Opsin genes of select treeshrews resolve ancestral character states within Scandentia.

Treeshrews are small, squirrel-like mammals in the order Scandentia, which is nested together with Primates and Dermoptera in the superordinal group Euarchonta. They are often described as living fossils, and researchers have long turned to treeshrews as a model or ecological analogue for ancestral primates. A comparative study of colour vision-encoding genes within Scandentia found a derived amino acid substitution in the long-wavelength sensitive opsin gene (OPN1LW) of the Bornean smooth-tailed treeshrew (Dendrogale melanura). The opsin, by inference, is red-shifted by ca 6 nm with an inferred peak sensitivity of 561 nm. It is tempting to view this trait as a novel visual adaptation; however, the genetic and functional diversity of visual pigments in treeshrews is unresolved outside of Borneo. Here, we report gene sequences from the northern smooth-tailed treeshrew (Dendrogale murina) and the Mindanao treeshrew (Tupaia everetti, the senior synonym of Urogale everetti). We found that the opsin genes are under purifying selection and that D. murina shares the same substitution as its congener, a result that distinguishes Dendrogale from other treeshrews, including T. everetti. We discuss the implications of opsin functional variation in light of limited knowledge about the visual ecology of smooth-tailed treeshrews.

Introduction

Treeshrews are small, squirrel-like mammals with a broad distribution from India and southern China through most of Southeast Asia (figure 1a). They comprise a single order, Scandentia, in which two families are recognized: Ptilocercidae, containing a nocturnal species, Ptilocercus lowii, and Tupaiidae, containing 22 diurnal species [4] in four traditional genera (Dendrogale, Anathana, Urogale and Tupaia; figure 1b). Scandentia is nested in the superordinal group Euarchonta together with Dermoptera and Primates (figure 1b), an affinity that invites the use of treeshrews as a model system for studying a wide range of human disorders, including myopia [6]. Treeshrews are also viewed as ‘living fossils’ [7] and therefore practical models or ecological analogues of ancestral and stem primates [8-14]. Accordingly, Melin et al. [15] sequenced the opsin genes of Bornean treeshrews to contextualize the origins of high-acuity colour vision in primates. In Dendrogale melanura—the earliest branching tupaiid in their sample—they found an amino acid substitution (A180S) that translates into a relatively red-shifted long-wavelength sensitive (LWS) opsin. Given that opsins are sensitive to ecological selective pressures [16], it is tempting to interpret this genotype as a novel visual adaptation; however, the full extent of its derivation in Tupaiidae is uncertain. To resolve this uncertainty, data are needed from treeshrew species outside Borneo.

Figure 1.

(a) Approximate distribution of treeshrew genera in the family Tupaiidae. Dendrogale is broadly sympatric with Tupaia in northern Borneo and Indochina. Huxley's Line generally corresponds to the edge of the Asian continental shelf and separates oceanic islands from landbridge islands in the Philippines. Redrawn from Roberts et al. [1]. (b) Phyletic relationships of genera historically recognized in the order Scandentia, which is sister to Dermoptera and Primates (Primatomorpha) in the superordinal group Euarchonta [2,3]. *Recently, Urogale has been subsumed into the genus Tupaia [2,4,5]. Original artwork by Priscilla Barrett, reproduced with permission.

Here we fill two voids by focusing on the only extant congener of D. melanura, the northern smooth-tailed treeshrew (Dendrogale murina) and the Mindanao treeshrew, a species described as Tupaia everetti in 1892 and elevated to a monotypic genus (Urogale) in 1905 on the basis of distinguishing morphological traits [17]. This latter nomen prevailed for a century until mounting molecular evidence favoured the subsumption of Urogale into Tupaia [1,2,18]. Tupaia everetti is therefore the senior synonym of U. everetti, and re-recognition of T. everetti is spreading in the literature [4,5]. Setting this taxonomic reversal aside, T. everetti holds interest because it is the earliest branch in crown Tupaia, splitting prior to the diversification of Tupaia across Southeast Asia (figure 1b). It is therefore crucial for pinpointing the origin of differential opsin sensitivities in Tupaiidae.

Accordingly, we examined the sequences of exons known to determine the spectral tuning of the LWS opsin gene (OPN1LW) as well as the short-wavelength sensitive opsin gene (SWS1, OPN1SW), which has been subject to different selective pressures among taxa within Scandentia and more generally across Euarchonta [15]. We then compared the sequences of D. murina and T. everetti to those of other treeshrews and primates in order to reconstruct the ancestral character states of opsin genes at two nodes within Tupaiidae.

Material and methods

Study species and sample collection

Dendrogale murina is one of two recognized species in the genus (figure 2a). It has a wide distribution throughout Vietnam, Thailand and Cambodia [19] in contrast to its congener (D. melanura), which is endemic to montane Borneo [20]. It is also more flexible ecologically, with records ranging from lowland plains to 1500 m across a wide range of habitat conditions, from evergreen forest (at varying stages of degradation), to mixed deciduous forest, to secondary bamboo fields lacking any dicotyledonous canopy, to streamside tangles in rocky savannah [19]. It not only uses the under- and mid-storeys, but also enters the canopy; recent observations come primarily from understorey tangles, especially of bamboo, almost exclusively 30–300 cm above ground level [19]. In Thailand, it has been observed on the branches of fruiting trees [19].

Figure 2.

Illustrations of study species: (a) the Northern smooth-tailed treeshrew, Dendrogale murina and (b) the Mindanao treeshrew, Tupaia everetti. Original artwork by Priscilla Barrett, reproduced with permission.

Tupaia everetti is the sole scandentian to inhabit the Mindanao Faunal Region, Philippines (figure 2b; [17,21]). It has a widespread distribution on Mindanao, preferring montane and mossy forests between 750 and 2250 m a.s.l., but it is also found at lower elevations on neighbouring islands [5]. Limited observations and stomach content analysis indicate a diurnal terrestrial niche and a mixed diet of ground-dwelling insects, other arthropods, fruits and other plant material [22,23]. It is readily differentiated from other tupaiids by its even-haired round tail, elongated snout and large, canine-like second incisors [17]; it is also the largest treeshrew, with captive adults ranging from 270 to 410 g (mean: 314 g; n = 7 [24]) and 190 to 342 g (mean: 276 ± 8.0 g; n = 10 males; mean: 252 ± 6.3 g; n = 18 females [25]). Wild-caught specimens are equally large (range: 235–315 g; n = 2 males [26]), but Heaney reported smaller values (mean: 190 g; n = 3; cf. Sargis [27]). Data are limited, but specimens of T. everetti from Dinagat and Siargao are smaller than those from Mindanao [28], raising the possibility of cryptic speciation on these islands [29].

We extracted DNA from the muscle tissues of museum specimens. The specimen of D. murina is accessioned in the University of Alaska Museum (catalogue no. UAM 103000). It was a wild-caught male from the Seima Biodiversity Conservation Area, Mondulkiri, Cambodia. Two specimens of T. everetti are housed at the Vertebrate Museum, Institute of Biology, University of the Philippines Diliman, Quezon City (catalogue nos. PNM7496 [adult male] and PNM7497 [juvenile male]). The animals were wild-caught in the Mt Apo Natural Park, Barangay Agco, Kidapawan City, Cotabato, Mindanao. Muscle tissues were biopsied in the field and stored in 99% ethanol.

DNA extraction, amplification and sequencing

Genomic DNA was extracted from muscle tissues using a DNeasy Blood and Tissue Kit (Qiagen) following the manufacturer's instructions. Amino acids at 10 sites on exon 1 of the OPN1SW gene determine the spectral tuning of the opsin, with three sites—86, 90 —primarily governing sensitivity in the violet-blue (400–450 nm) region of the light spectrum [30-32]. The λmax of the opsin encoded by OPN1LW is determined by five amino acid sites spanning exons 3–5 [33]. Three of these sites are variable in primates (exon 3: 180; exon 5: 277, 285; [34,35]); among treeshrews, genotypes at sites 180, 277 and 285 have recently been reported, with variation between taxa noted at site 180 [15]. We obtained partial opsin sequences for D. murina and T. everetti by amplifying exons 1, 2–3, 4 and 5 of the OPN1SW and exons 3 and 5 of the OPN1LW. Although the tuning sites of OPN1SW are located on exon 1, variation in gene functionality occurs within Scandentia, and we sequenced additional exons to rule out indels or premature stop codons leading to gene pseudogenization [15].

Polymerase chain reactions (PCRs) were conducted in 25 µl volumes containing 1× KAPA HiFi Readymix (Kapa Biosystems Inc., USA), 1.0 mM each of the forward and reverse primers (electronic supplementary material, table S1) and 200 ng template DNA. Molecular grade water was used as a negative control in all reactions. Thermocycler parameters were as follows: 3 min initial denaturation at 98°C; 35 cycles of 10 s denaturation at 98°C, 30 s annealing at either 58°C (for OPN1SW exons 1 and 4) or 60°C (for OPN1SW exons 2–3 and 5 and OPN1LW exons 3 and 5) and 30 s extension at 72°C, followed by a 5 min final extension at 72°C.

Amplification of target sequences was confirmed on a 1.5% agarose gel in 1 X Tris Borate EDTA (TBE) buffer. PCR products were purified using EXOsap-IT (Affymetrix, USA) following the manufacturer's protocol. If non-target sequences were also amplified, target amplicons were excised and purified using the Purelink Gel Extraction kit (Life Technologies Inc.). Purified PCR products were directly Sanger sequenced on sense and antisense strands at the University of Calgary Core DNA Sequencing Facility (Faculty of Medicine, University of Calgary, Alberta, Canada) using an Applied Biosystems 3730×l 96 capillary DNA Analyzer.

Data analysis

We assembled and edited OPN1SW and OPN1LW sequences in Geneious v. 10.0.3 (Biomatters) using the Clustal W function with manual refinement. We aligned them to opsin sequences from Homo sapiens and published treeshrew species: Ptilocercus lowii, Dendrogale melanura, and six species of Tupaia (T. minor, T. belangeri, T. tana, T. montana, T. longipes and T. gracilis; electronic supplementary material, table S2). We translated the coding regions into amino acid sequences, which we used to infer colour vision phenotype. Site numbers in our alignments correspond to the position of the amino acid in the human SWS1 and M/LWS pigments. We used the Protein Variation Effect Analyzer (http://provean.jcvi.org/; PROVEAN) tool to predict the functional implications of other nonsynonymous mutations [36,37]. PROVEAN predicts deleterious polymorphisms, but the functional effect of nonsynonymous mutations that change the amino acid will be predicted neutral when the properties of the amino acid do not change drastically (e.g. due to hydrophobicity). We tested for evidence of purifying or positive selection in OPN1SW and OPN1LW using codeml free-ratio branch models implemented in phylogenetic analysis by maximum likelihood (PAML) [38]. Outgroups for PAML analyses were two primates with functional opsin genes, Alouatta palliata (OPN1SW: AH005790.1; OPN1LW: AB809459.1) and Tarsius bancanus (OPN1SW: AB111463.1 and OPN1MW: AB675927.1) (electronic supplementary material, table S2). We also used codeml site models to test whether any sites, including the tuning sites are under positive selection. We did not include P. lowii in the PAML analysis of the OPN1SW gene due to low sequence coverage. This nocturnal species has a predicted OPN1SW pseudogene of ancient origin and relaxed selection pressures have previously been reported [15].

Results

We successfully sequenced partial exons 1–5 of OPN1SW, and exons 3 and 5 of OPN1LW for the D. murina and T. everetti specimens (electronic supplementary material, figure S1). Among scandentian sequences, the mean amino acid divergence (mean amino acid difference per sequence/total amino acids analysed, s.e. over 1000 bootstrap replicates) for OPN1SW was 1.02% divergence (3.524/347, s.e. = 0.822). The amino acid divergence for OPN1LW partial sequences was 0.96% divergence (1.311/136, s.e. = 0.567).

OPN1SW evolution

Relative to T. everetti and other Tupaia species, D. murina has four derived nonsynonymous variants causing amino acid differences: G111A and V119C in exon 1, T159A in exon 2, and S224T in exon 3 (electronic supplementary material, figure S1). The first two mutations are shared with D. melanura, but missing sequence data for D. melanura prevent assessing whether the last two are shared or unique to D. murina. At site 322, the OPN1SW amino acid sequence of D. murina has retained the ancestral state Cys322, relative to its congener, which has a derived nonsynonymous variant Phe322. All these nonsynonymous mutations are predicted to be neutral using PROVEAN, scores = −1.484 (site 111), 2.346 (site 119), −0.341 (site 159), 1.773 (site 224) and −1.903 (site 322); cut-off = −2.5). Tupaia everetti differs from Tupaia species in possessing L rather than Q at site 28. This may be the ancestral condition for Tupaiidae and possibly for Scandentia, as it is shared with D. melanura. Data at these sites from P. lowii would help to further resolve this but are presently unavailable. There are two derived substitutions unique to T. everetti: exon 1, A120T, exon 5, R323K (electronic supplementary material, figure S1). All mutations were predicted to be neutral (PROVEAN scores = 4.742 (site 28), 0.977 (site 120) and −0.299 (site 323); cut-off = −2.5). Interestingly, a derived mutation in T. longipes, F115V, may be deleterious (PROVEAN score, −5.951). Overall, based on broad conservation of amino acids, including the main spectral tuning sites Tyr86, Ser90 and Val93 (exon 1), the λmax of the SWS1 opsin protein of both D. murina and T. everetti is predicted to be around 444 nm [15,30,32] (figure 3).

Figure 3.

Phyletic relationships and divergence dates for primates and treeshrews were based on TimeTree [39] (accessed February 2019) and published estimates [2]. Branch colours correspond with the function and spectral tuning of opsin photopigments. Pseudogenization events are marked with a diagonally bisected circle. The inferred long-wave shift from orange to red sensitivity in the LWS opsin of the genus Dendrogale is marked with a star, along with the amino acid substitution responsible. The timing of this shift in sensitivity is unknown. Species sequenced in this study are indicated with an asterisk.

OPN1LW evolution

Two derived nonsynonymous mutations (A180S and V182I, exon 3) are present in the OPN1LW gene sequence of D. murina compared to other treeshrews, indicating a shared origin in the last common ancestor of extant Dendrogale species (electronic supplementary material, figure S1). Nonsynonymous substitutions at site 180 and 182 were predicted to be neutral (PROVEAN respective scores = −0.302 and 0.133; cut-off = −2.5). However, site 180 is a known opsin tuning site (figure 3). We did not detect any derived nonsynonymous mutations in the OPN1LW sequence of T. everetti. Based on the three-site composition of SYT at sites 180, 277 and 285, D. melanura and D. murina are predicted to be long-wavelength shifted, and have a λmax value of ca 561 nm, as opposed to shorter λmax values for other treeshrews. We predict that the T. everetti LWS opsin protein (three-site composition of AYT) has a similar λmax to Tupaia and Ptilocercus, ca 555 nm.

Evaluation of selective pressures

The codeml branch models for OPN1SW indicated purifying selection acting on all species (ω < 0.312), but supported a model of branch-specific differences in the strength of selection (p = 0.045, Likelihood Ratio (LR) = 29.25), showing stronger purifying selection acting on D. murina (ω = 0.0001) than on T. everetti (ω = 0.312). The codeml branch model did not find species-specific differences in the strength of selection acting on OPN1LW, indicating that the gene is under purifying selection (ω = 0.165) in all examined species, including D. murina and T. everetti. We did not find evidence of positively selected sites in our codeml site models for either OPN1SW (p = 1, LR = 0) or OPN1LW (p = 0.228, LR = 2.957).

Discussion

We found that the short-wavelength sensitive opsin gene (OPN1SW) is intact in both D. murina and T. everetti. The amino acid composition of 10 spectral tuning sites, including the three most influential (Tyr86, Ser90, Val93) predicts a λmax of 444 nm, an inferred phenotype that unites D. murina and T. everetti together with every tupaiid examined to date [15]. In addition, we found that the long-wavelength sensitive opsin gene (OPN1LW) is functionally variable in the two study species. Despite PROVEAN's neutral prediction, in vitro site-directed mutagenesis of the LWS pigment has demonstrated that the single site mutation A180S significantly shifts the λmax of the pigment [33]. Notably, D. murina shares the A180S substitution with its congener D. melanura [15]. The LWS opsins of Dendrogale are therefore predicted to have a red-shifted λmax of 561 nm, whereas those of all other scandentians, including T. everetti (present results), are predicted to have a λmax of 555 nm.

Our results suggest that the colour vision of T. everetti is practically identical to that of other Tupaia species despite its relatively large body size and restricted distribution. This result is consistent with recent shared ancestry [2], and it suggests comparable visual ecologies. Indeed, observations of habitat and resource use by these species speak to similar preferences for insects and fruit in the forest understorey [19,22,40,41]. It is possible that the 444/555 nm opsin combination is a ‘multi-purpose’ dichromacy with functionality across a wide range of understorey light conditions. Given the strict conservation of opsin genes in Tupaia, including the spectral tuning sites, it is rather surprising that Dendrogale is predicted to express a red-shifted LWS opsin, which suggests unique selective pressures on the colour vision of this lineage. The implications of this 6 nm red-shift are discussed below.

The split between Ptilocercidae and Tupaiidae occurred ca 60 Ma, preceding the emergence of Dendrogale, the earliest branching tupaiid, ca 35 Ma; divergence of the two extant congeners, D. melanura and D. murina, is estimated at ca 21 Ma [2]. Our results raise the possibility that the common ancestor of Dendrogale occupied a distinct visual niche that favoured the fixation of a red-shifted opsin, once this novel mutation arose. One possibility is that Dendrogale is adapted to relatively open habitats with greater exposure to unfiltered daylight. Such light is relatively enriched in longer wavelengths compared to foliage-filtered downwelling light [42,43]. In consequence, redder objects should appear brighter in open habitats illuminated with redder light; however, the nature of these putative objects (foods, russet-coloured predators) is uncertain.

Some support for this ‘open habitat’ hypothesis is evident in the natural history of D. murina—it is found in degraded evergreen forest, mixed deciduous forest, bamboo fields without a dicotyledonous canopy, as well as streamside tangles in rocky savannahs [19]. It is also evident in the differences between two sympatric montane species in Mount Kinabalu National Park, Sabah, Borneo [44]. Smooth-tailed treeshrews (D. melanura) are found around elevations of well above 900 m, the point at which another species of montane treeshrew, Tupaia montana, replaces its lowland congeners T. gracilis, T. longipes, T. minor and T. tana [40,44]. The elevational overlap of D. melanura and T. montana is telling, as the smaller D. melanura has a limited range, patchier distribution, and prefers relatively open areas around mossy boulders, whereas T. montana prefers greater forest cover and uses a larger range of montane habitats, as reflected by its higher abundance in trap records ([2,40]; K. Wells 2009–2010, personal observation). Importantly, our finding indicates deep antiquity for the red-shifted OPN1LW in Dendrogale. This interpretation argues against the possibility of recent character displacement; i.e. that the colour vision of D. melanura is the result of competition with T. montana and niche-divergence in diet, microhabitat or other aspects of visual ecology [15]. However, early opsin divergence between Dendrogale and other tupaiids may have facilitated current sympatry via different microhabitat preferences, i.e. habitat patches with less intensive canopy cover versus greener, denser canopy cover, a hypothesis that invites testing through behavioural observation and characterization of downwelling light. Overall, there is a need to better understand fine-scale habitat use by sympatric species at different times of the year and to characterize the light conditions in different microhabitats.

Finally, it is alternatively possible that the A180S substitution in Dendrogale did not impact the fitness of individuals, but spread through an ancestral population neutrally. If so, this may point to a bottleneck event in the history of this genus, as neutral mutations are unlikely to become fixed in large populations [45]. Future examination of the population genomics of extant Dendrogale species and estimations of NE and past population contractions and expansions may allow us to favour or rule out this hypothesis. Additionally, increased sampling of OPN1LW in these species will also allow us to rule out OPN1LW polymorphisms, which are common in neotropical primates, but unknown outside of the primate order [15,35,46].

Conclusion

We present new data on the opsin gene sequences of the northern smooth-tailed treeshrew, Dendrogale murina, and the Mindanao treeshrew, Tupaia everetti. The gene codings for both short-wavelength sensitive and long-wavelength sensitive opsins are under purifying selection in each species and presumed to be functional. Dendrogale murina shared a derived amino acid with its congener, D. melanura, at site 180 of the OPN1LW, which should cause a shift in the sensitivity towards reddish light. This may indicate the common ancestor of extant Dendrogale taxa occupied a relatively more open habitat—richer in light unfiltered by green foliage—and may contribute to present-day niche separation among sympatric diurnal treeshrews. We end by noting that lack of availability of DNA from Anathana precluded analysis of the opsins of the remaining genus in the order Scandentia, but in the future this would shed additional light on pressures shaping treeshrew visual ecology.