MRI reveals differential effects of amphetamine exposure on neuroglia in vivo

Abstract

How amphetamine affects the neuroglia in living brains is not well understood. In an effort to elucidate this effect, we investigated neuroglia in response to amphetamine exposure using antisense (AS) or sense (S) phosphorothioate-modified oligodeoxynucleotide (sODN) sequences that correspond to glial fibrillary acidic protein (GFAP) mRNA (AS-gfap or S-gfap, respectively) expression. The control is a random-sequence sODN (Ran). Using cyanine 5.5-superparamagnetic iron oxide nanoparticle (Cy5.5-SPION) labeling and fluorescent microscopy, we demonstrated that living neural progenitor cells (PC-12.1), as well as the cells in fresh brain slices and intact brains of male C57BL6 mice, exhibited universal uptake of all of the sODNs but rapidly excluded all sODN-Ran and most S-gfap. Moreover, transmission electron microscopy revealed electron-dense nanoparticles only in the neuroglia of normal or transgenic mice [B6;DBA-Tg(Fos-tTA, Fos-EGFP*)1MmayTg(tetO-lacZ,tTA*)1Mmay/J] that had been administered AS-gfap or Cy5.5-SPION-gfap. Subtraction R2* maps from mice with acute and chronic amphetamine exposure demonstrated, validated by postmortem immunohistochemistry, a reduction in striatal neuroglia, with gliogenesis in the subventricular zone and the somatosensory cortex in vivo. The sensitivity of our unique gene transcript targeted MRI was illustrated by a positive linear correlation (r(2)=1.0) between in vivo MRI signal changes and GFAP mRNA copy numbers determined by ex vivo quantitative RT-PCR. The study provides direct evidence for targeting neuroglia by antisense DNA-based SPION-gfap that enables in vivo MRI of inaccessible tissue with PCR sensitivity. The results enable us to conclude that amphetamine induces toxicity to neuroglia in vivo, which may cause remodeling or reconnectivity of neuroglia.

Figure 1.

A) Rat PC-12.1 neural progenitor cells, in F12 medium with 10% horse serum in a dish with a polylysine-free (noncoated) glass surface, retained antisense Cy3-AS-gfap in the nucleus (solid arrows) and cytoplasm (dashed arrows) for ≥2 h after transfection. B) We compared retention of 3 sODNs: sense sequence to gfap mRNA (Cy5-S-gfap, labeled with Cy-5); an 18-mer with a randomized sequence, no intracellular target (FITC-Ran); and a control antisense to histone deacetylase 5 (HDAC5) mRNA (Cy3-AS-hdac5; used to avoid the presence of a target for Cy5-S-gfap had Cy3-AS-gfap been used). PC-12.1 cells retained all sODN during transfection. C) FITC-Ran was excluded or degraded immediately at 20 min (C) after medium wash; Cy5-S-gfap was retained in the cells for in both the nucleus and cytoplasm for ≥70 min (not shown).

Figure 2.

Fresh living brain slices with thickness of 60 μm were harvested and incubated in F12 medium with 10% horse serum in a dish with a polylysine-free (noncoated) glass surface. A) We transfected 3 sODNs (each at 10 nM) as in Fig. 1. All sODNs were retained in neural cells (dashed arrows). After 30 min, sODN was removed by washing twice in sODN-free medium, and then the tissue was incubated in fresh medium (time 0). B, C) Photographs show different time points, 5 min (B) and 10 min (C) after sODN removal by washing in fresh medium. Solid arrows point to neural cells that exhibited slow exclusion or degradation of Cy5-S-gfap. A fast exclusion or degradation of FITC-Ran in these cells suggests the cells are viable. Scale bars = 12.5 μm.

Figure 3.

We transfected Cy5-AS-gfap (5 pmol/mouse, i.c.v.; n=2) to live transgenic (B6;DBA-Tg(Fos-tTA, Fos-EGFP*)1MmayTg(tetO-lacZ,tTA*)1Mmay/J) mice that produce GFP in neural cells expressing cFos. Because GFP induction requires c-Fos antigen, we administered amphetamine (4 mg/kg, i.p.) 3 h after Cy5-AS-gfap delivery to induce c-Fos antigen, and prepared brain samples 1 h later. Cy5-AS-gfap is shown as pseudocolor purple (A1); propidium iodide (red; A3) stains the nuclei (confocal microscopy). Neural cells with small nuclei (<5 μm; dashed arrows) take up Cy5-AS-gfap (A1–A4) and Cy5.5-SPION-gfap (B1–B3), express GFAP antigen (B, C), and retain SPION-gfap (D). The neuronal formation of the dentate gyrus (DG) shows neither GFAP antigen (C, solid arrow) nor signal reduction associated with SPION-gfap (D, solid arrow); however, signal reduction as the result of SPION-cfos retention is observed in the DG (E). No enhanced image contrast is observed in mice without SPION-sODN infusion (F). The cornu ammonis (CA) formation contains mixed neurons and neuroglia, and signal reduction appears for both SPION-gfap (D) and SPION-cfos (E) in MRI. All animals (n=2 each, A1–F) received 40 μg Fe/kg, 120 pmol sODN/kg, or saline (2 μl) by i.c.v. route.)

Figure 4.

SPION-gfap uptake in glia of 3 mouse samples (partially stained with uranyl acetate) was examined by TEM. A) Several EDNs (boxes) were present around a small vessel (perivascular uptake); no EDN was present in the vascular lumen. B–B2) Intercellular border of a neuron (B, B2; dashed arrows) and enlarged views of EDN in boxed areas in B (B1, B2; solid arrows). N, neuronal nucleus; G, glial nuclei. C, C1) One EDN (C1; solid arrow) within the endoplasmic reticulum in neuroglia. Fully stained samples (osmium, uranyl acetate, and lead citrate) are presented in Supplemental Fig S2.

Figure 5.

Optimal contrast for MRI in vivo. A) Representative R₂* maps at 2, 4, and 6 h after i.c.v. infusion of SPION-gfap with a scale bar ranging from 10 to 120 s⁻¹. B) ROIs from which R₂* values were extracted for panels C and D (32). C) ΔR₂* values in all ROIs at 2, 4, and 6 h after i.c.v. infusion of SPION-gfap; Δ R₂* is significantly elevated in most ROIs at 4 and 6 h. *P < 0.05; t test. D) CNR analysis indicates that although CNR was much improved at 4 h, regional variation among the ROIs was reduced significantly thereafter; i.e., probe distribution had reached steady state at 6 h.

Figure 6.

Amphetamine-induced changes in GFAP mRNA expression by GT tMRI and RT-PCR. At 2 wk before MRI, 6 mice were given saline (n=4) or amphetamine (4 mg/kg, i.p., n=2) injections, once every other day for a total of 7 doses. On the day of MRI, we administered SPION-gfap to all 6 mice. After 3 h, we subdivided the 4 mice in the saline group, and injected them with saline (SAL group; n=2) or amphetamine (A1 group; acute exposure, n=2); the animals in the chronic exposure paradigm (A8 group) were administered one final amphetamine dose (30); there was no drug withdrawal period in this study. We then acquired R₂* maps of the brains 3 h later (i.e., 6 h after i.c.v.), as described herein for optimal CNR, the earliest time point for sufficient SPION-gfap MR signal specificity. The entire protocol was repeated, as determined by power analysis (P=80%; α=0.05) of the data from the first MRI. A) MRI paradigm used on the day of experiment. B) Changes in ΔR₂* after amphetamine showed differential decrease across the SAL, A1, and A8 groups in all brain regions except hippocampus (hippo). C) Experimental paradigm used on the day of ex vivo gene activity measurements. D) RT-PCR of GFAP mRNA collected 30 min after amphetamine showed significant reduction in gfap mRNA levels in the striatal regions (CPu), but not in hippocampus; this was true for all paradigms. Means and se (error bars) from each group are shown. Western blotting showed a slightly, but not significant elevation of striatal GFAP levels in A1 group (n=3) and A8 group (n=5) (Supplemental Fig S6).

Figure 7.

Amphetamine-induced gliogenesis (dashed arrows) in the SVZ of A8 group. A) Subtraction map illustrating the percentage increase in the A8 group, compared to the A1 group [(A8 − A1)/A1 × 100%], scale = 0–100%. Solid arrows indicate gliogenesis in the SSC. B) Greater number of neuroglia (gliogenesis) in the SVZ in A8 mice (1 of 4 is shown) compared to the SVZ of a normal mouse. V, cerebral ventricle.