A nonsense mutation in TFEC is the likely cause of the recessive piebald phenotype in ball pythons (Python regius)

Abstract

13 Captive-bred ball pythons (Python regius) represent a powerful model system for studying the 14 genetic basis of colour variation and Mendelian phenotypes in vertebrates. Although hundreds 15 of Mendelian phenotypes (colour morphs) affecting colouration and patterning have been 16 described for ball pythons, the genes causing these colour morphs remain unknown. Here, we 17 used crowdsourcing of samples from commercial ball python breeders to investigate the 18 genetic basis of a classic phenotype found in the pet trade, the piebald [characterized by 19 dorsolateral patches of unpigmented (white) skin]. We used whole-genome sequencing of 20 pooled samples followed by population genetic methods to delineate the genomic region 21 containing the causal gene. We identified TFEC of the MIT-family of transcription factors as a 22 candidate gene. Functional annotation of SNPs identified a nonsense mutation in TFEC, which 23 we conclude is the likely causal variant for the piebald phenotype. Our work shows that ball 24 python colour morphs have the potential to be an excellent model system for studying the 25 genetic basis of pigment variation in vertebrates, and highlights how collaborations with 26 commercial breeders can accelerate discoveries. 27 28 29 30 31 32

Full text

A nonsense mutation in TFEC is the likely cause of the recessive piebald phenotype in ball
1
pythons (Python regius)
2
*Alan Garcia-Elfring, Heather L. Roffey, *Andrew P. Hendry, and *Rowan D. H. Barrett
3
4
5
6
*Redpath Museum and Department of Biology, McGill University, Montreal, QC, Canada
7
The corresponding author's email address: alan.garcia-elfring@mail.mcgill.ca
8
9
10
11
12

Abstract
13
Captive-bred ball pythons (Python regius) represent a powerful model system for studying the
14
genetic basis of colour variation and Mendelian phenotypes in vertebrates. Although hundreds
15
of Mendelian phenotypes (colour morphs) affecting colouration and patterning have been
16
described for ball pythons, the genes causing these colour morphs remain unknown. Here, we
17
used crowdsourcing of samples from commercial ball python breeders to investigate the
18
genetic basis of a classic phenotype found in the pet trade, the piebald [characterized by
19
dorsolateral patches of unpigmented (white) skin]. We used whole-genome sequencing of
20
pooled samples followed by population genetic methods to delineate the genomic region
21
containing the causal gene. We identified TFEC of the MIT-family of transcription factors as a
22
candidate gene. Functional annotation of SNPs identified a nonsense mutation in TFEC, which
23
we conclude is the likely causal variant for the piebald phenotype. Our work shows that ball
24
python colour morphs have the potential to be an excellent model system for studying the
25
genetic basis of pigment variation in vertebrates, and highlights how collaborations with
26
commercial breeders can accelerate discoveries.
27
28
29
30
31
32

Introduction
33
The study of colour variation has a long history in evolutionary biology and genetics (Caro 2017;
34
Cuthill et al. 2017; Harris et al. 2019). Over 150 years ago, crossing yellow and green peas
35
helped Mendel to discover that traits are inherited as discrete hereditary units or “factors”
36
(Mendel 1866; reviewed by Ellis et al. 2011). Once Mendel’s work was rediscovered in 1900,
37
invertebrate colour variation helped establish that Mendel’s “factors”, which we now call
38
genes, were located on chromosomes (Bridges 1919; Harrison 1920). Further, colour variation
39
in vertebrates has provided insights in many fields, particularly the biomedical sciences (e.g.
40
Pingault et al. 1998; Tachibana et al. 2003; Richards et al. 2019), developmental biology (e.g.
41
Cohen et al. 2016; Haupaix et al. 2018), and evolutionary biology (e.g. Schweizer et al. 2018;
42
Barrett et al. 2019; Burgon et al. 2020; reviewed by San-Jose and Roulin 2017). However, our
43
current knowledge of the genetics of vertebrate colouration comes primarily from studies on
44
model species, particularly the mouse (e.g. Baxter et al. 2004; Hoekstra 2006; San-Jose and
45
Roulin 2017; but see Ullate-Agote et al. 2020). This bias in the literature provides a limited view
46
of the ways in which genetic variation can affect vertebrate colouration.
47
Although mice, as all mammals and birds, have one pigment-producing cell type
48
(melanocytes) which synthesizes melanin, reptiles, like fish and amphibians, have three colour-
49
producing cells or chromatophores (reviewed by Bechtel 1978; Mills and Patterson 2009;
50
Olsson et al. 2013). In addition to melanin-producing cells, called melanophores, reptiles have
51
xanthophores and iridiophores that contribute to colouration. Xanthphores synthesize yellow
52
pteridine-based pigments and can also contain red carotenoid pigments that are obtained from
53

diet (often called erythrophores). Iridiophores do not contain pigments but instead produce
54
structural colouration by reflecting light on guanine nanocrystals (Teyssier et al. 2015). The
55
three colour-producing cells interact in 3D space (Grether et al. 2004; Saenko et al. 2013) to
56
generate the remarkable range of colour variation observed in reptiles.
57
We propose that the ball python (Python regius; Figure 1a) is an excellent model for
58
studying the genetic basis of vertebrate colouration (Irizarry and Bryden 2016). Ball pythons are
59
native to western Sub-Saharan Africa, with a range extending from Senegal to Uganda. With a
60
maximum length of approximately 6 feet long, ball pythons are small relative to their better-
61
known Asian congener, the Burmese python (Python bivittatus). Owing to their small size,
62
docility, and simple husbandry, ball pythons rose in popularity in the pet trade, particularly
63
among reptile enthusiasts. An additional factor that has contributed to the ball python’s
64
popularity was the discovery of rare phenotypes that affect colouration. In Ghana, claims of
65
rare wild ball pythons with altered patterning go back as early as 1966. In the early 1990s, one
66
such ball python was caught, a piebald. This individual had patches of unpigmented (white) skin
67
(e.g. Figure 1b-d), a pigment deficiency found in many vertebrates (Ahi and Sefc 2017),
68
including humans (Oiso et al. 2013). However, it would take a few years to rule out
69
environmental factors as a causal explanation for the piebald phenotype in the ball python
70
caught from the wild. By 1998, Peter Kahl had reproduced the piebald phenotype in captivity
71
(called ‘proving’ by commercial breeders) and showed it had a recessive mode of inheritance (A
72
brief history of Royal Ball Python morphs for beginners, n.d.). This pioneering work of breeders
73
and hobbyists spurred the discovery and propagation of numerous additional and highly diverse
74
colour morphs (e.g. Figure 1b-g). Today, 314 Mendelian phenotypes (i.e., phenotypes that
75

segregate as a single Mendelian factor or major quantitative trait locus) are bred in captivity
76
(known as ‘basic’ morphs; http://www.worldofballpythons.com/morphs), providing a valuable
77
resource for studying the genetics of vertebrate colouration in great detail. Indeed, commercial
78
breeders have taken advantage of epistasis and produced over 6000 unique phenotypes
79
(‘designer morphs’) with different combinations of the Mendelian colour factors.
80
Ball pythons therefore provide an outstanding opportunity to study the genetic basis of
81
Mendelian phenotypes related to pigmentation and patterning. For example, ball python colour
82
morphs can be characterized by changes to pattern or to pigmentation or both. Phenotypes
83
with changes to pigmentation but not pattern (Figure 1c) are presumably determined by
84
changes to genes with roles in a pigment synthesis pathway. On the other hand, aberrations to
85
pattern, like white-spotting, are thought to occur through mutations in genes with functions in
86
progenitor cell migration and differentiation from the neural crest (reviewed by Mills and
87
Patterson 2009). However, most research on colouration, vertebrate and invertebrate, has
88
focused on pigmentation, particularly melanin-based pigmentation (e.g. San-Jose and Roulin
89
2018), rather than on patterning (Mills and Patterson 2009). The extensive phenotypic variation
90
in pigment and pattern found in captive-bred ball pythons can therefore help correct the bias
91
(e.g. Ullate-Agote et al. 2020) associated with mammalian models and melanin-based
92
pigmentation in the scientific literature on vertebrate colouration.
93
In this study, our goal was to discover the candidate mutation(s) that produce the
94
Mendelian phenotype affecting pattern in ball pythons, the piebald (Figure 1b). We used
95
crowdsourcing of samples from commercial breeders, thereby bridging a gap between
96

academia and industry. Genes known to cause white-spotting in other vertebrates, like KIT,
97
MITF, EDNRB, and SOX10 (Fleischman et al. 1991; Baxter et al. 2004; Ahi and Sefc 2017)
98
represent a priori candidate genes, but given the differences in pigmentation development
99
between reptiles and previously studied model systems, we employed a whole-genome
100
approach to facilitate discovery of unknown candidate genes. We sequenced whole-genomes of
101
pooled samples and then applied population genetic methods to delineate the genomic region
102
of interest and identify candidate genes. We then functionally annotated SNPs to identify the
103
putative causal mutation. Given the character of the piebald phenotype (i.e., dorsolateral
104
patches of unpigmented skin), we expected the candidate gene to be expressed in the neural
105
crest and associated with chromatophore migration and differentiation.
106
107
Methods
108
Sampling and DNA extraction
109
We obtained ball python samples by crowd-sourcing from commercial breeders in Canada
110
(Mutation Creation, T. Dot Exotics, The Ball Room Canada, Designing Morphs), who supplied us
111
with shed skin. We then used shed skin samples for 47 piebald individuals (i.e., individuals
112
homozygous for the piebald variant; Table S1) and 52 non-piebald individuals (Table S2).
113
Although individuals from both sets of samples contained additional basic morphs, the only
114
consistent difference between the two was the piebald phenotype. We attempted to maximize
115
the number of individuals that come from different families to minimize the effects of
116
population structure, although there were some exceptions (Table S2). From each sample we
117

used approximately 0.1 g of shed skin, cut to small pieces using scissors, for DNA extraction. We
118
extracted DNA following a standard phenol-chloroform procedure, with a 24-hour proteinase-K
119
incubation time at 37 °C. Piebald and non-piebald samples were prepared on different working
120
days to avoid contamination. We quantified all samples using a Picogreen® ds DNA assay
121
(Thermo Fisher Scientific, Waltham, USA) on an Infinite® 200 Nanoquant (Tecan Group Ltd.
122
Männedorf, Switzerland).
123
Sequencing
124
After DNA extraction, we mixed DNA of individuals (according to phenotype) in equimolar
125
amounts to obtain a single pool for each phenotype, ‘piebald’ and ‘non-piebald.’ We used PCR-
126
based whole-genome libraries for both pools, which were prepared at the McGill University and
127
Genome Quebec Innovation Center, Montreal, Canada. We sequenced 150 bp paired-end reads
128
using two lanes of Illumina HiSeqX.
129
Bioinformatics
130
To process raw reads, we applied filters based on read quality and length, keeping reads with a
131
minimum quality of 20 (--quality-threshold 20) and a length of 50 bp (--min-length 50). We then
132
aligned processed reads to the Burmese python (Python bivittatus) draft assembly Pmo2.0
133
(Castoe et al. 2013) using the program NextGenMap (Sedlazeck et al. 2013). NextGenMap was
134
designed for aligning reads to highly polymorphic genomes or genomes of closely related
135
species. We used SAMtools (Li et al. 2009) to convert SAM files to BAM format and remove
136
reads with mapping quality below 20 (samtools view -q 20). We filtered for PCR duplicates using
137
the program MarkDuplicates of Picard Tools (Wysoker et al. 2013). We then created a mpileup
138

file (samtools mpilep -B) from which the synchronized (sync) file was produced using
139
Popoolation2 (Kofler et al. 2011b). The sync file contains read counts for all nucleotides
140
sequenced in the genome and it is used for subsequent downstream analyses (e.g. FST scan).
141
We also applied the same protocol as above but instead aligned reads to the chromosome-
142
length Burmese python reference genome, Python_molurus_bivittatus-5.0.2_HiC.assembly
143
(Dudchenko et al. 2017, 2018).
144
Identification of fixed SNPs and delineation of the candidate genomic region
145
We applied an FST scan to identify fixed SNPs across the two pools. For this procedure, we used
146
the fst-sliding.pl script of Popoolation2 (--min-count 10, --min-coverage 20, --max-coverage 500,
147
--min-covered-fraction 0, --window-size 1, --step-size 1, --pool-size 47:52, --suppress-
148
noninformative). We then identified candidate SNPs as those with FST estimates of 1 and
149
mapped them to genes. We used a custom BASH script to map fixed SNPs to genes in the gene
150
annotation file using scaffold and SNP position. Because the draft assembly is highly
151
fragmented (Castoe et al. 2013), we also applied the same FST scan on data aligned to the
152
chromosome-length genome assembly (Dudchenko et al. 2017, 2018) to get a better
153
delineation of the genomic region of interest within candidate SNPs. However, this latter
154
assembly is not annotated with genetic features – hence necessitating the use of both
155
assemblies.
156
Functional annotation of SNPs
157
We next functionally annotated genome-wide SNPs with the software snpEff (Cingolani et al.
158
2012) to aid in identifying the putative causal mutation for the piebald phenotype. SnpEff was
159

designed for annotating and predicting the effects of SNPs, such as amino acid changes. This
160
program provides an assessment of the impact of a SNP, including ‘HIGH’ (e.g. stop codon),
161
‘MODERATE’ (e.g. non-synonymous change), ‘LOW’ (e.g. synonymous change), or ‘MODIFIER’
162
(change in an intergenic area).
163
164
Results
165
SNP dataset
166
From the BAM files, we obtained an average depth of coverage of 50.5 and 52.6 for the piebald
167
and non-piebald pools, respectively. After filtering, our dataset consisted of 3,095,304 SNPs for
168
the draft assembly and 3,221,285 SNPs in the chromosome-length assembly.
169
FST scan for fixed SNPs and candidate genomic region
170
Across all SNPs, we found an average FST of 0.03456 – relatively low, indicating that population
171
structure was minimized. Using the draft assembly, we identified 129 fixed SNPs (FST = 1.0) and
172
369 SNPs with FST > 0.9. Plotting the FST values (Figure S1) shows peaks across the reference
173
genome, demonstrating the fragmented state of the draft assembly. In the chromosome-length
174
assembly, we found 131 fixed SNPs (FST = 1.0) and 372 SNPs with FST > 0.9. The chromosome-
175
length assembly also shows that 128/131 fixed SNPs and 365/372 SNPs with FST > 0.9 map to a
176
single region on scaffold 7, clearly delineating a genomic region of interest (Figure 2). The 128
177
fixed SNPs map to a region 8 Mb long (scaffold 7: 49526089-57612101). This region becomes
178
26.5 Mb when considering the 365 SNPs with FST > 0.9 (scaffold 7: 49246858-76754467).
179

However, most of these 365 SNPs cluster within the 8 Mb region demarcated by fixed SNPs,
180
with only 11 mapping beyond the bounds of this region (three upstream, eight downstream).
181
Candidate genes and SNP annotation
182
We found that 59 fixed SNPs map to 12 different protein-coding genes (BMT2, CAPZA2, CNTN1,
183
DOCK4, FOXP2, GPR85, KCND2, LOC103067393, LSMEM1, ST7, TES, and TFEC). We annotated
184
SNPs for predicted impact (see methods) and found that all fixed SNPs have the lowest impact
185
classification (MODIFIER), often assigned to mutations in non-functional regions, and most map
186
to introns. However, we found other impact classifications among SNPs with FST greater than
187
0.9. One synonymous (i.e., silent) SNP (LOW impact, NW_006539084.1 26867, FST = 0.92) is
188
located within the gene LSMEM1 and one nonsense SNP (HIGH impact, NW_006534020.1
189
160458, FST = 0.96) is located within the gene TFEC. This nonsense variant consisted of a C>T
190
mutation on Arg165, resulting in a premature Opal stop codon (Table S3).
191
192
Discussion
193
We regard the ball python as a powerful new model for understanding the genetic basis of
194
pigmentation in vertebrates. As proof-of-principle, we set out to discover candidate genes and
195
causal mutations for the classic colour morph found in the pet trade, the piebald. Over the last
196
thirty years, ball python breeders have discovered and successfully propagated in captivity
197
more than three hundred Mendelian phenotypes. Yet, researchers working on the genetics of
198
vertebrate colouration have been slow to appreciate the ball python as a powerful new model
199

(but see Irizarry and Bryden 2016). We set out to bridge this gap. To do so, we appealed to
200
commercial ball python breeders to join our efforts in discovery, starting with uncovering the
201
genetic basis of the piebald phenotype. Commercial breeders have a vested interest in
202
acquiring the ability to identify heterozygotes for recessive phenotypes (i.e., carriers of
203
recessive alleles affecting colouration), and piebald is one such phenotype, although subtle
204
phenotypic differences on heterozygotes are not uncommon (e.g. ring of unpigmented skin,
205
often near the vent).
206
To elucidate the genetic basis of the piebald phenotype, we combined a case-control
207
experimental design with whole-genome sequencing, population genetic methods, and
208
functional annotation of the candidate SNPs. We identified a premature stop codon in the
209
protein-coding region of the gene TFEC as the putative genetic basis for this Mendelian
210
phenotype. TFEC belongs to a family of transcription factors (MITF) known to be expressed in
211
the neural crest and pigment cells. However, functional validation through gene editing
212
techniques, like CRISPR cas9 (Rasys et al. 2019), is required to establish that the candidate
213
nonsense mutation is the causal variant.
214
With knowledge of the specific genetic variants underlying pigmentation variation in ball
215
pythons, gene editing could pave the way to produce colour morphs free of secondary
216
phenotypes that are undesired from a commercial or ethical perspective. Genes that contribute
217
to colouration are also expressed in many other tissues derived from neural crest cells
218
(Donoghue et al. 2008). Thus, mutations that alter colour or patterning often alter other traits,
219
such as behaviour (Ducrest et al. 2008; McKinnon and Pierotti 2010). Not surprisingly, there are
220

ball python colour morphs linked to morphological deformities and neurological dysfunction
221
(Rose and Williams 2014). For example, the incomplete dominant ‘spider’ morph is associated
222
with a neurological syndrome called the ‘head wobble’ (Fox and Hogan 2020). The propagation
223
of colour morphs with correlated traits have implications for the health and welfare of ball
224
pythons in captivity (Rose and Williams 2014) and are a source of current controversy among
225
hobbyists and commercial breeders (Riera 2015). In 2017, the International Herpetological
226
Society in the U. K. banned ball pythons with the spider mutation, which is thought to be
227
homozygous lethal, from their reptile shows (International Herpetological Society 2020).
228
However, it is not known if the correlated trait(s) for any particular colour morph is due to
229
pleiotropic effects or linked variation. If due to linked variation, targeted mutation on a
230
different genetic background (i.e., without the linked variant) could allow the reproduction of
231
colour phenotypes free of behavioural and morphological abnormalities. If due to pleiotropy,
232
targeted mutation at different nearby loci could achieve the same effect.
233
The wide variety of ball python phenotypes found in the pet trade offers the
234
opportunity to study how genes control vertebrate pigmentation and pattern formation.
235
Changes to pigmentation are often caused by mutations in genes involved in the pigment
236
synthesis pathway (Mills and Patterson 2009; Hubbard et al. 2010). For example, albinism
237
occurs when an individual inherits two copies of a gene involved in melanin synthesis (e.g.
238
OCA2) that is non-functional. Snakes with two non-functional OCA2 genes lack dark melanin-
239
based pigment though contain yellow pigments produced in the xanthophores (Saenko et al.
240
2015). Changes to pattern, on the other hand, are often attributed to genes with functions in
241
cell migration from the neural crest and differentiation into colour-producing cells (Mills and
242

Patterson 2009). However, Mendelian phenotypes characterized by changes to either
243
pigmentation or patterning are likely to map to the protein-coding regions of genes. Genomic
244
research on the most extensively studied vertebrate, humans, shows that Mendelian
245
phenotypes are generally caused by mutations arising in protein-coding regions and affecting
246
protein function (Cooper et al. 2010; Chong et al. 2015). The progenitor cells of
247
chromatophores originate in the neural crest and mutations to these genes result in partial or
248
total loss of pigment and structural colouration. Mutations to genes such as MITF, KIT, EDNRB,
249
and SOX10, which are expressed in the neural crest, results in white spotting or a complete lack
250
of pigment across a wide range of vertebrates, including fish, amphibians, birds, and mammals
251
(Baxter et al. 2004; Kelsh 2006; George et al. 2016; Ahi and Sefc 2017; Woodcock et al. 2017;
252
Camargo-Sosa et al. 2019; Goding and Arnheiter 2019). These genes represented a priori
253
candidates for the piebald phenotype in ball pythons, but did not show differentiation between
254
our phenotypic pools.
255
We identified a putative causal mutation for the piebald phenotype in ball pythons
256
located in a new candidate gene for vertebrate pigmentation: the gene TFEC (Figure 3, Table
257
S3). This variant, which produces a premature stop codon, was nearly fixed (FST = 0.96).
258
Complete differentiation between the piebald and non-piebald pools was prevented by a single
259
read for the reference allele that was sequenced in the piebald samples. It is possible this is the
260
result of a sequencing error or the inclusion of an unknown heterozygote. TFEC was not one of
261
the candidate genes expected a priori. This particular gene has not been implicated in
262
mammalian colouration. However, TFEC is part of the MIT-family of transcription factors known
263
to be involved in animal patterning. For example, this family consists of the gene MITF, which
264

forms heterodimers with the other three members, TFEB, TFE3, and TFEC (Goding and
265
Arnheiter 2019). In mouse models, double knockout mutations to MITF and TFE3 cause all-
266
white phenotypes (Steingrímsson et al. 2002). Interestingly, the same study showed that
267
knocking out TFEC did not result in a phenotype different from the wild-type mouse. There is
268
evidence that shows TFEC is expressed in the retinal pigment epithelium of mammals (Rowan et
269
al. 2004). No evidence suggests, however, that TFEC is expressed in cells that descend from the
270
mammalian neural crest, which includes pigment-producing cells. In contrast, recent evidence
271
from the zebrafish model shows that TFEC is highly expressed in neural crest cells, specifically
272
progenitor pigment cells and iridophores (Petratou et al. 2019; Saunders et al. 2019). In fact,
273
TFEC is expressed in two dorsolateral patches on the torso during embryogenesis (Lister et al.
274
2011; Petratou et al. 2018; Petratou et al. 2019), reminiscent of where the white-spotting
275
occurs in the piebald phenotype in ball pythons. The expression patterns of TFEC in zebrafish,
276
together with our results, makes a strong case that TFEC is the piebald gene in ball pythons and
277
that a premature stop codon is the likely causal mutation.
278
279
Conclusion
280
In this study, we aimed to uncover, for the first time, the genetic basis of a ball python colour
281
morph. We discovered that a premature stop codon in the gene TFEC is the likely cause for the
282
Mendelian piebald phenotype found in ball pythons from the pet trade. The lack of evidence for
283
TFEC affecting mammalian pigmentation highlights the need to study the genetic basis of
284
colouration in non-mammalian vertebrate models. Future work will be aimed at demonstrating
285

causality through direct functional manipulation. These efforts would produce the first
286
genetically engineered ball python, a step that would open the door to safe reproduction of
287
other colour morphs through targeted mutation.
288
289
Acknowledgments
290
RDHB was supported by an NSERC Discovery Grant and Canada Research Chair. We thank
291
Mutation Creation, T Dot Exotics, The Ball Room Canada, and Designing Morphs for supplying
292
samples.
293
294
Data accessibility
295
The data that support the findings of this manuscript will be uploaded to Dryad.
296
297
Contributions
298
AGE, HR, APH, and RDHB conceived and designed the study. HR collected samples. AGE
299
performed the molecular work. AGE performed the bioinformatics and analyzed the genomic
300
data. AGE wrote the manuscript with input from all authors.
301
302
REFERENCES
303

Ahi, E.P. and Sefc, K.M., 2017. A gene expression study of dorso-ventrally restricted pigment
304
pattern in adult fins of Neolamprologus meeli, an African cichlid species. PeerJ, 5,
305
p.e2843.
306
Barrett, R.D., Laurent, S., Mallarino, R., Pfeifer, S.P., Xu, C.C., Foll, M., Wakamatsu, K., Duke-
307
Cohan, J.S., Jensen, J.D. and Hoekstra, H.E., 2019. Linking a mutation to survival in wild
308
mice. Science, 363(6426), pp
309
Baxter, L.L., Hou, L., Loftus, S.K. and Pavan, W.J., 2004. Spotlight on spotted mice: a review of
310
white spotting mouse mutants and associated human pigmentation disorders. Pigment
311
Cell Research, 17(3), pp.215-224..499-504.
312
Bechtel, H.B., 1978. Color and pattern in snakes (Reptilia, Serpentes). Journal of Herpetology,
313
pp.521-532.
314
Bridges, C.B., 1919. The genetics of purple eye color in Drosophila. Journal of Experimental
315
Zoology, 28(2), pp.265-305.
316
A brief history of Royal Ball Python morphs for beginners. Retrieved from
317
http://bagnallsreptiles.weebly.com/history-into-morphs.html
318
Burgon, J.D., Vieites, D.R., Jacobs, A., Weidt, S.K., Gunter, H.M., Steinfartz, S., Burgess, K.,
319
Mable, B.K. and Elmer, K.R., 2020. Functional colour genes and signals of selection in
320
colour‐polymorphic salamanders. Molecular ecology, 29(7), pp.1284-1299.
321
Camargo-Sosa, K., Colanesi, S., Müller, J., Schulte-Merker, S., Stemple, D., Patton, E.E. and
322
Kelsh, R.N., 2019. Endothelin receptor Aa regulates proliferation and differentiation of
323
Erb-dependent pigment progenitors in zebrafish. PLoS genetics, 15(2), p.e1007941.
324
Caro, T., 2017. Wallace on coloration: contemporary perspective and unresolved insights.
325
Trends in ecology & evolution, 32(1), pp.23-30.
326
Castoe, T.A., de Koning, A.J., Hall, K.T., Card, D.C., Schield, D.R., Fujita, M.K., Ruggiero, R.P.,
327
Degner, J.F., Daza, J.M., Gu, W. and Reyes-Velasco, J., 2013. The Burmese python
328
genome reveals the molecular basis for extreme adaptation in snakes. Proceedings of
329
the National Academy of Sciences, 110(51), pp.20645-20650.
330
Chong, J.X., Buckingham, K.J., Jhangiani, S.N., Boehm, C., Sobreira, N., Smith, J.D., Harrell, T.M.,
331
McMillin, M.J., Wiszniewski, W., Gambin, T. and Akdemir, Z.H.C., 2015. The genetic basis
332
of Mendelian phenotypes: discoveries, challenges, and opportunities. The American
333
Journal of Human Genetics, 97(2), pp.199-215.
334

Cingolani, P., Platts, A., Wang, L.L., Coon, M., Nguyen, T., Wang, L., Land, S.J., Lu, X. and Ruden,
335
D.M., 2012. A program for annotating and predicting the effects of single nucleotide
336
polymorphisms, SnpEff: SNPs in the genome of Drosophila melanogaster strain w1118;
337
iso-2; iso-3. Fly, 6(2), pp.80-92.
338
Cohen, M.A., Wert, K.J., Goldmann, J., Markoulaki, S., Buganim, Y., Fu, D. and Jaenisch, R., 2016.
339
Human neural crest cells contribute to coat pigmentation in interspecies chimeras after
340
in utero injection into mouse embryos. Proceedings of the National Academy of
341
Sciences, 113(6), pp.1570-1575.
342
Cooper, D.N., Chen, J.M., Ball, E.V., Howells, K., Mort, M., Phillips, A.D., Chuzhanova, N.,
343
Krawczak, M., Kehrer‐Sawatzki, H. and Stenson, P.D., 2010. Genes, mutations, and
344
human inherited disease at the dawn of the age of personalized genomics. Human
345
mutation, 31(6), pp.631-655.
346
Cuthill, I.C., Allen, W.L., Arbuckle, K., Caspers, B., Chaplin, G., Hauber, M.E., Hill, G.E., Jablonski,
347
N.G., Jiggins, C.D., Kelber, A. and Mappes, J., 2017. The biology of color. Science,
348
357(6350), p.eaan0221.
349
Donoghue, P.C., Graham, A. and Kelsh, R.N., 2008. The origin and evolution of the neural crest.
350
Bioessays, 30(6), pp.530-541.
351
Ducrest, A.L., Keller, L. and Roulin, A., 2008. Pleiotropy in the melanocortin system, coloration
352
and behavioural syndromes. Trends in ecology & evolution, 23(9), pp.502-510.
353
Dudchenko, O., Batra, S.S., Omer, A.D., Nyquist, S.K., Hoeger, M., Durand, N.C., Shamim, M.S.,
354
Machol, I., Lander, E.S., Aiden, A.P. and Aiden, E.L., 2017. De novo assembly of the
355
Aedes aegypti genome using Hi-C yields chromosome-length scaffolds. Science,
356
356(6333), pp.92-95.
357
Dudchenko, O., Shamim, M.S., Batra, S., Durand, N.C., Musial, N.T., Mostofa, R., Pham, M.,
358
Hilaire, B.G.S., Yao, W., Stamenova, E., Hoeger, M., Nyquist, S.K., Korchina, V., Pletch, K.,
359
Flanagan, J.P., Tomaszewicz, A., McAloose, D., Estrada, C.P., Novak, B.J., Omer, A.D.,
360
Aiden, E.L., 2018. The Juicebox Assembly Tools module facilitates de novo assembly of
361
mammalian genomes with chromosome-length scaffolds for under $1000. bioRxiv
362
254797. https://doi.org/10.1101/254797.
363
Ellis, T.N., Hofer, J.M., Timmerman-Vaughan, G.M., Coyne, C.J. and Hellens, R.P., 2011. Mendel,
364
150 years on. Trends in plant science, 16(11), pp.590-596.
365

Fleischman, R.A., Saltman, D.L., Stastny, V. and Zneimer, S., 1991. Deletion of the c-kit
366
protooncogene in the human developmental defect piebald trait. Proceedings of the
367
National Academy of Sciences, 88(23), pp.10885-10889.
368
Fox, B.C. and Hogan, S., 2020. Hypopigmentation as the Cause for the Neurological Disorder
369
“Spider Wobble” in Python regius.
370
George, A., Zand, D.J., Hufnagel, R.B., Sharma, R., Sergeev, Y.V., Legare, J.M., Rice, G.M.,
371
Schwoerer, J.A.S., Rius, M., Tetri, L. and Gamm, D.M., 2016. Biallelic mutations in MITF
372
cause coloboma, osteopetrosis, microphthalmia, macrocephaly, albinism, and deafness.
373
The American Journal of Human Genetics, 99(6), pp.1388-1394.
374
Goding, C.R. and Arnheiter, H., 2019. MITF—the first 25 years. Genes & development, 33(15-
375
16), pp.983-1007.
376
Grether, G.F., Kolluru, G.R. and Nersissian, K., 2004. Individual colour patches as
377
multicomponent signals. Biological Reviews, 79(3), pp.583-610.
378
Harris RB, Irwin K, Jones MR, Laurent S, Barrett RD, Nachman MW, Good JM, Linnen CR, Jensen
379
JD, Pfeifer SP. The population genetics of crypsis in vertebrates: recent insights from
380
mice, hares, and lizards. Heredity. 2019 Aug 9:1-4.
381
Harrison, J.H., 1920. Genetical studies in the moths of the geometrid genus Oporabia (Oporinia)
382
with a special consideration of melanism in the lepidoptera. Journal of Genetics, 9(3),
383
pp.195-280.
384
Haupaix, N., Curantz, C., Bailleul, R., Beck, S., Robic, A. and Manceau, M., 2018. The periodic
385
coloration in birds forms through a prepattern of somite origin. Science, 361(6408),
386
p.eaar4777.
387
Hoekstra, H.E., 2006. Genetics, development and evolution of adaptive pigmentation in
388
vertebrates. Heredity, 97(3), pp.222-234.
389
International Herpetological Society. (2020) IMPORTANT NOTICE – BANNED MORPHS. June 19,
390
2020. https://www.ihs-web.org.uk/news/show-news/important-notice-banned-species/
391
Irizarry, K.J. and Bryden, R.L., 2016. In silico analysis of gene expression network components
392
underlying pigmentation phenotypes in the Python identified evolutionarily conserved
393
clusters of transcription factor binding sites. Advances in bioinformatics, 2016.
394
Kelsh, R.N., 2006. Sorting out Sox10 functions in neural crest development. Bioessays, 28(8),
395
pp.788-798.
396

Li, H., Handsaker, B., Wysoker, A., Fennell, T., Ruan, J., Homer, N., Marth, G., Abecasis, G. and
397
Durbin, R., 2009. The sequence alignment/map format and SAMtools. Bioinformatics,
398
25(16), pp.2078-2079.
399
Lister, J.A., Lane, B.M., Nguyen, A. and Lunney, K., 2011. Embryonic expression of zebrafish MiT
400
family genes tfe3b, tfeb, and tfec. Developmental Dynamics, 240(11), pp.2529-2538.
401
Mills, M.G. and Patterson, L.B., 2009, February. Not just black and white: pigment pattern
402
development and evolution in vertebrates. In Seminars in cell & developmental biology
403
(Vol. 20, No. 1, pp. 72-81). Academic Press.
404
McKinnon, J.S. and Pierotti, M.E., 2010. Colour polymorphism and correlated characters:
405
genetic mechanisms and evolution. Molecular ecology, 19(23), pp.5101-5125.
406
Mendel, G. (1866) Versuche u¨ber pflanzen-hybriden. Verhandlungen der naturfoschung
407
Vereins 4, 3–47
408
Oiso, N., Fukai, K., Kawada, A. and Suzuki, T., 2013. Piebaldism. The Journal of Dermatology,
409
40(5), pp.330-335.
410
Olsson, M., Stuart-Fox, D. and Ballen, C., 2013, June. Genetics and evolution of colour patterns
411
in reptiles. In Seminars in cell & developmental biology (Vol. 24, No. 6-7, pp. 529-541).
412
Academic Press.
413
Petratou, K., Subkhankulova, T., Lister, J.A., Rocco, A., Schwetlick, H. and Kelsh, R.N., 2018. A
414
systems biology approach uncovers the core gene regulatory network governing
415
iridophore fate choice from the neural crest. PLoS genetics, 14(10), p.e1007402.
416
Petratou, K., Spencer, S.A., Kelsh, R.N. and Lister, J.A., 2019. The MITF paralog tfec is required in
417
neural crest development for fate specification of the iridophore lineage from a
418
multipotent pigment cell progenitor. bioRxiv, p.862011.
419
Pingault, V., Bondurand, N., Kuhlbrodt, K., Goerich, D.E., Préhu, M.O., Puliti, A., Herbarth, B.,
420
Hermans-Borgmeyer, I., Legius, E., Matthijs, G. and Amiel, J., 1998. SOX10 mutations in
421
patients with Waardenburg-Hirschsprung disease. Nature genetics, 18(2), pp.171-173.
422
Rasys, A.M., Park, S., Ball, R.E., Alcala, A.J., Lauderdale, J.D. and Menke, D.B., 2019. CRISPR-Cas9
423
Gene Editing in Lizards through Microinjection of Unfertilized Oocytes. Cell reports,
424
28(9), pp.2288-2292.
425
Richards, J.R., Yoo, J.H., Shin, D. and Odelberg, S.J., 2019. Mouse models of uveal melanoma:
426
Strengths, weaknesses, and future directions. Pigment Cell & Melanoma Research.
427

Riera, D., 2015. BALL PYTHON GENETICS: MORPH COMBINATIONS TO BE CAUTIOUS OF.
428
https://www.rierareptiles.com/ball-python-genetics-anomalies/
429
Rose, M.P. and Williams, D.L., 2014. Neurological dysfunction in a ball python (Python regius)
430
colour morph and implications for welfare. Journal of Exotic Pet Medicine, 23(3),
431
pp.234-239.
432
Rowan, S., Chen, C.M.A., Young, T.L., Fisher, D.E. and Cepko, C.L., 2004. Transdifferentiation of
433
the retina into pigmented cells in ocular retardation mice defines a new function of the
434
homeodomain gene Chx10. Development, 131(20), pp.5139-5152.
435
San-Jose, L.M. and Roulin, A., 2017. Genomics of coloration in natural animal populations.
436
Philosophical Transactions of the Royal Society B: Biological Sciences, 372(1724),
437
p.20160337.
438
San-Jose, L.M. and Roulin, A., 2018. Toward understanding the repeated occurrence of
439
associations between melanin-based coloration and multiple phenotypes. The American
440
Naturalist, 192(2), pp.111-130.
441
Saenko, S.V., Teyssier, J., Van Der Marel, D. and Milinkovitch, M.C., 2013. Precise colocalization
442
of interacting structural and pigmentary elements generates extensive color pattern
443
variation in Phelsuma lizards. BMC biology, 11(1), p.105.
444
Saenko, S.V., Lamichhaney, S., Barrio, A.M., Rafati, N., Andersson, L. and Milinkovitch, M.C.,
445
2015. Amelanism in the corn snake is associated with the insertion of an LTR-
446
retrotransposon in the OCA2 gene. Scientific reports, 5, p.17118.
447
Saunders, L.M., Mishra, A.K., Aman, A.J., Lewis, V.M., Toomey, M.B., Packer, J.S., Qiu, X.,
448
McFaline-Figueroa, J.L., Corbo, J.C., Trapnell, C. and Parichy, D.M., 2019. Thyroid
449
hormone regulates distinct paths to maturation in pigment cell lineages. Elife, 8,
450
p.e45181.
451
Schweizer, R.M., Durvasula, A., Smith, J., Vohr, S.H., Stahler, D.R., Galaverni, M., Thalmann, O.,
452
Smith, D.W., Randi, E., Ostrander, E.A. and Green, R.E., 2018. Natural selection and
453
origin of a melanistic allele in North American gray wolves. Molecular biology and
454
evolution, 35(5), pp.1190-1209.
455
Sedlazeck, F.J., Rescheneder, P. and Von Haeseler, A., 2013. NextGenMap: fast and accurate
456
read mapping in highly polymorphic genomes. Bioinformatics, 29(21), pp.2790-2791.
457
Steingrímsson, E., Tessarollo, L., Pathak, B., Hou, L., Arnheiter, H., Copeland, N.G. and Jenkins,
458
N.A., 2002. Mitf and Tfe3, two members of the Mitf-Tfe family of bHLH-Zip transcription
459

factors, have important but functionally redundant roles in osteoclast development.
460
Proceedings of the National Academy of Sciences, 99(7), pp.4477-4482.
461
Tachibana, M., Kobayashi, Y. and Matsushima, Y., 2003. Mouse models for four types of
462
Waardenburg syndrome. Pigment cell research, 16(5), pp.448-454.
463
Teyssier J, Saenko SV, Van Der Marel D, Milinkovitch MC. Photonic crystals cause active colour
464
change in chameleons. Nature communications. 2015 Mar 10;6:6368.
465
Ullate-Agote, A., Burgelin, I., Debry, A., Langrez, C., Montange, F., Peraldi, R., Daraspe, J.,
466
Kaessmann, H., Milinkovitch, M.C. and Tzika, A.C., 2020. Genome mapping of a LYST
467
mutation in corn snakes indicates that vertebrate chromatophore vesicles are lysosome-
468
related organelles. Proceedings of the National Academy of Sciences, 117(42),
469
pp.26307-26317.
470
Woodcock, M.R., Vaughn-Wolfe, J., Elias, A., Kump, D.K., Kendall, K.D., Timoshevskaya, N.,
471
Timoshevskiy, V., Perry, D.W., Smith, J.J., Spiewak, J.E. and Parichy, D.M., 2017.
472
Identification of mutant genes and introgressed tiger salamander DNA in the laboratory
473
axolotl, Ambystoma mexicanum. Scientific reports, 7(1), pp.1-10.
474
Wysoker, A., Tibbetts, K. and Fennell, T., 2013. Picard tools version 1.90. http://picard.
475
sourceforge. net (Accessed 14 December 2016), 107(17), p.308.
476
477

Figures
478
479
Figure 1. A small sample of the phenotypic variation found in captive-bred ball pythons (Python regius).
480
(a) wild-type, (b) piebald, (c) banana piebald, (d) pastel piebald, (e) pastel HRA enhancer, (f) ultramel
481
clown, (g) banana champagne. Photo credit (b-g): Designing Morphs.
482
483
484
485
486
487

488
Figure 2. Fst plot between piebald and non-piebald samples on the chromosome-length assembly
489
(Python_molurus_bivittatus-5.0.2_HiC.assembly).
490
491

492
493
Figure 3. A premature stop codon on Arg165 is the
494
putative causal variant for the piebald phenotype.
495
496
497
498

Supplementary figures
499
500
Figure S1. Fst plot of SNPs mapped to the Burmese python draft assembly (Pmo2.0). The 129 fixed SNPs
501
in this plot were used to map to candidate genes.
502
503
504
505