Comparative transcriptomics of albino and warningly‐coloured caterpillars

Abstract Coloration is perhaps one of the most prominent adaptations for survival and reproduction of many taxa. Coloration is of particular importance for aposematic species, which rely on their coloring and patterning acting as a warning signal to deter predators. Most research has focused on the evolution of warning coloration by natural selection. However, little information is available for color mutants of aposematic species, particularly at the genomic level. Here, I compare the transcriptomes of albino mutant caterpillars of the aposematic wood tiger moth (Arctia plantaginis) to those of their full sibs having their distinctive orange‐black warning coloration. The results showed >290 differentially expressed genes genome‐wide. Genes involved in the immune system, structural constituents of cuticular, and immunity were mostly downregulated in the albino caterpillars. Surprisingly, higher expression was observed in core melanin genes from albino caterpillars, suggesting that melanin synthesis may be disrupted in terminal ends of the pathway during its final conversion. Taken together, these results suggest that caterpillar albinism may not be due to a depletion of melanin precursor genes. In contrast, the albino condition may result from the combination of faulty melanin conversion late in its synthesis and structural deficiencies in the cuticular preventing its deposition. The results are discussed in the context of how albinism may impact individuals of aposematic species in the wild.

acid identity, alignment length of < 100 bp amino acid length, and e-value ≤ 10− 5 were 163! removed. Gene ontology terms (GO) and information of protein family was obtained using 164! Blast2Go v.4.0 (Conesa, et al. 2005 To investigate gene expression differences between the two larval conditions, the high 169! quality reads from the four samples were first mapped to the wood tiger moth's reference 170! transcriptome (Galarza, et al. 2017) using bowtie2 v. 2.2.5 (Langmead and Salzberg 2012).

171!
The number of mapped reads for each sample was counted using SAMtools v.1.3.1 (Li 2009) 172! and merged into a count matrix which was then normalised by transcript length and 173! sequencing depth. Here, for each transcript, the number of mapped (Mr) reads was divided 174! by the total number of reads (Tr), multiplied by transcript length (Tl) scaled by a factor of a 175! million (i.e. (Mr/Tr)* 10^9 ). This procedure returns normalized counts as transcripts per every 176! million reads sequenced (TPM), in which the sum of all TPMs in each sample is the same, 177! thus allowing a direct comparison of normalized expression values across samples and 178! conditions. The R package edgeR (Robinson, et al. 2010) was used to test for differential 179! gene expression setting a P-value cut-off threshold of 0.05, with a Benjamin and Hochberg 180! correction (Benjamini and Hochberg 1995) for multiple testing. Subsequently, a functional 181! annotation of the differentially expressed genes between conditions was obtained by blasting 182! (BLASTx) the up-and down-regulated genes against non-redundant protein databases (nr) 183! (NCBI; last updated 15-06-2017) the Swiss-Prot (last updated 25-06-2017). All hits that 184! showed < 70% amino acid identity, alignment length of < 100 bp, and e-value ≤ 10− 5 were 185! excluded and the gene ontology terms (GO) and information of protein family was obtained 186! using Blast2Go v.4.0 (Conesa, et al. 2005 To validate gene expression results from RNA-seq data, a subset of 6 differentially 191! expressed gene transcripts was evaluated through quantitative PCR (qPCR). The expression 192! of three gene transcripts involved in larval cuticular processes and three other non-annotated 193! random gene transcripts, were selected for qPCR comparison to the RNA-seq data.  intron boundaries were first identified by aligning these gene transcripts to genomic data 195! previously obtained through 454 (Life Sciences) pyrosequencing using Mummer v.3.23 196! (Kurtz, et al. 2004). Bridging primers were then designed using Primer3 v. 4.0.0 197! (Untergasser, et al. 2012). As a normalization control (i.e. housekeeping gene), I selected one 198! transcript from the RNA-seq data, which showed a uniform expression level within and 199! between the two larval conditions. The software Normfinder v.5 (Andersen, et al. 2004) was 200! used to evaluate the normalized count matrix to find the transcript with the highest stability 201! value and lowest expression variation within and between the two conditions. The annotation 202! and primer sequences of the genes transcripts used for qPCR are presented in supplementary 203!

233!
Here I took a candidate gene approach to evaluate expression patterns in genes from 234! the melanin biosynthesis pathway in larvae from both conditions. The candidate genes 235! investigated were tyrosine hydroxylase (TH), yellow, laccase2, DOPA decarboxylase (Ddc), 236! arylalkylamine N-acetyltransferase (aaNAT), tan, and ebony. These genes or the enzymes 237! ! 10! they encode have been shown to impact insect black pigmentation and patterning (Fujii, et al. 238! 2013;Liu, et al. 2016;van't Hof and Saccheri 2010). In addition, I evaluate expression of 239! tetrahydrobiopterin (BH4), a cofactor outside the melanin pathway that impacts the 240! hydroxylation of tyrosine, the precursor of melanin synthesis (supplementary material S9). It 241! was recently found that mutations in 6-pyruvoyl-tetrahydropterin synthase (PTS), the gene 242! that encodes for BH4, can promote albinism in the silkmoth (Fujii, et al. 2013

298!
Four candidate genes (aaNAT, ebony, Ddc, TH) showed significantly higher 299! expression in the albino condition, whereas three others (yellow, laccase2, tan) were more 300! expressed in the coloured condition. PTS on the other hand, showed non-significant 301! differences (P > 0.05) in its expression levels between conditions ( Table 2). The most 302! differentially expressed genes were TH and tan, being highly expressed in the albino and 303! coloured conditions respectively ( Figure 6). 304!

307!
In this study, I report the transcriptome characterisation and comparison between 308! albino wood tiger moth caterpillars and their warningly coloured siblings. The results showed 309! substantial differences transcriptome-wide. Processes such as immune response and cuticle 310! formation were found significantly reduced in albino caterpillars. Albinism also appears to be 311! ! 13! associated with depressed immunity and with a high activity in processes essential for 312! genome replication and subsequent RNA transcription. Moreover, core-melanin genes (TH, 313! Ddc) showed a higher expression in albino larvae. However, a down-regulation of genes 314! (yellow, laccase2) involved in final melanin conversion was observed. Taken together, these 315! results suggest that caterpillar albinism may not be due to a depletion of melanin precursor 316! genes as it could be expected. In contrast, the albino condition may result from the 317! combination of faulty melanin conversion late in its synthesis and structural deficiencies in 318! the cuticle preventing its deposition. 319! 320! Albinism Frequency 321!

322!
Albinism in humans has been extensively studied and its frequency is well 323! documented for distinct populations among different geographic regions (Montoliu, et al. 324! 2014). For wild species however, little information is available about its prevalence.

325!
Albinism seems to be a common phenomenon in species living in lightless environments 326! such as caves or subterranean habitats (Bilandžija, et al. 2012;Bilandžija, et al. 2017;327! Oliveira and Aguiar 2008;Protas, et al. 2006). However, it can be observed also in 328! aboveground species. For instance, albino chorus frogs (Pseudacris triseriata) have been 329! reported in frequencies of 7% and 12% during two consecutive years from natural ponds 330! (Corn 1986). Likewise, scatter reports of albino Viperinae (Vipera ammodytes, V. aspis, V. 331! seoanei, and V. berus) in Europe have been collected ranging from 1 to 16 observations 332! depending on the species with an increase in frequency towards Nordic populations. (Krecsák 333! 2008). In Lepidoptera, cases of partial albinism have been observed in alpine butterflies, 334! Erebia cpiphron silesiana and E.sudeiica sudetica with a frequencies ranging from 0.03-335! 1.4% in E.cpiphron to 0.7-3.9% in E. sudeiica (Kuras, et al. 2001 (2006) found that white-black mutants of the aposematic firebug 348! (Pyrrhocoris apterus) were more attacked that the red-black wild-type, and equally attacked 349! as non-aposematic grey-black controls. On the other hand, in a later experiment, naïve, hand-350! reared P.major did not show any avoidance and attacked firebugs equally irrespective of 351! colour (Svadová, et al. 2009). Data from a field study in two alpine butterflies (E.cpiphron, E. 352! sudeiica) showed that predation marks and malformations were positively associate with 353! albinism in both species (Kuras, et al. 2001). It is worth noting that these studies examined 354! partial albinos having some degree of dark pigmentation. The effect of predation on wholly 355! albinos like those reported here remains to be examined. As suggested by the previous studies 356! (Exnerová, et al. 2006;Svadová, et al. 2009 In wood tiger moth larvae, the orange patch contrasts with the dark body conforming the 370! warning signal, which is more effective against visual predators when the orange patch is 371! large (Lindstedt, et al. 2009). The patch is made of clusters of chitin hairs pigmented with 372! eumelanin and diet-derived flavonoids that give it its orange colouration (Lindstedt, et al. 373! 2010). The role of the orange pigmentation in physiology is unknown, but it is likely 374! negligible. The black hairs on the other hand, contain only eumelanin , 375! a type of melanin that produces the black pigmentation. Melanin is central for insect 376! immunity (Tsakas and Marmaras 2010). Experiments with high-and low-melanin wood tiger 377! moth larvae have shown a better resistance to oral bacterial infections in high-melanin larvae 378! (Zhang, et al. 2012). Likewise, more melanised larvae showed a faster encapsulation 379! response to artificial implants than less melanised ones (Nokelainen, et al. 2013). By analogy, 380! the complete absence of melanisation in albino larvae hints to a suppressed immunity. This 381! notion is supported by the gene expression results showing a down-regulation of genes 382! involved in immune system, defence response, and antifungal responses in the albino larvae 383! ( Figure 2). However, it has recently been shown that wound-healing melanisation can still 384! occur in most albino cave-adapted adapted species, including insects (Bilandžija, et al. 2017 Melanic dark pigmentation not only correlates with insects' immunity, but also 390! protects from deleterious solar radiation, and at the same time helps in thermoregulation 391! (Ellers and Boggs 2004;Stoehr and Goux 2008). It has been shown that darker wood tiger 392! moth larvae are more efficient in thermoregulation than larvae with larger orange patches 393! (Lindstedt, et al. 2009). More recently, it was found that darker larvae absorb more heat, keep 394! higher body temperature, and actively avoid overheating by seeking shade sooner than less 395! melanised larvae (Nielsen et al. in prep.). Such behavioural variation relative to the amount of 396! dark pigmentation could be expected to be exaggerated in albino larvae. Accordingly, albinos 397! may need partial sheltered thermoregulation due to the lack of protective dark pigmentation.

398!
This could result in longer heat-up periods, which in turn may reduce foraging time and also 399! increase their exposure to non-visual predators. If true, albinos may face different predation 400! pressures than their warningly coloured counterparts, for which, an aposematic strategy may 401! be irrelevant.