Towards and understanding of maize adaptation to highlands using a combination of lipidomics and quantitative and population genetics

After domestication from lowland teosinte in the Mexican subtropics maize colonized the highlands of Mexico and South America. In both Mexican and South American highlands maize encountered lower temperatures and in several volcanic area soils with low phosphorus availability. We hypothesize that these two abiotic stresses were major selective forces during maize adaptation to the highlands and shaped glycerolipid metabolism since low temperature and P deficiency have opposite effects on the relative content of glycerolipid species. We are using this system to explore how metabolic pathways change during the process of local adaptation and how we can use metabolic phenotyping together with quantitative and population genetics to identify loci that might have been under selection during the local adaptation events.

I will present data on a couple of common garden experiments in highland and lowland Mexican field sites where we grew a maize RIL mapping population and a 120 landrace diversity panel composed of highland and lowland maize. From these plants we were able to identify around 125 glycerolipid species that we are using in different ways to to identify loci involved in glycerolipid metabolism I will present these results and in particular on an interesting QTL in ch3 where a candidate phospholipase gene seems to be responsible of phosphatidylcholine (PC) to lysophosphatidylcholine (lyso-PC) conversion.

Our phenotypic data together with population differentiation data suggests that genes controlling PC to lyso-PC conversion were under selection during the process of maize adaptation to the highlands. We are now trying to understand to what extent balancing PC to lyso-PC species might be important to local adaptation in the highlands, the possible convergence of this phenotype in Mexican and South American highland maize and the possibility to apply this knowledge understand how maize adapts to different temperatures and phosphorus levels.

Transcript

1. Towards and understanding of maize adaptation to highlands using a combination of lipidomics and quantitative and population genetics www.rrlab.org Rubén Rellán-Álvarez @rrellanalvarez
2. Plants use two strategies to increase P availability Phosphate Starvation ((Pi Sensing)) Local Response Systemic Response Root System Architecture Phosphate homeostasis
3. Local responses increase top soil root surface exploration area Rellan-Álvarez al . 2015 PubMedID 26287479 shoot/root
4. Recovery Transport Recycling Systemic responses lead to P transport to where is most needed Li et al . 2007 PubMedID 17592130 logFC -P/+P root shoot Sawers et al . Unpublished Veneeklas et al . 2012 PubMedID 22691045 P P P
5. Phospholipids 24 Ester-P 18 Nucleic Acids 40 Inorganic Pi 18 How is phosphorus partitioned in the cell? Veneeklas et al . 2012 PubMedID 22691045
6. Hessen et al . 2010 PubMedID 19796842 P deficiency leads to high selective pressure for high growth rate -> smaller genomes
7. High P Low P Pi Recycling Cold Adaptation Glycerolipids relative abundance change in low P and in low Temperature Lambers et al . 2012 PubMedID 22937909 Degenkolbe et al . 2012 PubMedID 23061922 Phospholipids Sulfolipids Galactolipids High T Low T Phospholipids Sulfolipids Galactolipids
8. Phospholipids are a class of glycerolipids and essential building blocks of membranes Glycerol Polar Head Precursors Fatty Acids Polar Lipids Phospholipids Galactolipids Essential Intermediates Phosphatidic A Diacylglycerol Sulfolipids adapted from: http://big.cea.fr/drf/big/english/PCV/LiPMB Basic components of glycerolipids lyso-Phospholipids
9. A B C D E F We are interested in understanding how environmental abiotic stresses shape nutrient (P) homeostasis and how this processes might determine plant local adaptation
10. Maize colonized highland territory after its domestication in tropical areas of Southwest Mexico Sawers et al . 2013 PubMedID 22303349 Relationships in the genus Zea Tripsicum Zea nicaraguensis Zea luxurians Zea diploperennis Zea perennis Zea mays ssp. huehuetenangensis Zea mays ssp. mexicana Zea mays ssp. parviglumis Zea mays ssp. mays After Hufford et al., 2012
11. Relationships in the genus Zea Tripsicum Zea nicaraguensis Zea luxurians Zea diploperennis Zea perennis Zea mays ssp. huehuetenangensis Zea mays ssp. mexicana Zea mays ssp. parviglumis Zea mays ssp. mays After Hufford et al., 2012 In México, there was significant gene flow between maize and highland teosinte mexicana Sawers et al . 2013 PubMedID 22303349 Hufford et al . 2013 PubMedID 23671421 maize mexicana low from mexicana in highland adaptation 3 parviglumis mexicana Geneflow from mexicana Adaptive? van Heerwaarden et al., 2011 Sawers & Sanchez, 2011 1750m parviglumis Gene flow from mexicana in highland adaptation 3 parviglumis mexicana Geneflow from mexicana Adaptive? van Heerwaarden et al., 2011 Sawers & Sanchez, 2011 1750m parviglumis
12. 3 parviglumis Geneflow from mexicana Adaptive? van Heerwaarden et al., 2011 Sawers & Sanchez, 2011 1750m parviglumis A second, independent event of highland adaptation occurred in In South America Gene flow from mexicana in highland a parviglumis Geneflow from mexicana Adaptive? van Heerwaarden et al., 2011 Sawers & Sanchez, 2011 parvig
13. Andosol probability % 0 50 100 Maize had to adapt to colder temperatures and soils with lower P availability
14. High P Low P Pi Recycling Cold Adaptation Glycerolipids relative abundance change in low P and in low Temperature Lambers et al . 2012 PubMedID 22937909 Degenkolbe et al . 2012 PubMedID 23061922 Phospholipids Sulfolipids Galactolipids High T Low T Phospholipids Sulfolipids Galactolipids
15. Hypothesis: glycerolipid metabolism was under high selective pressure and was important to maize adaptation to highlands in Mesoamerica and South America A B C D E F
16. Experimental approaches Common Garden Experiments in contrasting environments Bi-parental and diversity panel mapping populations Biochemical phenotyping Quantitative and population genetics Reverse genetics and heterologous expression
17. We grow our mapping populations in highland and lowland common garden fields Valle de Banderas (Nayarit) 25 masl Metepec (Edo. de México) 2600 masl
18. 50 masl GDD ~ 15
19. 2550 masl GDD ~ 5
20. pH ~ 5.1 [P] ~ 4 ppm 3570 masl
24. Meso/South America Highland/Lowland Paired design Diversity panel aka HiLo panel Mesoamerica highland 30 x lowland 30 x highland 30 x lowland 30 x South America Daniel Runcie Matt Hufford Sherry Flint-Garcia Ruairidh Sawers
25. B73 x Palomero Toluqueño (PT) Recombinant Inbred Lines B73 PT X F1 Backcross 100 BC1S5 families (25% Palomero Toluqueño on B73 background) 5 Self Generations
26. B73 x Palomero Toluqueño (PT) Recombinant Inbred Lines 100 BC1S5 families (25% PT on B73 background) Lowland site Highland site B73 PT B73 PT LOWLAND Valle de Banderas (50masl) HIGHLAND Metepec (2600masl) Local adaptation drives maize diversity B73 B73 PT PT
27. Separation Charged Surface Hybrid Technology Detection qTOF MS/MS Identification Quantification Internal Stds LipidBlast Oliver Fiehn, UC Davis Samples are collected in the field and then analyzed using HPLC-MS to identify and quantify glycerolipid species Extraction Field sampling: - V4-V6 plants - Youngest fully developed leaf - 600 samples collected in 80 minutes and flash frozen in liquid N2 Kind et al, 2013 PubmedID 25340521 Karla Juarez
28. Separation Detection qTOF MS/MS Identification Quantification Internal Stds LipidBlast We can identify around 120 glycerolipid species Extraction Kind et al, 2013 PubmedID 25340521 Charged Surface Hybrid Technology
29. PCs and LPCs are the compounds that show higher differences between highland and lowland lines of the diversity panel haring metabolomic features B)
30. PCs and LPCs are the species that show higher differences between highland and lowland lines, but we can see some differences between Meso and South America Mesoamerica South America Low High Low High Low High Low High Mesoamerica South America Meso_high SA_high Meso_low SA_low
31. Since PCs can be converted into LPCs and vice-versa we use ratios between them as a biochemical phenotype
32. Landraces with high PC/LPC ratios are mainly Mexican highland landraces ∑ ∑ ∑ Meso_high SA_high Meso_low SA_low
33. Landraces with high PC/LPC ratios are mainly Mexican highland landraces ∑ Meso_high SA_high Meso_low SA_low ∑ ∑ Meso_high SA_high Meso_low SA_low
34. PC/LPC ratios between Mexican higlands and lowlands is significative but not when we compare South American highlands and lowlands Meso_high SA_high Meso_low SA_low Mesoamerica South America ∑ ∑ Wilcoxon test p_val = 0.031 Low High Low High Wilcoxon test p_val = 0.626 Mesoamerica South-America
35. We use a B73 x Palomero Toluqueño (PT) Recombinant Inbred Lines mapping population to identify QTLs explaining glycerolipids variation B73 PT X F1 Backcross 100 BC1S5 families (25% Palomero Toluqueño on B73 background) 5 Self Generations
36. Palomero Toluqueño has lower levels of LPCs and higher PC/LPC ratios than B73 PT RILS B73
37. We run a QTL analysis using sum of PCs as the phenotype and identified a QTL in chr 3 chr3 8.6Mb ∑
38. Another QTL on the same region was identified when we used Lyso-PCs as phenotype chr3 8.6Mb ∑
39. Petersen et al, 2012 PubmedID 22672667 p-val = 9.094074e-09 PCs Metabolite ratios improve our ability to find associations between metabolite concentrations and genetic loci since they provide a better readout of enzymatic activity +
40. Metabolite ratios improve our ability to find associations between metabolite concentrations and genetic loci since they provide a better readout of enzymatic activity p-val = 9.094074e-09 p-val = 1.183811e-14 PCs PCs/LPCs Petersen et al, 2012 PubmedID 22672667 +
41. p-val = 9.094074e-09 p-val = 1.183811e-14 p-gain = 768203 PCs PCs/LPCs Metabolite ratios improve our ability to find associations between metabolite concentrations and genetic loci since they provide a better readout of enzymatic activity Petersen et al, 2012 PubmedID 22672667 +
42. PT and B73 alleles at the QTL peak have opposite effects on PCs and LPCs B73 B73 PT PT B73 B73 PT PT
43. RILS PCLPC BB = PT AA = B73 PT and B73 alleles at the QTL peak have opposite effects on PCs and LPCs
44. Heterozygous phenotype shows that the B73 allele is dominant B73 B73 PT PT B73 PT
45. Is there any candidate gene in that region that makes sense with our biochemical phenotype?
46. A phospholipase A1 with predicted PC to lyso-PC activity is right at the QTL peak GRMZM2G353444 + PLD PLC PLA1 PLA2
47. There are around 75 different genes with predicted phospholipase activity in the maize genome and almost 1/2 of them are PLA1 type phospholipases Stepflug et al, (2016) www.maizegdb.org/gene_center/gene/#rnaseq Tissue FPKM www.plantcyc.org 1 2 3 4 5 6 7 8 9 10
48. Highest Expression Levels of PLA1 s are found on vegetative leaves Tissue FPKM Stepflug et al, (2016) www.maizegdb.org/gene_center/gene/#rnaseq
49. The candidate gene is one of the most highly expressed PLA1 s GRMZM2G353444 Tissue FPKM Stepflug et al, (2016) www.maizegdb.org/gene_center/gene/#rnaseq
50. The gene is highly expressed in leaves (V3-V9) and is up-regulated in cold conditions in temperate inbred lines 2523 FPKM Roots Vegetative leaves Leaves AP/Node Reproductive tissue Seeds + Stepflug et al, (2016) www.maizegdb.org/gene_center/gene/#rnaseq data from Waters et al, 2016 PubMedID 28188666
51. We think this biochemical phenotype might be due a mutation that was selected for in PT and, probably, other Mexican highland material B73 B73 PT PT B73 B73 PT PT B73 B73 PT PT
52. The gene is equally expressed in PT and B73 homozygous RILS in the QTL region CDK control PT B73 PT PT B73
53. Several non-synonymous mutations (especially one in the flap-lid) in the PT allele may have an effect on the protein function 5’-UTR lipase domain 3’-UTR catalytic triad flap-lid nucleophilic elbow Non-syn SNP Syn SNP Ile > Leu
54. The flap/lid allows access of the substrate to the active site by alternating between close/open conformations CLOSE OPEN Khan et al . 2017 PubMedID 28337436
55. Is this mutation conserved in other highland maize? - 6 maize landraces from four highland regions and two lowland regions - Illumina HiSeq to 20-30X depth - Resequencing approach using genotype probabilities generated in ANGSD Li Wang (Hufford lab)
56. Allele differentiation (XP-CLR) values in the candidate gene region are higher in Mexican and SW-US populations than in the South American ones CDS Population XP-CLR nSNP Mexico 12.06 26 South West US 11.17 20 South America <95% 0 Li-Wang (Hufford lab)
57. The Ile > Leu mutation in the the flap/lid is conserved in Mesoamerican highland maize
58. Analysis of allele frequencies with PCadapt using ~4000 landraces indicates that the gene was under selection with increasing altitude Dan Gates (Ross-Ibarra lab)
59. Analysis of allele frequencies with PCadapt using ~4000 landraces indicates that the gene was under selection with increasing altitude Arg > Glu
60. Analysis of allele frequencies with PCadapt using ~4000 landraces indicates that the gene was under selection with increasing altitude Dan Gates (Ross-Ibarra lab) G het A
61. Analysis of allele frequencies with PCadapt using ~4000 landraces indicates that the gene was under selection with increasing altitude
62. www.rrlab.org Rubén Rellán-Álvarez @rrellanalvarez
63. Maize adapted twice to highland conditions, are there any signs of convergent evolution? Lowland Meso America Lowland South America Highland South America Highland Meso America Takuno et al . 2015 PubMedID 26078279
64. Very few genes show signs of convergent highland adaptation between Meso America and South America. Only Meso America (0.89 %) Only South America (0.65 %) Mesoamerica + South America Takuno et al . 2015 PubMedID 26078279
65. 32 pathways show evidence of convergent selection to highlands Only Meso America (0.89 %) Only South America (0.65 %) Mesoamerica + South America Takuno et al . 2015 PubMedID 26078279 A B C D E F A B C D E F Mesoamerica South America
66. 1/5 of them are glycerolipid related pathways GRMZM2G014981 GRMZM2G111632 GRMZM2G481755 CDP-diacylglycerol biosynthesis I CDP-diacylglycerol biosynthesis II Phosphatidylglycerol biosynthesis I Phosphatidylglycerol biosynthesis II Phospholipid biosynthesis II Triacylglycerol biosynthesis A B C D E F A B C D E F Mesoamerica South America
67. Two of these genes add acyl groups to LPC to form PC GRMZM2G014981 GRMZM2G481755 PC Mesoamerica South America PC +
68. This gene is expressed in leaves (V3-V9) and is down-regulated in cold conditions in temperate inbred lines Roots Vegetative leaves Leaves AP/Node Reproductive tissue Seeds Stepflug et al, (2016) www.maizegdb.org/gene_center/gene/#rnaseq data from Waters et al, 2016 10.1111/tpj.13414 526 FPKM +
69. The gene is highly expressed in leaves (V3-V9) and is up-regulated in cold conditions in temperate inbred lines 2523 FPKM Roots Vegetative leaves Leaves AP/Node Reproductive tissue Seeds Stepflug et al, (2016) www.maizegdb.org/gene_center/gene/#rnaseq data from Waters et al, 2016 10.1111/tpj.13414 +
70. In summary: High PC/LPC ratios seem to have been selected for in Mesoamerican highland maize and to a lesser extent in South American ones Genes coding for enzymes catalyzing PC -LPC conversions are differentiated between highlands and lowlands This high accumulation of PCs could be involved in highland adaptation /cold tolerance + +
71. Currently working on: - Heterologous expression of PT and B73 alleles in E.coli and A.thaliana to test functional differences of proteins
72. Currently working on: -Developing NILs and mutants of candidate genes to test the possible effect of PT alleles on highland adaptation and under cold/low P conditions. - Heterologous expression of PT and B73 alleles in E.coli and A.thaliana to test functional differences of proteins
73. Currently working on: -Identifying divergence of glycerolipids between highland and lowland maize using Qst-Fst. -Developing NILs and mutants of candidate genes to test the possible effect of PT alleles on highland adaptation and under cold/low P conditions. - Heterologous expression of PT and B73 alleles in E.coli and A.thaliana to test functional differences of proteins
74. Mesoamerica highland 30 x lowland 30 x highland 30 x lowland 30 x South America Runcie/Hilo Panel 120 landrace panel
75. Currently working on: -GWAS on glycerolipids and genes involved in glycerolipids pathways using RNA-Seq ASE on B73 x LRs F1s -Identifying divergence of glycerolipids between highland and lowland maize using Qst-Fst. -Developing NILs and mutants of candidate genes to test the possible effect of PT alleles on highland adaptation and under cold/low P conditions. - Heterologous expression of PT and B73 alleles in E.coli and A.thaliana to test functional differences of proteins
76. X Daniel Runcie highland site lowland site V4 stage Exome Capture RNA-Seq Lipids B73 GWAS on glycerolipids and genes involved in glycerolipids pathways using RNA-Seq ASE on B73 x LRs F1s landraces 120 F1s
77. Currently working on: -Building a Latitudinal Highland Magic Population (High-MAGIC) to analyze the genetic architecture of highland adaptation -GWAS on glycerolipids and genes involved in glycerolipids pathways using RNA-Seq ASE on B73 x LRs F1s -Identifying divergence of glycerolipids between highland and lowland maize using Qst-Fst. -Developing NILs and mutants of candidate genes to test the possible effect of PT alleles on highland adaptation and under cold/low P conditions. - Heterologous expression of PT and B73 alleles in E.coli and A.thaliana to test functional differences of proteins
78. We are building a Latitudinal Highland Magic Population (High-MAGIC) to analyze the genetic architecture of highland adaptation Azul Connor Palomero Bolita Sabanero Mishca Cpunti Pising Elevation Race Race Elevation 2413 Sabanero X Azul 2100 2164 Bolita X Mishca 2890 2030 Connor X Cpunti 3188 3488 Pissing X Palomero 2597 X F1s X X X F2s X X X 4 way crosses 2 rounds of inter-matting ~ 500 families to self
79. Currently working on: -U-Mu reverse genetics to study the effect of candidate genes on glycerolipid metabolism -Building a Latitudinal Highland Magic Population (High-MAGIC) to analyze the genetic architecture of highland adaptation -GWAS on glycerolipids and genes involved in glycerolipids pathways using RNA-Seq ASE on B73 x LRs F1s -Identifying divergence of glycerolipids between highland and lowland maize using Qst-Fst. -Developing NILs and mutants of candidate genes to test the possible effect of PT alleles on highland adaptation and under cold/low P conditions. - Heterologous expression of PT and B73 alleles in E.coli and A.thaliana to test functional differences of proteins
80. U-Mu reverse genetics to study the effect of candidate genes on glycerolipid metabolism Galactolipid synthesis Phospholipid synthesis Sulpholipid biosynthesis Phospholipid degradation Sofia Sánchez Juan Estevez
81. W22 PT X F1 Backcross x 3 300 RILS (BC3S6) 6 generations of selfing 300 RILS (BC3) To be used together with a W22 x PT RIL mapping population Matt Hufford Ruairidh Sawers
82. Funding www.highlandadaptation.org
83. It takes a whole village to grow maize… Karla Juarez Jeffrey Ross-Ibarra Matt Hufford Daniel Runcie Sherry Flint-García Juan Estevez Ruairidh Sawers Members of the Sawers and Rellan-Álvarez Labs www.highlandadaptation.org Oliver Fiehn Dave Jackson Christoph Benning Sofia Sánchez Denise Costich
84. Allele Specific Expression will allow us to confirm/discover allelic effects X 40x 40x 80 F1s Daniel Runcie highland site lowland site V4 stage Exome Capture RNA-Seq Lipids B73 www.rrlab.org Rubén Rellán-Álvarez @rrellanalvarez bit.ly/rra_ISU_2017