Tracking the spread of insecticide resistance in Anopheles gambiae populations

Alistair Miles (@alimanfoo)

MalariaGEN Resource Centre
University of Oxford / Wellcome Sanger Institute

6 June 2019 - WHO/Global Malaria Programme

These slides: http://alimanfoo.github.io/slides/20190606-who-geneva.html

## Use cases for genomic surveillance of malaria vector populations Pyrethroid resistance is widespread in primary African vector species. How can surveillance improve insecticide resistance management (IRM)?

## Use case (1): Procurement and deployment of next-generation LLINs * "Next-generation" LLINs available, e.g.: * Olyset Plus: permethrin + PBO (P450 synergist) * Olyset Duo: permethrin + pyriproxyfen (second insecticide) * More expensive than standard LLINs * How many to buy? * Where to deploy them?

## Use case (2): IRS deployment strategy * "Next-generation" IRS formulations available, e.g.: * Actellic 300CS: pyrimiphos methyl (organophosphate) * SumiShield 50WG: clothianidin (neonicotinoid) * Fludora Fusion: deltamethrin + clothianidin * Preemptive rotation? * Geographical mosaic? * Is it working?

## Use case (3): Cross-border coordination * Can countries take decisions in isolation about how to manage insecticide resistance? * When and where do decisions need to be coordinated across borders?

## The *Anopheles gambiae* 1000 Genomes Project (Ag1000G) * A consortial project using whole-genome sequencing to investigate genetic variation and evolution in natural mosquito populations * Create an open access data resource to accelerate research and surveillance * [www.malariagen.net/ag1000g](http://www.malariagen.net/ag1000g)

Ag1000G Consortium

Wellcome Sanger Institute / University of Oxford / Liverpool School of Tropical Medicine / Sapienza University of Rome / University of California, Riverside / Liverpool John Moores University / Broad Institute / Institut de Recherche pour le Développement / Virginia Tech / KEMRI Wellcome Trust Research Programme / New Mexico State University / Universidade Nova de Lisboa / University of Minnesota / Université d’Abomey–Calavi, Benin / Indiana University / University of Notre Dame / Washington State University / Imperial College / University of Oregon / University of North Carolina at Chapel Hill / University of Montana / Institut Pasteur / Instituto Nacional de Saúde Pública, Guiné-Bissau / Centre International de Recherches Médicales de Franceville, Gabon / Programa Nacional de Controle da Malária, Angola / Institut de Recherche en Sciences de la Santé, Burkina Faso / University of Bamako, Mali / Infectious Diseases Research Collaboration, Uganda / Organisation de Coordination pour la lutte contre les Endémies en Afrique Centrale, Cameroon

## Ag1000G sequencing methods * Sequence individual mosquitoes collected from the field * Use whole-genome Illumina (Hi-Seq) sequencing * Deep coverage (~30X) * Sequencing performed at and funded by Wellcome Sanger Institute

## Ag1000G population sampling * Aim for broad geographical coverage * 18 countries, ~1 site per country * *An. gambiae*, *An. coluzzii*, *An. arabiensis* * Sequence >30 individuals per site per species * Why 30? Statistical power to make inferences about populations (e.g., gene flow).

## Ag1000G data production * Raw sequence reads → * Alignment to reference genome → * Variant calling → * Variant filtering and annotation → * Haplotype phasing → * **Curated "analysis-ready" variant calls and haplotypes** * Validation, e.g., <1% FDR

## Ag1000G data releases * **Phase 1**: 765 mosquitoes; 8 countries; *An. gambiae*, *An. coluzzii* * Data released 2016 * **Phase 2**: 1,142 mosquitoes; 13 countries; *An. gambiae*, *An. coluzzii* * Data released 2017 * **Phase 3**: ~4,000 mosquitoes; 18 countries; *An. gambiae*, *An. coluzzii*, *An. arabiensis* * Data in production

## Ag1000G further information * www.malariagen.net/ag1000g * https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6026373/

## Gene flow * Between locations * Between species * Between generations (i.e., increasing in frequency, i.e., under selection)

Genes under selection

## Pyrethroid target-site resistance Spread of "knock-down resistance" (*kdr*) mutations in the voltage-gated sodium channel gene (*Vgsc*).

## *kdr* mutations * Two known *kdr* mutations in *Vgsc* codon 1014 * L1014F found throughout West and Central Africa * L1014S found throughout East and Central Africa * Are these mutations spreading? * Where is gene flow occurring?

## Inferring *kdr* gene flow * Analyse the genetic backgrounds on which *kdr* mutations occur ("*kdr* haplotypes") * Use all mutations within the *Vgsc* gene * 1,710 biallelic SNPs (mostly intronic) * Same *kdr* haplotype in two different locations: * ⇒ gene flow

Inferring kdr gene flow

## *kdr* haplotypes * Analysed data from Ag1000G phase 1 (765 mosquitoes, 8 countries) * L1014F - 5 major haplotypes (F1-F5) * L1014S - 5 major haplotypes (S1-S5)

kdr haplotypes

## *kdr* gene flow - further information https://doi.org/10.1101/323980

## Pyrethroid metabolic resistance Spread of copy number variations in cytochrome P450 genes.

## Cytochrome P450 genes * Known to metabolise pyrethroids * Neutralised by PBO synergist in next-gen LLINs * Multiple P450 genes in genome, e.g.: * *Cyp6p/aa* * *Cyp9k1* * Increased expression ⇒ pyrethroid resistance * Increased gene copy number ⇒ increased expression

## Detecting copy number variation

## P450 copy number variation * Analysed data from Ag1000G phase 2 (1,142 mosquitoes, 13 countries) * Gene amplifications are common at two P450 loci: * *Cyp6p/aa* * *Cyp9k1*

## *Cyp6p/aa* CNV gene flow * Dup1 - BFcol (8%), UGgam (58%) * Dup7 - BFcol (44%), CIcol (32%), GHcol (5%), GNcol (75%) * Dup8 - BFgam (3%), GNgam (3%) * Dup10 - BFcol (49%), GHcol (5%) * Dup11 - CIcol (41%), GHcol (5%) * Dup14 - BFcol (3%), CIcol (46%) * Dup15 - BFcol (1%), CIcol (39%)

## CNVs further information https://doi.org/10.1101/399568

## Summary & discussion * Target-site (*kdr*) and metabolic (P450 CNV) pyrethroid resistance are spreading via gene flow * Multiple independent outbreaks of resistance * Some spreading, some localised * Long distance gene flow, e.g.: * *kdr*-F1 found in GN, BF, CM and AO * *Cyp6p*-Dup1 found in BF and UG * *kdr*, *Cyp6p/aa* and *Cyp9k1* show different patterns of spread

Where to deploy PBO LLINs?

Cf. Ebola outbreaks

## Resistance outbreaks * Geographical origins? * Direction and routes of gene flow? * Timing? * Rate of movement?

## Next steps * Scale up genome sequencing of vector populations * Increase geographical coverage * Regular (seasonal) sampling * Other vector species (e.g., *An. funestus*)

## MalariaGEN Vector Observatory * Aim to sequence 10,000 mosquitoes per year * Coupled with routine ento surveillance * Follow sentinel sites over time * Link genomic and epi/ento data * Partnerships * PAMCA/BMGF, GAARDian, Target Malaria, ... * Open data * [Bridge the gap](https://yourshot.nationalgeographic.com/photos/8198269/) between research and implementation

Acknowledgements

## Extra slides

## *Cyp9k1* CNV gene flow * Dup4 - BFgam, CMgam, GHgam, GNgam * Dup7 - BFgam, CMgam * Dup10 - BFcol, GHcol * Dup11 - BFgam, CMgam, GHgam, GNgam * Dup12 - GM, GW * Dup13 - BFgam, GNgam * Dup15 - BFgam, GHgam, GNgam

## Gene drive use cases * Design of gene drive constructs - identification of ultra-conserved regions * Modelling gene drive deployment and spread - estimating rates, ranges and routes of migration * Spread of insecticide resistance as a model for spread of gene drive * Monitoring gene drive in the field - is it spreading as expected, is resistance emerging?