Inhibition of type 1 IFN induction, IFN signaling, and the establishment of an antiviral state are pivotal for efficient replication of HPIV1 and many other viruses 
. We have previously shown that WT HPIV1 is able to suppress IFN-β induction and signaling, while F170S HPIV1 is unable to do so 
. As a consequence, replication of F170S HPIV1 is restricted more than 100-fold in the respiratory tract of non-human primates 
. In the present study, we took a closer look at the differences in IFN signaling between WT and F170S HPIV1, aiming to define at what step the virus-host interactions differ between these viruses.
We used African green monkey Vero cells for all of our assays except for the co-immunoprecipitation study, where 293 T cells were used because of their high transfection and protein expression efficiency. Vero cells are unable to express type 1 IFNs but are fully able to respond to exogenous IFN. Thus, one can evaluate IFN signaling in a controlled fashion by adding exogenous IFN without the confounding effects of endogenously produced IFN. This is particularly important because WT HPIV1 and F170S HPIV1 differ greatly in their ability to block IFN-β induction 
, which would complicate the distinction between effects on induction versus signaling. Vero cells also represent a susceptible host for HPIV1 infection. We also performed every experiment except the co-immunoprecipitation experiment in the context of viral infection rather than cDNA expression, which would provide an authentic environment for evaluating protein function and distribution. In Vero cells, infection with WT HPIV1 but not F170S HPIV1 inhibited the induction of an antiviral state, an indication of the extent of signaling following the addition of exogenous IFN-α, IFN-β, or IFN-γ. The level of restriction of VSV-GFP following IFN treatment was similar in uninfected versus F170S HPIV1-infected cells, indicating that this single point mutation essentially ablated the ability of the virus to inhibit signaling.
Although WT HPIV1 and WT SeV C proteins have previously been shown to block type 1 IFN signaling, most of the available information was for SeV, and it remained controversial where this block occurs (Introduction
. Here, we did not observe a reduction in Stat1 or Stat2 accumulation in cells infected with WT or F170S HPIV1, in contrast to what is seen with Rubulavirus infection 
(also see ). This is in agreement with previous reports on WT HPIV1 in human MRC5 cells 
. For WT SeV, the situation is less clear, since the loss of Stat1 was observed in murine NIH 3T3 and BALB/c fibroblasts 
but not in human HeLa or MRC5 cells 
We also found that, in response to treatment with IFN-α, -β, and -γ, the accumulation of pStat1 and pStat2 was reduced in WT and F170S HPIV1-infected cells compared to mock-infected cells. Though WT HPIV1-infected cells showed marginally less phosphorylation for Stat2 than F170S HPIV1-infected cells, we were surprised to find that the F170S HPIV1 did not differ more drastically from WT HPIV1 in this regard. Thus we concluded that the inability of the F170S mutant to block signaling in response to IFN-α, -β, and -γ could not be explained at the level of phosphorylation of Stat1 and Stat2.
Following overnight exposure of Western blots, a small amount of pStat1 was detected in the absence of IFN treatment in WT HPIV1-infected cells, but not in F170S HPIV1-infected cells. A similar IFN-independent increase in pStat1 accumulation was previously reported for WT SeV and HPIV3 
. WT SeV infection or expression of WT SeV C protein from transfected plasmid in HeLa cells also inhibited dephosphorylation of Stat1 
. Garcin et al. confirmed that neither Stat2, nor a functional IFN receptor, nor Jak1 were required for the SeV-mediated increase in pY701-Stat1 accumulation 
, supporting the idea that the increase in pStat1 resulted from virus-mediated inhibition of dephosphorylation, with the phosphorylation signal probably stemming from a background level of IFN-independent phosphorylation. Thus, our results suggest that HPIV1, like SeV, also inhibits dephosphorylation of Stat1. Since this activity was lost in F170S HPIV1-infected cells, it likely is a function of the HPIV1 C protein itself. While these observations further illustrate the greater Stat1 binding of WT C proteins versus F170S C proteins, this small amount of pStat1 present in the absence of IFN treatment likely does not contribute to inducing an antiviral state, since it is complexed with the C proteins.
Using fluorescence microscopy, we detected marked differences between WT and F170S HPIV1-infected Vero cells with regard to Stat1 and Stat2 translocation to the nucleus. WT HPIV1-infected cells remained negative for nuclear Stat1 and Stat2 following IFN-β treatment, but F170S HPIV1-infected cells permitted translocation of Stat1 and Stat2 to the nucleus. Our data for WT HPIV1 agree with results from Bousse et al. in MRC-5 cells 
, but F170S HPIV1 was not examined by these authors. The finding that a single amino acid substitution in C permits translocation strongly suggests that for WT HPIV1 the C protein is responsible for the observed block. We also found that WT C protein, but not the F170S C protein, could be co-immunoprecipitated with Stat1, as has been reported for SeV 
. Furthermore, WT C protein co-immunoprecipitated with both the phosphorylated and non-phosphorylated forms of Stat1, while co-immunoprecipitation with Stat2 was not detected. Additionally, the ratio of pStat1 to Stat1 was noticeably higher in the precipitates than in the lysates, suggesting that pStat1 was preferably bound by C′ protein. Such preferential binding of the phosphorylated form of Stat1 would be of interest, because it suggests that the C proteins preferentially target the active form of Stat 1. This also raises the possibility that the C proteins might bind to pStat1 contained in complexes such as with Stat2 and destabilize these complexes. However, further investigation using methods more suitable to measure binding affinity would be needed to investigate possible stronger association with pStat1.
Unexpectedly, we found that most of the Stat1 and C proteins in WT and F170S HPIV1-infected cells co-localized in fairly large perinuclear granules in the cytoplasm. While these complexes were observed with both viruses, the signal was somewhat less granular and dense with the F170S virus. Furthermore, for both viruses, these complexes largely co-localized with M6PR, which is a widely used marker for late endosomes. We believe this is the first report of the association of Respirovirus C proteins with large aggregates associated with the late endosome. Takeuchi et al. noted high molecular weight C protein:Stat1 complexes in SeV-infected cells based on size exclusion chromatography 
, but these complexes were not directly visualized in infected cells. In contrast to the present report, the SeV C proteins have generally been described as being associated with the plasma membrane. Marq et al. previously proposed that the SeV C proteins might be anchored to the plasma membrane by an amphipathic helix at the N-terminus of the C protein 
. Also, Sakaguchi et al. reported co-localization of C proteins with Alix/AIP1 along the plasma membrane 
, suggesting that C proteins might recruit Alix to the plasma membrane to facilitate virus budding 
. However, the significance of Alix for SeV budding is still controversial 
. For HPIV1, most of the C protein and Stat1 protein in Vero cells infected with either the WT or F170S mutant appeared to be contained in these aggregates and not at the plasma membrane. Stat2 was distributed more evenly throughout the cytosol and in contrast to Stat1 did not seem to co-localize with M6PR. We note that two of the phenotypes that we do not detect, but which are described for SeV, namely Stat1 loss 
and pronounced localization of either C proteins or Stat1 
to the plasma membrane, have both been ascribed to the N-terminal 23 amino acids of the SeV C′ protein, a region that is poorly conserved between HPIV1 and SeV.
The structure of the aggregates containing the C proteins, Stat1, and M6PR remains to be defined. Since the HPIV1 C proteins appear to lack a sequence for translocation across membrane, and since Stat1 quickly relocated to the nucleus in F170S HPIV1-infected cells following IFN treatment, it seems likely that the C protein:Stat1 complexes are located on the cytoplasmic face of late endosomes, rather than within the vesicles. Our microscopy data also suggests that the C protein might change the distribution of the late endosome. In non-infected cells, the late endosome looks polarized and sits like a cap on the nucleus. In contrast, in infected cells, distinct vesicles are frequently distributed all around the nucleus.
Stat2 did not appear to co-localize in these perinuclear aggregates, based on several observations. First, in the absence of IFN-β treatment, Stat2 appeared to be diffusely distributed in WT or F170S HPIV1-infected cells, in contrast to the Stat1 aggregates that clustered in the perinuclear space. Second, the Stat2-containing aggregates were not as well defined and not as dense as Stat1 aggregates. Third, these granules did not co-localize for the most part with M6PR. The finding that the Stat1-containing granules do not appear to contain Stat2 suggests that the C proteins bind predominantly to monomeric Stat1 rather than to the ISGF3 complex (Stat1:Stat2:IRF9). This suggestion is supported by the finding that Stat2 did not co-immunoprecipitate with C proteins, as would have been observed if the C proteins bound to ISGF3 complexes.
We previously tried to identify C protein binding partners using yeast-two-hybrid assays or immunoprecipitation, size-separation and mass-spectroscopy (unpublished data), but neither method identified Stat1 as a C protein binding partner. Only when the C′ protein (the largest form of the C proteins) was over-expressed in 293 T cells and the washing conditions for the immunoprecipitation were adjusted, could we co-immunoprecipitate Stat1 (and pStat1) protein with the WT HPIV1 C′ protein. Based on these findings, we suggest that the HPIV1 C proteins bind Stat1 (and pStat1) with only modest affinity to create an equilibrium that permits the binding partners to be exchanged and passed on frequently, and that a certain fraction of Stat1 proteins remains unbound at any time. Our studies suggest that the F170S C protein has an even lower affinity to Stat1 than does the WT C protein since it did not detectably immunopreciptate Stat1 and did not prevent Stat1 from entering the nucleus, thus permitting the establishment of an antiviral state. A higher affinity of WT C proteins towards Stat1, as compared to F170S C proteins, also was suggested by the detection of residual pStat1 in WT HPIV1-infected cells in the absence of IFN stimulation, whereas no pStat1 was detected in F170S HPIV1-infected cells in the absence of IFN stimulation, as already noted.
Depending on the particular virus, members of Paramyxovirinae may express both V and C (e.g., SeV; members of genus Morbillivirus [e.g. measles virus]; and members of Henipavirus [i.e., Nipah and Hendra viruses]), or only V (members of genus Rubulavirus [e.g. HPIV2] and Avulavirus [e.g., Newcastle disease virus]), or only C (HPIV1 and possibly HPIV3, as noted in the Introduction
). Even though the C and V proteins are completely distinct, they can have similar effects in blocking host cell innate responses. However, the mechanisms involved can vary considerably between the two proteins and between different viruses, including the mechanisms for blocking signaling from the IFN-α/β receptor. As already noted, for SeV, IFN signaling appears to be blocked by the C proteins but not the V protein, involving inhibition of Stat phosphorylation and possibly degradation of Stat1 (Introduction
). For the Rubulaviruses, the V protein was shown to promote degradation of Stat1 (parainfluenza virus 5 
and mumps virus 
) or Stat2 (HPIV2 
, also see ). For the Avulavirus Newcastle disease virus, the V protein inhibits IFN signaling by targeting Stat1 for degradation 
. The V proteins but not the C proteins of measles virus inhibit IFN signaling by inhibiting Stat1 and Stat2 phosphorylation, but degradation of Stat1 or Stat2 was not observed 
. For Hendra and Nipah viruses, the V proteins inhibit signaling by binding to both Stat1 and Stat2, inhibiting their phosphorylation and creating cytoplasmic aggregates 
. Whether these Stat1 and Stat2 aggregates with the Henipavirus V proteins have any similarity to the aggregates between Stat1 and the HIPV1 C proteins reported in the present study is not known. Thus, there is little consistency with regard to the specific mechanisms associated with C or V or within most genera.
In summary, these studies showed that both the WT HPIV1 and the F170S mutant retain the ability to inhibit phosphorylation of Stat1 and, to a lesser extent, Stat2. Thus, the inability of the F170S mutant to block IFN signaling is not due to the loss of this ability. We found that the WT C proteins bind to Stat1 and pStat1 and sequester them in aggregates that co-localize with the late endosomal marker M6PR and are little affected by IFN treatment. This sequestration appears to be the mechanism by which the HPIV1 C proteins block signaling. Stat2 did not co-localize with M6PR or co-precipitate with C proteins, indicating that it was not contained in these aggregates. While the F170S C proteins retained the ability to aggregate Stat1 in perinuclear granules, they were unable to prevent nuclear translocation following IFN treatment. Co-immunopreciptation experiments indicated that this reflected lower-affinity binding due to the mutation. These results describe the mode of action of one of the major attenuating mutations present in a live attenuated HPIV1 vaccine candidate presently being evaluated in clinical trials (ClinicalTrials.gov ID NCT00641017).