HIV induces expression of complement component C3 in astrocytes by NF-κB-dependent activation of interleukin-6 synthesis

Adult human brain specimens were provided by the Manhattan HIV Brain Bank, a member of the National NeuroAIDS Tissue Consortium (U24MH100931) under an Institutional Review Board-approved protocol at Icahn School of Medicine at Mount Sinai. HIV and gene expression analyses were conducted on coded brain samples without subject identifiers under an “exempt” status approved by an Institutional Review Board of St. Luke’s-Roosevelt Hospital Center (presently Mount Sinai). Samples from four HIV-positive and three HIV-negative subjects were used in this study. All HIV-positive patients had evidence of HIVE; HIV-negative subjects (controls) had normal neurological function and unremarkable brain histology. Ages ranged from 30 to 63 years; five were male and two were female; and three were white, two hispanic, and two black. All samples used for analyses were from a similar location within gray matter of the frontal lobe.

RNA was extracted using the RNeasy Mini Kit (QIAGEN, Valencia, CA). cDNA was prepared from individual brain samples using the WT-Ovation™ RNA Amplification System (NuGEN Technologies, Inc., San Carlos, CA). QPCR was conducted using TaqMan chemistry and using probes from the Universal Probe Library (Roche Diagnostics, Indianapolis, IN) and primers designed with the online ProbeFinder software (https://​lifescience.​roche.​com/​en_​us/​brands/​universal-probe-library.​html). In the PCR reaction, 5 μl of cDNA obtained as previously described was combined with 10 μl of 2X Universal Master Mix (Thermo Fisher Scientific, Waltham, MA), 0.2 μl of each forward and reverse primers at 200 nM, 0.2 μl of probe at 100 nM, and RNAse/DNAase-free water. All reactions were performed in duplicate and were run in a 7500 real-time PCR system (Thermo Fisher Scientific). Data were normalized using glyceraldehyde-3-phosphate dehydrogenase (GAPDH). For immunohistochemical analysis, formalin-fixed blocks were used to construct tissue microarrays with three punches (diameter of 1 mm) from each block. Five-micrometer sections were cut and immunohistochemistry performed using a rabbit polyclonal anti-C3c complement antibody (1:3000, DakoCytomation) and either mouse anti-GFAP or anti-CD68 as the primary reagents and diaminobenzidine or amino ethyl carbazole as the chromogens.

HFA were isolated from second trimester human fetal brains obtained from elective abortions in full compliance with NIH guidelines as previously described [52]. Highly enriched populations of astrocytes were obtained by high-density cultures condition in the absence of growth factors in F12 Dulbecco’s modified Eagle’s medium (Thermo Fisher Scientific), containing 10% fetal bovine serum (Hyclone, Piscataway, NJ) antibiotics and fungizone. Cells were incubated at 37 °C in 5% CO₂/95% air. HIV-1 molecular clones used were NL4-3 (clone pNL4-3; M19921) from Dr. Malcolm Martin’s laboratory [53] that expresses all HIV-1 proteins. HIV/NL4-3 stocks were prepared by transfection of 293T cells. Cells were transfected by calcium phosphate co-precipitation method [54] and then purified by sedimentation as described; mock virus was sedimented in parallel from supernatants of 293T cells [52]. Residual plasmid DNA was removed post-transfection using DNAse I digestion (Sigma, St. Louis, MO). Early passages of HFA were cultured for 5–7 days until 75% confluence; the cells were washed in warm PBS and infected with cell-free HIV-1 at the indicated number of picograms per cell or comparable dilutions of mock virus for 2 h at 37 °C and were washed three times to remove the virus. For more detailed protocol, see [55] and [16].

Cell supernatants were collected as indicated and the levels of IL-6 were measured by ELISA (Raybiotech, Norcross, GA). To detect C3, a (indirect) sandwich ELISA was conducted using goat anti-human C3 polyclonal antibody (Sigma) as the antibody in solid phase, followed by cell supernatants. Bound C3 was detected using rabbit anti-human C3c complement antibody (Dako, Carpinteria, CA), as the detection antibody followed by peroxidase conjugated-monoclonal anti-rabbit IgG (γ-chain specific) (Sigma). The concentration of C3 protein was measured using complement C3 protein from human serum (Sigma) as a standard (in serial dilution).

The C3 promoter region was cloned by PCR amplification of human astrocyte DNA using primers: sense (5′ GCGCGCTAGCCTGCAATTTAGCCTGAGTGACAGAATG 3′) and antisense (5′ GCGCCTCGAGAGAGGGACAGAGGGACAGAGGGAGAGG 3′); the sense primer contains a NheI cleavage site and the antisense primer contains a XhoI cleavage site for insertion into the pGL3 vector (Promega, Madison, WI). The reaction was performed in a buffer containing 10 mM Tris-HCl, 50 mM KCl, 0.1% Triton X-100, 200 μM each dNTP, 2.5 mM MgCL₂, 4 units Taq polymerase (Promega), 0.5 μg DNA, and 0.6 μM each primer, in total reaction of 50 μL. The cycling parameters were 94 °C for 1 min, 55 °C for 1 min, and 72 °C for 2 min, for a total 25 cycles for both stages. Six C3 promoter deletion mutants were constructed by PCR amplification using sense primers containing a NheI cleavage site as follows:and an antisense primer containing a XhoI cleavage site (5′ GCGCCTCGAGAGCAGCGCCTGCTGGAGCTGGCTTTTTATC 3′). Six replacement mutations were constructed in specific response elements within the C3 promoter by PCR. The table contains the mutagenic primers and sequences altered. Reaction products were inserted into the pGL3 basic vector.

Dr. Gail Bishop of the University of Iowa (Iowa City, IA) kindly provided plasmids encoding luciferase whose expression is driven by the intact murine IL-6 promoter or a mutated promoter lacking the NF-κB recognition element [56]. Their function in HFA was assessed as described for constructs encoding C3 promoter-dependent luciferase expression.

RNA was isolated from astrocytes with RNeasy kits (QIAGEN), and cDNA was synthesized using the Superscript kit (Thermo Fisher Scientific), all according to manufacturer’s instructions. Primers for C3 and GAPDH were purchased from Applied Biosystems (Thermo Fisher Scientific). Primers for other human genes were designed using the Roche Universal Probe Library Assay Center and were purchased from Thermo Fisher Scientific; the Universal Probe Library (Roche) was used to provide probes. C3 gene expression in HFA was evaluated by fold change of Relative Quantification. In the human brain tissue study, QPCR reactions were prepared as described in [35]. All reactions were performed in duplicate. Raw data was analyzed using the 7900 System SDS Software (Thermo Fisher Scientific). Data was normalized using GAPDH expression values. Relative quantification employed the comparative threshold cycle method. Student’s t test was used to test significant versus control groups.

HFA were grown to 80% confluence in 12-well plates and transfected with plasmid DNA as follows: 1.5 μg C3 or IL-6 promoter driving firefly luciferase and 0.5 μg of Renilla luciferase vector, after 2.5 h of transfection using lipofectamine 2000 (Thermo Fisher Scientific), cells were washed and incubated 48 h with various stimuli, then lysed and both luciferase activities were measured using the Promega Dual Luciferase Assay kit according to the manufacturer’s instructions, firefly luciferase is reported as relative light units (RLU), normalized to Renilla luciferase activity.

Astrocytes were preincubated for 6 h with one of pharmacological inhibitors (EMD Chemicals, Gibbstown, NJ) of signal transduction pathways or with vehicle as indicated: AG17 2 μg/ml AG18 10 μg/ml, CAPE 0.5 μg/ml, genistein 25 μg/ml, JNK inhibitor II 1 μg/ml, PDTC 5 μM, SB 202190 10 μM, SB 203580 10 μM, U0126 10 μM, and wortmannin 0.1 μg/ml. After preincubation with inhibitor, cells were washed and then were cultured in 7.5% FBS DMEM with/or without inhibitor, followed by HIV infection or mock control. Alternatively, HFA were infected with adenovirus control or an adenovirus expressing super-repressor I-κBmt32 as described [57]; cells were then transfected with the C3-luciferase construct, followed by HIV or mock infection and luciferase activity measured.

For quantitation of nuclear content of NF-κB, nuclei were isolated using the Panomics Nuclear Extraction Kit and protein was measured using the Transbinding TM NF-κB Assay Kit according to the manufacturer’s instructions. Alternatively, astrocytes were cultured on two-well chamber slides, fixed with 4% formaldehyde, permeabilized with 0.1% Triton X-100 and after blocking nonspecific binding with 1% bovine serum albumin, stained with anti-p65 antibody (1:100; Santa Cruz Biotechnology, Santa Cruz, CA) overnight at 4 °C. Cells were then rinsed three times for 5 min each in PBS and incubated with Alexa488-conjugated anti-rabbit IgG (Thermo Fisher Scientific) for 1 h at room temperature. After three rinses for 5 min each in PBS, cells were mounted in Vectashield fluorescence mounting medium containing 4.6-diamidino-2-phenylindole (Vector Laboratories, Burlingame, CA). Images were taken with a Confocal Laser Scanning Microscope LSM Multiphoton 510 (Zeiss, Thornwood, NY).

Student’s t test was used to test significant differences in between two groups with asterisk indicating p < 0.05 (Figs. 1 and 8b). For all other studies (Figs. 2, 3, 4, 5, 6, 7, and 8a), one-way analysis of variance (ANOVA) was used to assess significant differences among the groups. Dunnett’s multiple comparison post hoc test was performed generally to compare all the groups to untreated mock-infected or untreated medium or in Fig. 6 to compare differences in values between the intact C3 promoter and deletion or point mutations in HIV infected. Significance is indicated by asterisk (p < 0.05). All the ANOVA and Dunnett’s statistical comparisons and values are included in Additional file 1: Table S1.

Our results and observations by other investigators indicate that HIV exposure of transformed or primary human astrocytes in culture leads to induction of C3 [23‐26]. To address the physiological significance of these findings in HIV neuropathogenesis, we measured the relative levels of the transcripts of C1qa, C1qb, C3, and C4a in brain tissue obtained at autopsy from individual patients with HIVE in comparison to brain tissue from uninfected individuals (Fig. 1A). C3, C1qb, and C4a transcripts, but not C1qa, were significantly upregulated in the brains of HIVE patients compared to controls. Other markers of inflammation, some previously implicated in HAD [23, 35, 58, 59], were also transcriptionally induced in the brains of HIVE patients including interferon-related genes IFIT1 and STAT1 and the macrophage inflammatory protein 2-alpha (MIP2-α), also known as the chemokine CXCL2.

To identify the cell types producing C3, we performed two-label immunohistochemistry staining on paraffin sections from control and HIVE brains for C3c and either GFAP for astrocytes or CD68 for microglia/macrophages (Fig. 1B). C3c was detected with AEC-conjugated antibody (red) and GFAP and CD68 with DAB (brown/black). Brain sections from an HIV-negative subject showed no C3c immunoreactivity indicating no baseline complement activation (Fig. 1Ba). In contrast, brain sections from a patient with HIVE (Fig. 1Bb–d) showed strong C3c staining, particularly in neurons (dark red cells in Fig. 1Bb and Bd) and some astrocytes (light red). Double staining for C3 and GFAP (Fig. 1Bc) confirmed the astrocytic expression of C3 during HIV infection in patients (red C3 staining in cell body and brown GFAP staining in cell processes; arrows). There was limited cellular co-localization of C3c and CD68 markers (Fig. 1Bd; arrow) suggesting that in this tissue, macrophages/microglia were not the major source of C3. These findings indicate that the environment of the HIV-infected brain generates signals to neurons and astrocytes to produce elevated C3. The remainder of this study is dedicated to identifying the specific triggers to this innate immune response.

We established a primary tissue culture system to assay the responses of HFA to HIV/NL4-3 exposure at multiple levels of C3 regulation: protein, RNA, and promoter activation (Fig. 2). Cells were exposed either to mock virus concentrates as negative controls or medium for baseline expression. C3 promoter activity was measured by activation of the C3 promoter-driven luciferase marker (Fig. 2a); C3 RNA was measured by QPCR, standardizing by GAPDH levels (Fig. 2b); and extracellular C3 protein was measured by ELISA (Fig. 2c). At each level studied, HIV-induced C3 in HFA: HIV activated the C3 promoter, C3 transcription, and C3 protein expression.

As a first step to understand the mechanism of C3 activation, we began investigation of the protein kinase signaling pathways in HFA required for C3 induction. HFA were pretreated with pharmacological inhibitors of specific protein kinases, then were infected by HIV/NL4-3, and transfected with the C3 promoter-luciferase construct to assay the activity of the C3 promoter (Fig. 3). NF-κB inhibitors (CAPE and PDTC) as well as inhibitors of p38 (SB203580, SB202190), protein tyrosine kinase (Genistein), JNK, platelet-derived growth factor receptor tyrosine kinase (AG17), and epidermal growth factor receptor tyrosine kinase (AG18) blocked the HIV activation of transcription from the C3 promoter. Based upon the maintenance of activation of the C3 promoter in the presence of specific inhibitors, PI3K and MEK are not involved in induction of C3 by HIV in astrocytes. Earlier studies using astrocytic cell lines implicated cyclic AMP, protein kinase C, and C/EBPδ but not NF-κB in the activation of C3 expression in astrocytes by HIV [24]. To supplement the evidence obtained using pharmacological inhibitors (Fig. 3), we performed three complementary assays to investigate the state of activation of NF-κB after exposure of HFA to HIV. First, nuclear localization of p50 monomer of NF-κB in astrocytes was measured by ELISA after various stimuli (Fig. 4a). Like the positive control IL-1β, HIV induced the entry of NF-κB into the nucleus and this event was inhibited by CAPE. Second, we employed an adenovirus vector of super-repressor I-κBmt32 with the expectation that the complex of NF-κB and I-κBmt32 will resist degradation and NF-κB will be retained in the cytoplasm and cannot function in transcription. HFA were infected by control adenovirus or adenovirus encoding I-κBmt32 and then infected with HIV and induction of the C3 promoter was measured by luciferase assay (Fig. 4b). Like the experiment shown in Fig. 2, HIV exposure activated transcription from the C3 promoter; the control adenovirus vector had no effect upon this activation. In contrast, expression of I-κBmt32 and inhibition of activation of NF-κB completely blocked C3 promoter induction by HIV in HFA (Fig. 4b). Finally, to unambiguously demonstrate that HIV activates NF-κB in astrocytes, we localized NF-κB in HFA by confocal microscopy. Cells were activated by various stimuli in the presence of control adenovirus or adenovirus encoding I-κBmt32 super-repressor of NF-κB. Cells were fixed and stained for the p65 chain of NF-κB, nuclei were visualized with DAPI (Fig. 5). In the absence of stimuli, NF-κB is mainly cytoplasmic in HFA; however, both HIV and IL-1β induced the translocation of p65 into the nucleus (upper row). In contrast, the expression of the super-repressor I-κBmt32 arrested NF-κB nuclear localization induced either by HIV or by IL-1β (lower row); control adenovirus had no effect (middle row). The findings in Figs. 3, 4, and 5 demonstrate unequivocally that HFA respond to HIV by NF-κB-linked C3 synthesis. Further studies dissect the elements in the C3 promoter that control HIV-induced C3 responses.

We prepared a series of deletion and point mutations in the C3 promoter in the luciferase construct and tested their response during activation of HFA by HIV (Fig. 6). The minimal promoters consisting of only 50 or 100 base pairs upstream of the initiation codon were inactive; restoring the promoter to 200 base pairs restored the response to HIV (Fig. 6a). The constitutive activity of this construct was somewhat higher than that of the intact promoter, suggesting that there is a negative regulatory element upstream. To identify essential sites within the promoter recognizing transcriptional modulators, we prepared five mutations (Fig. 6b). Mutation of the proximal IL-1β/IL-6 site (M1 or M2) abolished both constitutive and HIV-induced transcription directed by the C3 promoter. In contrast, mutation of the interferon-γ response element (M3), the more distal IL-6 element (M4), or either of two NF-κB sites (M5 or M6) had only modest effects upon the C3 promoter response to HIV by HFA. Studies in the previous section demonstrated that HIV-induced C3 through NF-κB, but here it is clear that the NF-κB sites in the C3 promoter are dispensable for the response to HIV. This apparent paradox can be resolved if NF-κB is required for induction of expression of an intermediate protein that itself induces C3.

We propose that HIV activates IL-1β or IL-6 expression through NF-κB and one or both of these cytokines then initiates a program to activate the C3 promoter in HFA. As shown in Fig. 1, both IL-1β and IL-6 are prominently expressed in the HIV-infected brain, consistent with the premise that these cytokines trigger C3 production by astrocytes. We have previously shown that HIV can induce IL-6 transcription and secretion [16] and Fig. 7a illustrates dose-dependent induction of IL-6 in HFA by HIV. Because HFA do not produce IL-1β under the same conditions (not shown), we confined further studies to activities of IL-6 upon HFA. To definitively link the production of IL-6 to the activation of C3 expression, we employed anti-IL-6 neutralizing antibody. HFA expressing the C3 promoter-luciferase construct were exposed to HIV in the presence of anti-IL-6 or control antibody (Fig. 7b). The ability of HIV to activate the C3 promoter was reduced by neutralization of IL-6 to background levels, indicating that IL-6 is an essential intermediate in C3 induction by HIV. To determine whether IL-6 employs NF-κB to activate the C3 promoter, we exposed cells to exogenous IL-6 and inhibited NF-κB through transduction of the super-repressor, I-κBmt32 (Fig. 7c). IL-6 increased the activity of the C3 promoter in HFA but infection with the adenovirus vector of I-κBmt32 entirely blunted this response; infection with control adenovirus had no effect upon the C3 promoter response. These findings strongly implicate endogenously produced IL-6 in the pathway of HIV activation of C3 by HFA. To determine the transcriptional pathway employed in HIV activation of IL-6 synthesis in HFA, we tested the induction of the IL-6 promoter by pharmacological inhibition of NF-κB or by mutagenesis. An IL-6 promoter construct directing synthesis of luciferase was tested for activation by HIV in the presence or absence of CAPE (Fig. 8a) or an intact IL-6 promoter, and a mutant in the κB site driving luciferase expression were used (Fig. 8b). HIV required NF-κB activity to activate the IL-6 promoter, the presence of CAPE or a mutated κB element abolished promoter function. These findings, coupled with those in Fig. 7, strongly suggest that HIV activates a circuit in astrocytes, first inducing NF-κB-dependent IL-6 synthesis; extracellular IL-6 then initiates a secondary signaling cascade resulting in C3 transcription and protein expression.