Among the numerous lineages of teleost fish that have independently transitioned from obligate water breathing to facultative air breathing, evolved properties of hemoglobin (Hb)–O2 transport may have been shaped by the prevalence and severity of aquatic hypoxia (which influences the extent to which fish are compelled to switch to aerial respiration) as well as the anatomical design of air-breathing structures and the cardiovascular system. Here, we examined the structure and function of Hbs in an amphibious, facultative air-breathing fish, the blue-spotted mudskipper (Boleophthalmus pectinirostris). We also characterized the genomic organization of the globin gene clusters of the species and we integrated phylogenetic and comparative genomic analyses to unravel the duplicative history of the genes that encode the subunits of structurally distinct mudskipper Hb isoforms (isoHbs). The B. pectinirostris isoHbs exhibit high intrinsic O2 affinities, similar to those of hypoxia-tolerant, water-breathing teleosts, and remarkably large Bohr effects. Genomic analysis of conserved synteny revealed that the genes that encode the α-type subunits of the two main adult isoHbs are members of paralogous gene clusters that represent products of the teleost-specific whole-genome duplication. Experiments revealed no appreciable difference in the oxygenation properties of co-expressed isoHbs in spite of extensive amino acid divergence between the alternative α-chain subunit isoforms. It therefore appears that the ability to switch between aquatic and aerial respiration does not necessarily require a division of labor between functionally distinct isoHbs with specialized oxygenation properties.
Facultative air-breathing fishes have always been of interest to comparative physiologists because of the versatility required to switch between aquatic and aerial respiration (Graham, 1997; Graham and Wegner, 2010; Bayley et al., 2019). Moreover, the physiological features that distinguish air-breathing fishes from their obligate water-breathing kin may provide clues about the types of phenotypic changes that facilitated the invasion of land by the shallow-water progenitors of modern tetrapods. In addition to evolutionary changes in branchial and cardiovascular function in fish that have evolved the capacity for aerial respiration, there has been much interest in associated changes in hemoglobin (Hb) function and respiratory gas transport (Damsgaard et al., 2014, 2015; Bayley et al., 2019).
Regardless of breathing mode, the physiologically optimal Hb–O2 affinity is dictated by the trade-off between O2 loading at the respiratory surfaces (gills, lungs or other air-breathing structures) and O2 unloading to respiring tissues (Brauner and Wang, 1997; Wang and Malte, 2011; Storz, 2019), and there are reasons to expect that this trade-off may be especially profound in facultative air-breathing fishes. Because air has a much higher O2 content than water, aerial breathing should generally place a lower premium on O2 uptake at the respiratory surfaces (as arterial O2 saturation is typically not a limiting factor in tissue O2 delivery). This has led some authors to suggest that the evolutionary transition from water to air breathing generally entailed a reduction in Hb–O2 affinity that permitted a concomitant increase in circulatory O2 delivery and aerobic metabolism (Johansen and Lenfant, 1972; Johansen et al., 1978; Powers et al., 1979). However, among amphibious fish that alternate between breathing modes, a high Hb–O2 affinity retains much utility under conditions of aquatic hypoxia because it improves O2 uptake at the gills, thereby enhancing blood O2 capacitance [the quantity of O2 unloaded to the tissues for a given difference in arterial and venous partial pressures of O2 (PO2)]. Indeed, among water-breathing and facultative air-breathing fish alike, increased Hb–O2 affinities are generally associated with adaptation to aquatic hypoxia (Powers, 1980; Jensen, 2004; Mandic et al., 2009; Wells, 2009; Fago, 2017; Harter and Brauner, 2017).
In addition to considerations related to differences in O2 capacitance between air and water and the challenges posed by aquatic hypoxia, the optimal Hb–O2 affinity in air-breathing fish may also relate to the diffusive conductance of air-breathing structures and design features of the cardiovascular system. Among most air-breathing teleosts, extrabranchial gas exchange occurs across the vascularized epithelia of the buccopharyngeal cavity, esophagus, intestine, stomach or swim bladder (Graham, 1997; Graham and Wegner, 2010). Such air-breathing structures generally have much lower diffusion capacities and lower gas exchange efficiencies than normal gills, so an elevated Hb–O2 affinity can promote O2 uptake by maintaining a steep PO2 gradient across the respiratory epithelium (Hlastala and Berger, 2001; Damsgaard et al., 2014). Finally, aerial breathing generally increases the PO2 of arterial blood in the ventral aorta that flows towards the gills and – in the absence of a non-respiratory branchial shunt – O2-rich arterial blood could desaturate during branchial passage (Randall et al., 1981; Olson, 1994; Ishimatsu, 2012; Bayley et al., 2019). Thus, a high Hb–O2 affinity may also be beneficial in facultative air breathers because it helps prevent transbranchial O2 loss, thereby ensuring adequate O2 delivery to metabolizing tissues. For this reason, Damsgaard et al. (2014) suggested that facultative air-breathing fish species that do not possess transbranchial shunts could be expected to have evolved higher Hb–O2 affinities than obligate air breathers with reduced gills.
Most teleost fish co-express multiple, structurally distinct Hb isoforms (isoHbs) in adult red blood cells, which are conventionally classified as ‘anodic’ or ‘cathodic’ on the basis of electrophoretic mobility (Weber, 1990, 2000). Anodic isoHbs tend to have relatively low O2 affinities and a large Bohr effect (reduced Hb–O2 affinity at low pH), whereas cathodic isoHbs tend to have relatively high O2 affinities, a heightened sensitivity to the affinity-reducing effects of organic phosphates (e.g. ATP and GTP), and a negligible Bohr effect in the presence of organic phosphates (Weber and Jensen, 1988; Weber, 1990, 2000; Jensen et al., 1998; Weber et al., 2000; Wells, 2009). According to the classification scheme of Weber (1990), ‘class I’ species such as plaice and carp express multiple anodal isoHbs with very similar O2-binding properties, whereas ‘class II’ species such as anguillid eels, salmonids and catfish additionally express one or more cathodal isoHbs that broaden the spectrum of O2 affinities and allosteric regulatory capacities. The isoHb multiplicity in class II species may permit a physiological division of labor whereby pH-insensitive cathodal isoHbs provide a reserve capacity for blood–O2 transport under conditions of environmental hypoxia or metabolic acidosis (Weber, 1990, 2000). In some teleost species, evidence suggests that regulatory switches in isoHb expression may play a role in the acclimatization response to environmental hypoxia (Rutjes et al., 2007), and in amphibious fish such as mudskippers it seems plausible that a physiological division of labor between isoHbs with different oxygenation properties could contribute to the versatility required to reversibly switch between aquatic and aerial respiration.
Here, we examine the structure and function of Hbs in the blue-spotted mudskipper (Boleophthalmus pectinirostris), an amphibious, facultative air-breathing fish that is distributed along the coastlines of China, Taiwan, the Korean peninsula and Japan in the northwest Pacific. This species routinely switches between breathing modes when shuttling between tidepools and mudflats in the intertidal zone (Martin and Bridges, 1999) and it uses an extensive respiratory surface for aerial breathing, including the buccal, pharyngeal, branchial and opercular cavities, in addition to the gills and integument (Graham, 1997). We also report sequence data for the full complement of α- and β-type globin genes from B. pectinirostris, we characterize the isoHb composition of adult red cells, and we integrate comparative genomic and phylogenetic analyses to shed light on the evolutionary origins of isoHb multiplicity in mudskippers.
MATERIALS AND METHODS
Sequence data and phylogenetic analyses
We annotated the full complement of α- and β-type globin genes in the genome assembly of the blue-spotted mudskipper, Boleophthalmus pectinirostris (Linnaeus 1758) (GenBank accession number: JACK00000000.1). For analyses of conserved synteny and phylogenetic relationships, we used genomic sequence data from a diverse set of bony fish and cartilaginous fish, including globin gene sequences that we annotated previously (Hoffmann et al., 2012; Opazo et al., 2013, 2015).
Protein coding sequences were translated and aligned using L-INS-i strategies in Mafft v7.3 (Katoh and Standley, 2013) and were then reverse-translated so that coding sequences were codon-aligned. We annotated several genes with truncated coding sequences [Hbb3-MN in platyfish (Xiphophorus maculatus), Hbb2-MN and Hbb6-MN in Nile tilapia (Oreochromis niloticus) and Hba1 in spotted gar (Lepisosteus oculatus)], but we did not include such sequences in the phylogenetic analyses. We reconstructed separate phylogenies for the α- and β-globin genes using maximum likelihood (ML). After partitioning the alignment into codon positions, we performed the ML analyses in RAxML v 8.1.3 (Stamatakis, 2006) using the GTRGAMMA substitution model. Support for each node in the best tree was based on 1000 bootstrap replicates.
Blood sampling and characterization of red cell isoHb composition
Prior to blood sampling, the fish were maintained in plastic aquaria with artificial seawater (15% salinity) at 27°C for 24 h. Two individual specimens of B. pectinirostris were anesthetized with 0.01% MS222 (Sigma-Aldrich, St Louis, MO, USA) and blood was collected via caudal venipuncture with a heparinized syringe. The animals were handled in accordance with a protocol approved by the Animal Ethics Committee and the Institutional Review Board on Bioethics and Biosafety of BGI (certification number FT15029).
Each blood sample was incubated for 5 min in a 5-fold volume of ice-cold buffer (10 mmol l−1 Hepes, 0.5 mmol l−1 EDTA, pH 7.75). Hemolysates were centrifuged to remove cell debris (30 min at 15,000 g, 4°C) and KCl was then added to the clear supernatant to a final concentration of 0.2 mol l−1 (Jendroszek et al., 2018). The sample was then passed through a 5 ml PD-10 (GE Healthcare) desalting column equilibrated with 10 mmol l−1 Hepes, 0.5 mmol l−1 EDTA, pH 7.7. This desalting step removes endogenous organic phosphates, which are mainly ATP and GTP in fish red cells.
Separation and purification of isoHbs
We purified individual isoHbs by means of anion-exchange fast-protein liquid chromatography (FPLC) using an Äkta Pure system (GE Healthcare). Stripped hemolysate was loaded on the prepacked HiTrap Q-HP column (GE Healthcare) and equilibrated with 10 mmol l−1 Hepes buffer (0.5 mmol l−1 EDTA, pH 7.66), and separation of individual isoHbs was achieved using a linear gradient of 0–80% of 0.2 mol l−1 NaCl, with a flow rate of 1 ml min−1. Fractions were concentrated using Amicon Ultra centrifugal filters (10 kDa) and subsequently dialyzed against 10 mmol l−1 Hepes, 0.5 mmol l−1 EDTA, pH 7.7, at 4°C, to remove NaCl. Purity of the isoHb fractions was verified by means of native polyacrylamide gel electrophoresis (PAGE) using a PhastSystem (GE Healthcare) in 10–15% gradient precast gels. Purified samples were stored at −80°C at a 1 mmol l−1 heme concentration.
Identification of isoHb subunit composition
The subunit composition of each eluted Hb fraction was determined by tandem mass spectrometry (MS/MS). For the MS/MS analysis, eluted fractions were separated on a 20% SDS-PAGE gel and stained with Coomassie Brilliant Blue. The individual α- and β-chain subunits of the purified isoHbs were excised and digested with trypsin. Following previously described protocols (Storz et al., 2011; Revsbech et al., 2013; Storz et al., 2015), peptide mass fingerprints derived from MS/MS were queried with a custom database that included amino acid sequences from all embryonic and adult-expressed α- and β-type globin genes that we annotated from the B. pectinirostris genome assembly. The queries were conducted with the Mascot data search system (version 1.9.0, Matrix Science, London, UK). Search parameters included no restriction on protein molecular weight or isoelectric point, and methionine oxidation was allowed as a variable peptide modification. Mass accuracy settings were 0.15 Da for peptide mass and 0.12 Da for fragment ion masses. We identified all significant protein hits that matched more than one peptide with P<0.05.
Measurement of Hb–O2 equilibria
We measured O2-equilibrium curves of purified isoHbs at 25°C in 0.1 mol l−1 Hepes buffer (pH 7.40) in the absence (stripped) and presence of 0.1 mol l−1 KCl and ATP (0.75 mmol l−1), and in the simultaneous presence of KCl and ATP. The 0.75 mmol l−1 ATP concentration falls within the range of values estimated for teleost red cells (Nikinmaa, 1982; Cadiz et al., 2019) and is therefore relevant to in vivo conditions. Heme concentration for each Hb solution was 0.3 mmol l−1, yielding a 10-fold molar excess of ATP to tetrameric Hb. This ATP:Hb4 ratio permits an assessment of the maximal possible effect of ATP on Hb–O2 affinity. To measure the Bohr effect, we took replicate measurements of stripped isoHbs under identical conditions at pH 6.7 and 7.4. The final pH of Hb solutions used in each experiment was measured at 25°C using an InLab Micro pH electrode (Mettler Toledo). We measured O2 equilibria of 3 µl thin-film samples in a modified diffusion chamber where absorption at 436 nm was monitored during stepwise changes in the equilibration of N2/O2 mixtures generated by precision Wősthoff gas-mixing pumps (Weber, 1992; Grispo et al., 2012; Weber et al., 2013; Natarajan et al., 2015, 2016). We estimated values of P50 and n50 (Hill's cooperativity coefficient at half-saturation) by fitting the Hill equation Y = PnO2/(Pn50 + PnO2) to the experimental O2 saturation data by means of nonlinear regression (where Y is the fractional O2 saturation and n is the cooperativity coefficient). The nonlinear fitting was based on four to nine equilibration steps between 30% and 70% oxygenation, and standard errors are reported for the resultant estimates of P50 and n50. Free Cl− concentrations were measured with a model 926S Mark II chloride analyzer (Sherwood Scientific Ltd, Cambridge, UK).
Genomic organization of the globin gene clusters
We used a genome assembly of the blue-spotted mudskipper, B. pectinirostris, to characterize the physical organization of the globin gene clusters, and we integrated analyses of conserved synteny with phylogenetic reconstructions to unravel the duplicative history of the mudskipper α- and β-type globin genes. We identified a total of four α-type globin genes and three β-type globin genes in the genome of B. pectinirostris. As in other teleosts examined to date, these genes were distributed between two unlinked clusters, designated ‘LA’ and ‘MN’ (the abbreviations refer to flanking genes that demarcate the 5′ and 3′ boundaries of each set of tandemly linked globin genes) (Fig. 1). The LA and MN gene clusters represent the paralogous products of the teleost-specific whole-genome duplication (TGD) (Opazo et al., 2013) that occurred ∼350 million years ago (Meyer and Van de Peer, 2005). The LA cluster of B. pectinirostris contains two α-type globin genes (Hba1-LA and Hba2-LA, from 5′ to 3′) with two β-type globin genes interleaved between them in the opposite orientation (Hbb2-LA and Hbb1-LA), whereas the MN cluster contains three α-type globin genes (Hba1-MN, Hba2-MN and Hba3-MN) (Fig. 2).
IsoHb composition of B. pectinirostris red cells
Red cell lysates from B. pectinirostris were subjected to anion-exchange FPLC, which revealed two major adult isoHbs, HbI and HbII, in approximately equal quantities and with similar anodic mobilities on native polyacrylamide gels. MS/MS analysis of the FPLC-purified isoHbs revealed that each of them incorporate the same β-chain (product of the Hbb1-LA gene) but different α-chains, as HbI incorporates the product of Hba1-LA and HbII incorporates the product of Hba1-MN (Fig. 3A). The two α-globin genes, Hba1-LA and Hba1-MN, are members of the two separate TGD-derived gene clusters (Fig. 2). Given the antiquity of the TGD, it is not surprising that the two α-globin paralogs are highly divergent at the amino acid sequence level, differing at 54 of 141 residue positions (Fig. 3B).
Phylogeny of α- and β-type globin genes
Consistent with the comparative genomic analysis of conserved synteny, phylogenetic analysis revealed that α- and β-type globin genes from a taxonomically diverse set of teleost species clustered into discrete LA and MN clades (Fig. 4). Within each main clade of LA- and MN-associated globin genes, the adult-expressed mudskipper globin genes were in some cases nested within subclades of genes from other species that are annotated as early-expressed embryonic globins. For example, within ‘LA Hba clade 2’ in the phylogeny of α-type globins (Fig. 4A), the adult-expressed Hba1-LA gene of mudskipper forms a subclade with the adult-expressed Hba1-LA gene from medaka (Oryzias latipes) and the early-expressed Hba2-LA gene from Nile tilapia (O. niloticus). Likewise, within the ‘LA Hbb clade 1’ in the phylogeny of β-type globins (Fig. 4B), the adult-expressed Hbb1-LA of mudskipper forms a subclade with the adult-expressed Hbb1-LA of medaka and the early-expressed Hbb1-LA of Nile tilapia, as well as the early- and late-expressed Hbb1-LA genes of green-spotted puffer (Dichotomyctere nigroviridis) and fugu (Takifugu rubripes).
O2-binding properties of mudskipper Hbs
Under identical experimental conditions, O2-equilibrium curves revealed that the two isoHbs exhibited virtually identical O2 affinities (as measured by P50, the PO2 at which Hb is 50% saturated) and cooperativities (as measured by n50, Hill's cooperativity coefficient at half-saturation) (Fig. 5, Table 1). O2 affinities of the two isoHbs were reduced (i.e. P50s were increased) in the presence of ATP but not in the presence of Cl− ions alone (added as 0.1 mol l−1 KCl) (Table 1, Fig. 5). Curves for both isoHbs measured in the joint presence of KCl and ATP revealed a slight decrease in P50 compared with curves measured with ATP alone, indicating that the binding of monovalent Cl− ions interferes with the allosteric binding of ATP in the central cavity of T-state deoxyHb. The O2 affinities of both isoHbs exhibited pronounced sensitivities to changes in pH: Bohr factors (ΔlogP50/ΔpH) were −0.79 and −0.81 for HbI and HbII, respectively.
Oxygenation properties of mudskipper Hbs
Boleophthalmus pectinirostris expresses multiple anodic isoHbs, each with a relatively high intrinsic O2 affinity, moderate sensitivity to organic phosphates, and a large Bohr effect – a ‘class I’ profile, according to the classification scheme of Weber (1990). This same basic pattern of isoHb multiplicity has been well documented in other teleosts, including facultative air breathers and obligate water breathers alike. The Hb system of B. pectinirostris appears functionally similar to that of the swamp eel (Monopterus albus), a facultative air breather in which gas exchange is largely restricted to the buccopharyngeal cavity (Damsgaard et al., 2014, 2015). Similar to B. pectinirostris, M. albus expresses multiple anodic isoHbs and no cathodic components. The two major isoHbs of M. albus exhibit high intrinsic O2 affinities [P50(stripped)=4.8−5.2 torr (25°C, pH 7.7)], similar to values measured for the two isoHbs of B. pectinirostris (Table 1), but the mudskipper isoHbs exhibited a higher sensitivity to ATP and estimated Bohr coefficients were 2- to 3-fold larger when measured under identical buffer conditions. The mudskipper isoHbs have P50 values similar to those reported for the anodic isoHbs of hypoxia-tolerant, water-breathing teleosts such as carp (Cyprinus carpio) (Gillen and Riggs, 1972; Weber and Lykkeboe, 1978; Jensen et al., 2017) and European eel (Anguilla anguilla) (Fago et al., 1997). In contrast to the mudskippers, another facultative air-breathing fish, the Amazonian armored catfish (Hoplosternum littorale), exhibits a far more pronounced level of isoHb differentiation, with anodic and cathodic isoHbs that differ substantially in intrinsic O2 affinity, phosphate sensitivity and Bohr effect (Weber et al., 2000).
The high Hb–O2 affinity of the mudskipper should promote branchial O2 uptake under conditions of aquatic hypoxia and – during terrestrial excursions – transepithelial O2 uptake in the pharyngeal air-breathing structures. The potential drawback of an increased Hb–O2 affinity is that it hinders O2 unloading in the tissue capillaries, but this can be mitigated to some extent by the Bohr effect. The large Bohr effect of B. pectinirostris and some other facultative air-breathing fish may enhance O2 delivery to tissues during metabolic or respiratory acidosis upon switching from aquatic to aerial respiration, where O2 uptake may be less impaired. Likewise, liberation of Bohr protons upon Hb oxygenation would greatly facilitate CO2 excretion. Measurements on whole blood have revealed numerous facultative air-breathing fish species that exhibit pronounced Bohr effects, but there is extensive interspecific variation (Shartau and Brauner, 2014; Bayley et al., 2019; Mendez-Sanchez and Burggren, 2019). Given the especially large Bohr effects of the mudskipper isoHbs, it is worth noting that the β-chain C-termini (HC3) have His (Fig. 3B), which is known to play a major role in deoxygenation-linked proton-binding, whereas the cathodic isoHbs of other teleosts such as Hoplosternum and Anguilla (which exhibit little or no Bohr effect, or even a reverse Bohr effect in the absence of organic phosphates) have non-ionizable Phe at the same residue position (Fago et al., 1995; Weber et al., 2000).
In addition to the hypothesis that obligate and facultative air-breathing fishes have evolved reduced Hb–O2 affinities compared with obligate water breathers (Johansen and Lenfant, 1972; Johansen et al., 1978; Powers et al., 1979), it has been suggested that facultative air-breathing fishes may have evolved Hbs with reduced Bohr effects to cope with CO2 retention and/or exposure to hypercarbic water (Carter, 1931). Available data do not appear to support either hypothesis. Among ray-finned fish, lobe-finned fish and amphibians, facultative air and water breathers tend to have higher blood–O2 affinities than obligate air breathers, and facultative air breathers tend to have slightly larger Bohr effects than members of the other two groups (Bayley et al., 2019). The high intrinsic O2 affinities and large Bohr effects that we measured for the purified mudskipper isoHbs appear to be consistent with both trends.
Genomic insights into the origins of Hb multiplicity in mudskippers
Given that the two adult-expressed isoHbs of B. pectinirostris are present at approximately equimolar concentrations in the red cell, it seemed reasonable to expect that the two distinct α-chain subunits would be encoded by a tandemly linked pair of genes under the transcriptional control of the same cis-acting regulatory elements. Such a linkage arrangement could also help explain why the two isoHbs exhibit such similar O2-binding properties, because tandem duplicates often have similar coding sequences owing to interparalog gene conversion (Hoffmann et al., 2008 a,b; Runck et al., 2009; Storz et al., 2010; Gaudry et al., 2014; Natarajan et al., 2015; Signore et al., 2019). Contrary to this seemingly reasonable expectation, the analysis of conserved synteny revealed that the genes that encode the α-type subunits of HbI and HbII are members of unlinked, paralogous gene clusters that represent the duplicated products of the teleost-specific genome duplication (Figs 1 and 2).
In the phylogenies of α- and β-type globins, the fact that adult-expressed mudskipper genes clustered with some embryonically expressed globin genes of other teleosts is not too surprising, as evolutionary changes in stage-specific expression during ontogeny have been well documented in teleosts (Opazo et al., 2013) and in many tetrapods as well (Opazo et al., 2008; Hoffmann et al., 2010, 2018; Storz, 2016, 2019).
IsoHb multiplicity and lack of functional differentiation
In the case of mudskippers, there is no appreciable difference in the functional properties of co-expressed isoHbs in spite of extensive amino acid divergence between the alternative α-chain subunit isoforms (Fig. 3B). It therefore appears that the ability to switch between aquatic and aerial respiration does not necessarily require a division of labor between functionally distinct isoHbs that are specialized for O2 transport under different conditions. The adult-expressed isoHbs of many obligate water-breathing fish are known to exhibit much higher levels of functional differentiation than those of mudskippers, including quantitative differences in O2 affinity and qualitative differences in the mode of allosteric regulation, especially in relation to the magnitude of the Bohr effect and sensitivity to organic phosphates (Weber, 1990, 2000). In many teleost species, the presence of specific isoHbs with functional specializations such as the Root effect (an extreme form of pH sensitivity that plays a key role in oxygen secretion and general tissue O2 delivery) exert an important influence on physiological capacities (Berenbrink, 2007; Rummer et al., 2013; Randall et al., 2014). However, the absence of functional isoHb differentiation in amphibious mudskippers and other facultative air breathers, and the often pronounced functional heterogeneity in the isoHbs of obligate water breathers, supports the conclusions of several authors (Fyhn et al., 1979; Ingermann, 1997; Wells, 2009; Storz, 2019) that the overall diversity of co-expressed isoHbs in fish red cells is not generally a strong determinant of physiological versatility or ecological niche breadth. In mudskippers, any changes in blood–O2 affinity that are associated with transitions between aerial and aquatic breathing are likely caused by changes in red cell pH and/or red cell concentrations of nucleotide triphosphates.
We thank Elin E. Petersen for assistance in the lab and two anonymous reviewers for helpful comments.
Conceptualization: J.F.S., A.F.; Methodology: C.N., M.K.G., M.V., F.G.H., A.F.; Formal analysis: J.S., C.N., M.V., F.G.H., A.F.; Investigation: C.N., M.K.G., M.V., F.G.H., A.F.; Resources: J.F.S., X.Y., B.V.; Writing - original draft: J.F.S.; Writing - review & editing: C.N., M.V., F.G.H., X.Y., B.V., A.F.; Supervision: J.F.S., A.F.; Project administration: J.F.S.; Funding acquisition: J.F.S.
This research was supported by a National Institutes of Health grant to J.F.S. (HL087216) and National Science Foundation grants to J.F.S. (OIA-1736249 and IOS-1927675). Deposited in PMC for release after 12 months.
The authors declare no competing or financial interests.