Antigenic materials were extracted from Campylobacter fetus subsp. jejuni strains by heating bacterial suspensions in saline at 1000C and by exposure to ethylenediaminetetraacetic acid. The antigens were heat stable at 1000C, capable of sensitizing sheep erytnrocytes for agglutination in antisera, and able to elicit production of specific antibody in rabbits; they occurred with different immunological specificities in 23 strains. Antisera against the 23 strains could be used for discriminating among isolates of the species when the passive hemagglutination technique was used for serotyping. Three serotypes were more common than others among a collection of human isolates.

Hippurate hydrolysis tests performed on the serotype reference strains of the serotyping scheme based on thermostable antigens under development for Campylobacter jejuni showed that 42 strains were Campylobacter jejuni and 17 were Campylobacter coli. Moreover, only four (0.2%) of 2025 hippurate positive Campylobacter jejuni isolates reacted in Campylobacter coli antisera and 12 (4.3%) of the 282 Campylobacter coli reacted in Campylobacter jejuni antisera. Evidently each species has its own array of antigenic specificities. Separate schemes for serotyping Campylobacter jejuni and Campylobacter coli are advocated.

Antisera were prepared against type strains of the original scheme of B. Perch (Acta Pathol. Microbiol. Scand. 25:703-714, 1948) and against newly defined types to produce separate schemes for O-grouping Proteus vulgaris and Proteus mirabilis. In assessing the schemes for their effectiveness it was found that 82% of 208 P. vulgaris isolates and 88% of 194 P. mirabilis isolates from two hospitals were typable. Only 3.4% of the P. vulgaris isolates agglutinated in P. mirabilis antisera, and 1.5% of the P. mirabilis agglutinated in P. vulgaris antisera, indicating that separation of the schemes would be more advantageous in routine typing. P. mirabilis of groups 03, 06, 010, 029, and 030 were most frequently isolated. Of the P. vulgaris isolates, 25% belonged to newly defied 0-groups, and one of these was the largest with 14% of all isolates of this species. The application of serotyping using separate schemes for each species was advocated in epidemiological studies.

Two hundred eighty-five isolates of Campylobacter jejuni-Campylobacter coli from children with gastroenteritis at The Hospital for Sick Children (Toronto, Canada) over a three-year period were biotyped by the hippurate hydrolysis test and serotyped on the basis of thermostable, soluble antigens by the passive hemagglutination technique. Hippurate-negative strains (C. coli) were only 3.2% of the isolates. Ninety-seven percent of the isolates were serotypable with 55 antisera. About half of the strains belonged to one of four serotypes (2, 4, 3, or 1); about three-quarters belonged to one of 10 serotypes. Serotype 2 was consistently the commonest serotype in each of the three years of the study, accounting for 15%-20% of all isolates tested.

Forty Campylobacter jejuni and 17 Campylobacter coli strains that constitute the set of reference strains for our serotyping scheme were each examined for the presence of plasmid DNA. Agarose gel electrophoresis of alkaline-extracted DNA showed the occurrence of 29 bands in 11 C. jejuni strains and 40 bands in C. coli strains. Plasmids ranged in size from 1.6 to 70 megadaltons. Most strains that carried plasmids had between 2 and 6 of them; however, one strain had 14 plasmids, and two strains contained only 1 plasmid each. Repeated electrophoresis demonstrated that all plasmid profiles were stable. A different plasmid profile was seen for each of the 19 plasmid-carrying strains, but it was clear that plasmids of the same or similar molecular weight could be found in different strains. On the basis of these findings, we are persuaded that plasmid profiles determined by a rapid procedure for DNA extraction will play a significant role in resolving complexities among strains that are difficult to serotype and could be useful in epidemiological studies in which the implicated isolates are plasmid bearers.

Examination of' 729 isolates of' Proteus rettgeri showed that 674 reacted positively in tests for phenvlalanine deamination, indole production, growth on citrate, and acid production from meso-inositol, and negativelv for L-ornithine decarboxvlation and acid production from lactose, maltose, D-xvlose, and L-arabinose. Only 51 isolates diff'ered in one, and four diff'ered in two of these ten reactions, which were taken as the core characteristics of' the species. On the basis of' additional tests (acid production from salicin, L-rhamnose, D-mannitol, adonitol, and D-arabitol), the 729 isolates could be separated into f'ive groups. Groups 1, 2, 3, and 4 could be f'urther separated on the basis of' the reaction with meso-erythritol, and group 5 could be subdivided on the basis of' reaction with D-mannitol. Two metabolically distinct kinds of' P. rettgeri were recognized. Isolates of the f'irst kind (groups 1, 2, 3, and 4) each utilized both adonitol and D-arabitol and most utilized meso-ervthritol. Isolates of' the other kind (group 5) were negative with the three polyhydric alcohols but resembled, in their reactions, some strains of' Prouidencia stuartil. These may be intermediates between P. rettgeri that catabolize these substrates and the Providencia. One of the gram-negative species of bacteria that has increased in importance as an infectious agent is Proteus rettgeri. Its importance in

The somatic (O-) antigens of the type strains of the providencia antigenic scheme were examined for their biochemical reactions and their O-specificities. The scheme of 62 O-antigens was reconstituted from 52 original type strains and 10 strains substituted for originals that either were biochemically atypical of the genus or showed inappropriate serological reactions. Thirty-six type strains showed no significant relations with other type strains, and antisera could be used for typing without absorption. Among 26 type strains, significant reciprocal relations were demonstrated, and each cross-reacting antigen was examined for specificity and for its distribution among the type strains. Antisera to these strains required absorption with cell suspensions of other type strains for production of specificity in O-typing. Each typing antiserum, at low dilution, was shown to agglutinate homologous, but not heterologous, cell suspensions of type strains, and this result demonstrated the required specificity for typing on the basis of the O-antigens.

Campylobacterjejuni from sporadic cases and outbreaks of gastroenteritis were serotyped on the basis of heat-extracted soluble thermostable antigens identified with the use of the passive hemagglutination technique. A total of 168 isolates were separated into 45 different types. The largest proportion of the isolates fell into three serotypes, each with 11 to 12.5% of the total number. Three less frequently occurring serotypes each included approximately 5%, and the remaining 50% of the isolates were distributed among 39 other serotypes. In most cases, serotyping demonstrated that epidemiologically linked isolates were of the same serotype, but the outbreak strains could belong either to frequently or to infrequently isolated serotypes. The high correlation between clinical findings and serotyping results confirmed the applicability of the serotyping scheme in epidemiological investigations of C. jejuni infections.