Prompt administration of PEP is the recommended course for an individual who has come in contact with a rabid animal. Since this prevents the development of clinical rabies it is impossible to be certain how many of the tens of thousands of people who receive PEP on an annual basis have actually been infected with the virus. It is also impossible to know how far the rabies virus may have spread before being cleared by the immune effectors provided or induced by PEP and the infection. The commonly held view that pathogenic wildlife rabies virus that has spread to the CNS cannot be cleared by immune mechanisms is supported by the absence of significant immune cell infiltration into the CNS tissues of individuals who die from rabies 
and the failure of PEP in individuals that have developed signs of rabies 
. Our studies in animal models of rabies suggest that this is a consequence of the inability of virus-specific immune effectors to cross the BBB and enter CNS tissues infected with pathogenic rabies viruses 
. The rabies virus-specific immune effectors that are raised in lethally infected mice are able to clear rabies virus from the CNS if provided access across the BBB. For instance, when the BBB is compromised by the induction of autoimmune CNS inflammation, rabies-specific immune effectors infiltrate CNS tissues and can clear the highly pathogenic SHRBV 
. In addition, the adoptive transfer of immune effectors recovered from mice lethally infected with SHBRV results in clearance of the attenuated CVS-F3 virus from the CNS tissues of mice lacking T and B lymphocytes 
. In contrast, the transfer of cells from mice clearing CVS-F3 has no impact on the outcome of SHBRV infection 
. Regardless of the infecting virus strain, elements of the innate immune response that are important for the early control of virus replication and for attracting immune cells into infected tissues are induced 
. These findings led us to speculate that functional changes at the BBB required to provide immune effectors access to the CNS tissues are induced during infection with attenuated rabies virus strains but not during pathological rabies virus infection 
. A key issue examined in this study is whether or not this is due to an inhibitory process triggered by infection with pathogenic rabies virus. If so, it may be expected that BBB integrity would be maintained during infection with both pathogenic and attenuated rabies viruses and the outcome would be lethal, but it is not. Infection with an attenuated rabies virus induces BBB integrity changes and immune effector entry into CNS tissues regardless of whether or not there is also an ongoing infection with pathogenic rabies virus. However, protection is not provided by immunization with killed virus. We therefore conclude that the generation of a rabies virus-specific immune response in the periphery is not sufficient to clear pathogenic rabies viruses from the CNS tissues. A mechanism selectively induced by infection with attenuated rabies virus, likely manifested at the BBB, is necessary to provide immune effectors access to CNS tissues.
To gain further insight into the mechanism of rabies virus clearance from the CNS tissues, we have used gene-deleted mice to study the role(s) of different antiviral immune effectors in the CNS tissues of mice clearing the attenuated rabies virus CVS-F3. Mice lacking T and B cells cannot clear this virus and die from the infection 
. CD8 T cells contribute to, but are not required for the clearance of CVS-F3 as clearance is merely delayed in mice without this cell population 
. On the other hand, JHD−/−
mice, which lack B cells but have functional CD4 and CD8 T cells, often survive CVS-F3 infection over extended periods despite being unable to clear virus from CNS tissues and exhibiting neurological symptoms. This leads us to conclude that elements of the T cell response, likely including IFN-γ production by CD4 and CD8 T cells, can control certain features of the infection that make significant contributions to its lethality but that antibody is required for virus clearance. To examine the contribution of circulating antibody to virus clearance from CNS tissues, we administered high levels of the rabies virus neutralizing mouse monoclonal 1112 antibody to CVS-F3-infected JHD−/−
mice during the stage of infection when BBB permeability is maximal. While leakage of a 150 kDa molecular weight marker from the circulation into the CNS tissues of CVS-F3-infected mice is minimal over a 4-hour period 
, antibodies present in the circulation over a more extensive period of time can evidently leak into the CNS tissues of the infected mice. 1112 antibody was found associated with the Purkinje cells in the cerebellum that express high levels of rabies virus antigen. The antibody was primarily localized in inclusion bodies which is consistent with previous in vitro studies showing that 1112 antibody is rapidly internalized by rabies virus-infected neuroblastoma cells where it accumulates in intracellular vesicles 
. Of note in our studies is that the intracellular inclusions of glycoprotein-specific 1112 are generally distinct from inclusions of the virus nucleoprotein. The amounts of antibody reaching rabies virus-infected cells in vivo
appears to be relatively low as considerably greater amounts of the antibody can bind to the cells when applied to tissue sections in vitro
. While it is possible that even low levels of virus-neutralizing antibody may impact the replication and spread of the virus while BBB permeability is enhanced, treatment of CVS-F3-infected JHD−/−
mice with 1112 antibody failed to clear the virus.
It should be noted with respect to the origin of the antibodies that participate in rabies virus clearance that serum rabies virus-specific antibody titers peak some time after BBB integrity has been restored 
. The presence of cells expressing the B cell phenotypic marker CD19 and mRNAs specific for κ- light chain in the CNS tissues of mice clearing CVS-F3 
led us to examine the possibility that rabies virus-specific antibodies are produced by infiltrating B cells. The current findings indicate that this is the case. Focal concentrations of antibodies can be readily detected in the CNS tissues of mice clearing CVS-F3 and a high proportion of B cells recovered from the tissues produce rabies virus-specific antibodies in vitro
. This leads us to conclude that the high levels of antibodies required for rabies virus clearance from the CNS tissues are produced at the site of infection rather than diffusing in from the circulation. In this case, passively administered antibody during PEP would primarily impact virus in the periphery and an active immune response leading to elevated BBB permeability and immune effector delivery to the CNS tissues would likely be required to clear virus from the CNS.
As certain of the aspects of BBB function required for immune cell infiltration are unchanged by CNS infection with pathogenic rabies viruses 
, once the virus has reached the CNS a PEP protocol capable of altering the BBB, so that virus-specific immune effectors can reach the infected tissues, is required. Inactivated CVS-F3 can induce rabies virus-specific T and B cells, but fails to promote recovery from CVS-N2c infection over a time frame during which the administration of live CVS-F3 is therapeutic. We consider that this is a consequence of the inability of the inactivated virus to induce the functional changes in the BBB that are required for antiviral immune effectors to enter CNS tissues. In our view, administration of a live-attenuated rabies virus vaccine is the most reasonable, currently available, approach to providing the appropriate immune effectors access to the CNS tissues. The results of our experiments with a new, highly attenuated recombinant rabies virus vaccine which expresses three copies of a mutated glycoprotein gene, strongly support this hypothesis 
. In these studies, the triple G vaccine was shown to promote immune effector delivery into CNS tissues and normal mice were found to survive the intracranial injection of a mixture of the vaccine virus and a highly pathogenic dog strain which was nearly 100% lethal when administered alone 
. The triple G vaccine also proved effective in the post-exposure treatment of mice infected with a highly pathogenic dog rabies virus several days previously 
. However, when UV-inactivated and given peripherally to mimic conventional post-exposure vaccination, there was little protective effect 
. In addition to boosting the antiviral response, attenuated rabies virus vaccines spread to the CNS where they trigger the mechanisms required for T cells and B cells to enter the tissues and clear, not only the attenuated, but also pathogenic rabies viruses. It is clear from the commonly lethal outcome of rabies that these mechanisms are not induced in a timely fashion in the context of the spread of a wildlife rabies virus to the human CNS.