Anopheles homing suppression drive candidates exhibit unexpected performance differences in simulations with spatial structure
Curation statements for this article:-
Curated by eLife
Evaluation Summary:
This is one of the most thorough assessments to date of suppression gene drives against mosquitoes. The models specifically consider the spatial dynamics of gene drives and whether a form of group selection may prevent the drive from eradicating the population, with mosquito ecology parameters. This manuscript will be of interest to those working in the technical development of gene drives, those predicting how such genetically modified insects would spread in the wild, and those evaluating the technology from regulatory and funding standpoints.
(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 #1 and Reviewer #3 agreed to share their name with the authors.)
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
Abstract
Recent experiments have produced several Anopheles gambiae homing gene drives that disrupt female fertility genes, thereby eventually inducing population collapse. Such drives may be highly effective tools to combat malaria. One such homing drive, based on the zpg promoter driving CRISPR/Cas9, was able to eliminate a cage population of mosquitoes. A second version, purportedly improved upon the first by incorporating an X-shredder element (which biases inheritance towards male offspring), was similarly successful. Here, we analyze experimental data from each of these gene drives to extract their characteristics and performance parameters and compare these to previous interpretations of their experimental performance. We assess each suppression drive within an individual-based simulation framework that models mosquito population dynamics in continuous space. We find that the combined homing/X-shredder drive is actually less effective at population suppression within the context of our mosquito population model. In particular, the combined drive often fails to completely suppress the population, instead resulting in an unstable equilibrium between drive and wild-type alleles. By contrast, otherwise similar drives based on the nos promoter may prove to be more promising candidates for future development than originally thought.
Article activity feed
-
-
Author Response
Reviewer #1 (Public Review):
Champer et al. evaluate two homing drives that have been developed in the Anopheles mosquito. Variants of one of these (zpg) are possibly being further investigated for an eventual release. Work with the other has seemingly been discontinued because of unintended fitness costs. The authors argue that this second drive may be in fact better if the experimental results are interpreted more favourably. An important point if true, but somewhat separate from the findings in the paper. To a large extent, this point could be made without any of the results in the paper. However, the authors do show through modelling that this difference may in fact be relevant.
This careful justification of the model parameters increases its relevance to the evaluation of those specific gene drives. The zpg …
Author Response
Reviewer #1 (Public Review):
Champer et al. evaluate two homing drives that have been developed in the Anopheles mosquito. Variants of one of these (zpg) are possibly being further investigated for an eventual release. Work with the other has seemingly been discontinued because of unintended fitness costs. The authors argue that this second drive may be in fact better if the experimental results are interpreted more favourably. An important point if true, but somewhat separate from the findings in the paper. To a large extent, this point could be made without any of the results in the paper. However, the authors do show through modelling that this difference may in fact be relevant.
This careful justification of the model parameters increases its relevance to the evaluation of those specific gene drives. The zpg drive will likely be extensively investigated and the specific relevance of this work is a valuable contribution. While a range of parameters is tested for each expression pattern, there are no step-by-step investigations of how the drive outcomes are effect by changes to the underlying DNA-repair/deposition/fitness parameters. So while a reader may learn one drive is better than the other, the ability to get a deeper understanding of the underlying relationship is limited. This means this work has a more limited scope and relies on the relevance of the chosen parameters. In that regard, there may be room for improvement. The chosen parameters for zpg and nos may not be completely fair in regards to the target site and I believe this needs to be addressed.
The second aspect of this paper is the comparison between the commonly used panmictic modelling approach and spacial models. This also somewhat relies on the drive parameters being chosen well, as a more comprehensive evaluation of the spacial approach has been done in prior work by this group. However, showing that these particular extremely efficient drives may still struggle when additional spacial factors are considered is useful and relevant. That a second Anopheles-specific spacial model further reduces the drive performance is a relevant finding. This is helped by a specific analysis of the effect of changes to the migration rates and the low-density growth rate. This spacial modelling also has relevant findings for the homing X-shredder design.
In our previous study (Champer, Kim, et al, 2021 in Molecular Ecology) that we reference, we varied some of these drive performance parameters, which may address some of the reviewer’s concerns. We view this study as building off that one, but with a more specific focus (mosquitoes and existing drives). We also now discuss how using parameters for a different target site may have affected our results (see below - nos may actually have been shortchanged since zpg performs better at dsx than at nudel).
Reviewer #2 (Public Review):
Champer and colleagues present forward simulations of several gene drive systems that have been designed to suppress the malaria mosquito, Anopheles gambiae. These gene drives have all been validated in laboratory cage experiments but have not yet progressed to field trials. The authors are particularly concerned with the phenomenon of "chasing," in which local success of the drive will lead to continuous cycles of recolonization by wildtype mosquitoes, preventing complete suppression of the population. In addition to their spatially-explicit model, they additionally present results from a model in which the parameters are tuned to the ecology of the mosquito.
Though there are a few additions that would improve the manuscript, the authors achieved these aims and their conclusions are supported by their modeling, which appears to be technically sound and well executed.
Strengths:
The work represents a useful, model-based comparison of the various Anopheles gene drives that have had success in laboratory conditions. With the incorporation of spatial dynamics, the authors are thereby able to focus on the problem of chasing, or a fluctuating equilibrium state that is impossible to study in laboratory colonies. Through a comparative framework, the authors additionally provide key information on the differences in predicted success between the various gene drive systems. For these reasons, the work will be a useful addition to the other published forecasts of gene drive success. Given the importance of the topic to diverse stakeholders that vary in their familiarity with gene drives and ecological modeling, I was glad to see the authors summarize their findings cogently and accessibly.
We thank the reviewer for these kind comments.
Weaknesses:
The main area in which the manuscript could be strengthened is the description of the Anopheles-specific model. Based in part on the differences observed between their discrete generation model and Anopheles-specific model, the authors correctly note that "the outcome of a drive release could be very sensitive to the precise ecological characteristics of the targeted population." It was sometimes unclear which model parameter choices were informed by literature and how much confidence was had in each. A more explicit summary or perhaps a table of parameters, references, and estimates with confidence ranges informed by the authors' knowledge of literature would strengthen this section.
We now have added Table S1 showing all model parameters. The parameters themselves often require more text for justification than just references, but we have improved our methods section throughout to increase the visibility and clarity of these sections. Note that mosquito ecology is a fairly understudied field, resulting in widely varying parameter ranges throughout different studies, so it is difficult to provide confidence ranges for our parameters, other than that they are designed to fall within estimates from different studies (see “Anopheles-specific spatial model” methods subsection).
In the framing of the work, the authors imply that their modeling study "suggest(s) an alternative interpretation of [the] performance [of homing gene drives]" from recent studies (e.g., Simoni et al. 2020 and Kyrou et al. 2018). I am not certain this framing is justified, given the original authors' circumspection in correctly noting their drives had success in the cage experiments, without claiming they would be successful in the wild. I would prefer this study be presented as building on those previous studies and extending their work.
In crafting this sentence, we had in mind very specific technical interpretations of drive performance mechanisms (paternal deposition vs. more somatic fitness cost, existing of somatic fitness cost in nos males) rather than general performance in any given environment. To more clearly convey our intended meaning here, we have adjusted our wording. The sentence now reads: “Here, we analyze data associated with each of these gene drives and consider both the original and alternative interpretations of these drives’ characteristics and performance parameters.”
Likely impact:
This work will be of interest to research scientists whose interests range from transgenic mosquitoes to ecological modelers to post-release assessment. The authors correctly note that additional refinement of the ecological parameters will increase the utility of the model, but the framework as it stands will be an important contribution to the literature. Given the timeliness of this topic, the subject is of interest to other stakeholders in the regulatory or policymaking realm, as well as governmental and funding agencies deciding between gene drive systems.
Reviewer #3 (Public Review):
This is a computational modeling study to evaluate the merits (likely success) of different 'suppression' gene drive systems. Gene drives offer a possible simple and low-effort means of suppressing or even extinguishing pest populations. Using CRISPR technology, several gene drive systems have been developed in the last decade for key mosquito vector species. As no gene drive has been approved for release in the wild, efforts to evaluate their likely success are limited to cage trials and modeling, the latter as done here. In contrast to some modeling studies, the effort here is to develop and analyze models that match the gene drive and mosquito biology closely. The models are thus parameterized with values representative of what is known about mosquito biology and of the various gene drive constructs that have been developed for lab studies.
In these models, gene drive success or failure in population suppression largely depends on (i) how well the drive spreads throughout the population, and (ii) whether the population persists because of a type of ongoing spatial 'group selection' in which local pockets invaded by the drive die out and are then repopulated by migrants lacking the drive. Formal evolution of functional resistance is not allowed. The numerical results show striking differences in suppression success with different gene drive constructions, and these differences are likely to be of use when designing drives for actual releases.
The basic group selection outcome that allows population persistence amid a suppression gene drive has been shown before, as cited in the ms. The novelty provided by the present study is to tie the models to the biology of known gene drive constructions. Given the high specificity of the models, the audience for this work is likely to be somewhat narrow, confined to those involved in gene drive design. The work is nonetheless significant in view of the strong potential of gene drives in global public health efforts.
The software used to generate the trials is freely available from one of the authors for anyone wishing to repeat the simulations. There is an extensive supplement of results referenced (but not otherwise included) in the main text.
We thank the reviewer for these comments, and we note that an analysis of functional resistance was performed in our previous study. Because the results of this study are likely to be fully applicable to our new results, we did not repeat it with our mosquito model and with our parameterized drives. However, it is certainly an important topic, and we explicitly mention in the discussion how the possibility of chasing requires further consideration in the acceptable rate of functional resistance allele formation. The text reads, “We did not consider the possibility of.... functional resistance, which can evolve more readily during lengthy chases11.”
-
Evaluation Summary:
This is one of the most thorough assessments to date of suppression gene drives against mosquitoes. The models specifically consider the spatial dynamics of gene drives and whether a form of group selection may prevent the drive from eradicating the population, with mosquito ecology parameters. This manuscript will be of interest to those working in the technical development of gene drives, those predicting how such genetically modified insects would spread in the wild, and those evaluating the technology from regulatory and funding standpoints.
(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 #1 and Reviewer #3 agreed to share their name with the authors.)
-
Reviewer #1 (Public Review):
Champer et al. evaluate two homing drives that have been developed in the Anopheles mosquito. Variants of one of these (zpg) are possibly being further investigated for an eventual release. Work with the other has seemingly been discontinued because of unintended fitness costs. The authors argue that this second drive may be in fact better if the experimental results are interpreted more favourably. An important point if true, but somewhat separate from the findings in the paper. To a large extent, this point could be made without any of the results in the paper. However, the authors do show through modelling that this difference may in fact be relevant.
This careful justification of the model parameters increases its relevance to the evaluation of those specific gene drives. The zpg drive will likely be …
Reviewer #1 (Public Review):
Champer et al. evaluate two homing drives that have been developed in the Anopheles mosquito. Variants of one of these (zpg) are possibly being further investigated for an eventual release. Work with the other has seemingly been discontinued because of unintended fitness costs. The authors argue that this second drive may be in fact better if the experimental results are interpreted more favourably. An important point if true, but somewhat separate from the findings in the paper. To a large extent, this point could be made without any of the results in the paper. However, the authors do show through modelling that this difference may in fact be relevant.
This careful justification of the model parameters increases its relevance to the evaluation of those specific gene drives. The zpg drive will likely be extensively investigated and the specific relevance of this work is a valuable contribution. While a range of parameters is tested for each expression pattern, there are no step-by-step investigations of how the drive outcomes are effect by changes to the underlying DNA-repair/deposition/fitness parameters. So while a reader may learn one drive is better than the other, the ability to get a deeper understanding of the underlying relationship is limited. This means this work has a more limited scope and relies on the relevance of the chosen parameters. In that regard, there may be room for improvement. The chosen parameters for zpg and nos may not be completely fair in regards to the target site and I believe this needs to be addressed.
The second aspect of this paper is the comparison between the commonly used panmictic modelling approach and spacial models. This also somewhat relies on the drive parameters being chosen well, as a more comprehensive evaluation of the spacial approach has been done in prior work by this group. However, showing that these particular extremely efficient drives may still struggle when additional spacial factors are considered is useful and relevant. That a second Anopheles-specific spacial model further reduces the drive performance is a relevant finding. This is helped by a specific analysis of the effect of changes to the migration rates and the low-density growth rate. This spacial modelling also has relevant findings for the homing X-shredder design.
-
Reviewer #2 (Public Review):
Champer and colleagues present forward simulations of several gene drive systems that have been designed to suppress the malaria mosquito, Anopheles gambiae. These gene drives have all been validated in laboratory cage experiments but have not yet progressed to field trials. The authors are particularly concerned with the phenomenon of "chasing," in which local success of the drive will lead to continuous cycles of recolonization by wild-type mosquitoes, preventing complete suppression of the population. In addition to their spatially-explicit model, they additionally present results from a model in which the parameters are tuned to the ecology of the mosquito.
Though there are a few additions that would improve the manuscript, the authors achieved these aims and their conclusions are supported by their …
Reviewer #2 (Public Review):
Champer and colleagues present forward simulations of several gene drive systems that have been designed to suppress the malaria mosquito, Anopheles gambiae. These gene drives have all been validated in laboratory cage experiments but have not yet progressed to field trials. The authors are particularly concerned with the phenomenon of "chasing," in which local success of the drive will lead to continuous cycles of recolonization by wild-type mosquitoes, preventing complete suppression of the population. In addition to their spatially-explicit model, they additionally present results from a model in which the parameters are tuned to the ecology of the mosquito.
Though there are a few additions that would improve the manuscript, the authors achieved these aims and their conclusions are supported by their modeling, which appears to be technically sound and well executed.
Strengths:
The work represents a useful, model-based comparison of the various Anopheles gene drives that have had success in laboratory conditions. With the incorporation of spatial dynamics, the authors are thereby able to focus on the problem of chasing, or a fluctuating equilibrium state that is impossible to study in laboratory colonies. Through a comparative framework, the authors additionally provide key information on the differences in predicted success between the various gene drive systems. For these reasons, the work will be a useful addition to the other published forecasts of gene drive success. Given the importance of the topic to diverse stakeholders that vary in their familiarity with gene drives and ecological modeling, I was glad to see the authors summarize their findings cogently and accessibly.
Weaknesses:
The main area in which the manuscript could be strengthened is the description of the Anopheles-specific model. Based in part on the differences observed between their discrete generation model and Anopheles-specific model, the authors correctly note that "the outcome of a drive release could be very sensitive to the precise ecological characteristics of the targeted population." It was sometimes unclear which model parameter choices were informed by literature and how much confidence was had in each. A more explicit summary or perhaps a table of parameters, references, and estimates with confidence ranges informed by the authors' knowledge of literature would strengthen this section.
In the framing of the work, the authors imply that their modeling study "suggest(s) an alternative interpretation of [the] performance [of homing gene drives]" from recent studies (e.g., Simoni et al. 2020 and Kyrou et al. 2018). I am not certain this framing is justified, given the original authors' circumspection in correctly noting their drives had success in the cage experiments, without claiming they would be successful in the wild. I would prefer this study be presented as building on those previous studies and extending their work.
Likely impact:
This work will be of interest to research scientists whose interests range from transgenic mosquitoes to ecological modelers to post-release assessment. The authors correctly note that additional refinement of the ecological parameters will increase the utility of the model, but the framework as it stands will be an important contribution to the literature. Given the timeliness of this topic, the subject is of interest to other stakeholders in the regulatory or policymaking realm, as well as governmental and funding agencies deciding between gene drive systems.
-
Reviewer #3 (Public Review):
This is a computational modeling study to evaluate the merits (likely success) of different 'suppression' gene drive systems. Gene drives offer a possible simple and low-effort means of suppressing or even extinguishing pest populations. Using CRISPR technology, several gene drive systems have been developed in the last decade for key mosquito vector species. As no gene drive has been approved for release in the wild, efforts to evaluate their likely success are limited to cage trials and modeling, the latter as done here. In contrast to some modeling studies, the effort here is to develop and analyze models that match the gene drive and mosquito biology closely. The models are thus parameterized with values representative of what is known about mosquito biology and of the various gene drive constructs that …
Reviewer #3 (Public Review):
This is a computational modeling study to evaluate the merits (likely success) of different 'suppression' gene drive systems. Gene drives offer a possible simple and low-effort means of suppressing or even extinguishing pest populations. Using CRISPR technology, several gene drive systems have been developed in the last decade for key mosquito vector species. As no gene drive has been approved for release in the wild, efforts to evaluate their likely success are limited to cage trials and modeling, the latter as done here. In contrast to some modeling studies, the effort here is to develop and analyze models that match the gene drive and mosquito biology closely. The models are thus parameterized with values representative of what is known about mosquito biology and of the various gene drive constructs that have been developed for lab studies.
In these models, gene drive success or failure in population suppression largely depends on (i) how well the drive spreads throughout the population, and (ii) whether the population persists because of a type of ongoing spatial 'group selection' in which local pockets invaded by the drive die out and are then repopulated by migrants lacking the drive. Formal evolution of functional resistance is not allowed. The numerical results show striking differences in suppression success with different gene drive constructions, and these differences are likely to be of use when designing drives for actual releases.
The basic group selection outcome that allows population persistence amid a suppression gene drive has been shown before, as cited in the ms. The novelty provided by the present study is to tie the models to the biology of known gene drive constructions. Given the high specificity of the models, the audience for this work is likely to be somewhat narrow, confined to those involved in gene drive design. The work is nonetheless significant in view of the strong potential of gene drives in global public health efforts.
The software used to generate the trials is freely available from one of the authors for anyone wishing to repeat the simulations. There is an extensive supplement of results referenced (but not otherwise included) in the main text.
-