A 4-kb transgene was separated from pNF-kB-Luc vector (BD Clontech, Palo Alto, CA, United States) and injected into the pronuclei of single cell fertilized embryos isolated from FVB/N mice. The NF-κB-RE (Responsive)-luc transgenic mice were maintained under conditions compliant with the rules and guidelines of the Institutional Animal Care and Use Committee of National Cheng Kung University. Female mice were used for excluding the autoactivation signal of NF-κB from testis.

Streptococcus pyogenes strain A20 (emm sequence type 28) was isolated from a blood sample from a patient with necrotizing fasciitis at the National Cheng Kung University Hospital. A20 Tn4001-8, a bioluminescent strain, was produced by J. J. Wu, Department of Medical Technology, National Cheng Kung University. A20 Tn4001-8 is a bioluminescent strain generated from A20 by the transformation with the plasmid pXen-luxCDABE (Xenogen Bioware, Alameda, CA, United States) and selected with kanamycin. The bioluminescent phenotype of A20 Tn4001-8 was screened using the xenogeny IVIS 200 (PerkinElmer, Santa Clara, CA, United States). A20 Tn4001-8 was grown in tryptic soy broth supplemented with 0.5% yeast extract and 400 μg/ml kanamycin.

The model of subcutaneous infection was modified as previously described (Lu et al., 2013). Female mice at 8–10 weeks of age were anesthetized and hair was then removed from a portion of their backs. An air pouch was produced by subcutaneous injection of 0.2 mL of bacterial suspension (2 × 10⁹ CFU of S. pyogenes strain A20 Tn4001-8). Dominant-negative (DN)-TNF (XENP1595) was a soluble selective inhibitor obtained from Xencor, Inc. Three dosages of DN-TNF (50 mg/kg) were intraperitoneally administered 30 min before infection, 3 h post infection, and 6 h post infection.

Mice were anesthetized and imaged using the xenogeny IVIS 200 (PerkinElmer, Santa Clara, CA, United States). To acquire images of the bacterial luciferase, emission filter wavelengths ranging from 500 to 540 nm were used. To acquire images of firefly luciferase, luciferin (150 mg/kg of animal body weight) was intraperitoneally injected 10 min prior to imaging and images were then acquired using emission filter wavelengths ranging from 620 to 660 nm. To differentiate signals from the bacterial and firefly luciferases, spectral unmixing was performed as previously described (Kadurugamuwa et al., 2005) and analyzed using Living Image software v3.1.

Total RNA was extracted from whole brains using REzol (PROtech, Taipei, Taiwan). Reverse transcription was performed using M-MLV reverse transcriptase (Invitrogen, Carlsbad, CA, United States) and oligo(dT) primers. Real-time PCR was performed using SYBR as implemented in the ABI StepOnePlus Real-Time PCR System (Applied Biosystems, Foster City, CA, United States). Expression of each gene was normalized to β-actin. The sequences of primers were as follows: forward: 5′- CATCTTCTCAAAATTCGAGTGACAA-3′ and reverse: 5′-TGGGAGTAGACAAGGTACAACCC-3′ for TNF; forward: 5′-GCAACTGTTCCTGAACTCAACT-3′ and reverse: 5′-ATCTTTTGGGGTCCGTCAAT-3′ for IL-1β; forward: 5′-CCCACTCACCTGCTGCTACT-3′ and reverse: 5′-TCTGGACCCATTCCTTCTTG-3′ for MCP-1; forward: 5′-GCCATCAGCAACAACATAAGCGTC-3′ and reverse: 5′-CCACTCGGATGAGCTCATTGAATG-3′ for IFN-γ; forward: 5′-GAGAACAACCTGGCTGCGTAT-3′ and reverse: 5′-GCCTCGTATTGAGTGCGAAT-3′ for GFAP; forward: 5′-AGCAGCGGCTCCATGACT-3′ and reverse: 5′-TCATGCGGCCTCCTTTGA-3′ for iNOS; forward: 5′-ACCTGAAACTGCCCACTGAC-3′ and reverse: 5′-ACCTGAAACTGCCCACTGAC-3′ for Ncf1; forward: 5′-GCAGTGGCCTACTTCCAGAG-3′ and reverse: 5′-ACCTCACAGGCAAACAGCTT-3′ for Ncf2; forward: 5′-TTCCTGTTGTCGGTGCCTGC-3′ and reverse: 5′-TTCTTTCGGACCTCTGCGGG-3′ for Cyba; forward: 5′-GGAGTTCCAAGATGCCTGGA-3′ and reverse: 5′-CCACTAACATCACCACCTCATAGC-3′ for Cybb; forward: 5′-CTGTTTGTTCGAGAGCATAAC-3′ and reverse: 5′-TAGGTACTTCTTCATGGCTG-3′ for MPO; forward: 5′-CCTGGAACTCACACGACATCTTC-3′ and reverse: 5′-TGGAAACTCACACGCCAGAA-3′ for MMP-9; forward: 5′-TCAAAGAGGAGAAGGCTGGAAA-3′ and reverse: 5′-CACCACAGCATACAGAATCGCA-3′ for TNFR1; forward: 5′-AAGGGTGGCATCTCTCTTCCA-3′ and reverse: 5′-AGGCACCTTGGCATCTCTTTG-3′ for TNFR2; forward: 5′-ACTGCCGCATCCTCTTCCTC-3′ and reverse: 5′-TGCCACAGGATTCCATACCC-3′ for β-actin.

Sodium fluorescein was used to assess BBB permeability as described previously (Esaki et al., 2010). Mice were intraperitoneally injected with 200 μL of 10% sodium fluorescein (Sigma-Aldrich, St. Louis, MO, United States). After 40 min, the mice were anesthetized and perfused with 10 mL of phosphate-buffered saline (PBS). The brain was then removed and homogenized in 50% trichloroacetic acid and centrifuged at 10,000 g for 10 min. The supernatant was diluted with 0.8 volume of 5 M NaOH and measured using a fluorimeter at an excitation of 485 nm and emission of 515 nm. Sodium fluorescein standard solutions (1–1000 ng/mL) were used to calculate the tissue content, which was normalized to the total amount of protein in the homogenate.

Serum cytokines were measured using a Milliplex MAP Mouse Cytokine Kit (Millipore, Billerica, MA, United States) on a Luminex 200 analyzer (Luminex, Austin, TX, United States). Data were evaluated by applying a 5-parameter logistic curve fit by using the Software Luminex IS 2.3.

Mice were transcardially perfused with 4% paraformaldehyde (PFA) in PBS. For Iba-1 immunofluorescence, the brain was removed, postfixed in 4% PFA for 24 h, and then transferred to 30% sucrose for 24 h. Cryostat brain sections (10-μm-thick) were fixed in ethanol, blocked with normal serum, and incubated overnight at 4°C with primary antibodies against Iba-1 (1:1000, Dako, Carpinteria, CA, United States), followed by secondary antibodies (Alexa Fluor^® 488, Life Technologies, Grand Island, NY, United States). Cell nuclei were counterstained with Hoechst. Images were visualized using a confocal microscope and processed using the EZ-C1 software (Nikon, Tokyo, Japan). For Fluoro-Jade C staining, the deparaffinized and rehydrated slides were transferred to 0.06% potassium permanganate solution for 10 min. After 1–2 min of water wash, the slides were transferred to a solution of 0.0001% Fluoro-Jade C dissolved in 0.1% acetic acid for 10 min. The slides were washed with distilled water, coverslipped, and subsequently imaged.

Nuclear fraction of the brain was isolated according the previous report (Rajendrasozhan et al., 2010). Total 20 μg of nuclear protein was separated on SDS-PAGE and transferred to polyvinylidene fluoride membranes. The membranes were blocked with 3% BSA for 1 h and incubated overnight with the primary antibodies against phosphor-p65 (#3033, Cell Signaling, Danvers, MA, United States) and lamin B1 (#ab16048, Abcam, Cambridge, MA, United States) at a 1:1,000 dilution in the blocking buffer. After being washed with TBST buffer (20 mM Tris-HCl, pH 7.4, 150 mM NaCl and 0.1% Tween-20) three times for 5 min each, membranes were incubated with the anti-rabbit antibody (#401315, Calbiochem, Gibbstown, NJ, United States) at a 1:30,000 dilution for 1 h and visualized by ECL (Millipore, Billerica, MA, United States).

Values are reported as means ± SEM. Statistical analyses were conducted using Student’s t-test or one-way analysis of variance followed by least significant difference post hoc tests. Kaplan–Meier survival curves were compared using the log-rank (Mantel–Cox) test. Differences were considered to be statistically significant at p < 0.05.

Transcription factor NF-κB signaling is considered to play a critical role in inflammation and innate immunity. To concurrently monitor the burden of GAS and host innate immune response in the same mouse, the reporter mice carrying firefly luciferase under transcriptional control of the NF-κB promoter were subcutaneously inoculated in the air pouch at the back site with a lethal dose of A20 carrying a lux operon and monitored using Living Image software at 3, 24, and 48 h post infection (hpi). Dual bioluminescent signals of bacterial and firefly luciferase in the same mouse were monitored with spectrum peaks at 490 and 610 nm, respectively. The bacterial signal was restricted at the back site of inoculation and decreased at 24 hpi, but reinforced at 48 hpi (Figure 1A, upper panel). Similarly, the firefly luciferase signal rapidly increased at 3 hpi and decreased at 24 hpi, but reinforced at 48 hpi at the inoculation site (Figure 1A, upper panel). Notably, in addition to the infection site, the brain region exhibited firefly luciferase signals, which increased exponentially in a time-dependent manner; however, the bacterial signal in the brain was negative (Figure 1A, upper panel). The dual bioluminescent signals at the back site and brain were quantified (Figure 1A, lower panel). Using non-luminescent GAS can also induce central nervous system inflammation (data not shown).

To confirm the phenomenon in the brain resulting from NF-κB activation, p65 phosphorylation and the expression of NF-κB-mediated downstream genes were detected. We observed that p65 phosphorylation (Figure 1B) and TNF, interleukin (IL)-1β, and monocyte chemoattractant protein (MCP)-1 mRNA expression levels (Figure 1C) were increased in the brain at 48 hpi, but not interferon gamma (IFN-γ) levels, indicating that NF-κB signaling was turned on and triggered neuroinflammation.

Because the time point of 48 hpi exhibited the strongest firefly luciferase signal, we examined whether any brain injury markers were detectable during this time point. We found that the inflamed brain abundantly expressed glial fibrillary acidic protein (GFAP), a specific predictor of brain damage (Figure 2A). The inflamed brain overexpressed inducible nitric oxide synthase (iNOS) (Figure 2B) and nicotinamide adenine dinucleotide phosphate (NADPH) oxidase components (Figure 2C), suggesting increased ROS production. Myeloperoxidase (MPO), mainly released by activated macrophages and neutrophils, was abundantly expressed in the cortex and hippocampus of infected mice (Figure 2D). Microglia are specialized innate immune cells in the CNS, which can perform proinflammatory effector functions on activation. Accumulating evidence indicates that microglia activation contributes to neuroinflammation in neurodegenerative diseases (Chen et al., 2016; Carniglia et al., 2017). Therefore, we examined the activation status of microglia at the 48 hpi time point. Immunostaining of brain sections for ionized calcium binding adaptor molecule 1 (Iba1), a marker for microglia, was used for detecting the activation of microglia. Infected mice exhibited activated microglial phenotypes with thicker and shorter processes in the cortex and hippocampus (Figure 2E). These results suggested that subcutaneous GAS infection promotes microglia activation and induces brain injury.

Blood–brain barrier is an important barrier to maintain brain homeostasis by blocking the entrance of toxins and pathogens (Abbott et al., 2010; Wong et al., 2013). Disruption of the BBB is believed to be an early and significant event in neuroinflammation (Banks et al., 2015; Varatharaj and Galea, 2017). We detected an increased matrix metallopeptidase (MMP)-9 expression level in the infected brain (Figure 3A); this is a critical molecule contributed to BBB disruption associated with neuroinflammation (Takata et al., 2011; Dal-Pizzol et al., 2013). Moreover, the permeability of the BBB to small molecules by quantitatively using intraperitoneal sodium fluorescein injection during GAS infection was significantly increased compared with the control group at 48 hpi (Figure 3B).

Sepsis-induced neuroinflammation has been reported to further lead to serious neuronal degeneration (Yokoo et al., 2012). To examine whether the inflamed brain with increased brain injury and BBB permeability finally causes brain pathological change, we performed Fluoro-Jade C staining to analyze neuronal degeneration. Fluoro-Jade C positive cells were identified in each different regions of the brain examined (Figure 4).

Our infection model demonstrated that circulating cytokines TNF, IL-1β, MCP-1, and IFN-γ were largely increased at both 24 and 48 hpi as a severe systemic inflammatory response (Figure 5A). Therefore, we suggested that neuroinflammation in GAS-infected mice was mediated by the peripheral effect. Peripheral TNF, the key mediator of sepsis, is an acute phase response that initiates a cascade of cytokines. Previous studies have demonstrated that peripheral TNF plays an important role in the pathogenesis of SAE and is involved in the disruption of the BBB in many brain diseases (Sharief and Thompson, 1992; Pan and Kastin, 2007; Alexander et al., 2008; Lv et al., 2010). We observed that TNFR1 and TNFR2 mRNA levels were increased in the brains of infected mice (Figure 5B). Thus, we hypothesized that peripheral TNF is an important mediator of GAS-induced central nervous system inflammation through TNFR signaling.

To determine whether circulating TNF blockage can attenuate subcutaneous GAS infection-induced central nervous system inflammation, we treated infected mice with three doses (30 min before, 3 h after infection, and 6 h after infection) of a dominant-negative inhibitor of TNF (DN-TNF) and detected the luciferase signal at 3, 24, and 48 hpi. We first detected the circulating TNF level after DN-TNF treatment and found that DN-TNF can successfully block circulating TNF, but not IL-1β, MCP-1, and IFN-γ levels at 6, 12, 24, and 48 hpi (Figure 6A–D). From the bioluminescence results, the bacterial signal exhibited no difference between the GAS and GAS plus DN-TNF treatment groups (Figure 7A), suggesting that the host defense mechanisms still functioned after DN-TNF treatment. However, the firefly luciferase signal from the infected brain was significantly decreased after DN-TNF treatment, particularly at 48 hpi (Figure 7A). The dual bioluminescent signals at the back site and brain were quantified. DN-TNF treatment reduced the inflammatory cytokine and MPO expression levels in the infected brain (Figure 7B,C). In addition, DN-TNF treatment significantly reduced the MMP-9 expression and sodium fluorescein content in the brain (Figure 7D). Results showed that the infected mice with DN-TNF treatment had a higher survival rate (median survival 11.5 days) than the infected mice without DN-TNF treatment (median survival 4 days) (Figure 7E).