The Proteome of TLR3-stimulated human immortalized fibroblasts; implications for susceptibility to herpes simplex encephalitis

Abstract

Background

Inborn errors in the Toll like receptor 3 (TLR3)-Interferon (IFN) type I and III pathway have been implicated in susceptibility to herpes simplex virus encephalitis (HSE) in children, but most patients studied do not carry mutations in any of the genes presently associated with HSE-susceptibility. Moreover, many patients do not display any TLR3-related fibroblastic phenotype.

Objective

This suggests the study of other signalling pathways downstream of TLR3 and/or other independent pathways which may contribute to HSE susceptibility.

Methods

We have used the stable isotope labelling of amino acids in cell culture (SILAC) proteomics methodology to measure changes in the human immortalized fibroblast proteome after TLR3 activation.

Results

Cells from healthy controls were compared to cells from a patient with a known genetic aetiology of HSE (UNC93B^−/−) and also to cells from an HSE patient with an unknown gene defect. Consistent with known variation in susceptibility of individuals to viral infections, substantial variation in the response level of different healthy controls was observed, but common functional networks could be identified, including upregulation of superoxide dismutase 2 (SOD2). The HSE patients show clear differences in functional response networks when compared to healthy patients and also when compared to each other.

Conclusions

The present study delineates a number of novel proteins, TLR3-related pathways and cellular phenotypes that may help elucidate the genetic basis of childhood HSE. Furthermore, our results reveal SOD2 as a potential therapeutic target for amelioration of the neurological sequelae caused by HSE.

INTRODUCTION

Herpes simplex virus (HSV-1) encephalitis (HSE) is a devastating infection of the central nervous system (CNS)¹. HSE is the most common form of sporadic viral encephalitis in Western countries, where it is estimated to occur in approximately 1 in 250,000 individuals per year. It peaks in childhood between the ages of 3 months and 6 years during primary infection with HSV-1, which reaches the CNS via a neurotropic route involving the trigeminal and olfactory nerves^{2, 3}. Treatment with acyclovir decreases the mortality rate in affected children, but significant neurological impairment is observed in most survivors, young children in particular. HSV-1 is widespread and typically innocuous in human populations. Childhood HSE has not been associable with known immunodeficiencies and its pathogenesis remained elusive until we identified the first three genetic etiologies of this condition^{4, 5}. Autosomal recessive UNC-93B deficiency abolishes Toll-like receptor (TLR) 3,7,8, and 9 signaling⁶, whereas autosomal dominant TLR3 deficiency specifically affects TLR3 signaling⁷. Recently an autosomal dominant deficiency in TNF receptor-associated factor 3 (TRAF3), an adaptor molecule implicated in the TLR3 pathway, and Toll/IL1R (TIR) domain-containing adaptor inducing IFN-β (TRIF) deficient patients has been described^{8, 9}. These studies suggested that childhood HSE may result from impaired interferon (IFN)-α/β and IFN-λ production in response to the stimulation of TLR3 by dsRNA intermediates of HSV-1 in the CNS. However, only a small fraction of children with HSE carry mutations in UNC93B1, TLR3, TRIF or TRAF3. The study of proteins implicated in the TLR3-IFN pathway for HSE patients enrolled in our laboratory (226 patients) has revealed that only a small proportion of patients bear a genetic defect in these proteins. A larger proportion of patients display an impaired production of IFN type I and III upon TLR3 stimulation of their fibroblasts. Other patients do not show the fibroblastic phenotype associated with impaired IFN production; in these patients the gene defect may affect other pathway(s) that are normally activated in fibroblasts after TLR3 stimulation. In the present study, as an approach to define new candidate genes for HSE, we have used proteomics methods to characterize TLR3-dependent pathways in fibroblasts that might be correlated with HSV-1 infection in the CNS.

METHODS

Subjects and patient P case report

The experimental protocol was approved by the Ethical Committee of Necker-Enfants Malades Hospital (Paris, France) and written informed consent was obtained for this study. P is a French boy of non-consanguineous descent with no family history of encephalitis. At the age of eight years, he suffered fever, convulsions and hemiparesis and he was diagnosed with HSE.

Cell purification

Primary human fibroblasts were obtained from biopsies of an UNC-93B-deficient patient⁶, three healthy controls, and one HSE patient with unknown gene defect. They were then transformed with an SV-40 vector, as previously described¹⁰, to create immortalized fibroblast cell lines: SV40-fibroblasts.

Stable isotope labelling with amino acids in cell culture (SILAC) labelling of human SV40-fibroblast

Cells were grown for six rounds of cell division in DMEM containing 13C6, 15N4 L-arginine, 13C6 L-lysine (Invitrogen, Carlsbad, California, USA) (heavy medium) supplemented with 10% dialysed FCS (Invitrogen) to ensure all the cellular proteins were labelled to saturation. The SILAC technique relies on the intrinsic metabolic machinery of the cell to incorporate “heavy” amino acids into all newly synthesized proteins. Immortalized fibroblast cell lines were used to achieve the full SILAC incorporation without detrimentally affecting the characteristics of the cell line. Our previous work has demonstrated that SV40 immortalised fibroblasts can successfully be used to study TLR3 signalling^7–9.

Labelled cells were activated in 24-well plates, at a density of 10⁵ cells/well, with 25 µg/ml polyinosinepolycytidylic acid (poly(I:C)), a TLR3 agonist (InvivoGen, San Diego, CA, USA), for 24 hours. Unlabelled cells were grown in parallel in medium containing normal amino acids (light medium) and were not stimulated. The SILAC labels were shown to be fully incorporated before the experiment was conducted (data not shown).

Protein Separation and In-Gel Enzymatic Digestion

Equal amounts of protein from unlabelled “light” (non-stimulated) and labelled “heavy” (stimulated) cells were mixed and subsequently separated by SDS-PAGE (BioRad Laboratories, Hercules, CA, USA). Proteins were visualized by silver-staining (Sigma, St Louis, MO, USA) and the gel lane was divided into approximately 34 equally sized pieces which were excised from the gel and destained (30mM K₃Fe(CN)₆; 100mM Na₂S₂O₃) prior to further processing. Gel processing was conducted with a Progest Investigator Instrument (DigiLab, Genomics Solutions, Cambs, UK) for reduction and alkylation according to established protocols¹¹. Briefly, the gel pieces were washed with three cycles of 25 mM NH₄HCO₃ pH 8.0 and acetonitrile. Finally, gel plugs were rehydrated in 20 µg/mL sequencing grade modified trypsin (Promega, Hamps, UK) and incubated overnight at 37C. Tryptic peptides were eluted, vacuum-dried, and resuspended in 0.1% formic acid.

XCalibur raw files were processed by Quant.exe and Identify.exe of the MaxQuant suite (version 1.0.13.13), in combination with the Mascot search engine (version 2.2, Matrix Science, U.K.), and searched against a concatenated International Protein Index (IPI) human protein database(version 3.54; containing 75,710 entries, including 262 commonly observed contaminants such as porcine trypsin and some human keratins). All proteins were filtered according to a false discovery rate (FDR) of 1% applied at both peptide and protein levels. Proteins were automatically quantified in the MaxQuant software. In the final results files, all Protein Groups with a normalized ratio SigB score of ≤0.05 and a normalised ratio > 1.5 were accepted for downstream analysis. Experiments were done induplicate and all proteins discussed in the text were significant in both replicates for each sample. For details of mass spectrometry analysis, protein inference and quantification, analysis of sample correlations and global functional analysis of protein networks, see Additional Methods at the Online Repository.

Transient transfections

SV40-fibroblasts were transfected with 2 µg of pBI-EGFP-MnSOD (SOD2) or pTRE-Tight-BI-AcGFP1 (Clontech, Palo Alto, CA, USA) as mock vector, in the presence of the FuGENE ®HD Transfection Reagent (Roche Applied Science, Indianapolis, IN, USA), according to manufacturer’s instructions. 24 hours post-transfection, SV40-fibroblasts were activated by incubation with 25 µg/ml poly(I:C) (InvivoGen) for 24 h. pTRE-Tight-BI-AcGFP1 was used as mock vector (pBI-EGFP has been replaced by AcGFP; it has 94% identical amino acid sequence and same biophysical properties).

Western blot

Total cell extracts were prepared from SV40-fibroblasts, both transfected and not transfected. Equal amounts of protein from each sample were separated by SDS-PAGE and blotted onto iBlot™ Gel Transfer Stacks (Invitrogen). Membranes were probed with anti-ICAM-1 rabbit monoclonal (Lifespan Biosciences, Seattle, WA, USA), anti-SOD2 and ITGA2 monoclonals and anti-ITGA5 and PPIF polyclonals (Abcam, Cambridge, MA, USA), anti-ANXA5/7 monoclonals (SantaCruz Biotechnology Inc., SantaCruz, CA), followed by a secondary anti-mouse or rabbit IgG (GEHealthcare, Buckinghamshire, UK). Membranes were stripped and reprobed with anti-GADPH (SantaCruz) to control for protein loading. Antibody binding was detected by enhanced chemiluminescence (ECL; Amersham-Pharmacia-Biotech).

Apoptosis analysis

Levels of SV40-fibroblast apoptosis after poly(I:C) stimulation were assessed by measures of caspase-3 and 7 activity. Cells were plated, in triplicate, in Microtest™ 96 well Assay Plate, Optilux™ Black/Clear Bottom (Falcon, Becton Dickinson, Franklin Lakes, NJ, USA) (1 × 10⁴ cells/well), in DMEM supplemented with 2% FCS; some cells were transfected with 150 ng per well of pBI-EGFP-MnSOD (SOD2) or pTRE-Tight-BI-AcGFP1 as mock vector. 24 hours later, cells were treated with poly(I:C) (25 µg/ml) and incubated for 24h. Caspase-3 and 7 activity was measured 48 hours post-transfection by Caspase-Glo® 3/7 Assay (Promega) as per manufacturer’s instructions.

IFN type I and III determination

SV40-fibroblast cell lines were activated in 24-well plates at a density of 10⁵ cells/well and stimulated with poly(I:C) at a concentration of 25 µg/ml for 24 h. Cells were grown at 37°C under an atmosphere containing 5% CO₂. Cell supernatants were recovered and an ELISA was performed for IFN-β (TFB, Fujirebio, Inc., Tokyo, Japan) according to manufacturer’s instructions. The ELISA for IFN-λ was performed as previously described⁷.

RESULTS

Identification of proteins up-regulated in human SV40 fibroblasts following TLR3 activation

Response to TLR3 activation was defined by altered levels of protein abundance in human SV40-fibroblasts when stimulated with poly(I:C) (used as TLR3 agonist^6–9) and was measured by SILAC/MS^12–14 monitoring of heavy/light isotope labelling ratios for peptides from proteins of stimulated/unstimulated cells (Figure 1A).

Figure 1.

(A) Illustration of the SILAC workflow. (B) Summary of samples used and TLR3 pathway state for the SILAC experiment. C1–3: healthy controls; C2NS: healthy control-2 non-stimulated; UNC-93B^−/−: UNC-93B-deficient patient; P: patient with unknown gene defect. WT: Wild-type HSE: herpes simplex encephalitis. (C) Overview of MS results obtained with MaxQuant software.^a Protein groups identified with protein and peptide FDR = 1%.^b Protein abundance ratios calculated as median values using a minimum of 3 quantifiable razor peptides.^c Protein ratios with SigB ≤ 0.05 and SILAC ratios ≥ 1.5. L: light medium; H: heavy medium; NS: non-stimulated cells; TLR3: Poly(I:C) stimulated cells.

Three Healthy samples (C1, C2 and C3) are positive controls that provide a measure of variability between individuals (Figure 1B). Negative control, C2NS, is a healthy sample with strong TLR3 response, but for which neither cell population (light/heavy) was stimulated (Figure 1B). The UNC-93B-deficiency blocks TLR3, 7, 8 and 9 pathways, so cells from the UNC-93B^−/− patient (sample UNC-93B^−/−) should not respond to TLR3 agonists⁶. We used SV40-fibroblasts from a homozygous UNC-93B^−/− patient rather than a patient with heterozygous TLR3 mutation due to the complete deficiency of TLR3 pathways observed in UNC93B^−/− cells⁶. Finally, we included cells from an HSE patient with unknown gene defect (P) (Figure 1B), which produce detectable amounts of IFN type I after TLR3 stimulation (Figure 2).

Figure 2.

Production of IFN-β by SV40-fibroblasts after poly(I:C) stimulation for 24 hours as assessed by ELISA. C1–C5: positive healthy controls. UNC93B^−/−: UNC-93B-deficient patient. Mean values ± SD were calculated from three independent experiments.

All six biological samples were subjected to duplicate MS analysis^{15, 16} and protein lists were obtained by combining both datasets prior to analysis of differential protein abundance (E Table 1–6). The MS data is summarized in Figure 1C, and E Figure 1 illustrates the quality of the MS data for a single SILAC peptide pair detected in a key protein of our study.

Figure 1C shows the number of proteins found to be significantly upregulated (normalized H/L (SILAC) ratio ≥ 1.5 and Significance B (SigB)¹⁶ < 0.05, see below) in the different cells following activation (except for C2NS) with poly(I:C). We have previously observed variability in response of healthy control fibroblasts upon TLR3 stimulation, particularly in terms of cytokine and interferon production (Figure 2). To take this into account, we included sample C3, from a healthy donor with known weaker response to TLR3 stimulation compared to C1 and C2 (Figure 2). In line with expectation, this control showed a reduced number of upregulated proteins (19 with SILAC H/L ratios between 1.50-fold and 2.15-fold; Figure 1C). Comparison of upregulated proteins between the control samples revealed variation in those proteins most strongly upregulated. This variability might arise from a) the conservative condition that protein SILAC ratios were calculated from at least three separate H/L ratio measurements, and/or b) differences in the extent to which different proteins change in abundance in cells taken from different individuals. A qualitative test of cellular variability amongst healthy controls displays biologically-relevant positive correlation distinguishable from the negative control (C2NS) (see Additional Results, E Figure 2, E Figure 3 and E Table 7 at the Online Repository).

Detection of Functional Networks

Given the observed variability in detection and measurement of specific protein regulation across experiments, we sought common functional networks amongst the most strongly upregulated proteins in order to characterise response to TLR3 stimulation and identify disease-related differences by comparison. We used MetaCore™ version 6.3 (GeneGo Inc.) to investigate pathways and biological functions represented by differentially regulated proteins in all samples (see Additional Results and E Figure 2 at the Online Repository).

The analysis indicated that the most significant GeneGo process networks for C1/C2/C3 are similar and are implicated in immune responses, such as IFN-signalling and antigen presentation pathways (Table 1A), known to be implicated following TLR3 stimulation^{17, 18}. Similar analysis was performed for the non-stimulated C2NS control sample. The majority of proteins in the non-stimulated control C2NS show SILAC ratios ≈ 1 (E Table 4) which represent “measurement noise”. As may be expected, the 8 proteins designated as upregulated (H/L = 1.50 to 3.07-fold) did not associate with canonical pathways or process networks implicated in immune response, and networks inferred from them were of limited statistical significance and scattered over diverse pathways and processes (Table 2A). The UNC-93B^−/− sample did not associate with the immune-related networks identified above, further suggesting that the TLR3 pathway is completely abolished by this deficiency (Table 2B). UNC-93B^−/− did, however, yield several upregulated proteins (E Table 5), with SILAC ratios of 1.5 to 3.95-fold, which may represent attempts by these cells to compensate for the genetic defect. Canonical pathways associated with these proteins were largely metabolic (glycolysis, gluconeogenesis, fructose metabolism), however, and process networks inferred were particularly diverse (Table 2B).

Table 1.

GeneGo pathway maps and process networks with their proteins associated. (A) Analysis of the individual data sets for C1–3. (B) Enrichment analysis for the union of the C1–3 data sets. Grey panels: pathways/networks known to be implicated in TLR3-IFN signalling. Blue panels: new pathways/networks activated. Threshold: 1.5 (SILAC ratio), SigB p-value < 0.05. Sorting method is statistically significant.

A
C1 SAMPLE		C2 SAMPLE		C3 SAMPLE
name	pValue	name	pValue	name	pValue
Ontology: Top GeneGo Pathway Maps		Ontology: Top GeneGo Pathway Maps		Ontology: Top GeneGo Pathway Maps
Immune response_Antigen presentation by MHC class I	0,0004167	Immune response_Antiviral actions of Interferons	5,215E-05	Immune response_MIF-mediated glucocorticoid regulation	0,02399
Immune response_IFN alpha/beta signaling pathway	0,007584	Immune response_Antigen presentation by MHC class I	0,0001645	Immune response_IFN alpha/beta signaling pathway	0,02615
Ontology: Top GeneGo Process Networks		Ontology: Top GeneGo Process Networks		Immune response_IL-27 signaling pathway	0,02615
Inflammation_Interferon signaling	4,681E-05	Immune response_Antigen presentation	0,0004169	Immune response_Role of integrins in NK cells cytotoxicity	0,04112
Immune response_Phagosome in antigen presentation	0,0006502	Immune response_Phagosome in antigen presentation	0,007849	Ontology: Top GeneGo Process Networks
Immune response_Antigen presentation	0,00111	Inflammation_Interferon signaling	0,01796	Inflammation_Interferon signaling	0,01668
				Immune response_IL-5 signalling	0,0779

B
UNION C1, C2 and C3 SAMPLES
Ontology: Top GeneGo Pathway Maps			Ontology: Top GeneGo Process Networks
name	pValue	Proteins associated	name	pValue	Proteins associated
Immune response_IFN alpha/beta signaling pathway	0,007584	ISG54, ISG15, PTP-1B	Inflammation_Interferon signaling	4,68E-05	TAP1, TAP2, ICAM1, ISG15, ISG54, IFI56
Immune response_MIF-mediated glucocorticoid regulation	0,02399	ICAM1	Immune response_Antigen presentation	0,000417	Tapasin, TAP1, TAP2, Calnexin, ICAM1
Immune response_IL-27 signaling pathway	0,02615	ICAM1	Inflammation_NK cell cytotoxicity	0,04988	Histone H1, ICAM1, TfR1, MHC class1
Ontology: Top GeneGo Pathway Maps			Ontology: Top GeneGo Process Networks
name	pValue	Proteins associated	name	pValue	Proteins associated
Transcription_Role of AP-1 in regulation of cellular metabolism	0,01016	ITGA2, ITGA5	Blood coagulation	0,000239	TFPI-2, PLAU, ITGA2, AnnexinV
Chemotaxis_CCR4-induced leukocyte adhesion	0,01171	ICAM1, Talin	Cell adhesion_Platelet-endothelium-leucocyte interactions	0,000597	Integrins, TFPI-2, PAI2, ICAM1, PLAU, FGF2
Apoptosis and survival_Endoplasmic reticulum stress response pathway	0,01918	ERP5, SOD2	Transcription_mRNA processing	0,01563	hnRNP F, SNRPA, HYPA, KHSRP, SAM68, DDX5, SF3B1, U2AF35
Development_Cross-talk between VEGF and Angiopoietin 1 signaling pathways	0,0283	ICAM1	Cell adhesion_Attractive and repulsive receptors	0,02102	ICAM1, Integrins, Cortactin
Chemotaxis_Inhibitory action of lipoxins on IL-8- and Leukotriene B4-induced neutrophil migration	0,0319	ICAM1, Talin	Proteolysis_Connective tissue degradation	0,03509	TFPI-2, IRAP, PAI2, PLAU
Inhibitory action of Lipoxins on neutrophil migration	0,03911	SOD2	Cell adhesion_Cell-matrix interactions	0,05649	Integrins
Apoptosis and survival_Anti-apoptotic action of nuclear ESR1 and ESR2	0,04006	SOD2	Cell adhesion_Integrin-mediated cell-matrix adhesion	0,05697	Integrin, Talin, CD82

Table 2.

GeneGo pathway maps and process networks after the analysis of the individual data sets for (A) control non-stimulated (C2NS), (B) UNC-93B deficient patient cells (UNC-93B^−/−) and (C) HSE patient cells with unknown gene defect (P). Threshold: 1.5 (SILAC ratio), SigB p-value<0.05 and the sorting method is statistically significant.

A		B		C
C2NS SAMPLE		UNC-93B−/− SAMPLE		P SAMPLE
name	pValue	name	pValue	name	pValue
Ontology: Top GeneGo Pathway Maps		Ontology: Top GeneGo Pathway Maps		Ontology: Top GeneGo Pathway Maps
Cell adhesion_Cadherin-mediated cell adhesion	0,00952	Glycolysis and gluconeogenesis (short map)	1,87E-09	Apoptosis and survival_Granzyme A signaling	0,00097
Development_Slit-Robo signaling	0,01098	Glycolysis and gluconeogenesis p.3 / Human version	6,96E-05	Transcription_P53 signaling pathway	0,0021
Neurophysiological process_EphB receptors in dendritic spine morphogenesis and synaptogenesis	0,0128	Glycolysis and gluconeogenesis p.3	6,96E-05	Transport_Macropinocytosis	0,00271
Cell adhesion_Tight junctions	0,01317	Glycolysis and gluconeogenesis p. 2	0,000152	Transport_RAB3 regulation pathway	0,0037
Transcription_P53 signaling pathway	0,01426	Glycolysis and gluconeogenesis p. 2/ Rodent version	0,000167	Immune response_Antiviral actions of Interferons	0,00479
Neurophysiological process_Receptor-mediated axon growth repulsion	0,01644	Glycolysis and gluconeogenesis p. 2 / Human version	0,000281	Normal wtCFTR traffic / Sorting endosome formation	0,00612
Ontology: Top GeneGo Process Networks		Fructose metabolism	0,001911	DNA damage_NHEJ mechanisms of DSBs repair	0,00681
Transcription_Transcription by RNA polymerase II	0,01626	Fructose metabolism/ Rodent version	0,00276	Delta508-CFTR traffic / Sorting endosome formation in CF	0,00991
Cytoskeleton_Actin filaments	0,01972	Muscle contraction_Delta-type opioid receptor in smooth muscle contraction	0,0035	Immune response_IFN alpha/beta signaling pathway	0,01076
Cell adhesion_Synaptic contact	0,02144	Transcription_Role of Akt in hypoxia induced HIF1 activation	0,003772	Cell adhesion_Endothelial cell contacts by non-junctional mechanisms	0,01076
Reproduction_Spermatogenesis, motility and copulation	0,03227	Ontology: Top GeneGo Process Networks		Ontology: Top GeneGo Process Networks
Reproduction_Male sex differentiation	0,03683	Muscle contraction	0,00024	Cytoskeleton_Cytoplasmic microtubules	0,00013
Cytoskeleton_Spindle microtubules	0,1304	Development_Skeletal muscle development	0,00127	Cell cycle_Mitosis	0,00035
Cytoskeleton_Cytoplasmic microtubules	0,1371	Transport_Iron transport	0,004586	Cytoskeleton_Spindle microtubules	0,00068
Development_Neuromuscular junction	0,1732	Protein folding_ER and cytoplasm	0,009973	Inflammation_Interferon signaling	0,00072
Cell adhesion_Cell junctions	0,1881	Cell adhesion_Cell-matrix interactions	0,03411	Cytoskeleton_Macropinocytosis and its regulation	0,00145
Cell adhesion_Attractive and repulsive receptors	0,2018	Cell adhesion_Integrin-mediated cell-matrix adhesion	0,03452	Cell adhesion_Integrin-mediated cell-matrix adhesion	0,00498
		Immune response_Phagocytosis	0,03703	Translation_Translation initiation	0,00665
		Development_Blood vessel morphogenesis	0,04052	Cytoskeleton_Actin filaments	0,00762
		Cell adhesion_Attractive and repulsive receptors	0,1177	Translation_Elongation-Termination	0,00786
		Signal transduction_Insulin signaling	0,1177	Cytoskeleton_Intermediate filaments	0,00823

The same analysis was also performed on the union of proteins upregulated in C1/C2/C3 (152 proteins, excluding 7 proteins in common withC2NS or UNC-93B^−/− samples), strengthening the association between TLR3 stimulation and IFN α/β signalling pathways (Table 1B, E Figure 4) and further suggesting a clear proteomic signature for activation of these pathways upon TLR3 stimulation.

The analysis of healthy controls also revealed other statistically significant proteins/pathways/processes (Table 1B) that do not appear in similar analysis of C2NS or UNC-93B^−/− samples (Table 2), and which have not been associated previously with TLR3 pathways. Such proteins included Intercellular adhesion molecule-1 (ICAM-1), integrins ITGA2 and 5, and Superoxide dismutase 2 (SOD2) (Table 1B). Furthermore, some proteins upregulated by TLR3 stimulation in healthy controls were not placed in pathways/networks by GeneGo, but are associated with a role in HSV infection in the CNS by literature evidence. These proteins include several Annexins and Peptidyl-propyl cis-trans isomerise mitochondrial (PPIF) (E Table 1–3), the majority of which were not identity as differentially regulated by analysis of UNC-93B^−/− (Figure 3A). E Table 8 shows MS data for individual peptides assigned to these proteins and confirms the quality of their detection and assignment. The expression of these proteins was confirmed by western blot (WB) (E Figure 5) and they are considered further in the discussion.

Figure 3.

(A) For proteins with potential biological significance in HSE, the SILAC ratio detected for healthy controls (C1–3), healthy control 2 non-stimulated (C2NS), UNC-93B-deficient patient (UNC-93B^−/−) and patient with unknown gene defect (P). SigB p-value < 0.05. (B) Illustration of potential biological significance in immunity against HSE for proteins up-regulated after TLR3 activation.

Finally, although not the primary focus of our analysis, the union of proteins commonly downregulated upon TLR3 stimulation in healthy controls C1/C2/C3 were subjected to similar pathway analysis. The results are summarised in E Tables 1–3 and reflect strongest association with cytoskeleton remodelling and negative regulation of cell cycle, in accordance with the apoptosis induced upon TLR3 stimulation and in agreement, for example, with upregulation of annexins identified above.

Proteomics of TLR3 stimulated fibroblasts reveals potential new treatments for HSE: SOD2 upregulation

SOD2 is highly upregulated in healthy patient samples C1, C2 and C3 after TLR3 stimulation but not in cells from a patient with known genetic deficiency UNC-93B^−/− and neither in unstimulated healthy control sample C2NS (Figure 3A). SOD2 is an antioxidant enzyme strongly upregulated after TLR3 activation in macrophages, protecting these cells from oxidative stress during microbial infection¹⁹. SOD2 is increased after HSV-1 infection and SOD2 levels in the CNS are associated with neuronal protection^{20, 21}. Other examples of the protective role of SOD2 have been shown in an animal model of complex I deficiency²², in which SOD2 gene transfer in vivo enhanced cellular resistance to reactive oxygen species and suppressed degeneration of the optic nerve²². This suggests SOD2 as a putative target for future treatment of HSE, in particular to reduce the neurological sequelae that patients suffer after encephalitis.

We used gene transfer to further investigate the role of SOD2 in human transformed fibroblasts. WB of healthy cells confirmed upregulation of SOD2 after poly(I:C) stimulation, and very low levels of SOD2 in UNC-93B^−/− cells suggested that the defect affects SOD2 expression (Figure 4A). We detected SOD2 overexpression after transient transfections of pBI-EGFP-MnSOD (SOD2), but not with mock transfected vector (Figure 4A). Measurement of apoptosis in SV40-fibroblasts transfected with SOD2 showed that healthy cells overexpressing SOD2 are protected against apoptosis and that this is also the case for stimulation with poly(I:C). Relative apoptosis measured in poly(I:C) stimulated UNC-93B^−/− cells against their non-stimulated counterpart was higher than similar measurement for healthy controls, indicating UNC-93B^−/− patient cells as more susceptible to apoptosis than healthy cells after TLR3 activation. When transfected with SOD2, however, UNC-93B^−/− patient cells were protected against apoptosis (Figure 4B). Together, these results suggest that pharmacological agents that upregulate SOD2 expression or activity could exert a protective anti-oxidant response mechanism to reduce the cell death associated with TLR3 stimulation in HSE patients.

Figure 4.

Human SV40 fibroblast non transfected (−), transfected with SOD2 or Mock vector. (A) WB analysis of SOD2 levels. (−): non stimulated; (+): poly(I:C) stimulated. Densitometry normalized with respect to GADPH, expressed as a fold induction over non stimulated/non transfected control cells. The panel is representative of three experiments. (B) Caspase-3/7 activity in non stimulated (NS) or poly(I:C) stimulated cells. Luciferase activity represented as percentage,100% is healthy control non stimulated. Mean values ± SD were calculated from three independent experiments. C: healthy control C3; UNC-93B^−/−: UNC-93B-deficient patient. Two-sided test were used and a p value less than 0.05 was considered statistically significant.

Analysis of a patient with unknown gene defect and without fibroblastic phenotype

30% of HSE patients analyzed do not show the fibroblastic phenotype characterized by reduced IFN type I and III production after TLR3 stimulation (data not shown). In these patients, study of the TLR3 stimulated proteome may help delineate the source(s) of the different cellular phenotype. We conducted a SILAC analysis of a patient whose genetic defect is unknown, but which produces significant amounts of IFN type I and III after TLR3 stimulation (E Figure 6A). The study identified many upregulated proteins shared with stimulated healthy control samples C1, C2 and C3, and which were not observed for UNC93B^−/−. The proteins shared those highlighted during earlier analysis of the stimulated healthy controls; with the exception of ICAM-1, which was not detected post-stimulation (Figure 3A and E Table 6) as confirmed by WB analysis (E Figure 6B). ITGA5 (H/L = 1.65) and ANX 11 (H/L = 1.29) are upregulated in patient sample P, but at a ratio significance just outside the threshold of SigB < 0.05 (E Table 6). GeneGo analysis also inferred process networks similar to those related to stimulation of healthy controls, involving immune responses such as IFN-signalling and antigen presentation pathways (Table 2C). This result concurs with previous studies that show cells from patient P to exhibit IFN type I and III production after TLR3 signalling (E Figure 6A). Given the extent of overlap between the analysis of TLR3 stimulation effect in patient P and that in healthy control samples, the non-detectable levels of ICAM-1 upon TLR3 stimulation in P, identified via proteomic analysis, provide an avenue in this patient for a candidate approach centred on ICAM-1 and related genes/proteins. This would be the first such approach in an HSE patient to involve pathways beyond TLR3-IFN signalling.

DISCUSSION

The quantitative SILAC analysis presented here provides important preliminary evidence to suggest that several proteins, pathways and processes could play important and novel roles in TLR3 response and HSE immunity.

ICAM-1 upregulation

After TLR3 stimulation, healthy controls C1, C2 and C3 displayed ICAM-1 upregulation, as described previously^{23, 24}, but there was no corresponding upregulation in UNC-93B^−/−, patient P or non-stimulated C2NS (Figure 3A and E Figure 6). ICAM-1 is an intercellular adhesion molecule implicated in HSV-1 infection in CNS^{25, 26}. The present results suggest that lack of ICAM upregulation may be an important factor in susceptibility to HSE, especially for those patients (e.g. P) which show substantial IFN production and lack the fibroblastic phenotype.

Upregulation of integrins ITGA2 and ITGA5

We detected ITGA2 and 5 upregulation after TLR3 stimulation in healthy cells but not in C2NS or UNC-93B^−/− (Figure 3A). Cells from patient P showed significant upregulation of ITGA2 (Figure 3A) and upregulation of ITGA5 (H/L = 1.65) just beyond the significance threshold employed (E Table 6). Some studies have demonstrated that integrins are necessary for HSV-1 entry/fusion in cells²⁷ and ITGA2 is upregulated in chronic hepatitis virus infection²⁸. The present results identify specific integrins upregulated following TLR3 activation and whose further investigation might offer new insights into HSE susceptibility.

Upregulation of Annexins

After TLR3 stimulation in healthy cells, we detected upregulation of several annexins (Figure 3A). Microglial cells, after activation in response to harmful stimuli, are able to induce neuronal cell death by ANX1²⁹ and ANX5 upregulation³⁰. It has been shown that TLR3 activation with poly(I:C) and HSV-1 infection induce apoptosis by enhancing ANX5 levels^31–33. The present data suggests possible roles for ANX5, 6, 7 in the activation and subsequent apoptosis of immune cells in the CNS upon viral infection.

PPIF upregulation

PPIF protein was upregulated after TLR3 stimulation in healthy and patient P cells, but not in those with UNC-93B^−/− defect (Figure 3A). PPIF, also called Cyclophilin D, is part of the mitochondrial permeability transition pore in the inner mitochondrial membrane. Activation of this pore is thought to be involved in the induction of apoptotic and necrotic cell death. HSV-1 infection of neurons is known to induce cell death and PPIF has been previously described as a key protein in neuronal cell death^34–36. This is the first report of a relationship between PPIF and TLR3, suggesting a possible role for PPIF in HSE immunity.

Previous studies have shown that the TLR3 pathway is essential for susceptibility to HSE^6–8. However the analysis of many proteins downstream of TLR3 implicated in the TLR3-IFN type I pathway has failed to reveal a genetic defect in most patients analyzed. In 226 patients enrolled in our laboratory, we have detected mutations in UNC93B1, TLR3 or genes downstream of the TLR3 pathway in a very low percentage. Conversely, study of IFN type I and III production after TLR3 activation in SV40-fibroblasts of HSE patients has shown that only 30%, of a total of 89 patients analyzed, have normal IFN type I and III production (data not shown). This suggests that despite the importance of the TLR3 pathway in HSE immunity, genetic defect(s) responsible for susceptibility to HSV-1 in the CNS could be due to TLR3-independent pathways, or other TLR3, IFN-dependent pathways that are activated after initial TLR3 activation. Overall, the present results suggest a rich diversity of pathways downstream of TLR3 activation that may be relevant for HSE susceptibility and which invite further investigation.

Proteomic analysis of a patient (P) with unknown gene defect but without the fibroblastic phenotype has revealed a lack of ICAM-1, delineating a cellular phenotype which may assist dissection of this genetic aetiology. Finally, an important problem in this pathology is post-infection neurological sequelae, for which the current treatment is aciclovir, an antiviral drug. However, this treatment is largely uneffective². The present study has now revealed SOD2 as a possible therapeutic target for prevention of the neurological sequelae suffered by HSE patients.

Proteomics has reached the stage where its methods are well developed and capable of generating large, robustly identified, high resolution quantitative datasets that provide a wealth of knowledge at the systems level. Even for a disease such as HSE, of apparent aetiological diversity, coupled with the cellular diversity evident amongst individuals, proteomics has provided an integrated overview of cellular responses that may be crucial for understanding the dynamics of HSE susceptibility.

CLINICAL IMPLICATIONS.

Quantitative SILAC-based proteome analysis of TLR3-stimutated human immortalized fibroblasts can delineate cellular phenotypes and may be used to dissect the genetic basis of childhood herpes simplex encephalitis (HSE), as well as new potential therapeutics targets.

Abbreviations

HSE

Herpes simplex virus encephalitis

SILAC

Stable isotope labelling of amino acids in cell culture

HSV-1

Herpes simplex virus

CNS

Central nervous system

TLR

Toll-like receptor

TRAF3

TNF receptor-associated factor 3

TRIF

Toll/IL1R (TIR) domain-containing adaptor inducing IFN-β

IFN

Interferon

poly(I:C)

Polyinosinepolycytidylic acid

SigB

Significance B

ICAM-1

Intercellular adhesion molecule-1

SOD2

Superoxide dismutase 2

PPIF

Peptidyl-propyl cis-trans isomerise mitochondrial

MS

Mass Spectrometry