AP-1 and the injury response of the GFAP gene

Abstract

Increased GFAP gene expression is a common feature of CNS injury, resulting in its use as a reporter to investigate mechanisms producing gliosis. AP-1 transcription factors are among those proposed to participate in mediating the reactive response. Prior studies found a consensus AP-1 binding site in the GFAP promoter to be essential for activity of reporter constructs transfected into cultured cells, but to have little to no effect on basal transgene expression in mice. Since cultured astrocytes display some properties of reactive astrocytes, these findings suggested that AP-1 transcription factors are critical for the upregulation of GFAP in injury, but not for its resting level of expression. We have examined this possibility by comparing the injury response in mice of lacZ transgenes driven by human GFAP promoters that contain the wild-type AP-1 binding site to those in which the site is mutated. An intact AP-1 site was found critical for a GFAP promoter response to the three different injury models used: physical trauma produced by cryoinjury, seizures produced by kainic acid, and chronic gliosis produced in an Alexander disease model. An unexpected additional finding was that the responses of the lacZ transgenes driven by the wild-type promoters were substantially less than that of the endogenous mouse GFAP gene. This suggests that the GFAP gene has previously unrecognized injury-responsive elements that reside further upstream of the transcription start site than the 2.2 kb present in the GFAP promoter segments used here.

Keywords: Alexander disease; RRID:IMSR_JAX:003487; RRID:IMSR_TAC:fvb; STAT3 transcription factor; aging; brain injuries; cold injury; female; genetic; glial fibrillary acidic protein; gliosis; kainic acid; mice; seizures; sex characteristics; transcription; transcription factor AP-1; transgenic.

FIGURE 1.. Transgene diagrams.

hGFAP is the human GFAP DNA fragment used to generate the 73-7 line used as the AxD model (Messing et al., 1998). The WT-1, -2 and -3 transgene is the previously described gfa2-nlac (Brenner et al., 1994); its complete sequence is available from AddGene as part of pGfa2-nLac (Plasmid #53126). WT-4 is identical to WT-1, except that the promoter starts at bp −1757 instead of −2163. The AP1m-1, -2 and -3 transgene is identical to WT-4, except that the AP-1 site has been changed from TGACTCA to TTCAGAA. Nucleotide positions are relative to the RNA start site as +1. The positions for the AP-1 and STAT3 binding sites are the first (5’) nucleotide. The diagrams are not drawn to scale.

Figure 2.. Injury response of the 73-7 transgene.

The four transgene negative (control) and four 73-7 positive pups were all eight day old littermates. The adults (age in months) were each from a separate litter. The mRNA levels for mGfap and hGFAP were measured as described in Methods. The bar graph presents the average mRNA levels relative to the average for the 73-7 eight day old pups, with each individual value represented by a small circle. Vertical lines at the top of each bar represent the standard deviation. Within each mRNA type, the statistical differences between either the control or the adult group and the 73-7 eight day old pups was P < 0.0001 (Tukey post hoc), indicated by *** above the bar value, except that the difference for mGfap between the eight day control and 73-7 pups was not significant (P = 0.402). Note that the hGFAP mRNA levels for the controls were too low to be visualized. Actual numerical values are presented in the inserted table.

FIGURE 3.. Effect of age on the injury responses of mGfap and lacZ.

A. Responses of the endogenous mGfap gene. Data are pooled for all lines. The Y-axis is the fold increase of mGfap mRNA compared to controls. B. Responses of the lacZ transgene of the WT-1, WT-2, WT-3 and WT-4 lines. The Y-axis is the increase in lacz mRNA as a percent of the increase in mGfap mRNA for the same animal. Since different lines respond to different extents, the responses for each line were divided by the average response for that line so the data could be pooled to achieve a higher n value. Thus the lacZ responses have an overall average of 1.0. C. Responses of the lacZ transgene of the AP1m-1, AP1m-2 and AP1m-3 lines. Data were normalized as described for panel B. Linear regression was performed using GraphPad Prism version 7.02.