AAV-delivered TurboRFP enables streamlined tracing and analysis of corticospinal tract sprouting Authors: Marc Hernaiz-Llorens, Camilo J Londoño, Ellen Wang, Felicia Cai, Junmi M Saikia, Carmine L Chavez-Martinez, Hugo J Kim, Chimuanya K Agba, Binhai Zheng Published: 2026 ABSTRACT SummaryQuantitative analysis of corticospinal tract (CST) sprouting after injury requires reliable labeling of long-range axons and fine collateral branches. Conventional biotinylated dextran amine (BDA) tracing has limited sensitivity and requires additional surgeries, while some viral-based approaches, although robust, rely on extensive tissue processing and signal amplification. Here, we describe a streamlined adeno-associated virus (AAV)-based workflow for CST sprouting analysis using TurboRFP that enables robust labeling of descending CST axons, including sprouting fibers after unilateral pyramidotomy. This approach allows direct visualization of fine CST axons without immunostaining or signal amplification, simplifying tissue processing and reducing experimental variability. Using a standardized workflow, we enable consistent CST labeling and reproducible quantification of CST remodeling. In addition, compatibility with co-delivery of other AAVs enables simultaneous circuit tracing and genetic manipulation within the same neuronal population. This workflow provides a practical platform that lowers technical barriers to axon repair research and potentially improves reproducibility across laboratories.Keywords: axon tracing, corticospinal tract, CST, axon sprouting, CNS injury, neural repair, pyramidotomy, AAV vectors, fluorescent labeling Introduction Quantitative analysis of long-range axonal projections requires reliable labeling of projection neurons and their axons. The corticospinal tract (CST), a major descending motor pathway, is widely used to study axonal plasticity and regeneration after spinal cord injury (SCI).1,2 Because CST axons extend over long distances and exhibit limited spontaneous regeneration, this system provides a stringent and well-established platform for evaluating axonal growth and remodeling after injury.3,4 Biotinylated dextran amine (BDA), a classical chemical anterograde tracer, has been extensively used to label CST axons.5,6 BDA tracing has supported major advances in CST biology, including enhanced regeneration following Pten deletion7 and extensive spontaneous corticospinal plasticity after primate SCI.8 Detection of BDA-labeled axons typically requires signal amplification, which, historically, was performed using chromogenic methods such as diaminobenzidine (DAB) and, more recently, involves fluorescent approaches, adding processing steps that can increase variability. Adeno-associated virus (AAV)-based methods have emerged as versatile tools for neural circuit tracing and manipulation, enabling targeted labeling and genetic access to specific neuronal populations.9,10 These approaches have been used to label CST axons and identify regulators of CST regeneration and remodeling.11,12 Efforts to improve AAV-based axon tracing have focused on optimizing viral serotypes and promoters to enhance labeling efficiency.13,14 However, robust detection of fine axons often still relies on immunostaining or signal amplification, which can be labor intensive and variable across experiments. Although AAV-mediated fluorescent tracing is widely used and prior studies have compared AAV-based labeling with BDA in the brain, fewer studies have addressed practical workflows for detecting sparse CST collateral and sprouting axons within the spinal cord after injury. Here, we evaluate an AAV1-hSyn-TurboRFP-based workflow for CST sprouting analysis after unilateral pyramidotomy, in comparison with conventional BDA tracing. Rather than introducing a new class of axonal tracer, this study establishes a simplified and reproducible pipeline for detecting and quantifying CST remodeling after injury. Our results demonstrate that AAV-TurboRFP enables robust CST labeling without immunostaining or signal amplification, providing a practical platform for quantitative analysis of axonal sprouting. Results AAV1-hSyn-TurboRFP robustly transduces CST neurons in the sensorimotor cortex AAVs are widely used gene delivery vectors characterized by broad but tunable neuronal tropism, low pathogenicity, and minimal immunogenicity.15 Different AAV serotypes exhibit marked differences in the transduction efficiency, cell type specificity, and axonal transport properties.16,17,18 Among these, AAV1 has been shown to be particularly effective for neuronal transduction and is among the most efficient serotypes for labeling CST neurons and their axons.13,14,19 In addition, the human synapsin (hSyn) promoter provides an optimal balance of neuronal specificity and expression efficiency for CST labeling.14,20 Based on these considerations, we used an AAV1-hSyn-TurboRFP construct (Addgene viral prep #105552-AAV1; RRID: Addgene_105552), in which the hSyn promoter drives expression of the dimeric red fluorescent protein TurboRFP.21 TurboRFP exhibits high brightness, rapid maturation, and the excitation/emission peak at 553/574 nm, while complementing commonly used GFP-based vectors. A schematic of the vector construct is shown in Figure 1A. Experimental design and TurboRFP expression in the mouse brain (A) AAV1-hSyn-TurboRFP vector design. (B) Experimental timeline for cortical AAV injection and tissue collection. (C) Schematic of the AAV1-hSyn-TurboRFP injection sites in the sensorimotor cortex. (D) TurboRFP expression across three cortical injection sites. (E) TurboRFP-labeled neurons in cortical layers II/III and V. (F and G) TurboRFP-labeled axons projecting through the corpus callosum (cc) (F) and internal capsule (ic) (G). (H) TurboRFP-labeled corticospinal axons descending through the medullary pyramid (Py) and pyramidal decussation (Dec). To label cervical-projecting corticospinal neurons (CSNs), we injected AAV1-hSyn-TurboRFP into the forelimb sensorimotor cortex of 6-week-old mice and collected tissue 6 weeks later (Figures 1B and 1C). Viral particles were delivered across three cortical sites to maximize consistency and coverage of the forelimb CSN territory (Figures 1C and 1D; see STAR Methods). Robust TurboRFP fluorescence was observed without the need for immunostaining or signal amplification across the motor cortex at all injection sites (Figure 1D). Labeled neurons were distributed predominantly in cortical layers II/III and V (Figure 1E), reflecting labeling of corticocortical neurons and layer V projection neurons, including CSNs. Labeled CST axons could be readily identified along their trajectory, including fiber bundles within the corpus callosum (cc) and internal capsule (ic) (Figures 1F and 1G). These axons converge at the medullary pyramids (Py) and decussate (Dec) and can be continuously traced caudally along the CST toward the spinal cord (Figure 1H).22 Fine axonal processes could be clearly resolved, and the pyramidal decussation was sharply delineated, highlighting the sensitivity of AAV1-mediated fluorescent tracing for resolving corticospinal projections (Figure 1H). AAV1-hSyn-TurboRFP provides more efficient cortical labeling than BDA and enables standardized CST axon quantification at the level of the pyramidal decussation BDA is a widely used anterograde tracer for mapping neural circuits and assessing axonal regeneration and sprouting, typically detected using avidin-biotin complex methods or fluorescently conjugated streptavidin.5 To compare viral and conventional tracing strategies, the mice received AAV1-hSyn-TurboRFP injections followed 4 weeks later by BDA injections at matching cortical coordinates; tissues were collected 2 weeks after BDA injection (Figures 2A and 2B). TurboRFP was detected by native fluorescence, whereas BDA was detected using fluorescently labeled streptavidin without amplification, enabling within-animal comparison under minimally processed conditions. Comparison of labeling efficiency between AAV-mediated TurboRFP expression and BDA (A) Experimental timeline for AAV1-hSyn-TurboRFP and BDA injections. (B) Schematic of the cortical injection sites. (C and D) Coronal brain sections showing TurboRFP (C) and BDA (D) labeling. (E) High-magnification and merged images showing TurboRFP and BDA labeling in the cortex. (F) Quantification of labeled neurons and cortical labeling area per injection site. Data are mean ± SEM; n = 8 animals, 3 sections per animal; paired t test. (G and H) Transverse medulla sections showing TurboRFP-labeled (G) and BDA-labeled (H) CST axons. (I) Merged TurboRFP and BDA labeling at the medullary level. (J) Quantification of total labeled CST axons in the medulla. Data are mean ± SEM; n = 8 animals per group, 3 sections per animal; paired t test. AAV1-hSyn-TurboRFP injections produced robust and widespread fluorescence across cortical layers without the need for secondary staining or signal amplification (Figure 2C). In matched sections, TurboRFP labeled more CSNs and covered a larger cortical area than BDA (Figures 2D and 2E). Immunodetection of RFP (detected with Alexa Fluor 488, in green) produced signals that largely overlapped with native TurboRFP fluorescence (in red) across the CNS (brain, medulla, and cervical spinal cord) but did not improve visualization of fine CST collaterals or varicosities in the cervical spinal cord and introduced substantial background signal under our experimental conditions (Figure S1); therefore, this approach was not pursued further. Quantitative analyses confirmed that both the number of labeled neurons (∼5-fold increase) and the cortical area occupied by labeled somata (∼7-fold increase) were significantly greater with TurboRFP than with BDA (Figure 2F). To establish a reliable normalization metric across animals, we quantified labeled CST axons at the level of the Py for both tracers (Figures 2G and 2H). TurboRFP consistently yielded higher axon counts (∼8-fold increase) than BDA (Figures 2I and 2J). These standardized axon counts served as an internal reference for downstream analyses of spinal projections in both injured and uninjured animals. Collectively, these results demonstrated that AAV1-hSyn-TurboRFP provides a simpler, more efficient, and more reproducible approach for labeling corticospinal projections than BDA, enabling robust within-animal normalization and improved quantification across experiments. TurboRFP-labeled CSNs show robust pS6 upregulation following Cre-mediated Pten deletion The study of axonal sprouting after pyramidotomy commonly employs Pten deletion as a positive control, as this manipulation has been well established to enhance axon regeneration and sprouting after injury.7,23 Conditional deletion of Pten, a negative regulator of the mTOR pathway, in CSNs leads to robust activation of mTORC1 signaling, which can be detected by increased pS6 levels.7,24 To induce Pten deletion, we injected AAV2-CAG-Cre into the forelimb region of the sensorimotor cortex of Ptenfl/fl mice, with AAV2-CAG-GFP as a control (Figures 3A and 3B). Cre- or GFP-expressing vectors were co-injected with AAV1-hSyn-TurboRFP. Four weeks later, BDA was injected at the same cortical sites, and tissues were collected two weeks later for histological analysis (Figure 3B). TurboRFP labeling and pS6 upregulation after AAV-Cre-mediated Pten deletion (A) Schematic of cortical co-injection of AAV1-hSyn-TurboRFP with AAV2-CAG-GFP or AAV2-CAG-Cre in Ptenfl/fl mice, followed by BDA injection. (B) Experimental timeline. (C) Representative coronal sections showing pS6 immunoreactivity in control and Pten-deleted hemispheres. (D) Quantification of pS6 fluorescence intensity, expressed as the injected/uninjected hemisphere ratio. Data are mean ± SEM; control, n = 7; Pten-deleted animals, n = 8; 3 sections per animal; unpaired t test. (E) Co-localization of pS6 with TurboRFP or BDA labeling in the injected cortex. Arrowheads indicate double-positive neurons. To enable co-injection, the volume of AAV1-hSyn-TurboRFP was optimized by testing multiple volumes (400, 200, 100, and 50 nL) and quantifying labeled CST axons at the level of the Py. Although no statistically significant differences were observed, a clear trend toward increased labeling with higher volumes was noted; 200 nL was, therefore, selected as a compromise between the labeling efficiency and injection volume for co-injections (Figure S2). As expected, pS6 immunoreactivity was strongly upregulated in the Cre-injected hemispheres (Figure 3C), and quantitative analysis confirmed increased pS6 signal relative to the contralateral side, consistent with activation of the mTORC1 pathway in CSNs (Figure 3D). Notably, pS6 upregulation closely overlapped with TurboRFP-labeled cortical regions and neurons, indicating efficient co-targeting of the same neuronal population for both genetic manipulation and neuronal labeling (Figure 3E). In contrast, BDA labeled a smaller subset of cortical neurons with limited overlap with pS6-positive cells. These findings demonstrated that Cre-mediated Pten deletion and TurboRFP labeling target the same and broader population of CSNs, supporting this approach as a unified platform for simultaneous genetic manipulation and axonal tracing. TurboRFP reliably detects enhanced CST sprouting in Pten-deleted mice after unilateral pyramidotomy To evaluate the ability of TurboRFP to trace and detect injury-induced CST plasticity, we applied our tracing strategy to a unilateral pyramidotomy model, a well-established paradigm for assessing axon sprouting from the spared CST.25 In this model, one Py is transected, leaving the contralateral CST intact to serve as a substrate for compensatory sprouting toward the denervated cervical spinal cord (Figure 4B). All animals were assessed for lesion completeness by immunostaining for PKCγ in the cervical spinal cord; animals exhibiting residual CST immunoreactivity were excluded (Figure S3). TurboRFP and BDA tracing of CST axons in Pten-deleted mice (A) Representative C5–C7 spinal cord sections from Ptenfl/fl mice showing TurboRFP-labeled CST terminals after cortical co-injection of AAV2-CAG-GFP or AAV2-CAG-Cre, with higher-magnification views of the denervated side. (B) Schematic of cortical injection sites, unilateral pyramidotomy, and cervical CST sprouting analysis. (C) Experimental timeline. (D) Quantification of the sprouting axon density index, calculated as contralateral/ipsilateral axon density within the gray matter. Data are mean ± SEM; n = 8 animals per group, 5–6 sections per animal; unpaired t test. (E) Quantification of the sprouting axon number index across Mid, Z1, and Z2 regions in the denervated gray matter. Data are mean ± SEM; n = 8 animals per group, 5–6 sections per animal; two-way ANOVA with Šidák’s multiple-comparisons test. See also Figures S3–S6. Conditional Pten deletion was induced in layer V CSNs of Ptenfl/fl mice via cortical injection of AAV2-CAG-Cre, whereas control animals received AAV2-CAG-GFP. Consistent with previous studies, Pten loss robustly enhances CST regeneration and axon sprouting after injury.7 To directly compare the tracing performance, both TurboRFP and BDA were used to label CST axons within the same animals (Figure 4C). Importantly, Pten deletion did not alter the total number of labeled axons at the level of the Py, indicating comparable labeling efficiency across groups (Figure S2). As expected, only limited spontaneous sprouting was observed in control animals, whereas AAV-Cre-injected mice exhibited a marked increase in CST collateralization within the denervated side of the cervical spinal cord (Figure 4A). TurboRFP labeling produced strong fluorescence signals that clearly delineated fine axonal branches, enabling robust visualization of sprouting axons. Although our cortical injections were targeted to the forelimb sensorimotor cortex to preferentially label cervical-projecting CST neurons, TurboRFP-positive CST axons and collateral fibers were also detectable at thoracic and lumbar spinal cord levels without immunostaining or signal amplification (Figures S4A and S4B). This likely reflected the spread of AAV around the injection site, together with overlapping coordinates between the forelimb and hindlimb cortical areas.23,26 As expected, the labeling intensity was reduced at more caudal spinal cord levels relative to the cervical cord. These data indicate that TurboRFP can also be effective for tracing distal CST axons and collateral sprouting. In contrast, BDA signal, detected using fluorescently conjugated streptavidin to allow a more direct comparison with native TurboRFP fluorescence, did not reliably label collateral and sprouting axons and was largely restricted to the main CST tract. Only sparse BDA+ fibers were observed in Pten-deleted animals (Figure S5, arrowheads) and appeared more abundant than in controls, but these were not quantified due to the low overall signal intensity. High-magnification views (Mid, Z1, and Z2) revealed numerous TurboRFP-labeled axons extending throughout the denervated gray matter (Figure 4A). Quantitative analysis of TurboRFP images revealed a significant increase in both the sprouting axon density index (Figure 4D) and the sprouting axon number index across all mediolateral regions examined (Figure 4E). The sprouting axon density index was defined as the CST axon density in the gray matter of the contralateral (denervated) side normalized to the ipsilateral (intact) side, whereas the sprouting axon number index represents the number of CST axons within three predefined regions (Mid, Z1, and Z2), normalized to the total CST axon counts in the medulla. Together, these results demonstrated that TurboRFP tracing enables rapid, high-contrast, and quantitative visualization of CST axon sprouting, with improved sensitivity and reliability compared with conventional BDA labeling. Notably, because Pten deletion may also enhance sprouting on the intact side, normalization to the ipsilateral (intact) side (as in sprouting axon density index) underestimates the overall magnitude of the sprouting effect. FeatureJ-based quantification enables standardized analysis of CST sprouting To enable objective and efficient quantification of CST sprouting after unilateral pyramidotomy, we implemented a semi-automated image-processing workflow in Fiji, using the FeatureJ plugin, which enhances the detection of tubular structures such as axons and facilitates sensitive identification of CST collaterals and sprouting fibers.27 Images of TurboRFP-labeled CST terminals in the cervical spinal cord were first subjected to background subtraction (Figure S6A), followed by application of the FeatureJ smallest Hessian eigenvalue filter (smoothing scale = 2) to enhance axon-like structures, while minimizing background signal (Figure S6B). The smoothing scale was empirically optimized using representative sections and then applied consistently across sections and groups to reduce user-dependent variability. Overexposed signals within the main CST were suppressed during this process, improving selective detection of collateral and sprouting axons (Figure S6C). Processed images were then thresholded and binarized to generate segmented masks for quantitative analysis (Figure S6D). From these binary images, two complementary metrics were derived. The sprouting axon density index was defined as the ratio of total axonal signal within the contralateral (denervated) gray matter relative to the ipsilateral (intact) side (Figure S6D). The sprouting axon number index quantified axonal intersections within a narrowly defined, line-shaped region of interest (ROI) positioned along the denervated gray matter (Mid, Z1, and Z2), normalized to total CST axon counts in the medulla (Figure S6E). This semi-automated pipeline enables reproducible quantification of CST sprouting across animals and conditions. Discussion In this study, we show that AAV1-hSyn-TurboRFP enables robust CST labeling and sprouting analysis after unilateral pyramidotomy without immunostaining or signal amplification. Combined with a standardized quantification workflow, this approach reduces tissue-processing complexity and supports reproducible assessment of CST remodeling after injury. Robust labeling of CSNs and CST axons using native fluorescence Vector injection into forelimb-targeting motor cortex robustly labeled layer V cortical neurons including CSNs, without the need for additional detection steps. The strong native fluorescence signal is in line with other AAV-based tracers, such as in a study with AAV-EGFP.11 However, robust labeling of CST axons, including fine collaterals, with AAV-mediated fluorescent protein expression often requires immunostaining, such as in the case of AAV-ChR2-mCherry and AAV-ChR2-YFP,12 or a GFP transgene expressed under the promoter of mu-crystallin (crym-GFP).28 Notably, TurboRFP has been employed in a number of studies to visualize neuronal processes in vitro29,30,31 and long-range axons in the brain and spinal cord,32,33,34,35,36,37,38 which underscores its reliability in tracing pathways that span large CNS distances. Specifically, AAV1-TurboRFP has been used to label regenerating CST axons but required immunostaining.36 It has also been used in an intersectional labeling strategy to identify specific CST populations, detected by native fluorescence.34 Together, these findings support that TurboRFP can provide sufficiently strong native signal for CST tracing under appropriate conditions. Advantages of TurboRFP over BDA for CST tracing and sprouting analysis BDA injections performed in the same animals labeled substantially fewer cortical neurons and subsequently fewer CST axons (Figures 2G–2J). Although amplification-based BDA detection can improve sensitivity, it adds processing steps and staining-dependent variability. Consistent with this, the BDA labeling efficiency can exhibit substantial variability, even within the same laboratory over time.23,26 BDA tracing also often requires a second cortical injection shortly before tissue collection in longer sprouting paradigms, which may reduce labeling efficiency when prior cortical injections have already been performed. In contrast, TurboRFP can be co-injected with genetic manipulation vectors and provides sustained labeling throughout the experiment. With native fluorescence, TurboRFP labeling captured a large number of CST axons, including thin or newly sprouting collateral fibers. This represents a major advantage for studies aiming to quantify axonal growth and response after injury, as these fine processes are often the most informative readout of plasticity. In this context, the key issue is not simply whether AAV-based tracers can label CST axons, but whether there is a practical and reproducible approach for reliably detecting sparse and fine sprouting CST axons within the spinal cord after injury. Prior studies comparing AAV-based tracers and BDA have primarily focused on the cortical transduction efficiency and anatomical projection mapping within the brain.39,40 Although these studies have demonstrated broad neuronal labeling with AAV-based approaches, they did not address the practical challenge of reliably detecting sparse CST collateral and sprouting axons within the spinal cord after injury. Similarly, while TurboRFP has previously been used to label CST axons,34 that study did not establish or evaluate a simplified workflow for quantitative analysis of CST sprouting. By combining amplification-free TurboRFP labeling with a streamlined quantitative analysis pipeline, the current workflow enables sensitive detection and reproducible quantification of CST sprouting after injury. Efficient viral delivery and compatibility with genetic manipulation The workflow is also compatible with genetic manipulation. TurboRFP produced robust CST labeling across reduced injection volumes, supporting co-injection with Cre- or GFP-expressing AAVs. This avoids additional BDA surgery and helps align the labeled and genetically manipulated neuronal populations, which is important for interpreting circuit remodeling. In Pten-deleted mice, this signal-to-noise ratio enabled detection of increased CST sprouting after pyramidotomy. Standardized and reproducible quantification Coupling TurboRFP tracing with a semi-automated FeatureJ pipeline enabled standardized quantification of cervical CST sprouting (Figure S6). Hessian-based filtering enhanced axon-like structures, and consistent application of empirically optimized parameters reduced user-dependent variability.27 Datasets with substantially different labeling patterns, such as sparse long-distance regeneration studies, may require modest adjustment of filtering or thresholding parameters. Conclusion and outlook In summary, this study establishes AAV1-hSyn-TurboRFP as a practical platform for CST sprouting analysis after unilateral pyramidotomy. Its main contribution is a simplified workflow that combines native fluorescence, compatibility with genetic manipulation, and standardized quantification to support reproducible assessment of corticospinal remodeling after injury. An expanding list of axonal tracers is emerging, such as codon-optimized membrane-embedded tracers (COMETs) built on existing fluorescent proteins,41 which will further accelerate axon tracing and neural repair studies. Limitations of the study Although effective for detecting collateral sprouting after unilateral pyramidotomy, this workflow has not yet been fully validated for true CST regeneration after SCI. In such paradigms, pre-labeling before injury could complicate interpretation because degenerating distal axons may persist for weeks and resemble regenerating fibers.42 Adapting AAV-TurboRFP to regeneration studies may, therefore, require optimization of viral delivery timing and post-injury survival intervals. We also did not assess TurboRFP performance in very long-term experiments, and variability across experimental paradigms cannot be excluded. Native fluorescence was robust in fixed tissue but appeared less reliable in unfixed ex vivo preparations in our hands (not shown). Finally, co-injection of different AAV serotypes may lead to incomplete overlap between the manipulated and labeled CSNs because serotypes differ in tropism and axonal transport.9,16,43 Using the same serotype for reporter and functional constructs, when compatible with the experimental design, may improve co-transduction precision. Resource availability Lead contact Requests for further information, resources, and reagents should be directed to and will be fulfilled by the lead contact, Binhai Zheng (bizheng@health.ucsd.edu). Materials availability This study did not generate new unique reagents. All commercially available reagents and viral vectors used in this study are listed in the key resources table. Data and code availability •All data reported in this paper will be shared by the lead contact upon request. All data reported in this paper will be shared by the lead contact upon request. •This paper does not report original code. This paper does not report original code. •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.