Activated Human T Cells, B Cells, and Monocytes Produce Brain-derived Neurotrophic Factor In Vitro and in Inflammatory Brain Lesions: A Neuroprotective Role of Inflammation?

Ag-specific CD4⁺ human T cell lines were isolated as described (13). PBMCs were separated into CD4⁺ T cells, CD8⁺ T cells, CD19⁺ B lymphocytes, and CD14⁺ monocytes using immunomagnetic beads (Dynal). The purity of the isolated subsets was >98% as determined by flow cytometry for relevant and irrelevant subset markers (not shown).

For proliferation and BDNF production assays, 5 × 10⁵ freshly isolated CD4⁺ T cells, CD8⁺ T cells, and B lymphocytes were cultured in RPMI 1640 medium (GIBCO BRL) supplemented with 5% FCS (GIBCO BRL) in 96-well plates (Nunc). After a resting period of 72 h, T lymphocytes were stimulated with 5 × 10⁴ irradiated allogeneic PBMCs and PHA (10 μg/ml; Sigma). B cells were stimulated with heat-inactivated Staphylococcus aureus Cowan strain 1 (SAC; Calbiochem) at a final dilution of 1:7,500. Monocytes were stimulated in the course of purification using anti-CD14–coated magnetic beads. Ag-specific T line cells (10⁵) were cultured in RPMI 1640 medium supplemented with 5% FCS plus irradiated autologous or HLA-matched PBMCs and the relevant Ag (10–30 μg/ml). Proliferation was assessed by [³H]thymidine uptake. Supernatants were removed at different time intervals after stimulation and analyzed for BDNF protein concentration and bioactivity. BDNF production by APCs was measured in separate wells and subtracted. The cytokine profile of CD4⁺ T cell lines was assessed by intracellular flow cytometry staining according to the manufacturer's protocol (PharMingen).

Total cellular RNA was extracted using two different RNA extraction systems (QIAGEN, and WAK Chemie). The RNA (0.5–2 μg) was transcribed with random hexamer primers (Boehringer-Mannheim) or oligo(dt) primers (GIBCO BRL), dNTP (MBI Fermentas), and Superscript™ Reverse Transcriptase (GIBCO BRL). All reactions were carried out in a total volume of 25 μl containing 1 U Taq polymerase (QIAGEN), 200 μM of each dNTP, cDNA, and 15 pmol of each primer for 35 or 40 PCR cycles with annealing at 60°C. The correct size of the bands was determined by comparison with a DNA mass standard (pUC MIX 8; MBI Fermentas). RNA samples transcribed in the absence of reverse transcriptase were used as negative controls to exclude genomic contamination. All PCR products were sequenced (Medigene). The primer sequences were as follows: β-actin forward: 5′-CGAGCGGGAAATCGTGCGTGA-3′ (position 622–642); β-actin reverse: 5′-CAGCGAGGCCAGGATGGAGCC-3′ (position 1057–1037); BDNF forward: 5′-AGCGTGAATGGGCCCAAGGCA-3′ (position 208–228); BDNF reverse 5′-TGTGACCGTCCCGCCCGACA-3′ (position 570–551).

BDNF protein concentration was determined with a sensitive sandwich ELISA. In brief, 96-well flat-bottomed plates (Immulon 4; Dynatech) were coated with a chicken anti–human BDNF Ab (Promega). Recombinant human BDNF (used as standard; Research Diagnostics, Inc.) and cell supernatants (all made up in RPMI, 5% FCS plus 0.05% Tween 20 [Sigma]) were used in serial dilutions. Bound BDNF was detected by incubating the plates with a mouse anti–human BDNF Ab (Research Diagnostics, Inc.) followed by peroxidase-labeled goat anti–mouse IgG (Dianova). The plates were developed using a liquid substrate system (Sigma), and the OD was determined at 450 nm. For enhanced detection of BDNF in biological samples, the cell supernatants were acid-treated for 15 min before the assays. This treatment did not change detection of the standard. Specificity of the BDNF ELISA was established using recombinant BDNF and control neurotrophic factors. The sensitivity of this ELISA ranges from 20 to 40 pg/ml. Quantification of BDNF was confirmed by a second highly specific ELISA.

Biological assays were performed using dissociated sensory neurons prepared from nodose ganglia of 8-d-old chick embryos (14). Neurons were plated in laminin/ polyornithine-coated 48-well plastic dishes. Exogenous BDNF (1 ng/ml) and immune cell supernatants were added in the presence or absence of the anti-BDNF mAb 4.D3.3A3, which neutralizes the biological activity of BDNF. After 24 h, the surviving neurons were counted using light microscopy. For each experiment, duplicate wells were evaluated.

Immunohistochemistry was performed on formalin-fixed, frozen, and paraffin-embedded human brain tissue using a mouse IgG1 mAb against BDNF (4.FL.1A1; 10 μg/ml) and a polyclonal anti-BDNF antiserum (N-20, used at 1:500 dilution; Santa Cruz Biotechnology) (15). Two actively demyelinating cases of multiple sclerosis, one case of acute disseminated (postinfectious) leukoencephalitis, and two controls without neurological disease were examined. Control sections were incubated without primary Ab or with control IgG1 Ab. Serial sections were labeled for CD3 (1:300; Serotec), CD68 (1:100; Dako), or human IgG (1:200; Amersham Pharmacia Biotech), using an avidin-biotin peroxidase method.

BDNF was detected in CD4⁺ T cells, CD8⁺ T cells, B cells, and monocytes (Fig. 1, a and b). CD4⁺ T cells and CD8⁺ T cells were stimulated with PHA and B cells with SAC. Monocytes were stimulated in the course of purification with anti-CD14–coated magnetic beads. Fig. 1 b shows the BDNF concentrations in supernatants of stimulated and nonstimulated cells after 72 h. Constitutive BDNF production was observed in T cells and B cells (white bars in Fig. 1 b). All cell types produced enhanced levels of BDNF after stimulation (black bars in Fig. 1 b). Reverse-transcription (RT)-PCR analysis showed BDNF transcripts in T cells, B cells, and monocytes (Fig. 1 a).

BDNF was also produced by Ag-specific CD4⁺ T cell lines of different Ag specificity and cytokine profile (Table I). Ag-induced BDNF secretion was observed both in Th1- and in Th2-type cells (Table I).

Bioactivity of the BDNF produced by human immunocompetent cells was assessed by measuring the ability to rescue cultured sensory neurons from cell death (14). CD4⁺ T cells, CD8⁺ T cells, B cells, and monocytes were stimulated with PHA, SAC, or anti-CD14–coated magnetic beads, respectively. After 96 h, the supernatants were collected and added to chicken embryonic nodose ganglia neurons to determine the neuronal survival rate (Fig. 2). Neuronal survival was supported by externally supplied recombinant BDNF and by supernatants of activated CD4⁺ T cells, CD8⁺ T cells, B cells, and monocytes (black bars in Fig. 2). The promotion of neuronal survival by immune cell supernatant was due to biologically active BDNF as demonstrated by inhibition with anti-BDNF mAb 4.D3.3A3, which blocks the biological activity of BDNF (white bars in Fig. 2).

Strong BDNF immunoreactivity was observed in inflammatory cells forming perivascular infiltrates in cases of disseminated (postinfectious) leukoencephalitis and multiple sclerosis (Fig. 3). Using serial sections, BDNF⁺ cells were found to correspond to infiltrating mononuclear cells (Fig. 3). In multiple sclerosis, BDNF⁺ lymphocytes and macrophages were not restricted to the perivascular localization but were found throughout the lesion (Fig. 3 g). Lesional areas with high numbers of macrophages actively involved in demyelination showed enhanced BDNF immunoreactivity. In inflammatory central nervous system disease and healthy controls, BDNF was also detected in various types of neurons, ependymal cells, and weakly in astrocytes. No immunoreactivity was observed in oligodendrocytes or ramified microglial cells. Similar staining patterns were observed with the monoclonal and polyclonal Abs.

We demonstrate that human immune cells express the neurotrophic factor BDNF both in vitro and in inflammatory brain lesions. BDNF is one of the most potent factors supporting neuronal survival and regulating neurotransmitter release and dendritic growth (2, 16, 17). Several studies have shown that the administration of BDNF protein or the BDNF gene can rescue injured or degenerating neurons and induce axonal outgrowth and regeneration (6–10). Furthermore, BDNF had beneficial effects in several animal models of neurodegenerative diseases (11). Difficulties in delivering sufficient amounts of BDNF to the site of central nervous system lesions have so far hampered the successful application of BDNF for treatment of human diseases (12).

One promising strategy for the delivery of neuroprotective factors relies on retroviral transduction of neurotrophic factors into T cell lines specific for Ag expressed in the nervous system (18). Our present results indicate that this experimental strategy has a natural counterpart: T cells and other immune cells homing to degenerative, infectious, or autoimmune lesions express a potent neurotrophic factor, BDNF, that may help to minimize neuronal damage. Consistent with this hypothesis, it has recently been demonstrated that the implantation of activated macrophages into experimental spinal cord lesions leads to partial functional recovery of paraplegic rats (19). In addition to its neurotrophic effect, locally produced BDNF could have an immunomodulatory function, e.g., by indirectly downregulating MHC class II expression on microglia (20).

NT-mediated effects on immune cells have previously been demonstrated for nerve growth factor (21). Thus far, however, we found no evidence that BDNF acts on peripheral human immune cells. This could be explained by the fact that immune cells express only the truncated form of the trkB BDNF receptor (our unpublished observations; consistent with previous reports that the expression of the full-length, signal-transducing form gp145trkB is restricted to neuronal cells, whereas the truncated gp95trkB form is widely expressed in nonneuronal tissues [5]). Further, we did not detect significant effects of BDNF on the proliferation, cytokine production, or apoptosis of human lymphocytes (our unpublished data). Clearly, these negative observations by no means rule out that BDNF has a role in the immune system, and the question of whether BDNF can affect elements of the immune system deserves more detailed study.

BDNF production by immune cells provides a novel example of the bidirectional interaction between the nervous system and the immune system (22, 23). It is noteworthy that upon stimulation with Ag a large proportion of immune cells, including CD4⁺ T cell lines specific for brain autoantigens, produce BDNF. Thus, in some circumstances, the immune infiltrates commonly found in inflammatory, ischemic, degenerative, and traumatic lesions of the nervous system may have a protective rather than destructive role. This could help to explain the presence of large numbers of autoreactive T lymphocytes in the healthy immune repertoire, a phenomenon invoked by Cohen's concept of the “immune homunculus” (24).

In conclusion, our demonstration that activated T cells, B cells, and monocytes produce BDNF in vitro and in inflammatory lesions raises the intriguing possibility that neuroinflammatory reactions have a neuroprotective (side) effect. This has obvious therapeutic implications for multiple sclerosis and other inflammatory diseases of the nervous system (25). Production of both neurotoxic and neuroprotective factors emphasizes the complex role of immune cells in inflammatory, degenerative, and regenerative processes of the nervous system.

The authors are grateful to M. Sölch for technical assistance; Drs. R. Lindert, T. Nickel, and E. Wilharm for support; and Drs. Y.-A. Barde, D. Jenne, and H. Neumann for valuable suggestions and critical reading of the manuscript.