INTRODUCTION

Top Abstract Introduction Materials and Methods Results and Discussion References

Members of the genus Klebsiella are responsible for nosocomial infections in adults and for outbreaks in newborn populations in the hospital (3, 7, 9). While these outbreaks are usually caused by Klebsiella pneumoniae or Klebsiella oxytoca, rare episodes caused by environmental strains have been reported in the literature (6, 11, 15). These environmental strains have also been isolated from sources where the significance of the isolate is unclear (13).

In 1981, Bagley et al. (2) and Izard et al. (10) described two new environmental species, Klebsiella planticola and Klebsiella terrigena, respectively. These species were differentiated from K. pneumoniae and K. oxytoca by using tests for temperature-dependent fermentation of glucose, acid production from melezitose and L-sorbose, utilization of carbon sources, and gas production from lactose at 44.5°C (fecal coliform test). Studies from France and Germany suggest that up to 19% of Klebsiella spp. identified in clinical settings are actually K. planticola (12, 13, 16, 17). K. planticola is not routinely identified because clinical identification protocols do not include the tests that are necessary for its identification. Test kits and automated methods have not been able to correctly identify these environmental organisms because they are not included in the identification databases of most diagnostic products nor are substrates included on panels that would differentiate them.

In an attempt to distinguish these environmental species from K. pneumoniae and the other clinical Klebsiella spp., Monnet and Freney developed a battery of supplemental tests as an adjunct to conventional clinical identification systems (11). These included four carbon substrate assimilation tests (ethanolamine, histamine, D-melezitose, and DL-3-hydroxybutyrate) and two conventional tests (indole production and ornithine decarboxylase).

More recently in Germany, a 1-year survey of newborns on a neonatal ward reported that of all the Klebsiella spp. recovered from oropharyngeal and rectal swab specimens, 72% were K. oxytoca and 8.7% were K. planticola (18).

To identify possible K. planticola strains among clinical Klebsiella isolates in the United States, and more specifically in newborns, this study consisted of two phases using the supplemental identification schemes described in the literature.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results and Discussion References

Bacterial strains. The 395 Klebsiella stock isolates taken from the culture collection of the Centers for Disease Control and Prevention (CDC) consisted of K. pneumoniae (205 strains), K. oxytoca (63 strains), K. terrigena (61 strains), K. planticola (26 strains), Klebsiella ornithinolytica (20 strains), Klebsiella ozaenae (10 strains), and Klebsiella rhinoscleromatis (10 strains). Most strains of K. terrigena were European environmental in origin and most K. planticola were from environmental sources in the United States. Of the 63 K. oxytoca isolates, 43 were from newborns. Strains from the CDC culture collection were removed from storage at 70°C. Each isolate was passed three times on 5% sheep blood agar plates (BD Biosciences, Sparks, Md.) at 35°C prior to testing. They were identified on conventional media as previously described (5) with some modifications by Hickman and Farmer (8). Incubations were at 35°C, and test results were read at 24 and 48 h and 7 days, unless otherwise noted. Commercial media were used whenever possible. This battery of media does not routinely contain the discriminators needed to distinguish K. planticola or K. terrigena from K. pneumoniae or K. oxytoca.

Carbon source utilization. The carbon substrates tested included ethanolamine, histamine, D-melezitose, and DL-3-hydroxybutyrate (Sigma Chemical Co., St. Louis, Mo.) according to the method of Monnet and Freney (11). Aqueous solutions of ethanolamine and histamine were prepared with 10 g of carbon substrate per liter. D-Melezitose and DL-3-hydroxybutyrate aqueous solutions were prepared by using 20 g of carbon substrate per liter. The solutions were filter sterilized through 0.22-µm-pore-size Millex filters (Millipore Corp., Bedford, Mass.) and were stored at 4°C in 50-µl aliquots in 13- by 100-mm sterile screw-cap tubes. A fifth tube without carbon substrate was used as a growth control. Two hundred microliters of AUX medium (bioMérieux, Inc.) was added to each tube prior to use. All tubes were inoculated with 100 µl of a 0.5 McFarland suspension of the respective strain made up in 0.85% saline. All tubes were incubated at 30°C and were examined at 24 and 48 h. Tubes with growth greater than or equal to the control tube were considered positive.

Pectate degradation. Liquefaction of pectate was determined by the method of Ewing (4).

Pigmentation on D-gluconate ferric citrate agar. Observation for brown pigment was determined according to methods used by Naemura (14).

Production of indole. Peptone water (2%) was inoculated, incubated at 35°C for 48 h, and developed with Kovac's reagent (4).

Fermentation of carbohydrates. Fermentation of D-melezitose, L-sorbose, and dextrose was determined by using Enteric Fermentation Base (BBL Microbiology Systems) with Andrade's indicator. A 0.5% filter-sterilized solution of either D-melezitose or L-sorbose was added to the base. Commercially obtained Andrade's broth with 1% dextrose (BBL) was also used. All tubes were inoculated with a light suspension of each respective organism in nutrient broth. All tests were incubated at 35°C and read daily for 7 days. For temperature-dependent tests at 5 ± 1°C, the tubes were incubated for 7 days in a refrigerator and read daily (4).

Ornithine decarboxylase. Commercially obtained Moeller's ornithine and paired base (BBL) was used. All tubes were inoculated with a light suspension of each respective organism in nutrient broth and were overlaid with approximately 1 in. of sterile mineral oil. They were incubated at 35°C and read daily for 7 days.

Fecal coliform reaction. EC broth (Difco) was inoculated with a light suspension of organism and was incubated for 24 h at 44.5 ± 0.5°C (1). Tubes producing gas were considered to be positive for fecal coliform.

	RESULTS AND DISCUSSION

Top Abstract Introduction Materials and Methods Results and Discussion References

Expected results of carbon source, indole, and ornithine testing for seven species of Klebsiella are shown in Table 3. The results of carbon source utilization, indole, and ornithine testing with our strains of seven Klebsiella species essentially agreed with those of Monnet and Freney (11) (Table 1). K. pneumoniae, K. oxytoca, and K. ornithinolytica were easily and accurately identified by using these methods. An organism was identified as K. planticola if it was negative in tests for ethanolamine and ornithine decarboxylase and was positive in tests for histamine. An organism was identified as K. terrigena if it was negative in tests for indole and was positive in tests for D-melezitose. After testing 205 stock K. pneumoniae strains by using the additional substrates described in this paper, only one strain was identified as K. planticola and none was identified as K. terrigena. We found carbon source utilization testing to be less suitable for the identification of K. ozaenae and K. rhinoscleromatis because of the similarity in test results. During the testing of 10 strains of each species, there was poor growth, making the tests difficult to read and interpret. Fortunately, these two species are not often encountered in the United States. None of the fresh clinical isolates originally identified at local hospitals as K. pneumoniae or K. oxytoca were reidentified as K. planticola or K. terrigena by using carbon source assimilation testing.

For those laboratories wishing to identify K. planticola and K. terrigena, the additional battery of tests proposed by Monnet and Freney is sensitive and specific.

Results of the biochemical tests on the 43 strains of K. oxytoca isolated from newborns (Table 2) confirmed the identities of 42 strains as K. oxytoca. We found only one strain that might possibly be identified as K. planticola, but it did not ferment glucose at 5°C, which it should have done. Only DNA hybridization would confirm its identity as K. planticola. An isolate was identified as K. planticola if it was negative in tests for pectate degradation, fermentation of melezitose, pigment production on ferric citrate agar, and utilization of m-hydroxybenzoate and failed to produce gas from lactose at 44.5°C. Our incidence does not agree with that of Podschun and Ullmann who reported that 9% of all Klebsiella spp. isolated from newborns in their institution were K. planticola (17). Because CDC is a reference laboratory that receives only problem isolates for identification, it is probable that we are not receiving a true surveillance set of isolates. Our entire collection of K. oxytoca (and, thus, of possible K. planticola isolates) numbers approximately 200 strains.

After retesting 479 strains, we could identify only two possible K. planticola strains. It appears that in the United States, K. planticola may not be present at the same rate as in other countries (0.5% versus 9 to 13%, respectively). At this time, the need for additional testing of Klebsiella isolates in clinical laboratories in the United States is not warranted.