A genetically targeted reporter for PET imaging of deep neuronal circuits in mammalian brains

Abstract Positron emission tomography (PET) allows biomolecular tracking but PET monitoring of brain networks has been hampered by a lack of suitable reporters. Here, we take advantage of bacterial dihydrofolate reductase, ecDHFR, and its unique antagonist, TMP, to facilitate in vivo imaging in the brain. Peripheral administration of radiofluorinated and fluorescent TMP analogs enabled PET and intravital microscopy, respectively, of neuronal ecDHFR expression in mice. This technique can be used to the visualize neuronal circuit activity elicited by chemogenetic manipulation in the mouse hippocampus. Notably, ecDHFR‐PET allows mapping of neuronal projections in non‐human primate brains, demonstrating the applicability of ecDHFR‐based tracking technologies for network monitoring. Finally, we demonstrate the utility of TMP analogs for PET studies of turnover and self‐assembly of proteins tagged with ecDHFR mutants. These results establish opportunities for a broad spectrum of previously unattainable PET analyses of mammalian brain circuits at the molecular level.

Mice expressing ecDHFR-EGFP or tdTomato (as control) from AAVs in forebrain were subjected to PET scans after peripheral administration of radioactive TMP analogs [ 11 C] TMP and [ 18 F]FE-TMP (40 MBq/mouse).
A Representative PET images (coronal, sagittal, and horizontal sections from left) generated by averaging dynamic scan data at 0-90 min after i.v. injection of [ 11 C]TMP. Arrowheads indicate areas of accumulation of radioactive ligand in animals carrying ecDHFR-EGFP (lower). White lines mark whole brain area as determined by MRI. B [ 11 C]TMP labeling kinetics. Volumes of interest (VOI) of fixed sizes were manually placed on paraventricular region exhibiting high-level radioactive signals. Data from control (n = 7) and ecDHFR-EGFP-expressing mice (n = 7) were plotted as mean AE SEM. F(1, 12) = 22.05; P < 0.01 (two-way ANOVA). C Ratios of averaged [ 11 C]TMP radioactive signals in ecDHFR versus control brains. D Representative PET images (coronal, sagittal, and horizontal sections from left) generated by averaging dynamic scan data at 0-180 min after i.v. injection of [ 18 F]FE-TMP. Arrowheads indicate areas of accumulation of radioactive ligand in animals carrying ecDHFR-EGFP (lower). White lines mark whole brain. E [ 18 F]FE-TMP labeling kinetics. VOI analysis was performed as described in panel c. Data from control (n = 6) and ecDHFR-EGFP-expressing mice (n = 6) were plotted as mean AE SEM. F(1, 10) = 326.1; P < 0.01 (two-way ANOVA). F Ratios of averaged [ 18 F]FE-TMP radioactive signals in ecDHFR versus control brains.
Source data are available online for this figure.

Masafumi Shimojo et al
The  Figure EV3. In vitro validation of TMP-HEX and [ 18 F]FE-TMP for ecDHFR.
A Representative images of striatal neurons expressing high-level ecDHFR-EGFP in fixed brain slices labeled with TMP-HEX. Note that incubation with excess amount of non-labeled TMP markedly inhibits fluorescence labeling. B In vitro autoradiography of mouse brain sections with [ 18 F]FE-TMP. Expression of transgenes in samples collected from mice treated with control and ecDHFR-EGFP vectors is fluorescently visualized with DAPI counterstaining (upper). Representative images of in vitro autoradiography using [ 18 F]FE-TMP demonstrate that this radioligand specifically labels brain areas overexpressing ecDHFR-EGFP but not tdTomato (middle), and that this radioligand binding to putative ecDHFR is profoundly blocked by an excess amount of non-labeled TMP (lower).

The EMBO Journal
Masafumi Shimojo et al

EV6
The EMBO Journal 40: e107757 | 2021 ª 2021 The Authors Figure EV5. Design and characterization of protein fragment complementation assay using a split DHFR system.

◀
A-D Cultured HEK293 cells expressing various constructs were incubated with 100 nM TMP-HEX and analyzed by fluorescent microscopy. TMP-HEX efficiently labeled cultured cells co-expressing NTF and CTF conjugated to a self-assembling leucine zipper (ZIP) motif, or a rapamycin-dependent heterodimerization motifs FKBP12rapamycin binding domain (FRB) and FK506 binding protein (FKBP) in the rapamycin-dependent manner, indicating that the cDHFR-PCA system functions under these conditions as designed. (A) Representative images illustrate a selective retention of TMP-HEX in cells carrying full-length ecDHFR (as positive control) or a combination of ZIP-tagged ecDHFR-NTF and ecDHFR-CTF. Insets demonstrate high-magnification fluorescence images overlaid with phase-contrast pictures of individual cells. Note that ZIP-tagged ecDHFR-NTF and CTF preferentially localize in nucleus. (B) Normalized fluorescence intensities of TMP-HEX in cells transfected with indicated constructs. Mean value of full-length ecDHFR (ecDHFR-FL) was set as 1. Note that only ecDHFR-FL and the split protein that reassembles via ZIP interaction retain the marker above background levels. Data from six independent experiments are plotted as mean AE SD. F(5, 28) = 21.12; *P < 0.05, **P < 0.01 (one-way ANOVA followed by Dunnett post hoc test). (C) HEK293 cells expressing various constructs were incubated with 100 nM TMP-HEX with or without 500 nM rapamycin (LC Laboratories) and imaged by fluorescence microscopy. Representative images illustrate selective labeling of TMP-HEX in cells co-expressing ecDHFR-NTF and CTF tagged with FRB and FKBP in the presence of rapamycin. Insets demonstrate high-magnification fluorescence photomicrographs overlaid with phase-contrast images. (D) Rapamycin (Rap)-induced FRB-FKBP interaction was assessed as TMP-HEX labeling efficiency in cells expressing various combinations of ecDHFR-NTF and ecDHFR-CTF fragments. Fluorescence intensities were normalized by mean value of labeling of full-length ecDHFR (ecDHFR-FL) with TMP-HEX. Data from four independent experiments are presented as mean AE SD. F(4, 15) = 55.08; **P < 0.01 (one-way ANOVA followed by Dunnett post hoc test). E A representative coronal PET image captured with [ 11 C]PBB3, a potent PET tracer for aggregated tau fibrils, in mice expressing TRD-NTF and TRD-CTF by AAV injection into one side of somatosensory cortex. Averaged image of dynamic scan data at 0-60 min after i.v. injection of [ 11 C]PBB3 is shown. A template MRI image was overlaid for spatial alignment of the PET image. Arrowhead indicates injection site. F Kinetics of [ 11 C]PBB3 in mouse brain during 60-min dynamic PET scan. VOIs were manually placed on ipsilateral and contralateral cortical areas for quantification. Data from four mice expressing TRD-NTF and TRD-CTF are plotted as mean AE SEM. Note there is no significant difference in radioactive signals of [ 11 C]PBB3 between ipsilateral and contralateral sides.
Source data are available online for this figure.