Journal of Medical Microbiology (2014), 63, 1575–1583
DOI 10.1099/jmm.0.076562-0
Interactions of uroseptic Escherichia coli with renal (A-498) and gastrointestinal (HT-29) cell lines Tara L. Vollmerhausen,1 Jasmine L. Woods,1 Joan Faoagali2 and Mohammad Katouli1 Correspondence Mohammad Katouli [email protected]
1
Inflammation and Healing Research Cluster, School of Health and Sport Sciences, Faculty of Science, Health, Education and Engineering, University of the Sunshine Coast, Maroochydore DC, Australia
2
Princess Alexandra Hospital, Brisbane, Queensland, Australia
Received 31 March 2014 Accepted 7 October 2014
We investigated the ability of Escherichia coli isolated from septic patients with urinary tract infection (UTI) to translocate through the gastrointestinal (GI) tract of the same patients using cell-culture models. Forty-seven hospitalized patients with urosepsis were included in this study. E. coli was isolated from their urine and blood (total 94 isolates) and investigated for genetic relatedness and interaction with the cell lines A-498 and HT-29. An initial comparison of the strains isolated from urine and blood showed that 44 out of 47 patients (94 %) had identical strains in their blood and urine. The blood isolates adhered to both cell lines, although their rate of adherence to A-498 cells was significantly higher than that to HT-29 cells (5.8±3.8 per cell vs 2.8±1.9; P,0.0001). The rate of translocation in A-498 cells was also significantly higher after 120 min (8.7¾105 vs 2.9¾105; P50.0006). Three non-identical blood isolates were unable to translocate in HT-29 cells, indicating that host immune factors might be more important than bacterial ability to translocate the GI epithelium in these patients. Our data showed that blood isolates from uroseptic patients are able to adhere to and translocate through both cell lines. This suggests that E. coli in patients with UTI may translocate from either the GI tract or the urinary tract, hence questioning the assumption that the urinary tract is the only source of septicaemia in these patients.
INTRODUCTION Escherichia coli is a common inhabitant of the large intestine of humans and can cause life-threatening intestinal and extra-intestinal infections (Johnson, 1991; Wilson & Blitchington, 1996). E. coli has been shown to translocate through the gastrointestinal (GI) epithelium to cause septicaemia (Berg, 1980; MacFie et al., 1999; O’Boyle et al., 1998). The disruption of gut flora equilibrium, such as during enteric bacterial overgrowth (Guarner et al., 1997; Katouli et al., 1994), physical disruption of the epithelial barrier (Macutkiewicz et al., 2008; Parks et al., 2000; Yu & Martin, 2000) or impairment of the host immune system (Xu et al., 1998), can lead to the development of bacterial translocation. a-Haemolysin-producing E. coli has also been shown to create small openings within the GI epithelium, known as focal leaks, as a novel mechanism to allow translocation via a paracellular pathway (Troeger et al., 2007). The gut origin hypothesis for septicaemia implies that the GI tract forms the source of septicaemic events in patients due to bacterial translocation Abbreviations: ID, identity; GI, gastrointestinal; TEER, transepithelial electrical resistance; UTI, urinary tract infection; VG, virulence gene.
076562 G 2014 The Authors
Printed in Great Britain
from this site (Alexander et al., 1990; Berg & Garlington, 1979; Katouli et al., 1994; Sedman et al., 1994). The urinary tract is also an important site for bacterial translocation and has been linked to the passage of pathogenic E. coli into the bloodstream, a process known as urosepsis (Gransden et al., 1990; Leblebicioglu et al., 2003; Yamamoto et al., 1997). Bacterial adherence to the host epithelium is an important step in translocation. There are a number of adhesins that help E. coli to attach to the epithelium, including type 1, P, DR, S and F1C fimbriae (Johnson & Stell, 2000). In view of the GI tract being a common reservoir for uropathogenic E. coli in patients with urinary tract infection (UTI), identification of the true source of bacterial translocation in septicaemic patients with a UTI is difficult. It has been shown that certain clonal groups of E. coli in the GI tract have a better ability to adhere to and colonize both the urinary tract and the GI epithelium (Katouli et al., 1997; Ramos et al., 2010; Tullus et al., 1992). In the present study, we hypothesized that, in hospitalized patients, E. coli may cause UTIs but does not necessarily translocate through kidney cells; instead, it may adhere to the GI epithelium and translocate through these cells into the bloodstream to cause a gut-associated septicaemia. In view of the above, 1575
T. L. Vollmerhausen and others
the present study was undertaken to investigate the interaction of a collection of E. coli isolates from hospitalized patients having both UTIs and septicaemia, using cell-culture models of renal (A-498) and GI (HT-29) epithelium to determine whether E. coli in urosepsis patients can translocate through the GI epithelium. We also investigated the prevalence of different virulence genes (VGs) associated with the adherence of E. coli strains causing UTIs among these isolates in order to identify whether there was any correlation between the presence of any of these VGs, adherence and/or translocation of these bacteria in either cell line.
METHODS Patients and strains. E. coli isolates were obtained from the blood and urine of 47 hospitalized patients (73 % female, median age 67 years, range 18–97 years) suffering simultaneously from septicaemia and a symptomatic UTI (termed urosepsis) at the Princess Alexandra Hospital, Brisbane, Australia, between September 2006 and August 2007. All patients developed UTI and septicaemia after admission, and none of the patients was diagnosed with pyelonephritis. Only two patients (patient nos PA29 and PA31) had urinary catheters at the time of the UTI. The characteristics of the patients and urine upon examination are outlined in Table 1. There was no mixed infection among any samples from blood or urine. All strains were transferred to the laboratory and purified on MacConkey agar no. 3 (Oxoid). Bacterial strains were stored at 280 uC in tryptic soy broth (TSB; Oxoid) containing 20 % glycerol.
correlations and coefficients and clustering, were performed using the PhPWin software. Phylogenetic grouping and PCR detection of VGs. A multiplex
PCR screening for the chuA and yjaA genes, and the TSPE4.C2 DNA fragment, was used to determine the phylogenetic grouping of the isolates, as described previously (Clermont et al., 2000). Isolates were tested for the presence of adhesin genes known to be associated with E. coli causing UTIs. These included genes coding for the production of P fimbriae (papAH, papEF, papC, papG allele II and papG allele III), type 1 fimbriae (fimH), the central region of DR antigen-specific fimbrial and afimbrial adhesion operons (afa/draBC) and the central region of S fimbriae and F1C fimbriae operons (sfa/ focDE). Genes were grouped into two multiplex primer sets and three uniplex sets in a method modified from that previously described by Johnson & Stell (2000), with appropriate positive and negative controls. The multiplex PCR volume of 25 ml consisted of 2 ml template DNA, 16 PCR buffer (Qiagen), 0.4 mM dNTPs (Fisher Biotec), 2.5 mM MgCl2, 0.6 mM primer (Invitrogen) for papAH, papEF, fimH, papG allele III and sfa/focDE, and 1 U HotStartTaq DNA polymerase (Qiagen). Amplification was conducted in a Mastercycler gradient (Eppendorf) using the following cycling conditions: denaturation at 95 uC for 15 min; 25 cycles of 94 uC for 30 s, 63 uC for 30 s and 68 uC for 3 min; and a final extension at 72 uC for 10 min.
covered onto nutrient agar (Oxoid) grown overnight at 37 uC. DNA extraction was performed by boiling a single colony in sterile water at 95 uC for 10 min. All strains were confirmed as E. coli by screening for the presence of the universal stress protein (uspA) gene, as described previously (Chen & Griffiths, 1998).
The uniplex PCR was performed in a final volume of 25 ml consisting of 2 ml template DNA, 16 PCR buffer (Bioline), 0.4 mM dNTPs (Fisher Biotec), 0.5 mM MgCl2, 0.6 mM each primer (Invitrogen) for papG allele II and afa/draBC (0.3 mM for papC), and 1 U BIOTAQ DNA polymerase (Bioline). Amplification was conducted in a Mastercycler gradient (Eppendorf) using the following cycling conditions: denaturation at 95 uC for 4 min; 25 cycles of 94 uC for 30 s, 63 uC for 30 s and 68 uC for 3 min; and a final extension at 72 uC for 10 min. All amplified PCR products were separated on a 2 % agarose gel (Amresco, Astral Scientific) in 0.66 TBE buffer. PCR products were visualized following ethidium bromide staining using a Syngene GeneGenius Gel Light Imaging System.
Typing of isolates. All strains were typed using a high-resolution
Cell-culture conditions and adherence assay. The human colonic
DNA extraction and confirmation of E. coli. Strains were re-
biochemical fingerprinting method specifically developed for typing of E. coli strains (PhP-RE; PhPlate AB) (Ku¨hn & Mo¨llby, 1993), as described previously (Vollmerhausen et al., 2011). The fingerprinting method was performed according to the manufacturer’s instructions. The biochemical fingerprints of blood and urine isolates were then compared pairwise and the similarity between each pair of strains was calculated as the correlation coefficient and clustered according to the unweighted pair group mathematical average method to yield a dendrogram (Saeedi et al., 2005; Sneath & Sokal, 1973). The identity (ID) level of 0.975 was established based on reproducibility of control strains tested in duplicate. Isolates showing similarity to each other above the ID level were regarded as identical and assigned to the same biochemical phenotype. All data handling, including calculations of
adenocarcinoma cell line HT-29 [HTB-38; American Type Culture Collection (ATCC)] was used to model the GI epithelium. This cell line has been used previously as a model for bacterial interactions with the human GI epithelium (Couto et al., 2007; Wold et al., 1992). HT-29 cells were grown in RPMI 1640 with 25 mM HEPES and Lglutamine (Gibco), supplemented with 10 % FBS (Gibco) and 1 % penicillin/streptomycin (Thermofisher, Australia). The human kidney carcinoma cell line A-498 (HTB-44; ATCC) was used to model the kidney epithelium. A-498 cells have been used previously as a model for bacterial interaction with the kidney epithelium (Hedlund et al., 2001; Kai-Larsen et al., 2010). A-498 cells were grown in Eagle’s minimum essential medium (Gibco), supplemented with 10 % FBS and 1 % penicillin/streptomycin. Cell-culture medium was replaced
Table 1. Characteristics of patients and urine samples at the time of examination for all 47 patients Characteristic Age No. No. No. No.
1576
(years) of bacteria in urine (c.f.u. ml21) of leukocytes (cells l21) of erythrocytes (cells l21) of epithelial cells (cells l21)
Mean±SD
Range
66±18 9.46107 3.56108±1.96108 7.26107±1.26108 1.26107±7.96106
18–97 16106–16108 26107–56108 16107–1.86108 1.06107–4.06107
Journal of Medical Microbiology 63
E. coli translocation in urosepsis with antibiotic-free medium prior to the infection experiments. For the adherence assays, cells were seeded onto round 12 mm diameter glass coverslips in 24-well tissue-culture plates (Nunc) and grown to semi-confluence. E. coli isolates were grown in TSB in 100 ml flasks at 37 uC with agitation for 4 h, centrifuged, washed three times in 5 ml PBS (pH 7.4) and resuspended in PBS, and 100 ml of the suspension (~1.06108 c.f.u.) was inoculated into each well. The plates were incubated for 90 min at 37 uC after which non-adherent bacteria were washed from the coverslips. The cells were then fixed for 5 min in 95 % ethanol (for HT-29 cells) or methanol (for A-498 cells), stained with Giemsa stain and observed by light microscopy. Adherence was determined by counting the number of adherent bacteria on 100 randomly selected cells per bacterial isolate, from which the mean number of bacteria per cell was calculated. Isolates were grouped into low-(¡3 bacteria per cell), medium-(4–7 bacteria per cell) and high(¢8 bacteria per cell) adherence groups using a modified method by Grey & Kirov (1993). E. coli strains HMLN-1 and JM109 were used as high- and low-adherence controls, respectively. Translocation assay. Bacterial translocation was assessed using a two-compartment system with cells grown in Millicell Inserts (Millipore), each with a permeable base containing 0.8 mm diameter pores in a 24-well tissue-culture plate. Inserts were seeded with 400 ml growth medium containing 105 HT-29 cells and 104 A-498 cells and were then incubated at 37 uC in 5 % CO2. Transepithelial electrical resistance (TEER) between the inner and outer wells was measured and calculated using the following equation: measured monolayer resistance 2 measured blank resistance (filter without cells)6area of the filter (0.6 cm2) 5 transmonolayer resistance (V cm22) (Maher & McClean, 2006; Snowden et al., 2006). A stable TEER was achieved after 4 days for HT-29 cells (45–85 V cm22) and after 6 days for A-498 cells (45–95 V cm22). One hundred microlitres of bacterial suspension was then inoculated into each inner well (final concentration ~1.06106 c.f.u. per well) and the cells were incubated at 37 uC in 5 % CO2. After 15, 30, 60 and 120 min, samples were collected from the outer chamber, serially diluted in PBS, plated onto MacConkey agar no. 3 and grown overnight at 37 uC to determine the number of translocated bacteria by c.f.u. In all assays, E. coli HMLN-1, which was isolated from the mesenteric lymph nodes and blood of a patient with fatal haemorrhagic pancreatitis, was used as the positive control (Ljungdahl et al., 2000; Nettelbladt et al., 2000). E. coli JM109 was used as a negative control due to its previous demonstrated inability to translocate (Snowden et al., 2006). Statistical analysis. Statistical analysis was performed using
GraphPad Prism version 6.02 for Windows software (GraphPad Software). A x2 test was used to compare low, medium and high adherence for HT-29 and A-498 cells. Fisher’s exact test was used when comparing the difference between the numbers of bacteria or VGs in two groups. The degree of translocation of strains was compared using an unpaired t-test with Welch correction. Differences were considered significant if P,0.05.
different sources (Fig. 1). As a result, in all subsequent experiments, E. coli strains isolated from both the blood and urine of these three patients were tested as two separate non-replicate isolates. Blood isolates have higher adherence to A-498 cells To investigate whether E. coli isolated from the blood of patients with urosepsis possessed the ability to translocate through the GI epithelium, we tested their interaction with gut (HT-29) and kidney (A-498) cells. Variations were seen in the adherence rate of E. coli to both cell lines and therefore isolates were grouped into low (¡3 bacteria per cell), medium (4–7 bacteria per cell) and high (¢8 bacteria per cell) adherence. The number of blood isolates adhering to A-498 cells was significantly (P,0.0001) higher for high (11 vs 0) and medium (20 vs 6) adherence than those adhering to HT-29 cells (data not shown). The number of blood isolates adhering to HT-29 cells with low adherence was significantly (41 vs 16) higher than the number adhering to A-498 cells. Blood isolates have a higher translocation rate with A-498 cells The ability of blood isolates from 46 patients to translocate through HT-29 and A-498 cells was measured at intervals over 120 min. The TEER value of both cell lines during this time remained relatively constant throughout all experiments, demonstrating the integrity of the monolayer (Fig. 2a, b). The degree of translocation of the blood isolates varied between cell lines, ranging from 1.86104 to 4.26106 c.f.u. ml21 for A-498 and from 0 to 3.26106 c.f.u. ml21 for HT-29 cells. Whilst the mean rate of translocation for all 46 blood isolates in the two cell lines did not differ after the first 30 min, their mean rate of translocation in A-498 cells after 120 min increased by 1 log yielding a significantly (P50.0006) higher number of translocating bacteria compared with HT-29 cells (8.76105 and 2.96105 c.f.u. ml21 for each cell line, respectively; Fig. 2c). The higher rate of translocation of blood isolates in A498 cells was also associated with their higher level of adherence to A-498 cells (5.8±3.8, mean±SD) compared with HT-29 cells (2.8±1.9; P,0.0001) after 90 min. Six blood isolates showed higher translocation in HT-29 cells compared with A-498 cells.
RESULTS Typing of blood and urine isolates
Non-identical blood isolates did not translocate through HT-29 cells
Typing of the 47 pairs of isolates from blood and urine from the same patients showed that 44 pairs (94 %) had identical biochemical phenotypes (Fig. 1). Three nonidentical pairs (patient nos PA30, PA43 and PA48) were found to be distantly related to each other with similarity levels below 0.681, indicating that the E. coli isolated from the blood and urine of these patients originated from
Typing of E. coli isolates from the blood and urine showed non-identical strains in three patients. To determine whether the GI tract was a potential site for translocation of these strains, we investigated their ability to adhere to and translocate through HT-29 cells. Whilst the strains isolated from the urine of these patients belonged to phylogenetic group B2, only one blood isolate belonged to
http://jmm.sgmjournals.org
1577
T. L. Vollmerhausen and others
Similarity coefficient 0.5 0.6 0.7
0.8
0.9
ID-Level
Patient Source ID PA1 PA1 PA36 PA36 PA31 PA31 PA37 PA37 PA40 PA40 PA6 PA6 PA15 PA15 PA18 PA18 PA30 PA24 PA24 PA4 PA4 PA11 PA11 PA43 PA12 PA46 PA46 PA45 PA12 PA14 PA14 PA45 PA20 PA34 PA34 PA41 PA41 PA20 PA30 PA9 PA9 PA32 PA32 PA38 PA38 PA48 PA22 PA22 PA19 PA19 PA48 PA3 PA3 PA27 PA8 PA8 PA28 PA28 PA13 PA13 PA17 PA17 PA5 PA27 PA5 PA23 PA23 PA35 PA35 PA42 PA42 PA29 PA29 PA21 PA21 PA47 PA47 PA2 PA2 PA44 PA25 PA44 PA26 PA26 PA25 PA39 PA39 PA7 PA7 PA16 PA43 PA16 PA10 PA10
1578
B U B U B U B U B U B U B U B U U B U B U B U U B B U B U B U U B B U B U U B B U B U B U U U B B U B B U U B U B U B U B U B B U B U B U U B B U B U B U B U B B U B U U B U B U B B U B U
Age
Gender
67 years
F
22 years
F
78 years
M
81 years
F
82 years
F
87 years
F
41 years
F
73 years
F
61 years
F
83 years
F
62 years
M
81 years 85 years 66 years
M F F
87 years
F
37 years
F
66 years 50 years
F F
81 years
F
70 years 79 years
F M
67 years
F
18 years
M
Strain pair
Similarity Level
PA30B PA30U
0.681
PA43B PA43U
0.588
PA48B PA48U
0.663
30U
43U
30B
48U 67 years 41 years
F F
72 years 64 years
M M
48 years
F
97 years
F
57 years
M
65 years
F
79 years 50 years
F F
60 years
F
90 years
F
45 years
M
72 years
F
41 years
F
61 years
F
80 years
M
85 years 82 years
F F
90 years
F
58 years
M
66 years
M
22 years 81 years
F M
79 years
M
48B
43B
Journal of Medical Microbiology 63
E. coli translocation in urosepsis
Fig. 1. Dendrogram showing the similarity between 47 blood and urine pairs of E. coli isolates from patients with urosepsis. The ID level set at 0.975 is indicated by the dotted line. Identical strains are connected to each other above the ID level. Blood (B) and urinary (U) isolates (n53) of the same patients showing dissimilarity have been marked with arrows. Similarity levels for the three non-identical blood and urine pairs are shown in the insert table. The age and gender [male (M) or female (F)] for each patient are indicated for blood isolates.
this phylogenetic group, and the blood isolates from the other two patients belonged to phylogenetic groups A and B1 (Table 2). The blood isolates from these three patients showed low adherence to HT-29 cells, and all three blood isolates failed to translocate HT-29 cells at all tested time points (Table 2). The three non-identical blood isolates varied in their adherence rate (one each had low, medium and high adherence) to A-498 cells. All non-identical blood and urine isolates were able to translocate through A-498 cells after 120 min. Adherence does not correlate with translocation of HT-29 cells or VGs The majority of blood isolates showed higher adherence and translocation in A-498 cells (Fig. 2c). There was no correlation between adherence of the strains and their translocation in A-498 and HT-29 cells (data not shown). There was also no correlation between the prevalence of any of the VGs tested and the adherence of the strains to A498 and HT-29 cells (Table 3).
DISCUSSION Bacteria that translocate through the epithelial layer to reach underlying tissues have a greater opportunity to cause severe diseases such as septicaemia (MacFie, 2004). Patients experiencing bacterial translocation are twice as likely to become septicaemic (MacFie et al., 2006; O’Boyle et al., 1998). Bacterial overgrowth is one mechanism that can yield high numbers of bacteria in the mesenteric lymph nodes, promoting bacterial translocation (Berg, 1981; Katouli et al., 1994). The GI tract is also a major reservoir of E. coli causing UTIs, and the dominance of uropathogenic E. coli in the gut microflora is associated with incidences of UTI (Moreno et al., 2008). Overall, 32 % of cases of severe sepsis have been attributed to infection from genito-urinary sources, compared with 22 % from GI sources (Moss, 2005). Previously, identical strains in the blood, urine and stools of patients with bacteraemia have been identified, suggesting the spread of these strains from the GI tract, possibly also via the blood, to the urine (Brauner et al., 1991). A recent study using genome
80
80
TEER (Ω cm–2)
(b) 100
TEER (Ω cm–2)
(a) 100
60 40 20 0
0
30
60
90
60 40 20 0
120
0
30
Time (min)
Number of translocating E. coli × 104 (c.f.u. ml–1)
(c)
60
90
120
Time (min)
100
A-498 HT-29
80 60 40 20 0
0
15
30
60
120
Time (min)
Fig. 2. Translocation of blood isolates is higher in A-498 cells after 120 min. (a, b) TEER values (V cm”2) of A-498 cells (a) and HT-29 (b) cell monolayers remained stable during the translocation assay (mean±SD). (c) Comparison of the mean rate of translocation of all blood isolates (n546) in A-498 and HT-29 cell monolayers over 120 min (mean±SEM). One strain was excluded due to a cytotoxic effect. http://jmm.sgmjournals.org
1579
T. L. Vollmerhausen and others
Table 2. Adherence to (bacteria per cell) and translocation (c.f.u.) through A-498 and HT-29 cells of non-identical E. coli blood and urine isolates The source of isolation and the phylogenetic group (PGG) membership of each isolate are shown. Patient
PA30 PA43 PA48
Source
PGG
Blood Urine Blood Urine Blood Urine
B1 B2 B2 B2 A B2
Translocation (c.f.u. ml”1¾104)
Adherence (mean no. bacteria per cell±SD) HT-29
A-498
1.86±1.04 1.49±0.95 1.99±1.24 2.76±1.94 2.12±1.35 2.76±2.1
3.82±2.41 4.13±2.72 8.91±9.47 9.53±4.56 2.29±1.62 3.14±2.17
sequencing identified the presence of non-identical blood and urine E. coli strains in a patient with urosepsis (McNally et al., 2013). To understand the routes of infection in urosepsis, our study employed a twocompartment model of translocation using renal and GI epithelial cell lines. In this study, the majority of E. coli isolates causing urosepsis were able to translocate in our GI epithelial cell model. This laboratory model of translocation enabled an insight into the understanding of the routes in which E. coli strains enter the bloodstream, making the GI tract, in addition to the kidney, a potential site for translocation. Using a high-resolution biochemical fingerprinting method, the uroseptic strains were shown to be identical between blood and urine samples for the majority of patients, suggesting the urinary tract as the source of bacterial translocation in 94 % of uroseptic patients. It has been shown previously that this typing method is as powerful as molecular typing methods including randomly amplified polymorphic DNA PCR (Ramos et al., 2010; Vollmerhausen et al., 2011) and enterobacterial repetitive intergenic consensus PCR fingerprinting (Ansaruzzaman
HT-29
A-498
0 110.4 0 0.016 0 23.4
172.5 138.05 1.75 7.78 33.27 0.05
et al., 2000). Two of the three non-identical pairs were shown to belong to different phylogenetic groups, and all three pairs varied in their ability to adhere to and translocate through the kidney and GI epithelial cell lines. The fact that, apart from infection of the urinary tract, there was no other focus of infection in these patients suggests that the presence of these isolates in the blood may be due to their translocation from the gut of these patients. E. coli strains isolated from the blood and urine in this study were identified clinically as monoclonal using classical culture techniques. Recent studies have highlighted the limitations of classical culture-based diagnosis of infectious disease (Levert et al., 2010; McNally et al., 2013), and therefore the possibility of polyclonal infections in our study cannot be ruled out. Up to 58 % of patients experiencing extra-intestinal E. coli infections have been shown to be infected by either several distinct clones or a single clone with microheterogeneity (Levert et al., 2010). Future phenotypic and genotypic investigation of multiple colonies from the site of infection may provide greater insight into the complexity of polyclonal extra-intestinal infections. Despite these limitations, our study highlights
Table 3. Prevalence of eight VGs among 47 E. coli strains isolated from uroseptic patients and their relationship with adherence to A-498 and HT-29 cells Isolates were grouped according to the mean number of bacteria adhering to cells into low (,3 bacteria per cell), medium (4–7 bacteria per cell) and high (.8 bacteria per cell) adherence. Virulence gene
A-498 adherence (%) Low (n516)
papAH papEF papC papG allele III papG allele II fimH afa/draBC sfa/focDE
1580
12 11 9 5 9 16
(75) (69) (56) (31) (56) (100) 0 4 (25)
Medium (n520) 16 13 12 3 12 20 1 6
(80) (65) (60) (15) (60) (100) (5) (30)
HT-29 adherence (%) High (n511) 8 9 7 2 8 11 1 3
(73) (81) (64) (18) (73) (100) (9) (27)
Low (n541) 31 28 25 9 25 41
(76) (68) (61) (22) (61) (100) 0 12 (29)
Medium (n56) 5 5 3 1 4 6 2 1
(83) (83) (50) (17) (67) (100) (33) (17)
Journal of Medical Microbiology 63
E. coli translocation in urosepsis
the potential for E. coli isolated from the blood of patients with urosepsis to enter the blood from either the kidney or the GI tract in the majority of patients. Alternatively, these isolates may contain adhesin molecules, such as P fimbriae, which has been shown to promote adherence of E. coli to the urinary tract epithelium better than the GI epithelium. Using animal models, many studies have shown that adherence is an important step and even a predictor of translocation (Katayama et al., 1997; Katouli et al., 1997; Nettelbladt et al., 1997). Contrary to previous findings, this study identified no correlation between the number of adhering bacteria and their rate of translocation in A-498 and HT-29 cells (Katayama et al., 1997; Katouli et al., 1994; Nettelbladt et al., 1997). Although the prevalence of the adhesive tip, FimH, has no correlation with uropathogenic strains (Johnson, 1991), its expression is crucial during the initial stages of infection to mediate binding to uroepithelial cells (Arthur et al., 1989; Holden & Gally, 2004) and plays an important role mediating invasion of host cells for the formation of intracellular bacterial communities. Type 1-piliated E. coli can mediate host-cell invasion through the recognition of integrins (Eto et al., 2007) to allow bacteria to colonize, multiply and persist as intracellular bacterial communities (Berry et al., 2009). From the urinary tract, bacteria can gain entry into the bloodstream via the kidney, with no evidence that bacterial translocation can occur from the bladder. The patients in our study were suffering from cystitis, with no clinical data to support a diagnosis of pyelonephritis. Clinically, diagnosis between cystitis and pyelonephritis based on symptoms, especially among elderly patients with septicaemia, might be difficult. In view of the GI tract being a major reservoir for E. coli, the potential exists for bacterial translocation across the GI epithelium. From the findings in this study, it is proposed that, in hospitalized patients with concurrent septicaemia and UTI, the source of translocation may in some cases be the GI tract. This is further supported by the fact that most septicaemic patients in this study were of a higher age group (median age 67 years), who often have a decreased immune system, a factor that has been linked with increased bacterial translocation (Grubeck-Loebenstein, 1997; MacFie et al., 1999; O’Boyle et al., 1998). We have previously shown a higher prevalence of uropathogenic VGs among E. coli isolated from the gut of healthy elderly adults compared with young adults (Vollmerhausen et al., 2011), which may be a risk factor for extra-intestinal infections among hospitalized elderly patients. In the absence of any study characterizing the E. coli population within the GI tract of such uroseptic patients, the translocation of E. coli from the GI tract of patients with urosepsis cannot be fully justified. In conclusion, we found that, whilst the majority of E. coli strains isolated from the blood and urine of the same patient were identical, it is possible that some patients suffered infection by two different E. coli strains in the http://jmm.sgmjournals.org
urinary tract and in their bloodstream. The fact that nonidentical pairs of isolates did not show the ability to translocate HT-29 cells may suggest that either this cell line is not as suitable as other cells such as Caco-2 for assessing translocation or as-yet-unidentified host factors may play a role in translocation from the GI tract in these patients. The fact that adherence to HT-29 cells was not correlated with translocation further suggests that factors other than adhesion are necessary for translocation of uropathogenic E. coli. Based on these results, it is suggested that future studies on the source of bacterial translocation in patients with urosepsis include testing of additional cell lines as well as characterizing E. coli populations of the gut at the time of hospitalization.
REFERENCES Alexander, J. W., Boyce, S. T., Babcock, G. F., Gianotti, L., Peck, M. D., Dunn, D. L., Pyles, T., Childress, C. P. & Ash, S. K. (1990). The process
of microbial translocation. Ann Surg 212, 496–510, discussion 511– 512. Ansaruzzaman, M., Albert, M. J., Nahar, S., Byun, R., Katouli, M., Ku¨hn, I. & Mo¨llby, R. (2000). Clonal groups of enteropathogenic
Escherichia coli isolated in case–control studies of diarrhoea in Bangladesh. J Med Microbiol 49, 177–185. Arthur, M., Johnson, C. E., Rubin, R. H., Arbeit, R. D., Campanelli, C., Kim, C., Steinbach, S., Agarwal, M., Wilkinson, R. & Goldstein, R. (1989). Molecular epidemiology of adhesin and hemolysin virulence
factors among uropathogenic Escherichia coli. Infect Immun 57, 303– 313. Berg, R. D. (1980). Mechanisms confining indigenous bacteria to the gastrointestinal tract. Am J Clin Nutr 33 (Suppl), 2472–2484. Berg, R. D. (1981). Promotion of the translocation of enteric bacteria
from the gastrointestinal tracts of mice by oral treatment with penicillin, clindamycin, or metronidazole. Infect Immun 33, 854–861. Berg, R. D. & Garlington, A. W. (1979). Translocation of certain
indigenous bacteria from the gastrointestinal tract to the mesenteric lymph nodes and other organs in a gnotobiotic mouse model. Infect Immun 23, 403–411. Berry, R. E., Klumpp, D. J. & Schaeffer, A. J. (2009). Urothelial
cultures support intracellular bacterial community formation by uropathogenic Escherichia coli. Infect Immun 77, 2762–2772. Brauner, A., Kaijser, B., Wretlind, B. & Ku¨hn, I. (1991). Characterization
of Escherichia coli isolated in blood, urine and faeces from bacteraemic patients and possible spread of infection. APMIS 99, 381–386. Chen, J. & Griffiths, M. W. (1998). PCR differentiation of Escherichia coli from other Gram-negative bacteria using primers derived from the nucleotide sequences flanking the gene encoding the universal stress protein. Lett Appl Microbiol 27, 369–371. Clermont, O., Bonacorsi, S. & Bingen, E. (2000). Rapid and simple
determination of the Escherichia coli phylogenetic group. Appl Environ Microbiol 66, 4555–4558. Couto, C. R., Oliveira, S. S., Queiroz, M. L. & Freitas-Almeida, A. C. (2007). Interactions of clinical and environmental Aeromonas isolates
with Caco-2 and HT29 intestinal epithelial cells. Lett Appl Microbiol 45, 405–410. Eto, D. S., Jones, T. A., Sundsbak, J. L. & Mulvey, M. A. (2007).
Integrin-mediated host cell invasion by type 1-piliated uropathogenic Escherichia coli. PLoS Pathog 3, e100. 1581
T. L. Vollmerhausen and others Gransden, W. R., Eykyn, S. J., Phillips, I. & Rowe, B. (1990). Bacteremia
due to Escherichia coli: a study of 861 episodes. Rev Infect Dis 12, 1008– 1018. Grey, P. A. & Kirov, S. M. (1993). Adherence to HEp-2 cells and
enteropathogenic potential of Aeromonas spp. Epidemiol Infect 110, 279– 287. Grubeck-Loebenstein, B. (1997). Changes in the aging immune
system. Biologicals 25, 205–208. Guarner, C., Runyon, B. A., Young, S., Heck, M. & Sheikh, M. Y. (1997). Intestinal bacterial overgrowth and bacterial translocation in
cirrhotic rats with ascites. J Hepatol 26, 1372–1378. Hedlund, M., Duan, R. D., Nilsson, A˚., Svensson, M., Karpman, D. & Svanborg, C. (2001). Fimbriae, transmembrane signaling, and cell
activation. J Infect Dis 183 (Suppl 1), S47–S50. Holden, N. J. & Gally, D. L. (2004). Switches, cross-talk and memory in
Escherichia coli adherence. J Med Microbiol 53, 585–593.
MacFie, J., Reddy, B. S., Gatt, M., Jain, P. K., Sowdi, R. & Mitchell, C. J. (2006). Bacterial translocation studied in 927 patients over 13 years.
Br J Surg 93, 87–93. Macutkiewicz, C., Carlson, G., Clark, E., Dobrindt, U., Roberts, I. & Warhurst, G. (2008). Characterisation of Escherichia coli strains
involved in transcytosis across gut epithelial cells exposed to metabolic and inflammatory stress. Microbes Infect 10, 424–431. Maher, S. & McClean, S. (2006). Investigation of the cytotoxicity of
eukaryotic and prokaryotic antimicrobial peptides in intestinal epithelial cells in vitro. Biochem Pharmacol 71, 1289–1298. McNally, A., Alhashash, F., Collins, M., Alqasim, A., Paszckiewicz, K., Weston, V. & Diggle, M. (2013). Genomic analysis of extra-intestinal
pathogenic Escherichia coli urosepsis. Clin Microbiol Infect 19, E328– E334. Moreno, E., Andreu, A., Pigrau, C., Kuskowski, M. A., Johnson, J. R. & Prats, G. (2008). Relationship between Escherichia coli strains causing
Johnson, J. R. (1991). Virulence factors in Escherichia coli urinary
acute cystitis in women and the fecal E. coli population of the host. J Clin Microbiol 46, 2529–2534.
tract infection. Clin Microbiol Rev 4, 80–128.
Moss, M. (2005). Epidemiology of sepsis: race, sex, and chronic
Johnson, J. R. & Stell, A. L. (2000). Extended virulence genotypes of
alcohol abuse. Clin Infect Dis 41 (Suppl. 7), S490–S497.
Escherichia coli strains from patients with urosepsis in relation to phylogeny and host compromise. J Infect Dis 181, 261–272.
Nettelbladt, C. G., Katouli, M., Volpe, A., Bark, T., Muratov, V., Svenberg, T., Mo¨llby, R. & Ljungqvist, O. (1997). Starvation increases
Kai-Larsen, Y., Lu¨thje, P., Chromek, M., Peters, V., Wang, X., Holm, A˚., Ka´das, L., Hedlund, K. O., Johansson, J. & other authors (2010).
the number of coliform bacteria in the caecum and induces bacterial adherence to caecal epithelium in rats. Eur J Surg 163, 135–142.
Uropathogenic Escherichia coli modulates immune responses and its curli fimbriae interact with the antimicrobial peptide LL-37. PLoS Pathog 6, e1001010.
Nettelbladt, C. G., Katouli, M., Bark, T., Svenberg, T., Mo¨llby, R. & Ljungqvist, O. (2000). Evidence of bacterial translocation in fatal
Katayama, M., Xu, D., Specian, R. D. & Deitch, E. A. (1997). Role of
O’Boyle, C. J., MacFie, J., Mitchell, C. J., Johnstone, D., Sagar, P. M. & Sedman, P. C. (1998). Microbiology of bacterial translocation in
bacterial adherence and the mucus barrier on bacterial translocation: effects of protein malnutrition and endotoxin in rats. Ann Surg 225, 317–326. Katouli, M., Bark, T., Ljungqvist, O., Svenberg, T. & Mo¨llby, R. (1994).
Composition and diversity of intestinal coliform flora influence bacterial translocation in rats after hemorrhagic stress. Infect Immun 62, 4768–4774. Katouli, M., Nettebladt, C. G., Muratov, V., Ljungqvist, O., Bark, T., Svenberg, T. & Mo¨llby, R. (1997). Selective translocation of coliform
bacteria adhering to caecal epithelium of rats during catabolic stress. J Med Microbiol 46, 571–578. Ku¨hn, I. & Mo¨llby, R. (1993). The PhP-RS-System: a simple microplate
method for studying coliform bacterial population. J Microbiol Methods 17, 255–259. Leblebicioglu, H., Esen, S. & Turkish Nosocomial Urinary Tract Infection Study Group (2003). Hospital-acquired urinary tract
infections in Turkey: a nationwide multicenter point prevalence study. J Hosp Infect 53, 207–210. Levert, M., Zamfir, O., Clermont, O., Bouvet, O., Lespinats, S., Hipeaux, M. C., Branger, C., Picard, B., Saint-Ruf, C. & other authors (2010). Molecular and evolutionary bases of within-patient genotypic
and phenotypic diversity in Escherichia coli extraintestinal infections. PLoS Pathog 6, e1001125. Ljungdahl, M., Lundholm, M., Katouli, M., Rasmussen, I., Engstrand, L. & Haglund, U. (2000). Bacterial translocation in experimental
shock is dependent on the strains in the intestinal flora. Scand J Gastroenterol 35, 389–397. MacFie, J. (2004). Current status of bacterial translocation as a cause of surgical sepsis. Br Med Bull 71, 1–11.
hemorrhagic pancreatitis. J Trauma 48, 314–315.
humans. Gut 42, 29–35. Parks, R. W., Stuart Cameron, C. H., Gannon, C. D., Pope, C., Diamond, T. & Rowlands, B. J. (2000). Changes in gastrointestinal
morphology associated with obstructive jaundice. J Pathol 192, 526– 532. Ramos, N. L., Saayman, M. L., Chapman, T. A., Tucker, J. R., Smith, H. V., Faoagali, J., Chin, J. C., Brauner, A. & Katouli, M. (2010).
Genetic relatedness and virulence gene profiles of Escherichia coli strains isolated from septicaemic and uroseptic patients. Eur J Clin Microbiol Infect Dis 29, 15–23. Saeedi, B., Ta¨rnberg, M., Gill, H., Ha¨llgren, A., Jonasson, J., Nilsson, L. E., Isaksson, B., Ku¨hn, I. & Hanberger, H. (2005). Phene Plate
(PhP) biochemical fingerprinting. A screening method for epidemiological typing of enterococcal isolates. APMIS 113, 603–612. Sedman, P. C., Macfie, J., Sagar, P., Mitchell, C. J., May, J., ManceyJones, B. & Johnstone, D. (1994). The prevalence of gut translocation
in humans. Gastroenterology 107, 643–649. Sneath, P. H. A. & Sokal, R. R. (editors) (1973). Numerical Taxonomy: the Principles and Practice of Numerical Classification. San Francisco: W. H. Freeman & Co. Snowden, L., Wernbacher, L., Stenzel, D., Tucker, J. R., McKay, D., O’Brien, M. & Katouli, M. (2006). Prevalence of environmental
Aeromonas in South East Queensland, Australia: a study of their interactions with human monolayer Caco-2 cells. J Appl Microbiol 101, 964–975. Troeger, H., Richter, J. F., Beutin, L., Gu¨nzel, D., Dobrindt, U., Epple, H. J., Gitter, A. H., Zeitz, M., Fromm, M. & Schulzke, J. D. (2007). Escherichia coli a-haemolysin induces focal leaks in colonic epithe-
lium: a novel mechanism of bacterial translocation. Cell Microbiol 9, 2530–2540.
MacFie, J., O’Boyle, C., Mitchell, C. J., Buckley, P. M., Johnstone, D. & Sudworth, P. (1999). Gut origin of sepsis: a prospective study inves-
Tullus, K., Ku¨hn, I., Ørskov, I., Ørskov, F. & Mo¨llby, R. (1992). The
tigating associations between bacterial translocation, gastric microflora, and septic morbidity. Gut 45, 223–228.
importance of P and type 1 fimbriae for the persistence of Escherichia coli in the human gut. Epidemiol Infect 108, 415–421.
1582
Journal of Medical Microbiology 63
E. coli translocation in urosepsis
Vollmerhausen, T. L., Ramos, N. L., Gu¨ndogdu, A., Robinson, W., Brauner, A. & Katouli, M. (2011). Population structure and
uropathogenic virulence-associated genes of faecal Escherichia coli from healthy young and elderly adults. J Med Microbiol 60, 574–581.
Xu, D., Lu, Q. & Deitch, E. A. (1998). Elemental diet-induced bacterial
translocation associated with systemic and intestinal immune suppression. J Parenter Enteral Nutr 22, 37–41.
Wilson, K. H. & Blitchington, R. B. (1996). Human colonic biota
Yamamoto, S., Tsukamoto, T., Terai, A., Kurazono, H., Takeda, Y. & Yoshida, O. (1997). Genetic evidence supporting the fecal-perineal-
studied by ribosomal DNA sequence analysis. Appl Environ Microbiol 62, 2273–2278.
urethral hypothesis in cystitis caused by Escherichia coli. J Urol 157, 1127–1129.
Wold, A. E., Caugant, D. A., Lidin-Janson, G., de Man, P. & Svanborg, C. (1992). Resident colonic Escherichia coli strains frequently display
Yu, P. & Martin, C. M. (2000). Increased gut permeability and bacterial translocation in Pseudomonas pneumonia-induced sepsis. Crit Care Med 28, 2573–2577.
uropathogenic characteristics. J Infect Dis 165, 46–52.
http://jmm.sgmjournals.org
1583
DOI 10.1099/jmm.0.076562-0
Interactions of uroseptic Escherichia coli with renal (A-498) and gastrointestinal (HT-29) cell lines Tara L. Vollmerhausen,1 Jasmine L. Woods,1 Joan Faoagali2 and Mohammad Katouli1 Correspondence Mohammad Katouli [email protected]
1
Inflammation and Healing Research Cluster, School of Health and Sport Sciences, Faculty of Science, Health, Education and Engineering, University of the Sunshine Coast, Maroochydore DC, Australia
2
Princess Alexandra Hospital, Brisbane, Queensland, Australia
Received 31 March 2014 Accepted 7 October 2014
We investigated the ability of Escherichia coli isolated from septic patients with urinary tract infection (UTI) to translocate through the gastrointestinal (GI) tract of the same patients using cell-culture models. Forty-seven hospitalized patients with urosepsis were included in this study. E. coli was isolated from their urine and blood (total 94 isolates) and investigated for genetic relatedness and interaction with the cell lines A-498 and HT-29. An initial comparison of the strains isolated from urine and blood showed that 44 out of 47 patients (94 %) had identical strains in their blood and urine. The blood isolates adhered to both cell lines, although their rate of adherence to A-498 cells was significantly higher than that to HT-29 cells (5.8±3.8 per cell vs 2.8±1.9; P,0.0001). The rate of translocation in A-498 cells was also significantly higher after 120 min (8.7¾105 vs 2.9¾105; P50.0006). Three non-identical blood isolates were unable to translocate in HT-29 cells, indicating that host immune factors might be more important than bacterial ability to translocate the GI epithelium in these patients. Our data showed that blood isolates from uroseptic patients are able to adhere to and translocate through both cell lines. This suggests that E. coli in patients with UTI may translocate from either the GI tract or the urinary tract, hence questioning the assumption that the urinary tract is the only source of septicaemia in these patients.
INTRODUCTION Escherichia coli is a common inhabitant of the large intestine of humans and can cause life-threatening intestinal and extra-intestinal infections (Johnson, 1991; Wilson & Blitchington, 1996). E. coli has been shown to translocate through the gastrointestinal (GI) epithelium to cause septicaemia (Berg, 1980; MacFie et al., 1999; O’Boyle et al., 1998). The disruption of gut flora equilibrium, such as during enteric bacterial overgrowth (Guarner et al., 1997; Katouli et al., 1994), physical disruption of the epithelial barrier (Macutkiewicz et al., 2008; Parks et al., 2000; Yu & Martin, 2000) or impairment of the host immune system (Xu et al., 1998), can lead to the development of bacterial translocation. a-Haemolysin-producing E. coli has also been shown to create small openings within the GI epithelium, known as focal leaks, as a novel mechanism to allow translocation via a paracellular pathway (Troeger et al., 2007). The gut origin hypothesis for septicaemia implies that the GI tract forms the source of septicaemic events in patients due to bacterial translocation Abbreviations: ID, identity; GI, gastrointestinal; TEER, transepithelial electrical resistance; UTI, urinary tract infection; VG, virulence gene.
076562 G 2014 The Authors
Printed in Great Britain
from this site (Alexander et al., 1990; Berg & Garlington, 1979; Katouli et al., 1994; Sedman et al., 1994). The urinary tract is also an important site for bacterial translocation and has been linked to the passage of pathogenic E. coli into the bloodstream, a process known as urosepsis (Gransden et al., 1990; Leblebicioglu et al., 2003; Yamamoto et al., 1997). Bacterial adherence to the host epithelium is an important step in translocation. There are a number of adhesins that help E. coli to attach to the epithelium, including type 1, P, DR, S and F1C fimbriae (Johnson & Stell, 2000). In view of the GI tract being a common reservoir for uropathogenic E. coli in patients with urinary tract infection (UTI), identification of the true source of bacterial translocation in septicaemic patients with a UTI is difficult. It has been shown that certain clonal groups of E. coli in the GI tract have a better ability to adhere to and colonize both the urinary tract and the GI epithelium (Katouli et al., 1997; Ramos et al., 2010; Tullus et al., 1992). In the present study, we hypothesized that, in hospitalized patients, E. coli may cause UTIs but does not necessarily translocate through kidney cells; instead, it may adhere to the GI epithelium and translocate through these cells into the bloodstream to cause a gut-associated septicaemia. In view of the above, 1575
T. L. Vollmerhausen and others
the present study was undertaken to investigate the interaction of a collection of E. coli isolates from hospitalized patients having both UTIs and septicaemia, using cell-culture models of renal (A-498) and GI (HT-29) epithelium to determine whether E. coli in urosepsis patients can translocate through the GI epithelium. We also investigated the prevalence of different virulence genes (VGs) associated with the adherence of E. coli strains causing UTIs among these isolates in order to identify whether there was any correlation between the presence of any of these VGs, adherence and/or translocation of these bacteria in either cell line.
METHODS Patients and strains. E. coli isolates were obtained from the blood and urine of 47 hospitalized patients (73 % female, median age 67 years, range 18–97 years) suffering simultaneously from septicaemia and a symptomatic UTI (termed urosepsis) at the Princess Alexandra Hospital, Brisbane, Australia, between September 2006 and August 2007. All patients developed UTI and septicaemia after admission, and none of the patients was diagnosed with pyelonephritis. Only two patients (patient nos PA29 and PA31) had urinary catheters at the time of the UTI. The characteristics of the patients and urine upon examination are outlined in Table 1. There was no mixed infection among any samples from blood or urine. All strains were transferred to the laboratory and purified on MacConkey agar no. 3 (Oxoid). Bacterial strains were stored at 280 uC in tryptic soy broth (TSB; Oxoid) containing 20 % glycerol.
correlations and coefficients and clustering, were performed using the PhPWin software. Phylogenetic grouping and PCR detection of VGs. A multiplex
PCR screening for the chuA and yjaA genes, and the TSPE4.C2 DNA fragment, was used to determine the phylogenetic grouping of the isolates, as described previously (Clermont et al., 2000). Isolates were tested for the presence of adhesin genes known to be associated with E. coli causing UTIs. These included genes coding for the production of P fimbriae (papAH, papEF, papC, papG allele II and papG allele III), type 1 fimbriae (fimH), the central region of DR antigen-specific fimbrial and afimbrial adhesion operons (afa/draBC) and the central region of S fimbriae and F1C fimbriae operons (sfa/ focDE). Genes were grouped into two multiplex primer sets and three uniplex sets in a method modified from that previously described by Johnson & Stell (2000), with appropriate positive and negative controls. The multiplex PCR volume of 25 ml consisted of 2 ml template DNA, 16 PCR buffer (Qiagen), 0.4 mM dNTPs (Fisher Biotec), 2.5 mM MgCl2, 0.6 mM primer (Invitrogen) for papAH, papEF, fimH, papG allele III and sfa/focDE, and 1 U HotStartTaq DNA polymerase (Qiagen). Amplification was conducted in a Mastercycler gradient (Eppendorf) using the following cycling conditions: denaturation at 95 uC for 15 min; 25 cycles of 94 uC for 30 s, 63 uC for 30 s and 68 uC for 3 min; and a final extension at 72 uC for 10 min.
covered onto nutrient agar (Oxoid) grown overnight at 37 uC. DNA extraction was performed by boiling a single colony in sterile water at 95 uC for 10 min. All strains were confirmed as E. coli by screening for the presence of the universal stress protein (uspA) gene, as described previously (Chen & Griffiths, 1998).
The uniplex PCR was performed in a final volume of 25 ml consisting of 2 ml template DNA, 16 PCR buffer (Bioline), 0.4 mM dNTPs (Fisher Biotec), 0.5 mM MgCl2, 0.6 mM each primer (Invitrogen) for papG allele II and afa/draBC (0.3 mM for papC), and 1 U BIOTAQ DNA polymerase (Bioline). Amplification was conducted in a Mastercycler gradient (Eppendorf) using the following cycling conditions: denaturation at 95 uC for 4 min; 25 cycles of 94 uC for 30 s, 63 uC for 30 s and 68 uC for 3 min; and a final extension at 72 uC for 10 min. All amplified PCR products were separated on a 2 % agarose gel (Amresco, Astral Scientific) in 0.66 TBE buffer. PCR products were visualized following ethidium bromide staining using a Syngene GeneGenius Gel Light Imaging System.
Typing of isolates. All strains were typed using a high-resolution
Cell-culture conditions and adherence assay. The human colonic
DNA extraction and confirmation of E. coli. Strains were re-
biochemical fingerprinting method specifically developed for typing of E. coli strains (PhP-RE; PhPlate AB) (Ku¨hn & Mo¨llby, 1993), as described previously (Vollmerhausen et al., 2011). The fingerprinting method was performed according to the manufacturer’s instructions. The biochemical fingerprints of blood and urine isolates were then compared pairwise and the similarity between each pair of strains was calculated as the correlation coefficient and clustered according to the unweighted pair group mathematical average method to yield a dendrogram (Saeedi et al., 2005; Sneath & Sokal, 1973). The identity (ID) level of 0.975 was established based on reproducibility of control strains tested in duplicate. Isolates showing similarity to each other above the ID level were regarded as identical and assigned to the same biochemical phenotype. All data handling, including calculations of
adenocarcinoma cell line HT-29 [HTB-38; American Type Culture Collection (ATCC)] was used to model the GI epithelium. This cell line has been used previously as a model for bacterial interactions with the human GI epithelium (Couto et al., 2007; Wold et al., 1992). HT-29 cells were grown in RPMI 1640 with 25 mM HEPES and Lglutamine (Gibco), supplemented with 10 % FBS (Gibco) and 1 % penicillin/streptomycin (Thermofisher, Australia). The human kidney carcinoma cell line A-498 (HTB-44; ATCC) was used to model the kidney epithelium. A-498 cells have been used previously as a model for bacterial interaction with the kidney epithelium (Hedlund et al., 2001; Kai-Larsen et al., 2010). A-498 cells were grown in Eagle’s minimum essential medium (Gibco), supplemented with 10 % FBS and 1 % penicillin/streptomycin. Cell-culture medium was replaced
Table 1. Characteristics of patients and urine samples at the time of examination for all 47 patients Characteristic Age No. No. No. No.
1576
(years) of bacteria in urine (c.f.u. ml21) of leukocytes (cells l21) of erythrocytes (cells l21) of epithelial cells (cells l21)
Mean±SD
Range
66±18 9.46107 3.56108±1.96108 7.26107±1.26108 1.26107±7.96106
18–97 16106–16108 26107–56108 16107–1.86108 1.06107–4.06107
Journal of Medical Microbiology 63
E. coli translocation in urosepsis with antibiotic-free medium prior to the infection experiments. For the adherence assays, cells were seeded onto round 12 mm diameter glass coverslips in 24-well tissue-culture plates (Nunc) and grown to semi-confluence. E. coli isolates were grown in TSB in 100 ml flasks at 37 uC with agitation for 4 h, centrifuged, washed three times in 5 ml PBS (pH 7.4) and resuspended in PBS, and 100 ml of the suspension (~1.06108 c.f.u.) was inoculated into each well. The plates were incubated for 90 min at 37 uC after which non-adherent bacteria were washed from the coverslips. The cells were then fixed for 5 min in 95 % ethanol (for HT-29 cells) or methanol (for A-498 cells), stained with Giemsa stain and observed by light microscopy. Adherence was determined by counting the number of adherent bacteria on 100 randomly selected cells per bacterial isolate, from which the mean number of bacteria per cell was calculated. Isolates were grouped into low-(¡3 bacteria per cell), medium-(4–7 bacteria per cell) and high(¢8 bacteria per cell) adherence groups using a modified method by Grey & Kirov (1993). E. coli strains HMLN-1 and JM109 were used as high- and low-adherence controls, respectively. Translocation assay. Bacterial translocation was assessed using a two-compartment system with cells grown in Millicell Inserts (Millipore), each with a permeable base containing 0.8 mm diameter pores in a 24-well tissue-culture plate. Inserts were seeded with 400 ml growth medium containing 105 HT-29 cells and 104 A-498 cells and were then incubated at 37 uC in 5 % CO2. Transepithelial electrical resistance (TEER) between the inner and outer wells was measured and calculated using the following equation: measured monolayer resistance 2 measured blank resistance (filter without cells)6area of the filter (0.6 cm2) 5 transmonolayer resistance (V cm22) (Maher & McClean, 2006; Snowden et al., 2006). A stable TEER was achieved after 4 days for HT-29 cells (45–85 V cm22) and after 6 days for A-498 cells (45–95 V cm22). One hundred microlitres of bacterial suspension was then inoculated into each inner well (final concentration ~1.06106 c.f.u. per well) and the cells were incubated at 37 uC in 5 % CO2. After 15, 30, 60 and 120 min, samples were collected from the outer chamber, serially diluted in PBS, plated onto MacConkey agar no. 3 and grown overnight at 37 uC to determine the number of translocated bacteria by c.f.u. In all assays, E. coli HMLN-1, which was isolated from the mesenteric lymph nodes and blood of a patient with fatal haemorrhagic pancreatitis, was used as the positive control (Ljungdahl et al., 2000; Nettelbladt et al., 2000). E. coli JM109 was used as a negative control due to its previous demonstrated inability to translocate (Snowden et al., 2006). Statistical analysis. Statistical analysis was performed using
GraphPad Prism version 6.02 for Windows software (GraphPad Software). A x2 test was used to compare low, medium and high adherence for HT-29 and A-498 cells. Fisher’s exact test was used when comparing the difference between the numbers of bacteria or VGs in two groups. The degree of translocation of strains was compared using an unpaired t-test with Welch correction. Differences were considered significant if P,0.05.
different sources (Fig. 1). As a result, in all subsequent experiments, E. coli strains isolated from both the blood and urine of these three patients were tested as two separate non-replicate isolates. Blood isolates have higher adherence to A-498 cells To investigate whether E. coli isolated from the blood of patients with urosepsis possessed the ability to translocate through the GI epithelium, we tested their interaction with gut (HT-29) and kidney (A-498) cells. Variations were seen in the adherence rate of E. coli to both cell lines and therefore isolates were grouped into low (¡3 bacteria per cell), medium (4–7 bacteria per cell) and high (¢8 bacteria per cell) adherence. The number of blood isolates adhering to A-498 cells was significantly (P,0.0001) higher for high (11 vs 0) and medium (20 vs 6) adherence than those adhering to HT-29 cells (data not shown). The number of blood isolates adhering to HT-29 cells with low adherence was significantly (41 vs 16) higher than the number adhering to A-498 cells. Blood isolates have a higher translocation rate with A-498 cells The ability of blood isolates from 46 patients to translocate through HT-29 and A-498 cells was measured at intervals over 120 min. The TEER value of both cell lines during this time remained relatively constant throughout all experiments, demonstrating the integrity of the monolayer (Fig. 2a, b). The degree of translocation of the blood isolates varied between cell lines, ranging from 1.86104 to 4.26106 c.f.u. ml21 for A-498 and from 0 to 3.26106 c.f.u. ml21 for HT-29 cells. Whilst the mean rate of translocation for all 46 blood isolates in the two cell lines did not differ after the first 30 min, their mean rate of translocation in A-498 cells after 120 min increased by 1 log yielding a significantly (P50.0006) higher number of translocating bacteria compared with HT-29 cells (8.76105 and 2.96105 c.f.u. ml21 for each cell line, respectively; Fig. 2c). The higher rate of translocation of blood isolates in A498 cells was also associated with their higher level of adherence to A-498 cells (5.8±3.8, mean±SD) compared with HT-29 cells (2.8±1.9; P,0.0001) after 90 min. Six blood isolates showed higher translocation in HT-29 cells compared with A-498 cells.
RESULTS Typing of blood and urine isolates
Non-identical blood isolates did not translocate through HT-29 cells
Typing of the 47 pairs of isolates from blood and urine from the same patients showed that 44 pairs (94 %) had identical biochemical phenotypes (Fig. 1). Three nonidentical pairs (patient nos PA30, PA43 and PA48) were found to be distantly related to each other with similarity levels below 0.681, indicating that the E. coli isolated from the blood and urine of these patients originated from
Typing of E. coli isolates from the blood and urine showed non-identical strains in three patients. To determine whether the GI tract was a potential site for translocation of these strains, we investigated their ability to adhere to and translocate through HT-29 cells. Whilst the strains isolated from the urine of these patients belonged to phylogenetic group B2, only one blood isolate belonged to
http://jmm.sgmjournals.org
1577
T. L. Vollmerhausen and others
Similarity coefficient 0.5 0.6 0.7
0.8
0.9
ID-Level
Patient Source ID PA1 PA1 PA36 PA36 PA31 PA31 PA37 PA37 PA40 PA40 PA6 PA6 PA15 PA15 PA18 PA18 PA30 PA24 PA24 PA4 PA4 PA11 PA11 PA43 PA12 PA46 PA46 PA45 PA12 PA14 PA14 PA45 PA20 PA34 PA34 PA41 PA41 PA20 PA30 PA9 PA9 PA32 PA32 PA38 PA38 PA48 PA22 PA22 PA19 PA19 PA48 PA3 PA3 PA27 PA8 PA8 PA28 PA28 PA13 PA13 PA17 PA17 PA5 PA27 PA5 PA23 PA23 PA35 PA35 PA42 PA42 PA29 PA29 PA21 PA21 PA47 PA47 PA2 PA2 PA44 PA25 PA44 PA26 PA26 PA25 PA39 PA39 PA7 PA7 PA16 PA43 PA16 PA10 PA10
1578
B U B U B U B U B U B U B U B U U B U B U B U U B B U B U B U U B B U B U U B B U B U B U U U B B U B B U U B U B U B U B U B B U B U B U U B B U B U B U B U B B U B U U B U B U B B U B U
Age
Gender
67 years
F
22 years
F
78 years
M
81 years
F
82 years
F
87 years
F
41 years
F
73 years
F
61 years
F
83 years
F
62 years
M
81 years 85 years 66 years
M F F
87 years
F
37 years
F
66 years 50 years
F F
81 years
F
70 years 79 years
F M
67 years
F
18 years
M
Strain pair
Similarity Level
PA30B PA30U
0.681
PA43B PA43U
0.588
PA48B PA48U
0.663
30U
43U
30B
48U 67 years 41 years
F F
72 years 64 years
M M
48 years
F
97 years
F
57 years
M
65 years
F
79 years 50 years
F F
60 years
F
90 years
F
45 years
M
72 years
F
41 years
F
61 years
F
80 years
M
85 years 82 years
F F
90 years
F
58 years
M
66 years
M
22 years 81 years
F M
79 years
M
48B
43B
Journal of Medical Microbiology 63
E. coli translocation in urosepsis
Fig. 1. Dendrogram showing the similarity between 47 blood and urine pairs of E. coli isolates from patients with urosepsis. The ID level set at 0.975 is indicated by the dotted line. Identical strains are connected to each other above the ID level. Blood (B) and urinary (U) isolates (n53) of the same patients showing dissimilarity have been marked with arrows. Similarity levels for the three non-identical blood and urine pairs are shown in the insert table. The age and gender [male (M) or female (F)] for each patient are indicated for blood isolates.
this phylogenetic group, and the blood isolates from the other two patients belonged to phylogenetic groups A and B1 (Table 2). The blood isolates from these three patients showed low adherence to HT-29 cells, and all three blood isolates failed to translocate HT-29 cells at all tested time points (Table 2). The three non-identical blood isolates varied in their adherence rate (one each had low, medium and high adherence) to A-498 cells. All non-identical blood and urine isolates were able to translocate through A-498 cells after 120 min. Adherence does not correlate with translocation of HT-29 cells or VGs The majority of blood isolates showed higher adherence and translocation in A-498 cells (Fig. 2c). There was no correlation between adherence of the strains and their translocation in A-498 and HT-29 cells (data not shown). There was also no correlation between the prevalence of any of the VGs tested and the adherence of the strains to A498 and HT-29 cells (Table 3).
DISCUSSION Bacteria that translocate through the epithelial layer to reach underlying tissues have a greater opportunity to cause severe diseases such as septicaemia (MacFie, 2004). Patients experiencing bacterial translocation are twice as likely to become septicaemic (MacFie et al., 2006; O’Boyle et al., 1998). Bacterial overgrowth is one mechanism that can yield high numbers of bacteria in the mesenteric lymph nodes, promoting bacterial translocation (Berg, 1981; Katouli et al., 1994). The GI tract is also a major reservoir of E. coli causing UTIs, and the dominance of uropathogenic E. coli in the gut microflora is associated with incidences of UTI (Moreno et al., 2008). Overall, 32 % of cases of severe sepsis have been attributed to infection from genito-urinary sources, compared with 22 % from GI sources (Moss, 2005). Previously, identical strains in the blood, urine and stools of patients with bacteraemia have been identified, suggesting the spread of these strains from the GI tract, possibly also via the blood, to the urine (Brauner et al., 1991). A recent study using genome
80
80
TEER (Ω cm–2)
(b) 100
TEER (Ω cm–2)
(a) 100
60 40 20 0
0
30
60
90
60 40 20 0
120
0
30
Time (min)
Number of translocating E. coli × 104 (c.f.u. ml–1)
(c)
60
90
120
Time (min)
100
A-498 HT-29
80 60 40 20 0
0
15
30
60
120
Time (min)
Fig. 2. Translocation of blood isolates is higher in A-498 cells after 120 min. (a, b) TEER values (V cm”2) of A-498 cells (a) and HT-29 (b) cell monolayers remained stable during the translocation assay (mean±SD). (c) Comparison of the mean rate of translocation of all blood isolates (n546) in A-498 and HT-29 cell monolayers over 120 min (mean±SEM). One strain was excluded due to a cytotoxic effect. http://jmm.sgmjournals.org
1579
T. L. Vollmerhausen and others
Table 2. Adherence to (bacteria per cell) and translocation (c.f.u.) through A-498 and HT-29 cells of non-identical E. coli blood and urine isolates The source of isolation and the phylogenetic group (PGG) membership of each isolate are shown. Patient
PA30 PA43 PA48
Source
PGG
Blood Urine Blood Urine Blood Urine
B1 B2 B2 B2 A B2
Translocation (c.f.u. ml”1¾104)
Adherence (mean no. bacteria per cell±SD) HT-29
A-498
1.86±1.04 1.49±0.95 1.99±1.24 2.76±1.94 2.12±1.35 2.76±2.1
3.82±2.41 4.13±2.72 8.91±9.47 9.53±4.56 2.29±1.62 3.14±2.17
sequencing identified the presence of non-identical blood and urine E. coli strains in a patient with urosepsis (McNally et al., 2013). To understand the routes of infection in urosepsis, our study employed a twocompartment model of translocation using renal and GI epithelial cell lines. In this study, the majority of E. coli isolates causing urosepsis were able to translocate in our GI epithelial cell model. This laboratory model of translocation enabled an insight into the understanding of the routes in which E. coli strains enter the bloodstream, making the GI tract, in addition to the kidney, a potential site for translocation. Using a high-resolution biochemical fingerprinting method, the uroseptic strains were shown to be identical between blood and urine samples for the majority of patients, suggesting the urinary tract as the source of bacterial translocation in 94 % of uroseptic patients. It has been shown previously that this typing method is as powerful as molecular typing methods including randomly amplified polymorphic DNA PCR (Ramos et al., 2010; Vollmerhausen et al., 2011) and enterobacterial repetitive intergenic consensus PCR fingerprinting (Ansaruzzaman
HT-29
A-498
0 110.4 0 0.016 0 23.4
172.5 138.05 1.75 7.78 33.27 0.05
et al., 2000). Two of the three non-identical pairs were shown to belong to different phylogenetic groups, and all three pairs varied in their ability to adhere to and translocate through the kidney and GI epithelial cell lines. The fact that, apart from infection of the urinary tract, there was no other focus of infection in these patients suggests that the presence of these isolates in the blood may be due to their translocation from the gut of these patients. E. coli strains isolated from the blood and urine in this study were identified clinically as monoclonal using classical culture techniques. Recent studies have highlighted the limitations of classical culture-based diagnosis of infectious disease (Levert et al., 2010; McNally et al., 2013), and therefore the possibility of polyclonal infections in our study cannot be ruled out. Up to 58 % of patients experiencing extra-intestinal E. coli infections have been shown to be infected by either several distinct clones or a single clone with microheterogeneity (Levert et al., 2010). Future phenotypic and genotypic investigation of multiple colonies from the site of infection may provide greater insight into the complexity of polyclonal extra-intestinal infections. Despite these limitations, our study highlights
Table 3. Prevalence of eight VGs among 47 E. coli strains isolated from uroseptic patients and their relationship with adherence to A-498 and HT-29 cells Isolates were grouped according to the mean number of bacteria adhering to cells into low (,3 bacteria per cell), medium (4–7 bacteria per cell) and high (.8 bacteria per cell) adherence. Virulence gene
A-498 adherence (%) Low (n516)
papAH papEF papC papG allele III papG allele II fimH afa/draBC sfa/focDE
1580
12 11 9 5 9 16
(75) (69) (56) (31) (56) (100) 0 4 (25)
Medium (n520) 16 13 12 3 12 20 1 6
(80) (65) (60) (15) (60) (100) (5) (30)
HT-29 adherence (%) High (n511) 8 9 7 2 8 11 1 3
(73) (81) (64) (18) (73) (100) (9) (27)
Low (n541) 31 28 25 9 25 41
(76) (68) (61) (22) (61) (100) 0 12 (29)
Medium (n56) 5 5 3 1 4 6 2 1
(83) (83) (50) (17) (67) (100) (33) (17)
Journal of Medical Microbiology 63
E. coli translocation in urosepsis
the potential for E. coli isolated from the blood of patients with urosepsis to enter the blood from either the kidney or the GI tract in the majority of patients. Alternatively, these isolates may contain adhesin molecules, such as P fimbriae, which has been shown to promote adherence of E. coli to the urinary tract epithelium better than the GI epithelium. Using animal models, many studies have shown that adherence is an important step and even a predictor of translocation (Katayama et al., 1997; Katouli et al., 1997; Nettelbladt et al., 1997). Contrary to previous findings, this study identified no correlation between the number of adhering bacteria and their rate of translocation in A-498 and HT-29 cells (Katayama et al., 1997; Katouli et al., 1994; Nettelbladt et al., 1997). Although the prevalence of the adhesive tip, FimH, has no correlation with uropathogenic strains (Johnson, 1991), its expression is crucial during the initial stages of infection to mediate binding to uroepithelial cells (Arthur et al., 1989; Holden & Gally, 2004) and plays an important role mediating invasion of host cells for the formation of intracellular bacterial communities. Type 1-piliated E. coli can mediate host-cell invasion through the recognition of integrins (Eto et al., 2007) to allow bacteria to colonize, multiply and persist as intracellular bacterial communities (Berry et al., 2009). From the urinary tract, bacteria can gain entry into the bloodstream via the kidney, with no evidence that bacterial translocation can occur from the bladder. The patients in our study were suffering from cystitis, with no clinical data to support a diagnosis of pyelonephritis. Clinically, diagnosis between cystitis and pyelonephritis based on symptoms, especially among elderly patients with septicaemia, might be difficult. In view of the GI tract being a major reservoir for E. coli, the potential exists for bacterial translocation across the GI epithelium. From the findings in this study, it is proposed that, in hospitalized patients with concurrent septicaemia and UTI, the source of translocation may in some cases be the GI tract. This is further supported by the fact that most septicaemic patients in this study were of a higher age group (median age 67 years), who often have a decreased immune system, a factor that has been linked with increased bacterial translocation (Grubeck-Loebenstein, 1997; MacFie et al., 1999; O’Boyle et al., 1998). We have previously shown a higher prevalence of uropathogenic VGs among E. coli isolated from the gut of healthy elderly adults compared with young adults (Vollmerhausen et al., 2011), which may be a risk factor for extra-intestinal infections among hospitalized elderly patients. In the absence of any study characterizing the E. coli population within the GI tract of such uroseptic patients, the translocation of E. coli from the GI tract of patients with urosepsis cannot be fully justified. In conclusion, we found that, whilst the majority of E. coli strains isolated from the blood and urine of the same patient were identical, it is possible that some patients suffered infection by two different E. coli strains in the http://jmm.sgmjournals.org
urinary tract and in their bloodstream. The fact that nonidentical pairs of isolates did not show the ability to translocate HT-29 cells may suggest that either this cell line is not as suitable as other cells such as Caco-2 for assessing translocation or as-yet-unidentified host factors may play a role in translocation from the GI tract in these patients. The fact that adherence to HT-29 cells was not correlated with translocation further suggests that factors other than adhesion are necessary for translocation of uropathogenic E. coli. Based on these results, it is suggested that future studies on the source of bacterial translocation in patients with urosepsis include testing of additional cell lines as well as characterizing E. coli populations of the gut at the time of hospitalization.
REFERENCES Alexander, J. W., Boyce, S. T., Babcock, G. F., Gianotti, L., Peck, M. D., Dunn, D. L., Pyles, T., Childress, C. P. & Ash, S. K. (1990). The process
of microbial translocation. Ann Surg 212, 496–510, discussion 511– 512. Ansaruzzaman, M., Albert, M. J., Nahar, S., Byun, R., Katouli, M., Ku¨hn, I. & Mo¨llby, R. (2000). Clonal groups of enteropathogenic
Escherichia coli isolated in case–control studies of diarrhoea in Bangladesh. J Med Microbiol 49, 177–185. Arthur, M., Johnson, C. E., Rubin, R. H., Arbeit, R. D., Campanelli, C., Kim, C., Steinbach, S., Agarwal, M., Wilkinson, R. & Goldstein, R. (1989). Molecular epidemiology of adhesin and hemolysin virulence
factors among uropathogenic Escherichia coli. Infect Immun 57, 303– 313. Berg, R. D. (1980). Mechanisms confining indigenous bacteria to the gastrointestinal tract. Am J Clin Nutr 33 (Suppl), 2472–2484. Berg, R. D. (1981). Promotion of the translocation of enteric bacteria
from the gastrointestinal tracts of mice by oral treatment with penicillin, clindamycin, or metronidazole. Infect Immun 33, 854–861. Berg, R. D. & Garlington, A. W. (1979). Translocation of certain
indigenous bacteria from the gastrointestinal tract to the mesenteric lymph nodes and other organs in a gnotobiotic mouse model. Infect Immun 23, 403–411. Berry, R. E., Klumpp, D. J. & Schaeffer, A. J. (2009). Urothelial
cultures support intracellular bacterial community formation by uropathogenic Escherichia coli. Infect Immun 77, 2762–2772. Brauner, A., Kaijser, B., Wretlind, B. & Ku¨hn, I. (1991). Characterization
of Escherichia coli isolated in blood, urine and faeces from bacteraemic patients and possible spread of infection. APMIS 99, 381–386. Chen, J. & Griffiths, M. W. (1998). PCR differentiation of Escherichia coli from other Gram-negative bacteria using primers derived from the nucleotide sequences flanking the gene encoding the universal stress protein. Lett Appl Microbiol 27, 369–371. Clermont, O., Bonacorsi, S. & Bingen, E. (2000). Rapid and simple
determination of the Escherichia coli phylogenetic group. Appl Environ Microbiol 66, 4555–4558. Couto, C. R., Oliveira, S. S., Queiroz, M. L. & Freitas-Almeida, A. C. (2007). Interactions of clinical and environmental Aeromonas isolates
with Caco-2 and HT29 intestinal epithelial cells. Lett Appl Microbiol 45, 405–410. Eto, D. S., Jones, T. A., Sundsbak, J. L. & Mulvey, M. A. (2007).
Integrin-mediated host cell invasion by type 1-piliated uropathogenic Escherichia coli. PLoS Pathog 3, e100. 1581
T. L. Vollmerhausen and others Gransden, W. R., Eykyn, S. J., Phillips, I. & Rowe, B. (1990). Bacteremia
due to Escherichia coli: a study of 861 episodes. Rev Infect Dis 12, 1008– 1018. Grey, P. A. & Kirov, S. M. (1993). Adherence to HEp-2 cells and
enteropathogenic potential of Aeromonas spp. Epidemiol Infect 110, 279– 287. Grubeck-Loebenstein, B. (1997). Changes in the aging immune
system. Biologicals 25, 205–208. Guarner, C., Runyon, B. A., Young, S., Heck, M. & Sheikh, M. Y. (1997). Intestinal bacterial overgrowth and bacterial translocation in
cirrhotic rats with ascites. J Hepatol 26, 1372–1378. Hedlund, M., Duan, R. D., Nilsson, A˚., Svensson, M., Karpman, D. & Svanborg, C. (2001). Fimbriae, transmembrane signaling, and cell
activation. J Infect Dis 183 (Suppl 1), S47–S50. Holden, N. J. & Gally, D. L. (2004). Switches, cross-talk and memory in
Escherichia coli adherence. J Med Microbiol 53, 585–593.
MacFie, J., Reddy, B. S., Gatt, M., Jain, P. K., Sowdi, R. & Mitchell, C. J. (2006). Bacterial translocation studied in 927 patients over 13 years.
Br J Surg 93, 87–93. Macutkiewicz, C., Carlson, G., Clark, E., Dobrindt, U., Roberts, I. & Warhurst, G. (2008). Characterisation of Escherichia coli strains
involved in transcytosis across gut epithelial cells exposed to metabolic and inflammatory stress. Microbes Infect 10, 424–431. Maher, S. & McClean, S. (2006). Investigation of the cytotoxicity of
eukaryotic and prokaryotic antimicrobial peptides in intestinal epithelial cells in vitro. Biochem Pharmacol 71, 1289–1298. McNally, A., Alhashash, F., Collins, M., Alqasim, A., Paszckiewicz, K., Weston, V. & Diggle, M. (2013). Genomic analysis of extra-intestinal
pathogenic Escherichia coli urosepsis. Clin Microbiol Infect 19, E328– E334. Moreno, E., Andreu, A., Pigrau, C., Kuskowski, M. A., Johnson, J. R. & Prats, G. (2008). Relationship between Escherichia coli strains causing
Johnson, J. R. (1991). Virulence factors in Escherichia coli urinary
acute cystitis in women and the fecal E. coli population of the host. J Clin Microbiol 46, 2529–2534.
tract infection. Clin Microbiol Rev 4, 80–128.
Moss, M. (2005). Epidemiology of sepsis: race, sex, and chronic
Johnson, J. R. & Stell, A. L. (2000). Extended virulence genotypes of
alcohol abuse. Clin Infect Dis 41 (Suppl. 7), S490–S497.
Escherichia coli strains from patients with urosepsis in relation to phylogeny and host compromise. J Infect Dis 181, 261–272.
Nettelbladt, C. G., Katouli, M., Volpe, A., Bark, T., Muratov, V., Svenberg, T., Mo¨llby, R. & Ljungqvist, O. (1997). Starvation increases
Kai-Larsen, Y., Lu¨thje, P., Chromek, M., Peters, V., Wang, X., Holm, A˚., Ka´das, L., Hedlund, K. O., Johansson, J. & other authors (2010).
the number of coliform bacteria in the caecum and induces bacterial adherence to caecal epithelium in rats. Eur J Surg 163, 135–142.
Uropathogenic Escherichia coli modulates immune responses and its curli fimbriae interact with the antimicrobial peptide LL-37. PLoS Pathog 6, e1001010.
Nettelbladt, C. G., Katouli, M., Bark, T., Svenberg, T., Mo¨llby, R. & Ljungqvist, O. (2000). Evidence of bacterial translocation in fatal
Katayama, M., Xu, D., Specian, R. D. & Deitch, E. A. (1997). Role of
O’Boyle, C. J., MacFie, J., Mitchell, C. J., Johnstone, D., Sagar, P. M. & Sedman, P. C. (1998). Microbiology of bacterial translocation in
bacterial adherence and the mucus barrier on bacterial translocation: effects of protein malnutrition and endotoxin in rats. Ann Surg 225, 317–326. Katouli, M., Bark, T., Ljungqvist, O., Svenberg, T. & Mo¨llby, R. (1994).
Composition and diversity of intestinal coliform flora influence bacterial translocation in rats after hemorrhagic stress. Infect Immun 62, 4768–4774. Katouli, M., Nettebladt, C. G., Muratov, V., Ljungqvist, O., Bark, T., Svenberg, T. & Mo¨llby, R. (1997). Selective translocation of coliform
bacteria adhering to caecal epithelium of rats during catabolic stress. J Med Microbiol 46, 571–578. Ku¨hn, I. & Mo¨llby, R. (1993). The PhP-RS-System: a simple microplate
method for studying coliform bacterial population. J Microbiol Methods 17, 255–259. Leblebicioglu, H., Esen, S. & Turkish Nosocomial Urinary Tract Infection Study Group (2003). Hospital-acquired urinary tract
infections in Turkey: a nationwide multicenter point prevalence study. J Hosp Infect 53, 207–210. Levert, M., Zamfir, O., Clermont, O., Bouvet, O., Lespinats, S., Hipeaux, M. C., Branger, C., Picard, B., Saint-Ruf, C. & other authors (2010). Molecular and evolutionary bases of within-patient genotypic
and phenotypic diversity in Escherichia coli extraintestinal infections. PLoS Pathog 6, e1001125. Ljungdahl, M., Lundholm, M., Katouli, M., Rasmussen, I., Engstrand, L. & Haglund, U. (2000). Bacterial translocation in experimental
shock is dependent on the strains in the intestinal flora. Scand J Gastroenterol 35, 389–397. MacFie, J. (2004). Current status of bacterial translocation as a cause of surgical sepsis. Br Med Bull 71, 1–11.
hemorrhagic pancreatitis. J Trauma 48, 314–315.
humans. Gut 42, 29–35. Parks, R. W., Stuart Cameron, C. H., Gannon, C. D., Pope, C., Diamond, T. & Rowlands, B. J. (2000). Changes in gastrointestinal
morphology associated with obstructive jaundice. J Pathol 192, 526– 532. Ramos, N. L., Saayman, M. L., Chapman, T. A., Tucker, J. R., Smith, H. V., Faoagali, J., Chin, J. C., Brauner, A. & Katouli, M. (2010).
Genetic relatedness and virulence gene profiles of Escherichia coli strains isolated from septicaemic and uroseptic patients. Eur J Clin Microbiol Infect Dis 29, 15–23. Saeedi, B., Ta¨rnberg, M., Gill, H., Ha¨llgren, A., Jonasson, J., Nilsson, L. E., Isaksson, B., Ku¨hn, I. & Hanberger, H. (2005). Phene Plate
(PhP) biochemical fingerprinting. A screening method for epidemiological typing of enterococcal isolates. APMIS 113, 603–612. Sedman, P. C., Macfie, J., Sagar, P., Mitchell, C. J., May, J., ManceyJones, B. & Johnstone, D. (1994). The prevalence of gut translocation
in humans. Gastroenterology 107, 643–649. Sneath, P. H. A. & Sokal, R. R. (editors) (1973). Numerical Taxonomy: the Principles and Practice of Numerical Classification. San Francisco: W. H. Freeman & Co. Snowden, L., Wernbacher, L., Stenzel, D., Tucker, J. R., McKay, D., O’Brien, M. & Katouli, M. (2006). Prevalence of environmental
Aeromonas in South East Queensland, Australia: a study of their interactions with human monolayer Caco-2 cells. J Appl Microbiol 101, 964–975. Troeger, H., Richter, J. F., Beutin, L., Gu¨nzel, D., Dobrindt, U., Epple, H. J., Gitter, A. H., Zeitz, M., Fromm, M. & Schulzke, J. D. (2007). Escherichia coli a-haemolysin induces focal leaks in colonic epithe-
lium: a novel mechanism of bacterial translocation. Cell Microbiol 9, 2530–2540.
MacFie, J., O’Boyle, C., Mitchell, C. J., Buckley, P. M., Johnstone, D. & Sudworth, P. (1999). Gut origin of sepsis: a prospective study inves-
Tullus, K., Ku¨hn, I., Ørskov, I., Ørskov, F. & Mo¨llby, R. (1992). The
tigating associations between bacterial translocation, gastric microflora, and septic morbidity. Gut 45, 223–228.
importance of P and type 1 fimbriae for the persistence of Escherichia coli in the human gut. Epidemiol Infect 108, 415–421.
1582
Journal of Medical Microbiology 63
E. coli translocation in urosepsis
Vollmerhausen, T. L., Ramos, N. L., Gu¨ndogdu, A., Robinson, W., Brauner, A. & Katouli, M. (2011). Population structure and
uropathogenic virulence-associated genes of faecal Escherichia coli from healthy young and elderly adults. J Med Microbiol 60, 574–581.
Xu, D., Lu, Q. & Deitch, E. A. (1998). Elemental diet-induced bacterial
translocation associated with systemic and intestinal immune suppression. J Parenter Enteral Nutr 22, 37–41.
Wilson, K. H. & Blitchington, R. B. (1996). Human colonic biota
Yamamoto, S., Tsukamoto, T., Terai, A., Kurazono, H., Takeda, Y. & Yoshida, O. (1997). Genetic evidence supporting the fecal-perineal-
studied by ribosomal DNA sequence analysis. Appl Environ Microbiol 62, 2273–2278.
urethral hypothesis in cystitis caused by Escherichia coli. J Urol 157, 1127–1129.
Wold, A. E., Caugant, D. A., Lidin-Janson, G., de Man, P. & Svanborg, C. (1992). Resident colonic Escherichia coli strains frequently display
Yu, P. & Martin, C. M. (2000). Increased gut permeability and bacterial translocation in Pseudomonas pneumonia-induced sepsis. Crit Care Med 28, 2573–2577.
uropathogenic characteristics. J Infect Dis 165, 46–52.
http://jmm.sgmjournals.org
1583
