Bi-Reporter Vaccinia Virus for Tracking Viral Infections In Vitro and In Vivo

ABSTRACT Recombinant viruses expressing reporter genes allow visualization and quantification of viral infections and can be used as valid surrogates to identify the presence of the virus in infected cells and animal models. However, one of the limitations of recombinant viruses expressing reporter genes is the use of either fluorescent or luciferase proteins that are used alternatively for different purposes. Vaccinia virus (VV) is widely used as a viral vector, including recombinant (r)VV singly expressing either fluorescent or luciferase reporter genes that are useful for specific purposes. In this report, we engineered two novel rVV stably expressing both fluorescent (Scarlet or GFP) and luciferase (Nluc) reporter genes from different loci in the viral genome. In vitro, these bi-reporter-expressing rVV have similar growth kinetics and plaque phenotype than those of the parental WR VV isolate. In vivo, rVV Nluc/Scarlet and rVV Nluc/GFP effectively infected mice and were easily detected using in vivo imaging systems (IVIS) and ex vivo in the lungs from infected mice. Importantly, we used these bi-reporter-expressing rVV to assess viral pathogenesis, infiltration of immune cells in the lungs, and to directly identify the different subsets of cells infected by VV in the absence of antibody staining. Collectively, these rVV expressing two reporter genes open the feasibility to study the biology of viral infections in vitro and in vivo, including host-pathogen interactions and dynamics or tropism of viral infections. IMPORTANCE Despite the eradication of variola virus (VARV), the causative agent of smallpox, poxviruses still represent an important threat to human health due to their possible use as bioterrorism agents and the emergence of zoonotic poxvirus diseases. Recombinant vaccinia viruses (rVV) expressing easily traceable fluorescent or luciferase reporter genes have significantly contributed to the progress of poxvirus research. However, rVV expressing one marker gene have several constraints for in vitro and in vivo studies, since both fluorescent and luciferase proteins impose certain limitations for specific applications. To overcome these limitations, we generated optimized rVV stably expressing both fluorescent (Scarlet or GFP) and luciferase (Nluc) reporter genes to easily track viral infection in vitro and in vivo. This new generation of double reporter-expressing rVV represent an excellent option to study viral infection dynamics in cultured cells and validated animal models of infection.

ABSTRACT Recombinant viruses expressing reporter genes allow visualization and quantification of viral infections and can be used as valid surrogates to identify the presence of the virus in infected cells and animal models. However, one of the limitations of recombinant viruses expressing reporter genes is the use of either fluorescent or luciferase proteins that are used alternatively for different purposes. Vaccinia virus (VV) is widely used as a viral vector, including recombinant (r)VV singly expressing either fluorescent or luciferase reporter genes that are useful for specific purposes. In this report, we engineered two novel rVV stably expressing both fluorescent (Scarlet or GFP) and luciferase (Nluc) reporter genes from different loci in the viral genome. In vitro, these bireporter-expressing rVV have similar growth kinetics and plaque phenotype than those of the parental WR VV isolate. In vivo, rVV Nluc/Scarlet and rVV Nluc/GFP effectively infected mice and were easily detected using in vivo imaging systems (IVIS) and ex vivo in the lungs from infected mice. Importantly, we used these bi-reporter-expressing rVV to assess viral pathogenesis, infiltration of immune cells in the lungs, and to directly identify the different subsets of cells infected by VV in the absence of antibody staining. Collectively, these rVV expressing two reporter genes open the feasibility to study the biology of viral infections in vitro and in vivo, including host-pathogen interactions and dynamics or tropism of viral infections. IMPORTANCE Despite the eradication of variola virus (VARV), the causative agent of smallpox, poxviruses still represent an important threat to human health due to their possible use as bioterrorism agents and the emergence of zoonotic poxvirus diseases. Recombinant vaccinia viruses (rVV) expressing easily traceable fluorescent or luciferase reporter genes have significantly contributed to the progress of poxvirus research. However, rVV expressing one marker gene have several constraints for in vitro and in vivo studies, since both fluorescent and luciferase proteins impose certain limitations for specific applications. To overcome these limitations, we generated optimized rVV stably expressing both fluorescent (Scarlet or GFP) and luciferase (Nluc) reporter genes to easily track viral infection in vitro and in vivo. This new generation of double reporter-expressing rVV represent an excellent option to study viral infection dynamics in cultured cells and validated animal models of infection.
better-studied poxviruses belong to the Orthopoxvirus genus, which includes both vaccinia virus (VV), the prototypic member in the poxvirus family, and variola virus (VARV), the causative agent of smallpox. While smallpox has been eradicated, VARV still remains a pathogen of concern because of its potential use as a bioterrorism agent (2)(3)(4). Presently, much of the United States (US) population has not been vaccinated against smallpox (routine vaccination was discontinued in the 1970s) and, therefore, is susceptible to VARV infection (3,5,6). There is also a concern that monkeypox or additional Orthopoxviruses, can emerge and cause zoonotic disease due to a lack of preexisting immunity within the human population (5). VV constitutes a model virus for basic and biotechnological studies, and modification of its viral genome to express reporter genes has been vital in the study of viral gene expression, viral replication and pathogenesis, virus-host interactions, and cell entry mechanisms (7,8). The development of reporter-expressing recombinant (r)VV has also been amendable to real-time and high-throughput screening (HTS) studies (9) and allow to easily assess viral infection without the use of laborious secondary methods to identify infected cells in culture or in live animals. However, to date, reporterexpressing rVV are limited to express a single reporter gene (e.g., fluorescence or luciferase proteins) or express distinctive fluorescent proteins using different promoters to monitor gene expression patterns (10,11). This is typically because the development of reporter viruses has been largely motivated by a specific type of study, such as in vitro quantification of viral entry and replication, or by antiviral or neutralizing antibody (NAb) screening assays (9,12,13). Although these rVV expressing either fluorescent or luciferase proteins have proven extremely useful, they also have limitations based on the intrinsic properties of the reporter genes. Fluorescent proteins represent an excellent option for in vitro studies to track viral infection compared to luciferase proteins (14)(15)(16)(17). However, luciferase proteins are more sensitive and convenient in quantitative analyses than fluorescent proteins (18)(19)(20)(21). In live animals, luciferase activity can be easily tracked longitudinally using in vivo imaging systems (IVIS) and can be used as a valid surrogate of viral infection without the need of sacrificing animals (22). Contrary, fluorescent proteins represent a better option to identify the types of cells infected by the virus as they can be easily detected using fluorescence microscopy and flow cytometry assays (15,(23)(24)(25).
Here, we describe the generation of novel and stable rVV expressing both luciferase (Nluc) and fluorescence (Scarlet or GFP) reporter genes optimally designed for the easy detection of viral infection by both luminescence and fluorescence expression, respectively; and thereby, circumventing the limitations of using rVV expressing a single luminescence or fluorescent reporter gene.

RESULTS
Generation of rVV Nluc/Scarlet and rVV Nluc/GFP. The rVV Nluc/Scarlet and rVV Nluc/GFP were generated by insertion of Nluc downstream of the F13L locus and either Scarlet (rVV Nluc/Scarlet) or GFP (rVV Nluc/GFP) downstream of the A27L locus ( Fig. 1). Insertion of foreign reporter genes was accomplished by a selection method based on plaque formation that allows simultaneous introduction of two genes into the VV genome (26). The corresponding genomic regions around those genes are depicted for the reference VV WR strain and for the recombinant viruses (Fig. 1A). Notably, the insertion is directed into intergenic regions (F13L/F12L and A27L/A26L) and did not cause any additional modifications of the viral genome (Fig. 1A).
We first characterized the rVV by Western blotting (Fig. 1B). Total cell lysates from mock-, WR-, rVV Nluc/Scarlet-, or rVV Nluc/GFP-infected (MOI 0.05) BSC-1 cells were collected at 24 h postinfection and they were analyzed using antibodies specific for Nluc, Scarlet (anti-mCherry), and GFP. As expected, Western blot analysis revealed specific bands with the predicted molecular size for Nluc only in cell extracts from BSC-1 cells infected with rVV Nluc/Scarlet or Nluc/GFP (Fig. 1B). Furthermore, specific bands for Scarlet or GFP were only detected in cell extracts from rVV Nluc/Scarlet-or rVV Nluc/ GFP-infected cells, respectively (Fig. 1B). Antibodies against VV F13L and cellular actin were included as loading controls for viral infection and cellular housekeeping proteins, respectively (Fig. 1B). As expected, we detected F13L only in VV-infected cell lysates, and the levels of F13L expression were similar between WR-, rVV Nluc/Scarlet-, and rVV Nluc/GFP-infected cells (Fig. 1B). These results confirmed that each rVV expressed two reporter genes.
In vitro characterization of reporter gene expression and viral replication of rVV Nluc/Scarlet and rVV Nluc/GFP. Next, we sought to assess whether the expression of reporter genes could be directly visualized by plaque assay and fluorescence microscopy (Fig. 2). We conducted plaque assays on BSC-1 cell monolayers infected (;20 plaque forming units [PFU]/well) with WR, rVV Nluc/Scarlet, or rVV Nluc/GFP. After 2 days postinfection, plaques were detected by either crystal violet staining, fluorescence (GFP and Scarlet), or bioluminescence (Nluc) ( Fig. 2A). Crystal violet staining revealed plaques with similar sizes for WR, rVV Nluc/Scarlet, and rVV Nluc/GFP ( Fig. 2A). As expected, only rVV Nluc/Scarlet and rVV Nluc/GFP, but not WR, displayed fluorescent positive plaques, respectively, under direct fluorescent imaging ( Fig. 2A). Moreover, in the presence of Nluc substrate, both rVV Nluc/Scarlet and rVV Nluc/GFP, but not WR plaques, were detectable using bioluminescence ( Fig. 2A). When representative images were merged, fluorescent plaques colocalized with Nluc-positive plaques, indicative of rVV Nluc/Scarlet and rVV Nluc/GFP expressing both reporter genes ( Fig. 2A). Moreover, Scarlet-, or GFP-, and Nluc-positive plaques colocalized with  plaques detected by crystal violet, as indicated by red and green arrows, respectively ( Fig. 2A). Fluorescence expression of Scarlet and GFP in viral plaques from cells infected with rVV Nluc/Scarlet and rVV Nluc/GFP but not with WR, was further confirmed by imaging individual plaques under a fluorescence microscope (Fig. 2B).
To evaluate the kinetics of reporter gene expression, BSC-1 cells were infected (MOI, 0.1) with WR, rVV Nluc/Scarlet, or rVV Nluc/GFP and red or green fluorescence expression in infected cells was directly observed under a fluorescence microscope at 24, 48, 72, and 96 h postinfection (Fig. 2C). Irrespective of the fluorescent protein used, signal was first detected after 24 h, progressively increased, and peaked at 72 h postinfection, with lower expression at 96 h postinfection, probably due to leakage from cells coincident with cytopathic effect (CPE) caused by viral infection (Fig. 2C). By brightfield microscopy, WR-infected cells displayed viral-induced CPE similar to that of the rVV Nluc/ Scarlet, or rVV Nluc/GFP but, as expected, fluorescence was only detected at background levels in WR-infected cells (Fig. 2C).
Cell culture supernatants were also collected at the indicated time points to quantify viral titers (Fig. 2D) and Nluc activity (Fig. 2E). Viral titers in the tissue culture supernatants of BSC-1 cells infected with rVV Nluc/Scarlet or rVV Nluc/GFP were identical and comparable to those of WR-infected BSC-1 cells, with all viruses achieving similar viral titers at all times postinfection (Fig. 2D). When we analyzed cell culture supernatants for Nluc activity, Nluc was detected as early as 24 h postinfection in BSC-1 cells infected with rVV Nluc/Scarlet or rVV Nluc/GFP (Fig. 2E). Notably, Nluc expression levels increased in a time-dependent matter, peaking at 96 h postinfection, most likely because the CPE caused by viral infection resulted in the release and accumulation of NLuc in the tissue culture supernatants from infected cells (Fig. 2E). As expected, BSC-1 cells infected with WR did not show detectable levels of Nluc in the tissue culture supernatant, like those of mock-infected cells (Fig. 2E).
Stability of rVV Nluc/Scarlet and rVV Nluc/GFP. A critical concern with reporterexpressing recombinant viruses is their genetic instability that might lead to the loss of correct marker gene expression. To test the stability of our bi-reporter constructs, we serially passaged both, rVV Nluc/Scarlet and rVV Nluc/GFP, as well as WR, in cultured cells for a total of 5 passages and evaluated progeny virus for reporter gene expression. Results indicate that rVV Nluc/Scarlet and rVV Nluc/GFP were genetically stable since both recombinant viruses displayed the expected fluorescent (Scarlet and GFP) and bioluminescence (Nluc) expression (not shown).
In vivo kinetics of rVV Nluc/Scarlet and rVV Nluc/GFP infection. Since detecting fluorescent proteins in vivo using in vivo imaging is challenging because its intensity is often quenched in live tissues (21,27), we focused on evaluating the dynamics of rVV Nluc/Scarlet and rVV Nluc/GFP infection in mice by bioluminescence (Nluc). BALB/c mice (n = 4/group) were i.n. infected with 10 4 , 10 5 , 10 6 , and 10 7 PFU/mice of rVV Nluc/ Scarlet, rVV Nluc/GFP, or WR, and bioluminescence was monitored every 2 days, for 14 days (Fig. 4). Bioluminescent imaging of the same mice infected with 10 4 ( . Nluc expression was readily detected in mice infected with the highest doses (10 6 and 10 7 PFU) of rVV Nluc/Scarlet and rVV Nluc/GFP as early as day 2 postinfection, although the Nluc intensity signal was higher in those animals infected with 10 7 PFU of rVV Nluc/Scarlet and rVV Nluc/GFP. Nluc expression in mice infected with 10 5 PFU of rVV Nluc/Scarlet and rVV Nluc/GFP was observed after 4 days postinfection, while Nluc signal was detected after 6 days postinfection in mice infected with 10 4 PFU of rVV Nluc/Scarlet and rVV Nluc/GFP. Peak of Nluc expression in mice infected with 10 7 PFU of rVV Nluc/Scarlet and rVV Nluc/GFP was observed at 4 days postinfection. In this case, mice could not be monitored at later times postinfection since all of them succumbed to viral infection ( Fig. 4D and E). In the case of mice infected with 10 6 PFU of rVV Nluc/Scarlet and rVV Nluc/GFP, maximum levels of Nluc expression were detected at days 4-6 postinfection and could not be monitored at later times postinfection since all of them succumbed to viral infection ( Fig. 4C and E). Mice infected with 10 5 PFU of rVV Nluc/Scarlet and rVV Nluc/GFP showed maximum levels of Nluc expression between days 4-8 postinfection, and we were able to monitor Nluc levels of expression during the entire experiment since all the mice survived infection ( Fig. 4B and E). Finally, mice infected with 10 4 PFU of rVV Nluc/Scarlet and rVV Nluc/GFP showcased detectable peak in Nluc expression at days 8-12 postinfection ( Fig. 4A and E), and all the mice survived viral infection.
In vivo fluorescence and Nluc expression, and correlation with viral infection. A major advantage of our dual reporter-expressing rVV is that they harbor both bioluminescent and fluorescent reporter genes. Therefore, either one could be used as a valid surrogate of viral infection. To demonstrate a direct correlation between bioluminescent (Nluc) and fluorescent (Scarlet or GFP) in vivo and ex vivo, respectively, we infected BALB/c mice (n = 4/group) with 10 4 (Fig. 5A), 10 5 (Fig. 5B), 10 6 (Fig. 5C), and 10 7 (Fig. 5D) PFU/mice of rVV Nluc/Scarlet, rVV Nluc/GFP, or WR. We then assessed Nluc expression at days 2, 4, and 6 postinfection using in vivo imaging. Immediately after imaging, mice were euthanized, and lungs were collected to quantitate Scarlet (rVV Nluc/Scarlet) or GFP (rVV Nluc/GFP) expression in the lungs using ex vivo imaging. Similar to our previous results, we did not detect Nluc expression in mice that were infected with 10 4 PFU of either rVV Nluc/Scarlet or rVV Nluc/GFP (Fig. 5A). Some mice infected with 10 4 PFU of rVV Nluc/Scarlet or rVV Nluc/GFP showed a minimal Nluc signal at day 6 postinfection (Fig. 5A). Likewise, we could not detect Scarlet or GFP expression in the lungs of mice infected with 10 4 PFU of rVV Nluc/Scarlet or rVV Nluc/GFP, respectively (Fig. 5A). Nluc was readily detected in mice inoculated with higher doses of rVV Nluc/Scarlet or rVV Nluc/GFP (10 5 , 10 6 , or 10 7 PFU). In mice infected with 10 5 PFU, Nluc was detected by day 4 postinfection (Fig. 5B). In contrast, Nluc was detected as early as 2 days postinfection in mice infected with 10 6 and 10 7 PFU, respectively ( Fig. 5C and D). Ex vivo imaging of fluorescent expression in excised lungs correlated with levels of Nluc expression. Scarlet and GFP expression were detected by day 4 postinfection in mice infected with 10 5 PFU of rVV Nluc/Scarlet or rVV Nluc/GFP (Fig. 5B); and as early as 2 days postinfection in mice infected with 10 6 and 10 7 PFU of rVV Nluc/Scarlet or rVV Nluc/GFP, respectively ( Fig. 5C and D). We also determined viral titers from homogenized lungs (Fig. 5E). As expected, viral titers were higher in mice infected with 10 7 PFU of rVV Nluc/Scarlet or rVV Nluc/GFP, while the lowest viral titers were observed in mice infected with 10 4 PFU of rVV Nluc/Scarlet or rVV Nluc/GFP FIG 4 In vivo kinetics of bi-reporter expressing rVV: Six-to-8-week-old female BALB/c mice (n = 4) were infected i.n. with 10 4 (A), 10 5 (B), 10 6 (C), or 10 7 (D) PFU of WR (left), rVV Nluc/Scarlet (middle), or rVV Nluc/GFP (right). Nluc activity was evaluated every 2 days for 14 days (representative images of a single mouse per time point are shown). Radiance (number of photons per second per square centimeter per steradian, p/sec/cm 2 /sr) is shown for each mouse and time point. Bioluminescence signal was quantified and expressed as the total flux (log 10 protons per second, p/s) (E). Error bars indicate the mean and standard deviation of each group of mice. One-way ANOVA was used for statistical analysis; *, P , 0.05; **, P , 0.01; ***, P , 0.001; ****, P , 0.0001; n.s., no significance. The line represents the geometric mean.  5E). Median Tissue Culture Infectious Dose (TCID 50 ) values in mice infected with 10 4 ,10 5 , 10 6 , and 10 7 PFU of rVV Nluc/Scarlet or rVV Nluc/GFP were within those observed in mice infected with the same viral doses of WR (Fig. 5E). Lastly, Nluc levels of expression correlated with both Nluc signal detected in whole mice and viral titers (Fig. 5F).
Direct visualization of rVV Nluc/Scarlet and rVV Nluc/GFP in mice lungs. One of the practical applications of viruses expressing fluorescent reporter genes is the rapid and easy detection of the pathogen without the need for additional reagents to identify the presence of the virus in infected cells. In addition, the resolution of microscopic imaging using fluorescent reporters allows identifying individual positive cells without the need for antibody staining. Thus, we decided to visualize the spatial distribution of the VV and their cellular targets in formalin-fixed, paraffin-embedded lung sections of BALB/c mice (n = 4) infected with 10 7 PFU of rVV Nluc/Scarlet or rVV Nluc/GFP at days 2 and 4 postinfection by examining Scarlet and GFP expression, respectively (Fig. 6). As expected, we did not find a fluorescent signal in the lungs of mock-infected mice or in mice infected with WR (Fig. 6A). However, lung sections from rVV Nluc/Scarlet-and rVV Nluc/GFP-infected mice displayed red and green fluorescent signals, respectively. Although some epithelial bronchi were virus-free, we observed a progressive increase in fluorescent signals in the airways between days 2 and 4 postinfection (Fig. 6A). As a result of the fixation process, in certain instances, fluorescence intensity might be drastically diminished. To overcome this problem, researchers have used antibodies specific for fluorescent proteins to identify sites of viral infection. Therefore, and to further confirm our initial results assessing direct expression of Scarlet and GFP in the lungs of infected animals, we stained lung sections of mice infected with rVV Nluc/Scarlet or rVV Nluc/GFP with antibodies specific for Scarlet (mCherry) or GFP, respectively (Fig. 6B). Of note, lung samples from mock-or WR-infected mice were negative for Scarlet and GFP (Fig. 6B). In contrast, Scarlet and GFP expression were enhanced by specific antibodies and were mainly located in the airways of the lung tissues (Fig. 6B). As we predicted, viral infection was detected using a specific antibody against VV in the lung sections of all mice groups infected with WR, rVV Nluc/Scarlet, or rVV Nluc/GFP, but not in mockinfected animals (Fig. 6C). The staining patterns observed after direct visualization of Scarlet or GFP expression from reporter-expressing viruses (Fig. 6A), incubation with antibodies against fluorescent proteins (Fig. 6B), or VV (Fig. 6C) were similar and indicative of epithelial cell infections in the bronchial airways ( Fig. 6A to C). Lastly, a morphometric analysis of the area infected by WR, rVV Nluc/Scarlet, or rVV Nluc/GFP at days 2 and 4 postinfection revealed a significant 4-fold increase in the percentage of area covered by the fluorescent signal on day 4 postinfection (Fig. 6D).
Pathological changes in mice lungs infected with dual reporter-expressing rVV. We next evaluated the ability of rVV Nluc/Scarlet and rVV Nluc/GFP to induce pathological changes in the lungs of infected (10 7 PFU) BALB/c mice (n = 4) and compared to those caused by WR infection, and mock-infected animals (Fig. 7). Lungs of mice infected with either rVV Nluc/Scarlet, rVV Nluc/GFP, or WR, showed mild to moderate bronchopneumonia (Fig. 7A). Blind assessment of pathological pulmonary features confirmed our preliminary observations and showed comparable inflammatory cell infiltration around bronchi and blood vessels ( Fig. 7A and B, blue arrows). Accumulation of inflammatory cells in the intra-alveolar space was significantly higher in WR-infected mice compared to rVV Nluc/ Scarlet and rVV Nluc/GFP infected animals at day 2 postinfection, but was not statistically   (Fig. 7A and C, green arrowheads). Bronchial epithelial cell necroses were similar in mice infected with WR and dual reporter expressing rVV Nluc/ Scarlet, or rVV Nluc/GFP ( Fig. 7A and D, black arrows).

FIG 6 Legend (Continued)
Nluc/GFP. Mice were sacrificed on days 2 and 4 postinfection and lungs were harvested to assess fluorescent Scarlet and GFP expression. Representative images from lungs of mock-infected and mice infected with WR, rVV Nluc/GFP, or rVV Nluc/Scarlet viruses on days 2 and 4 without immunostaining (A), stained with antibodies against Scarlet or GFP (B) or a rabbit polyclonal antibody against VV (C) are shown. Airways are noted with an asterisk. Scale bars 100 mm. Morphometric analyses of the infected bronchi in these experiments are also shown (D). One-way ANOVA was used for statistical analysis; ***, P , 0.001; n.s., no significance. The line represents the geometric mean. One-way ANOVA was used for statistical analysis; ***, P , 0.001; n.s., no significance. The line represents the geometric mean.
CD6 int/hi ) that have active viral infection by assessing GFP reporter expression (data not shown). Consistent with the progressive spread of viral infection in the airways, we found a significant increase in the proportion of CD45-stromal cells infected from days 2 to 4 postinfection (Fig. 8A, day 2; 2.7% CD45-GFP 1 versus day 4; 52% CD45-GFP 1 ). The frequency and total number of monocytes undifferentiated macrophages (day 2; 40.5% versus day 4; 31.3%), monocyte-derived dendritic cells (day 2; 31.5% versus day 4; 11.3%) and interstitial macrophages (day 2; 12.3% versus day 4; 3.4%) infected by rVV Nluc/GFP was lower at day 4 compared to day 2 postinfection (Fig. 8A and B). In contrast, infected macrophages (day 2; 8% versus day 4; 29.2%) and neutrophils (day 2; 7.5% versus day 4; 24.8%) increased at day 4 postinfection, which may be associated with the differentiation of monocytes into macrophages in the infected lungs ( Fig. 8A  and B). Yet, both rVV Nluc/GFP and WR efficiently induced an innate immune response in the lung of infected mice. rVV Nluc/GFP induced significant accumulation of macrophages and interstitial macrophages (day 2), monocyte-derived dendritic cells, and monocytes undifferentiated macrophages (day 4) compared to WR-infected mice. Intriguingly, we observed a decrease of total macrophages and neutrophils at day 4 postinfection that may be associated with the increase in viral infection (Fig. 8B).
Next, we performed immunofluorescence staining to identify the spatial distribution of innate cells at the site of infection. To visualize the location of neutrophils relative to the infected airways, we stained lung sections with antibodies specific for Ly6G, VV, and GFP to detect neutrophils, WR, and rVV Nluc/GFP, respectively. In the lungs of mock-infected mice, we observed most neutrophils accumulated in the interstitial space, while very few were near the airways (Fig. 8C, left panels). In contrast, many neutrophils accumulated around airways in mice infected with WR or rVV Nluc/GFP (Fig. 8C, middle and right panels). Consistent with our preliminary observations, the number of neutrophils significantly increased around the airways in mice infected with WR or rVV Nluc/GFP (Fig. 8D). The number of neutrophils infiltrating the peri-bronchial areas was similar in mice infected with either WR or rVV Nluc/GFP (Fig. 8D).

DISCUSSION
Previous studies, including ours, have described reporter-expressing rVV encoding either fluorescent or luciferase reporter genes to study the biology of VV and to evaluate the efficacy of new antivirals or NAbs to inhibit or neutralize, respectively, viral infection (24,25,28). However, and based on the different properties of fluorescent and luciferase proteins, the use of these reporter-expressing rVV is limited to the properties of the reporter gene (15,27,29). In vitro, fluorescent proteins represent a better option to detect the presence of rVV in infected cells (14)(15)(16)(17). However, luciferase proteins represent a better tool for quantitative purposes (19)(20)(21)30). In vivo, luciferase reporter genes represent the best option for whole-animal imaging (20,27), while fluorescent proteins are better for ex vivo imaging and identifying viral cellular targets (15,31). Thus, rVV expressing only one reporter gene does not allow to take advantage of the different properties of fluorescent and luciferase proteins and, therefore, the properties of the reporter fluorescent or luciferase proteins must be carefully considered in the experimental design.
Here, we have described the generation of novel rVV stably expressing two different reporter fluorescent and luciferase genes to overcome the current limitations of using single reporter-expressing rVV (Fig. 1). We previously demonstrated the feasibility and advantages of generating bi-reporter influenza A virus (IAV) (32,33) and the prototype arenavirus lymphocytic choriomeningitis virus (LCMV) (34). This new dual reporterexpressing rVV possess the advantages of both fluorescent and luciferase reporters, which could provide promising applications for the study of VV in vitro and in vivo.
We selected Nluc based on its intense brightness, small size and ATP-independence (18,23). We chose Scarlet or GFP, in combination with Nluc, to easily track viral infections based on the red or green fluorescent signal, respectively. The use of Scarlet (red) or GFP (green) fluorescent proteins would allow to track viral infections in genetically engineered cultured cells or animal models of infection. We selected GFP for its optimal brightness, and Scarlet for its excitation/emission at longer wavelengths, low toxicity and deeper light penetration in animal tissues (35). Notably, by introducing both fluorescent and bioluminescence reporter genes in the same virus, these rVV exploit the advantages of both fluorescent and bioluminescence proteins without the limitations of using one or the other.
We found that our bi-reporter-expressing rVV have similar growth kinetics and plaque phenotype in cultured cells than the parental WR strain (Fig. 2). Importantly, only rVV expressing both reporter genes were detectable by fluorescence microscopy or bioluminescence due to its expression of GFP or Scarlet, and Nluc, respectively. Infection with rVV expressing both reporter genes was visualized in real time, without the need of complex protocols (e.g., staining with antibodies) needed to detect the presence of parental VV strains in infected cells. Moreover, reporter gene expression correlated with viral replication, demonstrating the feasibility of using either the fluorescent or luciferase reporter genes as valid surrogates of infection. In 2018, the United States FDA-approved the use of tecovirimat (TPOXX) to treat poxvirus infections (3). However, additional antivirals for the efficient treatment of poxvirus infections are urgently needed. Thus, discovering novel antivirals and implementing rapid and sensitive screening assays amenable to HTS will accelerate the identification and characterization of novel anti-poxvirus compounds. Antiviral drug discovery against poxvirus infection will benefit of rVV expressing both fluorescent and luciferase proteins where identification of antivirals will be based on orthogonal assays based on inhibition of both reporter genes, similar to our previous studies with bi-reporter IAV (33) and LCMV (34). Likewise, these bi-reporter-expressing rVV represent an excellent option to assess the neutralizing capacity of antibodies in HTS settings, as previously described for IAV (33,36).
In vivo, the dual reporter-expressing rVV retained parental WR-like morbidity and mortality, with animals losing body weight and succumbing to viral infection, respectively, similar to those infected with WR (Fig. 3). Notably, the bi-reporter expressing rVV represent a powerful tool to visualize viral infection in vivo (Nluc) (Fig. 4) or ex vivo (Scarlet and GFP) (Fig. 5). When we imaged mice infected with rVV Nluc/Scarlet or rVV Nluc/GFP using IVIS, Nluc signal was easily detectable and quantifiable, revealing the dynamics of VV infection using different doses of viral inoculum (Fig. 4). Importantly, Nluc activity was time and viral dose dependent and allowed us to longitudinally follow and measure viral replications dynamics in the same live animals during the progression of natural viral infection. Likewise, the excised lungs from mice infected with rVV Nluc/Scarlet and rVV Nluc/GFP displayed Scarlet or GFP expression, respectively, where fluorescent signals were also time and viral dose-dependent (Fig. 5). In addition, fluorescent protein expression allowed us to identify individually infected cells using fluorescent imaging and flow cytometry (Fig. 6) and to assess the pathology of viral infections in the lungs of infected mice (Fig. 7). Importantly, we observed a spatial and temporal correlation between Nluc and fluorescent expression (Scarlet or GFP), which correlated with viral replication in the lungs of infected mice (Fig. 5). Finally, we identified the major primary cellular targets of VV in vivo by combining the identification of GFP1 cells with flow cytometry-based approaches (Fig. 8).
Altogether, rVV stably expressing both fluorescent and luciferase reporter genes represent an excellent tool for studying the biology of VV in vitro or in vivo. Expression of both reporter genes adds the advantages of using individual fluorescentor luciferase-expressing rVV and represent an excellent option for the identification of prophylactic and/or therapeutics against poxviruses.
Viruses. Vaccinia virus (VV) Western Reserve (WR) strain was obtained from the American Type Culture Collection (ATCC VR-119) and routinely propagated in BSC-1 cells. Previously described vDA27-DF13 is a VV mutant in which most of the coding sequence of genes A27L and F13L has been deleted (26).
Plasmids. Plasmid pRB-NLuc, designed to express the NLuc gene downstream of the F13L gene has been previously described (26). Plasmid pA.S-TagGFP2 for insertion of GFP downstream of the A27L gene was constructed previously (26). A similar construct containing the gene coding for the red fluorescent Scarlet protein, derived from mCherry (37), was subcloned from plasmid pRB-Scarlet (26,38) into the plasmid pA-LE-RGR (Sanchez-Puig and Blasco, unpublished), which contains A27L recombination sites flanking the A27L gene, to generate pA.S-Scarlet. For optimal expression, mScarlet was placed under a strong poxviral early-late synthetic promoter (39).
Generation of recombinant vaccinia viruses (rVV). rVV were generated as previously described (26). Briefly, BSC-1 cells (6-well plate format, 10 6 cells/well), were infected at a multiplicity of infection (MOI) of 0.05 PFU/cell with vDA27-DF13. After 1 h viral adsorption, virus inoculum was removed and cells were then transfected with 2 mg of plasmid DNA using FuGeneHD (Promega) according to the manufacturer's recommendations. The rVV Nluc/GFP was generated by transfecting pRB-NLuc and pA.S-TagGFP2 plasmids. The rVV Nluc/Scarlet was obtained by transfecting pRB-NLuc and pA.S-Scarlet plasmids. After 2 to 3 days, cells were harvested, and rVV Nluc/GFP and rVV Nluc/Scarlet were isolated by three consecutive rounds of plaque purification. Viral stocks were generated in BSC-1 cells and viral titers were determined by standard plaque assay (PFU/ml).
Plaque assay. To assess plaque phenotype of the rVV, BSC-1 cells (6-well plate format, 10 6 cells/well) were infected with approximately 20 PFU per well. After 1 h infection, the medium was removed, and the infection was maintained for 48 h at 37°C under semisolid medium consisting of EMEM-2% FBS containing 1% low-melting point agarose. For fluorescence imaging, plates were photographed using the ChemiDoc TM XRS 1 (Bio-Rad). Green and red fluorescent signals from GFP and Scarlet, respectively, were photographed using the software settings indicated for SYBR green using 1 s and 2 s exposure, respectively. For detection of Nluc, 1 ml of PBS with 100 ml of substrate (Nano-Glo luciferase assay system, Promega) was added per well and immediately photographed with the software option for Chemiluminescence with 40 s of exposure. Cell monolayers were subsequently fixed with 1 ml of 4% formaldehyde for 30 min at room temperature. After removing the semisolid medium, plaques were stained with a solution of 5% crystal violet in methanol (vol/vol) (40) and photographed using White Epi Illumination for 0.1 s. Viral growth kinetics. Viral growth kinetics were evaluated in BSC-1 cells (6-well plate format, 10 6 cells/well, triplicates) infected at an MOI of 0.01. Viral replication at various times postinfection (24,48,72, and 96 h), was evaluated by phase contrast and fluorescence microscopy. In addition, at each time point, tissue culture supernatants were collected to determine viral titers and Nluc activity. Viral titers were determined by standard plaque assay (PFU/ml) followed by staining with crystal violet (40). Nluc activity was quantified using the Nano-Glo luciferase assay (Promega) following the manufacturer's recommendations using an EnSight (Perkin Elmer) bioluminescence plate reader. Microsoft Excel software was used to determine the mean value and standard deviation (SD).
Fluorescence microscopy. GFP and Scarlet fluorescence expression were detected using a Nikon Eclipse TE-2000-E inverted microscope. GFP expression was assessed using 465-495 nm (excitation) and 515-555 nm (emission) filters. Scarlet expression was detected using 515-560 nm (excitation) and 600-650 nm (emission) filters. Images were acquired with a Photometrics PRIME SCMOS monochrome camera.
Mice experiment. All animal experiments were approved by the University of Rochester Institutional Biosafety (IBC) and Animal Care and Use (IACUC) Committees, which are in accordance with recommendations found in the Guide for the Care and Use of Laboratory Animals of the National Research Council (41). Six-to-8-weeks-old female BALB/c mice were purchased from the National Cancer Institute (NCI) and maintained at the University of Rochester animal care facility under specific-pathogen-free conditions. Mice were anesthetized intraperitoneally (i.p.) with ketamine-xylazine (100 mg/kg ketamine and 10 mg/kg xylazine) and infected intranasally (i.n.) with the indicated viral PFU. Mice (n = 6/group) were daily monitored for changes in morbidity (body weight loss) and mortality (survival) for 14 days (42)(43)(44). Mice losing more than 25% of their initial body weight were considered to reach their endpoint and were humanely euthanized with carbon dioxide (CO 2 ) and confirmed by cervical dislocation. The 50% mouse lethal dose (MLD 50 ) was determined using the Reed and Muench method (45).
An IVIS Spectrum multispectral imaging instrument (Caliper Life Sciences, Inc.) was used for in vivo bioluminescence imaging of live mice and ex vivo imaging of lungs collected from mock infected and VV infected animals. Six-to-8-weeks-old female BALB/c mice (n = 4/group/day) were mock-infected (PBS) or infected i.n. with the indicated PFU. At the indicated days postinfection, mice were anesthetized with ketamine-xylazine (100 mg/kg ketamine and 10 mg/kg xylazine) and then retro-orbitally injected with 100 ml of Nano-Glo luciferase substrate (Promega) diluted 1:10 in PBS. Subsequently, mice were placed in an XIC-3 isolation chamber (Perkin Elmer) and imaged. Bioluminescence data collection and analysis were conducted using the Living Image software (v4.5; PerkinElmer). Flux measurements (Log 10 photons per second (p/s) were obtained from the region of interest (ROI) around the whole body of each mouse). A bioluminescence intensity (radiance; number of photons per second per square centimeter per steradian; p/sec/cm 2 /sr) scale bar is displayed for each figure. Right after imaging, mice were removed from the XIC-3 isolation chamber and euthanized with a lethal dose of 2,2,2-tribromoethanol and exsanguination. Whole lungs were excised and washed with PBS. Scarlet or GFP expression were analyzed in the IVIS as previously described (15,32,33,42). Living Image (v.4.5) software was used to acquire and analyze images to determine the radiant efficiency of the ROI. Fluorescence signals were normalized to those collected from mock-infected animals. Excised lungs from days 2, 4, and 6 postinfection were homogenized, and viral titers were determined by Median Tissue Culture Infectious Dose (TCID 50 ).
Histopathology and pathology scoring. Six-to-8-weeks-old female BALB/c mice (n = 4) were infected (i.n.) with 10 7 PFU of the indicated viruses. Mice were sacrificed on days 2 and 4 postinfection, and lungs were slowly perfused with 10% neutral formalin and fixed for 2 days. Next, the lungs were washed three times in PBS before being gradually dehydrated by sequential immersion in 70%, 95%, and absolute ethanol. After that, lungs were transferred to xylenes and embedded in paraffin. Lungs were sectioned (5 mm) and placed on a hot plate to dissolve the paraffin, followed by sequential rehydration with 95% ethanol, 75% ethanol, and ddH 2 O. Once the lungs were rehydrated, slides were mounted with Vectashield media containing DAPI (Vector Laboratories) to stain the nucleus. Lung sections were incubated overnight with Alexa Fluor 448 rabbit anti-GFP (Thermo Fisher Scientific), mouse anti-RFP (Thermo Fisher Scientific) or a rabbit polyclonal antibody against VV A33R protein (BEI Resources, NR-628) to demonstrate that fluorescent reporter proteins and viruses can be alternatively detected with antibodies. The following day, slides were incubated with rabbit anti-AF488 to amplify the GFP signal, Cy3 donkey anti-mouse IgG (Jackson ImmunoResearch Laboratories) to detect mouse monoclonal against Scarlet and with Alexa Fluor 647 donkey anti-rabbit IgG to detect the rVV. Slides were washed in PBS and mounted with Vectashield with DAPI (H-1200, Vector Laboratories). Representative pictures were taken with an Axioplan Zeiss Microscope and recorded with a Hamamatsu camera. Five randomly chosen independent anatomical locations from the lung sections were microscopically evaluated for histopathologic lesion scoring as previously described (46,47). The histopathologic lesion scores included three different criteria (peri-bronchial & perivascular inflammation, intra-alveolar inflammation, and bronchial epithelial cell necrosis) and were graded on the basis of lesion severity as follows: grade 0 = no histopathological lesions; grade 1 = mild; grade 2 = moderate; grade 3 = marked; and, grade 4 = severe. A trained veterinary pathologist blindly performed all examinations at University of Rochester.
Genetic stability of rVV. To test the genetic stability of rVV, viruses were serially passaged 5 times in BSC-1 cells (6-well plate format, 10 6 cells/well) and incubated for 48 h until cytopathic effect (CPE) was observed. First passage was initiated using an MOI of 0.1 and the inoculum of the successive passages were 1:20 dilutions of the crude virus preparation from the previous passage. After 5 consecutive passages, viruses were monitored for the presence of fluorescence and luminescence using standard plaque assay. At least 50 plaques were counted for each virus.