DUAL EFFECTS OF TNFα ON NERVE FIBER FORMATION FROM VENTRAL MESENCEPHALIC ORGANOTYPIC TISSUE CULTURES

Abstract

Tumor necrosis factor alpha (TNFα) is toxic to dopamine neurons and increased levels of TNFα are observed in Parkinson’s disease. Dopamine nerve fiber outgrowth in organotypic cultures of fetal ventral mesencephalon occurs in two waves. The early appearing nerve fibers are formed in the absence of astroglia, while migrating astrocytes guide the late appearing dopamine nerve fibers. TNFα (40 ng/ml) was added to the medium of organotypic ventral mesencephalic tissue cultures between days 4–7 or 11–14. The cultures were evaluated at days 7 or 19 to study effects of TNFα on both types of nerve fiber formation. Tyrosine hydroxylase (TH) -immunohistochemistry demonstrated that the number of cultures showing non-glial-guided TH-positive outgrowth was reduced compared to controls, when TNFα was added at day 4. By contrast, the glial-guided TH-positive nerve fiber outgrowth and the astrocytic migration reached significantly longer distances by early TNFα treatment. Ki67-immunohistochemistry revealed that TNFα did not affect proliferation of astrocytes. Treatment with TNFα and antibodies against TNFα receptor 1 between days 4–7 revealed that the non-glial-guided TH-positive outgrowth reappeared. TNFα treatment between days 11–14 triggered neither the TH-positive glial-guided outgrowth, nor promoted the astrocytic migration to reach longer distances. The number of microglia was significantly increased after the late but not early TNFα treatment. In conclusion, TNFα is toxic for the non-glial dopaminergic nerve fiber outgrowth but stimulates the glial-guided outgrowth and the migration of astrocytes at an early time point. TNFα increased the number of microglia in VM tissue cultures after late but not after early treatment.

1. INTRODUCTION

In neurodegenerative diseases, such as in Parkinson’s disease, neuroinflammation plays a critical role. During neuroinflammation as well as in Parkinson's disease, the levels of proinflammatory cytokines are elevated (Boka et al. 1994; Mogi et al. 1994; Sawada et al. 2006). Tumor necrosis factor alpha (TNFα) belongs to the family of proinflammatory cytokines. The levels of TNFα are elevated in Parkinson's disease, and TNFα is known to induce apoptotic cell death (Kolesnick and Golde 1994; Mogi et al. 1994). TNFα is produced by activated microglia and the number of activated microglial cells are increased in the brains of parkinsonian patients (McGeer and McGeer 1997). Microglia are already activated during the early phase of Parkinson's disease, and the presence of activated microglia is correlated with decrease in dopamine nerve fiber density (Ouchi et al. 2005).

It has been shown that dopamine neurons are vulnerable to TNFα in tissue culture (McGuire et al. 2001). Furthermore, it has been documented that the dopaminergic neurotoxin 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) is less effective in TNFα deficient mice (Ferger et al. 2004), suggesting that TNFα takes part in the degenerative process. However, TNFα can also protect injured dopamine neurons in tissue culture (Shinpo et al. 1990). This neuroprotective effect is further confirmed in an in vivo study where the presence of TNFα at an early, transient event after injury of the dopamine system is beneficial for dopamine cell survival and regeneration (Gemma et al. 2007). Thus, there are indications that TNFα exerts a dual effect.

TNFα exerts its effect via two receptors, TNF receptor 1 (TNFR1) and TNF receptor 2 (TNFR2) (Lewis et al. 1991; Tartaglia et al. 1993; Tartaglia et al. 1991). These receptors are localized on neurons, including the nigral neurons, as well as on glial cells (Boka et al. 1994; Dziewulska and Mossakowski 2003; McGuire et al. 2001). TNFR1 induces cell death via a cascade of death proteins while TNFR2 instead stimulates survival and migration. Moreover, double mutant mice for TNFR1 and TNFR2, but not single mutants for either one of the two receptors, are protected against mechanical or toxic insults to the dopamine neurons (Rousselet et al. 2002; Sriram et al. 2002; Wessig et al. 2005). Thus, although TNFR2 is supposed to exert neuroprotective properties, both receptors need to be blocked to prevent degenerative effects of TNFα.

Astrocytes and microglia are mobilized after injury, infection, or in neurodegenerative diseases. Especially, the loss of dopaminergic neurons in Parkinson's disease is associated with an increased glial reaction. Astrocytes respond particularly to proinflammatory cytokines and it is believed that cytokines participate in astrocyte activation (McNaught and Jenner 2000). In organotypic slices cultures of fetal ventral mesencephalon (VM), dopamine neurons form their nerve fibers either in the absence of astroglial cell bodies or onto a monolayer of astrocytes (Johansson and Strömberg 2002). The two sequences of outgrowth are temporally separated, and the first formed nerve fibers appear prior to proliferation and migration of astroglia. When the astrocytes proliferate, they also migrate away from the tissue slice and form a monolayer to finally surround the tissue slice. As the astrocytes start to migrate, the initially formed nerve fibers are retracted and the secondary formed nerve fibers grow onto the astrocytes. Both types of outgrowth are often visible in the same culture over different areas of the periphery of the tissue slice, but over time, nerve fibers representing the initial wave of growth disappear and the newly formed nerve fibers are attached to astrocytes (Berglöf et al. 2007; Johansson and Strömberg 2003). Since proinflammatory cytokines are toxic to dopamine neurons, but leave a subset of dopamine neurons unaffected (McGuire et al. 2001), and affect astrocytes, which appear to exert important guiding cues on dopamine nerve fiber formation, the aim of this investigation was to study effects of TNFα on astrocytic migration and how this affected dopamine nerve fiber formation. Therefore, effects of TNFα were investigated during early or late treatment to determine possible diverging effects on nerve fiber formation with regard to their relationship to astrocytes.

2. RESULTS

In E14 ventral mesencephalic (VM) organotypic slice cultures, tyrosine hydroxylase (TH) -positive nerve fibers appeared in two morphologically different growth patterns, either in the presence or the absence of migrating astrocytes (Fig. 1). TH-positive nerve fibers were found in the absence of glial cells, as revealed by negativity to S100-immunoreactivity and DAPI-positive nuclei, while the TH-positive growth formed onto migrating astrocytes became obvious first when astrocytes had started to migrate to form a monolayer surrounding the tissue slice. The two growth patterns were often displayed over different areas of the periphery of the same tissue slice.

Fig. 1.

Dopamine nerve fiber outgrowth after treatment with TNFα. Dopamine neurons were visualized using tyrosine hydroxylase (TH) and astrocytes were visualized using S100, for cell nuclei staining, DAPI was used in VM organotypic tissue cultures. The VM cultures were treated with TNFα between days 4–7 (b) or between days 11–19 (d) and fixed either day 7 (a, b) or day 19 (c, d). The glial-guided TH-positive outgrowth is enhanced by the early treatment (b), and the astrocytic migration reached longer distances from the tissue slice compared to that measured in the control cultures at 7 DIV (a). TNFα treatment of VM cultures evaluated at 19 DIV (d) did neither demonstrate a difference for the glial-guided outgrowth nor for the astrocytic migration compared to control at day 19 (c). The non-glial-associated TH-positive outgrowth was determined by the absence of glial cells, seeing by the negativity to DAPI (e). In cultures processed for the astrocytic marker S100 and the proliferation marker Ki67 (f–h), the number of Ki67-positive cells did not differ between the control (f), cultures treated with TNFα (g), or TNFα plus antibodies against TNFR1 (h). Scale bar: a–d = 100 µm, e–h = 150 µm.

Effects of early TNF α treatment

VM cultures were treated with TNFα (40 ng/ml) and effects were examined at 7 or 19 days in vitro (DIV) using TH- and S100-immunohistochemistry. Treatment with TNFα during days 4–7 in vitro reduced the number of cultures with the non-glial-associated TH-positive growth. At 7 DIV, 16.0% of the cultures treated with TNFα displayed the non-glial-associated TH-positive growth while it was found in 36.8% of control cultures (Table 1). At 19 DIV, the number of cultures where the non-glial-associated growth was present was reduced to 31.6% in control cultures and remained at 16.7% after the early TNFα treatment (Table 1). The non-glial-associated TH-positive outgrowth often had a dotted appearance as if nerve fibers were degenerating at the longer time point.

Table 1.

The number of cultures expressing the non-glial- and glial-mediated TH-positive outgrowth in % of total number of cultures.

	7 DIV		19 DIV
	Control	TNFα days 4–7	Control	TNFα days 4–7	TNFα days 11–14
% of the cultures with the non-glial guided TH-positive outgrowth	36.8%	16.0%	31.6%	16.7%	0%
% of the cultures with glial-guided TH-positive outgrowth	78.9%	88.0%	94.7%	76.7%	96.2%

The early TNFα treatment (days 4–7) promoted the presence of glial-associated TH-positive outgrowth to an extent of 88.0% at 7 DIV compared to 78.9% in the control cultures (Table 1). The length of the glial-associated TH-positive outgrowth treated with TNFα between days 4–7 was significantly enhanced (p<0.001; F_6,115 =12.64; control n=30, TNFα days 4–7 n=27) compared to control cultures at 7 DIV (Fig. 1a, b and Fig. 2a). Nerve fiber outgrowth in treated cultures reached nearly twice the length compared to controls. At 19 DIV, there was no difference in the length of glial-associated TH-positive outgrowth after the early TNFα treatment (Fig. 2a). TH-positive neurons had no apparent change in morphology after TNFα treatment (Fig. 3a–d).

Fig. 2.

The distances for glial-guided TH-positive outgrowth and astrocytic migration from the periphery of the tissue slice. The distance of glial-guided TH-positive outgrowth in cultures treated with TNFα between days 4–7 (n=27) was significantly longer than that measured in control cultures at 7 DIV (n=30) (a). This difference seen at 7 DIV was no longer present at 19 DIV (control n=25, TNFα days 4–7 n=7, TNFα days 11–14 n=16). Thus, the enhanced outgrowth seen at the early time point in TNFα treated cultures, reached its maximal length before control cultures did. The migration of S100-positive astrocytes was significantly enhanced after TNFα treatment between days 4–7 at 7 DIV (n=28) (b). TNFα treatment between days 11–14 did no induce this effect when studying the cultures at 19 DIV (control n=30, TNFα days 4–7 n=10, TNFα days 11–14 n=17). ***p<0.001

Fig. 3.

Effects of TNFα on TH-positive cell morphology and on the presence of Iba-1-postive microglia. The appearance of TH-positive neurons (a–d) was not affected by the early (b) or late (d) TNFα treatment as compared to controls at 7 (a) or 19 (c) DIV. Microglia cells were more frequently found in cultures treated between days 11–14 (g, j) compared to the early treatment (f, i) and control cultures (g, h) at 19 DIV. A change in morphology of the microglia from a flat and rounded in control cultures and after the early treatment (e, f) to process-bearing microglia could be observed after TNFα treatment during days 11–14 (g). Scale bar: a–d = 50 µm, e–g = 75 µm, h–j = 230 µm.

Astrocytes migrated significantly longer distances (p<0.001; F_6,157 = 39.88; control n=34, TNFα days 4–7 n=28) from the periphery of the tissue slice after the early TNFα treatment when compared to control cultures at 7 DIV (Fig. 1a, b and Fig. 2b). At 19 DIV, no difference in astrocytic migration was observed between TNFα treated and control cultures.

Effects of late TNFα treatment

At 19 DIV, the non-glial-guided TH-positive outgrowth was absent in cultures treated with TNFα between days 11–14, while still present in 31.6% of control cultures (Table 1). There was no difference in the length of glial-associated TH-positive outgrowth after the late TNFα treatment when compared to control cultures at 19 DIV (Fig. 2a). Furthermore, TNFα treatment did not affect the distance that astrocytes migrated (Fig. 1c, d and Fig. 2b).

TNFα effects on the presence of microglia

The presence of Iba-1-immunoreactive microglia was studied after treatment with TNFα. The early TNFα treatment (days 4–7) did not affect the number of microglia that was present in the tissue slice at 7 DIV (Fig. 4a). At 19 DIV, the number of Iba-1-immunoreactive microglia was moderately reduced, but no significant difference compared to cell counts performed at 7 DIV or to control cultures at 19 DIV. Counting microglial profiles after the late TNFα treatment (11–14 DIV) demonstrated a significant augmentation in the number of microglia at 19 DIV, when compared to control cultures or cultures treated at the earlier time point (p<0.05; F_4,42 =1.94; control n=6; TNFα days 4–7 n=8; TNFα days 11–14 n=9; Fig. 3e–j and Fig. 4a).

Fig. 4.

Cell counts of microglia and astrocytic migration after blocking the effects of TNFα with antibodies against TNFR1, TNFR2, or TNFα. When counting the number of microglial cells using the pan-microglia marker Iba-1, no difference was found at 7 DIV (control n=11, TNFα days 4–7 n=13) (a). At 19 DIV the number of microglia was significantly increased after TNFα treatment between days 11–14 (n=9), compared to control (n=6) and to cultures treated between days 4–7 (n=8) (a). Astrocytic migration was determined in VM cultures treated with TNFα (n=28) and antibodies against either TNFR1 (n=4, 2.5 µg/ml) or TNFR2 (n=5, 2.5 µg/ml) alone, or TNFα together with the receptor antibodies between days 4–7 (b). The distance that astrocytes migrated from the tissue slice did not differ from control level (n=30) when blocking TNFα with antibodies against either TNFR1 (n=7) or TNFR2 (n=6) (b), while TNFα (n=28) treatment alone significantly enhanced astrocytic migration. Neutralizing TNFα by adding TNFα antibodies (n=7) resulted in reduced astrocytic migration compared to TNFα treated cultures. None of the treatments, except for TNFα, were changed from controls. *p<0.05, ***p<0.001

Effects of antibodies against TNFR1 or TNFR2

The effects of TNFα were investigated by adding antibodies against TNFR1 or TNFR2 to the medium to block the effects mediated through TNFR1 or TNFR2, respectively. VM slice cultures were treated between days 4–7 and TH- and S100-immunohistochemistry were performed at 7 DIV. First, single treatment with antibodies against TNFR1 and TNFR2 was investigated. Since the antibody solutions contained sodium azide, control cultures were given the same concentrations as in the antibody solutions, i.e. 12.5 µg/ml and 25.0 µg/ml sodium azide, and compared to untreated controls. The results revealed no differences in astrocytic migration, TH-positive outgrowth or distribution of non-glial vs. glial-associated outgrowths. Therefore control cultures were pooled to one group.

VM cultures treated with antibodies against TNFR1 or TNFR2 was performed in the concentrations of 2.5 and 5.0 µg/ml during 4–7 DIV. Both doses and antibodies affected TH-positive outgrowth such that the non-glial-mediated nerve fibers were present and measured similar lengths, around 1 mm, as compared to controls (1127± 114 µm for controls, 994±135 µm for TNFR1 antibody, 1117±86 µm for TNFR2 antibody using 5.0 µg/ml antibody). The glial-associated outgrowth could not be found after antibody treatment in any dose or treatment. The overall impression of TH-immunoreactivity in cultures treated with low (2.5 µg/ml) or high (5.0 µg/ml) dose of antibodies against TNFR1 or TNFR2 revealed that the neurons did not appear healthy, and the tissue contained many TH-immunoreactive dots as if neurites were degenerating. Most of the TNFR2 antibody-treated cultures in the high dose (5.0 µg/ml) came loose during the incubation. Therefore, no measurements were possible to be performed on those cultures. The high dose (5.0 µg/ml) of antibodies against TNFR1 reduced astrocytic migration from 459.6±30.9 µm in control cultures to 310.7±72.8 µm after treatment and the length became significantly shorter compared to TNFα treatment alone (p<0.05, F_9,115 = 10.19, control n=30, TNFR1 antibody n=7, TNFα n=28). The low dose of antibody treatment did not affect the attachment of the cultures to the substrate and both TNFR1 and TNFR2 antibody treatments reduced the lengths of astrocytic migration compared to TNFα treatment but did not differ from controls (Fig. 4b).

Effects of antibodies against TNFR1 or TNFR2 on TNFα treatment

At 7 DIV, cultures treated with TNFα in combination with antibodies against either TNFR1 or TNFR2 in two doses, a low (2.5 µg/ml) or a high (5.0 µg/ml) dose. TNFα treatment with antibodies against TNFR1 enhanced the number of cultures displaying the non-glial-associated growth to 50.0% (2.5 µg/ml) and 63.6% (5.0 µg/ml) from 16.0% of cultures treated with TNFα only. The addition of the high dose of antibodies against TNFR2 (5.0 µg/ml) did not affect the presence of non-glial-associated TH-positive growth compared to TNFα treatment only, 14.0% and 16.0%, respectively. However, when adding the low dose of TNFR2 antibodies (2.5 µg/ml) to TNFα treatment, the non-glial-associated growth was observed in 50.0% of the cultures (Fig. 5).

Fig. 5.

Blocking the effects of TNFα with antibodies against TNFR1, TNFR2 or TNFα. TH- and S100-immunoreactivity in control (a, d), and antibody treatment of TNFα (b, c, e), TNFR1 (f; 2.5 µg/ml), TNFR2 (h; 2.5 µg/ml) and TNFα plus antibodies against TNFR1 (g) and TNFα plus antibodies against TNFR2 (i) in VM cultures at 7 DIV. Blocking TNFα receptors promoted the presence of non-glial-associated TH-positive outgrowth. Neutralizing endogenous TNFα production by adding antibodies against TNFα abolished the presence of TH-positive nerve fibers (b). TH-positive neurons were found in cultures treated with antibodies against TNFα, but degenerating areas in the tissue slice was appearant by the areas of TH-immunoreactive dots (c) and S100-positive astrocytes was smaller in size (e) compared to controls (d). Scale bar: a, b, c = 100 µm, d, e = 50µm, f–i = 135 µm.

Blocking TNFR1 resulted in reduced number of cultures displaying the glial-associated TH-positive growth to 45.0% of the cultures after the early TNFα treatment. The distance these nerve fibers reached was significantly shorter (264.8±44.48 µm) than that measured for TNFα treated cultures (627.4±93.1 µm; p<0.01; F_8,131 =14,029; TNFα days 4–7 n=27; TNFα + TNFR1 antibodies n= 9), but within the same range as for control cultures at 7 DIV. In only 18.2% of the VM cultures treated with TNFα and antibodies against TNFR2 the glial-associated TH-positive growth was displayed. The distance these nerve fibers reached from the tissue slice was comparable to control cultures at 7 DIV. However, these TH-positive nerve fibers were visualized as dots rather than undisrupted nerve fibers and had no healthy appearance and were therefore never used for the statistics. The two doses used gave similar results.

By blocking either TNFR1 or TNFR2 (both doses) in VM cultures treated with TNFα between days 4–7, astrocytes migrated significantly shorter distances from the tissue slice compared to cultures treated with TNFα at 7 DIV (p<0.001; F_9,115 = 10.19; control n=30; TNFα days 4–7 n=28; TNFα + TNFR1 antibodies n=7; TNFα + TNFR2 antibodies n=6; Fig. 4b). There was no difference when compared to control cultures at the same time point.

Blocking TNFα function

Since the results from VM cultures treated with either antibodies against TNFR1 or TNFR2 alone or together with TNFα were similar, the next experiment was performed to neutralize any possible endogenous production of TNFα. Thus, adding TNFα antibodies to the medium resulted in no TH-positive outgrowth, although TH-positive neurons were present within the tissue slice. Furthermore, the TH-positive neurons were located to one cluster of neurons and no viable TH-immunoreactivity was found outside the area of TH-positive neurons (Fig. 5b, c). The S100-positive astrocytes migrated significantly shorter distances when compared to TNFα treated astrocytes (p<0.001; F_9,115 =10.19; TNFα antibody n=7: TNFα n=28), but did not differ from controls (Fig. 4b). In addition, the migrated S100-positive astrocytes appeared shrunken in size compared to astrocytes in any other treatment (Fig. 5d, e).

Effects of TNFα treatment on astrocyte proliferation

The distance that astrocytes migrated from the periphery of the VM tissue slice was significantly longer when treated with TNFα between days 4–7 compared to control cultures, an effect that was blocked by adding antibodies against TNFR1, TNFR2, or TNFα. Cultures were therefore processed for immunohistochemistry using the astrocytic marker S100 and for the antibody against Ki67. Ki67 is expressed in all proliferating cells during G1, S, M and G2 phases of the cell cycle. Cell counts of S100/Ki67-positive astrocytes revealed no difference in the number of cells in cultures exposed to TNFα or by adding TNFα along with TNFR1 antibodies as compared to control cultures 7 DIV (Fig. 1f–h and Fig. 6). Thus, TNFα stimulated the migration but not the proliferation of astrocytes.

Fig. 6.

Effects of TNFα on astrocytic cell division. In VM cultures treated with TNFα (n=6) or TNFα+antibodies against TNFR1 (n=9) between days 4–7 and studied at day 7 for S100 and the proliferation marker Ki67, there was no difference in the proliferation in the treated cultures compared to controls (n=13).

3. DISCUSSION

The main findings from this study demonstrate that TNFα exerts dual actions i.e. toxic and growth promoting. Thus, TNFα is toxic to the non-glial-associated nerve fiber production and the effect is mediated via TNFR1, while TNFα enhances the glial-guided TH-positive outgrowth during an early, but not late, event, which requires both TNFR1 and TNFR2. The triggered glial-guided nerve fiber formation might be an indirect effect caused by improved astroglia migration. Late TNFα treatment, between days 11–14, did not affect the glial-associated outgrowth but increased the number of microglia present in the cultures, while the early treatment did not affect the presence of microglia.

In previous studies, TNFα has shown to act detrimentally on dopamine neurons. For instance, intraparenchymal injection of TNFα into the medium forebrain bundle of the rat leads to dopamine neuron cell loss in ventral mesencephalon (Carvey et al. 2005). The toxic action has been suggested to be mediated via TNFR1 and data presented by Clarke et al. indicate improved survival of transplanted fetal dopaminergic neurons by including TNFR1 antibody in the cell suspension to block TNFα (Clarke and Branton 2002). The action of TNFα appears more complicated than regarding the effects either to be toxic strictly through TNFR1 or to act growth promoting via TNFR2 only, as previously suggested (Fontaine et al. 2002). For instance, TNFα levels are increased after MPTP treatment (Sriram et al. 2002). However, both TNFR1 and TNFR2 need to be deleted to protect against MPTP lesions (Sriram et al. 2002). Thus, one of the two receptors is not enough to block the toxic effect. In the present study, TNFα improved the glial-guided TH-positive outgrowth and migration of astrocytes. Separate blocking the exogenously added TNFα by each of the two receptors, equally affected the distances that migrating astrocytes and glial-guided outgrowth reached. It is known that TNFR1 is expressed in astrocytes (Figiel and Dzwonek 2007; McGuire et al. 2001). Furthermore, TNFα upregulates the expression of TNFR2, which normally is low in astrocytes (Choi et al. 2005). TNFα induces also astrogliosis (Sriram et al. 2002), which leads to enhanced production of astrocytic glial cell line-derived neurotrophic factor (GDNF) (Kuno et al. 2006). Interestingly, in VM cultures from GDNF knockout tissue, the astrocytic migration and glial-mediated nerve fiber outgrowth is impaired (af Bjerkén et al. 2006). Thus, blocking one of the two TNFα receptors might impair production of growth promoting factors. Furthermore, treating the VM cultures with antibodies against either of the two TNFα receptors alone abolished the glial-associated nerve fiber formation. Moreover, adding antibodies against TNFα to the cultures totally blocked all nerve fiber production. This indicates that the shortened glial-associated outgrowth seen after TNFα plus receptor antibody treatment was probably due to a dose effect. The results indicate that although adding TNFα to the cultures improved the glial-associated outgrowth there is an endogenous production of TNFα in the tissue, or the medium, which is important for nerve fiber formation during development, since no nerves were present after TNFα antibody incubation. Therefore, both TNFR1 and TNFR2 appear to be needed for TNFα to stimulate the migration of the astrocytes and the glial-guided TH-positive outgrowth at the early time point. Thus, TNFR1 appears in this situation to mediate growth-promoting effects.

The toxic effect of TNFα was demonstrated on the non-glial-associated outgrowth, which was reduced by half of the control cultures during the early treatment and absent after the late treatment. Blocking TNFR1 counteracted this effect. Thus, it appears that TNFα, when acting through TNFR2, promoted the presence of non-glial-associated nerve growth, especially since blocking TNFR1 actually enhanced the growth compared to control cultures. This was further demonstrated when blocking TNFR2 using the high dose, leaving TNFR1 active, which did not change the presence of non-glial-associated growth compared to treatment with TNFα alone. Thus, it seems that the subpopulation of TH-positive neurons producing the non-glial-associated nerve fibers are sensitive to TNFα toxicity via TNFR1. Furthermore, TNFα could promote the presence of the non-glial-associated growth when acting through TNFR2. Thus, in this case it appears as if TNFR1 and TNFR2 have direct opposite effects.

Although the presence of the non-glial-associated growth was affected by TNFα, it seemed that at 19 DIV, the TH-positive neurons producing the glial-associated nerve fibers were still robust, showing no difference from control cultures, when treated at the early or late time points. Other studies have shown a low, but still surviving subpopulation of cultured TH-positive neurons after treatment with TNFα (McGuire et al. 2001). Thus, subpopulations of dopamine neurons are affected differently by TNFα. Interestingly, it has been documented that TNFα treatment of cultures from E12.5 fetuses can enhance the number of dopamine neurons while the same treatment reduces the number of dopamine neurons in cultures from E14 and E16 fetuses (Doherthy 2007). This might be due to the fact that the survival time was much shorter in those experiments allowing no comparison for long-term culturing as in the present study. Thus, if TH-positive neurons producing the glial-mediated dopamine nerve fibers were lost during TNFα treatment, it did not affect the nerve fiber production at long-term survival.

When plating TH-positive neurons from fetal ventral mesencephalon, subpopulations of dopamine neurons from A8, A9, and A10 are included (Dahlström and Fuxe 1964). It is well known that A9 dopamine neurons are more vulnerable to toxic insults than A10 neurons (Barroso-Chinea et al. 2005; Gibb 1992; Gonzalo-Hernandez et al. 2004; Kish et al. 1988). It cannot be determined from this study whether TNFα treatment affected specifically A9 or A10 dopamine neurons. However, the late TNFα treatment killed 100% of the non-glial-mediated nerve fibers. A9 and A10 TH-positive neurons can form both the non-glial- and glial-guided TH-positive outgrowths (Berglöf et al. 2007), which suggests that the toxic effect of TNFα affected both A9 and A10 neurons and was not correlated to either subtype of dopamine neurons, but rather to the type of morphological nerve fiber growth. Indeed, the TNFα receptors are differently expressed over subgroups of lateral and medial substantia nigra dopamine neurons (Duke et al. 2007), which suggests that subgroups of A9 neurons need to be considered for differences in vulnerability. In conclusion, TNFα is toxic for the non-glial-associated dopaminergic nerve fiber outgrowth when mediated through the TNFR1, but not the TNFR2. Furthermore, TNFα stimulates the glial-guided outgrowth as well as the migration of astrocytes at an early time point, and this function requires both TNFR1 and TNFR2. Thus, TNFα exerts dual effects on VM dopamine neurons.

4. EXPERIMENTAL PROCEDURE

Dissection and tissue preparation

Fetal VM organotypic tissue cultures were prepared from tissue obtained from Sprague-Dawley pregnant rats. The animals were kept in a 12/12 h light/dark cycle and with access to food and water ad libitum. Pregnant rats were deeply anesthetized using isoflurane prior to neck dislocation, and the fetuses were obtained. The VM tissue was dissected from embryonic day (E) 14 fetuses under a dissection microscope and sterile conditions. Dulbecco’s modified Eagle’s medium (DMEM; Gibco) was used for dissection. The dissected tissue was sliced with a tissue chopper in 300 µm coronal sections and transferred into DMEM, and each tissue slice was cut in the midline. One such piece was used for each culture. The experiments were approved by the local ethics committee.

Tissue cultures

The tissue slices were plated on washed, autoclaved, and poly-D-lysine coated (5 mg/100 ml dH₂O; Sigma-Aldrich) 12 × 24 mm coverslips using the “roller-drum” culture technique (Gähwiler et al. 1997). Each tissue piece was placed in two drops of chicken plasma (Sigma-Aldrich). One drop of thrombin (1,000 units/ml; Sigma-Aldrich) was mixed with the chicken plasma and the tissue piece was placed in the center of the coverslip. The plasma/thrombin clot was dried for 15–20 minutes before the coverslips were placed in 15 ml Falcon tubes containing 0.9 ml of medium. The tubes were inserted into a “roller-drum” placed in an incubator. The cultures were continuously rotating at a speed of 0.5 turns per minute.

The culture medium contained 55% DMEM, 32.5% Hanks’ balanced salt solution (Gibco), 10% fetal bovine serum (Gibco), 1.5% glucose (Gibco) and 1% Hepes (Gibco). All ingredients were mixed and filtered through a sterile filter with pore size of 0.22 µm (Sterivex, Millipore). The medium was stored at −20°C and thawed to 37°C before use. Antibiotics were added to the medium to a final concentration of 1% (10,000 units/ml penicillin, 10 mg/ml streptomycin, 25 µg/ml amphotericin; Gibco) and used at plating. At the first medium change and thereafter, antibiotic was excluded. The medium was changed twice a week.

Rat tumor necrosis factor alpha (TNFα; 40 ng/ml; R&D Systems; see McGuire et al. 2001) was added to the medium between days 4–7 (n=65) or 11–14 (n=26) and the cultures evaluated at 7 (control n=102; TNFα days 4–7 n=47) or 19 (control n=36; TNFα days 4–7 n=18; TNFα days 11–14 n=26) days in vitro (DIV). Antibodies against TNFα receptor 1 (TNFR1; 2.5 µg/ml; n=7 and 5.0 µg/ml; n=16; goat polyclonal antibody; Santa Cruz Biotechnology, Inc., Santa Cruz, CA) or TNFα receptor 2 (TNFR2; 2.5 µg/ml; n=6 and 5.0 µg/ml; n=15; rabbit polyclonal antibody; Santa Cruz Biotechnology, Inc., Santa Cruz, CA, (Clarke and Branton 2002; Tacheuchi et al. 2006)) were added together with TNFα to the medium between 4–7 DIV and evaluated at 7 DIV. In addition, antibodies against TNFR1 and TNFR2 (2.5 µg/ml; TNFR1 n=4, TNFR2 n=5, or 5.0 µg/ml; TNFR1 n=7, TNFR2 n=6) were added without TNFα to test for specific effects of the antibodies per se. Since sodium azide was included in the antibody solutions, control cultures (n=14) also received the same concentration of sodium azide as given to the cultures including antibodies (12.5 µg/ml n=4; 25 µg/ml n=6). TNFα and TNFα + antibodies against TNFR1 or TNR2 were added once, when medium change was made at day 4.

Finally, neutralizing antibodies against TNFα (0.15 µg/ml; n=7; Serotec Oxford, UK) was added to the medium during days 4–7 in vitro. The dose was recommended by the manufacturer. The time points chosen for this study was based on when the two sequences of outgrowth appears: the non-glial-associated growth is pronounced already at day 3–5 and the glial-mediated outgrowth at day 10–14 (Johansson and Strömberg 2003).

Immunohistochemistry

The VM cultures were fixed at 7 or 19 DIV with 2% paraformaldehyde in 0.1 M phosphate buffer (pH=7.4) for 1 h. After fixation, the cultures were rinsed 3 times for 10 minutes with phosphate buffered saline (PBS; pH=7.4) before incubation in primary antibodies raised against the indirect marker for dopaminergic cells i.e. tyrosine hydroxylase (TH; diluted 1/300; mouse monoclonal; DiaSorin, MN, USA) followed by either S100 (diluted 1/400; polyclonal rabbit; Dako Patts, Denmark) to visualize astrocytes, Iba-1 (diluted 1/1000; rabbit polyclonal; Wako Chemicals, Germany) to visualize microglia, or Ki67 (diluted 1/150; mouse monoclonal; Novocastra Laboratories Ltd. UK), which is expressed in all proliferating cells during G1, S, M and G2 phases of the cell cycle (Schluter et al. 1993). Antibodies against Ki67 were applied after preincubation in PBS containing 0.3% Triton-x-100 for 1h before antibody application. Antibody incubations were performed in 48–72 h at 4°C. After rinsing in PBS, cultures were incubated with the secondary antibodies Alexa 488 (1/500; goat-anti-mouse or 1/200; goat-anti-rabbit; Molecular Probes, USA) and Alexa 594 (1/500; goat-anti-mouse or 1/500; goat-anti-rabbit; Molecular Probes, USA) for 1h at room temperature. Additionally, DAPI (1/50, Molecular Probes, USA) was used to stain the cell nuclei. Cultures were incubated in DAPI for 10 min at room temperature. All antibodies (except Ki67) and DAPI were diluted in 1% Triton X-100 in PBS. Incubations were performed in a humidified atmosphere. All cultures were triple labelled, which was performed in sequence with one antibody at the time. After additional rinsing the cultures were mounted in 90% glycerin in PBS.

Image analysis and statistics

Image analysis using Openlab software (Improvision) was utilized to analyze the cultures. The distances that the TH-positive nerve fibers had reached and the astrocytes had migrated were analyzed. The distances were measured from the periphery of the tissue slice to the distal end that the TH-positive nerve fibers or astrocytes had reached. The periphery of the tissue slice was defined based on the density of DAPI-positive cell nuclei. As the astroocytes started to migrate from the tissue piece, the density of DAPI-positive nuclei was drastically reduced. All cultures, where TH-positive nerve fiber outgrowth was determined, were processed for S100-immunohistochemistry. The TH-positive nerve fibers were defined as glial-associated when growing on astrocytes and as non-glial-associated when no astrocytic cells were present among the nerve fibers. Four measurements of glial-associated TH-positive growth as well as for astrocytic migration were performed for each culture. Cultures with less than 50 TH-positive neurons were excluded. The number of Iba-1-positive cells present in the tissue slices was counted using a standardized frame and a 20x objective in 2 areas per culture. Ki67-positive astrocytes were calculated in the area of migrating single layer of astrocytes using the same procedure as for the Iba-1-positive microglia. All measurements were performed randomly on a blind basis. Statistical analysis were performed on means per slice culture using one-way analysis of variance followed by Fisher post hoc analysis and expressed as means ± SEM.