Toxic and Phenotypic Effects of AAV_Cre Used to Transduce Mesencephalic Dopaminergic Neurons

Abstract

A popular approach to spatiotemporally target genes using the loxP/Cre recombination system is stereotaxic microinjection of adeno-associated virus (AAV) expressing Cre recombinase (AAV_Cre) in specific neuronal structures. Here, we report that AAV_Cre microinjection in the ventral tegmental area (VTA) of ErbB4 Cyt-1-floxed (ErbB4 Cyt-1^fl/fl) mice at titers commonly used in the literature (~10¹²-10¹³ GC/mL) can have neurotoxic effects on dopaminergic neurons and elicit behavioral abnormalities. However, these effects of AAV_Cre microinjection are independent of ErbB4 Cyt-1 recombination because they are also observed in microinjected wild-type (WT) controls. Mice microinjected with AAV_Cre (10¹²-10¹³ GC/mL) exhibit reductions of tyrosine hydroxylase (TH) and dopamine transporter (DAT) expression, loss of dopaminergic neurons, and they behaviorally become hyperactive, fail to habituate in the open field and exhibit sensorimotor gating deficits compared to controls microinjected with AAV_GFP. Importantly, these AAV_Cre non-specific effects are: (1) independent of serotype, (2) occur with vectors expressing either Cre or Cre-GFP fusion protein and (3) preventable by reducing viral titers by 1000-fold (10¹⁰ GC/mL), which retains sufficient recombination activity to target floxed genes. Our studies emphasize the importance of including AAV_Cre-injected WT controls in experiments because recombination-independent effects on gene expression, neurotoxicity and behaviors could be erroneously attributed to consequences of gene ablation.

Keywords: Cre recombinase; adeno-associated virus; dopamine; stereotaxic injection; toxicity; tyrosine hydroxylase; ventral tegmental area.

Figure 1

AAV1_Cre-GFP injections into the VTA of ErbB4 Cyt-1^fl/fl mice cause hyperactivity and sensorimotor gating deficits relative to AAV1_GFP injected ErbB4 Cyt-1^fl/fl controls. (A) Scheme visualizing bilateral injection of Cre (AAV1_Cre-GFP) and control (AAV1_GFP; both 10¹³ GC/mL; for details see Figure S1) into the VTA of adult ErbB4 Cyt-1^fl/fl mice. (B,C) Locomotor activity in a novel environment was analyzed in the open field (ctrl n = 8, Cre n = 9). Cre-injected mice became hyperactive compared to control-injected mice ((B), two-way ANOVA, F(5,75) = 6.871, p < 0.0001 ****; genotype: p = 0.0013 **) and tend to spend less time in the center over time ((C), two-way ANOVA, F(5,75) = 7.106, p < 0.0001 ****; genotype: p = 0.2769). (D) Anxiety assessed in the elevated plus maze is unchanged (two-way ANOVA, F(1,15) = 2.848, p = 0.1122; genotype: p = 0.3752; ctrl n = 8, Cre n = 9). (E) Spontaneous alternation in the Y-maze is unaltered (unpaired t test, p = 0.7095; ctrl n = 8, Cre n = 7). (F) Prepulse inhibition is impaired in Cre-injected mice compared to controls (two-way ANOVA, F(3,45) = 2.237, genotype: p = 0.0334 *; ctrl n = 8, Cre n = 9).

Figure 2

AAV9_Cre injections into the VTA of ErbB4 Cyt-1^fl/fl mice cause hyperactivity relative to AAV9-injected ErbB4 Cyt-1^fl/fl controls expressing GFP. (A) Scheme visualizing bilateral injection of AAV9_Cre/AAV9_DIO-GFP (Cre) and control AAV9_GFP (both 10¹³ GC/mL, for details see Figure S1) into the VTA of adult ErbB4 Cyt-1^fl/fl mice. (B) AAV9_Cre-injected mice become hyperactive compared to control-injected mice in the open field (ctrl n = 9, Cre n = 9; mixed-effects model, F(5,77) = 0.6211, p = 0.6841 and genotype: p = 0.0083 **) and spent less time in the center of the maze ((C); ctrl n = 9, Cre n = 9; two-way ANOVA, F(5,80) = 0.4144, p = 0.8374 and genotype: p = 0.0327 *). (D) Anxiety measured in the elevated plus maze (ctrl n = 9, Cre n = 9; two-way ANOVA, F(1,16) = 0.3410, p = 0.5674 and genotype: p = 0.6416). (E) Spontaneous alternation in the Y-maze (ctrl n = 9, Cre n = 9; unpaired t-test, p = 0.5718). (F) Prepulse inhibition did not differ between groups (ctrl n = 9, Cre n = 9; two-way ANOVA, F(3,48) = 1.431, genotype: p = 0.6086).

Figure 3

Loss of key dopamine regulatory proteins and appearance of lesions after AAV_Cre injections into the VTA of ErbB4 Cyt-1^fl/fl mice. (A,E) Schematic depiction of microinjections of AAV1 and AAV9 serotypes driving expression of either Cre-recombinase, GFP-tagged Cre or GFP controls (all at 10¹³ GC/mL) into the VTA of adult ErbB4 Cyt-1^fl/fl mice. (B) AAV1_Cre-GFP injection cause a dramatic loss of DAT (magenta) and TH (white) protein in the mesencephalon after four weeks, quantified in (C,D) as mean intensity (ctrl n = 8, Cre n = 9; unpaired t-test, p < 0.0001 ****). (F) Immunohistochemical visualization of AAV9_Cre/AAV9_DIO-GFP injections show unaltered protein levels. (F’) Magnification of DAPI-negative holes in AAV9_Cre injected mice. (G,H; ctrl n = 9, Cre n = 9; unpaired t-test, DAT p = 0.4419, TH p = 0.1922), but cause nuclei (DAPI)-negative lesions in the mesencephalon. Scale bars 500 µm in (B,F) and 100 µm in (F’).

Figure 4

DAT and TH loss in the VTA of AAV1_Cre-GFP-injected WT mice coincide with hyperactivity. (A) Scheme visualizing unilateral injection of Cre (AAV1_Cre-GFP) and AAV1_GFP controls at 10¹³ GC/mL into the VTA of adult WT C57BL/6J mice (for details see Figure S1). (B,C) Cre-injected WT mice become hyperactive compared to GFP control-injected mice in the open field (ctrl n = 5, Cre n = 6; mixed-effect analysis, distance: F(2.327,20.47) = 8.179, p = 0.4653 and genotype: p = 0.0465 *, center time: F(5,45) = 1.380, p = 0.2498 and genotype: p = 0.2969). (D) Immunostainings of unilaterally injected WT mice with either AAV1_GFP (ctrl; top panels) and AAV1_Cre-GFP (Cre; bottom panels) four weeks after injection. (E,F) Quantification revealed a massive loss of TH (white) and DAT (magenta) in the ipsilateral (ipsi, injected) side of Cre-injected mice compared to the contralateral (contra, non-injected) side and AAV1_GFP control-injected mice (ctrl n = 5, Cre n = 6; two-way ANOVA, DAT: F(1,9) = 12.12, p = 0.0006 ***; ipsi ctrl vs. Cre p = 0.0019 **; Cre ipsi vs. contra p = 0.0003 ***; TH: F(1,9) = 9.171, p = 0.0143 ***; ipsi ctrl vs. Cre p = 0.0011 **; Cre ipsi vs. contra p = 0.0006 ***). Scale bar 500 µm.

Figure 5

VTA injections of AAV9_Cre at lower titers do not alter behavioral or histological phenotypes. (A) Scheme visualizing bilateral injection of AAV9_Cre (10¹⁰ GC/mL) mixed with AAV9_DIO_GFP (2.99 × 10¹² GC/mL; mixture hereafter simply denoted as Cre) and control AAV9_GFP (3.0 × 10¹² GC/mL) into the VTA of adult WT C57BL/6J mice. (B,C) Distance traveled and center time spent in the open field task does not differ between Cre- and AAV9_GFP control-injected mice (n = 11/group; (B) two-way ANOVA, F(5,100) = 0.7684, p = 0.5747 and genotype: p = 0.5909; (C) mixed effects model, F(5,96) = 1.866, p = 0.1074 and genotype: p = 0.4474). (D) Anxiety assessed in the elevated plus maze is unaffected in Cre- and control-injected mice (n = 11/group; two-way ANOVA, F(1,20) = 0.00036, p = 0.9851 and genotype: p = 0.6528). (E) Spontaneous alternation measured in a Y-maze is slightly decreased in Cre- compared to control-injected mice (ctrl n = 7, Cre n = 8; unpaired t-test, p = 0.0442 *). (F) Sensorimotor gating is largely unaffected (ctrl n = 7, Cre n = 8; mixed effects model, F(2.184, 26.93) = 43.11, genotype: p = 0.5549, 66dB p = 0.0243 * (exclusion of two Cre values (~−30%) at 66 dB in outlier test, otherwise data would be non-significant). (G) Immunohistochemistry revealed normal histology in Cre-injected mice. (H,I) DAT and TH expression are comparable between Cre- and control-injected mice. Scale bars 500 µm.