Fast, high-throughput production of improved rabies viral vectors for specific, efficient and versatile transsynaptic retrograde labeling

Curation statements for this article:
  • Curated by eLife

    eLife logo

    Evaluation Summary:

    Rabies-mediated monosynaptic retrograde tracing is a powerful method to characterize the connectivity of neural circuits. The CVS-N2c strain of rabies virus shows significantly higher efficiency of transsynaptic spread and less toxicity than the more commonly used SAD B19 strain but has been limited in use by an arduous and lengthy packaging process and low resultant titers. Here, Sumser et al. present a method that significantly speeds up the production process while reducing off-target expression. They also introduce a suite of novel reagents (34 novel plasmids) for monosynaptic tracing with the CVS-N2c strain that they commendably, have already deposited with Addgene. The work is an important advance that will reinvigorate rabies-mediated circuit tracing.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #2 agreed to share their name with the authors.)

This article has been Reviewed by the following groups

Read the full article See related articles

Abstract

To understand the function of neuronal circuits, it is crucial to disentangle the connectivity patterns within the network. However, most tools currently used to explore connectivity have low throughput, low selectivity, or limited accessibility. Here, we report the development of an improved packaging system for the production of the highly neurotropic RVdG envA -CVS-N2c rabies viral vectors, yielding titers orders of magnitude higher with no background contamination, at a fraction of the production time, while preserving the efficiency of transsynaptic labeling. Along with the production pipeline, we developed suites of ‘starter’ AAV and bicistronic RVdG-CVS-N2c vectors, enabling retrograde labeling from a wide range of neuronal populations, tailored for diverse experimental requirements. We demonstrate the power and flexibility of the new system by uncovering hidden local and distal inhibitory connections in the mouse hippocampal formation and by imaging the functional properties of a cortical microcircuit across weeks. Our novel production pipeline provides a convenient approach to generate new rabies vectors, while our toolkit flexibly and efficiently expands the current capacity to label, manipulate and image the neuronal activity of interconnected neuronal circuits in vitro and in vivo.

Article activity feed

  1. Author Response

    Reviewer #3 (Public Review):

    This work addressed some of the limitations in the production of the CVS-N2c strain of the rabies virus. CVS-N2c exhibits lower cytotoxicity and more efficient transsynaptic spread than the more widely used SAD-B19 strain, but its use as a circuit tracing tool has been held back by its slow packaging process and low resultant titers. By demonstrating that rabies packaging cell lines do not affect retrograde labelling efficiency and by creating a pseudotyping cell line that can amplify pseudotyped virus from a small amount of starting material, Sumser et. al have achieved an improvement in the speed, titer, and native coat contamination of CVS-N2c preparations whilst generating a new set of viral vectors that will help to implement a range of circuit mapping tools.

    While many of the results from N2c evaluation experiments shown here (including bicistronic rabies usage, time courses of functional characterisation with GCaMP/channelrhodopsin, Cre-OFF labelling) have been previously demonstrated in other N2c and SAD-B19 rabies studies, the suite of vectors described in the manuscript will serve as a useful resource for the community. However, some key aspects of these vectors, specifically the propensity for the starter AAV for off-target labelling, are not characterized.

    We thank the reviewer for his / her positive comments (“improvement”, “useful resource”).

    1. The six DIO AAV vectors described here, and the Flp-dependent AAV-FRT-EF1a-TVA-2A-N2cG do not contain recombinase leak prevention mechanisms such as the "ATG-out" approach, where the initiating codon and Kozak sequence are moved outside of the recombinase recognition sites to reduce inverted ORF expression. Even with these measures in place, DIO constructs are prone to recombinase-independent reversion, proposed to occur during AAV production from spontaneous recombination (Fischer et al., 2019). This presents an issue for the sensitive TVA/EnvA system, where only a small amount of TVA expression can mediate off-target rabies infection in non-Cre expressing cells. The dilution of AAV vectors can have a strong effect on the amount of non-specific labelling (Lavin et al., 2020). As the bicistronic TVA-N2cG vectors used here do not allow for individual dilutions of TVA with respect to N2cG, which is required at higher expression levels for efficient transsynaptic spread, it is especially important to test these vectors for leak expression. A sensitive test for starter leak would be to inject the AAV and rabies virus in WT mice.

    We are aware of these potential issues with the AAV targeting system. We can confirm that in our hands, all vectors have been thoroughly vetted, and no leak has been found for any. For each new viral batch produced (37 in total) we have conducted experiments in which pseudotyped particles at work concentrations (1–5 × 108 TU/ml) were injected into naïve brains and have found only minimal non-specific labeling (less than 1–2 cells per 5–10 100-µm-thick sections examined).. We have added a sentence to the Materials and Methods section (p. 5 of the revised manuscript).

    Furthermore, we have tested the effects of extremely high viral titers are (100–500-fold higher than the titers used throughout the manuscript). We found that even with these extremely high virus concentrations, contamination was minimal. Furthermore, we found that the negligible contamination in these experiments likely arises from direct penetration of pseudotyped particles into damaged cells and cell processes along the needle tract. We have added a new Figure 1 – figure supplement 2 and added a sentence on p. 5 of the revised manuscript.

    We agree that the system might benefit from additional safety measures. However, these potential safety measures could also introduce new problems. In finding the right balance, one should take into account that such leak events are rare, and their random occurrence should make it possible to distinguish them from the main effects by replication.

    1. The manuscript reports the use of a bicistronic N-P system to express the optogenetic actuator ChIEF together with a fluorescence protein. While the results of the bicistronic experiments show that both proteins are successfully expressed, control experiments using other expression strategies would strengthen the claim that the bicistronic N-P system is superior.

    In our work, we compare the activation effectiveness of the optogenetic actuator to previously published results from two separate manuscripts (Reardon et al., 2016; Osakada et al., 2011), using comparable light intensity. Our results demonstrate that our vectors achieve higher AP success rates with substantially shorter light pulses. While we have not conducted a direct comparison, we think that the improvement presented in the bicistronic vectors is sufficiently substantiated, and the logic behind this improvement sufficiently sound. Since this is a lateral point in our manuscript but a relevant information for the community, we believe that this point is worth mentioning even without a direct comparison.

  2. Evaluation Summary:

    Rabies-mediated monosynaptic retrograde tracing is a powerful method to characterize the connectivity of neural circuits. The CVS-N2c strain of rabies virus shows significantly higher efficiency of transsynaptic spread and less toxicity than the more commonly used SAD B19 strain but has been limited in use by an arduous and lengthy packaging process and low resultant titers. Here, Sumser et al. present a method that significantly speeds up the production process while reducing off-target expression. They also introduce a suite of novel reagents (34 novel plasmids) for monosynaptic tracing with the CVS-N2c strain that they commendably, have already deposited with Addgene. The work is an important advance that will reinvigorate rabies-mediated circuit tracing.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #2 agreed to share their name with the authors.)

  3. Reviewer #1 (Public Review):

    The authors have identified a novel method to generate the N2C strain of the rabies virus in a significantly shorter period of time and with less contamination of WT rabies. Additionally, they have developed a suite of tools that can be used with the rabies. Lastly, they provide examples of the utilization of rabies encoding tools that can be used for functional studies post rabies infection. While the work is not extremely novel, it takes us past problems that were impediments to the use of the N2C strain which is more infectious than the SAD-B19 strain and will likely invigorate monosynaptic retrograde tracing, which is powerful for understanding circuit connectivity.

  4. Reviewer #2 (Public Review):

    Sumser et al. present and validate a suite of new reagents for monosynaptic tracing, most notably for the production of the required rabies viral vectors, and specifically of vectors of the CVS-N2c strain, which are superior to those of the more commonly used SAD B19 strain but which have so far been notoriously more difficult to make. The manuscript and many new reagents are significant contributions to the neuroscience community. Commendably, the authors have already deposited all of the 34 novel plasmids with Addgene. Reagents aside, the demonstration that ∆G CVS-N2c can be grown to high titers quickly on BHK cells co-expressing EnvA and TVA, with much less contamination by "unpseudotyped" virus and without losing their in vivo effectiveness, is an important advance.

  5. Reviewer #3 (Public Review):

    This work addressed some of the limitations in the production of the CVS-N2c strain of the rabies virus. CVS-N2c exhibits lower cytotoxicity and more efficient transsynaptic spread than the more widely used SAD-B19 strain, but its use as a circuit tracing tool has been held back by its slow packaging process and low resultant titers. By demonstrating that rabies packaging cell lines do not affect retrograde labelling efficiency and by creating a pseudotyping cell line that can amplify pseudotyped virus from a small amount of starting material, Sumser et. al have achieved an improvement in the speed, titer, and native coat contamination of CVS-N2c preparations whilst generating a new set of viral vectors that will help to implement a range of circuit mapping tools.

    While many of the results from N2c evaluation experiments shown here (including bicistronic rabies usage, time courses of functional characterisation with GCaMP/channelrhodopsin, Cre-OFF labelling) have been previously demonstrated in other N2c and SAD-B19 rabies studies, the suite of vectors described in the manuscript will serve as a useful resource for the community. However, some key aspects of these vectors, specifically the propensity for the starter AAV for off-target labelling, are not characterized.

    1. The six DIO AAV vectors described here, and the Flp-dependent AAV-FRT-EF1a-TVA-2A-N2cG do not contain recombinase leak prevention mechanisms such as the "ATG-out" approach, where the initiating codon and Kozak sequence are moved outside of the recombinase recognition sites to reduce inverted ORF expression. Even with these measures in place, DIO constructs are prone to recombinase-independent reversion, proposed to occur during AAV production from spontaneous recombination (Fischer et al., 2019). This presents an issue for the sensitive TVA/EnvA system, where only a small amount of TVA expression can mediate off-target rabies infection in non-Cre expressing cells. The dilution of AAV vectors can have a strong effect on the amount of non-specific labelling (Lavin et al., 2020). As the bicistronic TVA-N2cG vectors used here do not allow for individual dilutions of TVA with respect to N2cG, which is required at higher expression levels for efficient transsynaptic spread, it is especially important to test these vectors for leak expression. A sensitive test for starter leak would be to inject the AAV and rabies virus in WT mice.
    2. The manuscript reports the use of a bicistronic N-P system to express the optogenetic actuator ChIEF together with a fluorescence protein. While the results of the bicistronic experiments show that both proteins are successfully expressed, control experiments using other expression strategies would strengthen the claim that the bicistronic N-P system is superior.