Nrf2 target genes can be controlled by neuronal activity in the absence of Nrf2 and astrocytes

The transcription factor nuclear factor erythroid 2-related factor 2 (Nrf2) is a master regulator of antioxidant genes, and its activation in astrocytes confers noncell-autonomous neuroprotection. A recent interesting and provocative study published in PNAS proposes that neuronal activity stabilizes Nrf2 expression in astrocytes, leading to astrocytic Nrf2-mediated gene expression (1). However, we show here that it is possible for neuronal activity to activate expression of known Nrf2 target genes independently of both Nrf2 and of astrocytes.

Habas et al. (1) observe that in mixed neuronal/astrocyte rat hippocampal cultures, electrical activity induced by GABA_A receptor inhibition in the presence of the K⁺ channel antagonist 4-aminopyridine induces the known Nrf2 target genes Gclc (glutamate-cysteine ligase, catalytic subunit) and Nqo1 (NAD(P)H dehydrogenase, quinone 1). The authors attribute this gene induction to activation of Nrf2 signaling in astrocytes because it was stronger in a mixed culture (17% astrocytes) than in a neuronal culture with fewer (4%) astrocytes. However, the presence of a significant number of astrocytes in both preparations makes it difficult to make definitive conclusions. Moreover, in the absence of Nrf2 knockout/knock-down studies, it is not completely clear whether the induction of Gclc and Nqo1 was Nrf2-dependent, particularly as Nrf2 target genes can be controlled by other transcription factors.

For example, we and colleagues have studied the transcriptional regulation of sulfiredoxin 1 (Srxn1) and xCT (2, 3). These are both Nrf2 target genes, as confirmed below: the Nrf2 activator tert-Butylhydroquinone (tBHQ) induces expression of both Srxn1 and xCT (system X_c- transporter) in Nrf2^+/+ but not Nrf2^−/− mixed mouse cortical cultures (Fig. 1, Top) [antimitotic AraC added day in vitro 4 (DIV4): 10% astrocytes, 90% neurons]. Moreover, both genes are transcriptionally induced by synaptic activity triggered by GABA_A receptor inhibition (with bicuculline) in the presence of 4-AP (BiC/4-AP) treatment (2, 3). However, activity-dependent induction of these Nrf2 target genes requires neither astrocytes nor Nrf2. In a culture preparation nearly devoid of astrocytes (AraC added on day of plating, <0.2% astrocytes) (Fig. 1, Middle), Srxn1 and xCT induction remains robust compared with astrocyte-enriched neuronal cultures (no AraC added, ∼15% astrocytes). Furthermore, activity-dependent Srxn1 and xCT induction is Nrf2-independent, revealed by studying Nrf2^−/− cultures (Fig. 1, Bottom) (Srxn1; see ref. 3 for xCT). Activity-dependent induction of xCT and Srxn1 is mediated by ATF4 and AP-1, respectively (2, 3).

Fig. 1.

(Top) Nrf2^+/+ or Nrf2^−/− mouse cortical mixed astrocyte/neuronal cultures (DIV9) were treated with tBHQ (10 µM) for 8 h. Srxn1 and xCT mRNA expression was measured (relative to Gapdh) by quantitative RT-PCR as described previously (2, 3); *P < 0.05 (two-tailed t test). (Middle) Cortical mouse neuronal cultures were prepared with either a high proportion of astrocytes [by omitting any antimitotic AraC from the culture process: astrocyte-enriched (AE)] or virtually devoid of astrocytes [by adding an antimitotic agent AraC on the day of plating; <0.2% GFAP⁺ cells, astrocyte-restricted (AR)]. AE- and AR-neuronal cultures were treated in parallel with BiC/4-AP (as described in ref. 2) for 24 h, followed by quantitative RT-PCR analysis of Srxn1 and xCT mRNA expression. All values are relative to Gapdh and normalized to the unstimulated level expressed in AE-cultures; *P < 0.05 compared with Con (one-tailed paired t test, n = 4). (Inset) Example pictures of GFAP-immunofluorescent staining of AE-(no AraC) and AR-(AraC on DIV0) neuronal cultures at DIV9. (Scale bar, 25 µm.) (Bottom) Nrf2^+/+ or Nrf2^−/− mouse cortical mixed astrocyte/neuronal cultures (DIV9) were treated with BiC/4-AP for 8 h, followed by analysis of Srxn1 mRNA levels. *P < 0.05 compared with Con (two-tailed paired t test, n = 4).

Habas et al.’s (1) observation that rat hippocampal neuronal activity induces Nrf2 protein levels in astrocytes is intriguing, although some densitometric quantification would have been desirable. Using a different, more specific Nrf2 antibody (Fig. 2, Upper), we found that treatment of mixed mouse cortical cultures (no AraC, 15% astrocytes) with K⁺ (50 mM) or BiC/4-AP at time-points used by Habas et al. did not strongly affect Nrf2 levels, compared with Nrf2 activators sulforaphane or tBHQ (Fig. 2, Lower). Note that we used a different culture system from Habas et al. (mouse cortical vs. rat hippocampal cultures). Regardless, neuronal activity has the capacity to induce at least some Nrf2 target genes independently of Nrf2 and astrocytes within the neurons themselves. The potential cooperation of this with the activity-dependent induction of astrocytic Nrf2 reported by Habas et al. would be an intriguing topic for further investigation.

Fig. 2.

(Upper) To establish the specificity of anti-Nrf2 antibody (Cell Signaling D1Z9C; 1:500), Nrf2^+/+ or Nrf2^−/− astrocytes were treated with MG132 (MG, a proteasome inhibitor), tBHQ (TB), or sulforaphane (SF) as indicated for 16 h before Western analysis. Note that in Nrf2^−/− astrocytes, the knockout strategy (4) involved knocking LacZ cDNA in-frame into exon 5, resulting in a fusion protein comprising the N-terminal 301 amino acids of Nrf2 fused to LacZ that remains subject to Keap1-mediated degradation. The anti-Nrf2 antibody targets the N terminus of Nrf2 and so the induction of the Nrf2^NT-LacZ fusion can be seen in the Nrf2^−/− lanes. (Lower) Astrocyte-enriched mouse cortical neuronal cultures (no AraC, ∼15% astrocytes) were treated with BiC/4-AP (BiC) for 24 h or 50 mM K⁺ for 4 h or 24 h [K⁴, K²⁴, as previously described (5)], and Nrf2 levels measured. Positive controls were included (MG132, 10 µM; tBHQ, 10 µM; sulforaphane, 5 µM). Expression was normalized to β-actin loading control and expressed relative to the control level. n = 5 for all treatment except tBHQ (n = 3) and 24 h KCl (n = 4). *P < 0.05 (two-tailed t test). Two example blots are shown. Exposures used for analysis were relatively long, because they were selected to detect any changes relative to the control condition. As a result, the degree of induction triggered by tBHQ and sulforaphane may be underestimated, and the MG132-treated lane was saturated.