Determination of Enterococcus faecalis and Enterococcus faecium Antimicrobial Resistance and Virulence Factors and Their Association with Clinical and Demographic Factors in Kenya

Background Enterococci are clinically significant because of their increasing antibiotic resistance and their ability to cause severe infections due to an arsenal of virulence genes. Few studies in the developing world have examined virulence factors that may significantly impact patient outcomes. This study describes the antimicrobial resistance profiles and prevalence of five key Enterococcal virulence genes gelE, asa, cylA, esp, and hyl in forty-four clinical Enterococcus faecalis and E. faecium isolates in Kenya and their association with patients' demographic and clinical characteristics. Results All E. faecium isolates were obtained from hospital-acquired skin and soft tissue infections. While E. faecalis was associated with community-acquired urinary tract infections. All isolates were resistant to erythromycin, whereas 11/44 (27.5%), 25/44 (56.8%), 28/44 (63.6%), 37/44 (84.1%), 40/44 (90.0%), and 43/44 (97.5%) were susceptible to tetracycline, levofloxacin, gentamicin, ampicillin, nitrofurantoin, and teicoplanin, respectively. All isolates were susceptible to tigecycline, vancomycin, and linezolid. There was little difference in the antibiotic resistance profiles between E. faecalis and E. faecium. The prevalence of the virulence genes among the 44 isolates were 27 (61.4%) for gelE, 26 (59.1%) for asa1, 16 (36.3%) for esp, 11 (25.0%) for cylA, and 1 (2.3%) for hyl. 72.9% of E. faecalis isolates had multiple virulence genes compared to 57% of E. faecium isolates with no virulence genes. The hyl gene was only detected in E. faecium, while cylA and asa1 were only detected in E. faecalis. A significant correlation was observed between the presence of asa1 and esp virulence genes and tetracycline resistance (P=0.0305 and 0.0363, respectively). A significant correlation was also observed between the presence of virulence genes gelE and asa1 and nitrofurantoin resistance (P=0.0175 and 0.0225, respectively) and ampicillin resistance (P=0.0005 and 0.0008, respectively). Conclusion The study highlights the high levels of erythromycin resistance in E. faecalis and E. faecium, the demographic factors influencing the species distribution among patients, and the accumulation of multiple virulence genes in E. faecalis. The significant association of gelE, asa1, and esp virulence genes with drug resistance could explain the pathogenic success of E. faecalis and provides a guide for future studies.


Background
Enterococcal species bacteria are Gram-positive cocci typically found in the gut, bowel, throat, mouth, and vagina as commensals [1][2][3][4]. E. faecalis and E. faecium are the main pathogenic species of the eighteen-known species of enterococci. However, only the E. faecalis and E. faecium strains harboring virulence genes are associated with human infections [5], including urinary tract (UTI), pelvic, blood, intraabdominal, and skin and soft tissue infections (SSTI) [6]. Te main virulence factors that have been described in Enterococci are aggregation substance (asa1), gelatinase (gelE), cytolysin (cylA), enterococcal surface protein (esp), and hyaluronidase (hyl). asa1 is an aggregation substance that enhances adherence to renal tubular cells [1]. Gelatinase is a zinc metalloprotease that hydrolyzes fbrin, collagen, and other peptides [7] and has been linked to bioflm formation [8]. Te secretion of hemolysins such as cytolysin causes the breakdown of blood cells and has been linked with increased toxicity in human infection [9]. Hyaluronidase facilitates the colonization of host tissue by breaking down hyaluronic acid, a critical component of connective tissue [10]. Enterococcal surface protein (esp) is a high molecular weight surface protein associated with bioflm formation that is specifc to enterococci [11]. esp also plays a role in colonization and persistence in the urinary tract [12]. Te enterococcal virulence gene gelE is the most common in E. faecalis isolates, while hyl and esp are more common in E. faecium than in E. faecalis [13]. Most of what is known about Enterococci species virulence is based on studies outside Africa [14,15].
Tere is limited published data on the prevalence and distribution of virulence genes among clinical enterococci in Africa. However, studies in Ethiopia have demonstrated that Enterococcal isolates from animals carry the gelE virulence genes [16], contribute to hospital-associated infections [17] and profled their antimicrobial susceptibility [18]. Enterococci species are not major clinical pathogens in Africa as reports in Nigeria and Kenya indicate a point prevalence of 5-11% [19]. Despite this, resistance to glycopeptides, aminoglycosides, β-lactams, and fuoroquinolone antibiotic classes is on the rise globally [10,20] and could pose challenge to the treatment of these infections given that the remaining treatment options, such as linezolid, daptomycin quinupristin/dalfopristin, and vancomycin, are expensive with limited availability in primary care centers in Kenya. Although rare, there have been reports of vancomycin resistance in Kenya [19], indicating the possibility of growing resistance to these last-line drugs. Te interplay of virulence genes and antimicrobial resistance in clinical infections is worth examining as hypervirulent and multidrug-resistant Enterococci isolates leading to infections with adverse clinical outcomes could emerge. Zou et al. examined the correlation between erythromycin resistance and virulence genes and found a positive association between the presence of gelE and resistance to erythromycin [21]. Tis study was therefore conducted to examine clinical isolates of E. faecalis and E. faecium in Kenya to understand the prevalence and distribution of the virulence genes and determine if there is an association between antimicrobial resistance, clinical presentation, demographic factors, and virulence genes.

Study Design and Population.
Tis cross-sectional study is nested in a multi-hospital surveillance study where patients over two months old with UTIs or SSTIs were recruited between May 2015 and December 2019 from six hospitals in fve Kenyan counties. Urine samples, pus, and swabs from soft tissue infections were collected from study subjects after they consented to participate in the study. In addition, demographic and clinical data were collected for each patient, including gender, age, infection type, in-or outpatient status, infection acquisition in the community or hospital, and immune status.

Bacterial Isolation and Identifcation.
Samples were shipped to the Kenya Medical Research Institute (KEMRI) laboratories at 2-8°C for pus and wound swabs and at room temperature for urine samples in boric acid. All samples were inoculated on MacConkey (BD, New Jersey, United States of America), cysteine lactose electrolyte defcient agar (HIMEDIA, Mumbai, India), and sheep blood agar (SBA) and incubated for 24 to 48 hrs at 37°C. Colony morphology and culture characteristics were observed macroscopically. Preliminary identifcation of Enterococci was made based on observation of Gram-positive cocci in chains on Gram stain and a negative catalase test. Confrmation of the identifcation and Enterococcus speciation were performed on the VITEK 2 automated platform (bioMérieux, Marcyl'Étoile, France).  [13]. Each 25 μl of the PCR mixture consisted of 12 μl of Dream Taq DNA polymerase (Termofsher, Massachusetts, USA). 2.5 μl of DNA, 0.1 μl of cylA, asa1, gelE, and hyl. 0.2 μl of esp specifc primers [13]. Te positive and negative controls were E. faecalis ATCC 29212 and E. coli ATCC 25922, respectively. Te PCR conditions were: initial activation at 95°C for 15 min, 30 cycles of denaturation at 94°C for 1 min, annealing at 56°C for 1 min, extension at 72°C for 1 min, and fnally, one extension cycle at 72°C for 7 mins. Te PCR products were observed by running 15 μl of the PCR reaction on a 1% agarose gel alongside a 100 bp ladder (Termofsher, Massachusetts, United States) and visualizing on an E-box gel documentation station (Vilber, Marne-la-Vallée, France).

Data Analysis
Demographic and clinical data were extracted from study questionnaires and displayed in an Excel spreadsheet. Quantitative data were analyzed in Excel. Te association between virulence genes, demographic factors, and antimicrobial susceptibility phenotypes was assessed using the Fisher's exact test with a P ≤ 0.05 considered signifcant.

Demographic and Clinical Characteristics of Patients with
Enterococcal Infections. Enterococcus isolates were obtained from six hospitals within fve counties: Nairobi (16), Kisumu (18), Kericho (2), Kisii (7), and Kilif (1). Tirty-seven E. faecalis isolates and 7 E. faecium isolates were isolated from forty-four patients with a mean age of 37 years during the study period which represented only 1.89% of the total isolates obtained during the same period. Te majority of isolates 29/44 (65.1%) were from the skin and soft tissue infections (SSTIs) while 15/44 (34.9%) were isolated from urinary tract infections (UTIs). Ten out of forty-four patients (23%) were immunocompromised individuals, 28/44 (63.7%) were hospitalized, and 10/44 (23.0%) had hospitalassociated infections. E. faecalis was isolated more from females 21 (56.7%) than males. In contrast, 6 of the 7 E. faecium isolates were from males. All seven E. faecium isolates were from hospitalized patients and 6/7 from patients with SSTIs. Among the 44 patients, most had community-acquired enterococci infections, but E. faecium infections accounted for most of the hospital-associated infections. Only one UTI was hospital-associated compared with six SSTIs (Table 1).
Age was signifcantly associated with Enterococci infections (P � 0.0270). Te age group between 14 and 29 years contributed the largest proportion of infections, followed by the age group between 31 and 49 years and above 50 years. Tere was no association between gender and the species of Enterococcus or between immunocompromised status or type of infection with enterococcal species. Tere was a signifcant association between species type, the source of infections (P � 0.0367), and the patient status (inpatient/ out-patient). Community-acquired Enterococci infections were more likely to be caused by E. faecalis than E. faecium whereas E. faecalis were more likely to be HAI with all E. faecalis infections identifed in hospitalized patients (Table 1).

Detection of Virulence
Genes. All the fve virulence genes screened were detected among the E. faecalis and E. faecium isolates based on the presence of expected band sizes ( Figure 2). Data indicating the presence or absence of the virulence genes for all the isolates is shown in Supplementary  Table 1.
gelE was the most frequently detected gene at 27 (61.4%) followed by asa1 26 (59.1%), esp 16 (36.3%), cylA 11 (25.0%) and hyl 1 (2.3%) ( Tere was a signifcant association between tetracycline resistance and the presence of asa1 and esp (P � 0.0305 and 0.0363, respectively), nitrofurantoin resistance and the presence of gel E and asa1 genes (P � 0.0175 and 0.0225, respectively) and ampicillin resistance and the presence of gel E and asa1 (P � 0.0005 and 0.0008, respectively) (Table 3). Te associations between the other antibiotics were not tested because of complete or almost complete resistance or susceptibility to erythromycin, teicoplanin, nitrofurantoin, linezolid, vancomycin, and tigecycline.

Discussion
Tis study was conducted to evaluate the prevalence and distribution of known Enterococcus spp. virulence genes among clinical E. faecalis and E. faecium isolates obtained in a 5-year antimicrobial resistance surveillance study. Te study found that Enterococcus spp. are uncommon clinical pathogens in the sampled Kenyan population given that the E. faecalis and E. faecium isolates represented <2% of the isolates recovered from clinical samples in the parent study. Tis is consistent with previous research in Africa that showed a 2.7% prevalence of Enterococcus spp. infections among pediatrics [22] much lower compared to a 13.6% prevalence reported in East Asia. Te isolates in this study were obtained from SSTIs and UTIs consistent with the known infections caused by Enterococci spp. [23]. However, we observed that E. faecalis was the predominant pathogen of the two species with E. faecalis isolated fve times more than E. faecium.   [24], that identifed E. faecalis as the main uropathogen in community-acquired UTIs (CA-UTIs). Tis abundance of E. faecalis in CA-UTI could be attributed to the predominance of E. faecalis in the patient's commensal fora [25,26]. E. faecium though less frequently isolated, was found predominantly in male hospitalized patients with SSTIs, indicating its signifcance as a hospital-associated pathogen compared to other Enterococcus species. Despite having fewer of the virulence genes tested than E. faecalis, E. faecium infections were associated with inpatient infections which implies infections of greater severity. Tis apparent disparity was also reported by Higuita et al. [27], who observed that E. faecium caused more severe infections and had a higher mortality rate than E. faecalis. Based on our observations we hypothesize that hospitalassociated E. faecium infections could be opportunistic, afecting already vulnerable patients and leading to more adverse outcomes, and could have little to do with the presence or absence of virulence factors.
When we examined the antibiotic resistance patterns we found no signifcant diferences between the two species in contrast to studies that found that E. faecium isolates are more efcient in accumulating resistance genes [6], resulting in greater resistance to penicillin, ampicillin, piperacillin, imipenem, and ciprofoxacin than E. faecalis isolates Te lack of diference in resistance between species could refect the rarity of human enterococcal infections and the high number of community-acquired infections which do not experience the antibiotic pressure that would drive antibiotic resistance to drugs typically used in hospital settings. Te AST results indicated many available treatment options for the clinical management of Enterococcal infections since all the isolates tested were susceptible to vancomycin, tigecycline, and linezolid. Te samples also had high susceptibility to teicoplanin, nitrofurantoin, ampicillin, gentamycin, and levofoxacin. Variable E. faecalis antibiotic resistance results for rifampin (60.7%), tetracycline (17.9%), erythromycin (14.3%), and chloramphenicol (10.7%) have been documented in Kenya [27]. A high incidence of antibiotic  Journal of Pathogens 5 resistance to ampicillin (80%) and doxycycline (73.3%) have also been reported in Ethiopia [18]. Glycopeptide resistance which is mediated by the Van gene clusters has rarely been reported in studies from Kenya [28,29] and we did not observe any resistance in this study. Enterococci expressing the vanA gene are highly resistant to vancomycin and teicoplanin antibiotics, while enterococci expressing the vanB gene show high resistance to vancomycin and susceptibility to teicoplanin [30]. In this study, there was no resistance to vancomycin, but teicoplanin nonsusceptible isolates were observed in a few isolates and confrmed by repeat testing. Discordant resistance to the glycopeptide antibiotics is a rare but reported occurrence attributed in a study by Loong et al. [31] to novel point mutations and deletions in the regulatory regions for the Van genes located in the Tn1546 type transposon. Analysis of the presence of vancomycin resistance genes and the regulatory components could ofer clues on the mechanisms at play in the unusual discordance observed among the study isolates.
Te study observed high susceptibility (75%) to nitrofurantoin, the second most common drug to treat bacterial infections in the urinary tract in Kenya after beta-lactam drugs [28]. Tis high susceptibility in Kenya compared to other countries such as Iran with resistance levels of 35.5% [32] suggests that nitrofurantoin use is still not as common in Kenya as in other countries. Erythromycin is a relatively inexpensive broad-spectrum antibiotic used to treat many infections, so it was not surprising that resistance levels were high. Tis study shows that this drug is no longer efective for the treatment of Enterococcus spp. infections. A study in Nigeria also showed 100% erythromycin resistance, indicating that this trend is prevalent in more than one region of sub-Saharan Africa [33]. Tis study on clinical isolates found tetracycline resistance levels (72.5%) markedly higher than the 17.9% percent reported in the 2018 survey by Wambui et al. performed on slaughterhouse cattle [27] but consistent with studies done in a Kenyan hospital in 2009 that found tetracycline resistance rates of 80% for E. faecalis and 71% for E. faecium [34]. Tese high rates can be attributed to tetracycline being commonly used to treat infections because it is afordable and readily available. Te high tetracycline resistance rate has also been linked to the widespread use of tetracycline in livestock [27], evidenced by its presence in animal products [35,36] and among poultry in Europe [37] which could drive antibiotic resistance in bacteria including Enterococci.
Tis study reports gentamicin resistance of 36.4%, which suggests close to a two-fold increase in resistance in less than ten years based on a reported 19% resistance level in Kenya in 2012 [34]. Since aminoglycosides (e.g., gentamicin) monotherapy is known to have poor uptake into the cytoplasm, a combination of penicillin/gentamicin therapy is recommended for the treatment of patients with Enterococcal infections to improve the penetration of gentamicin through the bacterial cell-wall using the cell-wall active penicillins, resulting in synergistic activity [38]. A study done in Kenya in 2020 by Maina et al. [39] reported that combinations of penicillin and gentamicin were predominant in the neonatal unit (58%). Resistance to ampicillin was roughly comparable to resistance to gentamycin in this study, implying corresponding use.
After observing that the antibiotic resistance levels were low except for erythromycin and tetracycline we considered whether the pathogenicity of the organisms were a signifcant threat by examining the virulence gene profles. Te gene encoding gelatinase gelE was the most prevalent virulence gene (36.3%). Tis is consistent with the fndings of numerous studies conducted around the world, all of which found gelE to be more prevalent than the other genes measured in this study [40]. In terms of dominance, the agglutination substance gene asa1 came in second. Both genes were present only in E. faecalis. gelE and asa1 are the predominant virulence genes in similar studies performed among clinical isolates in India and Iran and also in pigs from a study in China [21,40,41]. Te hyl gene was found in only one E. faecium isolate, which was not a surprise given earlier studies that observed that the hyl gene was predominantly found in E. faecium isolates from clinical samples in the United States [6]. Although esp is known for its function in adhesion to the urinary tract wall [12] in this study, it was equally present in isolates from UTIs and SSTIs.
Te study also sought to test the association of virulence genes with antibiotic resistance. In addition to the association described by Zou et al. between gel E and erythromycin, this study has identifed other signifcant associations between the asa1 gene and esp and tetracycline and between nitrofurantoin, gelE, and asa1. Te asa1, gelE, and esp genes encode bioflm-forming proteins involved in adhesion to the host cells [8,42]. Te positive association between asa1, gelE, and esp genes with nitrofurantoin and tetracycline resistance is interesting as the two drugs are efective against bioflmforming isolates [43,44]. Bioflm-forming Enterococci bacteria are generally more resistant to antibiotics than nonbioflm-forming ones [45]. We hypothesize that exposure of bioflm-forming isolates to nitrofurantoin and tetracycline could drive specifc resistance to these agents. Te association of asa1, esp, and gelE in antibiotic-resistant bacteria is advantageous as it provides a double arsenal for causing and surviving the treatment of clinical infections.
Te study had a few limitations. First, there were only a few Enterococcus spp. isolates obtained in the study, which limited the statistical analysis and inferences that could be made. Second, some patients had coinfections with other pathogens, so the infections could not be solely attributed to Enterococcus spp. Tird, the study only screened for fve essential virulence genes, whereas more genes are involved in Enterococcal pathogenicity. Fourth, although most of the virulence genes tested are associated with bioflm formation, bioflm assays were not conducted to confrm the phenotype. A more extensive study combining phenotypic assays and whole-genome approaches would address virulence factors more comprehensively and provide the isolates' strain types to address any clonality issues that could skew the detected associations.

Conclusion
Te study has identifed E. faecium as a predominantly health-care-associated infection afecting male patients and E. faecalis as an important etiology of community-acquired urinary tract infections. Enterococcal infections can be well managed due to the low-level resistance observed for most antibiotics except tetracycline and erythromycin. All fve virulence genes tested were identifed among the Kenyan isolates with E. faecalis carrying more and multiple genes. Te importance of esp, asa1, and gelE virulence genes in virulence and their co-occurrence with antibiotic resistance could explain the clinical success of E. faecalis and provides an opportunity for further research.

Data Availability
Te graphs, fgures, and tables data used to support the fndings of this study are included within the article and also uploaded in the fgures fles section and supplemental fles section.

Ethical Approval
Tis study was approved by the Scientifc and Ethics Review Unit (SERU #2767) of the Kenya Medical Research Institute (KEMRI) and the Walter Reed Army Institute of Research (WRAIR #2089) IRB. Permission has been granted for the publication of this manuscript by the Director KEMRI.
Tere is no objection to its publication. Te investigators adhered to the policies for the protection of human subjects as prescribed in AR 70-25.

Consent
All subjects provided written consent to participate in the study and for their isolates and data to be used for research and publications.

Disclosure
Te opinions or assertions contained herein are the author's private views and are not to be construed as ofcial or as refecting the true views of the Department of Defense.

Conflicts of Interest
Te authors declare that they have no conficts of interest.