Challenging the concept that eumelanin is the polymorphic brown banded pigment in Cepaea nemoralis

The common grove snail Cepaea nemoralis displays a stable pigmentation polymorphism in its shell that has held the attention of scientists for decades. While the details of the molecular mechanisms that generate and maintain this diversity remain elusive, it has long been employed as a model system to address questions related to ecology, population genetics and evolution. In order to contribute to the ongoing efforts to identify the genes that generate this polymorphism we have tested the long-standing assumption that melanin is the pigment that comprises the dark-brown bands. Surprisingly, using a newly established analytical chemical method, we find no evidence that eumelanin is differentially distributed within the shells of C. nemoralis. Furthermore, genes known to be responsible for melanin deposition in other metazoans are not differentially expressed within the shell-forming mantle tissue of C. nemoralis. These results have implications for the continuing search for the supergene that generates the various pigmentation morphotypes.

or pink shells (Fig. 1C,D). We also detected the pheomelanin markers TDCA and TTCA in both brown-bands derived from yellow background and yellow background shell fractions in relatively high abundance, but again with no significant difference between the two (Fig. 1D). In pink shells we detected far less TDCA in both fractions while TTCA was below the limit of quantitation but could still be detected (Table 1).
To further investigate the presence of melanin in the brown bands of C. nemoralis shells we studied the expression of four known melanin synthesis pathway genes in regions of the mantle responsible for depositing the brown bands in to the shell ( Fig. 2A). We quantified gene expression via qPCR using reference genes we recently validated for C. nemoralis 15 . We selected the genes Yellow and Laccase 2 (known to be involved in melanin synthesis in insects [17][18][19][20][21]. Furthermore Miyashita and Takagi 22 recently showed that a tyrosinase is involved in the subtle pigmentation of oyster pearls, while Vicario et al. 13 previously suggested that either a Tyrosinase or a Tyrosinase related protein is involved in the melanic pigmentation of C. nemoralis shells. Phylogenetic analyses of the four genes that we studied are provided in the Supplementary Material ( Supplementary Files 1-3).
Despite considerable variation Cnem-Yellow-like was significantly differentially expressed in pigmented vs. non-pigmented regions of the mantle (p < 0.05 but not <0.01; Fig. 2B). However Cnem-Yellow-like transcripts were significantly more abundant (approximately three fold) in foot tissue relative to mantle tissue overall. www.nature.com/scientificreports www.nature.com/scientificreports/ This suggests that either Cnem-Yellow-like has pleiotropic roles in the foot and mantle, or that there is another rate-limiting enzyme or process that masks the differential activity of Cnem-Yellow-like between mantle and foot tissues, or that Cnem-Yellow-like is not involved in melanogenesis in C. nemoralis. Supporting this latter scenario the Cnem-Yellow-like homolog we identified here has relatively low sequence similarity (25% identity) with insect isoforms known to be involved in melanogenesis, and does not phylogenetically group with canonical insect Yellow proteins (Supplementary File S2). Cnem-Tyrosinase related protein was significantly upregulated in mantle vs. foot, however comparisons between pigmented and non-pigmented mantle tissue revealed no significant differences in expression (Fig. 2B). In addition, Cnem-Tyrosinase and Cnem-Laccase 2 were equally expressed Sample PDCA (µg/g) ± SD PTCA (µg/g) ± SD TDCA (µg/g) ± SD TTCA (µg/g) ± SD  Table 1. Melanin oxidation products normalised to shell weight. For each sample n = 3, means ± standard deviation. External calibration was set with 9-point calibration curves. Limit of quantitation (LOQ) was set at 10:1 signal-to-noise ratio (LOQ PDCA = 0.08 µg/ml), LOQ PTCA = 0.10 µg/ml, LOQ TDCA = 0.25 µg/ml, LOQ TTCA = 0.33 µg/ml). ddCt

Pigmented vs Unpigmented Mantle
Mantle vs Foot www.nature.com/scientificreports www.nature.com/scientificreports/ between foot and mantle tissues and also between pigmented and non-pigmented mantle tissue (Fig. 2). Although these qPCR results in themselves offer no support for the evidence of melanin synthesising activity in the mantle of C. nemoralis we must acknowledge that it remains possible that an inhibitory mechanism may differentially operate in pigmented and non-pigmented tissues.
While we do detect the presence of eu-and pheomelanin oxidation markers in shells of C. nemoralis, the spatially quantified LC-UV-MS and RT-qPCR data we present here suggests that eumelanin is not the dark banded pigment used by C. nemoralis. As far as we are aware the biochemical evidence that suggested melanin to be the responsible pigment in C. nemoralis was published more than 80 years ago 12 . Interestingly Comfort later states in regard to the banded pigment in C. nemoralis that "Attempts to attack the chemistry of the pigments with melanin inhibitors such as phenyl thiourea have not so far succeeded…. " and that "Very little can be said of these colour patterns in terms of chemistry. The black pigment ('melanocochlin') is alkali-soluble 12,23 ". Nonetheless, subsequent works refer to pigmented cells in the mantle as melanocytes 24 and have worked from the assumption that melanin is the brown-banded pigment. For example the upregulation of either tyrosinase or tyrosinase-related genes in C. nemoralis mantle tissue has been interpreted as evidence that melanin synthesis is occurring there 13 . However it is known that these enzymes also function to sclerotize the periostracum and are not necessarily involved in melanin synthesis 25,26 . We previously demonstrated that the brown-banded pigment in C. nemoralis is apparently not a chromoprotein 27 , and the results we present here indicate that a non-melanogenic mechanism is operating in the mantle tissue of C. nemoralis to deposit a dark-brown pigment in a banded fashion. A recent review on molluscan shell colour by Williams 28 highlighted the lack of data that unequivocally demonstrates the presence of melanin in molluscan shells, and our own recent survey of molluscan shells with brown/black pigments suggests that in many cases these pigments are not melanin 16 . What then are the brown-banded and background pigments in C. nemoralis shells? Currently our LC-UV-MS analyses do not reveal any hints as to the nature of these molecules, however a macromolecular composition may be supposed.
A chemical understanding of what the dark-banded pigment and the various background pigments are, together with all of their respective synthesis pathways and associated genes, would assist efforts in identifying the supergene that is thought to regulate the famous Cepaea colour polymorphism. Such a holistic understanding of this polymorphism would provide deep insight into the evolutionary biology of this system, and would allow C. nemoralis to join the short list of species for which a polymorphic trait is understood on molecular, phenotypic, population and ecological scales.

Cepaea nemoralis. Six living animals and approximately 100 empty shells of C. nemoralis were collected
at University of Göttingen, Germany (51°33′24.0″N 9°57′27.3″E). Empty shells were cleaned and dried, then crushed. Shell pieces were sorted according to replicate group and colour fraction. Tissue samples were taken from fresh material by careful dissection of mantle and foot tissue.

Sample preparation, melanin oxidation and LC-UV-MS analyses.
Two major morphs of C. nemoralis (pink banded and yellow banded) were investigated. Their corresponding colour-sorted shell fragments were as follows: pink background; brown band on pink background; yellow background; brown band on yellow background (see Fig. 1A). Three technical replicates of each colour group were performed, with each replicate comprised of up to 8 shells.
Analysis of eumelanin and pheomelanin oxidation products was carried out as previously described 14 (see main text): In brief, shells were cleaned in deionized water and weighted. Cleaned shell pieces were dissolved in 6 M HCl and centrifuged at 13,000 rpm for 15 min. Residues were washed with HPLC grade water. Samples were treated with proteinase K in 1 M Tris-HCl buffer at 37 °C for 2 h. Treatment was stopped by acidification with 6 M HCl, and samples were centrifuged and washed as described above.
Oxidation reactions and solid-phase extractions were performed as previously described 14 . For each sample, oxidation was performed for 20 h at 25 °C under vigorous shaking using 100 μL H 2 O, 375 μL 1 M K 2 CO 3 and 25 μL 30% H 2 O 2 . Remaining H 2 O 2 was decomposed by the addition of 50 μL 10% Na 2 SO 3 and the mixture was acidified with 140 μL 6 M HCl. Solutions were then centrifuged at 13,000 rpm for 40 min and the supernatants transferred to fresh tubes. Samples were then treated by solid-phase extraction on Phenomenex Strata-X 33 μm polymeric reversed phase columns under vacuum. Columns were first conditioned with 5 mL methanol (MeOH) followed by 5 mL H 2 O. Shell extracts were loaded onto the columns which were then washed with 5 mL 0.3% formic acid. The columns were then dried for 30 min and elution was carried out with 3 mL MeOH followed by 3 mL ethyl acetate. Solvents were removed under constant nitrogen stream at 40 °C and samples were dissolved in 200 μL H 2 O.
Measurements were also performed as previously described 14 . Briefly, a Thermo Fisher Scientific LC-MS system consisting of an Accela HPLC with a Finnigan Surveyor PDA Detector coupled to an LTQ Orbitrap XL mass spectrometer equipped with an electrospray ionisation (ESI) source was employed. Separation was performed on a Phenomenex Gemini C18 column (250 × 2 mm, 5 μm) at 45 °C using a flow rate of 0.2 mL/min. The mobile phase was 0.3% formic acid in H 2 O:MeOH (80:20). UV data were recorded between 200 and 400 nm. Mass spectra were acquired in negative ion mode over an m/z range of 120-220.

Reverse transcription quantitative pcR (qpcR) of melanin pathway genes in C. nemoralis.
Four genes known to be involved in melanin synthesis and dark pigmentation were chosen for qPCR testing: Cnem-Tyrosinase (Tyr), Cnem-Tyrosinase related protein (TyrRP), Cnem-Yellow-like (Yellow) and Cnem-Laccase 2 (Lacc2). These sequences were identified from within a C. nemoralis mantle tissue transcriptome data set using tBLASTx. Primers for qPCR were designed with Primer3 (https://primer3plus.com). Primer sequences and Genbank accession codes are listed in Table 2.

Scientific RepoRtS |
(2020) 10:2442 | https://doi.org/10.1038/s41598-020-59185-y www.nature.com/scientificreports www.nature.com/scientificreports/ The experimental design for RT-qPCRs followed the protocol described in Affenzeller et al. 15 : Six sub-adult individuals of C. nemoralis were collected at the University of Göttingen. Total RNA from pigmented mantle (producing the band in the shell), unpigmented mantle (producing background coloured shell) and foot tissue was extracted from each individual using Qiazol (Qiagen) according to the manufacturer's instructions resulting in a total of twelve RNA extractions. These underwent a DNase treatment (RQ1 RNase-free DNase, Promega) according to the manufacturer's instructions. Nanodrop and agarose gel electrophoresis were employed to verify quality and integrity of RNA. Synthesis of cDNA was carried out with 1 μg of total extracted RNA per sample using Promega M-MLV reverse transcriptase and oligo dTs. Reaction was run at 42 °C for 75 min, followed by 15 min at 70 °C to inactivate reverse transcriptase. The cDNA was stored at −20 °C until further use.
All qPCR runs followed a maximum sample layout, comply with the MIQE guidelines 29  qPCR. Raw fluorescence data were baseline and amplification efficiency corrected in LinRegPCR 30 . Inter run correction was performed using Factor-qPCR 31 . So gained corrected cycle threshold (Cq) values were used to calculate the geometric means of technical replicates. Normalisation and relative expression were calculated based on the Pfaffl method 32 with beta-actin (BACT) and EF1α serving as reference genes as previously tested for mantle tissue in C. nemoralis 15 .
Descriptive statistical analyses (mean and standard deviation calculations) of six biological replicates for each sample set (pigmented mantle, unpigmented mantle, foot) were carried out in Microsoft Excel for Office 365 MSO (16.0.11629.20192). Statistical comparisons between pigmented mantle and unpigmented mantle, as well as between all mantle samples and foot tissue, were run in PAST 3.15 as t-tests using Mann-Whitney as a significance measure (*p ≤ 0.05, **p ≤ 0.01).

Data availability
All data generated or analysed during this study are included in this published article or are available from the Dryad repository (https://doi.org/10.5061/dryad.gf1vhhmjs).  Table 2. Primer sequences and accession numbers.