SUMMARY
The primate brain possesses unique physiological and developmental features, yet its systematic investigation has been hampered by a paucity of transgenic germline models and tools. Here, we present a minimally invasive method to introduce transgenes widely across the primate cerebral cortex using ultrasound-guided fetal intracerebroventricular viral injections (FIVIs). FIVI enables efficient and long-lasting transgene expression following intrauterine delivery of recombinant adeno-associated viruses (rAAVs). In the marmoset, we demonstrate that adjusting gestational timing, rAAV serotype, and transcriptional regulatory elements enables selective targeting of defined cell populations, including layer-restricted labeling and Cre-dependent intersectional access. Pilot experiments in rats further demonstrate the potential of FIVIs for prenatal CRISPR-based gene editing and labeling of peripheral somatosensory and retinal pathways. By mimicking key desirable features of germline transgenic models, this efficient and targeted method for gene transfer into the fetal primate brain expands the experimental opportunities for basic and translational neuroscience research across the lifespan.
In brief
Ribeiro Gomes et al. develop a minimally invasive, ultrasound-guided method for delivering genes into fetal marmosets. Gestational timing, viral serotype, and upstream gene regulation afford control over expression in target cell populations that persists into adulthood.
Graphical Abstract
INTRODUCTION
The abundant use of genetically modified mouse lines has transformed the study of brain anatomy and function by enabling the systematic examination of molecules or circuit elements under genetic control1–4 Examples of analogous germline modification of rats,5,6 ferrets,7,8 and nonhuman primates9–14 have begun to broaden the repertoire of mammalian experimental and disease models. Among these, developing nonhuman primate genetic models is of particular importance for understanding primate-specific features and disorders of the human brain.15 However, the establishment of such models faces significant hurdles.16,17 For one, generating transgenic primates is costly and labor-intensive, requiring maintenance of sufficiently large colonies for efficient breeding. Furthermore, while molecular tools and genetic constructs are advancing rapidly, the long gestational and prepubertal periods of primates require many years of effort to establish a given genetic line. As a result, most emphasis has been placed on developing transgenic primate disease models,14,17,18 with fewer resources devoted to creating genetic lines for basic neuroscience research. Important research areas such as primate neurodevelopment, which stand to benefit from such tools and are highly relevant to human disease, remain largely unexplored. Thus, there is a need to consider complementary approaches to germline modification to harness the power of genetic methods in nonhuman primates for basic and translational neuroscience.
An alternative to germline modification is the somatic introduction of transgenes into a living organism. In experimental neuroscience, viral tools have been routinely used for gene delivery, either by inserting genetic material into the host chromosome or by creating persistent DNA concatemers in the nucleus (episomes) that support transcription.19 For systems neuroscience, transduction from most in vivo viral delivery methods is localized, confined to the tissue immediately surrounding an injection site. However, recent advances in viral vector administration have enabled broader coverage across the central nervous system.20 Notably, delivering recombinant adeno-associated viruses (rAAVs) through the cerebrospinal fluid (CSF) increases the breadth of transduction, particularly when the vector is infused in developing animals. Commonly used in mice, rAAV delivery into the CSF is most effective for achieving broad transgene expression within a brief postnatal window of a few days, after which transduction becomes less efficient and more localized.21 Given the experimental and biomedical opportunities potentially afforded by widespread gene transfer in the primate brain, we sought to translate this cerebroventricular injection approach to the marmoset monkey. Because the corresponding developmental stages in marmosets and other primates occur in utero,22 we designed a method to achieve gene transfer before birth, using ultrasound (US)-guided targeting to inject viral particles into the cerebroventricular system of fetal brains.
The present study describes a minimally invasive fetal intracerebroventricular viral injection (FIVI) procedure, initially developed in rats (Rattus norvegicus) and adapted to marmosets (Callithrix jacchus), to deliver rAAVs at early stages of brain development to achieve broad gene transfer across the primate cerebral cortex. We show that a single injection into the fetal CSF results in rapid, robust, widespread, and enduring transgene expression throughout the brain. By varying the gestational timing of injections and the rAAV serotypes, we demonstrate the systematic targeting of specific cell populations, thus opening the door to new modes of circuit-based study of the primate brain. We further describe experimental results, highlighting several areas of opportunity, including the use of FIVI for intersectional methods using Cre recombinase, CRISPR-based gene editing, and the capacity for gene transfer into sensory pathways outside the brain. We conclude that the FIVI method is straightforward and versatile, offering new opportunities to label developmentally defined cell populations in nonhuman primates for a broad range of experimental topics for systems and translational neuroscience research.
RESULTS
US-guided transabdominal injections enable targeted viral delivery into the fetal brain
To efficiently deliver material into the fetal brain, we developed an adjustable injection system consisting of an animal cradle, a US transducer, and independent micromanipulator arms to introduce a penetrating transcutaneous guide tube (Figure 1A). This apparatus was designed with multiple degrees of adjustment to accommodate the unpredictable orientation of the fetal cranium inside the uterus and to allow real-time US visual feedback and realignment of the injection path.
Figure 1. Fetal transduction through ultrasound-guided intracerebroventricular viral injection.
(A) A schematic of the fetal intracerebroventricular viral injection (FIVI) experimental setup and injection apparatus, highlighting its main components, designed for both rats and marmosets. The ultrasound machine, geared head, and translational stages permit real-time visualization for precise fetal orientation and injection (see STAR Methods and Figure S1 for details). The illustration is by NIH Medical Arts.
(B) Using high-frequency ultrasound linear transducers, the needle can be visualized advancing through a marmoset fetal skull at PC107 (left) and into the brain (right), with delivery of the viral particles into the midline ventricles (see also Video S3). The position of the guide tube tip, which is not visible due to low echogenicity, is marked by a green arrowhead, whereas the needle tip is indicated by a red arrowhead. lv, lateral ventricle; 3v, third ventricle. Scale bar: 5 mm.
(C) Timeline of experiments in rats and marmosets. FIVI was successfully applied across a broad range of gestational stages, involving significant variation in skull thickness and ventricle volume. ICV, intracerebroventricular; PC, postconception day. This image was created in BioRender.
(D) Reporter expression was visualized in vivo during the perinatal period. In rats (left), reporter gene expression (GFP, closed arrowhead; tdTomato, open arrowhead) is visible in the nervous system through the skull and skin after PC19 injections, with additional expression in muscle. As fur grows and skin thickens, visibility decreases under a flashlight, with expression most prominent in fur-free or thin-skinned areas, such as around the eyes and snout, as seen in a PC55 rat injected at PC19 (inset). In neonatal marmosets (postnatal day 0), following PC96 (middle) and PC116 (right) injections, strong reporter expression can be visualized inside the mouth and around the eyes. Fainter expression can also be observed in the body (e.g., neck; data not shown), though visualization is more challenging due to the darker skin and fur present at birth and varies depending on the injection date and serotype combination.
Details on the mechanical setup and step-by-step instructions on the procedure, including animal preparation and anesthesia, are given in the STAR Methods and Figure S1. Briefly, following isoflurane anesthesia induction, the pregnant female is positioned in the cradle in a supine position (Figure 1A). An aseptic environment is created over the shaved abdominal surface, which is covered with sterile US gel. Prior to positioning the transcutaneous guide tube, the experimenter initially determines the location and orientation of the fetus by adjusting the US transducer and observing the corresponding images on the display. The guide tube is supported over the abdomen by a custom micromanipulator arm. It is angled within the US plane, just underneath the edge of the transducer, which is supported by a separate micromanipulator arm (Figure S1D). This positioning allows for visualization of the guide tube within the imaging plane, facilitating precise alignment with the fetal target. The guide tube is thus visible as it is advanced slowly through the gel and pushed gently against the skin. Penetration through the skin is achieved using brief electrocautery pulses, applied exclusively to the uninsulated tip of the guide tube, while the body of the tube remains electrically insulated. A few briefly applied current pulses allow the guide tube tip to advance smoothly through the thick skin and abdominal wall without displacement of the fetus (see STAR Methods and Videos S1 and S2 for details). This straightforward step greatly simplifies and shortens the FIVI procedure, eliminating the need for skin incision or surgical exposure.
With the transcutaneous guide tube held firmly in position, the injection needle is manually introduced to the target by the experimenter under US guidance (Figure 1B). After the Hamilton syringe needle is introduced into the opening of the guide tube, it is advanced through the abdominal muscle and uterine wall into the uterine lumen adjacent to the fetal skull. At this stage, small micromanipulator adjustments are sometimes needed to optimize the positioning of the transducer, guide tube, or animal for the most effective targeting of the cerebral ventricles. Once optimized, the needle is pushed through the cranial wall toward the ventricular target. We found that this step is best achieved by initially applying a small-amplitude mechanical jolt that displaces the needle tip 1–3 mm into the intracranial space and subsequently advancing the needle into the ventricle (Video S3). Following this US-guided positioning of the needle tip, the viral particles are infused over 1–2 min. The needle is left in place for 2–3 min to allow fluid dispersion before retraction. We injected volumes of 5–10 μL in the rat and 10–60 μL in the marmoset. For both rats (typically targeting 6–9 fetuses) and marmosets (typically targeting 2–3 fetuses), the multiple-injection procedure usually requires between 1.5 and 2 h of isoflurane anesthesia, followed by the rapid recovery of the dam.
We successfully applied the FIVI procedure over a broad range of gestational ages (Figure 1C), during which the brain undergoes significant growth and morphological changes. In the marmoset, injections were carried out as early as postconception day (PC)62 (Videos S4 and S5) and as late as PC121. Similarly, in the rat, injections were carried out in fetuses between PC13 (Videos S6 and S7) and PC21, 2 days before birth. In the earliest injections, the ventricular lumen is more expanded, making it harder to precisely target a specific ventricular compartment. Therefore, we primarily targeted vesicles at the cranial end. For the later injections, the US-guided procedure allowed more selective targeting of the lateral or third ventricles. Importantly, the capacity to inject at different developmental stages allows for the targeted introduction of viral vectors into specific cell populations as they emerge in the developing brain.
Widespread rAAV-mediated transgene expression following FIVI delivery
In this section, we describe the widespread viral infection resulting from the FIVI method, revealed through the expression of fluorescent reporter genes (e.g., GFP and tdTomato). Unless stated otherwise, the experiments in this study used the rAAV9 serotype, which has proven effective for injections into developing animals, including the primate fetal brain,23,24 with reporter expression driven by the high-efficiency CAG promoter (derived from cytomegalovirus, chicken β-actin, and rabbit β-globin genes).
Following injections in rats, it was often possible to determine postnatally which of the pups in the litter received fetal injections by observing reporter fluorescence using a spectrally appropriate flashlight and matching barrier filter glasses (Figure 1D). In vivo fluorescence, evident in newborn rats across all gestational injection ages, was visible throughout the head and neck, sometimes extending caudally, and included extracranial labeling of muscle tissue and neurons of the peripheral nervous system. In the marmoset, postnatal in vivo evaluation was sometimes possible, though more challenging owing to the pigmentation of the skin, the presence of hair, and the more advanced developmental state at the time of birth relative to rats. When observed, fluorescence in newborn marmosets was most prominent in hairless regions, particularly in the face, around the eyes or in the ears, and inside the mouth (Figure 1D). Outside the nervous system, cell transduction was sometimes observed in skeletal muscle near the presumed injection path in the fetal head, with very sparse cell transduction in the liver and negligible transduction in the skin and kidney (Figure S2).
We used histological methods to evaluate the efficiency and cell-type specificity of reporter gene expression in the brain. In general, rAAV9 led to widespread labeling across the cerebral hemispheres. Figure 2 shows histological sections of a marmoset injected at PC87 (tissue collected at postnatal day 3) with rAAV9-CAG-GFP, with robust labeling of neurons in lower layers of the cerebral cortex. The transduced neurons were confined to these layers, consistent with the relatively early gestational age of injection (discussed in more detail below). In this example, transduced cells across regions were distributed within similar layers, although labeling density varied across cortical depth, with up to 80% of neurons labeled at selected cortical depths. Despite depth-dependent variations in labeling density, a prominent feature of transduction was labeling within a continuous laminar band across the cortical sheet, a pattern observed across cases (see Figure S3A for additional examples). It is worth noting that while cells in the cerebral cortex were well labeled with this serotype and promoter at PC87, thalamic cells were not (Figure 2H). The strong fluorescence signal observed in the thalamus likely reflects labeled axons originating from neurons in infragranular cortical layers (see Figures 2F–2I for examples of labeling in non-cortical regions).
Figure 2. Extensive cortical transduction following FIVI.
(A) Representative coronal sections in a 3-day-old marmoset show the widespread transduction pattern following a PC87 injection of AAV9-CAG-GFP. The side view of the extracted central nervous system highlights broad GFP labeling. Across all regions, the most abundant labeling is in the lower layers.
(B–E) Detailed view of the insets shown in (A), highlighting the labeling pattern in the temporal, prefrontal, frontal, and parietal regions. Most labeled cells in the cortex are neurons, based on morphology and co-expression of NeuN (B, white arrowheads in yellow inset). Labeling in more superficial layers is almost entirely restricted to apical dendrites from lower-layer neurons (B, pink inset). The density of transduced neurons varied across cortical depths (dashed boxes in B–E). GFP+/NeuN+ cells were located in similar layers across regions, but the proportion of transduced neurons varied with depth, with up to 80% of cells being transduced at certain depths.
(F–I) Cell labeling was observed in additional structures, such as the olfactory bulbs (OBs; A-1), striatum (Str; A-3, A-4, and F), hippocampus and choroid plexus (Hipp and CP, respectively; A-5 and G), and cerebellum (A-7 and I). Strong fiber labeling was observed in the thalamus (Th; A-5 and H) and white matter paths, such as corticofugal pathways within the internal capsule (F).
Scale bars: (A) 1 mm, (B–H) 500 μm, and (I) and insets in (G) and (H), 250 μm. FIVI volume: 60 μL.
In rats, the pattern of labeling was comparable to that of the marmoset. For example, we found similar infragranular labeling (albeit with some additional supragranular cortical neurons) when we injected the same serotype and construct at PC19 (Figure S4A). To evaluate the reliability and reproducibility of this method, we performed the same injection procedure in rat fetuses in 24 sessions using the rAAV9-CAG construct, with injections carried out at PC19 and euthanasia at PC44 (~21 postnatal days). While some experimental variables, such as the angle of approach to the ventricular system, necessarily varied with each fetus, we fixed other variables. For example, in the PC19 rat, we always injected 10 μL of the virus and aimed at the lateral ventricle or midline third ventricle. Although the specific patterns of labeling are subject to the idiosyncrasies of individual injections—for instance, some cases exhibited bilateral symmetrical transduction, while others showed more pronounced expression in one hemisphere—this method reliably led to widespread cortical transduction in multiple fetuses across sessions (Figures S4B and S4C).
In rats, the expression of reporter genes was first observed within a few days of the procedure (the shortest interval tested was 2 days) and persisted for at least several months (Figure S4D), suggesting that this method can be applied to longitudinal neuroscience experiments lasting into adulthood. In marmosets, transgene expression was observed in fetal tissue 9 days after the injection (Figure S5). While the maximum duration of transgene expression in marmosets remains unknown, strong expression was observed in adult marmosets’ cortex and non-cortical structures more than 2 years post-injection (Figure 3), with no overt neurological or behavioral deficits linked to long-term expression observed in transduced animals.
Figure 3. Long-term transgene expression in an adult marmoset following PC100 FIVI.
Representative section of an adult marmoset brain 27 months after FIVI procedure. Following transduction at PC100 with AAV8-hSyn-ChRmine-mScarlet-Kv2.1-WPRE, native fluorescent expression remained strong. The sustained transgene expression indicates that the FIVI approach can achieve long-term expression suitable for chronic primate studies, including longitudinal studies involving optogenetic stimulation and fluorescent microscopy using genetically encoded indicators. Scale bars: green and white insets, 500 μm and pink and blue insets, 50 μm. FIVI volume: 60 μL.
Strategies to refine targeting of cell subpopulations
Given the pattern of labeling described above, we investigated factors that might enable targeting different cell subpopulations. One important factor was the gestational timing of the rAAV injection. We compared the transduction patterns of rAAV9 FIVI across gestational ages in the marmoset (see Figures 4A, S3B, and S3C for sequential FIVIs in the same animal). We found that the laminar distribution of labeled cells closely followed the injection timing. Earlier injections labeled deeper layers, while later injections sequentially labeled more superficial layers, consistent with the known inside-out pattern of cortical neuron development (Figure 4B). Thus, adjusting the timing of the FIVI procedure provides a systematic way to target cell populations in the cortex.
Figure 4. Effects of varying the timing of postconceptional ICV injections.
(A) Exemplar perinatal pattern of transduction in the occipital, parietal, and prefrontal cortices following rAAV9 FIVI. While there are regional differences, a general trend emerges: earlier injections predominantly label deeper layers, whereas later injections label more superficial layers. For the injection dates featured here, a systematic laminar trend was most conspicuous in the occipital cortex, where labeling shifted from white matter/layer 6B at PC87 to layer 5/6A at PC99 and to layers 4 and 2/3 at PC114. In parietal and prefrontal areas, the PC114 injection led to transduction across all layers, with neurons intermingled with non-neuronal cells (arrowhead, non-neuronal cell in B histological inset), consistent with an earlier transition from neurogenesis to gliogenesis in these regions. Tissue collection: PC99 postnatal day (P)2, PC87 P3, and PC114 P0.
(B) Transduction patterns with rAAV9 injections follow the order of cortical histogenesis. The image was partially created in BioRender.
Scale bars: (A) 5500 μm and (B) 50 μm. FIVI volume: 60 μL.
The widespread, timing-dependent infection of neurons is influenced by the known asynchrony of laminar development across the different regions of the cortical mantle.25 This asynchrony results in moderate laminar differences in occipital, parietal, and prefrontal regions (Figure 4A). For example, the PC99 injection led to labeling of infragranular neurons in the occipital cortex but primarily layer 4 neurons in the parietal cortex, with minimal infragranular labeling. Further, the PC114 injection exclusively labeled neurons in the occipital cortex but labeled both neuronal and non-neuronal cells in parietal and frontal areas, suggesting an earlier shift from neurogenesis to gliogenesis in the latter. These interareal differences most likely stem from the known rostro-caudal gradient in neurogenesis timing across the cortex.26–29
Together, these observations indicate that varying the injection timing within a 30-day interval provides a means to achieve selective laminar transduction across the marmoset cerebral cortex. Furthermore, they highlight a potential principle underlying fetal rAAV9 infection in primates, which is that transduction only affects cortical cells born within a window of time relative to the injection procedure. In other words, neurons born after or before this time window do not express the transgene. The absence of labeling in neurons superficial to the densely labeled region is straightforward to understand, as these neurons may not have been born or are insufficiently mature for rAAV transduction at the time of the injection. However, the relative absence of deeper labeling is more puzzling. This observation suggests that, following a period in which neurons are receptive to rAAV transduction, they transition back to a non-permissive state. While the reason for this transition is unknown, it may stem from the physical position occupied by those radially migrating neurons at the time of injection. Alternatively, it may be due to a change in the cell surface molecules associated with normal maturation, such as the specific membrane glycoproteins known to act as co-receptors for rAAV infection.30
Thus, one possibility is that the selective transduction of neurons based on maturation stage derives from viral tropism, where only neurons at a given maturational stage are transduced by a given rAAV serotype expressing particular capsid proteins. To investigate this possibility, we compared the patterns of infection with two different serotypes, rAAV2 and rAAV9. Pilot studies have revealed these two serotypes to exhibit different laminar expression patterns in rats when injected at the same gestational age (Figure S4E). When rAAV2 and rAAV9 were injected into marmosets at comparable gestational ages using the FIVI method, we observed a pronounced serotype-specific laminar segregation across the cortex (Figure 5). Namely, rAAV9 labeled the earlier-born, deep-layer neurons, whereas rAAV2 labeled the later-born, superficial-layer neurons. This observed serotype-driven laminar transduction is consistent with an interaction between the intrinsic tropism of the virus and the developmental state of the cells.
Finally, for neurodevelopmental studies, the FIVI methodology provides a powerful tool to observe and manipulate cells as their identity is established and cell diversity emerges.31 For example, by employing rAAV intersectional strategies in marmosets, FIVI can be used to tag cells based on the transient prenatal expression of a specific gene. We applied this rationale to the Nestin gene, which encodes an intermediate filament protein expressed in neural stem cells as well as subsets of non-neural stem cells.32 It is also known that Nestin expression progressively declines shortly after cells differentiate.33 Using FIVI procedures at different gestational ages in two marmosets, we co-injected two viral constructs to tag neurons expressing Nestin at or after the injection procedure (Figure S6A). The first vector (AAV9-NestinTK-EGFP-iCre) expressed Cre recombinase under the control of a Nestin-specific enhancer in undifferentiated cells.34 The second vector (AAV8-nEF-Con/Foff 2.0-ChRmine-oScarlet), injected at the same time, led to the Cre-dependent expression of the fluorescent reporter oScarlet in Cre-expressing cells. With this combination, it was possible to identify cells that exhibited Nestin expression after either PC105 (Figures S6B and S6C) or PC86 (Figure S6D). This proof-of-concept intersectional experiment demonstrates the potential power of the FIVI method to identify and study developmentally defined neural subpopulations across the primate brain.
Prenatal gene editing using the CRISPR-Cas9 toolbox
The precise editing of a cell’s genome is an important advance in biology, with enormous prospects for human medicine.35 For somatic tissue, including the embryonic brain, gene editing can be achieved using the CRISPR-Cas9 toolbox delivered by rAAVs. To this end, we modified constructs previously validated in mice36 to insert the reporter EGFP into the rat Actin gene to create an EGFP-Actin fusion protein through CRISPR-Cas9-mediated homology-directed repair. This gene editing in rats was achieved by co-injecting a vector expressing SpCas9 (AAV9-EFS-SpCas9) along with a vector expressing a guide RNA targeting the start codon of rat Actin, as well as carrying a repair template (the coding region of GFP flanked by homologous arms specific for sequences flanking the CRISPR-induced double-strand break) (AAV9-rActin-EGFP-donor) (Figure 5A). Applying this two-virus approach resulted in EGFP expression among a sparse subset of neurons throughout the brain following FIVI at PC19 (Figure 5B). These neurons exhibited EGFP signals localized in known Actin-rich regions, such as the dendritic spines (Figures 5B, panels d and e). EGFP-expressing cells were also observed in the choroid plexus (Figure 5B, panel f), a tissue susceptible to rAAV infection and a potential source for widespread delivery of biotherapies to the brain, as well as in other regions amenable to rAAV9 serotype transduction at PC19.
Figure 5. Viral tropism-mediated selective transduction of the fetal brain.
Representative perinatal transduction patterns in the prefrontal, temporal, parietal, and occipital cortices of the marmoset following rAAV2 and rAAV9 FIVIs at similar gestation times. As in rats (Figure S4E), the serotypes exhibit complementary transduction patterns, with rAAV2 labeling neurons in more superficial layers compared to rAAV9. Notably, as shown in Figure 3, the transduction patterns vary across regions. Tissue collection: rAAV2 PC86 at P2 and rAAV9 PC87 at P3.
Scale bars: 500 μm. FIVI volumes: 60 μL.
Prenatal transduction of the peripheral nervous system and sensory pathways
Beyond labeling cells in the central nervous system, the FIVI approach can be used to transduce peripheral neurons and sensory pathways, as we illustrate with two examples, both of which use rAAV9 and a fluorescent reporter under the control of CAG. The first example, shown in Figures 5C and 5D, entails the selective labeling of retinal ganglion cells (RGCs) in rats at PC44 following US-guided targeting of the eye socket at PC19 with rAAV9-CAG-GFP. A single ocular injection led to prominent transduction of RGC projections exiting the eye, highlighting the many visual ganglion cell projection sites in the brain. Combined with more sophisticated viral approaches, this paradigm may be used to study the early development and childhood refinement of retinal pathways, including in nonhuman primates. The second example, shown in Figure 5E, illustrates the rAAV9-CAG-tdTomato labeling in the peripheral nervous system following a PC15 injection. In this case, the fetus was harvested 4 days later, at PC19, and histologically cleared. In the cleared fetus, tdTomato labeling was prominent in the dorsal root ganglia, trigeminal ganglia, and peripheral nerves. Across FIVI experiments, transduction of the spinal cord and sensory neurons was a common observation, in agreement with previous studies.37
DISCUSSION
This study describes a minimally invasive US-guided fetal intracerebroventricular delivery method offering new experimental and translational opportunities for widespread viral gene transfer, particularly in primates. We used the rat model to establish and refine the FIVI setup and procedure, as well as to pilot experiments that were subsequently extended to marmosets or that hold promise for future translation. In this initial study, we focused primarily on rAAV9, as this serotype has been shown to lead to efficient transgene expression following direct injection into the primate brain,20,38 including during early development.24,39,40
Extensive reporter expression following FIVI was prominent in the cerebral cortex. The continuous cortex-wide transduction paves the way for a range of neuroscience experiments facilitated by broader expression of opsins,41 calcium indicators,42,43 or other genetically encoded neuroscience tools. Investigations utilizing these methods, firmly established in mice, can now be translated to primates, particularly the marmoset with its lissencephalic cortex. The robust, long-lasting transgene expression observed in the adult marmoset years after rAAV FIVI (Figure 3) makes this approach suitable for long-term and longitudinal functional studies. Furthermore, since transgene expression starts during fetal development (Figure S5), this emerging combination of methods is well suited for longitudinal studies offering novel prospects for mapping whole-brain circuits and tracking physiological or behavioral changes across the lifespan. Outside the cerebral cortex, other structures were labeled (Figures 2F–2I). The patterns of transduction depended on the specific combination of rAAV serotype and injection timing relative to gestation.
It is useful to compare rAAV delivery with FIVI to other in vivo approaches. Intraparenchymal injections have become a standard method in primate neuroscience.44–46 Typically applied in adults, local injections allow for spatial restriction of gene transfer to regions of interest, due to the relatively short diffusion distance. While this facilitates precise targeting with high levels of transgene expression, it excludes experimental opportunities that come from widespread gene transfer across the entire brain. An emerging approach in adults for broader expression is the intravenous infusion of rAAV9. This strategy was initially based on the capacity of this recombinant serotype47–49 and later enhanced by its capsid-engineered variants50–52 to efficiently cross the blood-brain barrier and infect neural tissue. Successfully applied in nonhuman primates,24,53,54 intravenous delivery methods are conceptually straightforward. With improvements in vector design to decrease off-target transduction and immunological response, these approaches are becoming more commonly used. While holding promise for postnatal transduction, these methods have limited prenatal application, particularly for fetal transduction following infusions in the pregnant dam. To our knowledge, this approach does not lead to fetal transduction, most likely due to the inefficiency of the capsids in crossing the placenta. The minimal liver targeting observed with rAAV9 following FIVIs (Figure S2), compared with its widespread transduction after intravenous injections,54 suggests that there is limited diffusion of the viral vector outside of the nervous system following FIVIs. This indicates that FIVI can be a safe option for gene delivery, particularly when using widely available recombinant rAAV variants that can cause liver toxicity after systemic administration. Refining the constructs to include cell-type-specific promoters (e.g., synapsin) is expected to decrease off-target labeling in muscle and other structures, further enhancing the efficiency and specificity of nervous system transduction with rAAV FIVIs.
In fetal nonhuman primates, injections into the amniotic sac are technically simpler than injections into the cerebral ventricles. However, they typically lead to highly restricted transduction, with no detectable expression in the nervous system.55 rAAV administration through the umbilical cord and electroporation can both be used for broader transduction, but both methods pose significant risks for both the fetus and the dam, as they require major surgical procedures involving the exposure and manipulation of fetuses. Though the outcomes of these procedures can be improved,56–58 risk mitigation requires specialized training and large numbers of animals, thus severely restricting the number of research groups that could take such an approach. The FIVI method offers a more tractable solution to these challenges, as well as a number of unique advantages. Importantly, by enabling safe and flexible delivery of research and therapeutic products using viral tools or other emerging therapeutic vehicles,59,60 it keeps pace with rapid advances in delivery technologies, enabling the rapid evaluation and selection of effective tools for both fundamental and translational research. FIVI will facilitate the investigation of intricate mechanisms underlying neurodevelopment, such as cell history, fate, and specification, while also enabling new opportunities to study interventions during critical windows of prenatal development. This approach offers the potential for long-lasting and efficient genetic modulation, providing insight into the long-term effects of prenatal transduction and yielding information that is particularly valuable for disorders where in utero intervention can lead to improved outcomes.61,62 For restorative gene therapy, direct editing of the chromosomal DNA using methods such as CRISPR may prove most effective.35,62 The transduction of not only neurons but also important cellular targets for gene therapy, such as cells in the choroid plexus,63–65 suggests that the FIVI procedure can be a viable approach for therapeutic applications.
Limitations of the study
It is important to acknowledge the limitations of the current method and potential areas for future improvements. One key challenge in the application of viral vectors is the control over the specific tissues and cells being targeted and transduced. High cellular tropism specificity, with targeting of a narrow range of cell types, can be either desirable or limiting, depending on the intended transduction goal.66 The examples shown in the present study demonstrate both sides of this issue. For the broad entry and transduction of rAAVs across multiple cell types, followed by the restriction of transgene expression based on promoter or enhancer sequences, the tropisms observed with FIVI, such as the incomplete coverage of the cortical mantle, limit the extent to which all cell populations can be uniformly transduced. On the other hand, this method provides an additional strategy for cell selection based on combinations between gestation timing and serotype used, as observed in the complementary laminar patterns of neurons labeled at different times (Figure 4) or by two different serotypes (Figure 5). Significant gaps remain in our understanding of rAAV transduction in primates, particularly regarding the parameters that influence transduction efficiency. For instance, intracortical injections of nine different serotypes (AAV1, -2, -5, -6, -7, -8, -9, -rh10, and -DJ) at moderate titers predominantly resulted in transgene expression in neurons when using the ubiquitous CMV promoter.67 However, a 4-fold increase in titer led to a shift from neuronal to glial transduction.68 This gap in knowledge is especially pronounced in fetal brains. It is clear that the rAAV transduction patterns observed in this study differ significantly from those observed in adults. A possible explanation for this difference may be differences in the composition of rAAV binding sites in target cells during their maturation. Glycans, such as proteoglycans,69 serve as AAV receptors and play a crucial role in various mechanisms underlying nervous system development and maturation. Their expression is tightly regulated both spatially and temporally during development.70 Future studies on the spatial and temporal dynamic changes in the developing glycome71 may enhance our understanding of the host cell-rAAV binding, which may be the origin of differential transduction patterns. This knowledge will be valuable for designing experiments to optimize rAAV transduction based on the cell’s developmental state or for transducing cells at complementary stages (e.g., by combining serotypes), thereby expanding the coverage of gene transfer. Further studies are underway to assess the relationship between birthdate and rAAV transduction and to evaluate the reproducibility of patterns observed following FIVI at the same gestational age across litters. These analyses will help determine whether comparable cell populations can be prospectively targeted across animals based on cell birthdate and the postconception timing of rAAV FIVI administration.
One potential concern with fetal injections is the risk of inducing miscarriage or otherwise affecting pregnancy or fetal development. Because spontaneous abortion is common in marmosets, the reproductive losses observed in our colony were on par with those reported in the literature.72 The miscarriage rate was slightly lower in pregnancies that received FIVIs, and this difference was statistically significant (p = 0.0096, α = 0.01). This may reflect increased monitoring of injected animals relative to non-injected controls. For pregnancies resulting in miscarriage, potential contributing factors such as environmental stressors (e.g., abrupt changes in housing conditions)73 and litter size72 cannot be excluded as independent causes of pregnancy loss. Importantly, in this study, miscarriages occurred infrequently in FIVI pregnancies and did not occur immediately after the procedure.
Another potential concern with fetal injections is the possibility of morphological changes in brain development following FIVI. These could result from the interaction of multiple factors, including the FIVI methodology (e.g., anesthesia duration and intracranial targeting), parameters of rAAV delivery (e.g., needle gauge, injection pressure, dose, and volume), or characteristics specific to the injected subject (e.g., gestation timing and immunological response). We occasionally observed morphological changes (e.g., dilated cerebral ventricles and modifications in the typical sulci and gyri morphology) in the injected brains of rats and marmosets. For instance, enlarged ventricles were observed in 3 of the 12 brains analyzed histologically, 2 in brains that received sequential injections of rAAV (see neonate 3 in Figure S3A for an enlarged ventricle example). These brains also presented variations in the typical sulci and gyri morphology (e.g., an increase in the depth of the posterior occipital sulcus [pos] and the presence of a sulcus in the inferior posterior occipital cortex in neonate 2 in Figure S3A). Whether morphological variations result from the known variability in individual marmoset brain morphologies74–76 or are accentuated by biomechanical stress induced by the procedure or the genetic manipulation itself remains unclear and requires further investigation. Exploring the use of smaller injection volumes scaled proportionally to each fetus may help mitigate these changes by reducing CSF pressure and preserving CSF chemical balance. Investigating the use of lower titers, promoters with lower transcription rates, or inducible viral systems may also be considered if adverse effects of strong and long-term exogenous gene expression on cellular function are suspected. Diversifying delivery methods, such as US-guided retroorbital injections or injections into the fetal circulation, could provide complementary routes for fetal transduction. Optimizing rAAV vectors with increased transduction efficiency in progenitors77 may broaden the extent of transduction and help circumvent the need for repeated rAAV administrations to reach sufficient expression levels (e.g., for therapeutic effect). Further refinement of the methodology could improve targeting consistency, shorten the procedure, decrease anesthesia time, and minimize complications that may arise with longer procedures.
In summary, we presented a novel method for prenatal in vivo delivery of genetic technologies in nonhuman primates, emphasizing critical factors influencing rAAV vector transduction, such as gestational timing and rAAV serotype. The versatility of FIVI, which allows for routine testing and troubleshooting, holds significant potential for establishing best practices and successful protocols for in utero gene transfer. Its broad applicability across species, and particularly in primates, will further support the development of additional quasi-transgenic animal models for experimental and preclinical research.
RESOURCE AVAILABILITY
Lead contact
Requests for further information and resources should be directed to and will be fulfilled by the lead contact, David A. Leopold (leopoldd@mail.nih.gov).
Materials availability
Plasmids generated in this study have been deposited to Addgene: pAAV-NestinTK-EGFP-iCre, unique ID AAV916; pAAV-rActin-EGFP-donor, unique ID AAV942; and pAAV-EFS-SpCas9, unique ID AAV949.
Data and code availability
All data reported in this paper will be shared by the lead contact upon request.
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.
STAR★METHODS
EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS
All procedures were approved by the Animal Care and Use Committee (ACUC) of the Intramural Research Program of the National Institute of Mental Health (NIMH) and performed at the U.S. National Institutes of Health in accordance with institutional guidelines.
Time pregnant young adult Sprague Dawley rats (SD strain) were acquired from Charles River Inc. They were singly housed upon arrival with gestational ages varying between postconception (PC) day 3 and PC10 and acclimated to the vivarium for at least 72h before the procedure. Postconception day 1 corresponds to the day when an ejaculatory plug was observed. All rats were housed in a temperature and humidity-controlled environment, under 12h/12h reversed light/dark cycle with ad libitum access to food and water.
Fetal injections were done in sixteen adult common marmoset (Callithrix jacchus) females, with ages between 2 and 9 years old. All animals were paired and housed in family groups, with ad libitum access to food and water, and in a temperature and humidity-controlled environment with 12h/12h light/dark cycle. Marmoset dams were reinjected in successive pregnancies without an increased risk of miscarriage or fetal death. The histological data in this report (six neonatal males, one neonatal female, one adult male, and one fetus) is from the offspring of seven.
METHOD DETAILS
Marmoset ultrasound pregnancy evaluation and estimation of gestational day
Marmoset pregnancies were detected by transabdominal ultrasonography and monitored at least once monthly using the FUJIFILM VisualSonics Vevo MD UHF22 and UHF48 linear transducers. The females were hand restrained to minimize movement and given a positive reward (dried fruits or pediasure) before, during, and after each scan. The ventro-dorsal and transverse diameter of the uterus and uterine lumen, along with the biparietal diameter of fetus skull were measured. The PC dates for the procedures were estimated using the published values from Oerke and colleagues.78 Subsequently, these dates were re-estimated after birth by subtracting 143 days from birth, accounting for the average pregnancy length in marmosets, between 140 and 145 days.78 The difference between the estimated gestational day for each female before the procedure and the day after delivery varied between 0 and 9 days (n = 22 pregnancies). The median difference between PC days estimated by fetal ultrasound and post-delivery estimates is 0, while the median absolute difference is 3 days, indicating good agreement between the two approaches. The dates used in this report are the dates re-estimated following delivery, except for the PC114 case reported in Figure 4. This neonate was delivered via c-section at the estimated date of birth under veterinary guidance due to a history of dystocia in the previous pregnancies.
Viral constructs
All the constructs used in this study were from recombinant Adeno-Associated Viruses (rAAV) serotypes. The rAAV plasmids or viral preparations were acquired from Addgene or prepared by the National Institute on Drug Abuse Genetic Engineering and Viral Vector Core Facility (RRID:SCR_022969). Details on the viral vectors, including their sources and the investigators who gifted the items, can be consulted in Table S1.
Plasmid construction and vector packaging
Details on the plasmids, including their sources and the investigators who gifted the items, can be consulted in Table S2. All constructions were performed with ligation-independent cloning (In-Fusion, Takara), and transformed into NEB Stable cells. Isolated colonies were miniprepped and verified by fragment analysis and sequencing prior to viral packaging.
The plasmid pOTTC2234 (RRID:Addgene_228443) was constructed by amplifying the sequence corresponding to the human nestin 2nd intron enhancer with TK mini promoter and EGFP (Nestin-TKmini promoter-EGFP) from Addgene 38777 (RRID:Addgene_38777) and the sequence corresponding to iCre from the plasmid pOTTC1031 (pAAV CMV-IE eGFP-2A-iCre), followed by their insertion into a plasmid already containing a WPRE regulatory element, a TK65pA poly-adenylation signal, and two inverted terminal repeats (ITRs) for AAV (pOTTC2210).34 The open reading frames of EGFP and iCre are separated by a 2xglycine linker.
The plasmid pOTTC2360 (RRID:Addgene_228444) was constructed by amplifying the nucleotide sequences upstream (−971 to −1 bp) and downstream (4–1014 bp) with respect to the rat actin start codon and using them to replace the homology arms corresponding to mouse actin in pAAV-HDR-mEGFP-Actin (RRID:Addgene_119870).36 The guide RNA expression cassette was also retargeted from mouse actin to rat actin using the following seed sequence (TGTGCCTTGATAGTTCGCCA).
rAAV viral vectors were packaged using calcium phosphate precipitation to transfect HEK293 cells with a mix of the pAAV genomic cargo plasmid and two trans plasmids (pHelper and pAAV 2/9 (p0008) as previously described.79 Transfected cell pellets were thawed, then freeze-thawed two times, vortexing after each thaw. MgCl2 (2 mM final; Sigma-Aldrich, St. Louis, MO) and SAN HQ (400 U/mL final, Arcticzymes, 70920–150) were added to the cell solution and incubated at 37°C with continuous shaking for 1 h. The solution was centrifuged for 20 min at 2450 × g at 4°C. The supernatant was transferred to 75 mL of phosphate-buffered saline (PBS) with 2 mM MgCl2 and sequentially filtered through 5-, 0.45-, and 0.22-μm filters. This crude solution was then run through a 1 mL POROS GOPURE AAV9 Column (Life Technologies/Invitrogen, A36650) using an AKTA Pure25 FPLC (Cytiva) at a rate of 2 mL/min and eluted using a solution of 200mM Sodium citrate (pH2) + 400mM NaCl (Sigma-Aldrich) at a rate of 1mL/min. The fractions containing the peak of the UV absorbance (A254 and A280) were collected and dialyzed using a 10,000 MWCO dialysis cassette (Pierce, now Thermo Fisher Scientific, Rockford, IL) in 1 L of PBS containing 0.5 mM MgCl2 for three exchanges over 25–30 h. The equilibrated virus was aliquoted, snap-frozen, and stored at −70°C. Purified vectors were titered by droplet digital PCR with a probe-based assay that recognizes the EGFP or SpCas9.
ICV injection setup
To successfully perform fetal intracerebroventricular injections (FIVIs) without the need for surgery, we designed a system allowing repositioning of the pregnant dam, for better visualization and targeting of the fetuses, while maintaining stability and reducing fetal movements throughout the procedure. This system is fully mechanical, and most of its parts are commercially available. Only a few elements were custom designed by the NIMH Section on Instrumentation, such as the guide tube and transducer holders, and the cradle. Photographs of the experimental setup are provided in Figure S1. Drawings of the guide tube holder and cradle are available in the Data S1. Details of the FIVI setup components are listed in Table S3.
Figure 1A illustrates the assembly of the setup, highlighting its key components, as well as the general layout we use in our procedures, in which the different items are set up to maximize the use of smaller spaces. All supplies and equipment are portable, and their positioning depends on the specific procedure and the room in which it is carried out. All procedure rooms are equipped with a benchtop or a surgical table to place the setup and the electrosurgical unit and has enough space to accommodate a portable anesthesia machine, monitoring systems, and an ultrasound machine. The ultrasound machine is positioned near the person performing the injections. A main component of the FIVI setup is the cradle, composed by a 25 cm length hemicylindrical tube of 15 cm width. At one end of the cradle, a Loc-Line adjustable arm was connected to a stainless-steel extended spring clip for flexible yet stable fixation of the anesthesia mask during the procedure. In both rat and marmoset procedures, for inducing and maintaining anesthesia, we used an RC2 rodent circuit controller (VetEquip), fitted with a nylon-reinforced, 0.25-inch inner diameter, conductive gas supply hose for wall oxygen supply. Since this is a short procedure, oxygen can also be delivered using a high-pressure cylinder. The cradle is supported by a cube geared tripod head (Arca-Swiss C1), clamped to the cradle via a 3D printed dovetail rail fixable to the quick release clamp from the geared head. The length of the rail encompasses the full length of the cradle. This facilitates the initial positioning of the animal beneath the stereotaxic arms and allows for rapid repositioning of the dam over longer centimeters distances (e.g., quick shift between fetuses in the upper and lower abdominal quadrants). Depending on the layout of the room and dexterity of the person doing the injections, the cradle can be positioned with the head of the animal to the right or to the left. The 62° tilt-only axis of the geared head enables rotation of the cradle and allows fine adjustments to the fetus orientation. The geared head is mounted on two single axis translation stages, mounted on a breadboard. The bottom stage is used for x axis translation and, and the top stage, rotated 90° relative to the bottom stage, was added for y axis translation. These plates are equally useful for slower millimeter to centimeter distance adjustments, such as when moving from one rat fetus to another in a neighboring sac within the same uterine horn, or when readjusting the needle trajectory before and after penetration. Two handles were added to the breadboard base for easier transport.
The stereotaxic arms and micromanipulators are carried by an A/P bar attached to a travel vertical translational stage, which allows additional vertical adjustments. The placement of the stereotaxic arms behind the animal gives easy access to the abdomen for the injections. The approximate 8-centimeter space between the geared head and the bar allows the cradle to float without touching the A/P bar. This gives sufficient leeway for cradle rotations and y axis adjustments. We found that using a cradle, rather than a flat surface, was the best design for stabilizing the animal while moving and when rotated, eliminating the need for taping the animal. It also improves temperature maintenance provided by the heating pad throughout the procedure.
The guide needle and ultrasound holders are caried independently by two rotation adapters, fixed to an anterior-posterior (a.p.) slide attachment mounted onto stereotaxic arms. The rotation adapter carrying the ultrasound transducer holder, 3D printed and mounted on a stainless-steel rod, is used to angle the transducer. This provides space for positioning the guide beneath the transducer, allowing it to be visualized on the ultrasound images (Videos S1 and S2). The rotation adapter carrying the guide holder allows slight rotations of the guide. This feature is sometimes useful for correcting the injection trajectory after skin penetration, without the need to modify the angle of the stereotaxic arm. The a.p. slide attachments are used to further adjust the guide relative to the ultrasound transducer. Finer, millimeter or less, a.p. adjustments are achieved using the micromanipulators supporting the stereotaxic arms.
To penetrate the skin and inject the cranium, we used custom-made Hamilton Company needles. For the guide tubes we used 23-gauge or 24-gauge needles. To concentrate the small electric current delivered by the electrosurgical unit to the tip of the guide, the guides were sheathed with Palladium Pebax Heat Shrink Tubing using a heat gun. The blunt and beveled ends of the guide tube were uncoated. The guide tube was mounted on a custom-made holder, made of two anodized aluminum plates mounted on a stainless-steel rod. The guide is inserted into a small indentation between the two aluminum parts and secured in place by tightening the lateral screw. The current produced by the electrosurgical unit (cut mode) is transferred to the guide tube using a pair of alligator clamps. One end connects to the blunt end of the guide, and the other to the electrosurgical pencil. Using current greatly facilitates penetration of the skin (Videos S1 vs. S2).
To inject the fetuses, we use 31-gauge to 33-gauge small hub removable needles. Depending on injected volume, we used 5, 10, 50 or 100 μL Model 701 Hamilton syringes.
Injection procedure and ultrasound-guided injections
Before the procedure, the needle guide holder, forceps, guides, and needles were cleaned, disinfected, and sterilized with ethylene oxide. Unless bent, needle guides and injection needles were reused across procedures. At the start of the procedure, the ultrasound transducer was disinfected using germicidal disposable wipes (Sani-Cloth Prime). The room was divided in two spaces, one for preparing the animal and one for doing the injections. FIVIs in rats were made between PC13 and 21, while injections in fetal marmosets were done between PC62 and 121. Across species, ultrasound imaging was not used to determine fetal sex prior to injections, and fetuses were injected regardless of sex. At the start of the procedure, pregnant rats received a dose of meloxicam (1 mg/kg; 1.5 mg/mL (p.o.)) for preemptive analgesia, and pregnant marmosets received both meloxicam (0.1–0.2 mg/kg, 5 mg/mL (p.o., i.m or s.q.)) for preemptive analgesia and diazepam (0.25–1.0 mg/kg, 5 mg/mL (i.m)) for light sedation. Note that for marmosets, the procedure was performed after fasting to minimize the risk of complications associated with regurgitation and aspiration under anesthesia. Following premedication, the dam was anesthetized with isoflurane gas (2–3% induction, 1–2% maintenance), and transferred to the cradle in the prep area. Ophthalmic eye ointment was then applied to both eyes to prevent dryness. The abdominal fur was trimmed and, when necessary, then shaved with depilatory cream (Nair cream) to remove hair and enhance ultrasound visualization of the fetuses. A small patch of fur (about 2 × 3 cm) was trimmed on the back for electrode grounding. After removing all hair clippings, the dam was gently placed in the supine position over the electrosurgical grounding electrode in the cradle and transferred to the injection setup. Non-sterile ultrasound gel was applied between the trimmed back and the dispersive electrode to improve contact and to safely return the electrosurgical energy delivered through the guide, minimizing damage or burns to the female. The lateral sides of the pad were taped to the heating pad to secure the animal from sliding out of the cradle with larger angles. To prevent possible pressure on the inferior vena cava due to the weight of the uterus, the cradle was angled to elevate the head. Throughout the procedure, the animal’s respiratory rate was manually monitored every 10 min. Additionally, a physiological monitoring system (rat: Kent Scientific SomnoSuite; marmoset: IntelliVue MX500) was used to monitor the animal’s pulse oximetry, and to monitor and regulate body temperature. In marmosets, the system further allowed for monitoring of expired CO2, and continuous recording of respiratory and heart rates. In both species, supplemental heating was provided using a benchtop infrared heater and an overhead heating lamp to ensure regulation of the animal’s body temperature around 37°C and prevent hypothermia. Vital signs and toe-pinch response were monitored throughout the procedure and recorded every 10 min. The concentration of isoflurane was adjusted whenever necessary (e.g., slow or fast breathing), and the experiment was paused until stabilization of the values.
Once the physiological parameters stabilized, the shaved abdomen was sterilized with 3 alternating scrubs of 7.5% povidone-iodine (Betadine Surgical Scrub) and 70% alcohol, and sterile ultrasound gel (Sterile Aquasonic 100 gel) was applied. The sterile field was created, and thereafter, the only items that came into contact with the skin were the sterile guide and transducer. A critical aspect in this procedure is maintaining a straight approach to the target and avoiding the displacement of the fetus during penetration of the abdomen with the needle. Advancing a needle through the skin without displacing the underlying organs is challenging because of the toughness and elasticity of the dermal layer (Video S1). In pilot experiments, we determined that the most reliable and efficient means to achieve precise targeting with minimal tissue displacement was to implement a three-step penetration procedure. In the first step, the guide tube is advanced through the ultrasound gel and aligned to the ultrasound transducer over the fetal head using the setup’s translational stages and stereotaxic arm manipulators. Depending on the positioning of individual fetuses, the angle of the guide tube is adjusted to accommodate the planned trajectory. While it gently presses against the skin, a brief electric current is applied to the blunt non-insulated side of the guide tube to help insertion with minimal displacement of the fetus (Video S2). As the injection needle can penetrate through the abdominal wall and into the uterus, insertion of the guide tube beyond the subcutaneous compartment is not necessary. This cautery step typically introduces air bubbles at the site of skin penetration. However, these bubbles cause minimal disruption of the US visualization of the target, since the angular approach means that the bubbles concentrate outside the US field of view. In the second step, the internal needle is advanced smoothly by hand through the guide, into the abdominal wall and uterine muscle and then into the lumen adjacent to the fetal skull. The position and angle of the needle relative to the head is important, with the desired approach angle determined, in part, by skull thickness and size of the uterine lumen, parameters that varied with species and gestational age. In the third step, the skull is manually penetrated with a brief jolt that abruptly displaces the needle tip by 1–3 mm (Video S3). Penetration is most successful when attempted from an angle having maximum mechanical advantage, minimizing the risk of slipping along the convex skull surface and in a region of the cranium with thinner developing skull. Each attempt at penetration is evaluated in real time by the experimenter on the US display. In the case of trajectory misalignment or failed penetration, it is usually possible to readjust the geometrical approach to achieve successful entry without removal of the needle from the uterine lumen. A particularly effective strategy for this is to move the animal’s position and angle using the adjustments in the cradle setup with the guide tube still in place. Video S3 shows a P107 FIVI in marmoset.
The rAAV virus, stored at −80°C, was transported to the procedure room on wet ice. After allowing it to thaw, the vector was thoroughly mixed using a micropipette, and the amount to inject was aliquoted onto a sterile syringe cap and loaded into the injection syringe. In the rat we injected between 2 and 10μL (eye socket: 2 μL; cerebral ventricles: 5 and 10 μL) and in the marmoset 10 and 60μL. The presence of small air bubbles in the last portion of the injection served to increase the echogenicity of the injectate and thus allowed visualization and confirmation of the injected target. Following an injection, the backflow of rAAV was reduced by keeping the needle in place for up to 5 min. Upon retraction of the needle and the guide tube, the heart of the fetus was monitored to ensure proper functioning and stability. These steps were repeated for each fetus. When determining the best path for an injection, special care was taken to avoid injuring nipples (in rats), internal organs (such as bladder and intestines) and large vessels (such as the umbilical cord). Whenever possible, we avoided penetrating the placenta to reach the fetus by repositioning the pregnant dam using the geared head and translational plates. On occasion, particularly in marmosets, injections were made through the placenta. When doing so, we used the color doppler mode to minimize the risks of and evaluate any bleeding along the trajectory. For most injection sites, no blood was observed upon removal of the needle and needle guide. On rare cases, a minimal amount was seen when retracting the needle, without any presence of active bleeding. At the end of the procedure, a small amount of local anesthetic medication (lidocaine) was applied to each site to prevent pain, along with a triple antibiotic ointment to prevent infection. Before returning to the housing room, the pregnant dam recovered in a warmed environment and was only transferred after being bright, alert, and reactive. The use of current to insert the guide resulted in mild scab formation on the injection sites in the days following the procedure. Typically, within one to two weeks, the scabs resolved without infections nor other complications. After each procedure, the pregnant dam was provided with supplemental nutritional support and monitored for any signs of pain, miscarriage (e.g., blood discharge, contractions) and dystocia until delivery. Table S3 provides additional information on the tools and agents used.
Perinatal identification of positive animals
Being able to identify the progeny expressing a given transgene in an easy and timely manner is important to select animals for future experiments. For instance, the litters of rats injected in this procedure had up to 17 animals, and of those not all received FIVI injections. The high yield of endogenous reporter expression leads to brightness levels that can be observed using whole body imaging in living animals using LED flashlights. As for transgenic animals,80 this approach was also effective in retrieving animals transduced in utero with viral vectors leading to strong expression of the reporter gene. In rats, expression was most prominent in the brain, spinal cord, back muscle and in the face (peripheral nerve labeling). In marmosets, it could be observed in less pigmented and hairless regions. Since reporter proteins are spectrally distinct, their distribution can be detected using LED flashlights with appropriate excitation and emission filters. We used a portable, battery-operated dual fluorescence flashlight with high intensity LEDs and Royal Blue and Green excitation filters (Model DFP-1, Nightsea). For green shifted reporters we used the Royal Blue filter set, with excitation 440–460nm and 500–560nm bandpass barrier filter glasses. For red shifted reporters, we used the Green filter set, with excitation 510–540nm and 600nm bandpass barrier filter glasses. Transgene expression persists at strong levels through adulthood, remaining visible in hairless regions.
Histological processing
Perfusion and tissue collection
A 2–3 day variation in gestation is expected for the Sprague-Dawley SD strain. For this reason, to minimize the potential impact of age in the transduction patterns observed, the date of tissue collection in rats was determined based on postconception (PC) days rather than relative to birth. Unless otherwise indicated, all tissue collected from rats was obtained at PC44. At this age, approximately 21 days after birth, the number of neuronal cells in the rat brain, particularly in the cerebral cortex, is stable and maintained throughout adulthood.81 Rats were deeply anesthetized with isoflurane gas (3–4% induction, 4–5% maintenance), and transcardially perfused with heparinized normal saline or 1× PBS followed by 4% paraformaldehyde (PFA, pH 7.4) in 1× phosphate-buffered saline (PBS). Marmosets were first deeply anesthetized with isoflurane gas (3–5%), and were overdosed by intraperitoneal (i.p.) delivery of pentobarbital (at least 80–120 mg/kg). They were then transcardially perfused as described above. In both rats and marmosets, the brains were extracted and postfixed in 4% PFA in 1× PBS. Note that the histological data presented in this study were obtained from animals injected with 10 μL in rats and 60 μL in marmosets.
Tissue selection and processing. In the most recent cases, following brain extraction, images of the dorsal and ventral surfaces of the fixed brains were acquired with a customized MacroFluo Z6 APO with a 0.5× Planapo z series objective (nA max. 0.0585), a digital monochrome DFC3000G camera, a LED3000 RL ring light, a fluorescence illuminator LRF 4/22 and external light source EL6000 allowing for both brightfield and fluorescence imaging (Leica Microsystems Inc.). The pictures were then exported to TIF format using the Leica Application Suite X, loaded into stack into Adobe Photoshop, and compared. In general, the brains with higher expression levels were selected for further processing, and the remaining were transferred to 1× PBS with 0.02% of sodium azide at 4°C for longer term storage. Some examples of these images are shown in Figures S4B–S4D. For the experiments in which gene expression was sparse, such as those with CrispR/Cas9, samples of spinal cords were sectioned to confirm expression. The brains selected for further processing were then cryopreserved in gradients of glycerol or sucrose (10–20% or 10–20–30%) in 1× PBS. The postfixation and cryoprotection intervals were adjusted based on the quality of the fixation and cryoprotection. Once cryoprotected, brains were sometimes embedded in 15% gelatin in 1× PBS. Forty micrometers thick sections were cut on a sliding freezing microtome or cryostat. One in twelve sections were collected in 1× PBS for screening, and the remaining sections were collected in an antifreeze solution for long-term storage and subsequent processing.
For the PC87 case (Figure 2), additional PFA-fixed tissues including liver, intestines, kidneys, skeletal muscle, and adipose tissue were collected to evaluate transduction of non-neuronal cells. The blocks were cryoprotected in graded sucrose solutions (10%, 20%, and 30%), and 20 μm sections were directly collected for analysis on slides using a cryostat.
Immunofluorescent and microscopy imaging
Forty-micron thick free-floating sections were rinsed at room temperature (RT) with 1× PBS. Six 10-min washes were performed for antifreeze-stored sections, and three washes for sections collected in 1× PBS. To permeabilize the cells, the sections were incubated in three 10-min washes of PBST (1× PBS with 0.5% Triton X-100), followed by 30 min in 50% ethanol. After three 5-min washes in PBST, the sections were incubated in 10% Normal Goat Serum for 1 h to block non-specific sites. They were then transferred to a 10% PBST (1× PBS with 0.05% Triton X-100) solution containing the primary antibody and incubated overnight at 4°C under gentle shaking. Table S4 provides additional detail for the primary and secondary antibodies used in this study, including their working dilutions. For the marmoset fetal sections and rat CrispR/Cas9 and retinofugal projections cases, the endogenous reporter gene expression was enhanced with an anti-GFP antibody. The sections were then rinsed three times in 1× PBS and incubated in Alexa-fluor conjugated fluorescent secondary antibodies diluted with 1× PBS with gentle shaking at RT for 2h. Following three 10-min rinses in 1× PBS, the sections were mounted in charged microscope slides, stained with Dapi and coverslipped with Fluoromount-G Mounting Medium to minimize fluorescence quenching and photobleaching during microscopy imaging. Sections were imaged through 20× objectives on a VS200 slide scanner from Olympus to obtain widefield images. Tiled z-stacks of selected regions were taken on a Leica Stellaris 8 confocal microscope using 20× or 40× objectives. Fiji software was used to export the images after exposure adjustments and contrast enhancement, and to perform uniform background subtraction due to autofluorescence (e.g., incomplete perfusion) when needed. When necessary, further adjustments in orientation and in brightness and contrast were made in Photoshop.
Optical clearing and imaging of fetuses
A procedure based on the SHIELD82 (SHIELD fixation kit from Life Canvas, Inc) and CUBIC methods83 (reagents from Sigma-Aldrich, based on recipes in the cited manuscript) was used to evaluate labeling of the peripheral nervous system in intact fetuses. The fetus in Figure 5E was co-injected with AAV9-CAG-tdTomato (5 μL) and AAV2-CAG-GFP (5 μL) at PC15. Only the AAV9-CAG-tdTomato is shown, as it is the only one labeling the peripheral nervous system with injections at this gestation time. The pregnant dam was transcardially perfused with heparinized normal saline, followed by 200mL of SHIELD fixation solution, at PC19, four days after the injections. After perfusion, positive fetuses were collected and postfixed in SHIELD fixation solution for 6 days, incubated in SHIELD OFF solution for 6 days, and transferred to 37°C SHIELD ON solution for 24 h. Subsequently, they were transferred to 1× PBS with 0.1% sodium azide at 4°C until processing. The fetuses were passively cleared with Sodium dodecyl sulfate (SDS) clearing solution. Once cleared, they were transferred to 50% CUBIC-R+(N) overnight, and then 100% CUBIC-R+(N) for refractive index matching. Image acquisition was performed in silicone fluid (PM125, Clearco Products) using a 3i CTLS lightsheet microscope (1×, 0.25 NA objective, optical zoom 3.7). Image fields were merged using 3i Slidebook software and then background subtraction was performed using custom python scripts (adapted from de Rooi and colleagues84) on the NIH High Performance Cluster, followed by importing the images into Arivis Vision4D for visualization by adjusting orientation and look up tables.
QUANTIFICATION ANALYSIS AND STATISTICAL ANALYSIS
Quantification of NeuN and GFP neuronal co-labeling
Confocal images obtained using a Leica Stellaris 8 confocal microscope with a 20× objective were opened in Fiji. Regions of interest (ROIs) encompassing the pial surface to white matter (WM) were manually delineated, and a maximum-intensity z-projection was applied. The ROIs were then exported as TIF files and opened in NeuroInfo software (MBF Bioscience), where a “Detect Cells” pipeline was applied to each channel (NeuN: large diameter = 32.00 μm, small diameter = 8.00 μm, detection strength = 4; GFP: large diameter = 32.00 μm, small diameter = 10.00 μm, detection strength = 4). Following automated detection, images from each channel were reviewed, and false positives and false negatives were manually corrected. Colocalization analysis was performed by manually replacing the markers of double-labeled GFP+/NeuN+ cells with a third marker. The coordinates of labeled cell centers were exported and imported into RStudio. Using R, each ROI was divided into 20 bins of equal height, and the percentage of GFP+/NeuN+ neurons was plotted using the lattice package. The resulting graphs were exported and formatted in Photoshop for Figure 2 preparation. The vast majority of GFP-positive cells were NeuN-positive; only a minimal number were NeuN-negative.
Comparison of rates of miscarriage in pregnancies with or without FIVI
Miscarriage rates were compared between pregnancies with and without FIVI using the prop.test function from the stats package in R. The analysis tested whether the miscarriage rate was similar between groups (alternative = “two.sided”), or significantly lower (alternative = “less”) or higher (alternative = “greater”) in non-injected compared to injected pregnancies. A significance threshold of p < 0.01 was applied to reject the null hypothesis.
Supplementary Material
Supplemental information can be found online at https://doi.org/10.1016/j.celrep.2025.116756.
Figure 6. Proof-of-concept gene-editing and peripheral-labeling experiments using FIVI.
(A) CRISPR-Cas9-mediated genome editing in the rat through co-transduction with rAAV9-EFS-SpCas9 (5 μL) and rAAV9-rActin-EGFP-donor (5 μL).
(B) FIVIs at PC19 with dual-rAAV CRISPR-Cas9 system led to sparse gene editing of neurons in the cerebral cortex at PC44 (insets b and c). The EGFP reporter fused to actin fills the neurons and accumulates in dendritic spines, such as those highlighted in the insets (d) and (e). The inset (f) highlights edited cells in the choroid plexus. Scale bars: (a) 1 mm, (b) 500 μm, and (c–f) 10 μm. The diagram was adapted from Nishiyama and colleagues.36
(C) Transduced retinal ganglion cells (RGCs) labeled with a PC19 rAAV9-CAG-GFP (2 μL) injection into the right eye socket (red arrow). RGC axons decussate in the chiasma (open arrowhead) and project to targets in the contralateral hemisphere. Note the lack of a label in the contralateral optic tract (closed arrowhead).
(D) Sequential coronal sections (rostral to caudal) at PC44 showing the pathway and the central targets of RGC projections, such as the dorsal lateral geniculate nucleus (dLGN), the ventral LGN (vLGN), the intergeniculate leaflet (IGL), the medial terminal nucleus (MTN), and the olivary pretectal nucleus (OPN). Scale bar: 500 μm.
(E) FIVIs lead to both central and peripheral nervous system labeling. Optical clearing of a PC19 transduced fetus, injected at PC15 with rAAV9-CAG-tdTomato (5 μL), enables 3D visualization of reporter gene expression in the central (e.g., olfactory bulbs [OBs], cortical plate [CP], and spinal cord) and peripheral nervous system (e.g., dorsal root ganglion [DRG], trigeminal ganglion [TG], and vibrissal nerves [VNs]). This approach facilitates investigation of the spatial and temporal development of cell-type-specific projection pathways. The image was partially created in BioRender. Scale bar: 2 mm.
KEY RESOURCES TABLE
| REAGENT or RESOURCE | SOURCE | IDENTIFIER |
|---|---|---|
|
Antibodies | ||
| Chicken anti-Green Fluorescent Protein | Aves Labs | Cat# GFP-1020, RRID:AB_10000240 |
| Guinea pig anti-NeuN | Millipore | Cat# ABN90P, RRID:AB_2341095 |
| Mouse anti-NeuN | Millipore | Cat# MAB377, RRID:AB_2298772 |
| Mouse anti-parvalbumin | Swant | Cat# 235, RRID:AB_10000343 |
| Goat anti-Guinea Pig IgG (H + L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 555 | Thermo Fisher Scientific | Cat# A-21435, RRID:AB_2535856 |
| Goat anti-Chicken IgY (H + L) Secondary Antibody, Alexa Fluor 488 | Thermo Fisher Scientific | Cat# A-11039, RRID:AB_2534096 |
| Goat anti-Mouse IgG (H + L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 647 | Thermo Fisher Scientific | Cat# A-21236, RRID:AB_2535805 |
|
Bacterial and virus strains | ||
| AAV2-CAG-GFP | Addgene | 37825-AAV2 |
| AAV8-nEF-Con/Foff 2.0-ChRmine-oScarlet | Addgene | 137161-AAV8 |
| AAV9-CAG-GFP | Addgene | 37825-AAV9 |
| AAV9-CAG-tdTomato (codon diversified) | Addgene | 59462-AAV9 |
| AAV9-NestinTK-EGFP-iCre | NIDA GEVVC for this study; deposited on Addgene | AAV916 |
| AAV9-rActin-EGFP-donor | NIDA GEVVC for this study; deposited on Addgene | AAV942 |
| AAV9-EFS-SpCas9 | NIDA GEVVC for this study; deposited on Addgene | AAV949 |
|
Experimental models: Organisms/strains | ||
| Sprague-Dawley rats | Charles River | Strain Code: 001 |
| Common marmosets (Callithrix jacchus) | NIMH | N/A |
|
Recombinant DNA | ||
| pNestin-EGFP | Addgene | RRID:Addgene_38777 |
| pAAV-HDR-mEGFP-Actin | Addgene | RRID:Addgene_119870 |
| pAAV-EFS-SpCas9 | Addgene | RRID:Addgene_104588 |
| pAAV CMV-IE eGFP-2A-iCre | NIDA GEVVC | pOTTC1031 |
| pAAV JeT ConFon HA-hM4D(Gi)-mCherry TK65pA | NIDA GEVVC | pOTTC2210 |
| pAAV Nestin GFP-iCre | NIDA GEVVC for this study | pOTTC2234 |
| pAAV rActin EGFP donor | NIDA GEVVC for this study | pOTTC2360 |
| pHelper | PENN Vector Core | N/A |
| pAAV 2/9 | PENN Vector Core | N/A |
|
Software and algorithms | ||
| Fiji | ImageJ | https://imagej.net/software/fiji/ |
| NeuroInfo | MBF | https://www.mbfbioscience.com/products/neuroinfo/ |
| R | The R Project for Statistical Computing | https://www.r-project.org/ |
| Adobe Photoshop 2025 | Adobe | https://www.adobe.com/products/photoshop.html |
| 3i Slidebook | Intelligent Imaging | https://www.intelligent-imaging.com/slidebook |
Highlights.
rAAV delivery into the marmoset fetal brain enables safe, efficient gene transfer
Transgene expression begins before birth and persists into adulthood
Gestational timing and rAAV serotype are key factors modulating prenatal transduction
This method facilitates basic and translational neuroscience in primates
ACKNOWLEDGMENTS
This work was supported by funding from the Intramural Research Program of the National Institute of Mental Health (ZIAMH002898) to D.A.L. and the National Institute of Child Health and Human Development (P50HD103536), University of Rochester Intellectual and Developmental Disabilities Research Center (UR-IDDRC), to K.H.W. A.R.R.G. was funded by the National Institutes of Health’s Visiting Fellow Intramural Research Training Award. N.H. and N.W. were funded by the National Institutes of Health postbaccalaureate Intramural Research Training Award. We thank George Dold and members of the Section on Instrumentation of the Intramural Research Program of the National Institute of Mental Health for their support in designing and fabricating the custom-made parts used in this project. We are grateful for support from the Systems Neuroscience Imaging Resource of the Intramural Research Program of the National Institute of Mental Health for their allocation of imaging resources used in parts of this research. We thank the Genetic Engineering and Viral Vector Core from the Intramural Research Program of the National Institute of Drug Abuse for the development and production of genetic tools used in this study. We also thank the Veterinary Medicine and Resources Branch from the Intramural Research Program of the National Institute of Mental Health for their much-valued support with veterinary health care, animal husbandry, and technical assistance. Finally, we are grateful to Lenegereshe Baweke and Sean Kearney for technical assistance.
Footnotes
DECLARATION OF INTERESTS
The authors declare no competing interests.
REFERENCES
- 1.Luo L, Callaway EM, and Svoboda K. (2018). Genetic Dissection of Neural Circuits: A Decade of Progress. Neuron 98, 256–281. 10.1016/j.neuron.2018.03.040. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Huang ZJ, and Zeng H. (2013). Genetic approaches to neural circuits in the mouse. Annu. Rev. Neurosci 36, 183–215. 10.1146/annurev-neuro-062012-170307. [DOI] [PubMed] [Google Scholar]
- 3.Fenno L, Yizhar O, and Deisseroth K. (2011). The development and application of optogenetics. Annu. Rev. Neurosci 34, 389–412. 10.1146/annurev-neuro-061010-113817. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Madisen L, Zwingman TA, Sunkin SM, Oh SW, Zariwala HA, Gu H, Ng LL, Palmiter RD, Hawrylycz MJ, Jones AR, et al. (2010). A robust and high-throughput Cre reporting and characterization system for the whole mouse brain. Nat. Neurosci 13, 133–140. 10.1038/nn.2467. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Charreau B, Tesson L, Soulillou JP, Pourcel C, and Anegon I. (1996). Transgenesis in rats: technical aspects and models. Transgenic Res. 5, 223–234. 10.1007/BF01972876. [DOI] [PubMed] [Google Scholar]
- 6.Chenouard V, Remy S, Tesson L, Ménoret S, Ouisse LH, Cherifi Y, and Anegon I. (2021). Advances in Genome Editing and Application to the Generation of Genetically Modified Rat Models. Front. Genet 12, 615491. 10.3389/fgene.2021.615491. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Johnson MB, Sun X, Kodani A, Borges-Monroy R, Girskis KM, Ryu SC, Wang PP, Patel K, Gonzalez DM, Woo YM, et al. (2018). Aspm knockout ferret reveals an evolutionary mechanism governing cerebral cortical size. Nature 556, 370–375. 10.1038/s41586-018-0035-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Yu M, Sun X, Tyler SR, Liang B, Swatek AM, Lynch TJ, He N, Yuan F, Feng Z, Rotti PG, et al. (2019). Highly Efficient Transgenesis in Ferrets Using CRISPR/Cas9-Mediated Homology-Independent Insertion at the ROSA26 Locus. Sci. Rep 9, 1971. 10.1038/s41598-018-37192-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Yang SH, Cheng PH, Banta H, Piotrowska-Nitsche K, Yang JJ, Cheng ECH, Snyder B, Larkin K, Liu J, Orkin J, et al. (2008). Towards a transgenic model of Huntington’s disease in a non-human primate. Nature 453, 921–924. 10.1038/nature06975. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Sasaki E, Suemizu H, Shimada A, Hanazawa K, Oiwa R, Kamioka M, Tomioka I, Sotomaru Y, Hirakawa R, Eto T, et al. (2009). Generation of transgenic non-human primates with germline transmission. Nature 459, 523–527. 10.1038/nature08090. [DOI] [PubMed] [Google Scholar]
- 11.Niu Y, Shen B, Cui Y, Chen Y, Wang J, Wang L, Kang Y, Zhao X, Si W, Li W, et al. (2014). Generation of gene-modified cynomolgus monkey via Cas9/RNA-mediated gene targeting in one-cell embryos. Cell 156, 836–843. 10.1016/j.cell.2014.01.027. [DOI] [PubMed] [Google Scholar]
- 12.Park JE, Zhang XF, Choi SH, Okahara J, Sasaki E, and Silva AC (2016). Generation of transgenic marmosets expressing genetically encoded calcium indicators. Sci. Rep 6, 34931. 10.1038/srep34931. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Liu Z, Li X, Zhang JT, Cai YJ, Cheng TL, Cheng C, Wang Y, Zhang CC, Nie YH, Chen ZF, et al. (2016). Autism-like behaviours and germline transmission in transgenic monkeys overexpressing MeCP2. Nature 530, 98–102. 10.1038/nature16533. [DOI] [PubMed] [Google Scholar]
- 14.Izpisua Belmonte JC, Callaway EM, Caddick SJ, Churchland P, Feng G, Homanics GE, Lee KF, Leopold DA, Miller CT, Mitchell JF, et al. (2015). Brains, genes, and primates. Neuron 86, 617–631. 10.1016/j.neuron.2015.03.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Emborg ME (2017). Nonhuman Primate Models of Neurodegenerative Disorders. ILAR J. 58, 190–201. 10.1093/ilar/ilx021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Jennings CG, Landman R, Zhou Y, Sharma J, Hyman J, Movshon JA, Qiu Z, Roberts AC, Roe AW, Wang X, et al. (2016). Opportunities and challenges in modeling human brain disorders in transgenic primates. Nat. Neurosci 19, 1123–1130. 10.1038/nn.4362. [DOI] [PubMed] [Google Scholar]
- 17.Feng G, Jensen FE, Greely HT, Okano H, Treue S, Roberts AC, Fox JG, Caddick S, Poo MM, Newsome WT, and Morrison JH (2020). Opportunities and limitations of genetically modified nonhuman primate models for neuroscience research. Proc. Natl. Acad. Sci. USA 117, 24022–24031. 10.1073/pnas.2006515117. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Aida T, and Feng G. (2020). The dawn of non-human primate models for neurodevelopmental disorders. Curr. Opin. Genet. Dev 65, 160–168. 10.1016/j.gde.2020.05.040. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Nectow AR, and Nestler EJ (2020). Viral tools for neuroscience. Nat. Rev. Neurosci 21, 669–681. 10.1038/s41583-020-00382-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Campos LJ, Arokiaraj CM, Chuapoco MR, Chen X, Goeden N, Gradinaru V, and Fox AS (2023). Advances in AAV technology for delivering genetically encoded cargo to the nonhuman primate nervous system. Curr. Res. Neurobiol 4, 100086. 10.1016/j.crneur.2023.100086. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Kim JY, Ash RT, Ceballos-Diaz C, Levites Y, Golde TE, Smirnakis SM, and Jankowsky JL (2013). Viral transduction of the neonatal brain delivers controllable genetic mosaicism for visualising and manipulating neuronal circuits in vivo. Eur. J. Neurosci 37, 1203–1220. 10.1111/ejn.12126. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Clancy B, Darlington RB, and Finlay BL (2001). Translating developmental time across mammalian species. Neuroscience 105, 7–17. 10.1016/s0306-4522(01)00171-3. [DOI] [PubMed] [Google Scholar]
- 23.Rahim AA, Wong AM, Ahmadi S, Hoefer K, Buckley SMK, Hughes DA, Nathwani AN, Baker AH, McVey JH, Cooper JD, and Waddington SN (2012). In utero administration of Ad5 and AAV pseudotypes to the fetal brain leads to efficient, widespread and long-term gene expression. Gene Ther. 19, 936–946. 10.1038/gt.2011.157. [DOI] [PubMed] [Google Scholar]
- 24.Massaro G, Mattar CNZ, Wong AMS, Sirka E, Buckley SMK, Herbert BR, Karlsson S, Perocheau DP, Burke D, Heales S, et al. (2018). Fetal gene therapy for neurodegenerative disease of infants. Nat. Med 24, 1317–1323. 10.1038/s41591-018-0106-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Charvet CJ, and Finlay BL (2014). Evo-devo and the primate isocortex: the central organizing role of intrinsic gradients of neurogenesis. Brain Behav. Evol 84, 81–92. 10.1159/000365181. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Rakic P. (1974). Neurons in rhesus monkey visual cortex: systematic relation between time of origin and eventual disposition. Science 183, 425–427. 10.1126/science.183.4123.425. [DOI] [PubMed] [Google Scholar]
- 27.Rakic P. (1988). Specification of cerebral cortical areas. Science 241, 170–176. 10.1126/science.3291116. [DOI] [PubMed] [Google Scholar]
- 28.Rakic P. (2002). Pre- and post-developmental neurogenesis in primates. Clin. Neurosci. Res 2, 29–39. 10.1016/S1566-2772(02)00005-1. [DOI] [Google Scholar]
- 29.Kolk SM, and Rakic P. (2022). Development of prefrontal cortex. Neuropsychopharmacology 47, 41–57. 10.1038/s41386-021-01137-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Meyer NL, and Chapman MS (2022). Adeno-associated virus (AAV) cell entry: structural insights. Trends Microbiol. 30, 432–451. 10.1016/j.tim.2021.09.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Di Bella DJ, Dominguez-Iturza N, Brown JR, and Arlotta P. (2024). Making Ramon y Cajal proud: Development of cell identity and diversity in the cerebral cortex. Neuron 112, 2091–2111. 10.1016/j.neuron.2024.04.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Bernal A, and Arranz L. (2018). Nestin-expressing progenitor cells: function, identity and therapeutic implications. Cell. Mol. Life Sci 75, 2177–2195. 10.1007/s00018-018-2794-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Bott CJ, Johnson CG, Yap CC, Dwyer ND, Litwa KA, and Winckler B. (2019). Nestin in immature embryonic neurons affects axon growth cone morphology and Semaphorin3a sensitivity. Mol. Biol. Cell 30, 1214–1229. 10.1091/mbc.E18-06-0361. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Noisa P, Urrutikoetxea-Uriguen A, Li M, and Cui W. (2010). Generation of human embryonic stem cell reporter lines expressing GFP specifically in neural progenitors. Stem Cell Rev. Rep 6, 438–449. 10.1007/s12015-010-9159-9. [DOI] [PubMed] [Google Scholar]
- 35.Doudna JA (2020). The promise and challenge of therapeutic genome editing. Nature 578, 229–236. 10.1038/s41586-020-1978-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Nishiyama J, Mikuni T, and Yasuda R. (2017). Virus-Mediated Genome Editing via Homology-Directed Repair in Mitotic and Postmitotic Cells in Mammalian Brain. Neuron 96, 755–768.e5. 10.1016/j.neuron.2017.10.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Ayers JI, Fromholt S, Sinyavskaya O, Siemienski Z, Rosario AM, Li A, Crosby KW, Cruz PE, DiNunno NM, Janus C, et al. (2015). Widespread and efficient transduction of spinal cord and brain following neonatal AAV injection and potential disease modifying effect in ALS mice. Mol. Ther 23, 53–62. 10.1038/mt.2014.180. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Tremblay S, Acker L, Afraz A, Albaugh DL, Amita H, Andrei AR, Angelucci A, Aschner A, Balan PF, Basso MA, et al. (2020). An Open Resource for Non-human Primate Optogenetics. Neuron 108, 1075–1090.e6. 10.1016/j.neuron.2020.09.027. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Mattar CN, Waddington SN, Biswas A, Johana N, Ng XW, Fisk AS, Fisk NM, Tan LG, Rahim AA, Buckley SMK, et al. (2013). Systemic delivery of scAAV9 in fetal macaques facilitates neuronal transduction of the central and peripheral nervous systems. Gene Ther. 20, 69–83. 10.1038/gt.2011.216. [DOI] [PubMed] [Google Scholar]
- 40.Dehay B, Dalkara D, Dovero S, Li Q, and Bezard E. (2012). Systemic scAAV9 variant mediates brain transduction in newborn rhesus macaques. Sci. Rep 2, 253. 10.1038/srep00253. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Kim S, Moon HS, Vo TT, Kim CH, Im GH, Lee S, Choi M, and Kim SG (2023). Whole-brain mapping of effective connectivity by fMRI with cortex-wide patterned optogenetics. Neuron 111, 1732–1747.e6. 10.1016/j.neuron.2023.03.002. [DOI] [PubMed] [Google Scholar]
- 42.Song X, Guo Y, Li H, Chen C, Lee JH, Zhang Y, Schmidt Z, and Wang X. (2022). Mesoscopic landscape of cortical functions revealed by through-skull wide-field optical imaging in marmoset monkeys. Nat. Commun 13, 2238. 10.1038/s41467-022-29864-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Musall S, Kaufman MT, Juavinett AL, Gluf S, and Churchland AK (2019). Single-trial neural dynamics are dominated by richly varied movements. Nat. Neurosci 22, 1677–1686. 10.1038/s41593-019-0502-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Fredericks JM, Dash KE, Jaskot EM, Bennett TW, Lerchner W, Dold G, Ide D, Cummins AC, Der Minassian VH, Turchi JN, et al. (2020). Methods for mechanical delivery of viral vectors into rhesus monkey brain. J. Neurosci. Methods 339, 108730. 10.1016/j.jneumeth.2020.108730. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Lerchner W, Dash K, Rose D, Eldridge MAG, Rothenhoefer KM, Yan X, Costa VD, Averbeck B, and Richmond BJ (2023). Efficient viral expression of a chemogenetic receptor in the old-world monkey amygdala. Curr. Res. Neurobiol 4, 100091. 10.1016/j.crneur.2023.100091. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Yazdan-Shahmorad A, Tian N, Kharazia V, Samaranch L, Kells A, Bringas J, He J, Bankiewicz K, and Sabes PN (2018). Widespread optogenetic expression in macaque cortex obtained with MR-guided, convection enhanced delivery (CED) of AAV vector to the thalamus. J. Neurosci. Methods 293, 347–358. 10.1016/j.jneumeth.2017.10.009. [DOI] [PubMed] [Google Scholar]
- 47.Foust KD, Nurre E, Montgomery CL, Hernandez A, Chan CM, and Kaspar BK (2009). Intravascular AAV9 preferentially targets neonatal neurons and adult astrocytes. Nat. Biotechnol 27, 59–65. 10.1038/nbt.1515. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Gray SJ, Matagne V, Bachaboina L, Yadav S, Ojeda SR, and Samulski RJ (2011). Preclinical differences of intravascular AAV9 delivery to neurons and glia: a comparative study of adult mice and nonhuman primates. Mol. Ther 19, 1058–1069. 10.1038/mt.2011.72. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Samaranch L, Salegio EA, San Sebastian W, Kells AP, Foust KD, Bringas JR, Lamarre C, Forsayeth J, Kaspar BK, and Bankiewicz KS (2012). Adeno-associated virus serotype 9 transduction in the central nervous system of nonhuman primates. Hum. Gene Ther 23, 382–389. 10.1089/hum.2011.200. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Chan KY, Jang MJ, Yoo BB, Greenbaum A, Ravi N, Wu WL, Sénchez-Guardado L, Lois C, Mazmanian SK, Deverman BE, and Gradinaru V. (2017). Engineered AAVs for efficient noninvasive gene delivery to the central and peripheral nervous systems. Nat. Neurosci 20, 1172–1179. 10.1038/nn.4593. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.Deverman BE, Pravdo PL, Simpson BP, Kumar SR, Chan KY, Banerjee A, Wu WL, Yang B, Huber N, Pasca SP, and Gradinaru V. (2016). Cre-dependent selection yields AAV variants for widespread gene transfer to the adult brain. Nat. Biotechnol 34, 204–209. 10.1038/nbt.3440. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.Bedbrook CN, Deverman BE, and Gradinaru V. (2018). Viral Strategies for Targeting the Central and Peripheral Nervous Systems. Annu. Rev. Neurosci 41, 323–348. 10.1146/annurev-neuro-080317-062048. [DOI] [PubMed] [Google Scholar]
- 53.Chuapoco MR, Flytzanis NC, Goeden N, Christopher Octeau J, Roxas KM, Chan KY, Scherrer J, Winchester J, Blackburn RJ, Campos LJ, et al. (2023). Adeno-associated viral vectors for functional intravenous gene transfer throughout the non-human primate brain. Nat. Nanotechnol 18, 1241–1251. 10.1038/s41565-023-01419-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 54.Goertsen D, Flytzanis NC, Goeden N, Chuapoco MR, Cummins A, Chen Y, Fan Y, Zhang Q, Sharma J, Duan Y, et al. (2022). AAV capsid variants with brain-wide transgene expression and decreased liver targeting after intravenous delivery in mouse and marmoset. Nat. Neurosci 25, 106–115. 10.1038/s41593-021-00969-4. [DOI] [PubMed] [Google Scholar]
- 55.Garrett DJ, Larson JE, Dunn D, Marrero L, and Cohen JC (2003). In utero recombinant adeno-associated virus gene transfer in mice, rats, and primates. BMC Biotechnol. 3, 16. 10.1186/1472-6750-3-16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 56.Harrison MR, Anderson J, Rosen MA, Ross NA, and Hendrickx AG (1982). Fetal surgery in the primate I. Anesthetic, surgical, and tocolytic management to maximize fetal-neonatal survival. J. Pediatr. Surg 17, 115–122. 10.1016/s0022-3468(82)80193-0. [DOI] [PubMed] [Google Scholar]
- 57.Rakic P, and Goldman-Rakic PS (1985). Use of Fetal Neurosurgery for Experimental Studies of Structural and Functional Brain Development in Nonhuman Primates. In Perinatal Neurology and Neurosurgery, Thompson RA, Green JR, and Johnsen SD, eds. (Springer; Netherlands: ), pp. 1–15. 10.1007/978-94-011-7295-0_1. [DOI] [Google Scholar]
- 58.Adzick NS, Harrison MR, Glick PL, Anderson J, Villa RL, Flake AW, and Laberge JM (1986). Fetal surgery in the primate. III. Maternal outcome after fetal surgery. J. Pediatr. Surg 21, 477–480. 10.1016/s0022-3468(86)80215-9. [DOI] [PubMed] [Google Scholar]
- 59.An M, Raguram A, Du SW, Banskota S, Davis JR, Newby GA, Chen PZ, Palczewski K, and Liu DR (2024). Engineered virus-like particles for transient delivery of prime editor ribonucleoprotein complexes in vivo. Nat. Biotechnol 42, 1526–1537. 10.1038/s41587-023-02078-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 60.Hamilton JR, Chen E, Perez BS, Sandoval Espinoza CR, Kang MH, Trinidad M, Ngo W, and Doudna JA (2024). In vivo human T cell engineering with enveloped delivery vehicles. Nat. Biotechnol 42, 1684–1692. 10.1038/s41587-023-02085-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 61.Mallapaty S. (2025). Rare genetic disorder treated in womb for the first time. Nature 638, 869. 10.1038/d41586-025-00534-0. [DOI] [PubMed] [Google Scholar]
- 62.Newby GA, Yen JS, Woodard KJ, Mayuranathan T, Lazzarotto CR, Li Y, Sheppard-Tillman H, Porter SN, Yao Y, Mayberry K, et al. (2021). Base editing of haematopoietic stem cells rescues sickle cell disease in mice. Nature 595, 295–302. 10.1038/s41586-021-03609-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 63.Regev L, Ezrielev E, Gershon E, Gil S, and Chen A. (2010). Genetic approach for intracerebroventricular delivery. Proc. Natl. Acad. Sci. USA 107, 4424–4429. 10.1073/pnas.0907059107. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 64.Jang A, and Lehtinen MK (2023). In utero intracerebroventricular delivery of adeno-associated viral vectors to target mouse choroid plexus and cerebrospinal fluid. STAR Protoc. 4, 101975. 10.1016/j.xpro.2022.101975. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 65.Jang A, and Lehtinen MK (2022). Experimental approaches for manipulating choroid plexus epithelial cells. Fluids Barriers CNS 19, 36. 10.1186/s12987-022-00330-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 66.Chen SH, Haam J, Walker M, Scappini E, Naughton J, and Martin NP (2019). Recombinant Viral Vectors as Neuroscience Tools. Curr. Protoc. Neurosci 87, e67. 10.1002/cpns.67. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 67.Matsuzaki Y, Fukai Y, Konno A, and Hirai H. (2024). Optimal different adeno-associated virus capsid/promoter combinations to target specific cell types in the common marmoset cerebral cortex. Mol. Ther. Methods Clin. Dev 32, 101337. 10.1016/j.omtm.2024.101337. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 68.Watakabe A, Ohtsuka M, Kinoshita M, Takaji M, Isa K, Mizukami H, Ozawa K, Isa T, and Yamamori T. (2015). Comparative analyses of adeno-associated viral vector serotypes 1, 2, 5, 8 and 9 in marmoset, mouse and macaque cerebral cortex. Neurosci. Res 93, 144–157. 10.1016/j.neures.2014.09.002. [DOI] [PubMed] [Google Scholar]
- 69.Srivastava A. (2016). In vivo tissue-tropism of adeno-associated viral vectors. Curr. Opin. Virol 21, 75–80. 10.1016/j.coviro.2016.08.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 70.Bandtlow CE, and Zimmermann DR (2000). Proteoglycans in the developing brain: new conceptual insights for old proteins. Physiol. Rev 80, 1267–1290. 10.1152/physrev.2000.80.4.1267. [DOI] [PubMed] [Google Scholar]
- 71.Lee B, and An HJ (2024). Small but big leaps towards neuroglycomics: exploring N-glycome in the brain to advance the understanding of brain development and function. Neural Regen. Res 19, 489–490. 10.4103/1673-5374.380887. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 72.Rutherford JN, deMartelly VA, Layne Colon DG, Ross CN, and Tardif SD (2014). Developmental origins of pregnancy loss in the adult female. PLoS One 9, e96845. 10.1371/journal.pone.0096845. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 73.Tardif SD, Smucny DA, Abbott DH, Mansfield K, Schultz-Darken N, and Yamamoto ME (2003). Reproduction in captive common marmosets (Callithrix jacchus). Comp. Med 53, 364–368. [PubMed] [Google Scholar]
- 74.Ose T, Autio JA, Ohno M, Frey S, Uematsu A, Kawasaki A, Takeda C, Hori Y, Nishigori K, Nakako T, et al. (2022). Anatomical variability, multi-modal coordinate systems, and precision targeting in the marmoset brain. Neuroimage 250, 118965. 10.1016/j.neuroimage.2022.118965. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 75.Kita Y, Nishibe H, Wang Y, Hashikawa T, Kikuchi SS, U, M., Yoshida, A.C., Yoshida, C., Kawase, T., Ishii, S., et al. (2021). Cellular-resolution gene expression profiling in the neonatal marmoset brain reveals dynamic species- and region-specific differences. Proc. Natl. Acad. Sci. USA 118, e2020125118. 10.1073/pnas.2020125118. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 76.Shimogori T, Abe A, Go Y, Hashikawa T, Kishi N, Kikuchi SS, Kita Y, Niimi K, Nishibe H, Okuno M, et al. (2018). Digital gene atlas of neonate common marmoset brain. Neurosci. Res 128, 1–13. 10.1016/j.neures.2017.10.009. [DOI] [PubMed] [Google Scholar]
- 77.Kremer LPM, Cerrizuela S, Dehler S, Stiehl T, Weinmann J, Abendroth H, Kleber S, Laure A, El Andari J, Anders S, et al. (2021). High throughput screening of novel AAV capsids identifies variants for transduction of adult NSCs within the subventricular zone. Mol. Ther. Methods Clin. Dev 23, 33–50. 10.1016/j.omtm.2021.07.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 78.Oerke AK, Einspanier A, and Hodges JK (1995). Detection of pregnancy and monitoring patterns of uterine and fetal growth in the marmoset monkey (Callithrix jacchus) by real-time ultrasonography. Am. J. Primatol 36, 1–13. 10.1002/ajp.1350360102. [DOI] [PubMed] [Google Scholar]
- 79.Howard DB, Powers K, Wang Y, and Harvey BK (2008). Tropism and toxicity of adeno-associated viral vector serotypes 1, 2, 5, 6, 7, 8, and 9 in rat neurons and glia in vitro. Virology 372, 24–34. 10.1016/j.virol.2007.10.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 80.Tyas DA, Pratt T, Simpson TI, Mason JO, and Price DJ (2003). Identifying GFP-transgenic animals by flashlight. Biotechniques 34, 474–476. 10.2144/03343bm04. [DOI] [PubMed] [Google Scholar]
- 81.Bandeira F, Lent R, and Herculano-Houzel S. (2009). Changing numbers of neuronal and non-neuronal cells underlie postnatal brain growth in the rat. Proc. Natl. Acad. Sci. USA 106, 14108–14113. 10.1073/pnas.0804650106. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 82.Park YG, Sohn CH, Chen R, McCue M, Yun DH, Drummond GT, Ku T, Evans NB, Oak HC, Trieu W, et al. (2018). Protection of tissue physicochemical properties using polyfunctional crosslinkers. Nat. Biotechnol 37, 73–83. 10.1038/nbt.4281. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 83.Susaki EA, Tainaka K, Perrin D, Yukinaga H, Kuno A, and Ueda HR (2015). Advanced CUBIC protocols for whole-brain and whole-body clearing and imaging. Nat. Protoc 10, 1709–1727. 10.1038/nprot.2015.085. [DOI] [PubMed] [Google Scholar]
- 84.de Rooi JJ, Devos O, Sliwa M, Ruckebusch C, and Eilers PHC (2013). Mixture models for two-dimensional baseline correction, applied to artifact elimination in time-resolved spectroscopy. Anal. Chim. Acta 771, 7–13. 10.1016/j.aca.2013.02.007. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
All data reported in this paper will be shared by the lead contact upon request.
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.