The findings reported in this manuscript document a new viral transneuronal tracing approach that can be used to identify connections to neurons within a complex network whose axons collateralize to influence separate targets. To test the utility of this approach, we used a well documented dual infection animal model in which PRV recombinants that express unique reporters are injected into separate kidneys 
. The use of PRV-263 and PRV-267 in this model system provides unique insights into the synaptic organization of complex circuits that cannot be resolved in dual infection studies employing isogenic PRV recombinants that constitutively express unique reporters (e.g.
, PRV-152 & PRV-BaBlu). Particularly important in this regard is the ability to discriminate neurons presynaptic to dual infected neurons. Nevertheless, it is important to emphasize that the method does not permit a definitive identification of all neurons providing synaptic input to dual infected collateralized neurons and thereby provides a qualitative rather than quantitative approach for identifying these neurons.
The method builds upon recent studies in which we reported the construction and characterization of PRV-263 
and demonstrated the ability of lentivirus mediated Cre expression to produce conditional reporter expression from a Brainbow cassette 
carried by PRV-263 in targeted populations of neurons 
. Here we describe the construction and use of PRV-267, which serves both as a transneuronal tracer and a vector for circuit related expression of Cre. To our knowledge this is the first demonstration of the ability to deliver biologically active Cre in a circuit specific fashion across multiple synapses. The fact that Cre is expressed throughout the polysynaptic circuit infected by PRV-267 is validated both by the pattern and kinetics of conditional reporter expression observed within the CNS following separate injections of PRV-267 and PRV-263 into the kidneys. Importantly, the present data confirm the identity and organization of neurons within the preautonomic network previously shown to collateralize to regulate both kidneys 
. Considered with evidence that recombination of the Brainbow cassette only occurs in the presence of Cre, this is an important confirmation that PRV-267 is producing biologically active Cre in the neurons that it infects.
Importantly, the insights derived from the use of PRV-263 and PRV-267 in dual infection experiments are not limited to the ability to identify neurons that collateralize to influence separate targets. We observed neurons that expressed reporters of the recombined Brainbow cassette but not the unique reporter of PRV-267 infection (punctate VP26-mRFP). These neurons can only have been infected subsequent to Cre mediated recombination and transneuronal passage of the PRV-263 genome. Using in vitro
analysis Kobiler and colleagues demonstrated that Cre-mediated recombination occurs prior to replication of PRV-263 and that a remarkably small number of viral genomes – as few as seven – are expressed, replicated and assembled into virions 
. This interesting bottleneck may limit the population of virions that can spread transneuronally and express their genomes. In any case, even if PRV-263 and PRV-267 co-infect a single neuron, the data of Kobiler and colleagues indicates that the probability of second- and third-order neurons being infected by both recombinants drops after each transneuronal passage. Therefore, neurons displaying only cytoplasmic reporters of the recombined PRV-263 genome, and no reporters of PRV-267 infection (punctate VP26-RFP), likely represent neurons that were infected from the early transneuronal passage of progeny virus containing the recombined PRV-263 genome from a dual infected neuron. Similarly, early transneuronal passage of PRV-267, and not PRV-263 recombinants, from dual infected neurons would produce neurons only expressing the PRV-267 genome that are indistinguishable from neurons connected only to the PRV-267 infected kidney. Thus, data derived from this approach must be interpreted conservatively and conclusions on the synaptology of the circuit based only upon positive unequivocal results. In this regard, the singular expression of PRV-263 reporters of the recombined Brainbow cassette provides an unambiguous identification of neurons presynaptic to dual infected neurons.
As noted above, the in vitro data of Kobiler and colleagues demonstrated that Cre-mediated recombination of the PRV-263 genome occurs prior to replication of the virus. However, there is a chance that several incoming PRV-263 genomes will initiate replication before recombination can occur, even in the presence of PRV-267. This can result in neurons that were infected with both viruses expressing the default dTomato reporter along with the reporters liberated by Cre mediated recombination. Under these circumstances it is possible that a single dual infected neuron can replicate up to four different viral genomes (PRV-267, PRV-263red, PRV-263yellow, and PRV-263blue) and transneuronal infection of synaptically connected neurons would sample any combination of these replicated genomes. Indeed, we often observed neurons in vivo that expressed dTomato (a reporter of the uncombined PRV-263 genome) along with the mCerulean and EYFP reporters of recombination.
The functional implications of being able to identify neurons presynaptic to collateralized neurons are apparent in our data. Jansen and colleagues previously documented neurons within the preautonomic network that were co-infected by retrograde transneuronal transport of recombinant strains of PRV from the adrenal gland and superior cervical ganglion as a means of identifying “command” neurons instrumental in the initiation of the “fight-or-flight” response to stressful stimuli 
. The neurons identified in their investigation are among the dual infected neurons observed in our investigation and include areas that have been identified as important mediators of neural responses stress. The LC is among the regions identified in our analysis that were not included in the “command” neurons identified by Loewy and colleagues. Nevertheless, the LC is prominent among the cell groups activated by stressful stimuli and it has been postulated to play a prominent role in orchestrating behavioral and physiological responses to stressors 
. Importantly, available evidence indicates that the LC does not exert its influence upon sympathetic outflow through direct reticulospinal projections to SPNs in the IML 
. Rather, LC neurons project to components of the preautonomic network that, in turn, project directly to SPNs (e.g., RVLM & VMM) and also influence sympathetic outflow indirectly through projections to regions that influence affect 
. Our data suggest that the LC contains a large population of neurons presynaptic to dual labeled neurons, an observation consistent with a prominent role for the LC in the global activation of sympathetic outflow that is a cardinal feature of the fight-or-flight response. Our data are also consistent with a similar functional role for the hypothalamic PVN, which also contained prominent populations of neurons presynaptic to collateralized neurons. Definitive support for these hypotheses requires quantitative analysis of a larger sample size, but the possibility illustrates the potential power of the combined use of PRV-263 and PRV-267 in dual infection analysis of neural circuitry.
The ability to express Cre in a circuit related manner through PRV-267 infection and transneuronal passage also has other experimental applications. For example, PRV-267 can be used to mediate recombination of floxed genes in transgenic mice in a circuit-defined manner. Given the expanding list of floxed genes that are widely available (e.g.
, see the list on the web site of Andras Nagy at the Samuel Lunenfeld Research Institute at Mount Sinai Hospital; http://www.mshri.on.ca/nagy/
) this possibility markedly expands the utility of PRV-267 for functional studies in a variety of systems. Additionally, the virus can be used to produce circuit related conditional reporter expression in the nervous system of the Brainbow mouse 
In conclusion, we have described a new viral tracing method based on the polysynaptic tracing properties of PRV, the ability to express biologically active Cre from the PRV genome, and the conditional reporter capabilities of the Brainbow cassette. The method enables identification of neurons that collateralize within a complex network to exert regulatory control over distant separate targets. It provides a means of expressing Cre in a circuit specific fashion from a replication competent PRV recombinant (PRV-267) and relies upon Cre-dependent combinatorial expression of fluorescent reporters from a Brainbow cassette carried by second PRV recombinant (PRV-263). The unique reporter phenotypes produced in dual infection studies employing these recombinants provides unique insights into the synaptic organization and function of polysynaptic networks and increases the diversity of viral transneuronal tracing tools available for circuit analysis.