Transcriptome Profile During Rabies Virus Infection: Identification of Human CXCL16 as a Potential New Viral Target

Abstract

Rabies virus (RABV), the causative agent for rabies disease is still presenting a major public health concern causing approximately 60,000 deaths annually. This neurotropic virus (genus Lyssavirus, family Rhabdoviridae) induces an acute and almost always fatal form of encephalomyelitis in humans. Despite the lethal consequences associated with clinical symptoms of rabies, RABV limits neuro-inflammation without causing major histopathological lesions in humans. Nevertheless, information about the mechanisms of infection and cellular response in the central nervous system (CNS) remain scarce. Here, we investigated the expression of inflammatory genes involved in immune response to RABV (dog-adapted strain Tha) in mice, the most common animal model used to study rabies. To better elucidate the pathophysiological mechanisms during natural RABV infection, we compared the inflammatory transcriptome profile observed at the late stage of infection in the mouse brain (cortex and brain stem/cerebellum) with the ortholog gene expression in post-mortem brain biopsies of rabid patients. Our data indicate that the inflammatory response associated with rabies is more pronounced in the murine brain compared to the human brain. In contrast to murine transcription profiles, we identified CXC motif chemokine ligand 16 (CXCL16) as the only significant differentially expressed gene in post-mortem brains of rabid patients. This result was confirmed in vitro, in which Tha suppressed interferon alpha (IFN-α)-induced CXCL16 expression in human CNS cell lines but induced CXCL16 expression in IFN-α-stimulated murine astrocytes. We hypothesize that RABV-induced modulation of the CXCL16 pathway in the brain possibly affects neurotransmission, natural killer (NK) and T cell recruitment and activation. Overall, we show species-specific differences in the inflammatory response of the brain, highlighted the importance of understanding the potential limitations of extrapolating data from animal models to humans.

Keywords: CXCL16; brain; human; immune response; mouse; neuroinflammation; rabies; transcriptome.

Figure 1

Inflammatory transcription profile of murine CX and BSC upon RABV strain Tha infection at day 8 post-infection. Scatter plot shows gene expression fold changes in murine CX (n=6) and BSC (n=6), upon Tha infection. Fold changes were calculated based on normalization with the geomean of housekeeping genes Actb and Gapdh ( Supplementary Figure S1 ) as well as non-infected murine CX (n=6) and non-infected BSC (n=6), respectively. Differential expression was calculated by comparing the gene expression (ΔΔCT values) between murine CX and BSC with the Šídák’s multiple comparisons test (adjusted p-value ****< 0.0001, **< 0.01). Gene expression in the murine CX (⁺) or murine BSC ( ^□ ) that differed significantly from non-infected controls is marked accordingly. A detailed overview of the differential gene expression analysis can be found in Supplementary Table S4 .

Figure 2

Clinical presentation of Cambodian rabid patients from group 1 (n=10). (A) Description of the clinical signs and symptoms of Cambodian rabid patients. (B) Time between the onset of symptoms and death in days observed for Cambodian rabid patients. (A, B) A detailed overview of the clinical presentation of Cambodian rabid patients belonging to group 1 can be found in Supplementary Table S1 .

Figure 3

Human transcription profile upon RABV infection. (A) Gene expression fold changes in post-mortem brain biopsies of the BSC of rabid patients (n=10) normalized to non-rabid patients (n=4). Differential expression was calculated by comparing the gene expression (ΔCT values) between rabid patients (n=10) and non-rabid (control) patients (n=4) by using the Šídák’s multiple comparisons test (** adjusted p-value < 0.01). A detailed overview of the differential gene expression analysis can be found in Supplementary Table S4 . (B) Heatmap comparing the expression of inflammatory genes between RABV Tha strain-infected murine CX (n=6) and BSC (n=6), and RABV-infected human BSC (n=10). Heatmap presents gene expression fold changes calculated from normalized gene expression values (ΔΔCT) of infected animals and humans to the respective non-infected controls (murine CX [n=6]; murine BSC [n=6]; human BSC [n=4]). Gene names are indicated in the human gene name nomenclature although expression indicates quantification of human or murine genes depending on the tissue investigated. (C) Number of differentially expressed genes showing the same tendency (up- or downregulation) in the RABV Tha strain-infected murine CX (n=6) or BSC (n=6), and RABV-infected human BSC (n=10). Differential gene expression was calculated to non-infected mice (n=6) or human controls (n=4), respectively. A detailed overview of the differential gene expression analysis can be found in Supplementary Table S4 .

Figure 4

CXCL16 gene expression in human and murine brains and CNS cell types during RABV infection. (A) CXCL16 gene expression in brain from non-rabid patients and rabid patients (from group 1 and 2). Comparison of CXCL16 expression in brains from non-rabid (control) patients (n = 12; 12 samples) from Cambodia (IPC38, IPC39, IPC40, IPC41, IPC42) and France (P448, P447, P465, P521, P552, P523, P524) and from rabid patients (n=12; 15 samples) from Cambodia (IPC02, IPC08, IPC12, IPC15, IPC18, IPC23, IPC24, IPC28, IPC34, IPC36), Senegal (IPS01), Moldavia (1900105, 1900106, 1900107), and Madagascar (IPM01). A detailed overview of the additional included human samples can be found in Supplementary Table S5 . CXCL16 expression values (ΔCT) were compared by using an unpaired t-test. (B) CXCL16 expression in non-infected control human CX (n=4) and BSC (n=8) samples. CXCL16 expression values (ΔCT) between CX (P447, P465, P523, P524) and BSC (IPC38, IPC39, IPC40, IPC41, IPC42, P448, P521, P552) were compared by using an unpaired t-test (p=0.072). (C) Cxcl16 expression in the CX (n=6) and BSC (n=6) of non-infected mice. Cxcl16 expression values (ΔCT) between CX and BSC were compared by using a paired t-test (p=0.122). (D) CXCL16 expression in human CNS cell types (n = 3) infected with RABV Tha (48 hours post-infection). Human CNS cell types (SK-N-SH, HMC3, SVGp12) were infected with Tha (MOI 5) and treated with IFN-α at 24 hours post-infection. (E) Cxcl16 expression in murine neuroblastoma cells and cortical murine astrocytes (n=3) upon RABV Tha strain infection (48 hours post-infection). Murine CNS cell types were infected with Tha (MOI 5) and treated with IFN-α at 24 hours post-infection. Murine astrocytes were extracted from C57BL/6 wild type mice and their phenotype was validated by immunofluorescence and qPCR (Supplementary Figure S8) . (D, E) All cell culture experiments were performed in three independently replicates. Bars show average fold changed ± SD with a Tukey’s multiple comparison test (adjusted p-value ****< 0.0001, ***< 0.001, **< 0.01, *< 0.05). BSC, brainstem/cerebellum; CX, cortex; HMC3, human microglial cell line; mAstrocytes, cortical murine astrocytes; NI, non-infected; N2A, murine neuroblastoma cells; SK-N-SH, human neuroblastoma cell line; SVGp12, human astroglia cell line; ns, non-significant.

Figure 5

CXCL16 interaction network. (A) Physical network of CXCL16 indicating physical protein associations. Colours present five of the most upregulated KEGG pathways (strength > 2, FDR < 5.94e-21): Red = GABAergic synapse (hsa04727), yellow = serotonergic synapse (hsa04726), purple = cholinergic synapse (hsa04725), blue = glutamatergic synapse (hsa04724), green = dopaminergic synapse (hsa04728) (B) Full network of CXCL16 indicating functional and physical protein associations. Colours present four of the most upregulated biological processes from gene ontology (strength > 2, FDR < 0.00021): Red = T cell chemotaxis (GO:0010818), blue = regulation of T cell chemotaxis (GO:0010819), green = macrophage chemotaxis (GO:0048246), yellow = positive regulation of monocyte chemotaxis (GO:0090026). Networks were generated by using the STRING version 11.0. GNB, beta unit of guanine nucleotide-binding proteins; GNG, gamma unit of guanine nucleotide-binding proteins.