Identification of Nonlipophilic Corynebacteria Isolated from Dairy Cows with Mastitis

Abstract

Nonlipophilic corynebacteria associated with clinical and subclinical mastitis in dairy cows were found to belong to four species: Corynebacterium amycolatum, Corynebacterium ulcerans, Corynebacterium pseudotuberculosis, and Corynebacterium minutissimum. These species may easily be confused. However, clear-cut differences between C. ulcerans and C. pseudotuberculosis were found in their acid production from maltotriose and ethylene glycol, susceptibility to vibriostatic agent O129, and alkaline phosphatase. Absence of growth at 20°C and lack of α-glucosidase and 4MU-α-d-glycoside hydrolysis activity differentiated C. amycolatum from C. pseudotuberculosis and C. ulcerans. The mastitis C. pseudotuberculosis strains differed from the biovar equi and ovis reference strains and from caprine field strains in their colony morphologies and in their reduced inhibitory activity on staphylococcal β-hemolysin. C. amycolatum was the most frequently isolated nonlipophilic corynebacterium.

The lipophilic species Corynebacterium bovis is frequently isolated from milk samples in many dairy farms. It is associated with very mild forms of mammary inflammation. Slightly increased somatic (leukocyte) cell counts in the milk are usually the only manifestations of these infections. A new lipophilic oxidative species, Corynebacterium mastitidis, and a new fermentative species, Corynebacterium camporealensis, have been recognized and described from the milk of sheep (9, 10). These bacteria are also associated with subclinical mastitis.

Fermentative nonlipophilic corynebacteria are much less often found in mastitic milk. They usually remain unidentified or are named Corynebacterium ulcerans (23) or Corynebacterium pseudotuberculosis. The latter name is usually given when the bacteria are found in combination with cutaneous lesions. These identifications are tentative at best because C. ulcerans was not a recognized species until 1995 (21). Two other nonlipophilic fermentative species, Corynebacterium minutissimum and Corynebacterium amycolatum, associated with humans, were described in 1983 (5) and 1988 (6), respectively. They are often confused (11, 24, 27), and they have been identified erroneously as Corynebacterium xerosis or Corynebacterium striatum (16, 24).

In the present communication we report the isolation and identification of C. amycolatum and C. minutissimum, hitherto unknown in animals, from cows with mastitis, and we describe their differentiation from two other nonlipophilic corynebacteria, C. ulcerans and C. pseudotuberculosis, associated with the same type of animal infection. Special attention was given to the differentiation of the latter two species because the phenotypic characteristics described in the literature proved insufficient to allow reliable identification.

MATERIALS AND METHODS

Samples and strains.

Milk samples included samples from cows with clinical infections sent in for bacteriological diagnosis by veterinary practitioners and samples from the quarters of normal cows showing subclinical mastitis with increased somatic cell counts collected during routine subclinical-mastitis control examinations. The strains were isolated on Columbia blood agar base (Oxoid, Basingstoke, England) with 5% sheep blood. Nonlipophilic coryneform bacteria isolated in concentrations of over 1,000 CFU per ml of milk were selected for further study. The following reference strains were included: C. amycolatum CIP 103452, C. minutissimum ATCC 10288, C. pseudotuberculosis CIP 5297^T (biovar equi) and CIP 102968 (biovar ovis), and C. ulcerans CCUG 2708^T, CCUG 16556, CCUG 18647, CCUG 22327, and CCUG 22328. Additionally, four C. pseudotuberculosis field strains obtained from four herds of goats infected with cutaneous lymphadenitis were included to examine differences in hemolytic activity and colony morphology.

DNA hybridizations.

DNAs of the reference strains and three C. ulcerans and four C. pseudotuberculosis mastitis strains were extracted and purified following a previously described procedure (19). DNA from C. ulcerans CCUG 2708^T and from C. pseudotuberculosis CIP 5297^T was labeled in vitro by nick translation with tritium-labeled dCTP (Amersham International, Amersham, England). Hybridization experiments with labeled DNA and fragmented DNA preparations were carried out under optimal conditions at 60°C for 16 h in 0.42 M NaCl by the S1 nuclease-trichloroacetic acid method (19). The results were normalized to those of the homologous reaction, after the percentage of S1 nuclease-resistant nucleic acid in the control tube (DNA calf thymus) was subtracted.

Phenotypic identification.

Growth characteristics were recorded after aerobic and anaerobic incubation at 37°C in air or in 5% CO₂ in brain heart infusion (Oxoid) and on Trypticase soy agar (Gibco, Paisley, United Kingdom) supplemented with 5% sheep blood and with or without 1% Tween 80 (Merck, Darmstadt, Germany). The reverse CAMP test was carried out with a Staphylococcus aureus strain showing only β-hemolysin effects on sheep blood. Urease was tested with urea agar base concentrate (Difco, Detroit, Mich.) and 1.5% agar incubated for 10 days. Nitrate broth (Difco) in Durham tubes was incubated for 2 days. Amylase activity was recorded after 1 or 2 days on spot-inoculated Columbia agar (Biolife, Milan, Italy), which contains 1 g of starch/liter, on the same medium with a supplement of 0.5 g of soluble starch/liter, and additionally with the rapid amylase test (Biolife). Tyrosinase was investigated on nutrient agar (Oxoid) with 0.5 g of tyrosine/liter, spot inoculated and incubated up to 15 days. Filter strips impregnated with 0.1% N,N,N′,N′-tetramethyl-1,4-phenylenediammonium-dichloride (Merck) were used for oxidase testing, and 3% H₂O₂ was used for catalase testing. Esculin hydrolysis was recorded after 1 and 2 days of incubation on medium containing yeast extract, peptone, tryptone, agar, 0.5% esculin, and 0.5% ammonium ferric citrate. Growth at 20°C, glucose fermentation at 42°C, formate alkalinization, and ethylene glycol acidification were tested as described by Wauters et al. (25). Susceptibility to the vibriostatic agent O129 was tested with Rosco (Taastrup, Denmark) diagnostic tablets O129 (150 μg) on Isosensitest agar (Oxoid) with 5% sheep blood. API CORYNE, API 50 CH (BioMérieux, La-Balme-les-Grottes, France) galleries, and the BBL CRYSTAL gram-positive identification kit (Becton Dickinson, Cockeysville, Md.) were used according to the manufacturers’ instructions. The medium used for testing carbohydrate breakdown in API 50 CH contained 2% tryptone (Oxoid L42), 0.5% sodium chloride, 0.05% sodium sulphite, and 0.017% phenol red. Reactions were recorded after 2 days of incubation, except where otherwise indicated.

RESULTS

Samples and isolations.

Twenty strains identified as C. amycolatum (see below), seven C. pseudotuberculosis strains, six C. ulcerans strains, and three C. minutissimum strains were isolated in concentrations higher than 1,000 CFU/ml of milk and in pure or nearly pure culture from approximately 80,000 milk samples examined over a 3-year period. All were from different cows and their quarters from 34 different farms. Two C. amycolatum strains were from a farm with poor milking hygiene where 22 isolates were obtained on several monthly repeat visits from 14 quarters of 14 cows. These isolates were considered likely to be related representatives of a single strain and were not further studied. Five strains were obtained from approximately 3,000 cows with cases of clinical mastitis, 15 strains were from nearly 6,000 samples with high cell counts, and the remaining strains were from approximately 70,000 routine control examinations of udders with unspecified clinical status. Species distributions did not clearly differ among these samples of different origin.

DNA similarity.

DNAs of the C. ulcerans type strain CCUG 2708^T and the C. pseudotuberculosis type strain CIP 5297^T were labeled and tested by DNA-DNA hybridization against unlabeled DNAs of strains whose biochemical characteristics resembled those of these two species. The DNA relatedness between strain CCUG 2708^T and the mastitis strains A 203, B 204, and B 205 revealed similarity percentages of 78, 74, and 69, respectively, indicating that these strains are members of the species C. ulcerans. Field strains C 206, C 207, C 208, and C 209 showed hybridization percentages of only 23 to 26 with the same C. ulcerans strain, while they were related at the species level with the C. pseudotuberculosis type strain, CIP 5297^T, exhibiting DNA relatedness ranging from 63 to 67%.

Growth characteristics.

The colonies of the 21 C. amycolatum strains on aerobically incubated blood plates were dry and waxy, with crenate edges and whitish-pale elevated centers. Tween 80 did not influence their growth appreciably. Deposits, surface pellicles, and flakes were seen in broth. The anaerobic growth of A. amycolatum was particularly feeble, and colonies were scarcely visible with the naked eye after 2 days of incubation. Growth was better at 37°C than at 30 or 42°C. C. amycolatum strains did not grow at 20°C. Colonies of C. pseudotuberculosis reference and caprine field strains on aerobically incubated agar plates were smaller and more dry and crumbly than those of C. ulcerans. Colonies of C. pseudotuberculosis bovine mastitis strains had an appearance intermediate between these two types. The two species grew much better anaerobically than C. amycolatum, and they were able to grow at 20°C. In broth, they formed deposits with clumps or pellicles near the surface. C. minutissimum strains produced moist mucoid colonies, and in broth they showed uniform turbidity without clumps at the surface.

The C. pseudotuberculosis and C. ulcerans strains, except four of the C. ulcerans mastitis strains, differed from the others in being hemolytic. Hemolysis was much stronger in anaerobiosis than in aerobiosis. Only C. pseudotuberculosis and hemolytic C. ulcerans strains inhibited staphylococcal β-hemolysin in the reverse CAMP test. However, the inhibitory effects of the C. pseudotuberculosis mastitis strains were distinctly less vigorous than the effects of other strains.

Biochemical activity.

All strains fermented glucose. They were catalase positive and hydrolized l-phenylalanine–AMC, l-tryptophan–AMC, l-arginine–AMC, l-pyroglutamic acid–AMC, l-isoleucine–AMC, l-valine–AMC, and proline- combined with leucine-p-nitroanilide. They all remained negative in the following tests: gelatin hydrolysis, tyrosinase (except the C. minutissimum reference strain), oxidase, β-glucuronidase, pyrrolidonyl arylamidase, β-galactosidase, 4MU-β-d-glucoside (except one C. minutissimum and one C. amycolatum strain), 4MU-N-acetyl-β-d-glucosaminidine (except one C. pseudotuberculosis strain), 4MU-β-glucuronide, p-nitrophenyl-β-d-cellobioside, o-nitrophenyl-β-d-galactoside combined with p-ni-trophenyl-α-d-galactoside, and N-acetyl-β-glucosaminidase. None of the strains produced acid from adonitol, amygdaline, d- and l-arabinose, d- and l-arabitol, arbutine (except one strain of C. amycolatum and one C. minutissimum strain), cellobiose (except one C. amycolatum strain), dulcitol, erythritol, d- and l-fucose, β-gentiobiose, gluconate, 2- and 5-keto-gluconate, inositol, inulin, lactose (except one C. amycolatum strain), d-lyxose, mannitol, melezitose, melibiose (except one C. minutissimum strain), α-methyl-d-glucoside, α-methyl-d-mannoside, β-methyl-xyloside, d-raffinose, rhamnose, salicin (except one C. minutissimum and two C. amycolatum strains), sorbitol, l-sorbose, d-tagatose, d-turanose (except two weakly positive C. amycolatum strains), xylitol, or d- and l-xylose. Variable reactions, including reactions differentiating between species, are listed in Table 1.

TABLE 1.

Variable and differentiating characteristics of nonlipophilic corynebacteria isolated from mastitic bovine milk

Characteristic	No. of strains positive for characteristic
	C. amycolatum (21 strains)a	C. pseudotuberculosis (9 strains)a	C. ulcerans (10 strains)a	C. minutissimum (4 strains)
Homogeneous growth in broth	0	0	0	4
Growth at 20°C	0	9	10	0
Glucose fermentation at 42°C	7/7	1/5b	1/7b	4
Hemolytic	0	9	6	0
Reverse CAMP effect	0	9c	6	0
Susceptible to O129	0d	0	10	4
Strong amylase reaction	0	0	6e	0
Formate alkalinization	0/7	0/5	0/7	4
4MU-α-d-glucoside hydrolysis (BBL)	0	9	10	0
α-Glucosidase (API)	0	9	10	0
p-Nitrophenyl-β-d-glucoside hydrolysis	3	0	0	1
p-Nitrophenyl-α-d-maltoside hydrolysis	10	0	0	0
Arginine hydrolysis	3	9	10	3
Alkaline phosphatase	21	0	10	3
4MU-phosphate hydrolysis	21	0	8	1
p-Nitrophenyl phosphate hydrolysis	21	0	10	2
Urease	0f	9g	10g	0
Nitrate reduction	5	8	0	3 (2 very weak)
Pyrazinamidase	18	0	0	3
Acid from
Esculin	0	0	0	3 (2 very weak)
Ethylene glycol	2/7	0/5	7/7	1
N-Acetyl glucosamine	4	8	9	0
Glycerol	19	9	10	0
Glycogen	0	0	4	0
Galactose	10h	2	4	0
Fructose	6	9	10	4
Mannose	5h	9	10	4
Maltose	16	6h	7	4
Maltotriose	3	0	10	1
Ribose	21	9	9	1
Saccharose	2	0	0	4
Starch	1	0	5	0
Trehalose	0	0	7i	0

^aExcept when indicated (/). C. ulcerans CCUG 2708^T was not included in the phenotypic tests. 

^bReactions possibly weakly positive but difficult to read with the strains reported here as negative. 

^cVery weak in bovine mastitis strains. 

^dA partial inhibition zone may occur. 

^eNonhemolytic strains are only strongly positive in the plate test with Columbia agar with 0.1% starch. 

^fReference strain C. amycolatum CIP 103452 is urease positive. 

^gPositive in the API CORYNE system but variable in the BBL CRYSTAL galleries. 

^hMostly delayed positive (3 to 10 days). 

ⁱNegative in the CRYSTAL system. 

DISCUSSION

The four species discussed here are infrequently encountered in mastitis diagnostic bacteriology. Nevertheless, they appeared to be involved in a number of clinical or subclinical mastitis cases. C. xerosis, which resembles C. amycolatum in growth and biochemical characteristics, is very rare in human clinical specimens (11), and a discrepancy has been noted between the large number of C. minutissimum isolations in routine clinical bacteriology and the relatively small number of publications claiming a disease association (13, 20). C. amycolatum, on the other hand, has been reported as the most frequently encountered nonlipophilic corynebacterium in human clinical specimens (11, 13). This species has been recognized in recent years as possibly an important infectious agent (3, 8). The human clinical situation is reflected in the dairy material discussed here: C. xerosis was not found, and C. amycolatum was the most frequently isolated fermenting nonlipophilic corynebacterium.

Along with these strains, two other species which are known to occur in animals were found. C. ulcerans has been described as a cause of bovine mastitis (14, 23), but its pathogenic significance has remained unclear, possibly because of identification problems. C. pseudotuberculosis has long been known as a cause of caseous lymphadenitis, an important disease of goats, sheep, deer, and buffalo and, in certain regions, also of horses (4, 15). More rarely, at least in temperate climates, the organism causes severe disease in cattle, and in these animals a mastitic as well as a cutaneous form may develop (26). It is less well known as an exclusive agent of bovine mastitis without simultaneously occurring cutaneous lesions.

The results of the present study demonstrate that the differentiation of nonlipophilic corynebacteria from cows with mastitis is quite possible by taking advantage of a number of useful characteristics (Table 1). Unlike that of the lipophilic C. bovis, their growth on primary isolation plates extends outside the fatty zone of the milk spots inoculated on the plates. Apart from the rare C. minutissimum strains identified in the present study, the nonlipophilic corynebacteria associated with mastitis in cows all grow similarly in broth. They produce deposits and clumps near the surface. Their colonies vary from dry and crumbly for C. pseudotuberculosis or with a mat surface for C. amycolatum to smooth for C. minutissimum. The newly described C. camporealensis from sheep with subclinical mastitis differs from these strains in its positive CAMP reaction. C. mastitidis from the same origin differs in its oxidative and lipophilic nature. Other reactions differentiating both ovine species from the individual species in the present study are given in references 9 and 10.

C. amycolatum and C. minutissimum strains differ from C. pseudotuberculosis and most C. ulcerans strains in their negative urease and 4MU-α-d-glucoside reactions and their lack of hemolysis and absence of inhibitory activity on staphylococcal β-hemolysin. It should be recognized that although all C. amycolatum mastitis strains studied were urease negative, the type strain has urease activity (Table 1), and the inhibitory effect on β-hemolysin in the reverse CAMP test may be weak, as was the case with the C. pseudotuberculosis mastitis strains in the present study, or absent, as with the nonhemolytic C. ulcerans strains. The present results cannot be used to separate C. minutissimum from C. amycolatum because only four isolates of the first species were found in the mastitis samples studied. Carbon substrate assimilation tests can be used for that purpose (18).

As mentioned in the introduction, the differentiation of C. ulcerans from C. pseudotuberculosis is problematic when only characteristics reported in the literature are used. The urease, glycogen, sucrose, amylase, and alkaline phosphatase reactions have been indicated in Bergey’s Manual (7) and in the more recent descriptions of C. ulcerans (12, 21) as means to differentiate these two species. However, the species were urease positive in our tests, acid production from glycogen was variable in C. ulcerans, and acid from sucrose as well as alkaline phosphatase, both reported to be variable in C. pseudotuberculosis, were found to be negative in our strain collection. Only C. ulcerans was found to show strong amylase reactions, but the nonhemolytic mastitis strains did not produce large reaction zones on Columbia agar supplemented to contain 0.15% starch, and they were negative in the commercially available amylase tube test. Enzymatic hydrolysis of alkaline phosphate and 4MU-phosphate, acid production from maltotriose, and acidification of ethylene glycol, as well as susceptibility to O129, proved to be more useful differential characteristics of C. ulcerans. These findings were corroborated by the results of the DNA similarity tests. The improved identification of this species will also be helpful in elucidating its potential zoonotic implications.

With the exception of maltotriose and ethylene glycol, useful in the differentiation of C. ulcerans from C. pseudotuberculosis, and certain reactions of possible utility in the identification of C. minutissimum (Table 1), carbohydrate and polyalcohol breakdown tests had only limited discriminatory capacity for the species studied. It should be noted, however, that only four C. minutissimum strains were present in the strain collection studied. Moreover, this species is biochemically and genomically heterogenic (1, 13). One of the four strains studied here differed in several characteristics from the other three.

Although many results of individual tests contained in the biochemical galleries proved most helpful, the databases provided with these systems to interpret the biochemical activity profiles rarely gave reliable identifications. Only with C. pseudotuberculosis and the API CORYNE test were satisfactory results obtained. With other species and system combinations less than 10% of the identifications were exact. Undoubtedly, updates will yield considerably better results.

Certain host-associated (ecovar) differences appear to exist among C. pseudotuberculosis strains. Equine (biovar equi) strains, unlike caprine and ovine (biovar ovis) strains (2, 17, 22) and bovine strains from cutaneous lesions (26), reduce nitrate, as did the bovine mastitis strains we studied. The last group of strains also differ from biovar equi and biovar ovis strains as well as from caprine field strains in their colony morphologies and in their reduced inhibitory activities on staphylococcal β-hemolysin. These findings suggest that C. pseudotuberculosis strains involved in the exclusively mastitic form of corynebacterial infection may belong to a distinct group.