Highly Resistant Metabolically Deficient Dwarf Mutant of Escherichia coli Is the Cause of a Chronic Urinary Tract Infection

Abstract

We present the case of a 68-year-old diabetic woman who has been suffering from chronic urinary tract infections, recurring over a period of at least 5 years, caused by a slowly growing metabolically deficient dwarf mutant (MDD) of Escherichia coli. This MDD strain was auxotrophic for histidine, was resistant to multiple antibiotics, and showed atypical growth behavior. Colonies were tiny on routine media but were able to revert to normal growth after extended incubation. This strain was identified as E. coli by 16S ribosomal DNA sequencing, and virulence factor profiles were determined by PCR. Seven MDD isolates collected over the 5-year period were grown from midstream urine to significant colony counts and shown to belong to the same clonal group by pulsed-field gel electrophoresis and enterobacterial repetitive intergenic consensus PCR. These MDDs were repeatedly misidentified by biochemical methods due to their slow growth and atypical colony morphology. This case highlights the importance of recognizing MDDs of Enterobacteriaceae in patients with chronic infections. To our knowledge this is the first report of an MDD of E. coli causing a chronic urinary tract infection.

Urinary tract infections (UTIs) are among the most common infections in women, with uropathogenic Escherichia coli (UPEC) being the predominant pathogen (10, 28). Underlying host factors such as diabetes predispose individuals to UTIs with a more diverse etiology, including other Enterobacteriaceae, Pseudomonas aeruginosa, group B streptococci, Candida spp., and Enterococcus spp. (24). Recurrence of UTI is common in women (7, 12, 23), and the most effective treatment of these relapsing infections is prophylactic antibiotic treatment, but when antibiotics are discontinued, the UTI usually returns (27). Most recurrences are thought to be due to reinfection with the initially infecting strain, which may persist in the fecal flora after elimination from the urinary tract (9, 25), but it has also been suggested that the urinary tract can serve as a reservoir for recurring infections (6). Recent studies show that E. coli is able to invade bladder epithelial cells (15, 19) and replicate intracellularly forming large bacterial inclusions (20). Type 1 fimbriae and extracellular polysaccharides of UPEC are critical for uropathogenesis (1), and several virulence factors, including papA (adhesin), papG (adhesin), iha (iron-regulated gene homologue adhesin), and iutA (aerobactin), have been associated with multiple same-strain recurrences (13).

Wild-type E. coli is a prototroph that carries all genes necessary for synthesis of amino acids, vitamins, and growth factors. Metabolically deficient E. coli carrying mutations in amino acid biosynthesis have, however, been isolated from clinical specimens (2, 3, 16, 17, 29, 32). It has been reported that 1 to 5% of all UTIs might be caused by these metabolically deficient dwarf mutants (MDDs) (2, 16). Most MDDs described so far are auxotrophic for cysteine and have been reported to cause UTIs in elderly and debilitated patients (2-4). They are not known to cause chronic infections, and antibiotic resistance has not been associated with them.

CASE REPORT

We present the case of a 68-year-old woman, suffering from recurring urinary tract infections over a period of 7 years. The patient's past medical history revealed insulin-dependent diabetes mellitus (IDDM) since childhood, complicated by hypercholesteremia, hypertension, small-vessel disease with myocardial infarction in 1991, nephropathy, retinopathy, and polyneuropathy. Recurrent UTIs began in 1995 with our patient repeatedly complaining of symptoms typical of cystitis such as urgency, dysuria, and increased frequency. The presence of pyuria and significant bacterial colony counts (>10⁵/ml) in midstream urine specimens repeatedly confirmed the diagnosis. Searches for urinary tract abnormalities, including ultrasound, computed tomography scan, and intravenous urography, remained unremarkable. In August 1997, the patient was hospitalized at our University Medical Center for pyelonephritis and urosepsis. Since then she has been seen on a regular basis in our outpatient clinic for IDDM and recurrent UTIs. Several urine samples were examined each year at our bacteriological laboratory.

Between 1995 and 1997, UTIs were treated with various antibiotics (among them, documented 5-day courses each of trimethoprim-sulfamethoxazole, ampicillin, and ofloxacin), but UTI recurred shortly after the antibiotics were discontinued. In 1997 urosepsis was successfully treated with a 10-day course of intravenous meropenem; however, a few weeks later our patient presented to our outpatient clinic for another relapse of UTI. Therefore, she was put on nitrofurantoin (100 mg/day) for prophylactic treatment, but she has nevertheless suffered from several recurrences since then.

E. coli isolates from this patient repeatedly cultured between 1995 and 1997 (isolates not preserved) were initially susceptible to ampicillin, quinolones, trimethoprim-sulfamethoxazole, and tetracycline. Between 1997 and 2002, eight urine samples were sent to our laboratory for bacteriological examination. Initially, >10⁵ CFU was cultured and identified as E. coli by the API 20E system (isolate not preserved). They were resistant to ampicillin, piperacillin, tetracycline, trimethoprim-sulfamethoxazole, and quinolones. In September 1997, >10⁵ CFU of a slowly growing, lactose-negative, gram-negative rod was recovered (isolate 1111/97) for the first time. Due to weak biochemical reactivity, this isolate was first misidentified as Hafnia alvei by the API 20E system. However, 16S ribosomal DNA (rDNA) sequencing allowed its correct classification as a metabolically aberrant E. coli strain. This isolate was resistant to aminoglycosides in addition to the antibiotics named above. Since September 1997 several relapses of cystitis with phenotypically very similar isolates occurred (6/1999, 5/2001, 6/2001, and 9/2002). In August and December 2001, similar isolates were also recovered from urine and stool samples during an asymptomatic interval (2957/01, 4000/01, and 4001/01) (Table 1).

TABLE 1.

Characteristics of MDDs

Isolate no.	Source	Date of isolation (mo/day/yr)	Symptomatic UTI	Growth on M9 plus:		Virulence factor		API 20E code
				Histidine	Pool of 12 amino acidsa	iutA	FimH
1111/97	Urine	9/3/1997	+	+	+	+	+	5.004.000 (dwarf)
1448/01	Urine	5/2/2001	+	+	+	−	+	5.004.100 (dwarf)
								5.044.552 (revertant)
1852/01	Urine	6/5/2001	+	−	+	−	+	NDb
2957/01	Urine	8/30/2001	−	−	+	−	+	ND
3332/01	Urine	9/27/2001	+	−	+	−	+	ND
4000/01	Urine	12/19/2001	−	−	+	−	+	ND
4001/01	Stool	12/19/2001	−	−	+	−	+	ND
3236/02	Urine	9/24/2002	+	−	+	−	+	4.004.000 (dwarf)
								5.044.550 (revertant)

^al-Amino acids, including alanine, cysteine, histidine, isoleucine, leucine, lysine, methionine, phenylalanine, threonine, tryptophan, valine, tyrosine.

^bND, not determined.

MATERIALS AND METHODS

Bacterial identification.

Bacterial isolates 1111/97, 1448/01, 1852/01, 2957/01, 3332/01, 4000/01, and 3236/02 were recovered from cleanly voided midstream urine (isolation dates are given in Table 1). Specimens were cultured primarily on plates containing MacConkey and Columbia agar supplemented with 5% sheep blood (Difco Laboratories GmbH, Augsburg, Germany) at 37°C. Strain 4001/01 was isolated from a stool specimen. Auxotrophy for amino acids was tested on M9 medium (Sigma Aldrich Chemie GmbH, Taufkirchen, Germany) supplemented with 0.2% glucose and 1 mM MgSO₄. l-Amino acids (alanine, cysteine, histidine, isoleucine, leucine, lysine, methionine, phenylalanine, threonine, tryptophan, valine, and tyrosine) (Sigma Aldrich Chemie GmbH) were added at a concentration of 40 mg/ml or used to impregnate disks. Hemin chloride (Sigma Aldrich Chemie GmbH) in 10 mM NaOH was used at 25 μg/ml; thymidine (Fluka Chemie, Neu-Ulm, Germany) and thiamine hydroxide (Sigma Aldrich Chemie GmbH) were used at a concentration of 10 μg/ml. MEM amino acids (l-arginine-HCl, l-cysteine, l-histidine-HCl, l-isoleucine, l-leucine, l-lysine-HCl, l-methionine, l-phenylalanine, l-threonine, l-tryptophan, l-valine, and l-tyrosine) were purchased from Invitrogen (Karlsruhe, Germany). Bacteria were identified with the API 20E system (bioMérieux Ltd., Marcy l'Etoile, France). Serological typing of MDD isolates was performed by slide agglutination with polyvalent and monovalent anti-E. coli O and H antisera at the E. coli Reference Center (Hygiene Institute Hamburg, Hamburg, Germany).

Nucleic acid manipulations.

Analysis by enterobacterial repetitive intergenic consensus (ERIC) PCR was performed as previously described (31). Amplification and sequencing of the gene encoding 16S rRNA was performed as described previously (5) using the BigDye Terminator Cycle sequencing kit (Applied Biosystems Warrington, United Kingdom) and the ABI Prism 377XL DNA sequencer (Applied Biosystems GmbH, Darmstadt, Germany). Sequence data for comparison were obtained from GenBank (http://www.ncbi.nih.gov). UPEC virulence factor multiplex PCR was performed with primer mix I (hlyA, pathogenicity-associated island [PAI], fimH, sfa/focDE, iutA, ibeA) and primer mix II (hlyA, PAI, afa/draBC, cnf1, focG, sfaS) as previously described (14). Genomic DNA preparation and pulsed-field gel electrophoresis (PFGE) was described previously (25). Restriction enzyme cleavage was performed with XbaI or SpeI (Fermentas, St. Leon-Rot, Germany). PFGE was performed with the CHEF-DRIII system (Bio-Rad) in1% (wt/vol) agarose gels (FMC Incert agarose; Biozym, Hamburg, Germany).

Antibiotic susceptibility testing.

Antibiotic susceptibility was determined on Mueller-Hinton agar by disk diffusion following the guidelines of the National Committee for Clinical Laboratory Standards (20a) with the antibiotics (Oxoid Ltd., Basingstoke, United Kingdom) listed in Table 2. The MICs for several isolates were determined by E-test according to instructions of the manufacturer (Viva Diagnostika, Cologne, Germany). MICs for the dwarf colonies were determined after an extended incubation period of 48 h due to growth deficiency on Mueller-Hinton agar.

TABLE 2.

Antibiotic susceptibility profiles of MDDs

Antibiotic(s)	MICa (μg/ml) for:
	1111/97		3236/02
	Dwarf	Revertant	Dwarf	Revertant
Ampicillin	>256	>256	>256	>256
Amoxicillin-clavulanate	5.0	5.0	10	10
Piperacillin	>256	>256	>256	>256
Piperacillin-tazobactam	3.0	2.0	8.0	4.0
Meropenem	0.006	0.006	0.032	0.032
Cefuroxime	3.5	3.5	8.0	6.0
Ceftazidime	<0.5	<0.5	<0.5	<0.5
Cefotaxime	<0.25	<0.25	<0.25	<0.25
Ceftriaxone	0.023	0.023	0.094	0.064
Tetracycline	>256	>256	>256	>256
Gentamicin	12	8	32	24
Tobramycin	32	24	>256	>256
Trimethoprim- sulfamethoxazole	>32	>32	>32	>32
Levofloxacin	>32	>32	>32	>32
Ciprofloxacin	>32	>32	>32	>32

^aMICs were determined by E-test (Viva Diagnostika).

Determination of reversion frequency.

The frequency of reversion to normal growth was determined by resuspending a single small colony grown on MacConkey agar for 48 h in phosphate-buffered saline and culturing serial dilutions of this suspension on MacConkey agar plates at 37°C. After 48 h small and large colony types were counted and the reversion frequency was calculated.

RESULTS

Characterization of MDDs.

All eight MDD isolates exhibited typical growth behavior on MacConkey agar. After 24 h of incubation, colonies appeared tiny and lactose negative. After 48 h of incubation several larger lactose-positive colonies (revertants) appeared. All strains were cytochrome oxidase negative and positive for glucose fermentation. Gram stain showed gram-negative rods. Biochemical analysis with the API 20E system repeatedly showed biochemically weakly reactive bacteria that were misidentified as either diverse Yersinia, Burkholderia, Shigella, or Hafnia strains with unacceptable profiles (Table 1). Sequencing of 16S rDNA was performed with both colony types. In all cases 16S rDNA sequences (>400 nucleotides) were identical and matched the 16S rDNA sequence of E. coli (AE000474) 100%. Therefore, these isolates represent atypically growing mutants of E. coli. A closer look at the growth behavior revealed two types of isolates: early isolates (1111/97 and 1448/01) and late isolates (1852/01, 2957/01, 3332/01, 4000/01, 4001/01, and 3236/02). Colonies of the early isolates grew on MacConkey, Mueller-Hinton, and Columbia agars to diameters of 0.9, 1.2, and 2.7 mm, respectively, after 24 h. Larger revertants grew to diameters of 2.7 mm (MacConkey), 2.0 mm (Mueller-Hinton), and 3.5 mm (Columbia agar) (Fig. 1). The late isolates grew even more slowly with colonies of isolate 3236/02, reaching diameters of only 0.7 mm (MacConkey), 0.95 mm (Mueller-Hinton), and 1.8 mm (Columbia agar) (Fig. 2) These bacteria grew with either a smooth round convex phenotype or with a flat spreading “fried-egg” appearance (Fig. 2). The reversion frequency of several isolates was determined and varied between 1.1 × 10⁻⁵ for 1111/97 and 8 × 10⁻⁷ for 3236/02.

FIG. 1.

MDD E. coli isolate 1448/01 and revertants growing on MacConkey agar (B) and at a magnification of ×10 (A) after 48 h incubation at 37°C.

FIG. 2.

MDD E. coli isolate 3236/02 growing on MacConkey agar, showing both smooth round and “fried egg” phenotypes (magnification, ×10).

All MDDs represent a single strain.

The typical growth behavior of the MDDs already suggested that all isolates represent one strain. To verify this assumption, we performed ERIC PCR, which revealed identical patterns for all MDDs (results not shown). Furthermore, PFGE was performed on genomic DNA of MDDs cleaved with either XbaI or SpeI (Fig. 3). After XbaI digestion, isolates 1111/97, 1448/01, 3332/01, 4000/01, and 3236/02 showed identical patterns, whereas isolates 1852/01 and 2957/01 differed from these by only one cleavage site. Similarly, after SpeI digestion, isolates 1852/01, 2957/01, 4000/01, and 3236/02 showed identical patterns whereas isolates 3332/01, 1111/97, and 1448/01 differed from the other isolates by just one cleavage site. These changes were consistent with a single genetic event and allowed the classification of all isolates as clonal according to the criteria of Tenover et al. (30).

FIG. 3.

PFGE analysis of MDD isolates 1111/97 (lanes 1), 1448/01(lanes 2), 1852/01 (lanes 3), 2957/01 (lanes 4), 3332/01 (lanes 5), 4000/01 (lanes 6), and 3236/02 (lanes 7) (A) and isolate 3236/02 (lanes 7) and control strains E. coli clinical isolates (lanes C1 and C2) (B).

Auxotrophy of MDDs.

Auxotrophy of the MDD E. coli isolates was tested on M9 medium supplemented with glucose and amino acids. The early isolates 1111/97 and 1448/01 were auxotrophic for histidine only. The late isolates 1852/01, 2957/01, 3332/01, 4000/01, 4001/01, and 3236/02 grew on M9 medium only in the presence of a commercially available mixture of amino acids (MEM amino acids; Invitrogen) (Table 1). Attempts to determine the exact requirements were not successful, and it seems that at least three amino acids are necessary to stimulate growth of the late isolates. To determine if histidine auxotrophy was due to a large deletion in the histidine operon, we performed PCR analysis of hisB, hisIE, and hisG as well as adjacent wzzb, ugd, and gnd genes. We were, however, able to amplify DNA sequences within each of these genes, indicating that histidine auxotrophy is not due to a large deletion of this operon. Furthermore, MDDs were not auxotrophic for hemin, menadione, thiamine, or thymidine, indicating that the MDD phenotype was not due to one of the previously described mechanisms, interruption of electron transport chain (classic small colony variant [SCV]) or thymidine dependence (8, 22). Next we determined whether urine from our patient could support growth of MDDs on M9 agar. This was the case at least for isolates 1111/97 and 1448/01, whose growth defect on M9 agar could be complemented with sterile filtered urine from our patient. The other isolates were not able to grow around disks impregnated with sterile urine.

Characterization of serotype and pathotype.

To determine whether our isolates represent uropathogenic E. coli, we determined serotypes and virulence factor genotypes of several isolates. Isolates 1448/01 and 1852/01 belong to serotype O142:H−. This unusual serotype is not known for UPEC but has been described for enteropathogenic E. coli (21). To further characterize MDD isolates, we performed a UPEC virulence factor multiplex PCR (14). This revealed that all isolates harbor fimH (type 1 fimbrial adhesin). Only the oldest isolate, 1111/97, also showed a PCR product representing iutA (aerobactin). PCR was negative for the following virulence factors: sfaS (S fimbrial adhesin), hlyA (hemolysin), cnf1 (cytotoxic necrotizing factor), ibeA (invasion of brain endothelium) sfa/focDE (S and F1C fimbriae), focG (F1C fimbriae adhesin), afa/dra (DR antigen binding adhesin), and PAI.

Antimicrobial susceptibility testing.

All isolates were resistant to ampicillin, piperacillin, tobramycin, trimethoprim-sulfamethoxazole, tetracycline, levofloxacin, and ciprofloxacin. All isolates were susceptible to ceftazidime, cefotaxime, ceftriaxone, meropenem, and amoxicillin-clavulanate. Isolate 1111/97 was susceptible to cefuroxime, whereas the recent isolates were intermediate. Isolate 1111/97 was intermediate to gentamicin, whereas the recent isolates were resistant. For piperacillin-tazobactam, gentamicin, and tobramycin, the MICs for large colony revertants were slightly lower than those for the dwarfs. Furthermore, the MICs for recent isolates generally seem to be higher than those for older isolates (Table 2).

DISCUSSION

It is well known that auxotrophic E. coli is able to cause UTIs. Most of these infections are caused by cysteine-requiring E. coli (2, 16, 18) and occur mainly in older and debilitated patients (2, 16, 29, 32), but these strains have not been associated with antibiotic resistance and are not known to cause chronic infections (17). In this communication we describe the first case of a chronic UTI caused by an MDD E. coli. The same strain was repeatedly isolated in significant amounts from midstream urine from a 68-year-old diabetic patient over a period of 5 years. All isolates collected during this period were shown to be clonal by ERIC PCR and PFGE. This MDD E. coli strain grows slowly on routine media but has the ability to revert to normal growth after an extended incubation time, giving the appearance of a mixed culture. The early isolates of our MDDs were auxotrophic for only histidine, whereas the recent isolates were auxotrophic for multiple amino acids, indicating that the latter harbor a more complex genetic defect than the former. They were not auxotrophic for hemin, menadione, thiamine, or thymidine, indicating that these isolates are not SCVs, which are known to cause chronic infections (9, 26). Our strain developed resistance to multiple antibiotics over the course of several years, with the MICs for the dwarf colonies generally being slightly higher than those for the revertants. Also, the recent isolates, which grow more slowly than the early isolates, are generally more resistant, a characteristic which might be attributed to their reduced metabolism.

Not only were MDDs isolated from our patient during symptomatic UTI, but they could also be cultured during asymptomatic intervals. This indicates that the urinary tract might be a reservoir for recurring UTIs. Possibly MDDs could persist intracellularly in bladder mucosa, as has been shown for type 1-piliated E. coli (25). An intracellular location may protect these bacteria from the immune response of the host and antibiotic treatment. On the other hand, we were able to detect the same MDD in the fecal flora of our patient, indicating that colonization of the gut may also be responsible for recurring infections of the urinary tract. Urine is an adequate growth medium for UPEC, but several growth factors such as guanine, arginine, and glutamine are not abundant and must be synthesized de novo for optimal growth (11, 26). Histidine, nicotinamide, tryptophan, cysteine, and adenine are present in high concentrations in urine, and auxotrophs for these factors are able to readily grow in urine (11). In line with this, the early histidine-auxotrophic MDDs are able to grow on M9 medium supplemented with sterile urine from our patient. Surprisingly, however, growth of the recent MDDs is not supported by the urine from our patient despite the high colony counts that can be cultured from the same urine.

This MDD is not a typical UPEC according to serotype and virulence factor profile. Other MDDs (mostly cysteine auxotrophs) described so far have also been reported not to express the typical virulence factors of UPEC (18). Our dwarfs harbor only the gene encoding type 1 fimbrial adhesin (fimH), which is the preeminent virulence factor of E. coli and has been shown to be necessary for bacterial invasion of bladder epithelial cells (1, 20). Furthermore, the earliest isolate also harbors the gene encoding aerobactin (iutA), which has been associated with multiple same-strain recurrences (13). Other virulence factors typical of UPEC were, however, lacking, which suggests that infections by these bacteria are due to decreased resistance of the host rather than a pathogenic potential of MDDs. This is supported by the fact that our patient suffers from longstanding IDDM and that our MDD strain lost a virulence factor (iutA) between 1997 and 2001.

Our MDDs have probably colonized their host for a very long period and adapted their metabolism to the nutrients provided by the host. This could have led to a loss of biosynthetic genes that were no longer required while colonizing the host and might be responsible for the unusual colony morphology and slow growth of these bacteria. Diagnostic laboratories can easily misidentify or even overlook these MDDs. In the present case an MDD E. coli was repeatedly misidentified due to atypical appearance and weak biochemical reactivity. Therefore, this case highlights the importance of being aware of the presence and clinical significance of not only SCVs but also MDDs of enterobacteria in chronic infections.