Case Report

Top Case Report Methods Results Discussion Author & Article Info References

A 46-year-old woman was hospitalized on 10 July 1992 for acute cholecystitis, cholangitis, and fulminant pancreatitis. Her initial course was complicated by sepsis and adult respiratory distress syndrome requiring mechanical ventilation. She received broad-spectrum antimicrobial therapy and had several operations for intra-abdominal abscesses. On day 30 of hospitalization, she was transferred to the Surgical Intensive Care Unit at Thomas Jefferson University Hospital. Subsequently, nosocomial pneumonias developed and the patient needed prolonged mechanical ventilation. She also required several open and percutaneous drainage procedures, indwelling vascular and urinary drainage catheters, and prolonged broad-spectrum antimicrobial therapy that included vancomycin from hospital day 14 through day 61 and from day 74 through day 150 (Figure 1).

View larger version (15K): [in this window] [in a new window]   Figure 1. Summary of antibiotic exposures and microbiologic data for this patient. Day of hospitalization is shown on the horizontal axis. Open circles represent urine cultures that grew vancomycin-dependent Enterococcus faecalis. Also shown are cultures that grew nondependent E. faecalis isolates that were intermediately susceptible or resistant to vancomycin, including urine (\#9679;), blood (\#9670;), central venous catheter tip (\#9671;), stool (), wound or drainage (), abscess (+), and respiratory (x) cultures.

Beginning on day 68 of hospitalization, vancomycin-resistant E. faecalis was cultured from several sites (Figure 1). Ampicillin therapy was begun on hospital day 79 but discontinued on day 107 because of a severe rash; vancomycin and other antimicrobial agents were continued. Cultures from several sites continued to grow vancomycin-resistant enterococci, but no additional specific therapy for enterococcus infection was initiated despite persistent leukocytosis and low-grade fevers.

A urine specimen collected on hospital day 73 yielded vancomycin-resistant enterococci. The results of subsequent urine cultures, including culture of a specimen collected on day 79, were negative, although urinalysis revealed pyuria and bacteriuria. Gram stains showed elongated, gram-positive cocci in chains. When reviewed in the clinical microbiology laboratory, culture specimens yielded a few colonies of growth on primary plating onto blood agar, but colonies did not grow on subculture, suggesting a fastidious or nutritionally deficient organism. Ultimately we found that the nutritional requirement was specifically for the glycopeptide antimicrobial agent vancomycin. Subsequently, we isolated the vancomycin-dependent E. faecalis organism from five urine cultures collected from hospital day 79 through day 145.

On hospital day 145, bacteremia and sepsis with vancomycin-resistant E. faecalis developed in the patient. We discontinued vancomycin therapy and began therapy with imipenem, with subsequent sterilization of blood cultures, clearing of enterococci from urine and other sites, and resolution of fever and pyuria. Subsequent urine specimens did not grow either vancomycin-resistant or -dependent organisms.

Methods

Top Case Report Methods Results Discussion Author & Article Info References

Growth and Identification of Bacterial Isolates

Ten-microliter aliquots of urine specimens were plated on blood agar and incubated at 37 °C overnight. Subsequently, we subcultured single colonies from blood plates onto various enriched laboratory media with and without vancomycin supplementation. We identified bacterial colonies using the RapID STR (Innovative Diagnostic Systems, Inc., Atlanta, Georgia) and API 20S (Analytab Products, Plainview, New York) systems according to protocols provided by the manufacturer and using the AutoScan 4 version of the MicroScan system (Baxter Healthcare Corporation, West Sacramento, California). We did susceptibility testing using the MicroScan system and by standard Bauer-Kirby disc-diffusion methods.

Determination of Nutritional Requirements of the TJ310 Strain

We placed substrate-impregnated paper discs on antibiotic-free Mueller-Hinton or brain-heart infusion agar plates seeded with a suspension of TJ310 strain with turbidity equivalent to a 0.5 MacFarland standard. We incubated plates at 37 °C and monitored them for growth around the discs. For experiments in broth culture, we added substrates to test tubes containing 1 mL of a 10⁶ colony-forming unit per milliliter inoculum of strain TJ310 in antibiotic-free brain-heart infusion, and we monitored cultures for visible growth.

Plasmid and Total Genomic DNA Analysis

We extracted plasmid DNA from strains using the alkaline lysis method [4], digested them with the restriction enzyme Hind III, and analyzed them using gel electrophoresis in 0.8% agarose. Daniel Sahm, PhD (Washington University, St. Louis, Missouri), compared total genomic DNA using pulsed-field gel electrophoresis on strains we submitted to him in a blinded manner. We extracted genomic DNA [5], digested it with the restriction enzyme Sma I, and performed electrophoresis on a 1% agarose gel using contour-clamped homogeneous electric fields (CHEF-DRII, Bio-Rad Laboratories, Richmond, California). The pulse time was ramped from 1 to 20 seconds during 21 hours at 200 V.

Polymerase Chain Reaction for the VanB Gene

We performed the polymerase chain reaction (PCR) using oligonucleotides 5'GCTCCGCAGCTTGCATGGACA 3' (primer 1) and 5'ACGATGCCGCCATCCTCCTGC-3' (primer 2), which we chose to amplify a 529 base-pair internal fragment of the vanB vancomycin-resistance gene [6]. We performed PCR directly on 1 µL of overnight broth cultures of each strain grown in the brain-heart infusion. The PCR reaction mixture consisted of 0.6 µg each of primers 1 and 2 plus 200 µmol/L deoxynucleotide triphosphates, 2.5 mmol/L MgCl₂, and 1X PCR buffer (Perkin Elmer, Norwalk, Connecticut) in a total volume of 80 µL; we added 1 µL Taq DNA polymerase (Promega, Madison, Wisconsin) to begin amplification. We used a Perkin-Elmer Cetus Model 480 DNA thermocycler (Norwalk, Connecticut) to assay samples for 30 cycles, each consisting of denaturation at 95 °C for 1 minute, annealing at 55 °C for 2 minutes, and polymerization at 75 °C for 2 minutes. We visualized PCR products using ethidium bromide on 2% agarose gels.

Restriction Digests of vanB Polymerase Chain Reaction Products

We digested 10-µL aliquots of the PCR products containing the amplified vanB fragments overnight with 1 unit of the restriction enzyme DdeI (Boehringer-Mannheim, Indianapolis, Indiana) and visualized them using 2% agarose gels.

Results

Top Case Report Methods Results Discussion Author & Article Info References

We plated 10-µL aliquots of sterile-collected urine on blood agar and incubated them at 37 °C. After 24 hours, we observed a few gray colonies. Colonies were nonhemolytic and catalase negative. Gram stain showed elongated, gram-positive cocci in chains. We subcultured colonies but observed no growth after incubation in aerobic, anaerobic, or 5% CO₂-enriched atmosphere on various laboratory media, including blood and chocolate agar. We identified the isolate as an E. faecalis organism using biochemical profiles from the RapID STR and API 20S systems, which are growth-independent rapid identification panels. However, growth was inadequate for identification or susceptibility testing using the AutoScan 4 version of the MicroScan system. We performed direct susceptibility testing by disc diffusion on colonies isolated from the initial plates. After incubation at 37 °C for 24 hours, we observed growth only around the 30-µgrams vancomycin-impregnated disc, suggesting a growth requirement for vancomycin Figure 2, top).

View larger version (91K): [in this window] [in a new window]   Figure 2. Growth dependence of strain TJ310. Top. Thirty-µgram vancomycin (V) and teicoplanin (T) discs were placed on a brain-heart infusion agar plate seeded with a suspension of strain TJ310. After 48 hours of incubation, growth was observed only surrounding the vancomycin disc. Bottom. Discs impregnated with 50 µmol/LD-alanine (DA), 50 µmol/LL-alanine (LA), 50 µmol/LD-alanine-d-alanine (DA-DA), and 30 µg vancomycin were placed on a brain-heart infusion agar plate seeded with a suspension of strain TJ310. After 48 hours of incubation, growth was observed only surrounding the D-alanine-alanine and vancomycin discs.

To confirm the apparent vancomycin dependence of the isolate, we inoculated colonies from the primary plate in parallel onto chocolate agar and Thayer Martin agar (chocolate agar containing 3 µg/mL of vancomycin, colistin, and nystatin) and also onto brain-heart infusion agar with and without vancomycin (10 µg/mL). In each case, growth occurred only on vancomycin-containing media. After we added vancomycin to culture media, we were able to perform identification and susceptibility testing using the MicroScan system. We also did susceptibility testing using disc diffusion on Thayer Martin agar. We identified the organism using the MicroScan system as an E. faecalis (biotype number 67 743 654) and found that it was susceptible to penicillin, ampicillin, imipenem, and tetracycline but resistant to ciprofloxacin and vancomycin; we also found high-level resistance to gentamicin.

In broth culture, addition of vancomycin at concentrations of 0.25 µg/mL to 1024 µg/mL and addition of the closely related glycopeptide ristocetin sustained growth of the vancomycin-dependent strain, designated strain TJ310. However, we detected no growth after addition of the investigational glycopeptide compound teicoplanin or any other antimicrobial agents. The time to initiation of growth in broth cultures after addition of vancomycin depended on the concentration of vancomycin added, with most rapid onset of growth occurring after addition of 16 to 64 µg/mL of vancomycin. In addition to antimicrobial agents, we also screened other substrates for their ability to support growth of strain TJ310. We observed growth surrounding a disc containing 50 µmol/L of the dipeptide D-alanine-d-alanine Figure 2, bottom); this was the only compound other than the glycopeptide agents that supported growth of this strain.

To investigate the potential derivation of the vancomycin-dependent strain, we compared this isolate with two vancomycin-resistant but nondependent enterococcal isolates previously collected from the same patient. These included strain TJ282, cultured from an intra-abdominal abscess on hospital day 70, and strain TJ291, cultured from biliary drainage fluid on hospital day 67. Strains TJ282 and TJ291 each had a minimum inhibitory concentration to vancomycin of 16 µg/mL, and both strains were susceptible to teicoplanin. All isolates had the same biotypes when identified using the MicroScan system and had identical antimicrobial susceptibility profiles. We compared profiles of plasmid DNA extracted from all three strains and they appeared to be identical. We also compared total genomic DNA from these three strains using pulsed-field gel electrophoresis. Figure 3(top) shows pulsed-field patterns of DNA from strains TJ282, TJ291, TJ310, and several unrelated vancomycin-resistant E. faecalis strains. We interpreted the patterns for strains TJ310, TJ282, and TJ291 as highly related, with a coefficient of similarity (number of common bands x 2 x 100 divided by the total number of bands) of more than 90%.

View larger version (121K): [in this window] [in a new window]   Figure 3. Comparison of TJ310 with vancomycin-resistant but nondependent isolates from the same patient. Top. Pulsed-field electrophoresis patterns of Hind III digests of chromosomal DNA from E. faecalis TJ310 (lane 3) and two vancomycin-resistant nondependent E. faecalis isolates from the same patient, TJ282 (lane 2) and TJ291 (lane 5). Also shown are the unrelated vanB E. faecalis isolates V583 (lane 1) and vanA E. faecalis isolates TJ153 (lane 4) and TJ337 (lane 6). Bottom. DdeI digest products of vanB fragments amplified by PCR using internal primers for the vanB gene were visualized on a 2% agarose gel. The patterns of TJ310, TJ282, and TJ291 (lanes 4 to 6) were identical and differed from those of the vanB2 strains SF296 and WB319 (lanes 1 and 2) and the vanB1 strain V583 (lane 3).

Polymerase chain reaction is a sensitive and specific technique for determining the genotype of vancomycin-resistant enterococci [7]. Using PCR, we found that strains TJ282, TJ291, and TJ310 contained the vanB gene. We determined the amino acid sequences of the amplified vanB PCR fragments of TJ282 and TJ310 and found that they were identical to each other and showed 99% homology to the published sequence of the vanB2 variant of the vanB gene [8], but this sequence differed from the published sequence by a C-to-A substitution at base-pair 447 that altered a DdeI restriction site. DdeI restriction digest patterns of the amplified vanB fragment from TJ282, TJ291, and TJ310 were identical and differed from those of unrelated vanB1 and vanB2 strains Figure 3, bottom). This unique vanB restriction digest pattern has been observed only in a cluster of E. faecalis isolates from several hospitals in Philadelphia (Fraimow H. unpublished data) and further supports the close relation of TJ310 to the vancomycin-resistant, nondependent isolates from the same patient.

Discussion

Top Case Report Methods Results Discussion Author & Article Info References

We described the first clinical isolate of a bacterial strain that requires an antimicrobial agent to grow. Human pathogens with unusual nutritional requirements are well represented in the medical literature [1]. Antibiotic-dependent organisms, such as streptomycin-dependent strains of Escherichia coli, sometimes have been constructed or selected in the research laboratory, but no analogous strains have been reported from clinical material [9]. Glycopeptide antimicrobial agents would be unlikely candidates for nutritional factors because of their large and complex structure, a unique mechanism of action that does not rely on mimicry of known biochemical substrates, and their lack of intracellular penetration or metabolism [10, 11].

The ability of strain TJ310 to grow after direct plating of urine specimens onto standard media but not on subculture indicated that some component of the urine specimen was required for growth. This potential growth factor could have been a component of normal human urine or some substance unique to this patient's circumstance, such as a drug or drug metabolite. Subsequent growth of TJ310 only around a vancomycin-impregnated disc on direct susceptibility testing provided the critical clue to the unusual growth requirement. We took great care to verify the specificity of the growth requirement. Vancomycin from various sources provided growth supplementation, excluding the possibility that the growth factor was an impurity or contaminant of the vancomycin preparations. The concentration of vancomycin in the urine specimen from which strain TJ310 was recovered was determined by bioassay to be greater than 1500 µg/mL; transfer of small quantities of vancomycin with the initially plated urine specimen adequately explains the growth of a few colonies on the primary culture plates. Because of the difficulty in detecting this unusual requirement, the true prevalence of the vancomycin-dependent phenotype is unknown. Information on a vancomycin-dependent E. faecium isolate recovered from blood culture of a patient has been presented in part [12], and another similar strain has been isolated from a urine specimen from another institution (Kostman J. Personal communication).

Vancomycin-resistant enterococci are important nosocomial pathogens throughout the United States [2, 3, 6]. The patient we have described became infected during an intensive care unit outbreak of vancomycin-resistant E. faecalis. During that time, she required continuous treatment with intravenous vancomycin but never received adequate therapy for the E. faecalis. Comparison of strain TJ310 with other vancomycin-resistant but nondependent isolates from the same patient by other methods suggests that TJ310 is closely related to these isolates and probably evolved in situ from a vanB2 genotype, vancomycin-resistant isolate during prolonged exposure to the high concentrations of vancomycin in the urinary tract. The vancomycin-dependent phenotype of strain TJ310 appears to be unstable. Spontaneous reversion to a vancomycin-resistant but nondependent phenotype occurs in about 1 in 10⁷ organisms (unpublished data), which is consistent with the hypothesis that the change to the dependent phenotype may be due to a single mutation.

The mechanism of vancomycin dependence of strain TJ310 is difficult to explain based on usual growth factor requirements. Vancomycin does not penetrate the membrane of gram-positive bacteria and does not appear to be metabolized [10, 11]. Thus, the mechanism of growth may be related to an indirect effect of exposing the cells to vancomycin. Vancomycin-resistant enterococcal strains respond to vancomycin by producing a bacterial cell wall lacking the terminal dipeptide D-alanine-d-alanine, the bacterial target for vancomycin [13-15]. Induction of vancomycin resistance is associated with the expression of the abnormal bacterial ligases vanA or vanB, which are structurally related to D-alanine-d-alanine ligases but synthesize depsipeptides other than D-alanine-d-alanine [7, 13-16]. We hypothesized that the vancomycin-dependent strain may lack a functional D-alanine-d-alanine ligase; vancomycin could then function in strain TJ310 by inducing the synthesis of the vanB ligase to compensate for this otherwise lethal defect in cell wall synthesis. Consistent with this, TJ310 cells supplemented with D-alanine-d-alanine grew in the absence of vancomycin (see Figure 2, bottom). Growth did not occur after supplementation with either D-alanine or L-alanine, which enter the peptidoglycan biosynthesis pathway before the ligase reaction [17]. The glycopeptide teicoplanin, unlike vancomycin, does not induce expression of the vanB ligase in most vanB enterococcal strains [18, 19]. Thus, the failure of teicoplanin to support the growth of TJ310 also suggests that the production of the vanB ligase may be the critical factor required for the vancomycin-dependent strain to grow.

Vancomycin dependence is an unusual clinical phenomenon but shows the remarkable ability of bacterial organisms to adapt to the intense antimicrobial pressure of the nosocomial environment. Careful study of such organisms can provide important insights into basic mechanisms of antimicrobial action and development of resistance. In the patient we have described, it is difficult to determine how much morbidity can be attributed directly to the vancomycin-dependent strain, but this strain was repeatedly isolated from urine in association with pyuria and fevers and represented at least a potential pathogen. Detection of vancomycin-dependent organisms from body sites other than the urinary tract would also be considerably more difficult because of the lower vancomycin levels generally found at other sites. Clinicians and microbiologists must be prepared for the challenge of recognizing new, continually evolving bacterial pathogens.

Presented in part at the 1993 American Society for Microbiology Annual Meeting, Atlanta, Georgia, and at the 33rd Interscience Conference on Antimicrobial Agents and Chemotherapy, New Orleans, Louisiana.