The SPOTlight reporter design is based upon the 5’UTR of ATF4 mRNA, whose regulation by the ISR is extensively characterized (Dever et al., 2023; Lu et al., 2004; Pakos-Zebrucka et al., 2016; Vattem & Wek, 2004). The ATF4 5’UTR contains multiple uORFs: the very short uORF1 is self-contained upstream, whereas uORF2 overlaps out-of-frame with the ORF encoding the ATF4 protein and represses its translation when eIF2α is not phosphorylated. When the ISR is activated by eIF2α phosphorylation, the delivery rate of initiator methionyl tRNA to scanning ribosomes decreases. When this occurs, evidence supports that scanning ribosomes translate uORF1, then pass through uORF2’s start codon before receiving another initiator tRNA, resulting in less uORF2 translation inhibiting initiation at the main ATF4 coding ORF. In SPOTlight, enhanced green fluorescent protein (eGFP) cDNA is inserted after the uORF2 methionine start codon and tdTomato cDNA is inserted in place of the ATF4 coding sequence (Figure 1A).

To generate DIO-SPOTlight, we modified SPOTlight to allow for Cre-dependent expression using a LoxP DIO system (Sohal et al., 2009). This design reduces the leaky expression seen using a traditional lox-Stop-lox cassette (Smedemark-Margulies & Trapani, 2013). In addition to the transgenic mouse line, DIO-SPOTlight cassette was cloned into an AAV packaging vector available from Addgene (submission in process to Addgene) (Figure 1 – Supplement 1A-B).

Generation of the Rosa26-DIO-SPOTlight transgenic mouse line (Figure 1B) was guided by the targeting strategy used for the Ai9 mouse line (Madisen et al., 2010). The R26Sor5’-pCAG-DIO-SPOTlight-WPRE-bGHpA-AttB-NeoR-AttP-R26Sor3’ cassette was targeted to the Rosa26 locus in hybrid 129S6/C57BL/6N (G4) embryonic stem cells and inserted through homologous recombination. Correctly targeted clones were selected by neomycin resistance and microinjected into morulae from the Institute of Cancer Research (ICR) mouse strain to produce chimeric mice. Chimeric founder mice were identified using PCR and DNA sequencing. Founder colonies were expanded then crossed with Nestin-Cre mice (Tronche et al., 1999) to achieve pan-neuronal expression throughout the CNS (Figure 1B). Nestin-Cre::DIO-SPOTlight hemizygous mice showed broad expression in the brain and spinal cord consistent with the expected pan-neuronal distribution of Nestin-Cre (Figure 1C-D). In general, green fluorescence was the dominant signal in cells, consistent with the understanding that the ISR is not activated in most cells of healthy normal tissue under basal conditions.

While qualitative visual inspection of brains from 30-day-old (P30) Nestin-Cre::DIO-SPOTlight mice showed a majority of cells with green signal predominating, there were also rare distinctive cells that were strongly red fluorescent without substantial green fluorescence (“red hot” cells, Figure 2A-C). In many cases, these cells were distinguished by their unique SPOTlight readout from the majority of cells of similar cell type surrounding them. Importantly, these cells could also be identified in live tissue without fixation (Figure 2A).

In the dentate gyrus of hippocampus, visual scanning revealed that there were rare, dispersed cells distinguished by strong red SPOTlight signal but with morphology that was otherwise similar to neighboring granule cells. We show one striking example of a cluster of these cells (Figure 2B) providing an example of the utility of DIO-SPOTlight mice to highlight novel biology for follow-on studies. The dentate gyrus is unusual in that it is also a specialized area for neurogenesis. We wonder if these cells might share a “birthday” and/or might reflect a specific cell cycle state.

In cerebellum, we occasionally saw another unusual cluster of red-predominant cells, with distinctive morphology consistent with that of Bergmann glia (Figure 2C). This observation provides functional data to support a recent study that identified Bergmann glial cells with higher than normal ISR activation using gene expression profile data (Abbink et al., 2019). In addition, similar to what was previously observed with the viral reporter (Helseth et al., 2021), Nestin-Cre::DIO-SPOTlight mice revealed functional state variation across Purkinje cells.

Lastly, we recently reported that dorsal striatal cholinergic interneurons (CIN) have distinctive molecular hallmarks and functional SPOTlight activity indicative of steady-state ISR activity (Helseth et al., 2021). Here, we examined whether Nestin-Cre::DIO-SPOTlight mice reproduce those prior SPOTlight viral reporter observations and extended this analysis to consider whether this feature was present in other striatal subdivisions. Dorsal striatal subdivisions (medial, lateral, and tail) and ventral striatum (nucleus accumbens core and shell regions) each have distinctive input and output circuitry and functions. Using P30 Nestin Cre:DIO-SPOTlight mice, red and green fluorescence was measured in somata of cholinergic interneurons (ChAT+) and nearby non-ChAT+ cells in each striatal subdivision. Values obtained from Nestin-Cre::DIO-SPOTlight mice confirmed prior findings that dorsal striatal cholinergic interneurons show SPOTlight readouts characteristic of elevated ISR activity in comparison to neighboring non-ChAT neurons (Figure 2D-E). The mean ratio of red:green (R:G) fluorescence in dorsal striatal ChAT+ cells was more than twice the value of nearby non-ChAT cells (ChAT: 0.18 +/- 0.0074 SEM, n = 338 cells/6 mice, non-ChAT: 0.078 +/- 0.0082 SEM, n = 1154 cells/6 mice, P = 0.0001) yielding a log₂ fold change of 1.221 +/- 0.18 SEM (P = 0.0001). Second, we found that this property was similar across striatal subdivisions, with all subdivisions except for the nucleus accumbens core showing statistically significant increases in R:G ratios for ChAT interneurons relative to neighboring non-ChAT neurons (Figure 2 – Supplement 2). These data are shown normatively using log2 fold change of the R:G ratio in ChAT+ cells relative to non-ChAT neighboring cells (Figure 2D and Figure 2 – Supplement 2) (DLS-Tail: 1.588 +/- 0.18 SEM, ChAT(+) n = 41 cells/6 mice, non-ChAT n = 128 cells/ 6 mice, P = 0.0002; DLS: 1.187 +/- 0.22 SEM, ChAT(+) n = 143 cells/ 6 mice, non-ChAT n = 507 cells/6 mice, P = 0.0072; DMS: 1.112 +/- 0.18 SEM, ChAT(+) n = 154 cells/6 mice, non-ChAT n = 519 cells/6 mice, P = 0.001; NAc core: 0.548 +/- 0.43 SEM, ChAT(+) n = 93 cells/6 mice, non-ChAT n = 434 cells/6 mice, P= 0.172; NAc shell 0.655 +/- 0.18 SEM, ChAT(+) n = 112 cells/6 mice, non-ChAT n = 342 cells/6 mice, P = 0.019). Finally, a multiple comparison analysis showed no significant differences in R:G ratios or normalized fold change between the ChAT populations in each subdivision. Thus, SPOTlight reporter values indicate that ISR-regulated protein synthesis initiation dynamics are increased throughout the striatum, while trends suggest there might be modestly higher activity in dorsal versus ventral subdivisions.

In summary, these examples demonstrate the utility of DIO-SPOTlight mice to identify distinctive functional properties in rare, distributed cell populations – whether these sparse cells are unanticipated and brought to attention by their SPOTlight “red hot” signal or are sparse and of specific interest, like striatal cholinergic interneurons which make up only 5% of striatal neurons.

We next determined the performance of Nestin-Cre::DIO-SPOTlight mice in the setting of an acute stressor that activates the ISR. Tunicamycin (3 µg/g), a drug that inhibits N-linked glycosylation of proteins and is a potent trigger of ER stress, or vehicle was delivered to P12 mice in 2 subcutaneous injections that were 2 hours apart, followed by euthanasia and brain tissue fixation 16-18h later (Figure 3A). In this experimental design, the R:G ratio was not significantly different between conditions, but did trend higher in the tunicamycin group (vehicle: 0.042 +/- 0.0015 SEM, n = 799 cells/5 mice; tunicamycin: 0.055 +/- 0.0016 SEM, n = 770 cells/5 mice, P = 0.145, Figure 3B). We next examined the red and green components individually to provide further insight to their response kinetics under these conditions. Consistent with activation of the ISR, SPOTlight red fluorescence showed a highly significant increase to levels approximately 60% above vehicle (vehicle: 310 +/- 7.717 SEM, n = 799 cells/5 mice; tunicamycin: 505.6 +/- 9.297 SEM, n = 770 cells/5 mice, P = 0.0066, Figure 3C). There was no significant difference in green fluorescence, but there was a slight trend towards increased values with tunicamycin treatment (vehicle: 8483 +/- 135.9 SEM, n = 799 cells/5 mice; tunicamycin: 11113 +/- 176.25 SEM, n = 770 cells/5 mice here, P = 0.1218, Figure 3D). Representative images, including a SPOTlight transgene-negative mouse, are provided in Figure 3E.

In summary, the increases in R:G ratio and red fluorescence mean values are consistent with predicted effects of ISR activation. However, the trend of an increase in mean green fluorescence was opposite to the predicted response to ISR activation. Although we waited 16 – 18 h from stressor to evaluation timepoint, the time course of this tunicamycin experiment is relatively short in comparison to the half-lives for degradation of existing fluorophores (i.e. eGFP, t ½ approx. 24h). We can therefore expect that some pre-existing eGFP that has not yet been degraded contributes to the R:G ratio. Perhaps more importantly, tunicamycin is known to also activate the unfolded protein response (UPR). Prior studies have shown that tunicamycin induction of the UPR causes a transient inhibition of the ubiquitin-protease system (UPS) and increased accumulation of cytosolic and ER proteins (Menéndez-Benito et al., 2005; Shenkman et al., 2007). This acute effect might relate to the trend for increased eGFP observed. This acute experiment provides an example of how SPOTlight components and ratio can vary according to the nature and acuity of an intervention, where varying contributions of proteins synthesized earlier may not yet be degraded (Figure 3F).

Lastly, we wanted to provide an example of using Nestin-Cre::DIO-SPOTlight mice to comprehensively survey cell types across distant regions and with heterogenous local tissue properties. We returned to the consideration of cholinergic neurons, now including populations of forebrain, cranial nerve and spinal cord nuclei. In comparing across more diverse tissue than intra-striatal regions, we sought to control for variations in tissue autofluorescence/background and SPOTlight signal related to surrounding neuropil density by using non-ChAT neighboring cell reference measures. We found that even though there was some variation in non-ChAT neighbor means across the regions, the vast majority of non-ChAT cell populations show mean R:G ratios around 0.1, whereas many ChAT populations have values more than twice that (Figure 4A). To normalize for variations in the non-ChAT neighbor means, we present a summary graph in which we have ordered the regions according to their normalized fold increase values measured relative to their non-ChAT neighbors (Figure 4B, same order is used in Figure 4A for direct comparison). From this analysis, it becomes apparent that most cholinergic neuron populations throughout the CNS show significantly elevated R:G ratios compared to their non-ChAT neighbors.

Further collapsing all cells measured brain-wide, group means illustrate that ChAT neurons have a significantly increased R:G ratio that is more than double that of non-ChAT neighbors (ChAT: 0.23 +/- 0.004 SEM, n = 2084 cells/6 mice; non-ChAT: 0.096 +/- 0.01 SEM, n = 3641 cells/6 mice, P = 0.0001). The ratio increase was accompanied by significant differences in both fluorophores, with ChAT+ neurons having significantly lower eGFP and higher tdTomato levels (tdT: ChAT: 1038 +/- 76.74 SEM, non-ChAT: 804 +/- 50.33 SEM, P = 0.03; EGFP: ChAT: 6073 +/- 196.6 SEM, non-ChAT: 10389 +/- 562.6 SEM, P < 0.0001). These SPOTlight reporter findings are consistent with the functional consequences predicted for ISR activation on Atf4 mRNA reading frame dynamics (Dever et al., 2023). Finally, we did not observe any sex-dependent effects on SPOTlight measures, either in ChAT+ cells or non-ChAT cells (R:G ratios: ChAT male 0.24 +/- 0.045 SEM, n = 1155 cells/3 mice, female 0.23 +/- 0.005 SEM, n = 939 cells/3 mice, P = 0.437; Non-ChAT male 0.098 +/- 0.02 SEM, n = 1877cells/3 mice, female 0.094 +/- 0.013 SEM, n = 1764 cells/3 mice, P = 0.336, Figure 4 – Supplement 1A).

The original SPOTlight construct was published in Helseth et al., 2021. Briefly, SPOTlight was an in-house designed reporter consisting of recombinant DNA based on the known published sequence of murine Atf4 mRNA (NCBI Reference Sequence: NM_009716.3). To achieve Cre-recombinase dependent expression, SPOTlight was cloned into AAV vector driven by the CAG promoter in a double-floxed inverted orientation (DIO) containing two pairs of incompatible Lox sites (LoxP and Lox2722) by Vector Builder (https://en.vectorbuilder.com/).

Rosa26-DIO-SPOTlight mouse was produced by first cloning DIO-SPOTlight construct into Ai9 vector backbone in place of the Lox-STOP-Lox-tdTomato cassette (Ai9 was a gift from Hongkui Zeng; Addgene plasmid # 22799; http://n2t.net/addgene:22799; RRID:Addgene_22799) by Vector Builder (https://en.vectorbuilder.com/). Linearized Ai9-DIO-SPOTlight plasmid was then transferred to Duke Transgenic Mouse Core (Rodent Genetic Engineering Services). Integration into the Rosa26 locus was achieved through homologous recombination following electroporation into hybrid 129S6/C57BL/6N (G4) embryonic stem cells. Pan-neuronal expression of DIO-SPOTlight was achieved by crossing Rosa26-DIO-SPOTlight hemizygous transgenic founders (Founder lines ID: C1 and D8) with Nestin-Cre mice, obtained from Jackson Laboratories; JAX: strain #003771, RRID: IMSR_JAX:003771 (Tronche et al 1999). Two founder lines were maintained and used in experiments as follows: brain-wide regional analysis of cholinergic neurons n = 6 (5 founder D8 and 1 founder C1, age P30 – P33), spinal cord cholinergic neuron analysis n = 3 (2 founder D8 and 1 founder C1, age P30), and in vivo tunicamycin injection n = 10 (tunicamycin – 5 founder C1, vehicle – 4 founder C1 and 1 founder D8, age P13). All experimental procedures were approved by the institutional animal care and use committee at Duke University. Animals were maintained on a 12:12 hour light:dark cycle with ad libitum access to food and water.

Genotyping was performed using polymerase chain reaction of genomic DNA with the following primers: R26 wild type (WT) forward 5’-AAG GGA GCT GCA GTG GAG TA-3’; WT reverse 5’-CCG AAA ATC TGT GGG AAG TC-3’; DIO-SPOTlight forward 5’-AAC TGT GGC GTT AGA GAT CG-3’; DIO-SPOTlight reverse 5’-TTG AGC TCA GGA ACC TAT AAC T-3’. Cre genotyping was performed using the generic Cre protocol 22392 (The Jackson Laboratory, Bar Harbor, ME).

Tunicamycin (TM) was purchased from Tocris (Bristol, UK) and dissolved in dimethyl sulfoxide (DMSO) to a concentration of 5 mg/ml. On the day of the experiment, it was further diluted 1:25 (v/v) in 150 mM dextrose in PBS with 5% Tween-80 to maintain TM solubility and administered at a final concentration 3 µg/g bodyweight per mouse. Following methodology reported in Wang et al 2015, twelve-day old mice received two subcutaneous injections spaced two hours apart beginning at 3:00 PM. The mice were sacrificed 16 – 18 hours after the final injection, and the dissected brains were stored for further processing. Imaging and data analysis was performed blind to treatment groups.

Mice were anesthetized by using a drop jar method containing 3-5% aerosolized isoflurane. The thoracic cavity was then opened and transcardial perfusion with 1x PBS followed by fixation with 4% paraformaldehyde (PFA) was performed. The perfused brains were then removed and stored in 4% PFA + 20% sucrose for 24 hours at 4ᵒC. The brains were then transferred to 30% sucrose in PBS for an additional 24 hours before embedding in Tissue Tek optimal cutting temperature compound (OCT, Torrance, CA) and stored at -20ᵒC until sliced into 40 μm sagittal slices using a Leica CM 1950 cryostat. Brain slices were stored in PBS with 0.01% sodium azide at 4ᵒC.

After removing the brain, spinal cord dissection occurred using the following landmarks: cervical, base of skull to caudal T1 lamina; thoracic, 0.6 cm segment centered on T9 lamina; and lumbar, identified based on L1-L6 nerve roots. Spinal cord tissue was post-fixed using cyroPREP (Covaris, Woburn, MA) in 20% sucrose overnight, 30% sucrose for 5 days, then embedded in OCT, frozen on dry ice, and stored at –80°C. Tissue blocks acclimated at –20°C overnight prior to serial cryo-sectioning in the transverse plane to generate 20µm-thick cross sections using Microm HM 505N instrument.

Brain and spinal cord IHC differed only in that the brain slices were floating, and spinal cord sections were pre-mounted on slides. Any remaining free aldehyde groups were quenched with 0.1M glycine in 1x PBS for 30 min at room temperature (RT). Alkaline retrieval was then performed for 20 minutes at RT using freshly prepared 1% NaOH in 1x PBS while gently shaking. The sections were then washed three times for 10 minutes in PBS-T (1x PBS with 0.3% Triton X-100) and blocked for one hour in 10% normal donkey serum (NDS) plus 1% bovine serum albumin (BSA) in PBS-T, all while gently shaking on a VWR Standard Analog Shaker on low speed at RT. Slices were then incubated in primary antibodies on a shaker at 4ᵒC for 24 hours then washed three times in PBS-T at RT for 10 minutes each wash. Corresponding secondary antibodies were then added to the slices and incubated for 1 hours on a shaker at RT. Secondary antibodies were then removed and nuclei were then labeled for 10 minutes with Hoechst 33342 (Enzo Life Sciences) at a final concentration of 10 μg/mL in PBS-T. This was followed with two additional 10-minute washes in PBS-T at RT while on a shaker. PBS-T was exchanged for 1 x PBS and the slices were placed on glass slides and coverslips were sealed with Vectashield Vibrance Mounting Media (cat no. H-1700). All antibodies were prepared in dilution buffer containing 5% NDS plus 1% BSA in PBS-T. Commercially available antibodies used were anti-RFP (rabbit,1:1000, Rockland Antibodies and Assays, cat no. 600-401-379; RRID: AB_2209751), anti-GFP (chicken, 1:1000, Abcam, cat no. AB13970, RRID: AB_300798), and anti-ChAT (goat, 1:200, Millipore, cat no. AB144P; RRID: AB_2079751). Corresponding secondary antibody conjugates Alexa Fluor 488, Alexa Fluor 546, and Alexa Fluor 647 raised in donkey were obtained through Invitrogen (cat no. A78948, A10040, and A32849, respectively)

Fixed tissue sections were imaged using a Leica DMi8 Andor Dragonfly 202 unit equipped with Borealis illumination spinning disk confocal microscope and iXon 888 EMCCD camera. The full brain image was captured using a 10x objective lens plus a 1.5x camera zoom. All other regions of interest (ROI) were imaged using a 40x/1.3 oil emersion objective lens. For ratiometric SPOTlight analysis laser power and exposure time settings were held constant across acquisition channels for tdTomato and EGFP. Completed multi-framed images were stitched together using Fusion Stitcher (Oxford Instruments). Post-acquisition micrographs were analyzed using FIJI ImageJ analysis software (http://imagej.nih.gov/).

For measures of cholinergic and non-cholinergic neighboring neurons, 19 regions were imaged in P30 mice tissue. For acute tunicamycin experiments, methods are the same as described for the cholinergic and non-cholinergic analysis, with the exception that hippocampal pyramidal neurons from P13 mice were analyzed. Cholinergic neurons were used to create a threshold intensity cell body mask from a single z-plane using anti-ChAT as a cellular marker. ChAT negative neurons were selected using the SPOTlight reporter as a cell body mask. Given the two-fluorophore nature of the SPOTlight reporter and to limit biasing cell identification based on signal in one channel over the other, the image pixel values from the 561nm and 488nm channels were added together using FIJI’s image calculator option. Background subtraction for cell masking was performed by applying a 15-pixel gaussian blur on a duplicate image, subtracting the pixel values of blurred image from the original, and then applying a second 2-pixel gaussian blur on the resulting background subtracted image. Cell body ROIs were then selected using the wand selection tool and ROI manager window. In some cases, where cell bodies were clearly definable but high background prevented accurate thresholding, ROIs were drawn manually. Cytosolic signal analysis was achieved through nuclear exclusion in one of three ways. First, in many cases the nuclear compartment was clearly definable through the primary masking channel, selected, and added to the ROI manager. When this was not available, a nuclear mask was made from the Hoechst stain as described above for the primary masking channels. For both circumstances, the nuclear compartment was excluded from the analysis using the XOR function available in ROI manager’s “More” drop down window. The third method was used only when the primary cell body thresholding did not fully enclose the nuclear compartment within mask. The ROIs were then used to measure regions from the original image plane without need to exclude the nucleus since that region was not included in the ROI. Within slice background fluorescence was controlled for by measuring red and green fluorescence values in nearby cells where recombination did not occur (negligible red and green signal). These cells were identifiable in all regions, as Nestin-Cre recombinase efficiency was not 100%. Five such cells were selected for each region. The average red and average green signals from these SPOTlight-negative somata were subtracted from all cell measures.

To prepare acute brain slices, an adult animal (∼2 months old) was anesthetized and intracardially perfused with an ice-cold N-methyl-D-glucamine (NMDG)–based ACSF solution containing the following (in mM): 90 NMDG, 2.5 KCl, 1.2 NaH₂PO₄, 35 NaHCO₃, 20 HEPES, 25 glucose, 5 sodium ascorbate, 2 thiourea, 3 sodium pyruvate, 10 MgSO₄ heptahydrate, 0.5 CaCl₂ dihydrate, and saturated with 95% O₂ and 5% CO₂. The brain was then quickly removed and 250-μm sagittal sections were cut on a Leica VT1200S vibratome. Immediately after cutting, slices were placed in a 32°C recovery chamber containing NMDG ACSF for 10 to 15 min. After heated recovery, live tissue brain slices were transferred to and maintained in a holding chamber containing artificial cerebrospinal fluid (ACSF; 124 mM NaCl, 2.5 mM KCl, 1 mM MgCl₂, 2 mM CaCl₂, 26 mM NaHCO₃, 1.2 mM NaH₂PO₄, and 10 mM glucose, saturated with 95% O₂ and 5% CO₂, pH 7.4, 300 mOsm/l). Slices rested for approximately 1 hour prior to imaging.

For imaging, slices were placed in a perfusion chamber and superfused with oxygenated ACSF. Slices were imaged using a commercial two-photon microscope (Bruker Investigator Multiphoton Imaging System). Imaging of SPOTlight in live tissue slices was performed using 920nm light emitted from a Coherent Chameleon I Ti:Sapphire femtosecond laser with the laser power kept between 35-50 mW after the objective. Fluorescence signal was collected using a Nikon CFI75 LWD 16× W Objective and Quad GaAsP photomultiplier tubes (PMTs). Sample SPOTlight signal was split into red and green channels through a 565lpxr dichroic mirror and then passed through a 525/70 bandpass PMT filter. An image stack (512 x 512 resolution) of ∼200 µm total depth was collected using galvo-galvo scanning with frame rate of 1 Hz with 3-4 µm depth increments at either 1X or 4X optical zoom.