Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2024 Feb;27(2):373-383.
doi: 10.1038/s41593-023-01545-8. Epub 2024 Jan 11.

Long-term labeling and imaging of synaptically connected neuronal networks in vivo using double-deletion-mutant rabies viruses

Lei Jin  1   2 , Heather A Sullivan  1 , Mulangma Zhu  1 , Thomas K Lavin  1 , Makoto Matsuyama  1 , Xin Fu  1   3 , Nicholas E Lea  1 , Ran Xu  1 , YuanYuan Hou  1 , Luca Rutigliani  1 , Maxwell Pruner  1 , Kelsey R Babcock  1 , Jacque Pak Kan Ip  3   4   5 , Ming Hu  3   4   6 , Tanya L Daigle  7 , Hongkui Zeng  7 , Mriganka Sur  3   4 , Guoping Feng  1   3   8 , Ian R Wickersham  9
Affiliations

Long-term labeling and imaging of synaptically connected neuronal networks in vivo using double-deletion-mutant rabies viruses

Lei Jin et al. Nat Neurosci. 2024 Feb.

Erratum in

Abstract

Rabies-virus-based monosynaptic tracing is a widely used technique for mapping neural circuitry, but its cytotoxicity has confined it primarily to anatomical applications. Here we present a second-generation system for labeling direct inputs to targeted neuronal populations with minimal toxicity, using double-deletion-mutant rabies viruses. Viral spread requires expression of both deleted viral genes in trans in postsynaptic source cells. Suppressing this expression with doxycycline following an initial period of viral replication reduces toxicity to postsynaptic cells. Longitudinal two-photon imaging in vivo indicated that over 90% of both presynaptic and source cells survived for the full 12-week course of imaging. Ex vivo whole-cell recordings at 5 weeks postinfection showed that the second-generation system perturbs input and source cells much less than the first-generation system. Finally, two-photon calcium imaging of labeled networks of visual cortex neurons showed that their visual response properties appeared normal for 10 weeks, the longest we followed them.

PubMed Disclaimer

Conflict of interest statement

I.R.W. is a consultant for Monosynaptix, LLC, advising on design of neuroscientific experiments. The remaining authors declare no competing interests.

Figures

Fig. 1
Fig. 1. Second-generation monosynaptic tracing of inputs to corticostriatal neurons.
a,b, Corticostriatal neurons were retrogradely targeted by an AAV2-retro expressing either Cre or Flpo injected in dorsolateral striatum, with a Cre- or Flp-dependent helper virus combination injected in S1. EnvA-enveloped ∆GL RV expressing either Flpo or Cre was injected in S1 7 d later. The mice were crosses of the TRE-CB line described here with tdTomato reporter lines to report the activity of RVs. c, Results using the Flpo-expressing RV. At 2 weeks, few cells were found in input regions, whereas by 5 weeks, substantial numbers of labeled cells were found in ipsilateral secondary motor and somatosensory cortices, in contralateral S1, and ipsilateral thalamus, with or without doxycycline (note that doxycycline suppresses expression of mTagBFP2 as well as the viral genes G and L). Scale bars: 200 µm, apply to all images. d, Results using the Cre-expressing RV. Images in c and d are representative of four independent experiments that yielded similar results. e,f, Counts of labeled cells in contralateral S1 and ipsilateral thalamus for RV∆GL-Flpo (e) and RV∆GL-Cre (f). Counts are the total number of cells found in a series of every sixth 50-µm section spanning each brain so that the true number of labeled neurons in the entirety of each brain would be approximately six times the number shown here. Although the first-generation Flp-dependent system at 2 weeks had on average labeled more neurons in input regions than any other condition (f), almost no source cells were found in the first-generation conditions at 2 weeks (Extended Data Fig. 3), suggesting that almost all source cells had died by that point. g,h, Same data as in e and f but with the first-generation counts omitted and with the ranges of the y axes of the Flpo and Cre versions set to the same value for each input region. Box plots depict the mean (center), first and third quartiles (lower and upper box limits) and minima and maxima (bottom and top whiskers). One-way ANOVAs were used for all data analysis in this figure (e,f). **indicates 0.001 ≤ p < 0.01; *indicates 0.01 ≤ p < 0.05.
Fig. 2
Fig. 2. Second-generation monosynaptic tracing in Cre mice.
a, Experiment design for labeling inputs to dopaminergic cells in SNc. The Cre-dependent helper virus combination was injected into SNc of triple transgenic mice (DAT-IRES-Cre x TRE-CB x Ai65F); EnvA-enveloped ∆GL RV expressing Flpo was injected 7 d later. b, Labeled cells 5 weeks after RV injection, without doxycycline. i–v, injection site in SNc, with many cells coexpressing tyrosine hydroxylase (indicating the dopaminergic cells), EGFP (expressed by the first helper virus), mTagBFP2 (expressed by the second helper virus) and tdTomato (reporting activity of the Flpo-expressing RV). Scale bar in iii: 100 µm, applies to all images; vi, medium spiny neurons in striatum labeled by the ∆GL RV. c, Counts of labeled striatal cells (total found in a series of every sixth 50-µm section spanning each brain) in all conditions tested. d, Experimental design for labeling inputs to parvalbumin-expressing cells in primary somatosensory cortex (S1) in PV-Cre x TRE-CB x Ai65F mice. e, Example of labeled cells 5 weeks after RV injection, without doxycycline. i–v, injection site in S1, with many cells coexpressing parvalbumin, reporters of both helper viruses and tdTomato reporting activity of the RV. In addition to these source cells, many putatively presynaptic cells expressing tdTomato are present. Scale bar in iii: 100 µm, applies to all images; vi, relay neurons in ipsilateral thalamus labeled by the ∆GL RV. f, Counts of labeled thalamic cells (total found in a series of every sixth 50-µm section spanning each brain) in all conditions tested. The first-generation system labeled many more cells than did this implementation of the second-generation system, potentially due to the low efficiency of recombination by Flpo in these reporter mice. Box plots depict mean, first and third quartiles and minima and maxima. One-way ANOVAs were used for all data analysis in this figure. ***indicates p < 0.001. Images in b and e are representative of four independent experiments that yielded similar results.
Fig. 3
Fig. 3. Second-generation monosynaptic tracing in DAT-P2A-Flpo mice.
a, The design of the experiment was similar to that shown in Fig. 2a but using the Flp-dependent helper virus combination, and Cre-expressing RV, in DAT-P2A-Flpo x TRE-CB x Ai14 mice. b, Counts of labeled striatal neurons (total found in a series of every sixth 50-µm section spanning each brain) in all conditions tested (n = 4 independent experiments for each condition). Box plots depict mean, first and third quartiles and minima and maxima. Comparison of 5-week no dox and first-generation conditions: one-way ANOVA, P = 0.05095. c, Labeled cells 5 weeks after RV injection, without doxycycline. Many cells at the injection site in SNc were found to coexpress tyrosine hydroxylase and the fluorophores expressed by the helper viruses, as well as tdTomato, reporting activity of the Cre-expressing RV. Medium spiny neurons in striatum were labeled by the ∆GL RV. d, Example of a doxycycline (injection + food) mouse. The numbers of striatal input neurons were not significantly lower in this condition than without doxycycline, although comparisons were underpowered. e, Example label from the first-generation system. The numbers of striatal input neurons were not significantly higher than with the second-generation system, although comparisons were again underpowered. See also Extended Data Fig. 10 for whole-hemisphere image series for RV∆G-Cre and RV∆GL-Cre. f,g, Example of control conditions in which either G (f) or Flpo (g) was omitted. Scale bars: 100 µm, apply to all images. Images from c to g are representative of four independent experiments that yielded similar results.
Fig. 4
Fig. 4. Longitudinal structural two-photon imaging in vivo.
a, Experiment design. b,c, Two-photon images of tdTomato-labeled neurons in the V1 injection site at different time points. Arrows indicate example cells that are alive up to one time point (white arrows) but that are missing subsequently (green arrows). In this example, 46 of 215 cells were lost in the ‘no-dox’ mouse, whereas 12 of 305 cells were lost in the ‘dox’ mouse. Scale bar: 100 μm. d, Cell counts from all time points in a ‘no-dox’ mouse. Bar height for each color indicates the number of cells that were first visible at the corresponding time point that are still visible at the time point shown on the x axis. e, Cell counts from all time points in a ‘Dox’ mouse. f, Total number of imaged cells and number of cells present at the last imaging time point for each ‘no-dox’ mouse. g, Total number of imaged cells and number of cells present at the last imaging time point for each ‘dox’ mouse. h, Survival rates of cells that appeared at each time point for each mouse in both ‘dox’ and ‘no-dox’ groups. i, Aligned 12-week two-photon and postmortem confocal images from a ‘no-dox’ mouse. Scale bar: 20 µm. j, Aligned 12-week two-photon and confocal images from a ‘Dox’ mouse. Scale bar: 20 µm. k, Large-scale confocal image of labeled neurons in multiple visual cortical areas in the same ‘Dox’ mouse. Green rectangle indicates the FOV shown in j. Images from b, c, i–k are representative of three independent experiments that yielded similar results. l, Counts for each ‘no-dox’ mouse of the total number of imaged tdTomato+ cells (red), the number of cells that disappeared (black) and the number of cells found by postmortem confocal imaging to have expressed both tdTomato and mTagBFP2, considered surviving source cells (magenta). m, Counts for each ‘dox’ mouse of the total number of imaged tdTomato+ cells (red), the number of cells that disappeared (black) and the assumed number of total source cells (magenta). n, Estimated source cell survival rates for ‘dox’ and ‘no-dox’ groups.
Fig. 5
Fig. 5. Membrane properties of cells labeled by first- and second-generation monosynaptic tracing.
a,b, Experimental design. Beginning 35 d after RV injection, labeled neurons (input cells, defined as pyramidal neurons expressing tdTomato alone, or source cells, defined as neurons expressing both tdTomato and mTagBFP2) were targeted for ex vivo whole-cell recording. For AAV-only mice, so-called source cells denote neurons expressing mTagBFP2 and so-called input cells denote nearby unlabeled pyramidal neurons. c–h, Membrane properties of input cells (defined above). c, RMP; d, capacitance; e, rheobase and f, input resistance (Rm) of input cells. ΔGL, 26 cells from four mice; control, 13 cells from three mice; ΔG, 22 cells from three mice; RMP, one-way ANOVA, F(2, 58) = 0.35, P = 0.71. Capacitance, one-way ANOVA, F(2, 58) = 4.924, p = 0.011, Dunnett’s multiple comparisons test, ΔGL versus control, P = 0.93, ΔG versus control, *p = 0.023. Rheobase, one-way ANOVA, F(2, 58) = 1.81, p = 0.17. Rm, one-way ANOVA, F(2, 58) = 4.13, P = 0.021, Dunnett’s multiple comparisons test, ΔGL versus control, p > 0.99, ΔG versus control, p = 0.060. g, Representative traces of input cells’ firing in response to 250 pA current injection. h, Input cells’ firing frequency in response to sequential current step injection. Two-way ANOVA, F(2, 52) = 5.61, P = 0.0062. Dunnett’s multiple comparisons test, ΔGL versus AAV control, p = 0.94, ΔG versus AAV control, *p = 0.031. i–n, Membrane properties of source cells (defined above) in the ΔGL and control groups. Note that no surviving source cells could be found in the ∆G group at this survival time. i, RMP; j, capacitance; k, rheobase; l, Rm of source cells. ΔGL, 10 cells from four mice; control, 19 cells from three mice; RMP, unpaired t tests (two-tailed), *p = 0.016. Capacitance, unpaired t tests (two-tailed), p = 0.95. Rheobase, unpaired t tests (two-tailed), p = 0.47. Rm, unpaired t tests (two-tailed), *p = 0.040. m, Representative traces of source cell firing in response to 450 pA current injection. n, Source cells’ firing frequency in response to sequential current step injection. Two-way ANOVA, F(1, 27) = 1.42, p = 0.24. Error bars: mean ± s.e.m.
Fig. 6
Fig. 6. Functional two-photon imaging of transsynaptically labeled neurons’ visual response properties over 10 weeks.
a, The experimental design was similar to that shown in Fig. 4 but with a Flp-dependent jGCaMP7s AAV injected along with the RV. b, Following the virus injections, the injection site was imaged on a two-photon microscope while the awake mice were presented with drifting grating stimuli of different orientations and TFs, repeatedly for 10 weeks after RV injection. c, Representative two-photon field of view of neurons expressing jGCaMP7s (green channel) and tdTomato (red channel). Scale bar: 50 µm. Images from c are representative of two independent experiments that yielded similar results. d, Tuning curves of a jGCaMP7s-expressing neuron obtained with drifting gratings presented at 12 directions of motion and five TFs, repeated ten times (mean ΔF/F ± s.e.m.) at three different time points (left, week 3; middle, week 6 and right, week 10). e, Fraction of cells tuned at six different time points in two mice. The number of tuned neurons does not decline with time, suggesting intact response properties despite labeling by RV∆G. f, Tuning patterns at week 10 of 11 jGCaMP7s-expressing neurons showing clear preferred direction tuning, as well as the tuning pattern of an untuned neuron, representative of roughly half of imaged cells. g, Direction tuning patterns of six individual cells recorded at multiple time points (from week 1 to week 10). The top two cells became visible at week 1, the middle two appeared at week 2 and the bottom two cells appeared at week 3. Tuning patterns remain stable over the entire imaging period. h, Single-cell fluorescence time courses for 25 cells, showing activity over the first 120 s of visual stimulation. Cells are ranked in descending order of total activity. Scale bar: 10 s.
Extended Data Fig. 1
Extended Data Fig. 1. Intrinsic mCardinal fluorescence was not detected by confocal imaging, suggesting very low L expression in the conditions implemented.
Representative confocal images of injection sites in PV-Cre x Ai65F x TRE-CB mice, showing lack of clear mCardinal fluorescence induced by either of the laser lines (561 nm or 640 nm) flanking the excitation maximum on the Zeiss LSM 900 confocal microscope. Because of the lack of fluorescence in the channel otherwise used for AlexaFluor 647, specifically, we were able to use that channel for immunostaining for PV and TH (Figs. 2,3 and Extended Data Figs. 5,6). a, Injection sites of a FLEX AAV expressing tTA without a fluorophore, and b, injection sites of the FLEX AAV expressing TVA, EGFP, and tTA, with survival times of 3 and 4 weeks in each case. Left: EGFP channel; center-left: tdTomato channel; center-right: AlexaFluor 647 channel; right: ‘best case’ for imaging mCardinal: excitation with 561 nm and collecting all emitted light between 565 nm and 700 nm. Scale bars: 200 µm, apply to all panels. Images in this figure are representative of four independent experiments that yielded similar results.
Extended Data Fig. 2
Extended Data Fig. 2. Representative images of control experiments for corticostriatal experiments.
See Fig. 2 for explanation of experimental design and quantification. a,b, First-generation controls (7 d and 14 d survival times). c,d, Controls using second-generation viruses: no L, no G, and no Cre/Flpo controls (all without doxycycline), as well as the ‘dox (inj + food)’ condition for comparison. Scale bars: 200 µm, apply to all images. Images in this figure are representative of four independent experiments that yielded similar results.
Extended Data Fig. 3
Extended Data Fig. 3. Counts of corticothalamic ‘source’ and ‘starter’ cells and ratios of contralateral and thalamic cells to source and starter cells.
Counts are totals found in every sixth 50-µm section spanning each brain (n = 4 independent experiments for each condition). Box plots depict the mean (center), first and third quartiles (lower and upper box limits), and minima and maxima (bottom and top whiskers). One-way ANOVAs were used for all data analysis in this figure. a,b, Results using RV∆GL-Flpo(EnvA). a, Counts of source cells (left), defined as cells coexpressing tdTomato (marking activity of Flpo) and mTagBFP2 (marking expression of G), and ratios of these numbers of source cells to numbers of tdTomato+ cells in contralateral cortex (middle) and thalamus (right). Note the near-absence of source cells in the first-generation conditions at two weeks, suggesting that most had died by this time point. Note also that the low numbers of source cells in the second-generation conditions with doxycycline could reflect either cell death or simply the intended suppression of mTagBFP2 expression in these mice. For mice in which no source cells were found (either because they had died or because mTagBFP2 was suppressed by doxycycline), no ratio is included on the graphs. b, Counts and ratios of starting cells, defined here as cells coexpressing tdTomato (marking activity of Flpo) and EGFP (marking expression of TVA and tTA). c,d, Results using RV∆GL-Cre(EnvA). c, Counts and ratios of source cells, defined here as cells coexpressing tdTomato (marking activity of Cre) and mTagBFP2 (marking expression of G). d, Counts and ratios of starting cells, defined here as cells coexpressing tdTomato (marking activity of Cre) and EGFP (marking expression of TVA and tTA). ** indicates 0.001 ≤ p < 0.01; * indicates 0.01 ≤ p < 0.05.
Extended Data Fig. 4
Extended Data Fig. 4. Series of whole-brain images of labeled inputs to corticostriatal neurons using first- and second-generation rabies viral vectors expressing Cre.
Images are tiled confocal images of every sixth 50-µm coronal section spanning most of the rostrocaudal extents of the brains. a, Images from a mouse labeled using RV∆GL-Cre(EnvA) and associated AAVs (see Fig. 1b for experimental design), with a 5-week survival time and with doxycycline (administered in food only; see Methods). b, Images from a mouse labeled using RV∆G-Cre(EnvA) and associated AAVs, with a 1-week survival time. While the numbers of labeled neurons were not significantly different in either contralateral cortex or thalamus (see Fig. 1 and Supplementary Table 1), label is brighter in the ∆GL case, presumably because of greater tdTomato accumulation over the longer survival time. A similar difference can be seen when comparing the ∆G-Cre images in Extended Data Fig. 2b (top set of images) to the ∆GL-Cre ‘5 weeks with dox (food)’ images in Fig. 1d (bottom set of images).
Extended Data Fig. 5
Extended Data Fig. 5. Representative images of second-generation monosynaptic tracing of inputs to dopaminergic midbrain neurons using RV∆GL-Flpo.
See Fig. 1 for explanation of experimental design and for quantification. a, 3 weeks, no doxycycline, b, 3 weeks, with doxycycline in food, c, 5 weeks, with doxycycline in food, d, 3 weeks, with doxycycline in food and injected, e, 5 weeks, with doxycycline in food and injected. Control conditions are shown here: f, first-generation (∆G) system; g, no L; h, no G; i, no Cre, no doxycycline; j, no Cre, with doxycycline. Note that omitting L gave very similar results to omitting G. Images in this figure are representative of four independent experiments that yielded similar results.
Extended Data Fig. 6
Extended Data Fig. 6. Representative images of second-generation monosynaptic tracing of inputs to parvalbumin-expressing cortical interneurons.
See Fig. 1 for explanation of experimental design and for quantification. a, 3 weeks, no doxycycline; b, 3 weeks, with doxycycline in food (note that there is still blue fluorescence in this image, suggesting that, at one week after the mice were switched to dox food, the doxycycline has not completely suppressed G and presumably L expression); c, 5 weeks, with doxycycline in food; d, 3 weeks, with doxycycline in food and injected; e, 5 weeks, with doxycycline in food and injected. Control conditions are shown here: f) first-generation (∆G) system, g) no L, h) no G, i) no Cre, no doxycycline, j) no Cre, with doxycycline. Again, omitting L and omitting G gave very similar results. Images in this figure are representative of four independent experiments that yielded similar results.
Extended Data Fig. 7
Extended Data Fig. 7. Examples of extrinsic inputs to dopaminergic cells labeled with first- and second-generation RV vectors expressing Flpo and Cre.
a, Neurons labeled using ∆G and ∆GL rabies viruses expressing Flpo along with associated helper viruses (see Fig. 2a for experimental design). Top row: RV∆G-Flpo. Bottom row: RV∆GL-Flpo. b, Neurons labeled using ∆G and ∆GL rabies viruses expressing Cre along with associated helper viruses (see Fig. 3a for experimental design). Top row: RV∆G-Cre. Bottom row: RV∆GL-Cre. Scale bars: 100 µm, apply to all images.
Extended Data Fig. 8
Extended Data Fig. 8. Counts of dopaminergic and parvabuminergic source/starter cells and ratios of striatal and thalamic cells to source/starter cells in Cre mice.
Counts are totals found in every sixth 50-µm section spanning each brain (n = 4 independent experiments for each condition). Box plots depict the mean (center), first and third quartiles (lower and upper box limits), and minima and maxima (bottom and top whiskers). a,b, Results using RV∆GL-Flpo(EnvA) in DAT-IRES-Cre. a, Counts of source cells (left), defined as cells coexpressing tdTomato (marking activity of Flpo) and mTagBFP2 (marking expression of G), and ratios of these numbers of source cells to numbers of tdTomato+ cells in striatum (right). Note that the low numbers of source cells in the second-generation conditions with doxycycline could reflect either cell death or simply the intended suppression of mTagBFP2 expression in these mice. Note also that there is no graph for the ratio of labeled striatal cells to source cells for the ‘5 weeks dox (inj + food)’ because no source cells were found for that condition (that is, doxycycline appears to have fully suppressed mTagBFP2 expression). b, Counts and ratios of starter cells, defined here as cells coexpressing tdTomato (marking activity of Flpo) and EGFP (marking expression of TVA and tTA). c,d, Results using RV∆GL-Flpo(EnvA) in PV-Cre. c, Counts and ratios of source cells, defined here as cells coexpressing tdTomato (marking activity of Flpo) and mTagBFP2 (marking expression of G). Note that again there is no graph for the ratio of labeled contralateral cells to source cells for the ‘5 weeks dox (inj + food)’ because no source cells were found for that condition. d, Counts and ratios of starter cells, defined here as cells coexpressing tdTomato (marking activity of Flpo) and EGFP (marking expression of TVA and tTA).
Extended Data Fig. 9
Extended Data Fig. 9. Counts of dopaminergic and parvabuminergic source/starter cells and ratios of striatal cells to source/starter cells using RV∆GL-Cre(EnvA) in DAT-P2A-Flpo mice.
Counts are totals found in every sixth 50-µm section spanning each brain (n = 4 independent experiments for each condition). Box plots depict the mean (center), first and third quartiles (lower and upper box limits), and minima and maxima (bottom and top whiskers). a, Counts of source cells (left), defined as cells coexpressing tdTomato (marking activity of Cre) and mTagBFP2 (marking expression of G), and ratios of these numbers of source cells to numbers of tdTomato+ cells in striatum (right). Note that the low numbers of source cells in the second-generation condition with doxycycline could reflect either cell death or simply the intended suppression of mTagBFP2 expression in these mice. b, Counts and ratios of starter cells, defined here as cells coexpressing tdTomato (marking activity of Cre) and EGFP (marking expression of TVA and tTA).
Extended Data Fig. 10
Extended Data Fig. 10. Whole-hemisphere image series of labeled inputs to dopaminergic neurons using first- and second-generation rabies viral vectors expressing Cre.
Images are tiled confocal images of every sixth 50-µm parasagittal section spanning the injected hemispheres. a, Images from a mouse labeled using RV∆GL-Cre(EnvA) and associated AAVs (see Fig. 3a for experimental design), with a 5-week survival time and with doxycycline (administered in food; see Methods). b, Images from a mouse labeled using RV∆G-Cre(EnvA) and associated AAVs, with a 1-week survival time.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

References

    1. Wickersham IR, et al. Monosynaptic restriction of transsynaptic tracing from single, genetically targeted neurons. Neuron. 2007;53:639–647. - PMC - PubMed
    1. Zampieri N, Jessell TM, Murray AJ. Mapping sensory circuits by anterograde transsynaptic transfer of recombinant rabies virus. Neuron. 2014;81:766–778. - PMC - PubMed
    1. Wickersham IR, Finke S, Conzelmann KK, Callaway EM. Retrograde neuronal tracing with a deletion-mutant rabies virus. Nat. Methods. 2007;4:47–49. - PMC - PubMed
    1. Chatterjee S, et al. Nontoxic, double-deletion-mutant rabies viral vectors for retrograde targeting of projection neurons. Nat. Neurosci. 2018;21:638–646. - PMC - PubMed
    1. Reardon TR, et al. Rabies virus CVS-N2c(ΔG) strain enhances retrograde synaptic transfer and neuronal viability. Neuron. 2016;89:711–724. - PMC - PubMed