1. Developmental Biology
  2. Evolutionary Biology
Download icon

Long-distance communication by specialized cellular projections during pigment pattern development and evolution

  1. Dae Seok Eom
  2. Emily J Bain
  3. Larissa B Patterson
  4. Megan E Grout
  5. David M Parichy  Is a corresponding author
  1. University of Washington, United States
Research Article
  • Cited 27
  • Views 3,061
  • Annotations
Cite this article as: eLife 2015;4:e12401 doi: 10.7554/eLife.12401

Abstract

Changes in gene activity are essential for evolutionary diversification. Yet, elucidating the cellular behaviors that underlie modifications to adult form remains a profound challenge. We use neural crest-derived adult pigmentation of zebrafish and pearl danio to uncover cellular bases for alternative pattern states. We show that stripes in zebrafish require a novel class of thin, fast cellular projection to promote Delta-Notch signaling over long distances from cells of the xanthophore lineage to melanophores. Projections depended on microfilaments and microtubules, exhibited meandering trajectories, and stabilized on target cells to which they delivered membraneous vesicles. By contrast, the uniformly patterned pearl danio lacked such projections, concomitant with Colony stimulating factor 1-dependent changes in xanthophore differentiation that likely curtail signaling available to melanophores. Our study reveals a novel mechanism of cellular communication, roles for differentiation state heterogeneity in pigment cell interactions, and an unanticipated morphogenetic behavior contributing to a striking difference in adult form.

https://doi.org/10.7554/eLife.12401.001

eLife digest

Animals have very different patterns of skin pigmentation, and these patterns can be important for survival and reproduction. Zebrafish, for example, have horizontal dark and light stripes along their bodies, while a closely related fish called the pearl danio has an almost uniform pattern.

The dark stripes of the zebrafish contain cells called melanophores, while the lighter regions contain two other types of cells known as xanthophores and iridophores. These pigment cell types interact with each other to create stripes. The iridophores establish the lighter stripes and specify the position and orientation of the dark stripes. They also produce a protein called Csf1, which allows the xanthophores to mature. As the stripes form, melanophores present in lighter stripes move into nearby dark stripes. Pearl danios also contain these three types of pigment cells, but these cells remain intermingled giving the fish their uniform color.

Eom et al. have now used microscopy to image pigment cells in zebrafish and pearl danio to uncover how interactions between these cells differ in species with different pigment patterns. The technique involved tagging pigment cells with fluorescent markers and using time-lapse imaging to track them during the formation of the adult pigmentation pattern.

The experiments show that stripes form in zebrafish because the cells that make the xanthophores form long, thin projections that extend to neighboring melanophores. These so-called ‘airinemes’ deliver materials to melanophores and help to clear the melanophores from interstripe regions, partly by activating a cell communication pathway called Delta-Notch signaling. These cell projections are mostly absent from the cells that make xanthophores in the pearl danio due to differences in when Csf1 is produced. This alters the timing of when the xanthophores develop, leading to the loss of long-distance airineme signaling.

Eom et al.’s findings identify a new way in which cells can communicate and an unanticipated cell behavior that contributes to a striking difference in the pigmentation patterns of zebrafish and pearl danio. Future studies should further our understanding of these unique projections and reveal whether they are produced by other types of cells.

https://doi.org/10.7554/eLife.12401.002

Introduction

Genes contributing to phenotypic diversification are beginning to be identified, yet the morphogenetic mechanisms by which changes in gene activities are translated into species differences in form remain virtually unknown. Holding great promise for identifying such mechanisms are the diverse and ecologically important pigment patterns of fishes (Endler, 1980; Houde, 1997; Wang et al., 2006; Engeszer et al., 2008; Price et al., 2008; Seehausen et al., 2008; Roberts et al., 2009; Kelley et al., 2013) because pigment cell behaviors are observable as phenotypes unfold and because pigment pattern development is amenable to both experimental manipulation and theoretical modeling (Painter et al., 1999; Kondo and Miura, 2010; Miyazawa et al., 2010; Caballero et al., 2012; Yamanaka and Kondo, 2014; Bullara and De Decker, 2015; Volkening and Sandstede, 2015). In this regard, fishes of the genus Danio should be especially useful as their adult pigment patterns differ markedly among species and mechanisms of pattern formation are starting to be understood in zebrafish, D. rerio (Parichy, 2015; Parichy and Spiewak, 2015; Singh and Nusslein-Volhard, 2015; Watanabe and Kondo, 2015).

Zebrafish have dark stripes of black melanophores with light interstripes of yellow-orange xanthophores and iridescent iridophores (Figure 1A), all of which are derived from the neural crest, either directly, or through stem cell intermediates (Budi et al., 2011; Dooley et al., 2013; Mahalwar et al., 2014; McMenamin et al., 2014; Singh et al., 2014). Interactions among pigment cells are essential to pattern formation (Watanabe and Kondo, 2015). Iridophores differentiate in the prospective interstripe, and specify the positions and orientations of melanophore stripes (Frohnhofer et al., 2013; Patterson and Parichy, 2013). Iridophores also promote the differentiation of xanthophores in the interstripe by expressing Colony stimulating factor-1 (Csf1)(Patterson and Parichy, 2013). Subsequently, interactions between cells of melanophore and xanthophore lineages are required during a period of stripe consolidation, in which some melanophores initially in the interstripe join the stripes, and, simultaneously, the stripe borders become increasingly organized (Parichy et al., 2000; Maderspacher and Nusslein-Volhard, 2003; Parichy and Turner, 2003a, 2003b; Quigley et al., 2005; Takahashi and Kondo, 2008). Finally, this pattern is reiterated as additional iridophores invade the stripes, ultimately emerging on the other side, where they terminate the first stripes and initiate new interstripes and stripes dorsally and ventrally (Patterson et al., 2014; Singh et al., 2014).

Figure 1 with 3 supplements see all
Pigment cell projections.

(A) Zebrafish and pearl danio. Right, melanophores and xanthophores (arrows) after epinephrine treatment to contract pigment granules. (B) Long projections by zebrafish aox5+ cells of xanthophore lineage (arrows) with membraneous vesicles (arrowhead, inset). (C) Zebrafish aox5+ cells were more likely to extend projections than pearl, especially during early stripe development [7–8 SSL (Parichy et al., 2009); species x stage, χ2=103.4, d.f.=4, p<0.0001; N=929, 1259 cells for zebrafish and pearl; projections per cell: χ2=45.3, d.f.=1, p<0.0001]. (D) In zebrafish, projections were often long and fast. Bars indicate median ± interquartile range (IQR). (E) Extension and retraction (arrow) and release of vesicle (arrowhead) in zebrafish but not pearl. Scale bars: 5 mm (A, left); 50 µm (A, right); 10 µm (B); 50 µm (E).

https://doi.org/10.7554/eLife.12401.003

Despite an increasing appreciation for the phenomenology of interactions among zebrafish pigment cells through cell transplantation, genetic analyses, laser ablation, in vitro manipulations, and theoretical modeling (Maderspacher and Nusslein-Volhard, 2003; Nakamasu et al., 2009; Eom et al., 2012; Yamanaka and Kondo, 2014), the cellular and molecular mechanisms relevant to these interactions in vivo remain largely unknown. Moreover, the ways in which such interactions may have changed during the development of naturally occurring, alternative pattern states across species have yet to be explored. Interesting in this context is pearl danio, D. albolineatus, which has a nearly uniform pattern of fewer melanophores, more xanthophores and an intermingling of all three pigment cell classes (Figure 1A) (Quigley et al., 2005; Mills et al., 2007; Patterson et al., 2014; McCluskey and Postlethwait, 2015).

Here, we show by time-lapse imaging of cells in their native tissue environment that stripe consolidation in zebrafish requires a novel class of fast, long cellular projection, extended by cells of the xanthophore lineage to melanophores. These projections, which we call 'airinemes,' contribute to transducing a Delta-Notch signal that promotes the clearance of melanophores from the developing interstripe. We further show that production and targeting of airinemes are differentiation-state dependent. Finally, using interspecific cell transplantation and transgenic manipulations we demonstrate that evolution of the very different, uniform pattern of pearl danio has entailed the loss of long-distance airineme signaling, concomitant with modifications to xanthophore differentiation. Our results provide novel insights relevant to empirical and theoretical understanding of pigment pattern formation in zebrafish, as well as mechanisms of cellular communication and how they change during the evolution of adult phenotypes.

Results

Fast projections are extended frequently by cells of the xanthophore lineage in zebrafish but not pearl danio

To determine if species differences are associated with modifications to pigment cell interactions, we visualized cells using membrane-targeted fluorophores and time-lapse imaging during adult pigment pattern formation (Budi et al., 2011). In zebrafish, cells of the xanthophore lineage, marked by expression of aldehyde oxidase 5 (aox5; formerly, aox3) (Parichy et al., 2000; McMenamin et al., 2014), exhibited long, fast projections with distinctive, membraneous vesicles at their tips; vesicles were often left behind when the projection carrying them retracted or fragmented (Figure 1B,E; Video 1). Such projections were especially frequent at early stages of stripe formation (7–8 SSL; Figure 1C-left; Figure 1—figure supplement 1) (Parichy and Turner, 2003b; Parichy et al., 2009; Patterson and Parichy, 2013) and individual cells could extend several projections over 18 hr of time-lapse imaging (Figure 1C-right). Projections reached as far as 5–6 cell diameters and extended and retracted quickly (Figure 1D,E). Incidences of projection formation were not markedly altered in mutants lacking melanophores or iridophores, or in transgenic fish having supernumerary melanophores (Figure 1—figure supplement 2; Video 2).

Video 1
Projections by cells of xanthophore lineage in zebrafish but not pearl.

Left, in zebrafish D. rerio, an aox5+ cell in a mosaically labeled larva extends fast, thin cellular projections. Right, such projections are not apparent in two aox5+ cells in pearl. 10 min interval, 580 min total.

https://doi.org/10.7554/eLife.12401.007
Video 2
Fast projections of aox5+ cells persist in backgrounds with altered numbers of melanophores or iridophores.

Shown are representative fields illustrating fast projections (arrows) of aox5+ cells in the presence of excess melanophores (Kitlga++), an absence of melanophores (mitfa) and an absence of iridophores (ltk). 5 min interval, 375 min total.

https://doi.org/10.7554/eLife.12401.008

By contrast, fast projections were extended only rarely by aox5+ cells of pearl danio (Figure 1C,EFigure 1—figure supplement 1 Video 1). Likewise, fast projections were rare among melanophores of both species, marked by expression of tyrosinase related protein 1b (tyrp1b; Figure 1—figure supplement 3, Video 3) and were not observed for iridophores, marked by purine nucleoside phosphorylase 4a (pnp4a; Video 4) (Lang et al., 2009; Curran et al., 2010; McMenamin et al., 2014).

Video 3
Fast projections were only infrequently produced by cells of the melanophore lineage.

Melanophores labeled with tyrp1b:membrane-mCherry. Most cellular processes were robust in size and relatively slow moving, although some could extend over long distances (yellow arrowhead). Only rarely were fast cellular projections extended (white arrow near top of frame). M, lightly melanized, differentiating adult melanophore in the prospective stripe region. EL, brownish embryonic/early larval melanophore persisting at the horizontal myoseptum in the prospective interstripe region. *, macrophage carrying mCherry+ debris. 5 min interval, 785 min total.

https://doi.org/10.7554/eLife.12401.009
Video 4
Iridophores did not extend fast projections.

Iridophores that have started to differentiate in the prospective interstripe (and in the vicinity of persisting EL melanophores), labeled with membrane-GFP driven by the promoter of purine nucleoside phosphorylase 4a (pnp4a). Thin, short and relatively straight projections are often extended at cell edges (e.g., arrow). An individual pnp4a+ cell is observed dispersing from the mat of aggregated pnp4+ iridophores (arrowhead). 5 min interval, 475 min total.

https://doi.org/10.7554/eLife.12401.010

Thus, long, fast, and vesicle-containing projections were produced exuberantly by cells of the zebrafish xanthophore lineage during stages of stripe formation, but were not common to other pigment cell classes in this species or to cells of the xanthophore lineage in pearl danio.

Fast pigment cell projections are distinct in cytoskeleton and morphology from long filopodia

We asked whether fast projections of pigment cells were similar to previously described cellular projections. Fast projections were considerably longer than typical (<10 µm) filopodia, that lacked vesicles (e.g., Figure 1—figure supplement 3D; and below). Longer, often relatively straight, actin-based filopodia, or ‘cytonemes,’ function in intercellular communication in other systems (Miller et al., 1995; Ramirez-Weber and Kornberg, 1999; De Joussineau et al., 2003; Cohen et al., 2010; Caneparo et al., 2011; Danilchik et al., 2013; Massarwa and Niswander, 2013; Sanders et al., 2013; Gradilla et al., 2014; Luz et al., 2014; Roy et al., 2014; Stanganello et al., 2015). Similar to cytonemes, fast pigment cell projections contained F-actin, as evidenced by labeling with fluorescent reporters fused to the calponin homology domain of utrophin (UtrCH) as well as LifeAct (Burkel et al., 2007; Riedl et al., 2008; Sanders et al., 2013) (Figure 2A,B).

Fast projections harbor microfilaments and microtubules.

(A, B) aox5+ projections and puncta (arrowheads) labeled for F-actin, revealed by UtrCH-mCherry (A) and LifeAct-mKate (B). (C) Projections contained tubulin, as revealed by Tuba1b-mCherry. In some instances tubulin was absent from vesicles (arrowhead, upper) and in other instances was found in vesicles (arrow, lower). (D) Accumulations of microtubule end binding protein EB3 fused to GFP (arrowhead), were present in vesicles as well (arrow); aox5 here drives mCherry. Scale bars: 20 µm (A); 10 µm (B, C, D).

https://doi.org/10.7554/eLife.12401.011

In contrast to cytonemes, however, fast projections had highly meandering trajectories and the vesicles associated with them had diameters of 1.7 ± 0.1 µm (mean ± SE, n=22), considerably larger than the 30–200 nm exosome-like particles trafficked within cytonemes (Bischoff et al., 2013; Sanders et al., 2013; Gradilla et al., 2014). Also unlike cytonemes, filaments of fast projections contained tubulin, as indicated by localization of an alpha tubulin fusion protein; such labeling was present in some but not all projection-associated vesicles (Figure 2C). Suggesting active microtubule assembly, fast projections and vesicles also exhibited transient accumulations of microtubule plus-end binding protein EB3 (Figure 2D; Video 5).

Video 5
Microtubule plus-end binding protein EB3 localized transiently in airinemes.

Upper, membrane labeling with aox5:membrane-mCherry. Middle, EB3-GFP fusion protein driven by aox5 promoter. Lower, merge. EB3-GFP is present transiently as airinemes extend (arrow). Green spots in background are accumulations of xanthophore pigment in neighboring cells. 5 min interval, 190 min total.

https://doi.org/10.7554/eLife.12401.012

Fast projections were also 1–2 orders of magnitude longer than recently observed actin- and tubulin-containing projections of chick somitic epithelial cells (Sagar et al., 2015) and Drosophila germ line stem cells (Inaba et al., 2015), as well as cellular bridges that can transfer cytoplasmic materials between embryonic neural crest cells (McKinney et al., 2011). Finally, these projections differed markedly from long but slow and robust, pseudopodial-like processes sometimes extended by melanophores transiently during development (Video 3) and present in adult fish (Hamada et al., 2014).

Given their distinctive morphology and cytoskeletal composition, as well as their function (below), we termed these fast projections ‘airinemes,’ for Iris, who—fleet as the wind on golden wings—delivered messages for the gods (Homer, ~8th cent. BCE), as well as for Sir George Biddell Airy, who described limits on optical resolution (Cox, 2012).

Airinemes are produced by xanthoblasts within melanophore stripes, rather than xanthophores within interstripes

As a first step towards understanding potential roles for airinemes in stripe formation, we sought to further characterize the cells that extend them. In zebrafish, aox5+ cells occur in the prospective interstripe, where they differentiate as xanthophores [beginning ~6.5 SSL (Parichy et al., 2009; Patterson and Parichy, 2013)]. Yet, aox5+ cells also occur at lower densities in developing and completed stripes, where they remain unpigmented or lightly pigmented (Figure 3A); we refer to these incompletely differentiated cells as xanthoblasts (McMenamin et al., 2014).

Airinemes were produced by xanthoblasts within prospective stripe regions.

(A) Detail of developing pattern in zebrafish, illustrating pigmented aox5+ xanthophores of the interstripe as well as unpigmented aox5+ xanthoblasts of the prospective stripe (insets, boxed regions shown at higher magnification). Larva shown is 8.6 SSL, when xanthophore pigment is more readily visible, but after the peak of airineme production (Figure 1C). (B) aox5+ cells in prospective stripe regions were more likely to extend airinemes than aox5+ cells of the interstripe (χ2=28.6, d.f.=1, p<0.0001, N=295 cells). (C) Xanthoblasts (upper) had numerous membrane blebs (arrowhead), whereas xanthophores (lower) had smooth surfaces and more lobular edges (7.8 SSL). (D) Forced differentiation (TH++) reduced the incidence of cells that extended airinemes (left; χ2=12.2, d.f.=1, p<0.0001, N=123 cells) and the numbers of airinemes extended by each cell (right; TH++; χ2=12.0, d.f.=1, p<0.05), whereas differentiation-arrest increased airineme production (TH–; χ2=29.6, d.f.=1, p<0.0001; 7.5 SSL) Scale bars: 50 µm (A); 10 µm (C).

https://doi.org/10.7554/eLife.12401.013

To determine if interstripe and stripe populations of aox5+ cells differ in airineme production we compared behaviors of cells in these locations during the peak of airineme deployment (7.5 SSL; Figure 1C). We found that cells within stripe regions, presumptive xanthoblasts, were more likely to extend airinemes than were cells within the developing interstripe, presumptive xanthophores (Figure 3B). Inspection of aox5+ cells at high resolution further revealed that unpigmented xanthoblasts, but not pigmented xanthophores, had membrane blebs from which airinemes originated (Figure 3C; Videos 6, 7).

Video 6
Membrane blebs of xanthoblast.

Shown is a static image of a single aox5+ xanthoblast within the prospective stripe region, illustrating numerous membrane blebs, limited primarily to the superficial (epidermal-facing) surface.

https://doi.org/10.7554/eLife.12401.014
Video 7
Airinemes arose from membrane blebs of xanthoblasts.

Shown are corresponding views of same field exposed to highlight membrane blebs from which airineme originates (upper, red arrow) and airineme filament (lower). An additional bleb formed near the end of observation (yellow arrow, upper). 5 min interval, 300 min total.

https://doi.org/10.7554/eLife.12401.015

To test experimentally if airineme production was contingent upon differentiation state, we exploited the thyroid hormone (TH) dependence of xanthophore differentiation: aox5+ cells are differentiation-arrested in fish lacking TH, but differentiate fully across the flank in fish expressing excess TH (McMenamin et al., 2014). We found that aox5+ cells arrested as xanthoblasts (TH–) produced more airinemes than aox5+ cells forced to differentiate as xanthophores (TH++; Figure 3D; Video 8). These findings are consistent with the interpretation that airineme production is differentiation-state specific within the xanthophore lineage (and see below)

Video 8
Airineme production was differentiation-state dependent.

Left, a wild-type control aox5+ cell extended several airinemes. Middle, exuberant airineme production by differentiation-arrested aox5+ cells in hypothyroid (TH–) fish in which the thyroid had been ablated transgenically at 4 days post-fertilization (McMenamin et al., 2014). Right, forced aox5+ cell differentiation in hyperthyroid (TH++) mutant (opallus, tshrD632Y) resulted in failure of airineme production. Spots of autofluorescence are evident in neighboring cells lacking aox5:membrane-GFP. 5 min interval, 795 min total.

https://doi.org/10.7554/eLife.12401.016

Airinemes are target-specific and required for melanophore consolidation into stripes

To assess the significance of airinemes for pigment pattern formation we sought to block their production. Consistent with actin and tubulin cytoskeletal dependencies, airineme production was curtailed by acute treatment with the myosin II inhibitor blebbistatin (Kovacs et al., 2004) and the microtubule polymerization inhibitor nocodazole (Figure 4—figure supplement 1A), although longer term whole-fish treatments were lethal. By contrast, low levels of the Cdc42 small GTPase inhibitor ML141 (Surviladze et al., 2010) inhibited airineme production (Video 9) while allowing for extended whole-fish treatments. In larvae treated with ML141 through adult pigment pattern formation, melanophores occurred ectopically within the interstripe (Figure 4—figure supplement 1A,B).

Video 9
Pharmacological inhibition of airineme production.

Representative fields in which fast projections are extended by aox5+ cells in DMSO-treated controls, but not during acute administration of nocodazole, ML141 or blebbistatin. Note that cells continue to change shape and extend some slow moving processes, similar to controls. Stationary fluorescent puncta are pigment autofluorescence in neighboring cells lacking aox5:membrane-GFP. 5 min interval, 450 min total.

https://doi.org/10.7554/eLife.12401.017

To block airineme production specifically in the xanthophore lineage we constructed a TetGBD (‘Tet’) transgene (Knopf et al., 2010; Patterson and Parichy, 2013) driven by aox5 to express dominant negative Cdc42N17 (dnCdc42) (Kieserman and Wallingford, 2009), inducible in a temporally specific manner with dexamethasone and doxycycline (dd; Figure 4—figure supplement 2A). Cdc42 has roles in cytoskeletal organization and a variety of cell behaviors and its inhibition blocks filopodial extension in other systems (Etienne-Manneville, 2004; Wu et al., 2008; Chen et al., 2012; Sadok and Marshall, 2014; Stankiewicz and Linseman, 2014). We found that extended, low-level induction of dnCdc42 in aox5+ cells inhibited airineme production (Figure 4A; Video 10) without significantly affecting the production of short filopodia, lamellipodia-like protrusive activities, the numbers of differentiated xanthophores or melanophores, or the distributions of aox5+ cells (Figure 4—figure supplement 2B–F). Yet, in aox5:Tet:dnCdc42 fish inhibited for airineme production, melanophores persisted ectopically in the interstripe (Figure 4B), as for whole fish treated with ML141.

Figure 4 with 4 supplements see all
Airineme dependent patterning and airineme targeting specificities.

(A) dnCdc42 blocks airineme extension (χ2=16.4, d.f.=1, p<0.0001, N=43 cells total). (B) Interstripe melanophores persisted when airinemes were blocked with dnCdc42. Cell states are indicated by logos in lower left corners (X, aox5+ xanthophore lineage). Insets, brownish melanophores persisting from embryonic/early larval pattern and gray–black adult melanophores. (C) Airinemes contacting melanophores (arrows; Me, early larval; Mn, new; Mp, previously differentiated; xan, xanthophore). (D) Stabilization times (median ± IQR) of aox5+ airinemes on cells of melanophore (M) or xanthophore (X) lineages for zebrafish, cx41.8 mutant zebrafish, and pearl danio. aox5+ airinemes of wild-type zebrafish were less likely to stabilize, and stabilized more briefly (*, both p<0.0001) after contacting cells of the xanthophore lineage as compared to melanophores; this target specificity was altered in cx41.8 mutant zebrafish as well as pearl danio. Y-axis is split for clarity. (E) Zebrafish aox5+ airinemes were most likely to stabilize on Me and Mn (*, p<0.0001; median ± IQR). (F) cx41.8 mutant airinemes stabilized on aox5+ cells (arrow). (G) Vesicle transfer (arrow) to melanophore. Scale bars: 200 µm (B); 50 µm (C); 50 µm (F); 25 µm (G).

https://doi.org/10.7554/eLife.12401.018
Video 10
Airineme production was inhibited by dnCdc42.

Upon treatment with dd, non-transgenic control cells extended airinemes normally, whereas cells in fish transgenic for aox5:Tet:dnCdc42 failed to extend airinemes despite forming other processes. Stationary fluorescent puncta are pigment autofluorescence in GFP– cells. 5 min interval, 415 min total.

https://doi.org/10.7554/eLife.12401.023

These findings suggested that airinemes promote the consolidation of melanophores into stripes. During normal stripe development embryonic melanophores persisting from the early larval pigment pattern occur near the horizontal myoseptum, within the prospective interstripe. Many of these cells move short distances to join the developing stripes (Parichy et al., 2000; Parichy and Turner, 2003b; Takahashi and Kondo, 2008; Patterson and Parichy, 2013; Patterson et al., 2014), and their brownish color makes them distinguishable from gray–black adult melanophores that differentiate post-embryonically (Quigley et al., 2004; Parichy and Spiewak, 2015). In wild-type controls, these brownish embryonic melanophores had translocated to the edges of adult stripes, whereas in aox5:Tet:dnCdc42-expressing fish these melanophores remained in the interstripe (insets, Figure 4B).

Given these observations, we predicted that aox5+ airinemes interact directly with melanophores and so we time-lapse imaged aox5+ cells simultaneously with melanophores marked by tyrp1b. In these larvae we observed frequent, prolonged stabilization of aox5+ airinemes on melanophores (Figure 4c,d-left; Video 11). In contrast, airinemes retracted rapidly after contacting aox5+ cells (Figure 4d-left), indicating target specificity.

Video 11
Airinemes extended by cells of the xanthophore lineage stabilized on melanophores.

Shown are prospective dorsal stripe regions of two individuals; prospective interstripe is at bottom of frame. In each view, an aox5+ cell (green) extends an airineme (arrow) that stabilized (*) on a nearby tyrp1b:mCherry+ melanophore (magenta). EL, persisting embryonic/early larval melanophore. M, lightly pigmented adult melanophore. M*, more heavily pigmented adult melanophore. Video shows brightfield views before and after time-lapse. 5 min interval, 475 min total.

https://doi.org/10.7554/eLife.12401.024

Airineme targeting was further specific to subpopulations of melanophores. In addition to persisting embryonic melanophores (Me), newly differentiating adult melanophores (Mn) join the stripes, whereas previously differentiated adult melanophores (Mp) occur already within the stripes (Figure 4C) (Parichy et al., 2000; Parichy and Turner, 2003b; Takahashi and Kondo, 2008; Patterson and Parichy, 2013). Airinemes often stabilized on Me and newly differentiating, still migratory adult melanophores (Mn), but rarely on previously differentiated adult melanophores (Mp) (Figure 4E; Figure 4—figure supplement 3; Videos 11, 12). Moreover, airineme-delivered vesicles often remained on melanophores long after the airineme itself (Figure 4G; Video 13). In some instances, vesicles were transferred without evident airinemes (Video 14), a behavior that was also abrogated by dnCdc42.

Video 12
Airineme targeting differed among melanophore subpopulations.

Several previously differentiated tyrp1b+ melanophores (M; magenta) are shown as well as an initially unpigmented tyrp1b+ melanoblast (mb, upper left). Although numerous airinemes were extended by aox5+ cells in the vicinity of melanophores, only one (yellow arrow) stabilized (*) on a previously differentiated melanophore. Another airineme stabilized on the differentiating melanoblast (yellow arrowhead), which received additional aox5+ puncta. Video hows brightfield views before and after time-lapse and illustrates acquisition of melanin by as the melanoblast continued to differentiate. 5 min interval, 905 min total.

https://doi.org/10.7554/eLife.12401.025
Video 13
Airineme-delivered membraneous vesicles persisted on melanophores.

Left, aox5+ cell (green) extending airineme. Middle, tyrp1b+ melanophores (magenta). Right, merge. After the airineme contacted the melanophore it retracted, leaving behind a GFP+ vesicle that persisted on the melanophore membrane (arrow). 5 min interval, 310 min total.

https://doi.org/10.7554/eLife.12401.026
Video 14
Membrane vesicles could be transferred in the apparent absence of airinemes.

Left, aox5+ cell (green). Middle, tyrp1b+ melanophores (magenta). Right, merge. Several vesicles are indicated by yellow and white arrows. 5 min interval, 325 min total.

https://doi.org/10.7554/eLife.12401.027

To test whether aox5+ airineme targeting is genetically labile, we screened zebrafish pigment pattern mutants and also re-examined the very few airinemes produced in pearl danio. As compared to wild-type zebrafish, we found that airinemes in spotted connexin41.8 (cx41.8; leopard) mutants (Maderspacher and Nusslein-Volhard, 2003; Watanabe et al., 2006) were later-appearing and more promiscuous, stabilizing as extensively on aox5+ cells as on melanophores (Figure 4D-right, 4F; Figure 4—figure supplement 4A,B; Video 15). Such differences could reflect acute morphogenetic interactions, perhaps dependent on connexin hemichannels, or chronic defects in differentiation, as xanthophores of presumptive cx41.8 mutants are reported to contain melanin and melanosome-like vesicles (Kirschbaum, 1975). Interestingly, the rare airinemes of pearl danio also stabilized on xanthophores though inspection cx41.8 sequence and expression did not reveal gross differences between zebrafish and pearl (Figure 4D-right; Video 16; and data not shown).

Video 15
Altered airineme targeting altered by connexin mutation.

Upper, wild-type zebrafish airinemes (white arrow) retracted rapidly after contacting other aox5+ cells. Lower, leopard (cx41.8) mutant airinemes (red arrow) frequently stabilized on other aox5+ cells. 5 min interval, 450 min total.

https://doi.org/10.7554/eLife.12401.028
Video 16
Rare airinemes projected by pearl aox5+ cells stabilized on xanthophores.

Representative airinemes (arrowheads) stabilized on xanthophores, marked by orange pigment granule at cell centers (e.g., x). 5 min interval, 385 min total.

https://doi.org/10.7554/eLife.12401.029

Together these results suggest that, in wild-type zebrafish, airinemes extended by aox5+ xanthoblasts in prospective stripe regions contact melanophores still in the interstripe and promote their clearance during stripe consolidation, and that specificity of airineme targeting is genetically (and evolutionarily) labile.

Airinemes promote Delta–Notch signaling necessary for stripe consolidation

We sought to understand molecular bases for aox5+ airineme effects on melanophore patterning in zebrafish. A good candidate for mediating such interactions is Delta-Notch signaling. In amniotes, Notch promotes melanocyte migration and survival and also regulates differentiation (Moriyama et al., 2006; Schouwey et al., 2007; Aubin-Houzelstein et al., 2008; Kumano et al., 2008; Pinnix et al., 2009). In zebrafish as well as pearl danio, xanthophores express Delta genes whereas melanophores express Notch genes (Hamada et al., 2014) (Figure 5—figure supplement 1A). Consistent with a role for Notch in promoting melanophore migration, we found that expressing the constitutively active Notch1a intracellular domain in melanophores, using a promoter derived from the microphthalmia a (mitfa) gene (Lister et al., 1999), resulted in stripes that were significantly broader than in wild type (Figure 5—figure supplement 1B). Conversely, delta c (dlc) mutants exhibited melanophores ectopically in the interstripe, similar to the aox5:Tet:dnCdc42 phenotype (Figure 5—figure supplement 1C).

We predicted that if airinemes contribute to transducing a Delta-Notch signal from xanthoblasts to melanophores, then airinemes or their associated vesicles should harbor Delta proteins and perturbations to airineme production should impair Notch signaling within melanophores. We recombineered a 109 kb BAC containing zebrafish dlc coding sequence and regulatory elements to generate an mCherry fusion that was functional in rescuing dlc knockdown phenotypes; static and time-lapse imaging confirmed that DlC-mCherry was present in airineme vesicles (Figure 5A; Video 17). Technical limitations precluded corresponding analyses for a second Delta ligand expressed by xanthophores, Dll4 (Figure 5—figure supplement 1A).

Figure 5 with 5 supplements see all
Melanophore clearance requires airineme-dependent Notch signaling in melanophores.

(A) aox5+ airineme vesicles harbored DlC-mCherry (arrow). (B) Frequencies of Tp1+ Notch-responding melanophores were reduced in the absence of aox5+ cells (csf1ra; χ2=15.5, d.f.=1, p<0.0001, N=77 cells) or airinemes (dnCdc42; χ2=9.0, d.f.=1, p<0.005, N=46 cells). Epidermal and muscle cells were also Tp1+ (not shown) but within the hypodermis, where pigment cells reside (Hirata et al., 2003), we observed only melanophores to be Tp1+. (C) Interstripe melanophores persisted when Notch signaling was blocked within the melanophore lineage by dnSu(H) [melanophore logo with green outlining], similar to dnCdc42 blockade of aox5+ airinemes (compare with Figure 4B). (D) Quantification of interstripe melanophores (mean ± SE) in dd-treated non-transgenic fish as well as mitfa:Tet:dnSu(H) and aox5:Tet:dnCdc42 (overall, F2,12=58.6, p<0.0001). (E) When induced with threshold levels of dd, persisting interstripe melanophores (means ± SE) were threefold more abundant (t7=2.9, p<0.05) in fish doubly transgenic for mitfa:Tet:dnSu(H) and aox5:Tet:dnCdc42 as compared to singly transgenic fish (shown at 9.8 SSL). Scale bars: 15 µm (A, left); 10 µm (A, right); 20 µm (B); 200 µm (C); 200 µm (E).

https://doi.org/10.7554/eLife.12401.030
Video 17
Airineme vesicles harbored Dlc-mCherry.

Left, aox5+ cell (green) extending airineme. Middle, dlc:Dlc-mCherry (magenta). Right, merge. 5 min interval, 170 min total.

https://doi.org/10.7554/eLife.12401.036

To assess Notch responsiveness in melanophores, we used a synthetic Notch signaling activity detector, Tp1, consisting of multiple RBP-Jk-binding sites and a minimal promoter to drive fluorescent reporters (Parsons et al., 2009; Ninov et al., 2012). Destabilized and stabilized reporters were insufficiently sensitive to allow reliable real-time monitoring of signals upon airineme contact (not shown), yet immunohistochemistry for Tp1:H2BmCherry revealed that only ~50% of melanophores were marked by mCherry expression in wild-type fish (Figure 5B). To test whether this level of activity reflected signaling from cells of the xanthophore lineage, we examined colony stimulating factor-1 receptor a (csf1ra) homozygous mutants, which lack virtually all xanthophores and xanthoblasts (Parichy et al., 2000; Patterson and Parichy, 2013) (Figure 5—figure supplement 2), and found a marked reduction in the incidence of Tp1+ melanophores compared to wild-type siblings. To test for an airineme-dependence of such signaling, we examined airineme-inhibited fish expressing aox5:Tet:dnCdc42 and found a significant reduction in Tp1+ melanophores compared to dd-treated but non-transgenic siblings (Figure 5B).

If xanthoblast airinemes help transduce a Delta-Notch signal to melanophores important for stripe consolidation, then inhibition of Notch signaling generally, or within melanophores specifically, should cause defects in melanophore patterning similar to aox5:Tet:dnCdc42 and dlc mutant phenotypes. Indeed, global pharmacological inhibition of Notch signaling resulted in disruptions to melanophore stripe consolidation (Figure 5—figure supplement 3A). We therefore generated fish transgenic for a temporally inducible dominant negative Suppressor of Hairless [dnSu(H)] driven in the melanophore lineage with the mitfa promoter. mitfa:Tet:dnSu(H) reduced expression of Notch target gene her6 but did not affect total numbers of melanophores or xanthophores that differentiated (Figure 5—figure supplement 3B–D). Similar to effects of dnCdc42-airineme inhibition, however, melanophores persisted ectopically in the interstripe (Figure 5C,D). Concordant effects of aox5:Tet:dnCdc42 and mitfa:Tet:dnSu(H) on melanophore patterning were also observed in regenerating fin (Figure 5—figure supplement 4).

We reasoned that if airinemes contribute to Delta-Notch signaling then such interactions could be revealed by synergistic effects of aox5:Tet:dnCdc42 and mitfa:Tet:dnSu(H) transgenes. Accordingly, we induced transgenes at very low levels that were insufficient to cause a phenotype when transgenes were present individually. When transgenes were combined, however, ectopic melanophores persisted in the interstripe (Figure 5E). Conversely, fish induced to express aox5:Tet:dnCdc42 at standard levels had significantly fewer ectopic melanophores when transgenic simultaneously for mitfa:NICD1a (mean ± SE melanophores: dnCdc42, 26.4 ± 1.1; dnCdc42 with NICD1a, 9.2 ± 1.5; F1,8=84.5, p<0.001).

Because amniote melanocytes and most melanophores require the Kit (Kita) receptor tyrosine kinase for migration and survival (Tan et al., 1990; Parichy et al., 1999; Rawls and Johnson, 2003; Wehrle-Haller, 2003; Budi et al., 2011), we further speculated that airineme-dependent Notch signaling might act through Kita to promote melanophore clearance. Consistent with this idea, we found that melanophores of fish heterozygous for a kita null allele were sensitized to very low level induction of mitfa:Tet:dnSu(H) and aox5:Tet:dnCdc42 (Figure 5—figure supplement 5A,B). Abundances of kita transcript were also moderately reduced in melanophores, both in the absence of xanthophores and after inducing mitfa:Tet:dnSu(H) or aox5:Tet:dnCdc42 transgenes (Figure 5—figure supplement 5C). Indeed, melanophores of pearl danio, which migrate less than zebrafish (Quigley et al., 2005), exhibited similar reductions in Notch-responsive her6 and kita transcripts (Figure 5—figure supplement 5D).

These observations support a model in which aox5+ airinemes extended by xanthoblasts promote Notch signaling and Kita-dependent melanophore rearrangements during stripe consolidation in zebrafish.

Evolutionary changes in xanthophore differentiation alter the potential for airineme signaling

To better understand how species differences in airineme production evolved, we re-examined pearl danio (Figure 1A,C,E), in which initially dispersed melanophores fail to migrate, resulting in a nearly uniform pattern that obscures latent (but genetically detectable) stripes (Quigley et al., 2005; Mills et al., 2007). Contributing to this phenotype is an early, widespread differentiation of xanthophores that likely attenuates the positional information available to melanophores, in contrast to zebrafish, in which a biased differentiation of xanthophores at the interstripe provides a cue for melanophore patterning (Patterson et al., 2014). In light of these prior findings, and the differentiation state dependence of airinemes in zebrafish (Figure 3), we hypothesized that pearl danios exhibit fewer aox5+ airinemes owing to the precocious differentiation of their xanthophores. Consistent with this idea, experimental arrest of aox5+ cell differentiation in pearl danio (TH–) increased airineme production, albeit not to levels observed in zebrafish (Figure 6A-left).

Figure 6 with 1 supplement see all
Factors extrinsic to the aox5+ cells inhibit airineme production and signaling in pearl and models for airineme signaling and its evolution.

(A) Left, Pearl aox5+ cells extended more airinemes when differentiation-arrested (TH–; χ2=12.5, d.f.=1, p<0.0005, N=412 cells). Right, zebrafish aox5+ cells transplanted to pearl, or receiving excess Csf1 in zebrafish, extended fewer airinemes than comparably staged aox5+ cells in unmanipulated zebrafish (χ2=23.1, 22.1, d.f.=1, 1; p<0.0001, N=846 cells total). (B) In chimeras resulting from transplants of zebrafish donors (aox5:membrane-GFP, ubiquitous ubb:mCherry [Mosimann et al., 2011]) to pearl hosts, zebrafish aox5+ cells (arrow) were typically intermingled with pearl melanophores, as well as ubb+ zebrafish melanophores (arrowheads). (C) Working models for pigment cell interactions and pattern formation in zebrafish (left) and pearl danio (right). In zebrafish, xanthoblasts in stripe regions extend airinemes that signal to melanophores (Xb→M), promoting their clearance from the interstripe during stripe consolidation. Results of this study are consistent with interactions involving xanthoblast airineme dependent Delta (Dlc or possibly Dll4) activation of Notch signaling in melanophores, and the potentiation of Kita-dependent melanophore motility. Nevertheless, these data do not exclude roles for additional modes of Delta–Notch signaling, or the possibilities that airinemes transduce additional signals, or signals that cannot be distinguished from the Delta–Notch pathway using the experimental paradigms here employed. Analyses of mutant zebrafish further support roles for Cx41.8 in contributing to airineme-dependent communication, through modulation of target specificity or xanthophore lineage differentiation. In addition to xanthoblast–melanophore interactions, iridophores have attractive and repulsive effects on melanophores (I→M; I⊣M) (Frohnhofer et al., 2013; Patterson and Parichy, 2013) and express Csf1, promoting the differentiation of xanthophores (Xb→X) at the interstripe (Patterson and Parichy, 2013). Differentiated xanthophores repel melanophores (X⊣M) during normal development (Nakamasu et al., 2009) and are capable of repressing iridophore organization (X⊣I) [for details, see: (Patterson et al., 2014)]. In pearl danio, Csf1 is expressed at elevated levels by cells other than iridophores and this drives earlier and broader xanthophore differentiation than in zebrafish (Patterson et al., 2014). Precocious, widespread differentiation of xanthophores likely limits directional cues available to melanophores while simultaneously curtailing the potential for airineme signaling, as airineme competent xanthoblasts are depleted. Tissue contexts (lower panels) also show eventual death of some melanophores remaining in the interstripe in zebrafish (Parichy et al., 2000; Parichy and Turner, 2003b) and the higher overall incidence of melanophore death in pearl danio (Quigley et al., 2005); iridophores are omitted for clarity. Scale bar: 50 µm (B).

https://doi.org/10.7554/eLife.12401.037

To test if species differences in airineme production reflect evolutionary changes that are autonomous or non-autonomous to the xanthophore lineage, we transplanted cells between zebrafish and pearl danio (Parichy and Turner, 2003a; Quigley et al., 2004). Zebrafish aox5+ cells in pearl danio hosts extended airinemes at reduced frequencies, similar to that of pearl danio aox5+ cells (Figure 6A-right; Video 18). We interpret this observation, as well as aox5+ cell behaviors in reciprocal transplants, and adult pigment patterns of chimeras (Figure 6—figure supplement 1), as indicating species differences that are non-autonomous to the xanthophore lineage. These results are consistent with prior analyses that implicated cis-regulatory changes affecting environmentally produced xanthogenic Csf1 in the earlier and broader differentiation of xanthophores and altered melanophore pattern of pearl danio (Patterson et al., 2014). We therefore asked whether Csf1 alone could be responsible for species differences in airineme production. Consistent with this idea, overexpression of Csf1 in zebrafish resulted in a 22% reduction in airineme frequency within 24 hr and an 86% reduction after complete xanthophore differentiation, as in pearl (Figure 6A-right; Video 19).

Video 18
Zebrafish aox5+ cells transplanted to pearl hosts behaved like pearl aox5+ cells.

Zebrafish-derived cells expressed ubb:mCherry (magenta) and zebrafish cells of the xanthophore lineage were co-labeled with aox5:membrane-GFP (green). Airinemes were very rarely extended by zebrafish aox5+ cells in the pearl background; a single such example is shown (arrow). Also visible are ubb+ muscle cells deeper in the fish. Spots of green were autofluorescing host xanthophores, and melanophores shown in bright field were of host origin. 5 min interval, 370 min total.

https://doi.org/10.7554/eLife.12401.039
Video 19
Transgenic expression of hsp70l:Csf1a blocked airineme production in zebrafish.

Upper, an aox5+ cell extends an airineme (arrow) in a heat-shocked, non-transgenic control fish, that stabilized on an adjacent melanophore (magenta). Additional aox5+ vesicles were evident as well. Lower, in heat-shocked transgenic fish in which differentiation had been forced by Csf1a overexpression, airineme and vesicle production were not evident. 5 min interval, 810 min total.

https://doi.org/10.7554/eLife.12401.040

Together, these results support the interpretation that infrequent airineme extension, as well as pattern differences in pearl danio as compared to zebrafish, arise at least in part from evolutionary changes in factors extrinsic to the xanthophore lineage, and identify Csf1 as an especially good candidate for mediating these effects.

Discussion

Our findings suggest a revised model for how adult stripes form in zebrafish, and link evolutionary genetic modifications to morphogenetic behaviors occurring during the development of a very different, naturally occurring phenotype, the nearly uniform pattern of pearl danio (Figure 6C). More generally, our approaches have identified a novel mode of long distance cellular communication, and provide fresh insights into evolutionary changes in morphogenesis during pigment pattern development.

A particularly striking finding is the role in zebrafish stripe consolidation played by long-distance cellular projections—airinemes—extended by xanthoblasts in prospective stripe regions. Genetic analyses have shown that interactions between melanophore and xanthophore lineages are important for stripe formation and maintenance in zebrafish (Parichy et al., 2000; Maderspacher and Nusslein-Volhard, 2003; Parichy and Turner, 2003a), whereas laser ablation and in vitro studies suggest that such interactions can occur over both short- and long-range (Takahashi and Kondo, 2008; Nakamasu et al., 2009; Yamanaka and Kondo, 2014). By examining the behaviors of differentiated and undifferentiated cells in their native tissue environment using membrane-targeted fluorophores and high resolution time-lapse imaging, we identified novel interactions, as well as unanticipated differentiation-state dependencies to these interactions. Our data suggest a model in which xanthoblasts sparsely populating prospective stripe regions, but not xanthophores of the interstripe, extend Delta+ airinemes that contact melanophores, promoting Notch signaling and melanophore consolidation into stripes (Figure 6C-left). Airineme-delivered signals could provide directional information, or could simply promote motility, with directionality provided some other way. Indeed, differentiated xanthophores repel melanophores (Takahashi and Kondo, 2008; Yamanaka and Kondo, 2014), and these effects seem likely to synergize with airineme-mediated interactions to promote stripe consolidation. Additional approaches to testing this model (independent of dnCdc42 activities) should become feasible as mechanisms of airineme production and targeting are elucidated. The adult pattern defects we observed also illustrate how failures in even the subtle initial movements of adult melanophores during interstripe clearance and stripe consolidation (Parichy et al., 2000; Parichy and Turner, 2003a; Takahashi and Kondo, 2008) are amplified by the considerable somatic growth that occurs during pigment pattern formation (Parichy et al., 2009).

Our results have implications for theoretical models of pattern formation. Short-and long-range interactions between melanophore and xanthophore lineages are concordant with a Turing model (Meinhardt and Gierer, 2000; Nakamasu et al., 2009; Kondo and Miura, 2010; Watanabe and Kondo, 2015). Although such models classically envisaged diffusible substances, the same outcomes could, in principle, be achieved in many ways (Hamada et al., 2014). Our findings are likely consistent with the current formulation of a Turing model as applied to zebrafish pigment pattern. Nevertheless, our discovery that airineme production and targeting depend on precise states of differentiation reveals a previously unappreciated complexity to these signaling dynamics. Moreover, the nature of interactions seems likely to differ across stages: in adult fish, melanophore processes reach towards interstripe xanthophores, and may allow for long-range Delta–Notch trophic support (Hamada et al., 2014); yet, we did not observe long, stable melanophore processes during pattern development (e.g., Video 3). Expansions of existing theory to incorporate differentiation state heterogeneity, stage-specificity of interactions, and additional cell types would seem a valuable endeavor.

Our study adds to the known repertoire of specialized projections by which cells communicate (Miller et al., 1995; Ramirez-Weber and Kornberg, 1999; Caneparo et al., 2011; McKinney et al., 2011; Bischoff et al., 2013; Massarwa and Niswander, 2013; Sanders et al., 2013; Gradilla et al., 2014; Luz et al., 2014; Roy et al., 2014; Stanganello et al., 2015). In contrast to actin-based projections (e.g., cytonemes), airinemes required both microfilaments and microtubules, and delivered large, membrane-bound vesicles that persisted on target cells up to several cell diameters away. By analogy with neuronal dendrites, it seems plausible that airinemes participate not only in ‘forward’ signaling, shown here, but also ‘reverse’ signaling that could allow cells to sense their environment and regulate differentiation or morphogenesis accordingly (though we could not detect differences in xanthophore differentiation after increasing melanophore abundance or blocking airineme production; Figure 1—figure supplement 2B; Figure 4—figure supplement 2E). Cellular mechanisms of airinene production and targeting, as well as additional modalities for airineme signaling, are currently being investigated. It will be interesting to discover whether airinemes are used for signaling in other tissue contexts as well.

Finally, interspecific transplants and transgenic manipulations in this study provide new insights into the nearly uniform pattern of pearl danio, and together suggest a model in which enhanced Csf1 expression drives precocious, widespread xanthophore differentiation, limiting positional information available to melanophores (Patterson et al., 2014), as well as airineme production and the potential for melanophore migration and stripe consolidation (Figure 6C—right). These findings do not exclude roles for additional factors; e.g., melanophore-autonomous differences, or changes in iridophore patterning that may or may not themselves be xanthophore-dependent. Nevertheless, by focusing at high resolution on cellular behaviors our study has provided insights into the evolution of an alternative pattern state that could not have been anticipated from analyses of genetic variation and gene regulatory differences alone. These findings illustrate how an iterative approach with model organisms and closely related species can provide insights into mechanisms by which evolutionary changes in gene activity are translated through morphogenetic behaviors into species differences in form.

Materials and methods

Staging, rearing and stocks

Staging followed (Parichy et al., 2009) and fish were maintained at 28.5°C, 16:8 L:D. Zebrafish were wild-type ABwp or its derivative WT(ABb) and kitab5 (Parichy et al., 1999), csf1raj4e1 (Parichy et al., 2000), mitfaw2 (Lister et al., 1999), ltkj9s1 (Lopes et al., 2008), dlcb663 (from S. Amacher), hyperthyroid opallusb1071 (tshrD632Y), Tg(tg:nVenus-2a-nfnB)wp.rt8, Tg(aox5:palmEGFP)wp.rt22, Tg(tyrp1b:palm-mCherry)wp.rt11 (McMenamin et al., 2014),Tg(hsp70:kitga)wp.rt2, Tg(hsp70:csf1a-DrIRES-nlsCFP)wp.rt4 (Patterson and Parichy, 2013),Tg(Tp1:H2b-mCherry) (from N. Ninov, D. Stainer (Ninov et al., 2012)], Tg(aox5:TetGBD-TREtight:nVenus-2a-dnCdc42)wp.rt19 [‘aox5:Tet:dnCdc42’], Tg[mitfa:TetGBD-TREtight:nVenus-2a-dnSU(H)]wp.rt20 [‘mitfa:Tet:dnSu(H)’], Tg(mitfa:nVenus-2a-NICD)wp.rt21. Pearl danio D.aff. albolineatus (Quigley et al., 2005) were wild type, Tg(tg:nVenus-2a-nfnB)wp.at3 (McMenamin et al., 2014) and Tg(aox5:palmEGFP)wp.at4.

Transgenesis and transgenic line production

Cell-type and temporally specific expression of human dnCdc42 (from J. Wallingford) or dnSu(H) (from D. Stainier) used doxycycline and dexamethasone (dd) inducible TetON, TetGBD (Knopf et al., 2010) upstream of TREtight (Patterson and Parichy, 2013). Effectors were linked by 2a sequences to nuclear-localizing Venus (nVenus) and driven by 8 kb aox5 and 2.2 kb mitfa promoters in xanthophore and melanophore lineages. Inductions were performed on F0 mosaic fish and non-mosaic stable lines with similar results. Constitutive Notch activation used the intracellular domain of notch1a linked by 2a sequence to nVenus, and driven by the mitfa promoter. To visualize DeltaC localization under native regulatory elements, we inserted mCherry C-terminally using BAC CH73-208H7 with 34 kb 5’ and 69 kb 3’ to the open reading frame; DlC-mCherry functionality was verified in mosaic F0 fish by rescue of an embryonic segmentation defect in 47% of dlc loss of function embryos. Actin was examined with mCherry fused to the utrophin calponin homology domain [Addgene #26740; (Burkel et al., 2007)] or LifeAct-mKate [from M. Barna]. Microtubule labeling used subclones from Tuba1-mCherry [(Addgene #49149; (Friedman et al., 2010)] or EB3-EGFP (Michael Davidson, Addgene #56431).

Drug treatments and heat shock experiments

Fish were treated with drugs during dark cycles then changed to water without drugs at light cycles. TetGBD transgenics or non-transgenic siblings were given 25 nM doxycycline and 50 µM dexamethasone (dd; Sigma-Aldrich, St. Louis, MO) from 6.5 SSL. Tests for interactions with transgenes used 6 nM and 12.5 µM dd. Notch inhibitor LY411575 (3 µM) and Cdc42 inhibitor ML141 (1 µM) were prepared in DMSO and fish treated with drug or vehicle from 6.0 SSL. For acute drug administrations, ML141 (2 µM), blebbistatin (10 µM) or nocodazole (3 µg/ml) were added to ex vivoculture medium (below) 30 min prior to time-lapse imaging. TH– fish were generated by ablating thyroid follicles of Tg(tg:nVenus-2a-nfnB) at 4 d (McMenamin et al., 2014). For heat shock, Tg(hsp70:kitlga) and Tg(hsp70:csf1a-DrIRES-nCFP) larvae were exposed 1 hr to 38°C twice daily between 6.5–7.0 SSL and 7.5 SSL.

Time-lapse and still imaging

Ex vivo imaging of pigment cells in their native tissue environment followed (Budi et al., 2011; Eom et al., 2012), with images acquired at 5-min intervals for 18 hr at 10x using an Evolve (Photometrics, Tucson, AZ) camera mounted on a Zeiss Observer Z1 inverted microscope with CSU-X1 spinning disk (Yokogawa, Tokyo, Japan). Bright-field images were taken before and after imaging. Larvae were 7.5 SSL except where indicated. For analyses of aox5+ airinemes across stages, genetic backgrounds or both, labeled cells were examined in 79 wild-type zebrafish and 40 wild-type pearl; 4 opallus, 4 ltk, 10 mitfa, and 21 cx41.8 mutants; 11 Tg(hsp70:csf1a-DrIRES-nlsCFP), 4 Tg(hsp70:kitlga), 5 Tg(aox5:Tet:dnCdc42), 7 Tg(tg:nVenus-2a-nfnB) zebrafish and 13 Tg(tg:nVenus-2a-nfnB) pearl; 8 ML141-treated, 4 blebbistatin-treated, 4 nocodazole-treated, 3 DMSO control zebrafish. Analyses of tyrp1b+ airineme incidence used 56 wild-type zebrafish. Images in Figures 1B, 5D and Video 6 were taken with Zeiss (Jena, Germany) LSM880 or LSM800 scanning laser confocal microscopes with Airyscan detectors.

Cell counts and distributions

Pigment granules were contracted with epinephrine to facilitate counts and assessment of cell centroids using the Cell Counter plugin of ImageJ (NIH). To assess melanophore pattern, each cell was assigned to one of 20 bins from dorsal (0.0) to ventral (1.0), and total numbers of melanophores within each bin were averaged across individuals. Positions of interstripes were indicated by melanophore-free troughs in melanophore distributions of controls (positions 0.40, 0.45, 0.50); numbers of melanophores at these locations were compared for assessment of interstripe melanophore persistence.

RT-PCR

For all quantitative RT-PCR analyses, melanophores were isolated in 2.5 mg/ml trypsin in phosphate buffered saline (PBS) for 10 min at 36°C. Tissues were rinsed with PBS, and incubated at 28°C in 1 mg/ml collagenase I, 0.1 mg/ml DNase I, 0.1 mg/ml trypsin inhibitor in PBS while shaking. Suspension solutions were filtered, centrifuged 30x g for 10 min at 4°C in 50% Percoll density gradients to precipitate melanophores, assessed for purity at high magnification, then extracted for RNA with an RNAqueous-Micro kit (Thermo Fisher, Waltham, MA). cDNA was synthesized with the iScript cDNA Synthesis Kit (Bio-Rad, Hercules, CA). Quantitative PCR was performed on an Applied Biosystems StepOne Plus using custom TaqMan probes and primers spanning intron–exon boundaries and targeting regions of sequence identity between zebrafish and pearl (her6, AIRSA5P; kita, AILJKAN; rpl13a, AI5H3GZ). Quantitative PCRs were run with at least triplicate biological and technical replication.

For non-quantitative RT-PCR, 100–150 melanophores or xanthophores were picked by micro-manipulator from dissociated cells. Amplifications were 40 cycles (actb1, dlc, csf1ra, pmela, cx41.8, cx39.4) or 45 cycles (dll4) at 94°C, 30 s; 62°C, 20 s; 72°C, 20 s. Absence of cross-contamination was verified by expression of pmela or csf1r, specific to melanophores and xanthophores, respectively (Kelsh et al., 2000; Parichy et al., 2000). actb1: 5′- ACTGGGATGACATGGAGAAGAT3′, 5′- GTGTTGAAGGTCTCGAACATGA-3′; csf1r: 5′-CAGAGTGACGTCTGGTCTTACG-3′, 5′-GGACATCTGATAGCCACACTTG-3′; cx39.4: 5′- GACAGTCTTCCAAGCTACTCAA-3′, 5′-GGTGCTCTGCTTCTCAAACATA-3′; cx41.8: 5′-ACATCCGGTCAACTGCTACAT-3′, 5′- TTGTATCCGTGCACATACTTCC-3′; dlc: 5’-CGGGAATCGTCTCTTTGATAAT-3′, 5′-CTCACCGATAGCGAGTCTTCTT-3′; dll4: 5’- GCACTCACCTTACTCGGATCTA-3′, 5′-CACTTTGAACATCCTGAGACCA-3′ pmela: 5’-CTCGGAGTTCTGTTTTTCGTTT3′, 5′- AAGGTACTGCGCTTATTCCTGA-3′.

Immunohistochemistry

To verify transgene expression, 7.5-SSL larvae were fixed in 4% paraformaldehyde, embedded in OCT (Tissue-Tek, VWR, Radnor, PA), and cryosectioned at 20 µm. Sections were washed with 0.3% Triton X-100 in PBS (PBSTX), blocked, then incubated at 4°C overnight with primary antibodies. Primary antibodies were mouse anti-PAX7 (Developmental Studies Hybridoma Bank) for the xanthophore lineage (Minchin and Hughes, 2008) and rabbit anti-GFP (ThermoFisher) for Venus. Notch signaling reporter Tg(TP1:H2B-mCherry) was visualized with rat anti-mCherry (ThermoFisher).

Cell transplantation

Chimeric fish were generated by transplanting cells at blastula stages then rearing embryos reared through adult pigment pattern formation (Parichy and Turner, 2003a; Quigley et al., 2004) using fish that were transgenic for ubiquitous ubb:mCherry or actb1:EGFP, xanthophore-lineage aox5:membrane-GFP, or both.

Statistical analyses

Analyses were performed with JMP 8.0 (SAS Institute, Cary, NC). Frequency data for behaviors of individual cells or projections were assessed by single or multiple factor maximum likelihood. Continuous data were evaluated by t-test or analyses of variance, using ln-transformation in some instances to correct residuals to normality and homoscedasticity. Post hoc means were compared by Tukey-Kramer HSD.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
    The Iliad of Homer
    1. Homer
    (2011)
    ~8th cent. BCE, The Iliad of Homer, Lattimore R, Translator, Press Chicago University.
  26. 26
    Sex, Color, and Mate Choice in Guppies
    1. AE Houde
    (1997)
    Princeton, NJ: Princeton University Press.
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
    Untersuchungen über das farbmuster der zebrabarbe brachydanio rerio (Cyprinidae, Teleostei)
    1. F Kirschbaum
    (1975)
    Wilhelm Roux's Arch 177:129–152.
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
    Nacre encodes a zebrafish microphthalmia-related protein that regulates neural-crest-derived pigment cell fate
    1. JA Lister
    2. CP Robertson
    3. T Lepage
    4. SL Johnson
    5. DW Raible
    (1999)
    Development 126:3757–3767.
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
    Dynamics of thin filopodia during sea urchin gastrulation
    1. J Miller
    2. SE Fraser
    3. D McClay
    (1995)
    Development 121:2501–2511.
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
    An orthologue of the kit-related gene fms is required for development of neural crest-derived xanthophores and a subpopulation of adult melanocytes in the zebrafish, danio rerio
    1. DM Parichy
    2. DG Ransom
    3. B Paw
    4. LI Zon
    5. SL Johnson
    (2000)
    Development 127:3031–3044.
  58. 58
    Zebrafish sparse corresponds to an orthologue of c-kit and is required for the morphogenesis of a subpopulation of melanocytes, but is not essential for hematopoiesis or primordial germ cell development
    1. DM Parichy
    2. JF Rawls
    3. SJ Pratt
    4. TT Whitfield
    5. SL Johnson
    (1999)
    Development 126:3425–3436.
  59. 59
  60. 60
  61. 61
  62. 62
    Advancing biology through a deeper understanding of zebrafish ecology and evolution
    1. DM Parichy
    (2015)
    eLife, 4, 10.7554/eLife.05635.
  63. 63
  64. 64
  65. 65
  66. 66
  67. 67
  68. 68
  69. 69
  70. 70
  71. 71
    Zebrafish kit mutation reveals primary and secondary regulation of melanocyte development during fin stripe regeneration
    1. JF Rawls
    2. SL Johnson
    (2000)
    Development 127:3715–3724.
  72. 72
  73. 73
  74. 74
  75. 75
  76. 76
  77. 77
  78. 78
  79. 79
  80. 80
  81. 81
  82. 82
  83. 83
  84. 84
  85. 85
    Probe Reports From the NIH Molecular Libraries Program
    1. Z Surviladze
    2. A Waller
    3. JJ Strouse
    4. C Bologa
    5. O Ursu
    (2010)
     a potent and selective inhibitor of Cdc42 GTPase, Probe Reports From the NIH Molecular Libraries Program, Bethesda, MD.
  86. 86
  87. 87
  88. 88
  89. 89
    Search for basonuclin target genes
    1. J Wang
    2. S Zhang
    3. RM Schultz
    4. H Tseng
    (2006)
    Biochemical and Biophysical Research Communications 348:1261–1271.
    https://doi.org/10.1016/j.bbrc.2006.07.198
  90. 90
  91. 91
  92. 92
  93. 93
  94. 94

Decision letter

  1. Robb Krumlauf
    Reviewing Editor; Stowers Institute for Medical Research, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your work entitled "Long-distance communication by specialized cellular projections during pigment pattern development and evolution" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Diethard Tautz as the Senior Editor.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

The consensus view of the reviewers is that this study presents a series of genetic and experimental perturbations in zebrafish and pearl danio that provides new insight into the morphogenetic and developmental basis of periodic color pattern. They provide dynamic live imaging of pigment cells in developing zebrafish, leading to the discovery that (mostly) xanthophores exhibit an unusual type of cytoplasmic projection that appears to be involved in stripe consolidation. The authors characterize these projections, which they term airinemes, with regard to their development and composition. Transgenic experiments provide evidence that airinemes are an essential component of a xanthoblast to melanophore signal that promotes melanophore movement and stripe consolidation possibly via a δ-Notch signal. There are also associated changes in melanophore Kit expression. They also show that the pearl danio, which lacks melanophore stripes, also lacks airinemes.

Overall, the work is carefully and critically described, and will be of widespread interest and significance to developmental and evolutionary biologists. However, there is a general concern that the presentation or description in the text needs to be improved with regard to strength of evidence relative to interpretation of the data. The data in the first half of the paper is excellent but the second part of the work is suggestive not proven. This issue can primarily be addressed by changes to the presentation, and, possibly, an additional straightforward experiment.

To aid the authors in considering how to revise the text we are communicating the major points of individual reviewer opinions below.

1) The first half of the manuscript (identification, cell biologic and developmental characterization of airinemes) is strong, but the second half (Delta-notch, kita expression, differentiation dependence, xanthophore non-autonomy) is less so. The Delta-Notch experiments are an important addition to the manuscript, but the conclusions regarding a requirement for Delta-Notch signaling are not as strong as the conclusions regarding a requirement for airinemes. The existing evidence is enough to speculate about the involvement of Notch but in the absence of additional data it is essential that changes are made accordingly to the Discussion and the model in Figure 8 that take into account the relative strength of the conclusions.

a) The DLC-mCherry fusion protein appears to be found throughout xanthophores (cell body, projections, vesicles), which is not quite the smoking gun that one might have hoped for. What about Dll4? (Is there any genetic or molecular biologic information that makes dll4 a less likely candidate?) The Notch reporter experiment is interesting but it would be helpful to have some more context. Is the reporter expressed in any other pigment cells? Does the melanophore signal described by the authors stand out relative to other evidence of Notch signaling during pigment cell migration and stripe consolidation. Finally, while the result of the double mutant (dnCdc42; dnSu(H); low levels of dd) is interesting, it doesn't quite reach the level of rigor one would hope for in an epistasis analysis. A stronger experiment would be to show that increased Notch signaling in melanophores could rescue the phenotype of a xanthophore dnCdc42 transgene. Showing that constitutive notch (mitfa>NICD) suppresses aox5>dnCDC42 phenotype of interstripe melanocyte would be more direct (and more rigorous) than the experiments showing synergistic genetic interaction between notch pathway (dnSuH) in mels and CDC42 promoted airinemes in xans. Together, these two experiments would make the point of airinemes acting through Notch rather compelling. This seems like a fairly simple experiment with the available fish stocks already used in study.

b) The melanophore kita expression results are suggestive but not developed. Since the phenotype is a failure of mel migration, and kit is required for migration, it is justified to explore the role of kit in the phenotype. The results shown – stronger phenotype of genetic interactions (dnSuH, dnCDC42) in kit(-/+) than in kit(+/+), together with the evidence that Notch signalling controls kit expression is sufficient to tell us there is a role. The key question is to look for how kit is responsive to Notch signalling. As it stands in the paper kita is a candidate gene/pathway, the level of reduction is not very substantial, and it's not clear exactly how melanophores were purified/enriched (the Methods describes both Percoll purification and micromanipulation) or the degree of enrichment. If RNA-seq information is available on purified melanophores with and without airineme contact, that would add a lot. This data is not absolutely necessary for the manuscript to move forward but conclusions need to at least be toned down.

c) The TH and the transplantation experiments are similar to the Delta-Notch experiments – the results are interesting but not compelling.

2) It is surprising that this type of cell protrusion has not been described before. The authors themselves, and many other groups as well, have used live imaging of Xanthophores but never reported such a phenomena. The authors should comment in the paper why this is the case. Have they been simply overlooked so far or are technical advances now available that have led to this discovery?

3) How sure can the authors be that these structures are indeed different from previously identified cytonemes. There are cases of cytonemes that stain positive for tubulin. The paper also exaggerates a bit the absolute and unique roles of these "airinemes" in the text, while a closer look into the graphs and figures shows that such absolute statements are not correct. For example: "Thus, exuberant production of long, fast, and vesicle-containing projections was UNIQUE to the zebrafish xanthophore lineage…" Figure 1 however, shows protrusions both in melanophores and in danio pearl xanthophores – simply in lower numbers. Also, the difference seems to be restricted to stage 7-8. While this obviously could be enough to have an effect, a more veridic description of the observations throughout the text would be helpful.

https://doi.org/10.7554/eLife.12401.041

Author response

To aid the authors in considering how to revise the text we are communicating the major points of individual reviewer opinions below.

1) The first half of the manuscript (identification, cell biologic and developmental characterization of airinemes) is strong, but the second half (Delta-notch, kita expression, differentiation dependence, xanthophore non-autonomy) is less so. The Delta-Notch experiments are an important addition to the manuscript, but the conclusions regarding a requirement for Delta-Notch signaling are not as strong as the conclusions regarding a requirement for airinemes. The existing evidence is enough to speculate about the involvement of Notch but in the absence of additional data it is essential that changes are made accordingly to the Discussion and the model in Figure 8 that take into account the relative strength of the conclusions.

Thank you for raising this issue. We do believe that our several analyses, including a significant reduction in Notch signaling reporter activity in melanophores that results from airineme inhibition, as well as the phenotypes of the dlc mutant and several transgenic lines singly and in combination with one another, together support the notion that Delta-Notch signaling is important for stripe patterning and that airinemes very likely contribute to the transduction of such signals. Nevertheless, our original presentation was not intended to imply that airinemes are the only mechanism by which Delta–Notch signaling occurs between melanophore and xanthophore lineages (e.g., our own prior analyses in collaboration with Shigeru Kondo’s group in Hamada et al. 2014) or that airinemes exclusively transduce such a signal. The working model we presented and its original description were merely intended to represent our best current understanding of the system according to the evidence currently available. To minimize any potential for misunderstanding, we have made these points more explicit in the discussion and legend of Figure 6C (corresponding to original Figure 8).

a) The DLC-mCherry fusion protein appears to be found throughout xanthophores (cell body, projections, vesicles), which is not quite the smoking gun that one might have hoped for.

Although Dlc-mCherry is present widely over xanthoblast cell membranes, we do observe especially abundant concentrations associated with the blebs from which airinemes originate and we have replaced the original images to better illustrate this point. Nevertheless we did not intend to imply that Dlc becomes specifically concentrated at blebs or in airineme vesicles, merely that such structures could harbor the protein if it happens to already be at the cell surface, and our observations support this view. Indeed, we expect there are many ways in which melanophores could “obtain” a Delta signal from the xanthophore lineage, including the extension of processes from melanophores to xanthophore cell bodies as appears to happen at later stages (Hamada et al. 2014). Although we are interested in whether specific proteins are actively localized at sites of airineme emergence, and we are engaged in work to address this point, we feel this issue is beyond the scope of the present study as our interpretations do not require such a mechanism.

What about Dll4? (Is there any genetic or molecular biologic information that makes dll4 a less likely candidate?)

We suspect Dll4 could well be involved, too. Unfortunately, we have not been able to find a BAC clone containing Dll4 that is suitable for generating a fusion protein to test its localization in the context of native regulatory elements; so we have yet to pursue this candidate further. Nevertheless, our original intent with analyses of Dlc was merely to illustrate that at least one partner in Delta–Notch signaling can be found in airineme extensions and it was for this reason that we referred to potential interactions as involving generic “Delta” signals in the remainder of the manuscript, rather than specific “Dlc” signals. We have noted in the text that technical limitations have hampered analyses of Dll4.

The Notch reporter experiment is interesting but it would be helpful to have some more context. Is the reporter expressed in any other pigment cells? Does the melanophore signal described by the authors stand out relative to other evidence of Notch signaling during pigment cell migration and stripe consolidation.

We added more information about Notch reporter expression in the legend to Figure 5B and Notch pathway gene expression in the legend to Figure 5—figure supplement 1A. In short, we observe plenty of Notch gene expression and Notch signaling in other tissues, as would be expected given the diversity of Delta–Notch functions during development, but in the hypodermis we detect Tp1 only in melanophores. Ours is the first study to explicitly address levels of Notch signaling within melanophores so we do not know if the levels we observe are greater or less than what might be present at other stages of development.

Finally, while the result of the double mutant (dnCdc42; dnSu(H); low levels of dd) is interesting, it doesn't quite reach the level of rigor one would hope for in an epistasis analysis. A stronger experiment would be to show that increased Notch signaling in melanophores could rescue the phenotype of a xanthophore dnCdc42 transgene. Showing that constitutive notch (mitfa>NICD) suppresses aox5>dnCDC42 phenotype of interstripe melanocyte would be more direct (and more rigorous) than the experiments showing synergistic genetic interaction between notch pathway (dnSuH) in mels and CDC42 promoted airinemes in xans. Together, these two experiments would make the point of airinemes acting through Notch rather compelling. This seems like a fairly simple experiment with the available fish stocks already used in study.

We added these data on aox5:dnCdc42 phenotypes with and without mitfa:NICD in the background; a significant reduction in ectopic melanophores when Notch activity is restored autonomously to melanophores supports our original interpretation.

b) The melanophore kita expression results are suggestive but not developed. Since the phenotype is a failure of mel migration, and kit is required for migration, it is justified to explore the role of kit in the phenotype. The results shownstronger phenotype of genetic interactions (dnSuH, dnCDC42) in kit(-/+) than in kit(+/+), together with the evidence that Notch signalling controls kit expression is sufficient to tell us there is a role. The key question is to look for how kit is responsive to Notch signalling. As it stands in the paper kita is a candidate gene/pathway, the level of reduction is not very substantial, and it's not clear exactly how melanophores were purified/enriched (the Methods describes both Percoll purification and micromanipulation) or the degree of enrichment. If RNA-seq information is available on purified melanophores with and without airineme contact, that would add a lot. This data is not absolutely necessary for the manuscript to move forward but conclusions need to at least be toned down.

Our intent with analyses of kita was simply to assess whether the best known pathway promoting melanophore motility might be affected by alterations in airineme behaviors and Notch signaling. On-going studies are working to assess transcriptomic and other changes associated with airineme signaling but these analyses will require additional months or years of effort. In the meantime we have clarified that alterations in kita expression are moderate, as we would anticipate given that many factors likely contribute to activity of this pathway. We have also moved analyses of kita to supplementary materials to better reflect our own view of their relative import.

In the Methods we have further clarified that Percoll gradients were used for qPCR and cell picking by micromanipulator for non-quantitive RT-PCR. Purities of samples were assessed by microscopic examination at high magnification. Though we cannot absolutely exclude the possibility of some contaminating cells, the genes we have studied are already known to be expressed by melanophores from published analyses (and our own unpublished RNA-Seq studies on small numbers of individually picked cells) and we anticipate that, given the cell-type specificity of our manipulations, contaminating cells that might express the same genes would tend to reduce the sensitivity of our assays, biasing our results towards lower rather than higher levels of significance.

c) The TH and the transplantation experiments are similar to the Delta-Notch experiments – the results are interesting but not compelling.

Our only intent with TH and transplantation experiments were to use the former as a way to manipulate xanthophore differentiation and the latter to assess the cell autonomy of species differences in airineme production and pigment pattern. The results of these analyses were straightforward so it is not clear precisely how we would make them more compelling. We are indeed studying roles for TH in the xanthophore lineage more generally, and we are looking into additional factors that influence pattern differences between species, but these seem beyond the scope of the present study. Given concerns about these experiments, however, we have streamlined presentation of these analyses to ensure they are as concise as possible.

3) It is surprising that this type of cell protrusion has not been described before. The authors themselves, and many other groups as well, have used live imaging of Xanthophores but never reported such a phenomena. The authors should comment in the paper why this is the case. Have they been simply overlooked so far or are technical advances now available that have led to this discovery?

Our own previous live imaging focused on the behaviors of melanophore precursors using cytosolic reporters and fluorescence wide-field time-lapse imaging; other analyses have focused on behaviors of fully differentiated pigment cells using bright field imaging. To our knowledge, this is the first time that membrane targeted fluorophores and laser spinning disk imaging have been used to visualize dynamics of danio pigment cells and their precursors in the tissue environment. We have added a note to this effect in the Discussion.

4) How sure can the authors be that these structures are indeed different from previously identified cytonemes. There are cases of cytonemes that stain positive for tubulin.

Our reading of the literature suggests that projections of this size, with meandering trajectories, an actin and tubulin based cytoskeleton and very large vesicles that are delivered to other cells, are unique, or at the very least not widely known or previously named. We have also made informal enquiries with others studying long-range cellular communication, as well as prominent neuroanatomists; none have indicated that these projections resemble ones they have observed or heard about. Indeed, objections have been raised to our past use of the term ‘cytoneme’ in describing these projections in the context of seminar presentations, etc., precisely for the reasons listed above. We have therefore found it convenient, and hopefully less confusing, to name the structures rather than referring to them as “cytonemes”, “fast pigment cell projections”, “vesicle containing processes” etc. Regarding etymology, we could not pass up a once-in-a-career opportunity to reference Iris and The Iliad, and to validate the importance of a liberal arts education. We considered naming the processes “irinemes,” but we were concerned this might evoke “iridophores”, which are the one class of pigment cell that seems, so far, to lack them. Nevertheless, we believe the reference to fast message delivery is appropriate given our findings (of course “-neme” or “thread” honors the exciting biology of cytonemes and their discoverers). Finally, upon weighing the merits of several alternatives, we opted to hopefully avoid confusion with iridophores by modifying the name to “airineme,” which has the added advantage of honoring Sir George Airy’s contributions to optics.

The paper also exaggerates a bit the absolute and unique roles of these "airinemes" in the text, while a closer look into the graphs and figures shows that such absolute statements are not correct. For example: "Thus, exuberant production of long, fast, and vesicle-containing projections was UNIQUE to the zebrafish xanthophore lineage" Figure 1 however, shows protrusions both in melanophores and in danio pearl xanthophoressimply in lower numbers. Also, the difference seems to be restricted to stage 7-8. While this obviously could be enough to have an effect, a more veridic description of the observations throughout the text would be helpful.

Regarding our descriptions we have always taken great care to accurately portray the data presented in all plots and images, as well as the totality of our observations overall. Although the veracity of our descriptions was not intentionally lacking, we acknowledge that some statements may have been open to multiple interpretations. We have therefore made the following changes:

Original: “Thus, exuberant production of long, fast, and vesicle-containing projections was unique to the zebrafish xanthophore lineage and occurred during stages of stripe formation.”

Revised: “Thus, long, fast, and vesicle-containing projections were produced exuberantly by cells of the zebrafish xanthophore lineage during stages of stripe formation, but were not common to other pigment cell classes in this species or to cells of the xanthophore lineage in pearl danio.”

Original: “To test if airineme production depended on differentiation state we exploited the thyroid hormone (TH) dependence of xanthophore differentiation”

Revised: “To test experimentally if airineme production was contingent upon differentiation state, we exploited the thyroid hormone (TH) dependence of xanthophore differentiation”

Original: “kita transcript abundances were likewise reduced in melanophores upon inhibition of Notch signaling [mitfa:Tet:dnSu(H)] or airineme production [aox5:Tet:dnCdc42] (Figure 6C).”

Revised: “Abundances of kita transcript were also moderately reduced in melanophores, both in the absence of xanthophores and after inducing mitfa:Tet:dnSu(H) or aox5:Tet:dnCdc42 transgenes (Figure 5—figure supplement 5C).”

Original: “stripes in zebrafish require a novel class of thin, fast cellular projection to transduce a Delta-Notch signal over long distances”

Revised: “stripes in zebrafish require a novel class of thin, fast cellular projection to promote transduction of a Delta-Notch signal over long distances”

Original: “We found that xanthoblasts sparsely populating prospective stripe regions, but not xanthophores of the interstripe, extend Delta+ airinemes that contact melanophores to activate Notch signaling, upregulate kita expression and promote consolidation into stripes (Figure 8-left).”

Revised: “Our data suggest a model in which xanthoblasts sparsely populating prospective stripe regions, but not xanthophores of the interstripe, extend Delta+ airinemes that contact melanophores, promoting Notch signaling and melanophore consolidation into stripes (Figure 6C-left).”

https://doi.org/10.7554/eLife.12401.042

Article and author information

Author details

  1. Dae Seok Eom

    Department of Biology, University of Washington, Seattle, United States
    Contribution
    DSE, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    The authors declare that no competing interests exist.
  2. Emily J Bain

    Department of Biology, University of Washington, Seattle, United States
    Contribution
    EJB, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0428-4357
  3. Larissa B Patterson

    Department of Biology, University of Washington, Seattle, United States
    Contribution
    LBP, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    The authors declare that no competing interests exist.
  4. Megan E Grout

    Department of Biology, University of Washington, Seattle, United States
    Contribution
    MEG, Acquisition of data
    Competing interests
    The authors declare that no competing interests exist.
  5. David M Parichy

    1. Department of Biology, University of Washington, Seattle, United States
    2. Institute for Stem Cell and Regenerative Medicine, University of Washington, Seattle, United States
    Contribution
    DMP, Conception and design, Analysis and interpretation of data, Drafting or revising the article
    For correspondence
    dparichy@uw.edu
    Competing interests
    The authors declare that no competing interests exist.

Funding

National Institute of General Medical Sciences (NIH R01 GM096906)

  • David M Parichy

National Institute of General Medical Sciences (NIH R01 GM111233)

  • David M Parichy

National Institute of Child Health and Human Development (T32 HD007183)

  • Emily J Bain

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

For assistance or discussions we thank Inseok Hwang, Jiae Lee, Tracy Larson, Thao Pham, and Jay Parrish.

Ethics

Animal experimentation: This study was performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All animals were handled according to approved institutional animal care and use committee (IACUC) protocol (#4094) of the University of Washington.

Reviewing Editor

  1. Robb Krumlauf, Stowers Institute for Medical Research, United States

Publication history

  1. Received: October 18, 2015
  2. Accepted: December 23, 2015
  3. Accepted Manuscript published: December 23, 2015 (version 1)
  4. Version of Record published: February 15, 2016 (version 2)

Copyright

© 2015, Eom et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 3,061
    Page views
  • 731
    Downloads
  • 27
    Citations

Article citation count generated by polling the highest count across the following sources: Scopus, Crossref, PubMed Central.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

  1. Further reading

Further reading

    1. Cell Biology
    2. Developmental Biology
    Harri Elamaa et al.
    Research Article
    1. Cell Biology
    2. Developmental Biology
    Ehsan Sadeghipour et al.
    Research Article