PRV infects and replicates in dissociated SCG neurons
Previous studies investigating changes in electrophysiology of PRV infected neurons used widely varying techniques, including intracellular sharp recordings of infected SCG neurons in situ
and patch clamp recordings in brain slices of infected animals. In these studies, it was difficult to determine if and when a recorded neuron was infected. To overcome these limitations, we used cultures of dissociated rat sympathetic ganglion (SCG) neurons, free of replicating epithelial and support cells. After one week in culture, (see methods
), the cell bodies of some SCG neurons had divided once or twice before terminal differentiation and were typically found either in clusters of 2–6 cells. These neurons also formed an extensive network of axons (). We infected cultures with sufficient virus to ensure infection of all cell bodies. In most of our studies we used PRV 151, a virulent PRV Becker recombinant that expresses diffusible GFP. Fluorescence was visible by 4 hpi in all neurons across each culture and remained strong for at least three days following infection ().
Elevated rates of electrical activity observed in PRV infected neurons.
PRV infection induces significant electrophysiological changes in infected neurons in vitro
recordings from rat SCG tissue infected with PRV 151 revealed significant changes in electrophysiological properties, including highly elevated rates of spontaneous AP firing and synchronous activity on pre- and post- ganglionic nerves 
. To determine whether similar changes would be seen in cultured PRV infected SCG neurons, we examined action potential (AP) firing rates using whole cell patch recordings in current clamp mode. In control mock-infected neurons, AP firing was rare, and slow rising and decaying excitatory post-synaptic potential (epsp)-like depolarizations were only occasionally observed on top of the resting membrane potential (). As PRV 151 infection progressed, neurons began to fire APs at elevated rates compared to mock-infected neurons (). Infected neurons also displayed smaller spikelet-like events and had significantly hyperpolarized resting potentials, despite increases in AP firing rates (). These results demonstrate that electrophysiological changes qualitatively similar to those observed in tissue were observed under our culture conditions.
PRV infected neurons fire APs at elevated rates by 8 hours post infection
We next characterized the time course and magnitude of these changes. We first recorded from neurons at specific times following infection (4–8,8–10,14–16,18–20 and 24–46 hpi) and compared their AP firing rates with that observed under mock-infection. Mock-infected SCG neurons had a mean AP firing rate of 0.3±0.2 Hz (). This same low firing rate was seen in PRV 151 infected neurons between 4 and 8 hpi (). However, by 8 hpi PRV 151 infected neurons began to fire APs at a mean elevated rate of 4.0±1.4 Hz (). By 14–16 hpi the PRV 151 infected neurons were firing at rates 10–25 fold greater than uninfected neurons; most had rates greater than 5 Hz (86.2%), with a mean rate of 7.9±0.8 Hz (). Infected neurons continued to fire APs at elevated rates at 24 hpi and did so until at least 72 hpi, the longest incubation period assayed (, data not shown).
Onset of PRV-induced elevated AP and spikelet-like event rates.
PRV infected neurons show signs of electrical coupling
In addition to elevated AP firing rates, PRV infected neurons showed brief sub-threshold depolarizations similar to spikelets observed in electrically coupled neurons 
(, gray arrows). These putative electrical coupling events had fast rise and decay rates and were often followed by a shallow after-hyperpolarization. They were distinctly different from smaller more slowly rising and decaying, chemically mediated, excitatory postsynaptic potentials (epsps) observed in mock-infected neurons (, blue vs black traces) 
. We identified these events based on rate of rise (see Methods
) to distinguish them from any slower rising and decaying epsps. In the following, we will refer to these events as “spikelet-like events”. Typically when spikelet-like events were observed, they occurred in several size classes, similar to previously described spikelets (, blue traces) 
. Spikelet-like events began to occur at elevated rates above mock-infection by 4 hpi in PRV 151 infected neurons, and continued to occur at high rates until at least 24 hpi ().
Elevated AP and spikelet-like event rates are unaffected by somatic current injections
Current injections through the patch electrode were used to examine if PRV 151 infection affected the relationship of steady state voltage and AP firing rate of SCG neurons. In mock-infected neurons, AP firing rate could be elevated by injection of depolarizing currents (). A mean depolarization of 6.1±0.8mV was required to reach AP threshold (n
22, ). Between 4 and 8 hpi, PRV 151 infected neurons responded much like mock-infected neurons to current injections and required the same level of voltage to reach AP threshold (5.5±2.4mV, n
Despite large cell body current injections, elevated AP rates remain relatively unaffected.
By 8–10 hpi, PRV 151 infected neurons with elevated AP and spikelet-like event rates had a weaker response to current injections: hyperpolarizing current typically was unable to silence the neuron (). As a result, no threshold could be calculated for these neurons.
At 14–16 hpi, PRV 151 infected neurons showed no relationship between AP firing rate and steady state voltage achieved by current injection (). In a few cases, strong cell body hyperpolarization was able to reduce the AP amplitude. Full APs were replaced by spikelet-like events of large amplitude (, arrow). As a result, these neurons did exhibit a weak dependence of AP firing rate on injected current, as explained in Figure S1A–F
By 24–26 hpi, PRV 151 neurons showed AP firing rates that were largely independent of injected current over a broad range (PRV 151 5/7, ). As mentioned above, the replacement of full APs with large spikelet-like events during strong hyperpolarization accounts for the apparent dependence of AP rate on injected current (Figure S1D–F
). The combined rate of spikelet-like events and full APs remained the same as found when no current was injected infected neurons starting at 14–16 hpi (Figure S1G–J
The appearance of spikelet-like events and AP firing rates independent of steady state voltage during current injections strongly suggests that APs did not initiate in the cell bodies, but rather occurred in synchrony with large spikelet-like depolarizations in electrically isolated axons.
PRV infected neurons have hyperpolarized resting membrane potentials and altered AP shapes
We next determined whether PRV infection induced changes in resting membrane potential or AP shapes. Early in infection (4–10 hpi), neurons infected with PRV 151 had resting potentials and AP shapes comparable to mock-uninfected neurons. By 14–16 hpi, despite high firing frequency, resting potentials of PRV 151 infected neurons were strongly hyperpolarized compared to uninfected control neurons (−59.5±1.9mV vs −49.5±1.1mV, p<0.01). Infected neurons remained significantly hyperpolarized until at least 24 hpi (). Additionally, the input resistance of infected neurons was reduced compared to mock infected neurons as infection progressed ().
Time course of PRV induced changes in resting membrane potential and AP shape.
The shape of spontaneous APs also was changed by PRV 151 infection. AP shape was quantified by measuring the AP amplitude (difference in voltage between the absolute peak of AP), AHP amplitude (difference in voltage between resting membrane potential and the voltage 11 ms after AP peak) and total AP height (difference between the absolute peak of AP and the voltage 11 ms after AP peak) (). During the first 10 h of PRV infection, the shape of spontaneous APs remained nearly identical to that of mock-infected neurons (). However, by 14–16 hpi, PRV 151 infected neurons showed altered AP shapes. Infected neurons' AHPs were significantly reduced in amplitude (−3.2±5.3mV range −13.1 to 4.4mV vs −10.0±3.0mV range −13.9 to −4.0mV) after 14 hours of infection, corresponding with the hyperpolarization of membrane potentials (). Total AP height remained unchanged throughout infection up to 24 hpi ().
Since these changes in shape coincided with marked hyperpolarization, we explored whether hyperpolarization alone was responsible for the change in shape by injecting hyperpolarizing and depolarizing current. As seen in , during depolarizing current injections infected neuron APs showed larger AHP amplitudes and smaller AP amplitudes, while further hyperpolarization led to smaller AHP amplitudes and larger AP amplitudes (). However, the total AP height was unchanged. This also held true of neurons across all treatments during no current injection, as shown in . These results suggest that the differences in AP shape observed during PRV infection are an indirect result of the changes in resting membrane potential.
In addition, the AP shape at threshold of AP initiation had a generally sharper inflection in infected neurons. This change is evident by inspection of the right versus left traces in . Sharp inflection points at the onset of AP initiation have also been observed in cell body recordings of APs back propagated from axons 
. The observation of brief rapid deflections supports our contention that APs were initiating in the axons of infected neurons firing at elevated rates.
PRV infection induces small and large molecular weight dye coupling, correlated membrane potentials, and late neuronal syncytia formation
The shapes and characteristics of the spikelet-like events were suggestive of electrical coupling analogous to that observed with gap junction mediated electrical synapses between neurons. We used well characterized dye transfer methods used to characterize gap junctions 
to determine if we could detect any direct diffusion pathways between two infected neurons. Both Lucifer Yellow (457 MW), the traditional low molecular weight dye used in gap junction characterization, and Texas-red conjugated dextrans of much larger molecular weight (3,000, 10,000, and 40,000 MW) were used to fill individual neurons from the patch pipette 
. In addition to dye transfer measurements, dual patch electrodes were used to record from two neurons simultaneously to analyze the temporal correlation of dye transfer, APs, and spikelet-like events in the two neurons. The effect of current injected in one cell on the membrane potential of the second cell was also evaluated to determine electrical coupling ratios.
For Lucifer Yellow (LY) dye transfer measurements, whole cell recordings using dye-filled electrodes were obtained from randomly chosen cell bodies at various time points after infection. Recordings were maintained for 10–15 minutes to allow dye to diffuse into the cell body from the electrode. The electrode was then removed and a one-hour incubation period was provided to allow time for LY to diffuse into its processes and to any connected neurons. Because LY fluorescence is not easily distinguished from GFP, we used the parent strain PRV Becker, which exhibits identical electrophysiological changes identical to those induced by PRV 151 infection (data not shown).
When mock-infected cell bodies were filled with LY containing electrodes, cell bodies and axonal processes (>20um, range 1–4 2.3±0.3) were filled with dye, but no other cell bodies or processes were filled (). Similarly, neurons infected for less than 8 hours with PRV Becker retained LY within their cell bodies and prominent processes with no evidence of transfer into other cells. However, by 9 hpi, randomly filled PRV Becker infected neurons began to share LY with other non-adjacent cell bodies (). At this time the mean number of processes labeled with dye was similar to that of mock infected neurons (range 1–15, 4.5±1.1). By 24–26 hpi, many PRV Becker infected neurons shared LY with adjacent and distant cell bodies (). In addition to dye continuity with other cell bodies, LY filled infected neurons also had a larger number of contiguous labeled processes by 24–26 hpi (range 2–19, 7.0±2.5, p<0.05). Some of these processes are likely primary axons from distant cell bodies. Dye may have transferred from the filled neurons to the axons passing over the cell body. In general, these LY transfer results demonstrate that PRV infection produced a gradual development of small pores between infected neurons that enabled flow of small molecular weight dyes.
Increasingly larger molecular weight dyes transfer between neurons as infection proceeds.
By measuring transfer of large molecular weight dyes (Texas-red dextran conjugates) known to be too large to pass between gap junctions (>2,000 MW), we observed that larger pores formed later in PRV 151 infection. Cell bodies were filled with various Texas-red dextran conjugates by diffusion from a patch electrode (3,000–40,000 MW) for 10–15 minutes prior to imaging. Because Texas-red fluorescence was easily distinguished from that of GFP, PRV 151 was used in these experiments.
In both mock-infected and early PRV infected (<24hrs) neurons, no transfer of the Texas red conjugates was obvious (). However, by 24 hpi, conjugates of at least 40,000 MW were readily transferred from filled PRV 151 infected neurons to adjacent infected neurons (, ).
Large MW dye transfer between PRV 151 infected SCG neurons.
To relate dye transfer to electrophysiological activity, dual whole-cell patch recordings were performed on several pairs of infected neurons. As shown in during the time period of elevated firing rate, recorded pairs that did not share large molecular weight Texas red- dextrans had correlated events in the two recordings. Fast depolarizing potentials exhibited by one neuron corresponded to full AP firing of the adjacent, non-filled neuron, and vice versa nearly 100% of the time (3/3, ). In these cases, DC current that was injected into one cell produced a small change in membrane potential in the second cell (Figure S2A
). The coupling ratios, a measure of the change in membrane potential in the second cell divided by that produced in the current-injected cell were >0.1. However, when large molecular weight Texas red dextrans transferred between adjacent cell bodies, the membrane voltage traces of the two recorded neurons were identical and had a coupling ratio close to 1 (, Figure S2B
). From an electrical standpoint, the membranes of the two cell bodies were contiguous.
Large MW dye transfer between infected neurons corresponds with near complete electrical synchrony.
At late times after infection (>18 hrs) signs of membrane fusion were visible by bright field microscopy. Specifically, membrane contours of the junction between adjacent cell bodies of some infected neurons relaxed. A distinct vertex is normally observed where the membranes of two adjacent cell bodies come into contact; however, as infection progressed, this vertex became increasingly difficult to resolve. This change suggested that cell bodies were fusing. To quantify the extent of fusion across a population of neurons in a culture dish, we examined the relationship between dye transfer and the angle at the vertex formed by the outer contours of adjacent neuronal cell bodies that either shared or did not share large MW dye. As shown in Figure S3A
, outer angles greater than 125 degrees between pairs unambiguously corresponded to non-dye sharing pairs, while those with larger angles indicated large MW dye sharing (examples, , arrow and Figure S3B
). We therefore used vertex angle measurements from bright field images of a population of neurons in randomly chosen fields of view as an unbiased indirect estimate (compared to selecting pairs for dual electrode recording) of fusion. As infection progressed, there was an increasing fraction of fused cells (). These data suggest clusters of infected neurons transitioned from non-dye coupled, to small MW dye coupled, to large MW dye sharing and as infection proceeded. This process continued and resulted in formation of cell body syncytia by 78 hpi ().
PRV infection leads to fusion of adjacent cell bodies into multi-nucleated syncytia.
The viral membrane protein gB is necessary for PRV induced elevated rates of AP firing and spikelet-like events, electrical coupling and neuronal fusion
As shown above, PRV infected neurons become both electrically and dye coupled leading to the formation of cell body syncytia. Syncytia formation and cell-cell fusion are well known to occur in non-neuronal cells as a result of PRV infection. This process requires the viral fusion glycoproteins gB, gH, and gL 
. A null mutation in any of these genes blocks the entry of free viral particles to initiate infection, as well as eliminates cell-cell spread and the formation of multi-nucleated syncytia. Therefore, we infected neurons with the gB null mutant virus 233, derived from PRV Becker, expressing GFP, to determine the role of virally induced fusion in the above observations 
. This mutant is propagated in gB complementing cells so viral particles produced contain gB on their host derived membrane envelopes and are able to enter a cell. Once inside, viral infection proceeds normally, but no gB is expressed and despite particle production, no further spread can occur 
Strikingly, gB null infected neurons showed no evidence of elevated rates of AP firing or spikelet-like events as late as 24 hpi, the last time point assayed (). These infected neurons had normal resting potentials until 20 hpi, but by 24 hpi resting potentials were mildly hyperpolarized compared to mock-infected neurons (−54.0±1.2mV vs −49.6±1.0mV P<0.05). However, gB null PRV infected neurons were significantly less hyperpolarized than PRV 151 infected neurons at 24 hpi (−60.8±2.3mV and −62.0±2.2mV P<0.05). The input resistance of gB null PRV infected neurons also was slightly reduced by 24 hpi, but was not significantly different from mock-infected neurons (137±28 vs 206±12MΩ, slope at −70mV n
4,14). Like the AP firing rate for mock infected neurons, AP firing rates of PRV 233 infected neurons had a positive relationship to steady state voltage (or injected current) (data not shown).
The viral membrane fusion protein gB is required for elevated rates of AP firing and fusion of infected neurons.
Individual gB null PRV infected neurons showed no signs of pore formation or fusion events. Infected neurons filled with dextran conjugated Texas red (≥3000 MW) showed no dye transfer to cell bodies with closely opposing membranes (0/4 at 18–20 hpi and 0/4 at 24–26 hpi, ). These infected neurons also showed no cell body-cell body fusion by bright field microscopy, as late as 78 hpi, the latest time point assayed ().
The gB null mutant results strongly suggest that virally mediated membrane fusion events produce pores that allow ions and dyes to flow between cell bodies. These pores increase in size over time and their appearance correlates with the induction of rapid AP firing and spikelet like events. gB null infected neurons are not hyperpolarized at early times after infection, but do exhibit modest hyperpolarization at late times.
Strain dependent onset of PRV induced elevated AP firing rates
The attenuated PRV Bartha strain and its derivatives (e.g., PRV 152) have been widely used to identify synaptically connected neurons. PRV Bartha derivatives spread only from post-synaptic to pre-synaptic neurons (retrograde spread only) and all infected animals have markedly reduced symptoms compared to virulent strains. The PRV Bartha genome carries several characterized mutations that affect anterograde spread of infection between neurons, for efficient spread of infection in the retrograde direction, and for virulence 
. It the context of this report, it is important to stress that PRV Bartha has no known mutations in the gB coding sequences and is expresses gB at equivalent levels 
Given that previous reports had indicated that in animals infected with PRV 152, neurons showed no signs of abnormal electrophysiology 
, we used our in vitro
system to compare PRV 151 and PRV 152 infected neurons. SCG neurons infected with PRV 152 produced titers of infectious virus comparable to those infected with PRV 151 and the time course of infection (as measured by GFP expression) was comparable (data not shown).
Unlike PRV 151 infections, PRV 152 infected neurons did not show elevated rates of AP firing by 8 or 16 hours post infection (). However, 10 hours later (18 hpi) PRV 152 infected neurons showed elevated AP firing rates, indistinguishable from PRV 151 infected neurons (). Similarly, PRV 152 induced AP firing rates, like those induced by PRV 151, continued unabated until at least 72 hpi.
Delayed onset of PRV induced changes by attenuated PRV 152.
The onset of PRV 152 spikelet-like events was also delayed by 10 hours, beginning at 14–16 hpi. By 18 hpi 66.7% of PRV 152 infected neurons showed spikelet-like events at rates greater than 10 Hz (vs. 0% PRV 151 infected), and 86.7% by 24 hpi (vs. 35.7% PRV 151). These rates were higher than those of PRV 151 infected neurons at comparable time points. However, the elevated mean AP firing rates were lower during periods in which PRV 152 infected neurons showed higher rates spikelet-like events compared to PRV 151 infected neurons. Despite this difference, the total event rate, including APs and spikelet-like events, in PRV 152 infected was not significantly different compared to PRV 151 infected neurons at 18 or 24 hpi ().
AP firing rates independent of steady state voltage during current injection was also delayed by 10 hours in those infected with PRV 152 (). At 4–8 and 8–10 depolarization required to reach AP threshold was not altered compared to mock infected neurons (4.1±1.9mV, 5.3±2.4mV vs 6.0±0.8mV, n
5,5,22). At 14–16 hpi, PRV 152 infected neurons started to show a similar trend as PRV 151 infected neurons 8–10 hpi, namely that hyperpolarizing currents were typically unable to completely silence the neurons (6/19). By 24–26 hpi the majority of PRV 152 infected neurons showed AP firing rates with no dependence on steady state voltage during current injection (5/8, ). A weak relationship between absolute injected current level and AP firing rate after increased AP firing rates began, explained in (Figure S4A–F
). As seen in PRV 151 infected neurons, the combined AP and spikelet-like event rate could not be controlled current injection.
PRV 152 infected neurons also showed changes in resting potential and AP shape similar to PRV 151 infected neurons. By 14–16 hpi PRV 152 infected neurons became significantly hyperpolarized compared to mock-infected neurons (−52.5±1.0mV, p<0.05, ), showed a similar change in AHP amplitude (−7.5±1.1mV range −15.9 to 7.0mV, ) and showed a sharp infection in rise to AP at threshold. Changes in AP shape corresponded with hyperpolarization of resting membrane potential. Input resistance of PRV 152 infected neurons also was significantly reduced after infection ().
Membrane fusion events occurred in PRV 152 infected neurons, but were significantly delayed. Neurons infected with PRV Bartha (the PRV 152 parent) showed no evidence of LY transfer at 9–12 hpi (0/8, ). By 18 hpi, both LY and Texas red conjugates transferred between infected neurons (, ). While transfer of large MW fluorescent dye was observed earlier between PRV 152 infected neurons than those infected with PRV 151 (18 vs 24 hpi), transfer was only observed between adjacent and not distant cell bodies (). Similarly to those infected with PRV 151, by 24 hpi PRV 152 infected neurons showed LY dye transfer to a larger number of adjacent processes (9–12 hpi, range 2–4, 2.9±0.6, 24–26 hpi range 2–19, 8.0±2.7). And, when Texas red conjugates were shared, neurons showed complete electrical synchrony (3/3 vs 0/2 not sharing). We also noted that in bright field microscopy, PRV 152 infected neurons showed more cell body-cell body fusion at 24 hpi as compared to PRV 151 infected neurons, consistent with earlier adjacent cell body dye transfer ().
PRV 152 infected neurons follow a similar progression of fluorescent dye transfer and fusion of cell bodies.
Large MW dye transfer between PRV 152 infected SCG neurons.
These experiments indicate that infection with PRV Bartha delayed, but did not eliminate increased neuronal activity and coupling. Since there is no published evidence that PRV Bartha has mutations in gB or in membrane fusion activity, these results indicate that other mutations in the PRV Bartha genome modulate action or localization of the viral membrane fusion complex in neurons.