Page 334

an observation that illustrates the difficulties of using animal models to represent human disease.

Older mice are relatively resistant to respiratory infection; infant or suckling mice have reproducible symptoms and mortality from pertussis pneumonia, and the disease resembles the disease in humans (Pittman et al., 1980; Sato and Sato, 1988; Sato et al., 1981). Infection induced by intranasal inoculation (Pittman et al., 1980) has been reported to be less reproducible than that induced by aerosol inhalation (Sato and Sato, 1988). The strain of mice used can affect the results (Pittman et al., 1980).  Survivors of a sublethal dose of organisms can develop a chronic infection that lasts for weeks or months (Dolby et al., 1961; Sato et al., 1981; Weiss et al., 1984).

Using intranasal inoculation of infant mice, Weiss and colleagues (1983, 1984) showed that mutant strains of B. pertussis lacking pertussis toxin (PT) or extracytoplasmic adenylate cyclase were much less virulent than the wild-type (naturally occurring) organism. A mutant deficient in filamentous hemagglutinin was nearly as virulent as the wild-type strain. The results obtained with these carefully engineered strains raise a question about the contribution of filamentous hemagglutinin to virulence. Such a contribution had been suggested by data from other models. These and other considerations warrant reservations about the general applicability of the results obtained with this or the other models to the disease in humans.

Mice infected intracerebrally have been the most widely used animal model for pertussis. To achieve this model, anesthetized mice are injected with various numbers of organisms, in some cases after immunization with bacteria or bacterial products (usually given intraperitoneally). Only one strain of B. pertussis, strain 18-323, works well in the model, which raises further questions regarding the applicability of this model to the natural disease in humans. In fact, analysis of isoenzyme patterns suggests that this bacterial strain is genetically more closely related to Bordetella bronchiseptica than it is to other strains of B. pertussis (Musser et al., 1986). In mice, the bacteria attach to the ciliated cells of the ependymal lining of the ventricles (Berenbaum et al., 1960), which simulates attachment to the respiratory cilia in humans with whooping cough. However, this infection within the skull otherwise deviates rather markedly from the presentation of the disease in humans. Despite these obvious differences from the infection in humans, protection in this model has correlated with vaccine efficacy in humans (Medical Research Council, 1959; Standfast, 1958).

STANDARDIZED ANIMAL TESTS OF VACCINE MATERIALS

The intracerebral mouse protection test (Kendrick et al., 1947, 1949) has served importantly in the progress in vaccine development that has been made to date. The test uses a standardized strain of bacteria (strain 18-323)