A transposable element insertion is the switch between alternative life history strategies

TL;DR: The genetic basis of the ALHS switch in Colias crocea is mapped to a transposable element insertion downstream of the Colias homolog of BarH-1, a homeobox transcription factor, which arises via recruitment of a transcription factor previously known for its function in cell fate determination in pigment cells of the retina.

Abstract: Tradeoffs affect resource allocation during development and result in fitness consequences that drive the evolution of life history strategies. Yet despite their importance, we know little about the mechanisms underlying life history tradeoffs in wild populations. Many species of Colias butterflies exhibit an alternative life history strategy (ALHS) where females divert resources from wing pigment synthesis to reproductive and somatic development. Due to this reallocation, a wing color polymorphism is associated with the ALHS: individuals have either yellow/orange or white wings. Here we map the genetic basis of the ALHS switch in Colias crocea to a transposable element insertion downstream of the Colias homolog of BarH-1, a homeobox transcription factor. Using CRISPR/Cas9 gene editing, antibody staining, and electron microscopy we find morph-specific specific expression of BarH-1 suppresses the formation of pigment granules in wing scales. Lipid and transcriptome analyses reveal physiological differences associated with the ALHS. These findings characterize a novel mechanism for a female-limited ALHS and show that the switch arises via recruitment of a transcription factor previously known for its function in cell fate determination in pigment cells of the retina.

Summary

Author Contributions

• AW conducted butterfly rearings and lab work, analysed the data, and wrote the manuscript with CWW and input from the coauthors.
• AW, MWP, KT, and CWW conducted the CRISPR/Cas9 knockout experiment.
• Raw data was cleaned and high quality reads were used as input for the AllPaths-LG (v. 50960) 43 assembly pipeline.
• High molecular weight DNA was extracted from two more Alba females from the above mentioned cross (i.e full siblings).

Bulk segregant analyses (BSA):

• The female informative cross data and mapping protocol described in Woronik and Wheat, 2017 46 was applied to the high quality reference genome to identify the contigs that made up the Alba chromosome.
• SAMTOOLS v1.2 48 was used to filter (view -f 3 -q 20), sort and index the bam files and generate mpileup files for the two pools and the orange mother.
• DNA was prepared as described above for 26 Alba and 28 orange female offspring resulting in two DNA pools.
• Library preparation (TruSeq PCR-free) and Illumina sequencing (150 bp paired-end reads with 350bp insert, HiSeqX), was performed at Science for Life Laboratory (Stockholm, Sweden).
• A site was considered an Alba SNP if 1) it was homozygous in the orange pool and 2) the allele frequency difference in the Alba pool compared to the orange was 0.45-0.55.

Antibody Generation and Staining:

• A Rabbit-anti-Bar antibody was generated against the full length sequence of the Vanessa cardui Bar homolog.
• Protein was generated by GenScript (Piscataway, NJ) and purified to >80% purity.
• Antibody staining was performed as described previously for Drosophila and butterfly tissues 54 .
• Staged pupal wings and retinas were dissected and fixed 48 hours post-pupation.
• Images were captured using standard confocal microscopy on a Leica SP5.

CRISPR/Cas9 knockouts:

• The guide-RNA (gRNA) sequences were generated using the protocol described in Perry et al.
• Full gRNA constructs had the following configuration: an M13F region, a spacer sequence, a T7-promotor sequence, the Target specific sequence, a Cas9 binding sequence, and finally a P505 sequence.
• They were then mixed with Cas9-NLS protein (PNA Bio, Newbury Park, CA, USA) and diluted to a final concentration of 125-250 ng/µl.
• C. crocea females (n > 40) from Aiguamolls de l'Empordà, Spain were captured and kept in morph-specific flight cages in the lab at Stockholm University where they oviposited on alfalfa (Medicago sativa).

CRISPR/Cas9 validation:

• To validate the mutation, Cas9 cut sites were PCR-amplified and a ~370bp region, centered on the intended cut site were sequenced using Illumina MiSeq 300bp paired-end sequencing.
• Sequences were amplified and ligated with Illumina adapter and indexes in a two-step process following the protocol certified by peer review) is the author/funder.
• Samples were coated in gold for 80 seconds using an Agar sputter coater and imaged under 5 kV acceleration voltage, high vacuum, and ETD detection using a scanning electron microscope (Quanta Feg 650, FEI, Hillsboro, Oregon, USA).
• The authors counted the number of pigment granules within each square and took the average, then conducted a two sample t-test in R.
• The total lipid content (nmol per abdomen) was calculated as a sum of pmol certified by peer review) is the author/funder.

Transcriptome assembly and differential expression analysis:

• Offspring from a wild caught Alba female from Catalonia, Spain were reared at Stockholm University.
• Pupae were dissected in PBS solution, and the abdomen and wings were flash frozen in liquid nitrogen and stored at -80 o C. RNA was extracted from the abdomen and wing tissues using Trizol.
• Raw data was cleaned and reads from all libraries were used in a de novo transcriptome assembly (Trinity version trinityrnaseq_r2013_08_14 with default parameters) 57 .
• SAMTOOLS v1.2 48 idxstats was then used to calculate the read counts per gene for each of the sorted bam files.
• BarH-1 is heterogeneously expressed in the scale building cells within this region.

Fig. 4. Physiological differences between female morphs of C. crocea.

• A) The mass corrected total neutral lipid content for female morphs in two temperature treatments.
• However there is an interaction between morph and temperature as the difference is only significant in the cold treatment.
• The X-axis is the log of the fold change (FC), positive log(FC) indicates the gene is upregulated in Alba individuals.
• C) Volcano plots to visualize gene expression differences between female morphs in pupal wing tissue.
• Certified by peer review) is the author/funder.

Full text

A transposable element insertion is the switch between alternative life history strategies 1
2
Authors: Alyssa Woronik*
1,2
, Kalle Tunström
1
, Michael W. Perry
2,3
, Ramprasad Neethiraj
1
, 3
Constanti Stefanescu
4,5
, Maria de la Paz Celorio-Mancera
1
, Oskar Brattström
6
, Jason Hill
1,7
, 4
Philipp Lehmann
1
, Reijo Käkelä
8
, Christopher W. Wheat*
1
5
6
Author affiliations: 7
1
Department of Zoology, Stockholm University, S106 91 Stockholm, Sweden 8
2
Department of Biology, New York University, New York, New York 10003, USA 9
3
Division of Biological Sciences, University of California San Diego, La Jolla, California 92093, 10
USA 11
4
Museum of Natural Sciences of Granollers, Granollers, Catalonia 08402, Spain 12
5
CREAF, Cerdanyola del Valles, Catalonia 08193, Spain 13
6
Department of Zoology, University of Cambridge, Cambridge CB23EJ, United Kingdom 14
7
Department of Medical Biochemistry and Microbiology, Uppsala University, Uppsala, Sweden 15
8
Helsinki University Lipidomics Unit, Helsinki Institute for Life Science (HiLIFE) and Molecular 16
and Integrative Biosciences Research Programme, University of Helsinki, FI00014 Helsinki, 17
Finland 18
19
Correspondence to: AW alyssa.woronik@zoologi.su.se and CWW chris.wheat@zoologi.su.se 20
21
Tradeoffs affect resource allocation during development and result in fitness consequences that 22
drive the evolution of life history strategies. Yet despite their importance, we know little about 23
the mechanisms underlying life history tradeoffs in wild populations. Many species of Colias 24
butterflies exhibit an alternative life history strategy (ALHS) where females divert resources from 25
wing pigment synthesis to reproductive and somatic development. Due to this reallocation, a 26
wing color polymorphism is associated with the ALHS: individuals have either yellow/orange or 27
white wings. Here we map the genetic basis of the ALHS switch in Colias crocea to a 28
transposable element insertion downstream of the Colias homolog of BarH-1, a homeobox 29
transcription factor. Using CRISPR/Cas9 gene editing, antibody staining, and electron 30
microscopy we find morph-specific specific expression of BarH-1 suppresses the formation of 31
pigment granules in wing scales. Lipid and transcriptome analyses reveal physiological 32
differences associated with the ALHS. These findings characterize a novel mechanism for a 33
certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was notthis version posted July 24, 2019. ; https://doi.org/10.1101/424879doi: bioRxiv preprint

female-limited ALHS and show that the switch arises via recruitment of a transcription factor 34
previously known for its function in cell fate determination in pigment cells of the retina. 35
36
A life-history strategy is a complex pattern of co-evolved life history traits (e.g. number of 37
offspring, size of offspring, and lifespan
1
), that is fundamentally shaped by tradeoffs that arise 38
because all fitness components cannot simultaneously be maximized. Therefore, finite 39
resources are competitively allocated to one life history trait versus another within a single 40
individual, and selection acts on these allocation patterns to optimize fitness
2
. Evolutionary 41
theory predicts that positive selection will remove variation from natural populations, as 42
genotypes with the highest fitness go to fixation
3
. However, across diverse taxa alternative life 43
history strategies (ALHSs) are maintained within populations at intermediate frequencies due to 44
balancing selection
4
. Life history theory was developed using methods such as quantitative 45
genetics, artificial selection, demography, and modeling to gain significant insights into the 46
causes and consequences of genetic and environmental variation on life history traits. Yet 47
despite these advances, a key challenge that remains is to identify the proximate mechanisms 48
underlying tradeoffs, especially for ecologically relevant tradeoffs that occur in natural 49
populations
5
. Here, we identify the mechanism underlying one such ALHS in the butterfly Colias 50
crocea (Pieridae, Lepidoptera) (Geoffroy, 1785). 51
52
Colias butterflies (the “clouded sulphurs”) are common throughout the Holarctic and can be 53
found on every continent except Australia and Antarctica
6
. In approximately a third of the nearly 54
90 species within the genus, females exhibit two alternative wing-color morphs: yellow or 55
orange (depending on the species) and white
6,7
(Fig. 1A). The wing color polymorphism arises 56
because during pupation the white morph, also known as Alba, reallocates larval derived 57
resources from the synthesis of energetically expensive colored pigments to reproductive and 58
somatic development
8
. This tradeoff has been well characterized in Colias crocea, the Old 59
World species that we focus upon in this work, via radio-labelled metabolite tracking in pupae
9
60
as well as in the New World species Colias eurytheme
8
(Pieridae, Lepidoptera) (Boisduval, 61
1852) using ultraviolet spectrophotometry. As a result of the resource reallocation, Alba females 62
have faster pupal development, a larger fat body, and significantly more mature eggs at 63
eclosion compared to orange females
10
. However, despite these developmental advantages 64
and the dominance of the Alba allele, the polymorphism is maintained by several abiotic and 65
biotic factors
10-14
. For example, males preferentially mate with orange females, as wing color is 66
an important cue for mate recognition
10,12,13
. This mating bias likely has significant fitness costs 67
certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was notthis version posted July 24, 2019. ; https://doi.org/10.1101/424879doi: bioRxiv preprint

for Alba females because males transfer essential nutrients during mating, and multiply mated 68
females have more offspring over their lifetime
15,16
. The mating bias against Alba females is 69
strongest in populations that frequently co-occur with other white Pierid butterfly species due to 70
interference competition
13
. Also, Alba’s development rate advantage is temperature dependent, 71
with Alba females having faster development in cold temperatures
10
. Field studies confirm Alba 72
frequency and fitness increases in species that inhabit cold and nutrient poor habitats, where 73
the occurrence of other white Pierid butterflies is low. While in warm environments with nutrient 74
rich host plants and a high co-occurrence of other white species, orange females exhibit 75
increased fitness and frequency
12-14
. Previous work has also suggested Alba females have a 76
higher sensitivity to viral infections
9
. In all Colias species where it has been investigated (n=6), 77
the switch between the Alba or the orange strategy is controlled by a single, autosomal locus
6
. 78
This fact, along with ancestral state reconstruction
7
, has led to the assumption that the Alba 79
locus is conserved within the genus Colias, and potentially across the subfamily Coliadinae. Yet, 80
despite over a century of research on various aspects of Alba biology the mechanism underlying 81
this polymorphism remained unknown. 82
83
Using a de novo reference genome for C. crocea that we generated via Illumina and PacBio 84
sequencing, and three rounds of bulk segregant analyses (BSA) using whole genome 85
sequencing from a female and two male informative crosses for Alba, we mapped the Alba 86
locus to a ~3.7 Mbp region (Supplementary Fig.1, & Supplementary Information). Then, with 87
whole genome re-sequencing data from 15 Alba and 15 orange females from diverse population 88
backgrounds, a SNP association study fine mapped the Alba locus to a ~430 kb contig that fell 89
within the ~3.7 Mbp locus identified using the BSA crosses (Fig. 1B and Supplementary 90
Information). The majority of SNPs significantly associated with Alba (n=70 of 72) were within or 91
flanking a Jockey-like transposable element (TE) (Fig. 1C). We determined that the TE insertion 92
was unique to the Alba morph in C. crocea by assembling orange and Alba haplotypes for this 93
region, then quantifying differences in read depth between morphs within and flanking the 94
insertion (Supplementary Information and Supplementary Figs. 2, 3, & 4). We then used PCR to 95
validate the presence or absence, respectively, of the insertion in 25 Alba and 57 orange wild-96
caught females (Supplementary Fig. 7). We also found no evidence of a TE insertion in the 97
homologous region of other butterfly genomes (Danaus plexippus & Heliconius melpomene) 98
(Supplementary Fig. 2). 99
100
certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was notthis version posted July 24, 2019. ; https://doi.org/10.1101/424879doi: bioRxiv preprint

The Alba-specific insertion was located ~30 kb upstream of a gene encoding a DEAD-box 101
helicase, and ~6kb downstream of the Colias homolog of BarH-1, a homeobox transcription 102
factor (Fig. 1C). BarH-1 was an intriguing find as it affects color via pigment granule 103
development within eyes of Drosophila melanogaster
17
. To investigate BarH-1 expression in 104
developing C. crocea wings, we used in situ hybridization of BarH-1 on wings from two day old 105
pupae of orange and Alba females. We found the BarH-1 protein is expressed in scale building 106
cells within the white wing regions in Alba females (Fig. 2B). We did not observe BarH-1 in scale 107
building cells from orange areas of the wing in orange females (Fig, 2C). Interestingly however, 108
we found BarH-1 is expressed in scale building cells within black regions for both morphs (Fig, 109
2A&D). To validate the functional role of BarH-1 in the Alba phenotype, we generated 110
CRISPR/Cas9-mediated deletions within exons 1 and 2 using a mosaic knockout (KO) 111
approach (Supplementary Information). BarH-1 KO gave rise to a white/orange color mosaic on 112
the dorsal side of the wings in females with an Alba genotype (i.e. TE insertion +) (Fig. 1D), 113
while KO males and orange females displayed no white/orange mosaic on the wing. These 114
results indicate BarH-1 expression suppresses orange coloration in the wings. We also 115
observed black and green mosaic coloring of eyes in KO males and females of both morphs, 116
where green eyes are the wild type color (Fig. 1E). These results indicate BarH-1 also plays a 117
role in Colias eye development. 118
119
We next investigated how the Alba color change manifests within wings. Butterfly wing color can 120
arise either due to the absorption of light by pigments deposited within the scales, or by the 121
scattering of light via regularly arranged nanostructures in the scales
18
. Colias butterflies have 122
pteridine pigments. These pigments are synthesized within the wings and previous work using 123
ultraviolet spectrophotometry in C. eurytheme found Alba females exhibit dramatic reductions in 124
colored pteridine pigments compared to orange
8,9
. In insects, pteridines are synthsized in 125
pigment granules and pigment granules are concentrated within wing scales of Pierid 126
butterflies
19,20
. However, whether morphs differed in wing scale morphology was unknown. To 127
investigate wing morphology, we used scanning electron microscopy and found white scales 128
from Alba individuals exhibited a dramatic and significant reduction in pigment granules, 129
compared to orange scales (t
5.97
= 2.93, p = 0.03) (Fig 3 A&B). These results indicate the color 130
change to white is caused by reduced pigment granule formation. Congruent with this 131
interpretation, CRISPR KO Alba individuals exhibited significantly less pigment granules in 132
scales from the white wild-type region compared to scales in orange BarH-1 KO regions (t
5.45
= 133
10.78, p < 0.001) (Fig. 3C). To further test whether reduction in pigment granule amount alone 134
certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was notthis version posted July 24, 2019. ; https://doi.org/10.1101/424879doi: bioRxiv preprint

was sufficient for the orange to white color change, we chemically removed the pigment 135
granules from the wing of an orange C. crocea female. This resulted in formerly orange regions 136
turning white (Fig. 3D). Wings likely appear white after granule removal due to the scattering of 137
light from the remaining non-lamellar nanosctructures
21
. These results demonstrate that BarH-1 138
suppresses pigment granule formation in wing scales, resulting in the white color of Alba 139
females in C. crocea. Thus, we propose the resource tradeoff between color and development 140
arises due to a classic Y reallocation model, wherein limited resources are competatively 141
allocated and increased investement in one trait results in a decreased investment to another
22
. 142
Within the energetically closed system of a developing pupa, reduced pigment granule 143
formation would likely result in reduced pigment synthesis, which would in turn leave more 144
resources free to be used for other developmental processes. Finally, we also observed scale 145
building cells in black regions of both morphs express BarH-1 and also lack pigment granules 146
(Fig 2 A&D and Fig 3 A&B), but these scales appear black due to melanin deposition within the 147
scale
18
. These results suggest BarH-1 may also repress pigment granule formation within black 148
scales. 149
150
The Alba mechanism is assumed to be conserved across Colias. Therefore, we wished to test 151
whether Alba females from the New World species Colias eurytheme also exhibited significantly 152
less pigment granules than orange females. Indeed, we found orange C. eurytheme scales 153
exhibited abundant pigment granules while Alba scales almost entirely lacked granules (Fig. 3 154
E&F). These results demonstrate white wing color arises via the same morphological 155
mechanism within Colias and corroborate previous assumptions that Alba is conserved across 156
the genus. To further validate that other aspects of the Alba/orange alternative life history 157
strategy are conserved across the genus we tested whether one of the physiological tradeoffs of 158
Alba reported for New World species was also seen in C. crocea. In C. eurytheme, Alba females 159
have larger fat bodies than orange females and the strength of the Alba advantage increased in 160
cold temperatures
10
. To compare abdominal lipid stores between morphs in C. crocea, we 161
conducted high performance thin layer chromatography on two day old adult females reared 162
under two temperature treatments (Hot: 27
o
C vs. Cold: 15
o
C during pupal development). Adults 163
were not allowed to feed before samples were taken, therefore these measurements reflect 164
larval stores, where the putative energetic tradeoff should be more clearly visible. We found 165
Alba females had larger abdominal lipid stores than orange in both temperature treatments, 166
though the difference was only significant in the cold treatment (cold: n=32, t
29.12
= 3.42, P = 167
0.002, hot: n=25, t
22.71
= 0.67, P = 0.51) (Fig. 4A). These results are consistent with previous 168
certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was notthis version posted July 24, 2019. ; https://doi.org/10.1101/424879doi: bioRxiv preprint