X. tropicalis tmem16a is expressed in mucin-producing small secretory cells in tadpole skin
Functional studies of Tmem16a in Xenopus are limited to its role in the oocyte and the prevention of polyspermy [39, 40, 41]. RT-PCR expression analysis in adults shows expression in various tissues [42, 43], while bulk RNA sequencing in developing embryos shows expression at Nieuwkoop-Faber (NF) stages NF1, NF9 and NF24-NF42 [42]. We therefore sought tissue-specific expression during skin development.
Briggs [43] generated a developmental time series of single-cell transcriptomes in X. tropicalis embryos, covering early development from pre-gastrulation (NF8) to early tailbud (NF22), and disaggregated multiple cell lineages. Online interrogation of this dataset [44] permits analysis of gene expression in specific cell types. Visualised as an ‘all stages’ SPRING plot [45], the mucin-producing small secretory cell (SSC) lineage (Fig. 1a, upper) was originally identified by expression of met (Fig. 1a, lower left) [43]. tmem16a is expressed within this met-marked SSC lineage (Fig. 1a, lower right). No other clusters of expression in other cell lineages between NF8-NF22 were observed in this ‘all stages’ SPRING plot. To confirm specific expression in the SSC lineage and not other epidermal lineages, we analysed epidermal lineages in ‘tree view’. Over developmental time, the SSC lineage differentiates at NF14 from a pool of non-neural ectodermal cells that, from NF11 onwards, give rise to the major cell lineages of the tadpole skin. Analysis of tmem16a expression over differentiation of these epidermal lineages revealed expression only in SSCs in the developing epidermis, from NF18 and increasing to NF22 (Fig. 1b).
Using chromogenic mRNA in situ hybridisation, we tested tmem16a expression over a later time course of tadpole development (NF22, NF36 and NF43). We identified tmem16a expression in a punctate pattern in the tadpole skin at these later stages, matching our published distribution of SSCs (Fig. 1c-e) [46]. To confirm tmem16a expression is in SSCs at these later stages, we performed fluorescent in situ hybridization and staining with peanut agglutinin lectin (PNA), a lectin that binds to the Gal(β1–3)GalNAc moieties of mucin O-glycans and robustly marks the mucin-containing vesicles of SSCs [46]. Confocal imaging colocalised tmem16a with PNA in SSCs (Fig. 1f-h).
The above expression analyses show that X. tropicalis tmem16a is expressed exclusively in SSCs in the tadpole skin. In support, a single-cell transcriptome dataset analysing lineage patterns in the related species Xenopus laevis found that tmem16a is detectable only in the SSC lineage of the developing skin [47]. As SSCs produce the gel-forming mucin MucXS [38], these data suggest that X. tropicalis Tmem16a protein likely has a developmental and/or functional role in the mucociliary epidermal surface of the Xenopus tadpole.
The mucin-producing SSCs are equivalent to mammalian goblet cells
We have shown that tmem16a expression is apparently exclusive to the SSCs in the developing X. tropicalis tadpole skin. However, in addition to SSCs, the X. tropicalis epidermis has another mucin-producing cell type termed goblet cells (GCs) [48], implying equivalence with mammalian airway GCs that typically express TMEM16A and the mucin MUC5AC [49]. Thus, the expression pattern for X. tropicalis tmem16a generates a question regarding the homology of cell types in the Xenopus skin with those in the mammalian airway.
Returning to the single-cell developmental transcriptome dataset, we captured SSC and GC lineages at NF14 (the developmental time at which SSCs differentiate from non-neural ectoderm) and, using the platform tools with default parameters, plotted the two discrete gene expression clusters (Fig. 2a, left). We also captured SSC- and GC-specific lineages over developmental time (Fig. 2a, right). The SSC and GC lineages are marked by expression of met and itln1, respectively, with lineage-specific expression evident in NF14 clusters and from NF14 to NF22 (Fig. 2b). We and others have previously shown that the canonical goblet cell marker foxa1 is expressed in SSCs from NF14 and is necessary for the development of this cell population [46, 50]. Within the transcriptome dataset, foxa1 expression was near-exclusive to the SSC lineage at NF14 and from NF14 to NF22 (Fig. 2b).
We then sought expression of the transcription factor spdef, another mammalian GC marker [51]. We identified spdef expression in the NF14 SSC cluster and increasing levels of expression in SSCs from NF14 to NF22, but not in GCs at any stage analysed (Fig. 2b). In cultured human airway cells, FOXA1 has been shown to directly regulate transcription of SPDEF [52]. Here, we find that X. tropicalis SSCs express foxa1 earlier (at NF14) than spdef (from NF16), suggesting the same transcriptional dynamic in tadpole skin SSCs.
To confirm spdef expression in SSCs in the developing tadpole skin, we used chromogenic in situ hybridisation in NF22 embryos and found punctate expression in the skin, typical of SSC distribution, and strong expression of the spdef gene in the cement gland, a mucin-producing organ located in the anterior region of the developing head (Fig. 2c and inset). Fluorescent in situ hybridisation in combination with PNA staining at NF36 localised spdef expression to SSCs containing PNA-positive secretory vesicles (Fig. 2d-f).
These data show early differentiation of the two known secretory cells types in the X. tropicalis skin. Further, it is the SSCs and not the skin cells termed GCs that express markers more typical of mammalian airway GCs. Thus, tmem16a expression in the X. tropicalis skin is in the cell type (SSCs) likely to replicate the biology of mammalian GCs.
X. tropicalis Tmem16a protein localises to the plasma membrane of mucin-producing SSCs
TMEM16A has been localised to the apical plasma membrane in the mammalian airway epithelium [53] in GCs secreting MUC5AC [54], particularly during inflammation [55]. However, roles at the basolateral compartment of the plasma membrane in intestinal cells have also been described, as has the intracellular location of other paralogues of the TMEM16 family [56, 57]. We hypothesised that a functional role in mucin secretion/processing in the tadpole skin would most likely arise from apical expression of Tmem16a.
We investigated the cellular localisation of X. tropicalis Tmem16a protein in different planes of individual SSCs in the tadpole skin at NF36 by confocal microscopy, using an antibody against the human TMEM16A protein. The SSCs are filled with large, mucin-containing vesicles, visible by scanning electron microscopy (SEM) as bulges at the apical surface of the cell (Fig. 3a). At the SSC apical surface, Tmem16a appeared to surround the PNA-positive vesicles (Fig. 3b, arrowhead), also observable in the mid-plane of the cell (Fig. 3c). However, deep into the cell, expression was absent from the vesicle boundaries and appeared adjacent to the PNA-positive vesicles (Fig. 3d). Given the typical bulging of vesicles from the apical surface, we hypothesised that Tmem16a is, in fact, apical plasma membrane expression disrupted by this vesicle bulging, and this is supported by Tmem16a localisation on orthogonal (side) view (Fig. 3e).
To confirm plasma membrane localisation, we overexpressed membrane-localising EGFP (mEGFP) mRNA, targeting the ventral blastomeres fated to develop into skin to optimise signal. Tmem16a colocalised with mEGFP at the apical plasma membrane of PNA-positive SSCs (Fig. 3f-i), but not with mEGFP at the plasma membrane of other epidermal cell types (for example, Fig. 3i, arrowhead). Neither Tmem16a nor mEGFP signal was evident when imaging above the apical surface of the skin, while PNA-positive vesicles were evident in this plane (Fig. 3j-m).
We conclude that Tmem16a is found in the SSCs in the apical plasma membranes, and not the boundaries of mucin-containign vesicles. Thus, cellular localisation of Tmem16a is appropriate for a functional role in mucin secretion/processing.
X. tropicalis Tmem16a has calcium-activated, voltage-dependent chloride channel activity
Previous studies have shown that X. laevis Tmem16a is a calcium-activated chloride channel that is voltage-dependent [26] and sensitive to inhibitors [41]. However, the pharmacology of X. tropicalis Tmem16a has not been characterised, and these data are relevant to understand the potential of the tadpole skin as a model for TMEM16A in human health.
We cloned and expressed full-length X. tropicalis tmem16a in HEK293 cells. At 48 hours post-transfection, channel conductance and pharmacological characteristics were examined by whole-cell voltage clamp. Tmem16a channel activity was evoked using membrane depolarisation in the presence of intracellular calcium, and resultant whole-cell currents measured using chloride-selective buffers. A 1 second-step depolarisation from − 70 to + 70 mV in the presence of 338 nM free [Ca2+]i evoked a large current (2.15 ± 0.51 nA, n = 18), which was slow to activate (tauact 146 ± 34 ms, n = 18) and deactivate (taudeact 90 ± 21 ms, n = 18), and was completely inhibited by the presence of 10 µM Ani9 (Fig. 4a) [58]. These activation kinetics and Ani9 sensitivity are key similarities that X. tropicalis Tmem16a shares with human TMEM16A, which distinguish both from the closest human homologue TMEM16B [59].
Varying the level of free [Ca2+]i whilst using a fixed depolarisation voltage showed that current conductance was dependent on intracellular calcium (EC50 = 227\(\pm\)43 nM), was absent when [Ca2+]i was 0 nM and was inhibited at all calcium concentrations by 10 µM Ani9 (Fig. 4b). Conversely, varying the voltage of the depolarising step between − 90 and + 90 mV whilst maintaining intracellular free [Ca2+]i at 338 nM showed activated current to be strongly outwardly-rectifying, with the reversal potential coinciding with the calculated chloride equilibrium potential of -20 mV (Fig. 4c). The calcium and voltage-dependency of this chloride-mediated current match the characteristics described for human TMEM16A and differ significantly from any background chloride conductance observed in sham HEK-293 transfection (Supplementary Figure S1).
X. tropicalis Tmem16a has a comparable pharmacological profile to human TMEM16A
We have shown that, like human TMEM16A, X. tropicalis Tmem16a currents were sensitive to inhibition by Ani9. We extended our analysis of sensitivity to Ani9. Under conditions of maximal conductance (338 nM free [Ca2+]i combined with steady-state depolarisation to + 70 mV), the stepwise extracellular application of increasing concentrations of Ani9 caused a concentration-dependent inhibition of X. tropicalis Tmem16a current (Fig. 4d). The response of X. tropicalis Tmem16a to Ani9 was identical to that of human TMEM16A in terms of potency (Table 1), with current being blocked over a range of voltages (Fig. 4e). Ani9 inhibited both inward and outward chloride flux, with IC50 values at + 70 mV of 0.098 ± 0.009 µM and at -70 mV of 0.066 ± 0.012 µM (n = 13; Fig. 4f, normalised for comparison of large currents at + 70 mV vs very small currents at -70 mV).
Table 1
Pharmacological profile comparison of X. tropicalis Tmem16a with published data from human TMEM16A (abc and acd isoforms).
Compound | IC50 (n) X. tropicalis Tmem16a | Published human TMEM16A whole-cell patch clamp comparisons |
Ani9 | 0.098 ± 0.009 µM (13) | 0.066 µM, human TMEM16Aacd [61] 52% inhibition at 0.05 µM, human TMEM16Aabc [58] 0.068 µM, human TMEM16Aabc [64] |
CaCCinhA01 | 12.91 ± 1.96 µM (9) | 7.84 µM, human TMEM16A [62] 7.57 µM, human TMEM16Aacd [63] 1.7 µM, human TMEM16Aacd [60] |
Benzbromarone | 3.30 ± 1.45 µM (10) | 3.05 µM, human TMEM16Aabc [64] 4.09 µM human TMEM16Aacd [63] 2.35 µM, human TMEM16Aacd [61] |
Niflumic acid | 85.12 ± 19.06 µM (4) | 8.34 µM, human TMEM16Aacd [61] 12.1 µM, human TMEM16Aacd [60] 8.54 µM human TMEM16Aacd [63] |
CFTR Inh-172 | > 30 µM (5) | |
Idebenone | > 30 µM (4) | 54% and 90% inhibition by 10 µM and 30 µM respectively, human TMEM16A [61] |
MONNA | 11.89 ± 3.45 µM (6) | 13.60 µM human TMEM16Aacd [63] |
T16inh-A01 | > 30 µM (6) | > 30 µM human TMEM16Aacd [63] 1.51 µM, human TMEM16Aacd [60] |
9-AC | 145.9 ± 65.8 µM (5) | 57.7 µM, human TMEM16Aacd [60] |
Under the same conditions of maximal conductance, we defined concentration-inhibition relationships for a selection of chloride channel inhibitors known to inhibit human TMEM16A [60, 61, 62, 63, 64]. IC50 values for X. tropicalis Tmem16a are given in Table 1, alongside literature values for human TMEM16A from similar whole-cell patch-clamp studies. Side-by-side comparison of these values shows the compounds are similarly potent (within 3-fold of human TMEM16A values) except for niflumic acid, which was found to be markedly less potent at inhibiting X. tropicalis Tmem16a. There is some dispute in the literature of the effectiveness of T16inh-A01 at blocking human TMEM16A, which in our assay did not directly inhibit X. tropicalis Tmem16a channel function.
With these electrophysiology studies, we show that X. tropicalis Tmem16a exhibits the same functional and pharmacological hallmarks as human TMEM16A in terms of its activation by intracellular calcium, voltage-sensitivity, kinetics and relative sensitivities to commonly-used inhibitor compounds.
Loss of Tmem16a alters mucin secretion and its macromolecular properties
We have shown that X. tropicalis Tmem16a is present in the SSCs of tadpole skin, and that it functions as a voltage-sensitive, calcium-activated chloride channel in a comparable way to its human counterpart. We hypothesised that X. tropicalis Tmem16a, presumably via regulation of ion balance, functions in mucin secretion and/or affects its expansion and re-modelling post-secretion from SSCs. To test this, we used a morpholino oligonucleotide (MO) knockdown strategy to deplete Tmem16a from the tadpole skin.
We targeted the donor splice site of exon 2 in tmem16a pre-mRNA (Fig. 5a), predicting that full or partial intron retention would result in a premature termination codon after 21 amino acids (Ensembl ENSXETG00000001994) [65]. We tested for disrupted pre-mRNA splicing by RT-PCR across the tmem16a target site in cDNA from NF25 embryos (Fig. 5b). In MO control-injected (MOC) embryos, normal splicing of tmem16a pre-mRNA was evident by a single amplicon of the predicted size (Fig. 5b, arrowhead). After injection of 15 ng tmem16a splice MO into the fertilised egg, we observed a large decrease in signal intensity for the band corresponding to normal splicing of tmem16a pre-mRNA and concomitant appearance of a longer amplicon (abnormal splicing; Fig. 5b, asterisk) that likely corresponds to intronic retention and inclusion of a premature termination codon shortly downstream of exon 2. There was no difference in RT-PCR amplification of the housekeeping mRNA ornithine decarboxylase (odc), indicating equivalence of samples.
Tmem16a-depleted morphant embryos developed normally until tailbud stages. At NF38, morphant tadpoles displayed a bent anterior-posterior axis, delayed head development, and small edemas around the developing heart (Fig. 5c). However, the morphant tadpoles had a superficially healthy epidermis, and we predicted the observed developmental defects would not impede further investigation. To simultaneously confirm depletion of Tmem16 protein and examine the development of the SSCs in the skin, we performed wholemount immunofluorescence for Tmem16a on MOC-injected and morphant tadpoles at NF38 (Fig. 5d-e). Tmem16a was evident in the plasma membrane of PNA-stained SSCs in MOC-injected tadpoles (Fig. 5d) and this signal was completely lost in morphant tadpoles (Fig. 5e). Orthogonal views (Fig. 5d-e, insets) demonstrated that Tmem16a was not detected at any cellular plane in MO-injected embryos, and that its apparent absence is not an artefact of imaging plane. In these morphant tadpoles, PNA staining of SSCs indicated no detectable impact on the presence or location of mucin vesicles at the apical membrane in these cells. These data show that loss of Tmem16a in the developing tadpole does not impede the development of SSCs.
We next determined whether Tmem16a depletion affected mucin secretion and/or its macromolecular properties. First, after ten minute exposure to the secretagogue ionomycin, we detected (via slot blotting and immunostaining with an anti-MucXS antibody) an increase in MucXS in the tadpole media, to levels greater than secreted by MOC-injected embryos (Fig. 6a). This suggests that Tmem16a in the epidermis may be required to restrict mucin secretion. To assess the impact of Tmem16a depletion on the macromolecular properties of secreted MucXS, media from batches of ionomycin-exposed tadpoles was subjected to rate zonal centrifugation on sucrose gradients (Fig. 6b). This facilitates mucin separation by size/shape, with more compacted mucins moving further through the gradient than expanded mucins. In MOC-injected tadpoles, MucXS typically appears as a discrete density peak. However, in Tmem16a morphants, MucXS signal was detected across the entirety of the density gradient, with variability between biological replicates. In two of three replicates, MucXS signal was shifted towards the lower density fractions (Fig. 6b, MO repeat 1 and repeat 3). These data indicate that secreted MucXS is altered when X. tropicalis Tmem16a is depleted, and tends to be smaller and/or less compact than MucXS secreted by control tadpoles, although the potential for more tightly-compacted or larger, aggregated MucXS is evident in the shift towards higher-density fractions in one replicate (Fig. 6b, MO repeat 2).
We then examined the size and morphology of the secreted mucins from Tmem16a morphants by transmission electron microscopy (Fig. 6d-i). Mucins secreted from MOC-treated tadpoles are routinely detected in compact (Fig. 6d, within the oval) or semi-expanded (Fig. 6e, arrowheads) forms, consistent with secreted human mucin morphologies previously described [38, 66]. In contrast, mucins secreted by Tmem16a morphants were very rarely detected in compact form and only occasionally detected in semi-expanded form (Fig. 6f). We identified a few examples of compact mucins that were aggregated with dense, amorphous structures (Fig. 6g, within the oval), similar to those described in human cystic fibrosis saliva [67]. Although in a compact form, the mucin strands here are visibly thinner than those observed in compact mucins from MOC-injected tadpoles. Overall, the majority of mucins secreted from Tmem16a morphants appeared as expanded, linear molecules (Fig. 6h, asterisk and 6i). Together, these data suggest that Tmem16a is required to limit secretion of mucins and for post-secretory remodelling of mucin molecules.