A single-vector intersectional AAV strategy for interrogating cellular diversity and brain function

Abstract

As discovery of cellular diversity in the brain accelerates, so does the need for tools that target cells based on multiple features. Here we developed Conditional Viral Expression by Ribozyme Guided Degradation (ConVERGD), an adeno-associated virus-based, single-construct, intersectional targeting strategy that combines a self-cleaving ribozyme with traditional FLEx switches to deliver molecular cargo to specific neuronal subtypes. ConVERGD offers benefits over existing intersectional expression platforms, such as expanded intersectional targeting with up to five recombinase-based features, accommodation of larger and more complex payloads and a vector that is easy to modify for rapid toolkit expansion. In the present report we employed ConVERGD to characterize an unexplored subpopulation of norepinephrine (NE)-producing neurons within the rodent locus coeruleus that co-express the endogenous opioid gene prodynorphin (Pdyn). These studies showcase ConVERGD as a versatile tool for targeting diverse cell types and reveal Pdyn-expressing NE⁺ locus coeruleus neurons as a small neuronal subpopulation capable of driving anxiogenic behavioral responses in rodents.

Advances in sequencing technologies have supported a rapid expansion in cellular taxonomy that highlights the importance of considering complex molecular diversity when defining neuronal cell types¹. To target cells based on multiple features, such as gene expression and projection target, several transgenic^2,3 and virally based^4–10 intersectional strategies have been developed, each with unique strengths and weaknesses. Transgenic approaches can better accommodate reporter transgenes that utilize large stop cassettes and may also be preferable for driving intersectional expression during development or in difficult-to-access cell populations. On the other hand, they typically lack regional and temporal specificity and are expensive and time-consuming to generate. Adeno-associated virus (AAV)-based reporter tools are an alternative approach that offer regional and temporal specificity and are quick and inexpensive to generate. As a result of these benefits, a variety of intersectional AAV tools has recently emerged^4–10. Although useful, these tools also present challenges that may discourage their widespread adoption, including complex design parameters, use of large stop cassettes that impede the limits of viral-packaging capacity, dependence on multiple vectors for payload expression or reliance on tetracycline-based expression systems that have reported issues with toxicity or leak expression^10–12.

To address these limitations, we developed a new, single-AAV-vector-based intersectional targeting approach called Conditional Viral Expression by Ribozyme Guided Degradation (ConVERGD), which relies on conditional, hammerhead ribozyme (HHR)-mediated messenger RNA inactivation¹³. Although self-cleaving ribozymes have previously been incorporated into AAV vectors to restrict transgene expression^14,15, ConVERGD is unique in that ribozyme expression is controlled by recombinase activity. The design of ConVERGD introduces several improvements over current intersectional strategies. Chiefly, its simple design avoids splitting transgenes into multiple components to achieve intersectionality. The small size of the ribozyme (92 bp) also enables inclusion of larger promoters and transgenes while remaining within the limits of AAV-packaging capacity, a feature that may be out of reach for vectors using bulkier polyadenylation stop cassettes (>600 bp) for transgene repression. This space-saving feature also provides efficient payload silencing within a single vector, eliminating the need to inject multiple viral vectors into targeted tissues.

To test the functionality of ConVERGD in vivo, we focused on the locus coeruleus (LC), a population of brainstem norepinephrine (NE; also called noradrenaline) neurons. Activation of LC neurons promotes a wide variety of arousal-related behavioral states, ranging from enhanced memory formation and attentiveness to stress-related responses and anxiety¹⁶. Yet, how these diverse behavioral states are promoted by LC neurons remains largely unexplored. Using single-cell RNA sequencing (scRNA-seq), we identified a subset of LC neurons that co-express prodynorphin (Pdyn), an endogenous opioid implicated in the promotion of aversive emotional states¹⁷. To determine how Pdyn⁺/Dbh⁺ LC neurons contribute to the array of behaviors broadly associated with LC activity, we developed an arsenal of ConVERGD-based viral tools to identify the anatomical connectivity and behavioral influence of these cells. In addition, these studies showcase ConVERGD as a new and easily amenable strategy for selectively targeting cell types defined by multiple features in vivo.

Results

Developing ConVERGD

Intersectional AAV-based tools often rely on complex design parameters, bulky silencing cassettes or co-injection of multiple viruses to achieve specific targeting^4–10, limiting their ability to be easily modified or accommodate new and more complex payloads. ConVERGD combines a small, loxP-flanked, self-cleaving type-III HHR (T3H48)¹⁴ with a Flp-dependent FLEx switch¹⁸ (FRT/FRT5) that inverts the transgene-coding region (Fig. 1a). In the presence of Flp, the transgene is directionally aligned with the promoter, but transcribed mRNA is immediately destabilized by ribozyme-mediated cleavage to prevent protein production (Fig. 1b). Only when the ribozyme is excised via Cre-mediated recombination does the mRNA remain stable and become translated. We tested this strategy by transfecting mouse Neuro2a (N2a) cells with a ConVERGD construct expressing enhanced green fluorescent protein (eGFP) (pAAV-hSyn-ConVERGD-eGFP-W3SL) alone or with recombinase expression plasmids containing Cre, Flp or both Cre and Flp (pDIRE). Under these conditions, eGFP expression occurred only in cells co-transfected with ConVERGD-eGFP and pDIRE (Fig. 1c–e), confirming that ConVERGD-based plasmids yield intersectional expression only in the presence of both Cre and Flp recombinases in vitro.

Fig. 1 |. ConVERGD-based expression requires the presence of both Cre and Flp recombinases.

a, Schematic of ConVERGD. In the absence of recombinases or in the presence of Cre recombinase (yellow arrow), the transgene is not transcribed owing to the inversion of the transgene-coding region. In the presence of Flp recombinase (blue arrow), the transgene is transcribed but the mRNA is destabilized by ribozyme-mediated cleavage and degraded (gray arrow). In the presence of Flp and Cre, transcribed mRNA is stabilized to facilitate subsequent protein synthesis. b, Schematic of ribozyme-mediated cleavage and degradation of transcribed mRNA. c, Representative images showing hSyn-ConVERGD-eGFP (green) expression in DAPI (blue)-counterstained N2a cell transfections. d, Dot plots of hSyn-ConVERGD-eGFP N2a cell transfections quantified by FACS analysis. e, Histogram compiling FACS dot plots in d. The histogram represents all the cells counted; the y axis has been capped at 500 cells to emphasize the difference between the curves. Images in c represent 20 independent transfection experiments. FACS data in d and e represent pooled data collected across six separate transfection experiments. Scale bars, 100 μm. F, FRT; F5, FRT5; L, loxP; Ri, ribozyme.

The initial version of ConVERGD utilized the human synapsin (hSyn) promoter and the optimized posttranscriptional regulatory element W3SL¹⁹ to conserve space for larger payloads. To test how ConVERGD performs when paired with other promoters (CAG, EF1a and nEF) and posttranscriptional regulatory elements (WPRE), we designed additional ConVERGD vectors and assessed their performance against a single recombinase vector (Flp-dependent pAAV-hSyn-FLEx(FRT)-eGFP-W3SL) and the original intersectional vector (Cre- and Flp-dependent pAAV-hSyn-INTRSECT-Con/Fon-EYFP-WPRE⁴ where EYFP is enhanced yellow fluorescent protein) (Extended Data Fig. 1a,b). Cells co-transfected with ConVERGD-eGFP and pDIRE showed similar eGFP expression to cells co-transfected with FLEx(FRT)-eGFP and INTRSECT-Con/Fon-EYFP, indicating that ConVERGD’s in vitro performance is on a par with field-standard single recombinase and intersectional constructs. Pairing ConVERGD with different posttranscriptional regulatory elements did not influence expression (Extended Data Fig. 1a,b). The use of stronger promoters increased eGFP expression in samples transfected with ConVERGD and pDIRE, but expression was also increased in the Flp-alone conditions, albeit to a much lesser extent (Extended Data Fig. 1a,b).

This result could suggest that promoter strength influences the ribozyme’s ability to sufficiently cleave all transcribed mRNA. However, nonspecific transgene expression can also arise from other sources, including aberrant plasmid recombination or the presence of partially recombined mRNA products. Although it is unknown to what extent these nonspecific recombination events contribute to unwanted protein synthesis, one should still be aware of their prevalence. Toward this end, we performed PCR on pAAV-ConVERGD-eGFP-W3SL and pAAV-hSyn-INTRSECT-EYFP plasmids using primers designed to detect various recombination events (Extended Data Fig. 2a). These experiments revealed the presence of spontaneous recombination for both plasmids, despite using growth conditions recommended for retainment of plasmid integrity (Methods and Extended Data Fig. 2b). Additional primer pairs were then designed to detect ConVERGD partial recombination isoforms in transfected N2a cells via quantitative (q)PCR (Extended Data Fig. 2c). Unexpectedly, we repeatedly observed ConVERGD-specific PCR product amplified in samples that had not undergone reverse transcription (RT), indicating persistence of ConVERGD plasmid even after successful elimination of genomic DNA with DNase I (Extended Data Fig. 2d). Thus, although recombined ConVERGD transcripts were observed in both Flp- and pDIRE-transfected cells, we cannot resolve their amount or whether the amplified product arose from transcribed mRNA or the originating plasmid. This ambiguity motivated us to focus on detection of leaky expression at the protein level for subsequent tool validation.

We reasoned that ConVERGD might benefit from the implementation of alternative design features, such as recombinase recognition sites with reduced homology that are reported to improve the specificity of recombinase-dependent AAV vectors²⁰. As ConVERGD vectors are leakiest in the presence of Flp, we hypothesized that aberrant recombination at the loxP sites or incomplete transcriptional repression by the ribozyme might be the cause (Fig. 1 and Extended Data Fig. 1). Surprisingly, replacement of the loxP sites with alternative versions (lox43/44) substantially reduced expression in the pDIRE condition, with only a modest effect on leaky ConVERGD expression in the presence of Flp (Extended Data Fig. 3a,b). The use of optimized FRT sites (FRT_min/FRT5) did not affect expression in transfected N2a cells (Extended Data Fig. 3a,b). As these alterations did not drastically improve leak, we focused subsequent toolkit expansion and in vivo experiments on ConVERGD constructs containing the hSyn promoter, loxP and FRT/FRT5 sites and W3SL posttranscriptional regulatory element.

Expanding ConVERGD Boolean logics and transgene expression

In addition to AND (CreANDFlp) Boolean logic, NOT logic (CreNOTFlp; FlpNOTCre) is useful in situations requiring access to intermingled cell populations. ConVERGD NOT logic constructs were achieved by placing the transgene-coding sequence and the ribozyme in opposite orientations and within nested FLEx switches (Fig. 2a,c). ConVERGD-ConFoff (CreNOTFlp) (Fig. 2a,b) and ConVERGD-CoffFon (FlpNOTCre) (Fig. 2c,d) constructs showed their highest expression levels in the presence of a single recombinase. Expression decreased in the presence of a second recombinase, although not to control levels (Fig. 2b,d and Extended Data Fig. 4a). Residual transgene expression in NOT logic vectors has also been observed with INTRSECT^4,5 and, thus, is not unique to ConVERGD. Although imperfect ribozyme-mediated cleavage could be contributing to this residual expression, we predict that the incomplete silencing stems largely from the different and imperfect recombination efficiencies of Cre and Flp²¹.

Fig. 2 |. ConVERGD can be expanded to additional Boolean expression logics.

a, ConVERGD-based ConFoff (CreNOTFlp) expression. Representative images show hSyn-ConVERGD-ConFoff-eGFP (green) expression in DAPI (blue)-counterstained N2a cell transfections. b, hSyn-ConVERGD-ConFoff-eGFP N2a cell transfections quantified by FACS analysis. Histogram represents all cells counted. The y axis has been capped at 500 cells to emphasize the difference between the curves. c, ConVERGD-based CoffFon (FlpNOTCre) expression. Representative images show hSyn-ConVERGD-CoffFon-eGFP (green) expression in DAPI (blue)-counterstained N2a cell transfections. d, hSyn-ConVERGD-CoffFon-eGFP N2a cell transfections quantified as in b. e, ConVERGD-based, triple-AND ConFonvCon (CreANDFlpANDvCRE) expression. Representative images show hSyn-ConVERGD-ConFonvCon-eGFP (green) expression in DAPI (blue)-counterstained N2a cells. f, The hSyn-ConVERGD-ConFonvCon-eGFP N2a cell transfections quantified as in b. Images in a, c and e represent four independent transfection experiments per construct. FACS results in b, d and f represent pooled data from four, five and four separate transfection experiments, respectively. Scale bars, 100 μm. L2272, Lox2272; LN, LoxN; vL, vloxP.

Intersectional strategies largely exist as dual-feature tools, a property that may be limiting as scientific fields move toward characterizing more complex cell types. The INTRSECT toolkit recently developed a triple intersectional construct, called Triplesect, which allows payload expression in the presence of three orthogonal recombinases⁵. However, like all INTRSECT-based vectors, expression specificity is achieved via a complex genomic engineering pipeline that is unique to each payload and challenging to implement. We reasoned that the small size and simplicity of ConVERGD could provide an amenable framework for further expansion of multi-feature intersectional tools. ConVERGD-ConFonvCon (CreANDFlpANDvCre), -ConFonvConNon (CreANDFlpANDvCreANDNigri) and -ConFonvConNonVon (CreANDFlpANDvCreANDNigriANDVika) constructs were created by placing three, four or five ribozymes, respectively, each flanked with unique recombinase sites, after the transgene. In vitro experiments confirmed the functionality of these vectors (Figs. 2e,f and 3a–c and Extended Data Fig. 4b,c). Even with expanded intersectionality, these ConVERGD constructs require less genetic space than Triplesect; ConVERGD-ConFonvCon requires 508 bp compared with Triplesect’s 901 bp (Supplementary Table 1), a design feature that allows these tools to be paired with larger transgenes and promoters.

Fig. 3 |. ConVERGD can be expanded to quadruple and quintuple intersectionality.

a, ConVERGD-based quadruple-AND ConFonvConNon (CreANDFlpANDvCreANDNigri) expression. b, FACS quantification of N2a cells co-transfected with pAAV-hSyn-ConVERGD-ConFonvConNon-eGFP alone (control; gray bar) or with all combinations of recombinase-expressing plasmids (colored bars). Data points represent the fold-change of the MFI of individual transfections compared with the average control MFI. Bars represent the mean of all experiments. The number of transfections performed for each condition is noted above its corresponding bar in the graph. Error bars show the s.e.m. c, ConVERGD-based quintuple-AND ConFonvConNonVon (CreANDFlpANDvCreANDNigriANDVika) expression. All images show transfected N2a cells counterstained with DAPI (blue) and represent results observed across four (ConVERGD-ConFonVonNon-eGFP) and two (ConVERGD-ConFonvConVonNon-eGFP) transfections. Scale bars, 100 μm. Scale bars apply to all images. N, Nox; V, Vox.

Finally, the CreANDFlp ConVERGD toolkit was expanded to include several constructs relevant to in vivo study of neural circuits, such as chemogenetic activators and inhibitors (hM3Dq and hM4Di), optogenetic activators and inhibitors (ChR2, ArchT and stGtACR), components for trans-synaptic rabies tracing (TVA receptor and N2c glycoprotein), calcium-imaging tools (GCaMP8m) and a dual-labeling construct to visualize axonal projections and presynaptic sites (synaptophysin-GreenLantern-T2A-GAP43-mScarlet) (Extended Data Fig. 5a,b). Although these constructs represent only a small fraction of the tools that are commonly used in neuroscience research, they highlight the amenability of ConVERGD to promote intersectional expression of diverse and complex transgenes with minimal design considerations.

In vivo validation of ConVERGD viral vectors

ConVERGD vectors were next tested in vivo through targeted viral injections into the LC of transgenic mice where the NE-synthesizing gene dopamine-β-hydroxylase (Dbh) drives expression of either Flp or Cre recombinase. Dbh^Flp mice received injections of AAV-hSyn-ConVERGD-eGFP or a mixture of ConVERGD-eGFP and AAV-EF1a-mCherry-IRES-Cre (Fig. 4a). ConVERGD-injected Dbh^Flp mice showed eGFP expression only in LC neurons that also received Cre-expressing AAV (Fig. 4a). Similar results were achieved when ConVERGD-eGFP and AAV-EF1a-mCherry-IRES-Flp were co-injected into the LC of Dbh^Cre mice (Fig. 4b). These results show that ConVERGD vectors support intersectional payload expression in vivo, selectively in cells where Cre and Flp recombinase are present.

Fig. 4 |. ConVERGD shows specific expression in vivo in the presence of Cre and Flp recombinases.

a, Representative images of the LC (TH; cyan) of a Dbh^Flp mouse injected with AAV-hSyn-ConVERGD-eGFP (green) into one LC nucleus (top) and AAV-hSyn-ConVERGD-eGFP + AAV-Ef1a-Cre-mCherry (Cre(mChr); magenta) into the other LC nucleus (bottom). b, Representative images of the LC (TH; cyan) of a Dbh^Cre mouse injected with AAV-hSyn-ConVERGD-eGFP (green) into one LC nucleus (top) and ConVERGD-eGFP + AAV-Ef1a-FlpO-mCherry (Flp(mChr); magenta) into the other LC nucleus (bottom). Images in a and b represent independent experiments in five (Dbh^Flp) and three (Dbh^Cre) animals, respectively. Scale bars, 100 μm. mChr, mCherry.

To ensure that ConVERGD vectors yield specific expression beyond the LC, we performed additional validation experiments in the hippocampus of Calb1^Cre;Slc17a7^Flp double transgenic mice, as well as single transgenic and wild-type (WT) animals. Injection of single-recombinase-dependent viral tools into the hippocampus of Calb1^Cre or Slc17a7^Flp mice produced robust labeling of cells within the dentate gyrus region of the hippocampus (Extended Data Fig. 6a). Meanwhile, expression of AAV-hSyn-ConVERGD-eGFP into the hippocampus of these animals was largely undetectable. Instead, ConVERGD-eGFP expression was visible when injected in Calb1^Cre;Slc17a7^Flp double transgenic mice, similar to results obtained using AAV-hSyn-INTRSECT-EYFP (Extended Data Fig. 6b, upper two rows). Immunostaining revealed additional leak for both constructs; for ConVERGD, leak expression appeared in Flp-expressing samples, whereas INTRSECT leak was more prevalent in Cre-expressing samples (Extended Data Fig. 6b, lower two rows). These experiments suggest that ConVERGD can be applied to additional brain regions and cell types, with similar or improved efficacy when compared with other intersectional viral tools.

We also used this experimental setup to test the functionality of additional ConVERGD vectors in vivo. The triple ConVERGD tool, AAV-hSyn-ConVERGD-ConFonvCon-eGFP, produced expression in Calb1^Cre;Slc17a7^Flp mice only when co-injected with an AAV-expressing vCre (Extended Data Fig. 6c). In addition to demonstrating the utility of the triple intersectional ConVERGD reporter in vivo, these experiments suggest that flanking sequential ribozymes with discrete recombinase sequences is a feasible design strategy for future expansion of the ConVERGD toolkit. However, when implementing ConVERGD vectors with expanded intersectionality, one should consider that the presence of repeated ribozyme sequences could affect viral packaging and performance.

ConVERGD NOT logic tools were also tested in vivo via injection of AAV-hSyn-ConVERGD-ConFoff-eGFP into the hippocampus of Calb1^Cre or Calb1^Cre;Slc17a7^Flp mice and injection of AAV-hSyn-ConVERGD-CoffFon-eGFP into Slc17a7^Flp or Calb1^Cre;Slc17a7^Flp mice (Extended Data Fig. 6d,e). Qualitatively, both viral tools reflected in vitro results: the presence of a second recombinase drastically reduces transgene expression but does not fully eliminate it. This issue is not unique to ConVERGD^4,5, emphasizing that optimization of recombinase efficiency would benefit a wide range of molecular toolkits.

ScRNA-seq of LC neurons

To assess ConVERGD’s utility for characterizing discrete cell types in vivo, we turned to the LC, a neuromodulatory cell population with understudied molecular diversity. LC neurons are traditionally defined by their production of NE, but are also reported to express other neuromodulators with important roles in the brain^22–25. Thus, LC neurons could alter their functional impact through the co-release of additional signaling molecules, although the extent to which this molecular heterogeneity exists, and how it relates to the organization and function of LC-related neuronal circuits, is not comprehensively understood. To address this, we first performed single-cell transcriptome profiling (scRNA-seq) of adult mouse LC neurons. Fluorescently labeled LC neurons were isolated directly from brain sections obtained from adult male and female Dbh^Cre; Ai14 transgenic mice (Fig. 5a). Smart-seq2 was used to generate scRNA-seq libraries and sequencing was performed to a depth of >1 million uniquely mapped reads per cell (median 16.7 million, mapping rate 94.6%), resulting in a median of 9,000 genes detected per cell. Contaminated samples (14 cells), samples with <1 million uniquely mapped reads (22 cells) and samples without Dbh expression (34 cells) were excluded from further analyses (Fig. 5b). This resulted in 201 LC neurons for which normalized gene expression was reviewed to identify enriched genes, both in a nonbiased manner (Extended Data Fig. 7) and using specified gene ontology (GO) terms to identify transcripts associated with neuronal signaling (Fig. 5c). In addition to Dbh, samples had robust expression of tyrosine hydroxylase (TH) (Th; an enzyme in the NE-synthesizing pathway) and solute carrier family 6 member 2 (Slc6a2; an NE transporter), supporting their classification as NE⁺ neurons. Strong expression of galanin was also observed in most LC cells (Fig. 5c), consistent with other reports^22,24. ScRNA-seq revealed co-expression of neurotransmitter-synthesizing and neuropeptide genes within Dbh-expressing samples, such as calcitonin-related polypeptide α (Calca) which encodes the calcitonin gene-related neuropeptide (CGRP) and γ-aminobutyric acid receptor subunit θ (Gabrq), a mediator of inhibitory synaptic transmission. Smaller subsets of LC neurons expressed glutamate decarboxylase 2 (Gad2), an enzyme involved in the synthesis of the inhibitory neurotransmitter γ-aminobutyric acid (GABA), cocaine and amphetamine-regulated transcript protein (Cartpt), a preprotein known to modulate feeding, reward and stress response, and prodynorphin (Pdyn) and proenkephalin (Penk), genes encoding precursor peptides for the endogenous opioids dynorphin and enkephalin (Fig. 5c). Fluorescence in situ hybridization confirmed that populations of Dbh-expressing LC neurons contained Calca, Cartpt, Gabrq, Gad2, Pdyn and Penk (Fig. 5d). Collectively, these data show that LC neurons heterogeneously express transcripts encoding a variety of neurotransmitters and neuropeptides in addition to NE, potentially diversifying their functional impact in downstream brain areas.

Fig. 5 |. Subsets of NE LC neurons co-express genes for other neurotransmitters and neuropeptides.

a, Workflow for LC molecular profiling. Fluorescently labeled LC neurons were manually isolated from adult mouse brain tissue for sequencing (n = 14 mice). Full-length cDNA and sequencing libraries were generated for each cell using a Smart-seq2-based protocol. b, Of 271 cells, 201 identified as NE⁺ LC neurons owing to the presence of mRNA for the NE-synthesizing gene dopamine-β-hydroxylase (Dbh). c, Heatmap of scaled transcript abundance (transcripts per million (t.p.m.)) for LC neurons. The top expressing genes from GO term categories for neurotransmitters (red), transporters (dark blue), neuropeptides (orange), synaptic transmission (green), neurotransmitter receptors (light blue), opioid signaling (purple) or selected control genes (gray) are displayed. d, Fluorescence in situ hybridization of selected neuronal signaling molecules (magenta) within Dbh-expressing LC neurons (white). Transcripts for calcitonin-related polypeptide α (Calca), cocaine- and amphetamine-regulated transcript protein (Cartpt), GABA type A receptor subunit θ (Gabrq), glutamate decarboxylase 2 (Gad2), prodynorphin (Pdyn) and proenkephalin (Penk) were expressed within LC cells. Images represent independent experiments from three (Calca, Cartpt and Gabrq), four (Gad2 and Penk) and seven (Pdyn) animals. Scale bars, 100 μm for larger images and 20 μm for insets.

Validation of ConVERGD to access Pdyn-expressing LC neurons

Activation of the LC promotes diverse behavioral responses¹⁶, but how this nucleus, with its broad connectivity, maintains such a rich functional repertoire remains an open question in the field. One option is that subsets of LC neurons form distinct neural circuits with specialized functions, an idea supported by experiments where selective activation of certain LC projections promoted anxiogenic or anxiolytic-like behaviors^26–29. Our sequencing data revealed a subpopulation of LC cells that co-express both Dbh and Pdyn, a neuropeptide generally associated with aversive behavioral responses related to stress^30–32. As activation of the LC also modulates stress-related responses^16,33, we were motivated to perform additional characterization of these cells using ConVERGD tools. We first confirmed that transgenic animals were sufficient for targeting Pdyn⁺ LC neurons by crossing Pdyn^Cre mice with a Cre-dependent nuclear eGFP reporter line, Sun1-eGFP. eGFP⁺ cells were observed adjacent to and within the LC in a pattern consistent with Pdyn expression reported in the Allen Institute for Brain Science mouse brain in situ datasets³⁴. A subset of eGFP⁺ cells within the LC also expressed TH, confirming their identity as NE⁺ neurons (Fig. 6a).

Fig. 6 |. ConVERGD shows specific expression in Pdyn^Cre;Dbh^Flp mice.

a, Representative images of eGFP (green)-labeled Pdyn⁺ cells in and around the LC (TH; white) in the transgenic line Pdyn^Cre;Sun1-eGFP. b, Representative images of the LC (TH; white) with AAV-hSyn-ConVERGD-eGFP injected in WT, Pdyn^Cre, Dbh^Flp and Pdyn^Cre;Dbh^Flp mice. eGFP expression (green) was observed only in the LC of Pdyn^Cre;Dbh^Flp mice. c, Quantification of ConVERGD-eGFP-labeled cells in and around (~200-μm radius) the LC across different genotypes. Points represent the cell counts across six 50-μm LC brain sections. d, Representative images of multiplexed labeling of eGFP (green; immunostain), Dbh (magenta; in situ hybridization) and Pdyn (cyan; in situ hybridization) within a ConVERGD-eGFP-injected Pdyn^Cre;Dbh^Flp brain section. e, Representative images of AAV-hSyn-ConVERGD-mCherry (magenta) injected into the LC (TH; white) of Dbh^Flp;Pdyn^Cre;Sun1-eGFP (GFP; green) mice. f, Quantification of ConVERGD-mCherry-labeled cells in the LC and their co-localization with Cre-expressing cells. Points represent the total cell counts across five 50-μm LC brain sections. g, Quantification of the fraction of ConVERGD-mCherry labeled cells in the LC that co-localized with Cre-expressing cells or TH-expressing cells. Points represent the average cell counts across five 50-μm LC brain sections. Bars represent the mean of the data for c, f and g. Error bars show the s.e.m. for c, f and g. Images in a represent independent experiments in three Pdyn^Cre;Sun1-eGFP mice. Images in b represent independent experiments in two (WT), three (Pdyn^Cre), three (Dbh^Flp) and four (Pdyn^Cre;Dbh^Flp) mice. Images in d represent independent experiments in two Dbh^Flp;Pdyn^Cre mice. Images in e represent independent experiments in four Dbh^Flp;Pdyn^Cre;Sun1-eGFP mice. Scale bars, 100 μm for the larger images in a, b and e, 20 μm for the inset in a and 50 μm in d. CV, ConVERGD.

We next assessed whether ConVERGD could selectively target Pdyn⁺ LC neurons by injecting AAV-hSyn-ConVERGD-eGFP into the LC of Pdyn^Cre;Dbh^Flp mice or control animals (WT, Pdyn^Cre and Dbh^Flp mice). Although GFP expression was not observed in control mice, a subset of labeled LC cells was present in Pdyn^Cre;Dbh^Flp mice (Fig. 6b,c). Multiplexed labeling confirmed the co-expression of Pdyn and Dbh mRNA in eGFP-containing cells (Fig. 6d). To quantify the in vivo specificity of ConVERGD, AAV-hSyn-ConVERGD-mCherry was injected into the LC of Pdyn^Cre;Dbh^Flp;Sun1-eGFP triple transgenic mice (Fig. 6e). Almost all mCherry-expressing neurons within the LC co-expressed TH (98%) and most of these cells were also labeled with GFP (76%) (Fig. 6f,g). The imperfect co-localization between mCherry and GFP signal could be the result of aberrant expression of ConVERGD or reduced efficiency of the Sun1-eGFP reporter, although it is not straightforward to distinguish between these possibilities owing to a lack of antibodies that robustly label Pdyn and Cre proteins.

ConVERGD’s performance in vivo was also compared directly with INTRSECT. AAV-hSyn-INTRSECT-Con/Fon-EYFP was injected into the LC of Pdyn^Cre, Dbh^Flp and Pdyn^Cre;Dbh^Flp mice at the same titer used for ConVERGD injections (~7.87 × 10¹¹ genome copies per ml (g.c.p. ml⁻¹)). As expected, we observed a considerable number of INTRSECT labeled neurons in Pdyn^Cre;Dbh^Flp samples, but labeling was also seen in Pdyn^Cre and Dbh^Flp samples, suggesting that a nonspecific payload expression had occurred (Extended Data Fig. 8a,b). Collectively, these experiments establish ConVERGD as a viable intersectional approach for targeting discrete cell populations in vivo with improved selectivity over current field-standard approaches.

Rabies-based connectivity mapping of Pdyn⁺/Dbh⁺ LC neurons

ConVERGD tools were used next to investigate the connectivity of Pdyn⁺/Dbh⁺ LC neurons. Although several studies have highlighted the modular afferent and efferent patterns of LC neurons^33,35,36, the input–output organization of molecularly diverse subpopulations of LC neurons remains understudied, partially as a result of the lack of sufficient tools to access these subsets. Using ConVERGD-based helper AAVs expressing the TVA receptor and the N2c glycoprotein, we performed intersectional, monosynaptic input tracing from Pdyn⁺/Dbh⁺ LC neurons (Fig. 7a). A subset of TH⁺ cells had overlapping ConVERGD-TVA(mChr) and RABV-CVS-N2c^ΔG-H2B-eGFP labeling within the LC, indicative of ‘starter cells’ that support the trans-synaptic spread of the rabies virus to their presynaptic partners (Fig. 7b). Whole-brain quantification revealed inputs to Pdyn⁺/Dbh⁺ LC neurons in areas ranging from pontine nuclei to frontal cortical regions (Fig. 7c). From posterior to anterior, we found that the dorsal and lateral paragiganto-cellular reticular nucleus (PGRNd and PGRNl, respectively), midbrain reticular nucleus (MRN), periaqueductal gray (PAG), hypothalamus, central amygdala (CeA), bed nucleus of the stria terminalis (BNST), orbitofrontal cortex (OFC) and prelimbic cortical area (PLC) had the highest concentration of rabies-labeled presynaptic inputs (Fig. 7d). These data suggest that Pdyn⁺/Dbh⁺ LC cells receive inputs from regions spanning the anterior–posterior axis of the brain, with the highest proportion of these afferents arising largely from stress-related brain areas such as the BNST, hypothalamus and CeA.

Fig. 7 |. ConVERGD-based monosynaptic rabies tracing reveals inputs on to Pdyn⁺/Dbh⁺ LC cells.

a, ConVERGD-based rabies helper AAVs expressing TVA receptor (magenta) and N2c glycoprotein (brown) unilaterally co-injected into the LC of Pdyn^Cre;Dbh^Flp mice 3 weeks before injection of RABV-CVS-N2c^ΔG-H2B-eGFP. b, Representative images of TVA(mChr) expression (magenta) and RABV-GFP expression (green) within the LC (TH; cyan). The inset image highlights a TH⁺ LC starter cell (yellow arrowhead) containing eGFP (from the rabies virus) and mCherry (from ConVERGD-TVA(mChr)). c, Quantitative distribution of eGFP⁺ cells across the brain. Highlighted regions represent areas containing peaks of at least 3% of the total eGFP⁺ cells counted. Colors correspond to the regions specified in d. The gray crosshatched region was not quantified owing to nonspecific rabies expression known to occur locally at viral injection sites. Light gray traces represent the distribution of eGFP⁺ cells from individual brains. The bold line represents an average distribution of all brains. Gray shading represents the s.d. from the mean. d, Representative images of eGFP⁺ cells in major input regions. Cell nuclei are counterstained with DAPI (blue). e, Representative images of axon projections from TVA(mChr)⁺ (magenta) LC cells. Nuclei are counterstained with DAPI (blue). MPO, medial preoptic area of the hypothalamus. Images in b, d and e represent independent experiments in five Dbh^Flp;Pdyn^Cre animals. Scale bars, 100 μm in b, d and e and 25 μm in inset in b.

The CVS-N2c^ΔG-modified rabies virus was chosen for these studies because it has been reported to have improved tracing efficiency and reduced cellular toxicity over the B19 strain³⁷. Indeed, we observed that the cellular integrity of the starter cells was maintained, allowing tracing of their afferent axonal projections via their expression of TVA-mCherry. Regions differed in their level of innervation. For instance, CeA received sparse innervation from Pdyn⁺/Dbh⁺ LC cells, whereas the lateral hypothalamus (LHA), hippocampus (HC) and olfactory bulb (OB) were more densely innervated (Fig. 7e). Together, these data highlight nonuniform projection patterns of a molecularly distinct LC subpopulation, whereby Pdyn⁺/Dbh⁺ LC neurons preferentially innervate certain brain structures but still broadly collateralize their axons to target many other brain regions.

Pdyn⁺/Dbh⁺ LC neurons promote selective anxiogenic behavior

Dynorphin signaling and LC activation are separately implicated in promoting anxiogenic behaviors^{17,24,38–40} and our connectivity data revealed enriched connections between Pdyn⁺/Dbh⁺ LC neurons and brain areas centrally involved in the stress response. Based on this, we wondered how the selective activation of just Pdyn⁺/Dbh⁺ LC cells, which represent a small portion (<20%) of the LC, might influence behavior. To test this, we developed a ConVERGD-based excitatory DREADD construct to enable transient activation of Pdyn⁺/Dbh⁺ LC neurons during behavioral testing. Bilateral injection of AAV-hSyn-ConVERGD-(HA)hM3Dq into the LC of WT, Pdyn^Cre, Dbh^Flp and Pdyn^Cre;Dbh^Flp mice induced excitatory DREADD receptor expression in a subset of LC neurons in Pdyn^Cre;Dbh^Flp mice (Fig. 8a). The functionality of hM3Dq expression was validated by giving clozapine-N-oxide (CNO) intraperitoneally (i.p.) before perfusion and immunolabeling sections against the immediate early gene cFos to assess neuronal activation. Pdyn^Cre;Dbh^Flp mice expressing ConVERGD-hM3D showed enhanced cFos expression in TH⁺ LC cells compared with Pdyn^Cre and Dbh^Flp mice (Fig. 8b). To test the behavioral effect of activating Pdyn⁺/Dbh⁺ LC neurons, we assessed performance during open field testing (OFT), an assay used to quantify locomotion, exploration and anxiety-like behavior in rodents. All mice received bilateral injections of AAV-hSyn-ConVERGD-(HA)hM3Dq into the LC at least 3 weeks before behavioral testing and received CNO 30 min before placement in the open field, where distance and time spent in the corner and center regions were measured (Fig. 8c,d). Control mice included WT, Pdyn^Cre and Dbh^Flp animals after statistical analysis (analysis of variance (ANOVA); Methods) revealed no behavioral differences between these genotypes. Mice in the control and experimental groups traveled similar total distances within the arena with no difference in the time that each group spent in the corners or center portion of the open field (Fig. 8e–g), indicating that selective activation of Pdyn⁺/Dbh⁺ LC neurons was not sufficient to drive behavioral alterations during OFT. We also activated Pdyn⁺/Dbh⁺ LC neurons while mice were placed in an elevated zero maze (EZM), an assay used to test anxiety in rodents. As in OFT, mice were given CNO 30 min before being placed in the EMZ, where distance and time spent in closed and open regions were measured. Control and experimental groups traversed all areas of the EZM during their sessions (Fig. 8h) with no difference in the total distance traveled or distance traveled in the open regions (Fig. 8i,j). However, compared with control mice, hM3Dq-expressing Pdyn^Cre;Dbh^Flp mice spent significantly less time in the open regions of the EZM compared with controls (Fig. 8k). These behavioral tests suggest that activation of Pdyn⁺/Dbh⁺ LC cells, which represent a small percentage of cells within the LC, is sufficient to drive anxiety-like behaviors in EZM but not OFT. These findings differ from previous studies where activation of all LC neurons promoted anxiety-like behavior in both OFT and EZM^24,39–41 and could suggest that activation of Pdyn⁺/Dbh⁺ LC neurons has a greater behavioral influence in acutely stressful situations, such as in the EZM.

Fig. 8 |. ConVERGD-based chemogenetic activation of Pdyn⁺/Dbh⁺ LC neurons promotes an anxiogenic phenotype in EZM but not OFT.

a, Representative images of hM3Dq expression (HA; magenta) observed in a subset of LC cells (TH; white) in Pdyn^Cre;Dbh^Flp mice but not in control (WT, Pdyn^Cre, Dbh^Flp) mice. b, Representative images of cFos (cyan) expression after CNO injection in Pdyn^Cre,Dbh^Flp and Pdyn^Cre;Dbh^Flp mice injected with AAV-hSyn-ConVERGD-(HA) hM3Dq. c, Behavioral setup and timeline. AAV-hSyn-ConVERGD-(HA)hM3Dq was bilaterally injected into the LC of WT, Pdyn^Cre, Dbh^Flp and Pdyn^Cre;Dbh^Flp mice at least 3 weeks before behavioral tests. d, Diagram of the open field chamber with quantified ROI center (white) and corner (black) divisions. Perimeter (gray) regions were not quantified. e, Total distance traveled during OFT. f, Percentage of time spent in the center region of the open field chamber. g, Percentage of time spent in the corners of the open field. h, Diagram of EZM (top) and combined traces of control and hM3Dq-expressing mice (bottom). i, Total distance (sum of closed (shaded) and open (white)) traveled in the EZM. j, Percentage of distance traveled in the open region of the EZM. k, Percentage of time spent in the open region of the EZM. The hM3Dq-expressing mice spent significantly less time in the open regions compared with control mice. *P = 0.0314, two-tailed Student’s t-test; n = 15 control animals, 11 hM3Dq animals. All bars represent the mean of the data. Error bars show the s.e.m. Images in a represent independent experiments in 5 (WT, Pdyn^Cre, Dbh^Flp) and 11 (Pdyn^Cre;Dbh^Flp) animals. Images in b represent independent experiments in three (Pdyn^Cre), four (Dbh^Flp) and three (Pdyn^Cre;Dbh^Flp) animals. Scale bars, 100 μm. Data point shape legend in e applies to data points in e–g and j,k and represents the genotypes of the mice.

Discussion

Unique features of ConVERGD and future directions

Current intersectional strategies are at risk of being outpaced by the need to target increasingly specific cell populations with increasingly complex payloads. To address this, ConVERGD facilitates straightforward insertion of intact transgenes with minimal design constraints, with an overall goal to expedite the development of diverse intersectional viral tools for a broader research community (Extended Data Fig. 5). Still, there are potential areas for improvement. For instance, pairing ConVERGD with stronger promoters increased intersectional transgene expression but also caused a modest increase in nonspecific gene expression (Extended Data Fig. 1). We hypothesize that this may be the result of an inability of ribozyme-mediated cleavage to keep pace with mRNA production if transcription is too robust. Although this issue is circumvented by promoter choice, future iterations of ConVERGD may benefit from efforts currently taking place in the field of synthetic biology to optimize ribozymes and other RNA switches^14,42–44. Low levels of nonspecific expression were also observed in ConVERGD NOT logic and with certain combinations of recombinases with the ConVERGD quintuple intersectional vectors (Figs. 2 and 3). We speculate that this may arise from suboptimal recombinase performance, highlighting that ConVERGD, and indeed all recombinase-dependent vectors, would benefit from further optimization of recombinases with improved efficiency and specificity.

To our knowledge, ConVERGD is unique among neuroscience tools for its pairing of ribozymes with recombinase-based repression to promote conditional gene expression, positioning it as a new platform for future conditional tool design. For instance, ConVERGD vectors could utilize ribozymes to provide conditional expression in combination with morpholino binding¹⁴ or drugs such as neomycin⁴⁵ and tetracycline^46,47. Thus, it should be feasible to expand the ConVERGD toolkit to include temporal intersectionality. Temporal control could also be achieved by combining ConVERGD vectors with activity-dependent genetic labeling methods⁴⁸, allowing end-users to target distinct cellular populations based on their molecular identity and in vivo activity patterns. Ultimately, we reason that ConVERGD provides a starting point for building diverse intersectional vectors that encompass molecular and temporal dimensions, beyond what is possible with current strategies.

ScRNA-seq reveals molecular diversity in LC

Although advances in single-cell sequencing technologies are rapidly improving our understanding of cellular diversity, these methods have not been uniformly applied to all brain regions. A prime example of this is the LC, a brain region traditionally considered to be molecularly homogeneous because almost all neurons within it produce NE. The fact that LC neurons collectively connect with most areas of the brain to release NE has further perpetuated a view that these cells uniformly and broadly modulate brain state. Yet, activation of LC neurons can promote diverse and even opposing behaviors related to arousal^26–28, suggesting underlying heterogeneity within this circuit. Although not explored in the context of the LC, other neuromodulatory cell types, such as dopaminergic and cholinergic neurons, can co-release neuromodulators and neurotransmitters that diversify their influence on downstream brain areas⁴⁹. Expression of neuropeptides and neurotransmitters has been reported within the LC, both in classic histological studies using targeted probes and, more recently, through bulk and single-nucleus transcriptomic studies performed on rodent LC neurons^23,25,50. Still, we wondered whether molecular heterogeneity within the LC could be further resolved using alternative approaches. For instance, we reasoned that collecting the entire LC cell body, rather than just nuclei, would provide a richer pool of mRNA for sequencing. Our approach, although laborious, facilitated the use of Smartseq-2, which is inherently low throughput owing to reagent cost and intensive protocol steps. From these experiments, we profiled 201 LC neurons (Fig. 5 and Extended Data Fig. 7), revealing differential expression of many neurotransmitters, neuropeptides, transporters and receptors, the presence of which in the LC can now be more fully characterized using intersectional strategies such as ConVERGD. Furthermore, as Smartseq-2 allows recovery of full-length complementary DNA, this dataset could be useful for further interrogation of splice variants within subsets of LC neurons, which might not be feasible with other LC-sequencing datasets. We also believe that this manual collection and sequencing strategy could be useful for a variety of transcriptomic-based studies where cells of interest exist in low numbers. Ultimately, as experiments continue to uncover heterogeneity within neuronal populations such as the LC, we will need adaptable tools to functionally characterize these cells. Development of ConVERGD allowed us to leverage insights gained from our LC single-cell sequencing experiments to perform this functional exploration.

Methods

Molecular cloning

Building ConVERGD.

An AAV-backbone plasmid (pAAV-hSyn-FLEx(FRT)-eGFP-W3SL) containing the minimal hSyn, an antisense eGFP-encoding region within a Flp-dependent FLEx switch, the W3SL 3′-posttranscriptional regulatory cassette (which consists of sequences encoding WPRE3 (‘W3’) followed by the SV40 late polyadenylation sequence with its upstream element (‘SL’)) and multiple-cloning sites was synthesized (Azenta Life Sciences). A customized oligo was also synthesized (Azenta) that contained an hSyn promoter, a loxP-flanked (floxed) type-III HHR (T3H48)¹⁴ and an eGFP-coding region, all within a Flp-dependent FLEx switch. This ConVERGD-eGFP oligo was ligated into a pAAV-hSyn-FLEx(FRT)-eGFP-W3SL backbone at ApaI and HindIII restriction sites using Sticky-End ligation (NEB). This plasmid (pAAV-hSyn-ConVERGD-eGFP-W3SL) was used for subsequent development of ConVERGD vectors. For expansion, plasmids were transformed into stable competent Escherichia coli (NEB) and grown at 30 °C for 16–20 h for maxiprep (Nucleobond Xtra Maxi kit, Macherey-Nagel).

Creating ConVERGD backbone variants.

The hSyn promoter was excised from the pAAV-hSyn-ConVERGD-eGFP-W3SL backbone via NotI and KpnI restriction enzymes and replaced with the Ef1a promoter (from Ef1a-SplitATG-eGFP-WPRE; gift from the L. Luo laboratory, Stanford University), the CAG promoter (from pAAV-CAG-TVA(mChr)-WPRE; Addgene, cat. no. 48332; a gift from L. Luo) and the nEF promoter (synthesized customized oligo based on the map of nEF-Con/Fon-ChR2(ET/TC)-EYFP; Addgene, cat. no. 137139). The W3SL 3′-posttranscriptional regulatory element was excised with HindIII and BstEII restriction enzymes and replaced with the WPRE sequence (from pAAV-CAG-TVA(mChr)-WPRE) for WPRE-containing ConVERGD variants.

Creating ConVERGD with alternative recombination sites.

To create the loxP-modified ConVERGD variant, a customized oligo (Azenta) containing a ribozyme flanked by lox43/44 sites and an antisense eGFP-coding region was ligated into the pAAV-hSyn-ConVERGD-eGFP-W3SL backbone at AflII and AgeI restriction sites. To create the FRT-modified ConVERGD variant, a customized oligo (Azenta) containing a floxed ribozyme and an antisense eGFP-coding region within a Flp-dependent FLEx switch containing FRT_min/FRT5 sites was ligated into the pAAV-hSyn-ConVERGD-eGFP-W3SL backbone at NheI and HindIII sites. Recombination site variants were combined by ligating the lox43/44 containing an oligo into the ConVERGD FRT_min/FRT5 variant at AflII and AgeI restriction sites.

Key sequences.

Sequences of recombination site variants mentioned above:

loxP: 5′-ATAACTTCGTATAGCATACATTATACGAAGTTAT-3′

lox43: 5′-ATAACTTCGTATAGCATACATTATAGGTACCGAG-3′

lox44: 5′-AATGCATGCTATAGCATACATTATACGAAGTTAT-3′

FRT: 5′-GAAGTTCCTATTCCGAAGTTCCTATTCTCTAGAAAGTATAGGAACTTC-3′

FRT_min: 5′-GAAGTTCCTATTCTCTAGAAAGTATAGGAACTTC-3′

FRT5: 5′-GAAGTTCCTATTCTTCAAAAGGTATAGGAACTTC-3′.

Expanding ConVERGD expression logic.

Customized oligos (Azenta) containing ConVERGD-eGFP-encoding sequence and intersectional machinery in ConFoff (CreNOTFlp), CoffFon (FlpNOTCre) and ConFonvCon (CreANDFlpANDvCre) logics were ligated into pAAV-hSyn-ConVEGD-eGFP-W3SL at NheI and HindIII restriction sites. Quadruple (ConFonvConNon; CreANDFlpANDvCreANDN-igri) and quintuple (ConFonvConNonVon; CreANDFlpANDvCre-ANDNigriANDVika) ConVERGD variants were created by ligating customized oligos (Azenta) containing sets of uniquely flanked ribozymes into pAAV-ConVERGD-eGFP-ConFonvCon at EcoRI and HindIII restriction sites.

Exchanging ConVERGD transgenes.

Transgenes of interest were PCR amplified and inserted into ConVERGD vectors at EcoRI and SpeI restriction sites using NEBuilder HiFi Assembly. Transgenes inserted included: (HA)-hM3Dq (Addgene, cat. no. 125147; a gift from C. Ritchie), (HA)-hM4Di (Addgene, cat. no. 125146; a gift from C. Ritchie), ChR2(GFP) (Addgene, cat. no. 58880; a gift from E. Boyden), ArchT-eGFP (Addgene, cat. no. 28307; a gift from E. Boyden), stGtACR1(FusionRed) (Addgene, cat. no. 105678; a gift from O. Yizhar), TVA-(mCherry) (Addgene, cat. no. 48332; a gift from L. Luo), N2cG (Addgene, cat. no. 73481; a gift from T. Jessell), GCaMP8m (Addgene, cat. no. 162378; a gift from the GENIE Project) and Synaptophysin-GreenLantern-GAP43-mScarlet (generated in the laboratory of L.A.S.).

Comparing the genetic size of ConVERGD and INTRSECT intersectional machinery.

To directly compare the genetic space needed for intersectional machinery between ConVERGD and INTRSECT constructs, we generated a genetic size estimate by counting the base-pairs used within the intersectional components of each strategy. ConVERGD numbers included recombinase sites, spacer sequences between recombinase sites and ribozymes. INTRSECT numbers were based on double intronic intersectional constructs and included recombinase sites, spacer sequences between recombinase sites and intronic sequences (acceptors, donors and spacers). INTRSECT numbers were calculated from annotated sequences available at https://web.stanford.edu/group/dlab/optogenetics/sequence_info.html#intrsect.

Virus production

ConVERGD AAVs (serotype 2/8) were synthesized using the St. Jude Viral Vector Core with standard methods (detailed protocol available on request). Titration of AAV vectors was performed using a modified droplet digital PCR protocol using the following primer-probe set to amplify the AAV ITR sequence: ITR-F: 5′-GGAACCCCTAGTGATG GAGTT-3′, ITR-R: 5′-CGGCCTCAGTGAGCGA-3′ and ITR-Probe: 56-HEX/CACTCCCTC/ZEN/TCTGCGCGCTCG/3IABkFQ. The vector titer per milliliter was calculated by dividing the number of copies per reaction well by 5× the dilution factor. RABV-CVS-N2c^ΔG-H2B-eGFP rabies virus was propagated in the laboratory of L.A.S. from an aliquot obtained from the Allen Institute for Brain Science using cell lines and protocols described in Reardon et al.³⁷.

In vitro testing

Mouse neuroblastoma (N2a) cells (American Type Culture Collection) were co-transfected with ConVERGD and recombinase-expressing plasmids. Cells were cultured in Eagle’s minimal essential medium + 5% penicillin–streptomycin + 10% fetal bovine serum (FBS; HyClone) and seeded on either CC2-treated eight-chamber slides for qualitative analyses or cell-culture-treated, 24-well plates for collection and quantitative FACS analyses. After 1 d, cells were transfected using Lipo2000 (Invitrogen) with standard reagent ratios. For chamber slide transfections, 50 ng of each of the sample and recombinase plasmid DNA was used. For 24-well plate transfections, 250 ng of the each of the sample and recombinase plasmid DNA was used except for ConVERGD-ConFonvConNon-eGFP transfections, where we observed adverse effects on cell health, possibly caused by the large quantity of transfected plasmids. Instead, we used 125 ng of each of the ConVERGD vector and recombinase-expressing plasmids for these experiments. Then, 2-d posttransfection, cells in the chamber slides were fixed with 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS) and stained with the appropriate antibody, if necessary. All fixed slides were counterstained with 1:10,000 of 5 mg ml⁻¹ of DAPI (Sigma-Aldrich) in PBS. Fixed slides were imaged with a Leica DM6 automated fluorescence microscope for qualitative analysis. For quantitative FACS analysis, cells in 24-well plates were trypsinized 3-d posttransfection and collected into individual 1.5-ml tubes. Cells were pelleted at 200g for 5 min, resuspended in 200 μl of 1× PBS and transferred to filtered FACS tubes on ice. DAPI was added as a counterstain to exclude dead cells immediately before FACS analysis (BD FACSAria Fusion running BD FACSDiva v.8.0.1). FlowJo was used to analyze and quantify the results. Cells were positively gated for high-density populations plotted by side scatter (SSC-A) and forward scatter (FCS-A) (to isolate cells from debris), positively gated for DAPI-negative cells (to isolate live from dead cells), positively gated for high-density populations plotted by FSC-A and SSC pulse width (SSC-W) (to isolate single cells), and positively gated for GFP⁺ cells (to isolate GFP⁺ cells). Samples with <7,500 live single cells were excluded from analysis. Normalized control values were calculated by averaging the median fluorescence intensity (MFI) readings of all control samples for each construct. Fold-change values reported represent the fold-change of the MFI of the live single cells in each sample from these normalized control values. Percentage positive values represent the percentage of live single cells that were gated to be GFP⁺. FACS data are reported from at least three independent transfection experiments.

PCR and RT–PCR

The following primer pairs were designed to assess nonspecific recombination events in ConVERGD and INTRSECT plasmids:

eGFP2: tctcgttggggtctttgctc; eGFP4: ctaccccgaccacatgaagc

eGFP4: ctaccccgaccacatgaagc; WPRE2: ccacatagcgtaaaaggagcaaca

eGFP4R: gcttcatgtggtcggggtag; WPRE2: ccacatagcgtaaaaggagcaaca.

PCR reactions were conducted using 10 ng of maxiprepped plasmid DNA following the manufacturer’s protocols (ECONOTaq Plus, Biosearch Technologies) and 7 μl of each PCR reaction was resolved on a 2% agarose gel.

The following primer pairs were designed to assess nonspecific and partial recombination of ConVERGD transcripts in transfected N2a cells:

eGFPf1: gtgctcaggtagtggttgtc; eGFPr1: agcagaagaacggcatcaa

ribof1: gatcactctcggcatggac; ribor1: agctggatgtacgcgttt

ribof1: please see above; wPREr1: cagaggttgattatctcgggca.

N2a cell transfections were conducted as described above for 24-well plate transfections, with 125 ng of ConVERGD vector and 125 ng of recombinase-expressing plasmid used for each transfection. After 2 d, cells were lysed and mRNA was extracted using a RNeasy micro-extraction kit (QIAGEN, cat. no. 74004). The cDNA was generated from eluted mRNA samples (10 ng of mRNA per sample) using the iScript gDNA CLR cDNA kit (BioRad, cat. no. 1725035), which includes additional DNase steps to remove residual genomic DNA from mRNA samples. Reactions without RT were also generated in parallel. From these reactions (RT and no RT) 1 μl was used as a template for subsequent PCR reactions using specified primer pairs. PCR products were resolved on a 2% agarose gel along with 1 kb of Plus DNA ladder (Invitrogen).

Animals

Adult (aged 7–37 weeks) male and female mice were housed in sterile micro-isolator cages on ventilated racks (up to five mice in a cage with corncob bedding and access to nestlets) and kept on a 12-h light:dark cycle (lights on at 06:00), 40–60% relative humidity and free access to food and water. All experiments were conducted in accordance with a protocol approved by the Institutional Animal Care and Use Committee of St. Jude Children’s Research Hospital, an AAALAC-accredited facility, and conform to US National Institutes of Health (NIH) guidelines. Detailed information about the specific strains used in the present study, as well as the originating publications related to their development, are referenced in Supplementary Table 2.

Stereotaxic surgery

Mice were anesthetized and prepared for surgery following institutional guidelines. Craniotomies were performed to deliver virus directly to targeted brain regions via a microinjection pump (World Precision Instruments) and pulled glass electrode. The LC was targeted with 350 nl of virus at (from lambda): −0.8 mm (posterior), ±0.8 (lateral) and 3.2 mm (ventral from the brain surface). ConVERGD viruses were diluted to a titer of ~8 × 10¹¹ g.c.p. ml⁻¹ to minimize nonspecific expression. For expression comparisons between ConVERGD and INTRSECT, a working titer of ~8 × 10¹¹ g.c.p. ml⁻¹ was used for all injections of both viruses. The titer of RABV-CVS-N2c^ΔG-H2B-eGFP was ~2.1 × 10⁸ infectious particles per ml based on serial dilutions of the virus stock followed by infection of the 293-TVA800 cell line. For hippocampal injections, hippocampus was targeted with 300 nl of virus at (from bregma): +1.5 mm (anterior), ±2 mm (lateral) and 1.5 mm (ventral from the brain surface). ConVERGD and INTRSECT AAVs were injected at a titer of ~3.0× 10¹³ g.c.p. ml⁻¹.

Behavioral assays

Adult male and female mice, aged 10–19 weeks at the time of behavior, were used. All mice were bilaterally injected with 350 nl of AAV8-hSyn-ConVERGD-(HA)hM3Dq-W3SL into the LC at least 3 weeks before behavioral experiments. Mice were divided into control and experimental (hM3Dq-expressing) groups. The control group consisted of WT (two female, three male), Pdyn^Cre (two female, three male) and Dbh^Flp (two female, three male) mice. Behavioral data for control genotypes (WT, Pdyn^Cre and Dbh^Flp mice) were combined after ordinary one-way ANOVA tests with a post hoc Tukey’s multiple-comparison test showed no statistical difference between genotypes; EZM control group P values: percentage of time in open region: WT versus Pdyn^Cre versus Dbh^Flp: 0.3656 (ANOVA), WT versus Pdyn^Cre: 0.6926 (adjusted, Tukey’s test), WT versus Dbh^Flp: 0.3358, Pdyn^Cre versus Dbh^Flp: 0.7975 (adjusted, Tukey’s test); percentage of distance in open region: WT versus Pdyn^Cre versus Dbh^Flp: 0.9578 (ANOVA), WT versus Pdyn^Cre: 0.9599 (adjusted, Tukey’s test), WT versus Dbh^Flp: 0.9710 (adjusted, Tukey’s test) and Pdyn^Cre versus Dbh^Flp: 0.9991 (adjusted, Tukey’s test). OFT control group P values: percentage of time in center: WT versus Pdyn^Cre versus Dbh^Flp: 0.3320 (ANOVA), WT versus Pdyn^Cre: 0.8471 (adjusted, Tukey’s test), WT versus Dbh^Flp: 0.7131 (adjusted, Tukey’s test), Pdyn^Cre versus Dbh^Flp: 0.9682 (adjusted, Tukey’s test); percentage of time in corners: WT versus Pdyn^Cre versus Dbh^Flp: 0.4785 (ANOVA), WT versus Pdyn^Cre: 0.8764 (adjusted, Tukey’s test), WT versus Dbh^Flp: 0.4517 (adjusted, Tukey’s test) and Pdyn^Cre versus Dbh^Flp: 0.7385 (adjusted, Tukey’s test). The experimental (hM3Dq-expressing) group consisted of Pdyn^Cre;Dbh^Flp (four female, seven male) mice, which had bilateral hM3Dq expression in the LC confirmed via histology after behavioral experiments had been completed. Pdyn^Cre;Dbh^Flp mice lacking hM3Dq expression were considered to have missed injections and were excluded from analysis.

Elevated zero maze.

The EZM apparatus was a 70-cm wide, circular track 52 cm in diameter and 60 cm above the ground. The track contained two closed regions on opposite sides of the maze with 14-cm-tall opaque walls that each ran a quarter of the circumference of the track. Mice were handled for 4 d of acclimation: day 1: the experimenter placed their hands in the home cage for 5 min; day 2: the experimenter held each mouse for 5 min; day 3: mice were weighed, scruffed and given an i.p. saline injection; day 4: mice were weighed, scruffed and given an i.p. injection of saline 30 min before being placed on the EZM for ~12 min. On the test day, mice were acclimated to the testing room for at least 15 min before being weighed, scruffed and given an i.p. injection of 5 mg kg⁻¹ of CNO (HelloBio). Then, 30 min after the CNO injection, the mice were placed on the EZM at the border of the open and closed region facing the open region. Mice were video recorded from above for the entirety of the 10-min trial. Centroid tracing data were acquired with a customized Bonsai program and analyzed using a customized MATLAB script (included with Source data). To determine the time spent in each region, Bonsai was used to draw regions of interest (ROIs) around each open region and the amount of time that the ROI was triggered was recorded in real time. This method was very sensitive and counted any triggering of the ROI, including head pokes into the open region, as time spent in the open region. However, the threshold on the video analysis was set such that the tail was not visible by the ROI. This ensured exclusion of false triggers that could be caused by the mouse’s tail lying across the closed/open threshold. To calculate the distance traveled in each region, MATLAB was used to analyze centroid data obtained from the Bonsai tracking. ROIs were drawn around the closed regions and the cumulative distance between centroid points within these regions was calculated.

Open field testing.

The open field chamber consisted of a 40 × 40 cm² plastic chamber with 35-cm-tall transparent walls enclosing the chamber. Mice were video recorded from above for the entirety of the 10-min trial. The same mice used in the EZM experiments were used for OFT. Then, 1 week after EZM testing, mice were acclimated to the OFT room for at least 15 min, weighed, scruffed and given an i.p. injection of 5 mg kg⁻¹ of CNO 30 min being placed in the open field chambers. CleverSys TopScan software was used to track and analyze OFT behavior.

ScRNA-seq

Adult Dbh^Cre;Ai14 mice (aged ~8–20 weeks) were euthanized with tribromoethanol (Avertin). Brain tissue was immediately extracted and placed in ice-cold artificial cerebrospinal fluid (aCSF; 2.5 mM KCl, 7 mM MgCl₂, 0.5 mM CaCl₂, 1.3 mM NaH₂PO₄, 110 mM choline chloride, 25 mM NaHCO₃, 1.3 mM sodium ascorbate, 20 mM glucose and 0.6 mM sodium pyruvate and bubbled in 95% O₂/5% CO₂). Submerged brain tissue was placed on a VT1200 vibratome (Leica) and 200-μm coronal slices that included the LC were collected. Slices were transferred to a 32 °C holding chamber containing oxygenated aCSF, where they recovered for at least 1 h. Cells were visualized using an Olympus BX-51 fluorescence microscope and tdTomato-expressing LC neurons were manually extracted from the slice using glass pipettes pulled from borosilicate glass (World Precision Instruments) to create an ~50-μm tip. The ends of the glass pipette tips containing picked cells were snapped into individual wells of a 96-well plate containing 4 μl of lysis buffer: 0.1 μl of RNase inhibitor (Thermo Fisher Scientific), 0.1 μl of ERCC RNA spike-in (Invitrogen), 0.02 μl of 10% Triton (Sigma-Aldrich), 1 μl of 10 mM dNTP (Invitrogen), 0.1 μl of 100 μM dT (IDT; 5′-AAGCAGTGGTATCAACGCAGAGTACTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVN) and 2.68 μl of H₂O. After collection, plates were spun down, sealed and stored at −80 °C. Samples next underwent primer annealing (72 °C for 3 min) before placing on ice. Then, 6 μl of reverse transcription reagent was added to each sample: 0.14 μl of H₂O, 0.25 μl of RNase inhibitor, 2 μl of 5× First Strand Buffer (Clontech), 0.5 μl of 100 mM dithiothreitol (Invitrogen), 2 μl of 5 M Betaine (Sigma-Aldrich), 0.06 μl of 1 M MgCl₂ (Invitrogen), 0.1 μl of 100 μM TSO (QIAGEN; 5′-AAGCAGTGGTATCAACGCAGAGTACATrGrG+G) and 0.95 μl of SmartScribe enzyme (Clontech). Samples were incubated at 42 °C for 90 min, 70 °C for 5 min and then held at 4 °C. DNA preamplification solution (15 μl) was added to each sample: 2.1375 μl of H₂O, 12.5 μl of 2× Kapa HiFi Hot Start Ready Mix (Kapa Biosystems), 0.25 μl of 10 μM IS_PCR Primer (IDT; 5′-AAGCAGTGGTATCAACGCAGAGT) and 0.1125 μl of Lambda Exonuclease (NEB). DNA cleanup, quantification, dilution, tagmentation and barcoding were performed by the Hartwell Center at St. Jude Children’s Research Hospital using standard methods. Libraries were sequenced on a HiSeq4000 or MiSeq Sequencing Systems (Illumina) using 2× 100-bp or 2× 75-bp ends, respectively. Sequences were de-multiplexed using bcl2fastq and aligned to a mouse mm10 genome (with Cre and tdTomato genes added). Paired-end reads from Illumina were trimmed for adapters by cutadapt (v.1.15) and aligned to mm9 (MGSC v.37; National Center for Biotechnology Information (NCBI)) using STAR (v.2.5.2b). LiftOver was performed on GENCODE (v.M9) gene database (from mm10 to mm9) using CrossMap (v.0.1.5) to improve gene annotation. Aligned reads were then counted by HTSEQ (v.0.6.1p1) for genes annotated in GENCODE database. Trimmed reads were also run through FastQ Screen (v.0.9.5) against Human, Mouse, Adapters and Univec databases (December 2016). Samples with larger unique hits to human over mouse genomes and putative doublets were excluded from the analysis. We further excluded samples with <1 million reads and those that did not express Dbh. We then conducted an unbiased analysis to identify the top 100 genes that appeared most frequently across LC samples. Next, we downloaded gene lists from GO for selected terms and plotted heatmaps (z-score) for the top expressed genes in each GO term using the ComplexHeatmap R package.

Histology

For all histology experiments, mice were euthanized with tribromoethanol (Avertin).

In situ hybridization.

Cryosections, 16–20 μm, of fresh brain tissue containing the LC were collected. Samples were labeled using the RNAscope Multiplex Fluorescent Reagent Kit v.2 (ACDBio) following the supplied protocol and using the following probes: Calca-C1 (cat. no. 420361), Cartpt-C1 (cat. no. 432001), Dbh-O1-C3 (cat. no. 464621-C3), Gabrq-C1 (cat. no. 523941), Gad2-C1 (cat. no. 439371), Pdyn-C1 (cat. no. 318771) and Penk-C1 (cat. no. 318761). The probe signal was developed using Opal dyes (Opal 520 and 570 Reagent Pack, Perkin Elmer) at a dilution of 1:1,000 and counterstained with DAPI. Fluorescent images were taken using a Leica LSM780 laser-scanning confocal microscope.

Combined in situ hybridization and immunostaining.

Cryosections, 20 μm, of fresh brain tissue containing the LC were collected for Molecular Instruments-based hybridization chain reaction in situ hybridization. Samples were labeled using customized probes (Dbh-B2 and Pdyn-B5) following the supplied protocol. After RNA amplification and labeling, slides were washed with PBS (3×, 15 min), blocked (PBS with 0.05% Triton X-100 and 3% bovine serum albumin) for 1 h at room temperature and incubated overnight at room temperature with GFP–Rb antibody (1:500). The following day, the slides were washed with a PBS solution containing 0.1% Tween-20 before application of secondary antibodies (1:500) for 2 h at room temperature. All sections were counterstained in DAPI before mounting. Fluorescent images were taken using a Leica LSM780 laser-scanning confocal microscope.

Immunostaining.

Euthanized mice were transcardially perfused with PBS and 4% PFA. Brain tissue was removed, placed in 4% PFA at 4 °C overnight and then equilibrated to a 30% sucrose solution at 4 °C (2–4 overnights). Cryosections, 50 μm, containing relevant brain regions were collected and washed with PBS (3×, 10 min). Slide-attached sections were stained with the appropriate primary antibody (1:1,000 dilution in PBS + 2% normal donkey serum (NDS) + 0.2% Triton X-100) either at 4 °C for 3 overnights (slide attached) or at room temperature overnight with shaking (floating sections). Sections were stained with the appropriate secondary antibody (1:500 dilution in PBS + 2% NDS) at room temperature for 2 h and counterstained in DAPI before mounting. ConVERGD-eGFP, INTRSECT-YFP and Pdyn^Cre;Sun1-eGFP samples were stained with GFP-Ch and Th-Rb (for LC sections). For cFos staining, every other section of ConVERGD-(HA)hM3Dq-injected brains was stained with TH-Rb and human influenza hemagglutinin (HA)-Ms or TH-Ms and cFos-Rb. As a result of the overlap of antibodies against TH, we did not stain against TH, HA and cFos together. Fluorescent images were taken using a Nikon C2 or Leica LSM780 laser-scanning confocal microscope or a Leica DM6 automated fluorescence microscope.

Fluorescent image analysis.

To compare fluorescently labeled cells between ConVERGD-eGFP and INTRSECT-EYFP, six 50-μm sections spanning the anterior–posterior axis of the LC were selected from at least three mice per genotype (Pdyn^Cre, Dbh^Flp and Pdyn^Cre;Dbh^Flp) per virus group. All sections were immunostained against GFP as described above and the images were taken with a Leica LSM780 laser-scanning confocal microscope. Fluorescently labeled cells within these images were manually counted. To quantify rabies-tracing experiments, every other section throughout the whole brain was collected. Images of every other collected section were taken using a ×5 objective on a Leica DM6 fluorescent microscope. GFP⁺ nuclei were manually counted for every image and each image was assigned a percentage of the total counted GFP⁺ nuclei within each brain. Fluorescence images were then manually registered to the corresponding image in a reference atlas (Paxinos and Franklins). Using Rstudio, data were grouped by the registered atlas image number and an average percentage of the total number of labeled cells was determined for each registered image section. Peaks in the average labeled cell count of at least 3% of total counted cells were chosen to obtain example images from the major input regions. Example input images were taken with a Nikon C2 fluorescent confocal microscope. To visualize axon projections, we analyzed sections that had been collected for rabies quantification. These sections were immunolabeled with an anti-red fluorescent protein antibody to amplify the signal from AAV8-hSyn-ConVERGD-TVA(mC hr)-W3SL expression. Qualitative images of axon projections within ROIs were taken using a Nikon C2 fluorescent confocal microscope. Axon projection patterns to brain regions represented in Fig. 7 were observed in multiple samples.

Statistics and reproducibility

No statistical methods were used to predetermine sample size. Data were excluded in certain conditions (contaminated cell samples for sequencing, dead cells in FACS experiments, missed viral injections for behavioral experiments) and are noted in the methods corresponding to those experiments. Data distribution was assumed to be normal but this was not formally tested. Investigators were not blinded to mouse genotype during experiments or quantification of data. However, care was taken to ensure that a similar number of male and female mice were included across genotypes and experimental conditions.

Considerations for optimizing ConVERGD expression specificity in vivo

A low level of nonspecific expression was routinely observed for both ConVERGD and INTRSECT viral vectors (Fig. 6 and Extended Data Figs. 6 and 8). For ConVERGD, this occurred more frequently when paired with transgenically expressed recombinases and in the presence of Flp alone. We found that this Flp-specific leak could be exacerbated at higher viral titers (for example, 10¹³ g.c.p. ml⁻¹), but it also seemed to be influenced by the transgenic line that was used. For instance, more leak was qualitatively observed in Dbh^Flp mice than in Slc17a7^Flp mice. An early tester of ConVERGD from an independent laboratory also reported enhanced Flp-specific leak when animals were homozygous for Flp transgenes. Thus, it is important for end-users to validate ConVERGD within the experimental paradigms in which they intend to apply it.

Extended Data

Extended Data Fig. 1 |. Comparison of ConVERGD-based constructs with varying promoter and posttranscriptional elements.

a, FACS quantification of N2a cells co-transfected with an EYFP- or eGFP-expressing plasmid alone (Control, grey bars) or with recombinase plasmids expressing Cre (yellow bars), Flp (blue bars), or Cre and Flp (pDIRE, green bars). ConVERGD was tested in pAAV backbones containing different promoters and 3′ posttranscriptional regulatory elements. b, FACS quantification as in a but represented as percent of live, single cells that were counted positive for fluorescence. Bars represent the mean of all experiments. Error bars are SEM. CV - ConVERGD. Data points represent independent transfections with the following constructs: 56 (FLEx(FRT)eGFP alone or +pDIRE), 6 (CV-eGFP-W3SL +Flp, +pDIRE; CV-eGFP-WPRE, +Flp, +pDire; Ef1a-CV-eGFP-WPRE, +pDIRE) 5 (INTRSECT-EYFP, +Cre, +Flp, +pDIRE; CV-eGFP-W3SL, +Cre; CV-eGFP-WPRE+Cre; nEF-CV-eGFP-WPRE, +pDIRE) 4 (CV-EYFP, +Flp; CAG-CV-eGFP-W3SL, +Cre, +Flp, +pDIRE; CAG-CV-eGFP-WPRE; Ef1a-CV-eGFP-W3SL, +Cre, +Flp, +pDIRE; Ef1a-CV-eGFP-WPRE+Cre, +Flp; nEF-CV-eGFP-W3SL, +Cre, +Flp, +pDIRE; nEF-CV-eGFP-WPRE+Cre, +Flp) 3 (CV-EYFP+Cre, +pDIRE; CAG-CV-eGFP-WPRE+Cre, +Flp, +pDIRE).

Extended Data Fig. 2 |. Assessment of leaky expression from ConVERGD and INTRSECT constructs.

a, Schematics for ConVERGD and INTRSECT constructs where no recombination has occurred, or upon Cre, Flp, or Cre and Flp-mediated recombination. Expected band sizes via PCR with specified primers are listed below. b, PCR of pAAV-hSyn-ConVERGD-eGFP-W3SL and pAAV-hSyn-INTRSECT-eYFP plasmids using the primer pairs described in a. c, Schematics for ConVERGD vectors where no recombination has occurred, or upon Cre, Flp, or Cre and Flp-mediated recombination. Schematics for vectors undergoing partial Flp-mediated recombination are also included. Expected band sizes via PCR with specified primers are listed below. d, Amplified PCR product from N2a cells transfected with hSyn-ConVERGD-eGFP-W3SL alone or with Cre, Flp, or Cre and Flp (pDIRE) expressing plasmids. mRNA extracted from these samples underwent a reverse transcriptase (RT) reaction to generate cDNA, or a no reverse transcriptase (no RT) reaction as a control. Each gel displays PCR products from template arising from the RT and no RT reactions. The gels are representative of results obtained across four independent sets of transfections. CV - ConVERGD; INTR – INTRSECT.

Extended Data Fig. 3 |. Different recombination sites did not improve ConVERGD performance.

a, FACS quantification of N2a cells co-transfected with ConVERGD-eGFP construct variants containing different recombinase recognition sites alone (Control, grey bars) or with recombinase plasmids expressing Cre (yellow bars), Flp (blue bars), or Cre and Flp (pDIRE, green bars). Data represented as the fold change of median fluorescence intensity (MFI) of transfection condition compared to the average control MFI for each construct. b, The same FACS quantification as in a but represented as percent of live, single cells that were counted positive for eGFP fluorescence. Data points represent individual transfection experiments. In a and b, results represent data from 7 (FRT5/FRT;loxP), 3 (FRT5/FRT;lox43/44), 6 (FRT5/FRT(min);loxP), 6 (FRT5/FRT(min);lox43/44) separate transfection experiments. Bars represent the mean of all experiments. Error bars are SEM.

Extended Data Fig. 4 |. Assessment of ConVERGD expression as percent of live N2a cells.

a, FACS quantification of N2a cells co-transfected with ConVERGD-ConFoff-eGFP (left) or ConVERGD-CoffFon-eGFP (right) either alone or with recombinase-expressing plasmid. b, FACS quantification of N2a cells co-transfected with ConVERGD-ConFonvCon-eGFP either alone or with recombinase-expressing plasmid. c, FACS quantification of N2a cells co-transfected with ConVERGD-ConFonvConNon-eGFP either alone or with recombinase-expressing plasmid. Data points in all panels represent the percent of live cells that contain eGFP from individual transfections. Data points represent independent transfections with the following constructs: 3 (CV-ConFoff-eGFP control; CV-ConFonvConNon-eGFP +Flp/Nigri, +vCre/Nigri, +vCre/Flp/Nigri, +Cre/Flp/vCre/Nigri), 4 (CV-ConFoff-eGFP +Cre, +Flp, +pDire; all conditions for ConFonvCon; CV-ConFonvConNon-eGFP +Cre/Flp/Nigri), 5 (CV-CoffFon-eGFP all conditions; CV-ConFonvConNon-eGFP +Cre/vCre, +Cre/Nigri, +Cre/Flp/vCre), 6 (CV-ConFonvConNon-eGFP, +Cre, +Flp, +vCre, +Cre/vCre/Nigri), 7 (CV-ConFonvConNon-eGFP +Nigri, +Flp/vCre), and 9 CV-ConFonvConNon-eGFP +Cre/Flp. Bars represent the mean of all experiments. Error bars are SEM.

Extended Data Fig. 5 |. ConVERGD-based constructs are easily amenable and allow specific expression of diverse transgenes.

a, ConVERGD-based toolkit for modulating neuronal activity. b, ConVERGD-based toolkit for trans-synaptic rabies tracing. c, ConVERGD-based construct for in vivo calcium imaging (GCaMP8m; GC8m). d, ConVERGD-based construct for a dual-expressing transgene that labels pre-synaptic sites and axons (synaptophysin-GreenLantern and GAP43-mScarlet). All images show transfected N2a cells counterstained with DAPI (blue) and are representative of results observed across at least two separate transfections. FR - FusionRed; mChr - mCherry; GL - GreenLantern; mSc - mScarlet. Scale bar in a is 100μm and applies to all images.

Extended Data Fig. 6 |. ConVERGD shows specific, intersectional expression in the hippocampus of Calb1^Cre; Slc17a7^Flp mice.

a, Representative images of labeled cells in the hippocampus upon injection of Cre-dependent eGFP AAV in Calb1^Cre mice (left) or Flp-dependent mCherry AAV in Slc17a7^Flp mice (right). b, Representative images of AAV-hSyn-ConVERGD-eGFP or AAV-hSyn-INTRSECT-Con/Fon-EYFP injected into the hippocampus of Calb1^Cre, Slc17a7^Flp, and Calb1^Cre;Slc17a7^Flp mice. Tissue sections in the top two rows reflect endogenous fluorescence while tissue sections in the bottom two rows were immunostained with GFP antibody. c, Representative images of AAV-hSyn-ConVERGD-ConFonvCon-eGFP injected into the hippocampus of Calb1^Cre;Slc17a7^Flp mice in the absence (left) or presence (right) of vCre-expressing AAV. Tissue sections were immunostained with GFP antibody. d, Representative images of AAV-hSyn-ConVERGD-ConFoff-eGFP injected into the hippocampus of Calb1^Cre (left) or Calb1^Cre;Slc17a7^Flp (right) mice. Tissue sections were immunostained with GFP antibody. e, Representative images of AAV-hSyn-ConVERGD-CoffFon-eGFP injected into the hippocampus of Slc17a7^Flp (left) or Calb1^Cre;Slc17a7^Flp (right) mice. Tissue sections were immunostained with GFP antibody. Images in a, c, d, and e are representative of results across 2 mice for each genotype; images in b are representative of results across 3 mice for each genotype. Scale bars are 100μm. WT - wild-type; Calb1 - calbindin 1; Slc17a7 - solute carrier family 17 member 7; INTR - INTRSECT; CV - ConVERGD.

Extended Data Fig. 7 |. Top 100 most frequently detected genes in LC neurons using a Smart-seq2-based sequencing platform.

a, Heatmap of scaled (by cell) transcript abundance (transcripts per million, TPM) for the top 100 genes most frequently detected in 201 LC neurons by single-cell transcriptomic sequencing.

Extended Data Fig. 8 |. Increased single-recombinase induced expression observed with INTRSECT.

a, Representative images showing the LC (TH, white) of mice injected with AAV-hSyn-INTRSECT-Con/Fon-EYFP. All genotypes showed some level of YFP (green) expression. b, Quantification of INTRSECT-EYFP labeled cells in and around (~200μm radius) the LC across different genotypes. Points represent cell counts across 6 50μm LC brain sections. Bars represent the mean of the data. Error bars are SEM. All sections were immunostained against GFP. Images in a are representative of results observed across 4 (Pdyn^Cre), 4 (Dbh^Flp), and 5 (Pdyn^Cre;Dbh^Flp) animals. Scale bars are 100μm. TH - tyrosine hydroxylase; LC - locus coeruleus; Pdyn - prodynorphin; Dbh - dopamine-β-hydroxylase.

Supplementary Material

The online version contains supplementary material available at https://doi.org/10.1038/s41593-024-01659-7.

Data availability

RNA-seq data are deposited in the NCBI Gene Expression Omnibus database with accession no. GSE224285. Source data are provided with this paper.