance efforts. Furthermore, it can also inform the best antibiotics that can be utilised for the management of infection caused by the bacterium or aid in the design of newer and safer antibiotics. To understand this diversity and evolution of AMR in S. aureus, we examined the spread and diversity of the ARGs in S. aureus whole genome data deposited from Africa in GenBank using bioinformatics and phylogeography.

Materials and methods

Retrieval and preparation of datasets

Deposited S. aureus whole genome sequences (see supplementary dataset (Table 1a) for accession numbers) for African countries were retrieved from the NCBI (https://www.ncbi.nlm.nih.gov/). The deposited genome sequences were downloaded in FASTA format country by county in Africa. For each of the sequences retrieved, metadata (country of origin, source of sample and date of isolation) were also collected. The obtained sequences in FASTA format were then labelled using their accession numbers, species from which the sequences were gotten from, first three letters of each country, and date of isolation. From the retrieved sequences, two sets of datasets were obtained. The first dataset had whole genomes (sequences with at least 2.6 million bases) for each country (n = 1) that was used to perform the BEAST analysis. The second dataset comprised all the whole genome sequences (n = 95) for all the African countries (n = 11). However, for Tanzania, only one whole genome sequence was used, while for Kenya, 21 sequences were utilised in performing the AMR gene analysis.

AMR gene analysis

Using the second dataset (n = 95 whole genomes), the AMR profile in the various sequences for each of the countries was carried out using the CARD tool13. In the tool, the uploaded nucleotide sequences were first converted into open reading frames (ORF), from which protein sequences linked to the AMR genes were predicted. The analysis revealed the various AMR gene families, ARGs, their mechanisms of action, and antibiotics associated with the genes. The CARD run was validated using a second repeat run for all the sequences (n = 95).

BEAST analysis

The BEAST analysis tool used to infer the evolution or spread of the AMR genes had its speed and accuracy enhanced using BEAGLE14,15. Due to the size of the sequences, we utilised the first set of datasets (n = 11) for the BEAST run. The sequences were arranged alphabetically using Notepad, and each sequence label had a three-letter abbreviation of country of origin (first three letters), source of the isolate, and date of deposit in the Genbank. For each sequence, the coordinates of the capital city for each country were retrieved online and used in the preparation of the trait file (See the trait data in Supplementary Table 1). The sequences were first aligned using the MAFT alignment tool of NG phylogengy16. The aligned sequence file and the traits were uploaded using the command-line version of Beauti tool17. In Beauti, the continuous file was generated. In generating the continuous (.xml) file, the tips dates button was used, the sampling dates were specified, and this was followed by the setting of the clock models18,19. In setting the clock model, the HKY substitution model was selected, and other parameters were kept at default20,21,22. The uncorrelated relaxed clock model was then selected, and the relaxed distribution was kept as lognormal. Next, tree priors were maintained at a coalescent constant size, and a random starting tree was selected. Using the partition button, the sites were then relaxed. The operators were then set at 100,000,000 and 50,000 chain length and echo to screen intervals, respectively23. The generated.mxl files were then used to perform BEAST runs using the command-line version. Following a successful BEAST run, the generated files (log and tree) were utilised for further analysis. The resulting files were analysed using Tracer v1.7.224, TreeAnnotator v1.10.425 and SPREAD v426.

Statistical data analysis

The extracted abundance of the various AMR genes, classes of antibiotics, gene families, and mechanisms of action were plotted using heat maps in R. Furthermore, the extracted metadata were transformed into barcharts using Microsoft Excel (USA) package.

Results

A total of 95 S. aureus complete genome sequences were retrieved, and they were distributed as shown in Fig. 1 (left bar chart). The sources of the S. aureus isolates revealed diverse sources (right bar chart). Overall, Nigeria (n = 20) and Kenya (n = 21) had the highest number of sequences retrieved, while Tanzania had the least number of sequences (n = 1). Samples from humans dominated the sample types from which S. aureus sequences were obtained. The sample types observed for humans included blood, pus, wounds (including surgery wounds), vagina, endocervix, faeces, urine, kidney, and cancer samples. Other sample types from which S. aureus were obtained were bat, camel, fly, water, cattle, and pig in decreasing order of abundance.

Fig. 1
Fig. 1
Full size image

Metadata of the retrieved sequences. Key: ALG = Algeria, BUR = Burkina Faso, CAM = Cameroon, COM = Comoros, EGY = Egypt, GHA = Ghana, KEN = Kenya, NIG = Nigeria, SOU = South Africa, SUD = Sudan, and TAN = Tanzania.

Figure 2 shows the spatial spread of the S. aureus containing AMR genes in their whole genomes deposited across the continent. Countries where the missiles (arrows in red) did not hit indicate the absence of deposited whole genome sequences and loosely the absence of AMR genes data. The result further indicates that the West and East African regions were the busiest when it comes to S. aureus genomes with AMR genes. Figure 3 and supplementary Table 2 show the various annotated resistance mechanisms in the S. aureus genomes in this study. Overall, six (6) resistance mechanisms were observed, and these were antibiotics efflux pump (AE), antibiotics inactivation (AI), antibiotics target protection (ATP), antibiotics target replacement (ATR), and antibiotics target protection/target replacement (A-A). The result further indicates that the antibiotics efflux pump (AE) was the dominant mechanism in all the sequences (see the colour legend). Nigeria, Kenya, Ghana, Burkina Faso, and Algeria returned the highest frequencies of AE with values of 214, 179, 160, 107 and 97, respectively. Overall, Nigeria had the highest amount of all the annotated mechanisms apart from ATA/ATR and ATP that were not detected from the deposited sequences.

Fig. 2
Fig. 2
Full size image

Phylogeography spread of the S. aureus whole genomes in Africa revealed by SPREAD v4. Key: ALG = Algeria, BUR = Burkina Faso, CAM = Cameroon, COM = Comoros, EGY = Egypt, GHA = Ghana, KEN = Kenya, NIG = Nigeria, SOU = South Africa, SUD = Sudan, TAN = Tanzania.

Fig. 3
Fig. 3
Full size image

Annotated antibiotics resistance mechanisms in the various S. aureus for the various countries visualised using a heatmap. Key: ALG = Algeria, BUR = Burkina Faso, CAM = Cameroon, COM = Comoros, EGY = Egypt, GHA = Ghana, KEN = Kenya, NIG = Nigeria, SOU = South Africa, SUD = Sudan, TAN = Tanzania. The red colour in the colour gradient denote highest abundance, while blue denotes the least.

A total of 22 AMR families were annotated from the S. aureus sequences from the various African countries (Fig. 4 and supplementary Table 3). As indicated in the result, some of the AMR gene families were annotated more than once. The AMR families were antibiotic-resistant GlpT, antibiotic-resistant murA transferase, ABC-MFS, fosfomycin thiol transferase, glycopeptide resistance gene cluster; vanT, kdpDE, MFS antibiotic efflux pump, methicillin resistant PBP2, multidrug and MATE transporter, small multi-drug resistance (SMR) antibiotic efflux pump, target protecting FusB-type protein conferring resistance to fusidic acid, fluoroquinolone resistant parC, rifamycin-resistant beta-subunit of RNA polymerase (rpoB), trimethoprim resistant dihydrofolate reductase dfr, BlaZ beta-lactamase, fluoroquinolone resistant gyrA, APH(3'), Erm 23S ribosomal RNA methyltransferase, streptothricin acetyltransferase (SAT), tetracycline-resistant ribosomal protection protein, aminoglycoside bifunctional resistance protein, and ANT(6)/ANT(9).

Fig. 4
Fig. 4
Full size image

Annotated ARM families in the various S. aureus whole genome sequences. Key: ALG = Algeria, BUR = Burkina Faso, CAM = Cameroon, COM = Comoros, EGY = Egypt, GHA = Ghana, KEN = Kenya, NIG = Nigeria, SOU = South Africa, SUD = Sudan, and TAN = Tanzania. The red colour in the colour gradient denote highest abundance, while blue denotes the least.

Among the ARG families, the MFS antibiotic efflux pump was found in all the countries, with Nigeria having the highest abundance of 134, followed by Ghana with 96 (see supplementary Table 3 and red colour in Fig. 4). Some of the AMT families were unique to only one country, and these included the ANT(6)/ANT(9), tetracycline-resistant ribosomal protection protein, and target-protecting FusB-type protein conferring resistance to fusidic acid that were unique to South Africa, Kenya, and Algeria, respectively. The rest of the AMR gene families were detected in at least 2 countries.

From the various S. aureus whole genome sequences obtained for this study, 33 AROs were detected (Fig. 5 and supplementary Table 4). The most prevalent ARO was norC, which was observed in all the sequences, with the highest number observed in Nigeria (n = 40) and Ghana (n = 31). This was followed by arlR and mgrA that were annotated in all the sequences except for Cameroon.

Fig. 5
Fig. 5
Full size image

Annotated ARO in the various S. aureus whole genome sequences. Key: ALG = Algeria, BUR = Burkina Faso, CAM = Cameroon, COM = Comoros, EGY = Egypt, GHA = Ghana, KEN = Kenya, NIG = Nigeria, SOU = South Africa, SUD = Sudan, and TAN = Tanzania. The red colour in the colour gradient denote highest abundance, while blue denotes the least. BP = Bifunctional Protein; WMCRF = With Mutation Conferring Resistance To Fosfomycin; CRF = Conferring Resistance to Fluoroquinolones; MCRR = Mutants Conferring Resistance to Rifampicin.

Twenty (28) classes of antibiotics were observed in the sequences to which the various ARO and resistance mechanisms were annotated to as shown in Fig. 6 and supplementary Table 5. In addition to antibiotics in clinical use, other agents to which S. aureus displayed resistance to were also observed and these were acriflavine, disinfectants and antiseptics. The most prevalent antibiotics to which S. aureus showed resistance to was the fluoroquinolones antibiotics followed by the disinfectants and antiseptics classes.

Fig. 6
Fig. 6
Full size image

Annotated antibiotic classes in the various S. aureus whole genome sequences. Key: ALG = Algeria, BUR = Burkina Faso, CAM = Cameroon, COM = Comoros, EGY = Egypt, GHA = Ghana, KEN = Kenya, NIG = Nigeria, SOU = South Africa, SUD = Sudan, and TAN = Tanzania. The red colour in the colour gradient denote highest abundance, while blue denotes the least. CAN = Ciprofloxacin; Acriflavine; Norfloxacin; ECNO = Enoxacin; Ciprofloxacin; Norfloxacin; Ofloxacin; ESKLTTCF = Erythromycin; Streptomycin; Kanamycin A; Linezolid; Trimethoprim; Chloramphenicol; Florfenicol; FCPTPDA = Fluoroquinolone Antibiotic; Cephalosporin; Penam; Tetracycline Antibiotic; Peptide Antibiotic; Disinfecting Agents And Antiseptics; FDAA = Fluoroquinolone Antibiotic; Disinfecting Agents And Antiseptics; GTA = glycylcycline; tetracycline antibiotic; MAODA = macrolide antibiotic; aminoglycoside antibiotic; oxazolidinone antibiotic; diaminopyrimidine antibiotic; phenicol antibiotic; MLSSS = Macrolide Antibiotic; Lincosamide Antibiotic; Streptogramin Antibiotic; Streptogramin A Antibiotic; Streptogramin B Antibiotic and MCATDMCNSM = Methicillin; Ciprofloxacin; Acriflavine; Tetracycline; Daptomycin; Moxifloxacin; Cefotaxime; Norfloxacin; Sparfloxacin; Moenomycin.

Vancomycin was also only detected in the sequences from three countries namely the Comoros, Ghana, and Tanzania, while trimethoprim was only detected in Ghana. On the other hand, tigecycline was detected in only sequences in Ghana and Tanzania. Tetracycline antibiotic was only detected in Nigeria and Kenya. Rifamycin antibiotic was detected only in Cameroon and Kenya. On the other hand, phosphonic acid antibiotics had more coverage as they were detected in Algeria, Cameroon, Comoros, Ghana, Nigeria, and South Africa.

Discussion

Staphylococcus aureus is an important pathogen of public health concern for which newer and safer antibiotics are urgently needed6,7. However, our understanding of the diverse antimicrobial genes the pathogen could harbour is limited; hence, this study. The generated metadata revealed diverse isolation sources for the pathogen that included humans, bats, camels, cattle, flies, pigs, and water. This aligns with previous reports that assert that S. aureus is a normal flora of mammals5,12,27. Our metadata further revealed that the niche occupied by this important bacterium is increasing, as the pathogen was isolated from diverse samples, including faeces, blood, nasal swabs, wounds, and water. This indicates that the pathogen has evolved the ability to move from humans (its primary reservoir) to other mammals and the environment. The evolution of S. aureus, especially the antibiotic-resistant variants, is probably driven by urbanisation, agricultural practices, and global use/abuse of antibiotics28,29. The presence of AMR genes and resistant pathogens in the environment is well known30,31. Their presence in the environment could potentially be the route for the introduction of strains that have undergone genetically driven changes28 and the spread of resistance genes32. In addition to the diverse sources of isolation, the whole genome sequences were available for 11 countries in studies all aimed at isolating multi-drug resistant S. aureus isolates, further highlighting the ubiquitous nature of the pathogen. The absence of whole genome sequences in more than two-third of the African countries aligns with a report which indicates a gross absence of infrastructures and/or AMR surveillance studies11. Given the importance of the AMR genes, an understanding of their diversity in S. aureus is important in the search for new antibiotics and can also inform future studies.

To provide an insight into the diversity and spread of AMR genes harboured by S. aureus whole genomes in Africa, we conducted a phylogeography and bioinformatics analysis of AMR genes using the CARD and BEAST tools. Our phylogeography analysis revealed the spread of S. aureus harbouring AMR genes in their whole genome sequences in 11 countries, with the West and East Africa regions being the busiest regions. Phylogeography allows for the visualisation of genotypes using various scales, which could be temporal or spatial. It has been used in various studies to understand the spread of pathogens such as H5N133,34,35. Our findings indicate that the West and East Africa regions were the busiest on the continent as far as the spread of AMR genes in S. aureus from whole genome sequences was concerned. An earlier study aimed at understanding the position played by Africa in the spread of H5N1 showed West Africa as the region with the highest burden of the virus, acting as a hotspot for the spread of the virus. Coincidentally, the same region has the highest burden of AMR5. This correlates with our findings, as the same region was the busiest. The widespread ofAMR genes in the region has been attributed to many factors, including the exit of pharmaceutical giants and the absence of AMR surveillance in the region36,37. In addition to the spread of S. aureus harbouring AMR genes, there is a need to understand the diversity of these genes in the whole genomes of S. aureus.

The diversity of the AMR genes in our study was examined in terms of their mechanisms of action, the various AMR families, the annotated ARO, and the various antibiotics involved. The top ARO or ARG genes annotated were norC, arlR, mrrA, sepA, S. aureus LmrS, kdpD, the vanT gene in the vanG cluster, S. aureus norA, and S. aureus FosB. Our findings of AROs in the various whole genomes align with an earlier report in the USA that showed the presence of at least eight resistance genes in the S. aureus genomes in their study, which were also annotated in our study. This further confirms the universality of the ARGs12. The NorC belongs to the MFS antibiotic efflux pump family and has been observed to elicit resistance against antiseptics, disinfectants, fluoroquinolone, and lesser resistance to macrolides and tetracyclines38,39. In addition, the arlR and arlS genes were also detected, and these also belong to the MFS antibiotic efflux pump family, which has activities against fluoroquinolone antibiotics, disinfectants, and antiseptics40,41. ArlR is reportedly a component of the regulatory system of S. aureus and controls the expression of various virulence factors and resistance mechanisms42. On the other hand, ArlS in S. aureus encodes a sensor kinase that indirectly influences the expression of resistance-related genes and virulence factors thus impacting synergistically on the bacterium’s resistance profile and pathogenicity43. Furthermore, ArlS works in tandem with the arlR gene (the response regulator) to regulate various aspects of the bacterium’s physiology, including the elaboration of proteins involved in virulence and resistance44,45. In addition, the kdpD was also annotated. kdpD is a member of the kdpDE family of AMR genes that encodes resistance against the aminoglycosides via effluxing46. In S. aureus, kdpD reportedly encodes a sensor kinase that plays a role in the regulation of various physiological processes which includes antibiotic resistance47.

Also detected were the sepA and FosB genes. SepA is a member of the ARG family that acts using the mechanism of the small multi-drug resistance (SMR) antibiotic efflux pump and has been reported to elicit resistance to antiseptics and disinfectants48,49. The protein reportedly encodes staphylococcal epidermal cell adhesion protein (SepA), a key virulence factor that enhances the bacterium’s ability to attach to the host’s tissues and circumvent immune response50, and in the formation of biofilms51. On the other hand, FosB is a member of the AMR gene family fosfomycin thiol transferase that has been observed to promote resistance against phosphonic acid antibiotics such as fosfomycin52,53. It encodes for the fosB enzyme, which inactivates fosfomycin through enzymatic modification by catalysing methyl group addition to fosfomycin, leading to the formation of a modified compound, eventually rendering the antibiotic ineffective54. Fosfomycin is often used as a last-line treatment for certain infections, and resistance to it as displayed by S. aureus further limits this treatment option55. Furthermore, the mrrA, S. aureus LmrS, vanT gene in the vanG cluster, and S. aureus norA genes were also among the top ten ARO detected. LmrS is a member of the ARG family, the MFS antibiotic efflux pump that facilitates resistance to phenicol, diaminopyrimidine, oxazolidinone, aminoglycosides, lincosamides, streptogramins, and macrolide antibiotic classes56. The lmrS gene encodes an efflux pump that actively transports antibiotics out of the bacterium, lowering their cellular concentrations inside the cell and diminishing their efficacy57. The vanT gene in the vanG cluster represents a group of genes that are involved in peptidoglycan synthesis and presents with reduced compatibility to vancomycin, that is, alters the target58. ArlRS and MgrA together regulate virulence in methicillin-resistant S. aureus involved in skin infection via the control of virulence factor expression59. MgrA belongs to the MarR/SarA family proteins60. The S. aureus norA protein is an efflux pump protein that belongs to the MFS family and drives resistance to several antibiotics, such as the fluoroquinolones as well as the antiseptics61. In addition to the top 10 abundant AMR genes, other genes were also detected. The norA gene works via the efflux pump mechanism, and it drives resistance to fluoroquinolone antibiotics such as ciprofloxacin62. On the other hand, S. aureus rpoB mutants conferring resistance to rifampicin arise from a mutation to the rpoB gene. This implies that in the event of a co-infection with the S. aureus strain carrying the rpoB mutation with tuberculosis, it could worsen the clinical outcome, as rifamycin is a first-line drug for the management of tuberculosis. Resistance to fosfomycin was further observed to be via mutations in GlpT and murA genes. GlpT is an active transporter that drives the transportation of phosphorylated sugars into the cell63. Mutations in the GlpT and murA proteins have been linked to fosfomycin resistance among methicillin-resistant S. aureus isolates63. The MecA gene, despite its low abundance compared to other genes, was also annotated. MecA is located in the staphylococcal cassette chromosome mec (SCCmec), where it directs the synthesis of PBP2a that reduces the affinity of the β-lactam antibiotics, including the cephalosporins64,65,66. The aforementioned ARGs are linked with diverse resistance mechanisms.

The S. aureus whole genomes elaborated diverse mechanisms with which they resist diverse antibiotics used in the management of infections they cause. The mechanisms included antibiotic efflux pumps (AE), antibiotic inactivation (AI), antibiotic target protection (ATP), antibiotic target replacement (ATR), and combined antibiotic target protection/target replacement (A-A). The annotation of these mechanisms in all the genomes of S. aureus suggests their utilisation of multiple mechanisms in resisting antibiotics in clinical practice, and this aligns with previous reports of diverse mechanisms57,67,68. The top three mechanisms were AE, AI and ATR, and these align with previous reports that implicated them as the most common mechanisms57,67,68. Efflux pumps, as their name implies, are pumps or protiens used by bacteria to transport antibiotics from their intracellular milieu and, in the process, reduce their concentration in the cell and the effectiveness of the antibiotics69. These pumps are encoded by genes such as norA, mepA, and tetK, which contribute to multidrug resistance (MDR) in S. aureus and were also annotated in our study70. In Nigeria, AE mechanisms are highly prevalent. A meta-analysis revealed that S. aureus resistance to antibiotics varies between 13 and 82%, with efflux pumps playing a dominant role71. On the other hand, AI implies the utilisation of enzymes to inactivate drugs and eventually decrease their intracellular concentration and overall effectiveness. Examples of AMR genes that work via AI include the β-lactamases, which hydrolyse β-lactam antibiotics, and aminoglycoside-modifying enzymes that alter aminoglycoside antibiotics72. As a mechanism, ATP acts by protecting antibiotic targets and, in the process, prevents the binding of the antibiotics without modifying the target structure. This is observed in S. aureus strains carrying the tetM gene, which confers resistance to tetracyclines73. ATR involves the replacement of antibiotic target sites, as seen in the highly prevalent methicillin-resistant S. aureus (MRSA) strains, which possess the mecA gene74. This gene encodes an alternative penicillin-binding protein (PBP2a) with reduced affinity for β-lactam antibiotics74. The A-A mechanism, which combines ATP and ATR, allows bacteria to simultaneously protect and replace antibiotic targets, enhancing resistance57,67,68.

Conclusion

Despite the limitations of the study, our findings provide a number of valuable insights into the diversity of ARGs in S. aureus whole genomes in Africa. The metadata of the deposited sequences implicates clinical and environmental samples as the major sources of the S. aureus. Our study indicates the presence of diverse AMR genes and families previously not captured by PCR-based studies. Our findings further validate the universality of resistance genes, especially those linked the efflux (MFS, MATE, SMR, and ABC-MFS) pumps. The most annotated antibiotics were the fluoroquinolones, followed by the antiseptics and disinfectants. Other antibiotics annotated were phosphonic acid antibiotic, vancomycin, trimethoprim, tigecycline, Our evolution analysis indicated that the West and East African regions are hubs for the spread of the AMR genes. Given the ubiquity of this pathogen, similar studies for other regions of the world need to be conducted to have a holistic picture of the situation. This, if done, will provide a complete view of the diversity of AMR genes and inform further studies aimed at discovering newer and safer antibiotics for multi-drug-resistant S. aureus.

Limitations

The study had some limitations that we hereby acknowledge. First, there was sampling bias in the collection of the sequences, as only 11 countries deposited whole genome sequences of S. aureus in GenBank. Furthermore, in running the CARD analysis, all the sequences from the 11 countries were included in the analysis except for Kenya, which had the highest number of sequences, and we only had to utilise 21 of them, while Tanzania had one. We conducted the AMR analysis solely using the CARD tool. In running the BEAST analysis, only one sequence per country was utilised due to the computational constraint the size of each whole genome imposes, while for countries such as South Sudan and Sudan, only one sequence was utilised.