**3. Results**

### *3.1. Astrocytes Are Readily Infected by both Field Viruses and Lab Strains in Mixed Primary Brain Cell Cultures*

To investigate whether field and lab-adapted RABVs di ffer in their ability to infect primary neurons and non-neuronal astrocytes, hippocampal brain cells were prepared from neonatal rats and were cultivated as mixed cultures containing neurons and glial cells. After 13 days of cultivation, the cultures were infected with 10<sup>3</sup> infectious units of the recombinant field viruses rRABV Dog and rRABV Fox, and the lab-adapted strains rCVS-11 and SAD L16. After 24 h of infection, the cells were fixed and analyzed by confocal laser scanning microscopy using indirect immunofluorescence stainings against RABV nucleoprotein N, neuron marker MAP2 (microtubule-associated protein 2) and astrocyte marker GFAP (glial fibrillary acidic protein). All viruses infected both MAP2-positive neurons and GFAP-positive astrocytes, as demonstrated by the formation of RABV-positive cytoplasmic inclusion bodies (Figure 1). These data indicated that there was no obvious di fference between the tested field and the lab-adapted viruses in their ability to infect neurons and non-neuronal astrocytes.

**Figure 1.** In vitro infection of primary hippocampus neurons and astrocytes by field and lab rabies virus (RABV). (**<sup>a</sup>**–**d**) Indirect immunofluorescence detection of RABV nucleoprotein N (red) and microtubule-associated protein 2 (MAP2) (green). Nuclei were counterstained with Hoechst33342 (blue). (**<sup>e</sup>**–**f**) Indirect immunofluorescence detection of RABV nucleoprotein N (red) and glial fibrillary acidic protein (GFAP) (green). Nuclei were counterstained with Hoechst33342 (blue). Negative controls for RABV detection are provided in Supplementary Figure S1.

Although all four viruses were able to infect GFAP-positive cells in the hippocampal cell cultures, the majority of the infected cells were MAP2-positive neurons. To quantitatively compare astrocyte infection between viruses, the number of RABV N-positive cells was counted for an area of 25 mm2, and the percentage of GFAP-positive and -negative RABV-infected cells was determined (Table 1). All

four viruses resulted in infection of GFAP-positive cells at levels ranging from 4.9% to 6.7%. Although a higher percentage of GFAP-positive cells could not be excluded because a high number of cells could not be classified into either positive or negative for GFAP, these results indicated that all four viruses could readily establish an infection in cultivated astrocytes and did not substantially differ in their ability to do so.

**Table 1.** Infection of GFAP- and MAP2-expressing cells in hippocampus cell cultures. RABV-infected cells were counted after indirect immunofluorescence detection of RABV N protein and subdivided in cells that were either negative (−) or positive (+) for GFAP/MAP2. Because of overlaying positive and negative cells in GFAP and MAP2 stainings, a clear assignment was not always possible. Such cells were classified as uncertain. CVS: Challenge Virus Street.


*3.2. Field RABV Infects Neurons and Astrocytes In Vivo as Demonstrated by High-Resolution 3D Analysis of Infected Brain Tissue*

In order to test whether the primary cell culture infection experiments are comparable to the more complex in vivo situation, a sample preparation and imaging pipeline was established, which allows three-dimensional, high-resolution confocal laser scanning image acquisition from RABV-infected tissue for 3D reconstruction and quantification of infected cells. To this end, 1 mm-thick brain slices (Figure 2a) from clinically diseased rRABV Fox field virus-infected mice (i.m. infection route) were immunostained for RABV P, GFAP, and NeuN, optically cleared, and imaged. Acquisition of z-stacks by confocal laser scanning microscopy resulted in z-volumes of 400 μm × 400 μm × 50–100 μm (x, y, and z) from different areas of the infected brain. RABV-infected neurons of variable morphologies were detected by presence of NeuN and RABV P protein accumulation in neuronal cell bodies and neurites (Figure 2b–e; details in Supplementary Figure S2). Astrocytes were detected by filamentous GFAP-positive structures (Figure 2b–g).

Notably, besides accumulation of RABV P in NeuN-positive neurons, P also accumulated at GFAP-positive filaments (Figure 2b,c,f,g; white arrows), indicating a robust infection of astrocytes by field virus rRABV Fox. Evaluation of brain areas with differing localization patterns of the highlighted cellular subpopulations (Figure 2b,c) revealed that astrocyte infections were not restricted to particular regions of the brain but appeared in all imaged neuron layers or brain areas.

Because of the complex 3D morphology of both neurons (long axons in Figure 2b,c) and astrocytes (filamentous structure of GFAP signals in Figure 2f,g), 3D reconstruction of the imaged tissues (Figure 2d,e,g; Supplementary Videos S1 and S2) was performed in order to achieve reliable visualization and quantification of cells in downstream analyses.

**Figure 2.** 3D immunofluorescence imaging of field RABV-infected neurons and astrocytes in a mouse brain. (**a**) Workflow for 3D immunofluorescence imaging, including vibratome sectioning into 1 mm slices, pretreatment, immunostaining, and subsequent optical clearing with organic solvents. Confocal imaging and acquisition of cleared tissue slices was done in custom-made imaging containers (see lower image). (**b**,**<sup>c</sup>**) Maximum z- and (**d**–**<sup>e</sup>**) 3D projections of z-stack (x, y, z = 400 μm, 400 μm, 59 μm for D and 400 μm, 400 μm, 103 μm for E) after indirect immunofluorescence for RABV phosphoprotein P (red), GFAP (green), and NeuN (blue). White arrows in Figure 2b indicate RABV P and GFAP-positive astrocytes. (**f**) Maximum projection of detail from Figure 2b (see white box) with GFAP-positive cell (green) and associated RABV P fluorescence (red). (**g**) 3D projection of detail view from Figure 2f.

### *3.3. Similar Levels of Field Virus Infection in both Neurons and Astrocytes in the Infected Mouse Brain*

Whereas immunofluorescence analysis in Figure 2 clearly demonstrated infection of astrocytes and neurons, they also revealed that only a fraction of both cell types was positive for RABV P. To quantify the ratio of infected and non-infected neurons and astrocytes, 3D object segmentation and counting was performed for NeuN- (Figure 3a,b) or GFAP (Figure 3d,e)-positive cells. The object map was merged with their respective RABV fluorescence (Figure 3c,f; Supplementary Video S3), and the number of RABV-positive neurons and astrocytes was determined by manual counting of RABV P- and cell marker-positive cells. For the z-stack shown in Figures 2b and 3, total numbers of 762 neurons and 272 astrocytes were counted (Table 2, region 2), of which 16 and 18 were RABV positive, respectively.

**Figure 3.** Quantification of RABV-infected neurons and astrocytes. (**a**) Maximum z-projection of objects map for NeuN-positive neurons generated from a confocal z-stack (see Figure 2b). Individual neurons in the z-stack were identified by NeuN-specific fluorescence and converted to objects. Numbers indicate individual cell counts (for improved legibility of the individual numbers, refer to the enlarged details in Supplementary Video S3). n = 762 neurons in a volume of 400 μm × 400 μm × 59 μm. The object colors indicate different z-positions (darker colors in the back and brighter colors in the front). (**b**) Detail of area indicated by white box in Figure 3a. Green arrows indicate RABV-positive neuron cell bodies. (**c**) Overlay of objects map (greyscale) with RABV P signals allows identification and counting of rRABV Fox-infected neurons. (**d**) Maximum z-projection of objects map for GFAP-positive astrocytes generated from a confocal z-stack (see Figure 2b). Individual astrocytes in the z-stack were identified by GFAP-specific fluorescence and converted to objects. Numbers indicate individual cell counts. n = 272 astrocytes in a volume of 400 μm × 400 μm × 59 μm. The object colors indicate different z-positions (darker colors in the back and brighter colors in the front). (**e**) Detail of area indicated by white box in Figure 3d. Green arrows indicate RABV-positive astrocytes. (**f**) Overlay of objects map (greyscale) with RABV P signals allows identification and counting of rRABV Fox-infected astrocytes.



Analysis of six z-stacks from di fferent RABV-infected brain areas of the same sample led to the detection of 11,089 neurons and 2042 astrocytes, of which 3.9% and 7.0% were RABV positive, respectively (Table 2). The fraction of RABV-positive cells ranged from 2.1% to 6.1% (SD: ± 1.4) for neurons and 1.8% to 17.0% for astrocytes (SD: ± 4.9) (Table 2). These data indicated that rRABV Fox infects astrocytes and neurons in the mouse brain to comparable levels.

Observation of three di fferent rRABV Fox-infected mice revealed that the infection level for astrocytes di ffered between single animals from 1.2% to 15.6% (SD: ± 6.7) (Supplementary Table S1). Nevertheless, in spite of some variance between individual mice and/or between di fferent analyzed brain regions, detection of infected non-neuronal astrocytes in all animals at surprisingly high levels indicates that astrocyte infection by RABV has been underestimated thus far.

### *3.4. Astrocyte Infection by RABV Depends on the Type of Virus (Field vs. Lab-Adapted).*

To compare the astrocyte tropism of field and lab-adapted viruses after i.m. inoculation, brains of mice infected with three di fferent field viruses (rRABV Fox, rRABV Dog, and rRABV Rac) and two lab-adapted viruses (rCVS-11 and ERA) were analyzed. SAD L16 was excluded from these analyses since it is not able to induce clinical signs after i.m. inoculation (Supplementary Figure S3d).

Whereas rRABV Fox and rRABV Dog caused disease even after infection with a very low virus dose [42], infections with rRABV Rac, rCVS-11, and ERA did not show any clinical signs at a dose of 10<sup>2</sup> TCID50 (Supplementary Figure S3). High dose infection with the ERA strain (10<sup>5</sup> TCID50) led to 100% disease development (Supplementary Figure S3c), whereas high dose infections with rRABV Rac and rCVS-11 only caused disease in 50% and 16.7% of the infected mice, respectively (Supplementary Figure S3a,b). Accordingly, available tissue samples for the latter two viruses were limited to two and one infected brain. Imaging and quantification of at least six confocal z-stacks from di fferent RABV-infected regions per infected brain was performed.

RABV infections were readily detectable in the brain for all five viruses (Figure 4; Supplementary Video S4). Whereas RABV P antigen accumulated at GFAP-positive structures in rRABV Fox, rRABV Dog, and rRABV Rac-infected brains (Figure 4a,c,e), similar accumulations were not observed in rCVS-11 or ERA-infected brains, in which only infection of NeuN-positive neurons was detected (Figure 4b,d).

Infections with rRABV Fox, rRABV Dog, rRABV Rac, rCVS-11, and ERA led to a mean of 8.3%, 7.3%, 18.9%, 6.9%, and 15.1% RABV-positive neurons (Figure 5; Supplementary Table S1), respectively, demonstrating a comparable level of neuron infection by all five viruses in clinically diseased mice. However, significant di fferences were observed for the astrocyte infections. Whereas rRABV Fox, rRABV Dog, and rRABV Rac infected 7.6% (SD: ± 6.7), 10.1% (SD: ± 7.7), and 16.5% (SD: ± 15.0) of the astrocytes, no RABV-positive astrocytes were detected in rCVS-11 and ERA-infected samples (Figure 5; Supplementary Table S1). These data indicated that, in contrast to the tested field viruses, the lab-adapted RABVs rCVS-11 and ERA did not infect astrocytes to detectable levels in vivo after i.m. inoculation.

### *3.5. Confirmation of the Specific Astrocyte Tropism of Field RABV, ERA, and SAD L16 after i.c. Inoculation and Route-Dependent Astrocyte Infection by rCVS-11*

To test whether the inoculation route a ffects astrocyte infection and whether the highly attenuated SAD L16 virus is comparable to the SAD vaccine progenitor strain ERA, brain samples from two animals i.c.-infected with rRABV Fox, rRABV Dog, rCVS-11, or SAD L16 and one infected with the ERA strain were analyzed.

With astrocyte infections at frequencies of 10.9% (SD: ± 10.3), 11.6% (SD: ± 5.6), and 27.2% (SD: ± 12.8) (Figure 6; Supplementary Table S2), rRABV Fox, rRABV Dog, and rRABV Rac led to robust astrocyte infection via the i.c. route and thus confirmed their ability to establish astrocyte infection in vivo (Figure 7a,c,e). Lack of detectable astrocyte infection for the ERA and SAD L16 further confirmed that these viruses are not able to infect astrocytes to a detectable level (Figure 7d,f and Figure 6), even after direct virus administration into the brain.

**Figure 4.** Comparison of field and lab RABV-infected brains after intramuscular (i.m.) infection with rRABV Fox, rRABV Dog, rRABV Rac, rCVS-11, and Evelyn Rokitnicki Abelseth (ERA). (**a–e**) Maximum z-projections of z-stacks [x, y = 400 μm, 400 μm (**a–e**); z = 59 μm (**<sup>a</sup>**, rRABV Fox), 66 μm (**b**, rCVS-11), 49 μm (**c**, rRABV Dog), 100 μm (**d**, ERA) and 89 μm (**<sup>e</sup>**, rRABV Rac)] after indirect immunofluorescence for RABV phosphoprotein P (red), GFAP (green), and NeuN (blue). To improve visualization of the maximum z-projections, some z-stacks were reduced in thickness. For the full z-stacks, refer to Supplementary Video S4. Insets in Figure 4a,c,e show RABV P accumulation (red) at GFAP-positive cells (green). Insets in Figure 4b,d show NeuN- (blue) and RABV P (red)-positive neurons. For the individual channels of the detail images, see Supplementary Figure S4.

**Figure 5.** Percentage of field and lab RABV-infected neurons and astrocytes after i.m. inoculation. Per virus, 3 to 9 × 10<sup>3</sup> astrocytes and 1.5 to 4 × 10<sup>4</sup> neurons were counted in 19 (rRABV Fox), 20 (rRABV Dog), 12 (rRABV Rac and ERA), and six (rCVS-11) independent confocal z-stacks in three (rRABV Fox and rRABV Dog), two (rRABV RAC and ERA), and one (rCVS-11) animal (see Supplementary Table S1). Each dot represents the frequency of infected astrocytes or neurons in an analyzed z-stack. Mean values are provided as horizontal lines. \* *p* ≤ .05; \*\* *p* ≤ .01; \*\*\* *p* ≤. 001 (two-way ANOVA with Tukey's multiple comparison test).

**Figure 6.** Percentage of field and lab RABV-infected neurons and astrocytes after i.c. inoculation. Per virus, 5 to 9 × 10<sup>3</sup> astrocytes and 1.3 to 3 × 10<sup>4</sup> neurons were counted in 12 (rRABV Dog, rRABV Fox, rRABV Rac, rCVS-11, and SAD L16) and six (ERA) independent confocal z-stacks in two (rRABV Fox, rRABV Dog, rRABV Rac, rCVS-11, and SAD L16) and one (ERA) animal (see Supplementary Table S2). Each dot represents the frequency of infected astrocytes or neurons in an analyzed z-stack. Mean values are provided as horizontal lines. \* *p* ≤ .05; \*\* *p* ≤. 01; \*\*\* *p* ≤. 001; n.s. = not significant (two-way ANOVA with Tukey's multiple comparison test).

**Figure 7.** Comparison of field and lab RABV-infected brains after i.c. infection. (**<sup>a</sup>**–**f**) Maximum z-projections of z-stacks [x, y = 400 μm, 400 μm (**<sup>a</sup>**–**f**); z = 21 μm (**<sup>a</sup>**, rRABV Fox), 48 μm (**b**, rCVS-11), 74 μm (**<sup>c</sup>**, rRABV Dog), 98 μm (**d**, ERA), 75 μm (**<sup>e</sup>**, rRABV Rac) and 67 μm (**f**, SAD L16)] after indirect immunofluorescence for RABV phosphoprotein P (red), GFAP (green), and NeuN (blue). Insets in Figure 7a,b,c,e show RABV P accumulation (red) at GFAP-positive cells (green). To improve visualization of the maximum z-projections, some z-stacks were reduced in thickness. Insets in Figure 7d,f show NeuN- (blue) and RABV P (red)-positive neurons. For the individual channels of the detail images, see Supplementary Figure S5.

Notably, in contrast to the i.m. inoculation route and to the other i.c.-inoculated lab-adapted RABVs SAD L16 and ERA, 13.4% (SD: ± 10.4) of the astrocytes were RABV-positive in rCVS-11-infected animals, indicating that, in the case of rCVS-11, the infection route has a substantial influence on astrocyte infection in the brain. However, the ratio of rCVS-11-infected neurons (8.7%; SD: ± 4.0) remained comparable to i.m. infections.
