Virulence Factors in Klebsiella pneumoniae: A Literature Review

Abstract

Klebsiella pneumoniae, a member of the autochthonous human gut microbiota, utilizes a variety of virulence factors for survival and pathogenesis. Consequently, it is responsible for several human infections, including urinary tract infections, respiratory tract infections, liver abscess, meningitis, bloodstream infections, and medical device-associated infections. The main studied virulence factors in K. pneumoniae are capsule-associated, fimbriae, siderophores, Klebsiella ferric iron uptake, and the ability to metabolize allantoin. They are crucial for virulence and were associated with specific infections in the mice infection model. Notably, these factors are also prevalent in strains from the same infections in humans. However, the type and quantity of virulence factors may vary between strains, which defines the degree of pathogenicity. In this review, we summarize the main virulence factors investigated in K. pneumoniae from different human infections. We also cover the specific identification genes and their prevalence in K. pneumoniae, especially in hypervirulent strains.

Introduction

Klebsiella pneumoniae, a member of the Enterobacterales order and Enterobacteriaceae family [1], is a rod-shaped, gram-negative, and member of the autochthonous human gut microbiota also found in other environmental niches [2]. Although the literature addresses K. pneumoniae as an opportunistic bacterium [3–5], in this case, referring to the classical strains, there are hypervirulent strains capable of causing disease in previously healthy people [6–8].

The presence and type of virulence factor may vary between strains, and this characteristic defines the degree of pathogenicity [9]. Classical (or opportunistic) K. pneumoniae comprises most strains isolated from infections in immunocompromised patients, usually acquired after hospitalization [9, 10]. The main infections caused by classical K. pneumoniae are respiratory tract infection (RTI), bloodstream infection (BSI), and urinary tract infection (UTI) [11]. Classical strains are of low virulence and frequently exhibit high rates of antibiotic resistance [12, 13]; consequently, these infections are often difficult to treat.

Hypervirulent K. pneumoniae is highly invasive and is commonly found in community-acquired infections [14–18], mainly liver abscesses, meningitis, and disseminated infections [11, 19, 20]. Although more common in the community, several studies have reported that these strains cause hospital outbreaks [21, 22]. In contrast to classical strains, hypervirulent isolates are usually sensitive to antibiotics [6, 12, 13]; however, the convergence of multidrug resistance and hypervirulence is emerging [23–25], which is a global concern. This review aims to focus on the prevalence of virulence factors in K. pneumoniae and point out which characterizes hypervirulent strains.

Virulence Factors in Klebsiella pneumoniae

The main studied virulence factors in K. pneumoniae are capsule-associated, fimbriae, siderophores, Klebsiella ferric iron uptake, and the ability to metabolize allantoin (Fig. 1).

Fig. 1.

Schematic representation of the main virulence factors of Klebsiella pneumoniae

The target genes used to search for virulence factors in Klebsiella pneumoniae from different human infections are following: capsule-associated factors—magA for serotype K1, k₂A for serotype K2, and rmpA/rmpA2 for the regulator of mucoid phenotype A; fimbriae—fimH for type 1 fimbriae and mrkD for type 3 fimbriae; siderophores—entB for enterobactin, iucA, iucB, iucC, and iucD for aerobactin, iroB, and iroN for salmochelin, and ybtA, ybtS, irp1, irp2, and fyuA for yersiniabactin; Klebsiella ferric iron uptake—kfu; allantoin-utilizing capability—allS (virulence factor not shown in the figure).

Capsule

The capsule consists of an extracellular polysaccharide matrix surrounding the bacteria and is the most relevant and studied virulence factor in K. pneumoniae. The role of the capsule in K. pneumoniae is to prevent phagocytosis, hinder the antibacterial activity of antimicrobial peptides, block complement components preventing complement-mediated lysis and opsonization, and prevent fulminant activation of the immune response [9]. Although the capsule has historically been a target for vaccines in Gram-negative and Gram-positive bacteria, no licensed vaccine or therapy targeting the K. pneumoniae capsule is currently available to prevent or treat infections. However, studies show that the K. pneumoniae capsule has significant potential as a vaccine target [26–29].

Capsule biosynthesis involves a set of genes in the capsule synthesis locus, whose central region is highly variable and responsible for the variety of capsular serotypes [30, 31]. Seventy-seven capsular serotypes (referred to as K-types) were identified in K. pneumoniae by phenotypic studies; however, many other serologically non-typeable strains are now typed based on genomic analyses of the capsule synthesis loci (K-loci), with at least 134 distinct K-loci identified (referred to as KL-type) [31]. Nevertheless, K1 and K2 are the most common serotypes in hypervirulent strains [15, 32–34]. The magA gene is a specific marker of the K1 capsular serotype [35, 36]. K. pneumoniae strains with this serotype are strongly associated with liver abscesses [32, 37–39]. This association was confirmed in vivo mouse experiments, where the K. pneumoniae serotype K1 magA-deficient mutant lost the ability to cause liver abscess compared to the wild-type K1 strain [40]. The specific allele of serotype K2 is referred to as k₂A. Similar to K1 serotypes, K. pneumoniae K2-deficient lost the ability to cause a liver abscess in vivo mouse experiments compared to wild-type K2 strains [40].

Tan et al. [37] found an association among pyogenic liver abscess, K. pneumoniae serotype K1, and a positive string test in Singapore. The string test is a phenotypic test that determines the hypermucoviscosity phenotype [41] and was widely used as a hypervirulence marker. However, the molecular detection of the genes _prmpA/_prmpA2 (regulator of mucoid phenotype A/A2), iroB (salmochelin siderophore), and iucA (aerobactin siderophore) by conventional PCR is currently the best diagnostic method to identify hypervirulent K. pneumoniae strains, with diagnostic accuracy of > 95% for each gene [42]. In comparison, the detection accuracy of hypervirulent strains using the string test was 90% in the mentioned study.

The prevalence of liver abscess caused by K. pneumoniae serotype K1 was 64.3% (45/70) in an ethnically diverse population (Chinese, Malay, Indian, and Caucasian) [43] and 39.2% (20/51) in a study in China [39]. K. pneumoniae serotype K2 is the second most commonly found in liver abscess [32, 39, 43, 44]. The prevalence of K2 strains in pyogenic liver abscesses in a population of mainland China was 31.4% (16/51) [39], and in another study involving Chinese, Malay, Indian, and Caucasian people, it was 20% (14/70) [43]. In both studies, K1 was the dominant serotype.

K. pneumoniae serotype K1 is also common in bacteremias, and in many cases, pneumonia and liver abscesses are the primary sources [32, 37, 45]. The prevalence of magA in bacteremia isolates in a Japanese study was 47.3% (61/129) [46]. In two Chinese studies, the prevalence of K1 in BSI was 11% (38/344) [47] and 4.7% (9/190) [48]. In the last-mentioned study, serotype K2 was the most prevalent, 11.58% (22/190). Lastly, the prevalence of the K1 serotype was 13.3% (30/225) in bacteremia cases in Taiwan [49]. Although the prevalence appears low, K1 was the most prevalent serotype in almost all studies, followed by serotype K2, even with 134 known KL-types.

The prevalence of serotype K1 was 15.2% (5/33) in meningitis isolates in a retrospective study in different regions of Taiwan [50]. K2 strains were more prevalent than K1 in the same study, reaching 33.3% (11/33) of cases [50]. However, it is noteworthy that studies aimed at determining K. pneumoniae serotypes from meningitis are rare, making this actual prevalence a gap for future studies.

Regulator of mucoid phenotype A

The rmpA gene (regulator of mucoid phenotype A) regulates capsular synthesis, increases capsule production, and causes the hypermucoviscous phenotype [51]. This gene is a critical marker of hypervirulence in K. pneumoniae [34, 42], with a detection accuracy of > 95% [42].

Several studies report a high prevalence of the rmpA gene in K. pneumoniae isolated from a liver abscess [37, 44, 52]. All isolates from a liver abscess from a Taiwanese population had the rmpA gene regardless of serotype [32]. In the study by Lin et al. [44] in Hong Kong, Singapore, and Taiwan, 100% of K2 strains from liver abscesses presented the rmpA gene. In isolates from meningitis cases in a Taiwanese population, the prevalence of the rmpA gene was 69.7% (23/33), and that of rmpA2 was 60.6% (20/33) [50].

Although the distribution of capsular serotypes in community-acquired UTI by K. pneumoniae is not defined, the prevalence of rmpA was 29.6% in strains with the hypermucoviscous phenotype in a retrospective study in Taiwan [53]. On the other hand, the prevalence of the rmpA gene in UTI cases was 2.3% (4/170) in an Indian study [54]. In Portugal, the magA and rmpA genes were not found in the 81 UTI isolates in the study (50 community-acquired and 31 hospital-acquired isolates) [55]. The low occurrence of this virulence factor with UTI isolates is expected, as it is a marker of hypervirulence [42].

In community-acquired pneumonia, the prevalence of the rmpA gene was 69% (20/29) in sputum isolates in Taiwan [53]. The authors found an association of the rmpA gene with pneumonia compared to fecal isolates from healthy adults, where the prevalence was approximately 11.8% (9/76) (CI 95% 5.9–42.2; p < 0.0001). In a study of older adults in China with lower RTI caused by carbapenem-resistant K. pneumoniae, rmpA was present in 74.19% (23/31) of the isolates [56]. On the other hand, another study in India reported the prevalence of rmpA in 18.4% (7/38) of RTI [54]. In a study involving people from six countries (Australia, Indonesia, Laos, Singapore, Vietnam, and the United States), rmpA/rmpA2 were significantly associated with invasive infection (rmpA: odds ratio 15; p < 0.0001; rmpA2: odds ratio 9.1; p < 0.0001) [57].

The prevalence of rmpA in K. pneumoniae from blood cultures ranges from 6.6 to 37.3% in studies in Asian countries: 37.3% (84/225) in Taiwan [49], 26.8% (51/190) and 23% (79/344) in China [47, 48], 13.2% (17/129) in Japan [46], and 6.6% (2/30) in India [54]. The design of the studies may have been the reason for the wide range in prevalence. Additionally, the literature does not present data on the prevalence of this gene in K. pneumoniae isolated from this site in other countries outside the Asian continent, which makes an association of this virulence factor difficult with bacteremia cases.

Fimbriae

K. pneumoniae is believed to have ten types of fimbriae based on gene clusters identified in its genome. Recently, seven possible fimbrial gene clusters were identified and named kpa to kpg [58]; however, these clusters are still poorly characterized. The ecpABCDE operon responsible for expressing Escherichia coli common pilus (ECP), a highly studied fimbriae in E. coli, was recently found in the genome of K. pneumoniae [59]. However, little is said about it in the literature. On the other hand, type 1 and 3 fimbriae are the main adhesins in K. pneumoniae and are well characterized [60, 61].

Type 1 and 3 fimbriae are membrane-bound and are divided into structural and adhesive subunits. Each subunit has a specific gene organized in the fim (fimABCDEFGHIK-type 1 fimbriae) and mrk (mrkABCDF-type 3 fimbriae) gene clusters [60, 62]. However, of clinical and epidemiological interest, the adhesive subunits in the extremities are the most relevant in determining virulence [63, 64]. The genes responsible for these subunits are the most investigated in prevalence studies involving these virulence factors [55, 65–68].

Type 1 Fimbriae

In type 1 fimbriae, the fimH gene expresses the FimH adhesive subunit [69]. Through the FimH adhesin, K. pneumoniae recognizes and adheres to mannosylated residues in urothelium glycoproteins, uroplaquin Ia [70–72]. The urothelium is a tissue rich in uroplaquin [73]; for this reason, K. pneumoniae that expresses the fimH gene is considered uropathogenic and is frequently found in UTI [55, 65, 67, 68].

All K. pneumoniae isolates from UTI in the study by Eghbalpoor et al. [68] (n = 140) in Iran presented the fimH gene. The same occurred in cases of UTI in Algeria (n = 26) [67] and Portugal (n = 76) [65]. A study with type 1 fimbriae mutants showed that this adhesin is an essential virulence factor in UTI compared to the wild-type strain [74]. In this study, fim-deficient mutants had impaired urovirulence, and complementation of the fim gene cluster via plasmids restored this characteristic. Notably, almost all K. pneumoniae have FimH, which contributes to the infection of different sites along with other coexisting virulence factors. In a study, for example, the fimH gene was found in 100% of bloodstream isolates (n = 11), pus (n = 11), lung (n = 4), cerebrospinal fluid (n = 1), and peritoneal fluid (n = 1) [67].

Type 3 Fimbriae

The mrkD gene expresses the type 3 fimbriae adhesive subunit [64]. Although the mrkD gene has been reported in uncomplicated cases of UTI by K. pneumoniae strains [55, 65, 68], type 3 fimbriae do not make a significant contribution to this infection, which has already been demonstrated in a study with deficient mutants [74]. FimH adhesin, usually present with MrkD in isolates, is primarily responsible for urovirulence [74].

Several studies have demonstrated the ability of K. pneumoniae to form biofilms [66–68, 75–77]. K. pneumoniae type 3 fimbriae mediate biofilm formation, which has already been demonstrated by comparing mrk-deficient mutants and the wild-type strain [74, 76]. The prevalence of the mrkD gene in biofilm-producing K. pneumoniae was 100% (n = 43) in a study in China [77]. This same study found the fimH gene in 88.4% (38/43) of the isolates; therefore, the MrkD adhesin does not depend on FimH to form a biofilm. In Portugal, Bandeira et al. [66] also found biofilm-producing K. pneumoniae negative for fimH. However, in K. pneumoniae containing type 1 and 3 fimbriae, adhesins may contribute to UTI initiation (colonization process) and persistence [78].

Due to the ability to adhere to biotic and abiotic surfaces, such as glass, polyvinyl chloride (PVC), and silicone tubes [78, 79], K. pneumoniae is commonly found in medical devices such as urinary and intravascular catheters and endotracheal tubes [79, 80]. Therefore, K. pneumoniae is frequently reported in infections associated with these devices [80–82]. For example, a study in Algeria showed that 91.7% of K. pneumoniae isolates from urinary catheters and endotracheal tubes had the mrkD gene. Importantly, patients who were using these devices had severe infections [79]. Biofilm formation confers a protective barrier against the host immune system and is a physical barrier against antibiotics [66, 83]. In a biofilm, antibiotic resistance tends to be 10 to 1000 times higher than in the planktonic form [66, 79, 84].

Siderophores

Iron is an essential nutrient for both humans and microorganisms. For this reason, the concentrations of free iron in the human body are small because they are stored in proteins such as hemoglobin, haptoglobin, ferritin, and transferrin [85]. This concentration can be more intensely decreased with the help of other proteins (e.g., lactoferrin and hepcidin) in response to infection to inhibit bacterial growth [85]. Conversely, K. pneumoniae has mechanisms that capture and compete for this low free iron in the host. Siderophores are proteins with a high affinity for iron secreted to chelate and deliver iron to bacteria, essential for their growth and metabolism [86]. The ability to capture iron is higher in these proteins than in the host [9]; therefore, this is a relevant virulence factor.

Enterobactin

Among the siderophores that K. pneumoniae can secrete, enterobactin, encoded by the entABCDEF gene cluster where the entB gene was a marker in several studies [15, 54, 87, 88], is widespread in classical and hypervirulent strains [89]. Nonetheless, this siderophore is irrelevant in virulence because, when released, it is completely inhibited by the host’s innate immune protein lipocalin-2 [89]. Therefore, enterobactin is not associated with specific clinical manifestations.

Although enterobactin is ubiquitous in K. pneumoniae, the coexistence of different siderophores is common in hypervirulent strains. Studies show that hypervirulent K. pneumoniae quantitatively produce more siderophores than classical strains [90, 91]. Yersiniabactin, aerobactin, and salmochelin are siderophores associated with high virulence, and unlike enterobactin, these siderophores are not inhibited by lipocalin-2 [92, 93]. Peculiarly, enterobactin is the siderophore with the greatest iron affinity [90, 94].

Yersiniabactin

Yersiniabactin is the second most common siderophore in K. pneumoniae. The prevalence of yersiniabactin is approximately half of the classical strains but in almost all hypervirulent isolates [57, 95]. The gene cluster responsible for the expression of this siderophore is formed by ybtAEPQSTUX (also called ybt), irp1, and irp2, responsible for synthesis, and fyuA, whose product is the receptor [96, 97]. Among these genes, ybtA [98, 99], ybtS [54, 100], irp1 [88, 101], irp2 [102, 103], and fyuA [104, 105] were used as markers in several studies.

As is common in classic and hypervirulent K. pneumoniae, yersiniabactin is present in strains that cause different infections, coexisting with other virulence factors. However, in a study involving six countries (Australia, Indonesia, Laos, Singapore, Vietnam, and the United States), yersiniabactin was a strong predictor of invasive infection (OR 7.4; IC: 95% 2.2–40; p = 0.0001) [57].

In a study conducted in China, Bachman et al. [89] observed that yersiniabactin is a relevant virulence factor during pulmonary infection through evasion of lipocalin-2. Additionally, these authors described that yersiniabactin was significantly more prevalent in RTI isolates than in other sites, such as blood and urine. The prevalence of this virulence factor in K. pneumoniae isolated from RTI in an Indian study was 60.5% (23/38), the most prevalent siderophore after enterobactin [54]. In the study by Yan et al. [95] in China, the prevalence of the ybtS gene in classical and hypervirulent isolates from ventilator-associated pneumonia was 83.6%. In the mentioned study, 88.6% (31/35) of the classical isolates and 71.6% (10/14) of the hypervirulent isolates had yersiniabactin, and this difference in prevalence was not statistically significant (CI 95% 0.8–12.0; p = 0.202).

The prevalence of ybtS in K. pneumoniae from blood cultures in an Indian study was 63.3% (19/30) [54]. In two studies conducted in China, ybtS was present in 55.8% (192/344 and 106/190) of K. pneumoniae isolated from BSI [47, 48]. The prevalence of this virulence factor was 29.5% (38/129) in bacteremia isolates in Japan [46]. These findings suggest a possible correlation between ybtS and this infection, but this should be analyzed with caution, given the lack of similar studies in countries outside Asia.

Aerobactin

Aerobactin is a critical virulence factor often found in hypervirulent K. pneumoniae strains but rare in classical strains [34, 42, 90, 95, 106, 107]. For this reason, it is a critical marker in identifying hypervirulent K. pneumoniae [42, 106], with an accuracy of 96% [42]. The iucABCD gene cluster is responsible for this siderophore [108], and the iutA gene encodes the outer membrane receptor for aerobactin. The iutA gene is frequently used as a marker for this siderophore in K. pneumoniae [65, 87, 109]. iucA [110, 111], iucB [88], iucC [55, 101], and iucD [112] are also markers. However, several studies do not specify the gene used as a target in detecting aerobactin [34, 37, 39].

Aerobactin has been reported in K. pneumoniae isolated from different infections. The prevalence of aerobactin from blood cultures ranges from 6.6% to 50.9% in studies conducted in Asian countries. A study in China reported a prevalence of 50.9% (175/344) of the iucA gene in K. pneumoniae from BSI [47]. Namikawa et al. [46] found the iutA gene in 12.4% (16/129) of bacteremia isolates in Japan. In an Indian study, aerobactin was prevalent in only 6.6% (2/30) of the isolates from blood cultures [54]. Regarding the presence of aerobactin in hypervirulent isolates, 100% (14/14) of K. pneumoniae with a hypermucoviscous phenotype arising from bacteremia presented this siderophore in a study in Korea [34]. The rmpA gene was also present in 100% of the cases of these isolates, which explains the hypermucoviscosity. On the other hand, the prevalence of aerobactin in non-hypermucoviscous strains was 10.5% (2/19) in mentioned study. In Singapore, aerobactin was the most found virulence factor among those related to hypervirulence in bacteremic episodes (48.8%, 63/129) [37]. Together with rmpA and the positive string test, aerobactin was significantly associated with primary liver abscess. Aerobactin was the most prevalent virulence factor (86.3%, 44/51) in K. pneumoniae isolated from liver abscesses among the virulence factors investigated in the study by Luo et al. [39] in a Chinese population. The liver is an iron-rich organ, as is the blood [113]. Therefore, this characteristic is believed to explain the high prevalence of aerobactin in K. pneumoniae isolated from these sites.

The prevalence of aerobactin in K. pneumoniae from community-acquired (20%, 10/50) and hospital-acquired UTI cases (16.1%, 5/31) appears to be low [55]. Of 170 UTI isolates from a study in India, only 1.76% (3/170) had aerobactin [54]. In another study in Korea, the prevalence of aerobactin was only high in hypermucoviscous strains, with aerobactin being the most prevalent virulence factor (80%, 8/10). On the other hand, the prevalence in non-hypermucoviscous isolates was 15.5% (11/71). In this case, there was an association between the presence of aerobactin and the hypermucoviscous phenotype (CI 95% 4.0–106.3; p < 0.0001) [114]. In meningitis cases of a retrospective study performed in different regions of Taiwan, the prevalence of aerobactin in K. pneumoniae was 66.7% (22/33), and this siderophore was associated with primary meningitis compared to post-craniotomy meningitis (90%, 18/20 vs. 30.8%, 4/13, respectively; CI 95% 3.3–106.7; p = 0.0007) [50].

Salmochelin

Like aerobactin, the prevalence of salmochelin is high in hypervirulent K. pneumoniae and rare in classical strains [42, 95, 115, 116]. For this reason, it was an excellent marker of hypervirulence in one study, with 97% accuracy [42]. Salmochelin and its receptor are encoded by the iroA gene cluster, composed of the iroBCDE-iroN genes [117, 118]. iroB [88, 93, 95] and iroN [15, 119] are the most used target genes investigating this siderophore.

Salmochelin was statistically associated with invasive infection [57]. In a mouse model of sepsis, the iroA gene cluster was correlated with increased virulence and infection [92]. This siderophore is common in K. pneumoniae associated with liver abscess and appears to correlate with RTI. In E. coli, salmochelin promoted lung infection in an avian air sac model [120]. In K. pneumoniae, it was associated with worsening pulmonary infection compared to an enterobactin-producing strain alone in a murine pneumonia model [121]. However, the role of salmochelin in pulmonary infection remains unclear, and further studies are needed to fill this knowledge gap.

The prevalence of iroB in hypervirulent K. pneumoniae from ventilator-associated pneumonia was 100% (14/14) in a study in China [95]. This siderophore was absent in classical strains (n = 35) in the same study. Although we know of the association between yersiniabactin and ventilator-associated pneumonia and that ybtS was the most prevalent in that same study, it must be agreed that the coexistence of salmochelin potentiated the virulence of K. pneumoniae isolates. Despite the possible correlation with RTI, further studies on the prevalence of salmochelin in cases of K. pneumoniae pneumonia should be carried out.

Of 61 cases of liver abscess due to bacterial causes acquired in the community in a study conducted in China, 53 were due to K. pneumoniae. Of these, 88.7% (47/53) of the isolates were hypervirulent, and the prevalence of the iroB and iucA (aerobactin) genes was 100% (n = 47) [122]. In another study in China, the prevalence of salmochelin in isolates from pyogenic liver abscesses was 24.5% (40/163), and the most significant number of cases was associated with non-K1/K2 serotypes (p < 0.001) [119]. On the other hand, all isolates of K1 (n = 45) and K2 (n = 14) serotypes from liver abscesses in an ethnically diverse population presented salmochelin, and this high prevalence was associated with K1 serotype compared with non-K1 serotypes (100% vs. 84%; p = 0.014, for K1 and non-K1, respectively). The authors found salmochelin in 94.3% (66/70) of the isolates, the most prevalent siderophore [43].

Iron Uptake System

Host iron uptake can also be done by the Kfu system (Klebsiella Ferric Uptake), an ABC transporter that is encoded by the kfuABC operon [123]. The contribution of this factor to the increase in virulence involves the acquisition and metabolism of iron from the host, which ensures bacterial growth. Deleting this system (ΔkfuABC) reduced the virulence of K. pneumoniae in the mice model study, and its presence was associated with invasive infection [123]. There is a strong correlation between the kfu gene and hypervirulence in K. pneumoniae. Perhaps this is due to the high prevalence of this cluster in K. pneumoniae serotype K1 compared to non-K1 serotypes [33, 34, 39, 43, 100, 124].

The kfu gene is common in K. pneumoniae isolated from liver abscesses [37, 43, 100]. In a study in China, 100% (20/20) of K1 strains from liver abscesses had kfu, while only 6.25% (1/16) of K2 strains and 33.3% (5/15) of other serotypes had this virulence factor [39]. The prevalence of kfu in liver abscess isolates from a Chinese, Malay, Indian, and Caucasian population was also 100% (45/45) in K1 strains [43].

A study in Iraq found kfu in 54% (7/13) of K. pneumoniae isolated from UTIs [125]. The prevalence of kfu in isolates of a study in Portugal was 52.6% (10/19) in hospital-acquired UTI isolates and 42.1% (24/57) in community-acquired UTI isolates [65]. In India, this virulence factor was found in 19.4% (33/170) of K. pneumoniae isolates in urine samples [54]. In a Brazilian study, the prevalence of kfu in K. pneumoniae from UTIs was 14.6% (7/48) [126]. Although there are no studies with kfu-deficient mutants in K. pneumoniae from UTI, it is noted that this virulence factor is common in isolates from this site. However, there is variability in its prevalence in different countries.

The prevalence of kfu in K. pneumoniae identified in blood cultures was 43.3% (13/30) in an Indian study [54]. In a Japanese study, the prevalence of kfu in isolates from bacteremia was 30.2% (39/129) [46]. Another study in the same country found a similar prevalence, 29.9% (26/87) [124]. In respiratory secretion isolates from an Indian population, kfu was found in 31.5% (12/38) of the isolates [54]. In patients with ventilator-associated pneumonia in China, kfu was only associated with hypervirulent strains in 50% of cases (7/14) compared to classical K. pneumoniae (2.9%, 1/35) (CI 95% 4.1–389.6; p = 0.0003) [95]. In addition to these findings, Ku et al. [50] reported this virulence factor in 30.3% (10/33) of the isolates from meningitis in Taiwan. The lack of other studies is a limiting factor that does not allow us to correlate these findings.

Allantoin Metabolism

Allantoin, a product of purine catabolism, can be degraded by K. pneumoniae as a nitrogen and carbon source. Several enzymes are involved in this metabolic pathway, expressed by three operons and two transcriptional regulators, AllR (repressor) and AllS (activator) [127]. The allS gene is essential in activating this system, and for this reason, it is used as a marker for this virulence factor [15, 37, 54, 98].

Deleting the allS gene in K. pneumoniae significantly decreased the virulence and capacity to cause liver abscess in vivo in mice [127]. Despite this association, the prevalence of allS in liver abscesses varies between studies. allS was prevalent in 65.7% (46/70) of K. pneumoniae isolates from liver abscesses in a Chinese, Malay, Indian, and Caucasian population [43] and 41.7% (21/51) in a study in China [39]. One hundred percent (7/7) of the isolates in another study in France had the allS gene, and all were serotype K1 belonging to ST23 [128].

The allS gene has a low prevalence in K. pneumoniae from UTI in different populations: 4.2% (2/48) in a study in Brazil [126], 2.6% (2/76) in a study in Portugal [65], and 0.58% (1/170) in UTI isolates in an Indian study [54]. In Korea, Kim et al. [114] found an association between the allS gene and the hypermucoviscous phenotype in K. pneumoniae from UTI (40%, 4/10 vs. 9.9%, 7/71; CI 95% 1.6–25.7; p = 0.026). The authors also found a higher frequency of allS in K. pneumoniae from bacteremia secondary to UTI than strains from patients without bacteremia (54.5%, 6/11 vs. 7.1%, 5/70; CI 95% 3.8–58.9; p = 0.0005). However, the prevalence of this virulence factor in UTI isolates is correlated with only hypermucoviscous strains. For example, the global prevalence of this gene in the Kim et al. [114] study was 13.6% (11/81).

The allS gene was not found in K. pneumoniae (n = 30) from blood cultures in India [54]. In Japan, allS was found in only 4.6% (6/129) of bacteremia isolates [46]. In another study in the same country, the prevalence of allS in K. pneumoniae from BSI was 24.1% (21/87) [124]. This virulence factor was found in 55.8% (192/344) [47] and 47.4% (90/190) [48] of bacteremia isolates in two studies in China. Unfortunately, we did not find similar studies in countries outside Asia.

Concluding Remarks

The most investigated virulence factors in K. pneumoniae from different types of infections were capsule-associated (serotypes K1, K2, and regulator of mucoid phenotype A), fimbriae (type 1 and type 3), siderophores (yersiniabactin, salmochelin, and aerobactin), Klebsiella ferric iron uptake, and the ability to metabolize allantoin. It has already been shown that the regulator of mucoid phenotype A (rmpA and rmpA2 plasmidial genes), salmochelin (iroB), and aerobactin (iucA) had > 95% diagnostic accuracy for identifying K. pneumoniae hypervirulent strains, whereas yersiniabactin (irp2) and serotypes K1 and K2 had an accuracy of 79%, 77%, and 57%, respectively [42].

Various lines of evidence suggest that different virulence factors of K. pneumoniae are associated with specific infections (Table 1): serotypes K1 and K2 and the ability to metabolize allantoin with liver abscess; yersiniabactin and salmochelin with RTI; type 1 fimbriae with UTI; type 3 fimbriae with infections associated with medical devices through biofilm formation; and salmochelin and Klebsiella iron uptake system with invasive infection. Studies have shown that the inactivation of these virulence factors reduced the ability to cause infection in mouse models. Although this association, even in humans, the prevalence of these virulence factors varies and is not always present in isolates. We observed that many virulence factors coexist in hypervirulent strains, such as the hypervirulence markers aerobactin, salmochelin, and regulator of mucoid phenotype A, as well as those commonly found in classical strains (yersiniabactin, type 1 fimbriae, and type 3 fimbriae), which possibly increases the ability to cause infection.

Table 1.

Main virulence factors in Klebsiella pneumoniae reported in different infections

Virulence factor	Type/function	Target genes	Associated/correlated infection	References
Capsule	Capsular serotype K1; prevents phagocytosis, opsonization, the action of antimicrobial peptides and prevents activation of the immune system	magA	Liver abscessa	[40]
	Capsular serotype K2; prevents phagocytosis, opsonization, the action of antimicrobial peptides and prevents activation of the immune system	k2A	Liver abscessa	[40]
	Regulator of mucoid phenotype A; increases capsule production and is responsible for the hypermucoviscosity phenotype	rmpA/rmpA2	Liver abscessb	[32]
			Meningitisb	[50]
			Invasive infectiona	[57]
			Respiratory tract infectionb	[56]
Fimbriae	Type 1 fimbriae; adhesion	fimH	Urinary tract infectiona	[74]
	Type 3 fimbriae; adhesion and biofilm formation	mrkD	Infections associated with medical devicesa	[74, 76, 79]
Siderophores	Aerobactin; iron uptake	iucA, iucB, iucC, and iucD	Liver abscessa	[37]
			Bacteremia/Bloodstream infectionb	[47]
			Meningitisb	[50]
	Yersiniabactin; iron uptake	ybtA, ybtS, irp1, irp2, and fyuA	Bacteremia/Bloodstream infectionb	[47]
			Invasive infectiona	[57]
			Respiratory tract infectiona	[89]
	Salmochelin; iron uptake	iroB and iroN	Invasive infectiona	[57, 92]
			Liver abscessb	[43]
			Respiratory tract infectiona	[92, 120, 121]
Klebsiella Iron uptake system	Involved in the acquisition of iron	kfu	Invasive infectiona	[123]
			Liver abscessb	[43]
			Urinary tract infectionb	[65]
Allantoin metabolism	Involved in utilizing allantoin as an alternative source of nitrogen and carbon	allS	Liver abscessa	[127]
			Bacteremia/Bloodstream infectionb	[47]

^aVirulence factor associated with infection based on in vitro or in vivo studies

^bPresence of virulence factor in > 50% of cases

New studies should better characterize K. pneumoniae virulence, avoiding the limitation of focusing solely on a few virulence mechanisms. It was noted that salmochelin, a crucial virulence factor, has been little investigated in isolates from different infection types. Additionally, isolates from uncommon infection sites should be thoroughly characterized, and the prevalence of virulence factors must be estimated. For instance, there is a scarcity of studies that have explored the virulence of K. pneumoniae from meningitis cases. Finally, most studies that characterize virulence factors in K. pneumoniae described to date are in Asia, especially China, Japan, and Taiwan. Therefore, more studies are required to determine the real prevalence of virulence factors in K. pneumoniae from different infections in different geographic regions.

Author Contributions

Adriano de Souza Santos Monteiro did the literature search, data analysis, and drafted it; Soraia Machado Cordeiro had the idea of the article and drafted it; and the work was critically revised by Joice Neves Reis Pedreira.

Funding

Not applicable.

Data Availability

Not applicable.

Code Availability

Not applicable.

Declarations

Conflict of interest

The authors declare no conflicts/competing interests.

Ethical Approval

Not applicable.

Consent to Participate

Not applicable.

Consent for Publication

Not applicable.