Insights & Perspectives hn again Think
Paracrine signaling mediated at cell cell contacts
Sougata Roy*,† and Thomas B. Kornberg
Recent findings in several organ systems show that cytoneme-mediated systems. However, recent work that we signaling transports signaling proteins along cellular extensions and targets discuss here describes paracrine signal- cell-to-cell exchanges to synaptic contacts. This mechanism of paracrine ing that is instead contact-mediated and signaling may be a general one that is used by many (or all) cell types in many (or dependent on transient synapses that all) organs. We briefly review these findings in this perspective. We also non-neuronal cells make. These synap- describe the properties of several signaling systems that have previously been ses form at sites where specialized signaling filopodia called cytonemes interpreted to support a passive diffusion mechanism of signaling protein extend to contact target cells. dispersion, but can now be understood in the context of the cytoneme mechanism.
Keywords: The classical model of .cytonemes; filopodia; morphogen; paracrine signaling; synapse; TGF-b paracrine signaling assumes that signals disperse by passive Introduction so that signals are within only 15 20 nm diffusion of their target receptors when they are Animal cells communicate over long released. Paracrine signaling, the third There are many paracrine signaling distances in various ways. Endocrine general mechanism, may be considered proteins that have been characterized. cells signal systemically by releasing to be a variant of endocrine signaling, They include the Fibroblast Growth hormones that disseminate in the vas- functioning at relatively short range Factors (FGFs) and other proteins that culature. Neurons also release signals, when secreted signals move limited activate Receptor Tyrosine Kinases, but they exchange information at syn- distances by passive diffusion in extra- TGF-b family members, Wnt proteins, apses that form where their axons and cellular fluid. The evidence that has Hedgehog (Hh) proteins, chemokines, dendrites contact target cells. Some supported this mechanism of paracrine as well as cytokines and other ligands axons extend over distances of several signaling has been obtained over many that activate the Jak-STAT pathway. All meters, effectively bridging distant cells years in many different experimental act at a distance, presumably by binding to receptors on target cells after diffus- ing from the producing cells that secrete DOI 10.1002/bies.201400122 them. In many cases, their distributions and patterns of signaling activation Cardiovascular Research Institute, University of extracellular matrix; EGF, epidermal growth indicate that they have spread out from California, San Francisco, CA, USA factor; FGF, fibroblast growth factor; FRAP, source cells, generating concentration fluorescence recovery after photobleaching; *Corresponding author: GAG, glycosaminoglycan; GFP, green fluores- gradients that decline with increasing Sougata Roy cent protein; GRASP, GFP reconstitution across distance, and cells that are the targets of E-mail: [email protected] synaptic partners; Hh, hedgehog; HSPG, heparan the signals respond in a concentration- sulfate proteoglycan; Ihog, interference hedge- dependent manner. The question we †Current Address: Department of Cell Biology and hog; LAP, latency-associated peptide; LTBP, Molecular Genetics, University of Maryland, latent TGF-b binding protein; Ptc, patched; address is not whether developmental College Park, MD, USA RTK, receptor tyrosine kinase; Shh, sonic hedge- fields have concentration gradients of hog; TGF-b, Transforming growth factor-b; Tkv, signaling proteins that regulate growth Abbreviations: thickveins. ASP, air sac primordium; A/P, anterior/posterior; and patterning, but how these proteins Btl, breathless; Dpp, decapentaplegic; ECM, disperse to distant cells.
Bioessays 37: 25–33, ß 2014 WILEY Periodicals, Inc. www.bioessays-journal.com 25 S. Roy and T. B. Kornberg Insights & Perspectives.....
The first report of non-autonomous reflected in the many names that have on the columnar cells, concentrating instructive signaling described a region been coined (e.g. thick filopodia, thin mostly on the region that produces the of the amphibian gastrula (the blastopore filopodia, growth cone filopodia, den- wing blade. dorsal lip) that induced gastrulation and dritic spines, invadopodia, gliapodia, Columnar cells in the wing blade embryogenesis after transfer to ectopic and myopodia). Cytonemes are the primordium that are outside of the Dpp locationsinrecipientembryos[1].This specialized filopodia that have been signaling center and respond to Dpp inductive capacity was not a general shown to traffic signaling proteins such extend cytonemes that orient toward property of embryo cells, suggesting that as morphogens, growth factors and cell the signaling center (Fig. 1A), and the the cells at the blastopore lip are specially determination factors. They have been properties of these cytonemes are con- endowed to make something (an “induc- observed in both vertebrate and inverte- sistent with the idea that they bridge the
Think again er”) that instructs and patterns outlying brate systems and have been character- distance between Dpp producing and cells. This group of cells has been ized most extensively in Drosophila receiving cells, and move Dpp between branded a “developmental organizer”, larval tissues [4]. them [3, 20 22]. The presence and and the simplest mechanism that has In the Drosophila larval wing imagi- orientation of these cytonemes depend been considered is the synthesis and nal disc, cells in several signaling on active Dpp signaling under con- secretion of an inducer molecule that centers produce signaling proteins that ditions of reduced Dpp expression disseminates by diffusion. The search for activate signal transduction in cells that (dppts at restrictive temperature), their such inducers took many forms and express receptors for these signaling number declines and they appear disor- succeeded only after many decades. proteins. One of the signaling centers dered and lack consistent orientation. Although the inducers were assumed to expresses the TGF-b family member They increase in number and appear to be small organic molecules that diffuse Decapentaplegic (Dpp); its Type 1 and have random orientations under condi- rapidly and pass efficiently from cell to Type 2 receptor subunits, Thickveins tional activation of ubiquitous, uniform cell, the known inducers are proteins (Tkv) and Punt, are expressed by all Dpp over-expression (Heat shock in- such as TGF-b,FGF,WntandHh. wing disc cells (reviewed in [9]). The duced expression of Dpp), and they Mathematical analyses and experimental Dpp signaling center is a 8 10 cell-wide orient toward somatic clones that over- studies show that diffusion may, within row that flanks the anterior side of the express Dpp ectopically. There is also certain parameters of diffusion coeffi- anterior/posterior (A/P) compartment evidence that they make direct contact cient, geometric tortuosity, viscosity, and border (Fig. 1A). The A/P border bisects and synapse with the cells in the Dpp extracellular volume, generate the ob- the wing disc along the entire dorso- signaling center. GFP reconstitution served gradients in the times available ventral axis. Cells in the Dpp signaling across synaptic partners (GRASP) is a during development (reviewed in [2]). center both produce and respond to technique that was developed to identi- The discovery of filopodia that serve as Dpp, although it is not known whether fy stable cell cell contacts that juxta- conduits that move signaling proteins their signal transduction responses pose cell membranes at distances between source and target cells adds a are due to autocrine, juxtacrine, or 20 nm, such as the synapses that new dimension and a radically different paracrine signaling. Outside of the neurons make. GRASP employs two mechanism for dissemination ([3] and signaling center and as far away as extracellular, membrane-tethered frag- reviewed in [4]). the disc flanks, cells also respond to ments of GFP that can self-assemble to Dpp, and evidence that Dpp moves generate fluorescent protein. The points from the cells that express it in order to of GRASP fluorescence at the Dpp Cytonemes mediate bind to Tkv on distant cells include signaling center in discs that express both GFP fluorescence and antibody one of the complementing GFP frag- paracrine signaling by staining in discs that overexpress Dpp: ments in the dpp domain and the other bridging signaling cells GFPinthesignalingcenter[10 14]. in the disc flanks (Fig. 1B) indicates that Dpp distributes in concentration gra- cells in these distant locations make Filopodia are thin cellular extensions dients that decline with increasing direct contact despite their separation that have been observed in many cell distance from the signaling center; by as much as 40 mm. The Dpp-depen- types. As described in excellent reviews the question how Dpp moves across dent cytonemes contain Tkv that con- [5 8], they have been assigned different the disc has been studied and dis- centrates in motile puncta and appear to roles to account for their presence in cussed extensively [15 19]. be specific for Dpp. various contexts (e.g. cell migration, cell The wing disc is a flattened sac Although these studies establish a adhesion, force generation, wound heal- composed of two connected epithelial strong correlation between the Dpp- ing, environmental sensing, antigen sheets, each one cell deep. The cells of dependent disc cytonemes and Dpp presentation, and neuronal pathfind- the columnar layer generate most of the signaling in the disc, direct evidence ing). Although their physical properties adult wing and thorax, and numbering for a role in Dpp trafficking has been vary (2 400 mm in length, 0.1 0.3 mm >40,000, are the most numerous. They obtained for cytonemes that extend diameter), all are actin-based, they are highly elongated along their apical/ from a wing disc-associated tracheal extend and retract at velocities that basal axis. The cells of the opposing branch called the Air Sac Primordium have been measured as much as 25 mm/ peripodial layer are relatively flat, (ASP). Growth and differentiation of the minute, and their tips can contact other described as squamous, and number ASP depends on Dpp that it receives cells. Their different shapes and roles are <2,000. Signaling studies have focused from the Dpp signaling center of the
26 Bioessays 37: 25–33, ß 2014 WILEY Periodicals, Inc...... Insights & Perspectives S. Roy and T. B. Kornberg hn again Think
Figure 1. Cytonemes of the Drosophila wing disc and histoblasts. A: The wing blade (Btl) and synapses with FGF-producing primordium is depicted as a columnar monolayer subdivided by the anterior/posterior disc cells [21, 22]. The signaling depen- compartment border (yellow) and a strip of Dpp-expressing cells (red). Concentration dence and apparent plasticity of both gradients of Dpp protein and Dpp signal transduction (red) spread from the Dpp expression types of cytonemes are similar to the domain and cell clones that express GFP have cytonemes (green) that orient along the apical surface toward the Dpp expressing cells. B: GFP reconstitution (GRASP) generates Dpp-dependent cytonemes of the wing fluorescence (green dots) in the Dpp expression domain in wing discs that express disc. Moreover, it has been possible to complementary parts of GFP as membrane-tethered external proteins in the disc flanks and show that the Tkv-containing cyto- the Dpp expression domain. Fluorescence marks points of cell-cytoneme contact. C: The nemes pick up and traffic Dpp from tracheal air sac primordium (ASP; purple) adjoins the basal surface of the wing disc and the disc and that Dpp signal transduc- extends cytonemes toward both Dpp-expressing (red) and FGF-expressing (green) cells; the tion in the ASP depends on the contact- gradient of Dpp signal transduction in the ASP is shown in red. D: Mutant ASP cells that do dependent uptake (Fig. 1D). Although not extend cytonemes that synapse with the disc cells do not activate Dpp signal transduction and the ASP is morphologically abnormal. E: Basal cytonemes (green) that these studies cannot rule out a role carry Hh extend from both Hh-expressing cells (blue) in the posterior compartment and from for Dpp that might be secreted in a Hh-receiving cells (red) in the anterior compartment. F: The rows of Hh-expressing cells diffusible form from source cells, only (blue) in monolayered epithelium of abdominal histoblasts extend Hh-carrying cytonemes to cytoneme-associated Dpp has been Hh-receiving cells (pink) that express the Hh receptor Patched (Ptc). detected in the space between Dpp producing and receiving cells, and the ASP does not take up and respond to disc (Fig. 1C) [22]. Its growth and two types of cytonemes one that Dpp or develop normally if its cyto- differentiation also depends on Branch- contains Tkv and synapses with Dpp- nemes are defective and unable to less/FGF received from a FGF signal producing disc cells and another that make functional contacts with Dpp source in the disc [23]. The ASP extends contains the FGF receptor Breathless source.
Bioessays 37: 25–33, ß 2014 WILEY Periodicals, Inc. 27 S. Roy and T. B. Kornberg Insights & Perspectives.....
Hh signaling is cytoneme- tems. They appear to mediate Hh enhancers are available that drive dependent signaling in Drosophila ovary germline expression specifically in either of the cells [26] and larval lymph glands [27], two different organs (i.e. wing disc and Hedgehog (Hh) is another paracrine to mediate Notch-dependent lateral trachea), making it possible to separate- signaling system that regulates growth inhibition in the wing disc [28], and to ly label and specifically tailor both cell and differentiation of Drosophila epi- deliver Epidermal growth factor (EGF) in types. The power of this system is thelia and for which signal trafficking the leg disc [29]. In zebrafish embryos, evident in the study described above is cytoneme-mediated (Fig. 1E). In the Wnt8a has been observed localizing to which showed that Dpp produced by the wing disc, cytonemes with motile Hh- cytoneme-like processes that extend wing disc neither moves to nor activates containing vesicles that extend from the from Wnt8a-expressing cells [30]; the signal transduction in the ASP if the ASP
Think again basal surface of Hh-expressing colum- best extant evidence that links cyto- is incapable of extending cytonemes nar cells orient toward Hh-receiving nemes to signaling in vertebrates has that synapse with the disc (Fig. 1D). cells, and their properties are consistent been reported for the chick limb bud in Without Dpp reception and signaling with the idea that they deliver Hh to Hh- which Sonic hedgehog (Shh) moves in the ASP, the ASP does not develop responding cells. Genetic conditions several hundred microns across the normally, but the wing disc, which is that reduce the number and length of mesenchyme [31]. Shh is produced by dependent on Dpp, develops normally the cytonemes also reduce the level and a small group of cells at the posterior in the absence of a proper ASP. Dpp expanse of Hh signaling [24]. Cyto- margin, but activates Hh signal trans- signaling in these discs is unaffected by nemes also extend from the basal duction across the developmental field the defective ASP, indicating that Dpp surface of the columnar cells that of the limb bud to specify the number production is also normal. receive and respond to Hh. These and identity of digits. Dynamic cyto- This conclusion has fundamental cytonemes have motile vesicles that nemes emanate from the Shh express- implications: If the Dpp-producing disc contain the Hh receptor Patched (Ptc) ing cells and transport Shh along their cells signal to cells that make cytonemes as well as Hh and Interference hedge- length, which have been measured as (other disc cells) but not to cells that hog (Ihog), and Hh signal transduction much as 150 mm. The properties of these cannot make cytonemes (the ASP cells), is reduced if formation of these cyto- extensions are consistent with a traf- then Dpp is not normally secreted nemes is compromised [25]. Thus in the ficking role. Filopodia also appear to except in the context of cytoneme wing disc, Hh movement is mediated mediate long range Notch/Delta signal- contacts. There is ample evidence that both by cytonemes that deliver Hh to ing between xanthophores and mela- cells that express receptors for a ligand Hh-receiving cells as well as cyto- nophores in zebrafish embryos [32]. respond when bound by the ligand nemes that take Hh up from producing whether the ligand arrives by its normal cells. Again, these results implicate route or by exogenous administration. cytonemesasessentialconduitsfor Perspective Neurons in culture or in ex vivo explants moving signaling proteins from source are activated by neurotransmitter added to target cells. The findings described in this abbrevi- from a pipet. Cultured cells respond to Hh-trafficking cytonemes have also ated summary provide strong experi- recombinant ligand produced in bacte- been characterized in the histoblast mental evidence for paracrine signaling ria and added to culture medium. Cells cells that generate the abdominal epi- that is cytoneme-mediated and contact- in developing tissues respond to recom- dermis [24]. Analysis of the histoblasts dependent. The most informative sig- binant ligand that emanates from beads has taken advantage of the fact that the naling systems have been Dpp in the that are impregnated with ligand. The abdominal epithelium is visible in intact Drosophila wing disc and the wing disc- point is that receptors discriminate animals (through a window cut out associated ASP and Hh in the Drosophi- ligand not route of administration, of the pupal case), and growth and la wing disc and abdomen, all of which and the fact that cells respond to ligand development continues over periods have genetic and physical attributes that arrives by diffusion is not evidence of observation that have exceeded that are favorable for these studies [21, that the ligand normally diffuses in 14 hours. In contrast, wing discs must 22, 24]. Cytonemes are thin (<200 nm extracellular space. The analysis of the be dissected and mounted in culture diameter) and only weakly fluorescent ASP showing that cytoneme contacts medium in order to image cytonemes for when marked with proteins such as are essential for Dpp uptake and shorter period of time, and there is no membrane-tethered eGFP, are sensitive signaling should therefore be under- evidence that isolated wing discs con- to physical manipulations, and most do stood in the context of ASP cells that tinue to develop ex vivo. Cytonemes not survive fixation. Imaging is there- express Tkv and are primed to respond carrying Hh emanate from Hh-express- fore challenging and only possible if to Dpp. Because Dpp moves from ing histoblasts (Fig. 1F). They are background fluorescence is low, if producing to receiving cell only at dynamic, extending across the entire fluorescent markers are expressed in a cytoneme synapses, the lack of re- domain of Hh-responding cells, they mosaic fashion in a portion of the cells sponse by ASP cells suggests that Dpp appear to contact the responding cells so that cytonemes extend over a non- is not near their receptors in the absence with their tip, and they rapidly retract. fluorescent (dark) background, and if of cytoneme contacts, that Dpp is not Cytonemes or specialized signaling the cytonemes orient along a suitable released constitutively, and that Dpp filopodia have been reported in several optical plane. The ASP system has the does not disperse across tissues by free other Drosophila and vertebrate sys- additional feature that transcriptional diffusion.
28 Bioessays 37: 25–33, ß 2014 WILEY Periodicals, Inc...... Insights & Perspectives S. Roy and T. B. Kornberg hn again Think
Figure 2. Restricted diffusion and cytoneme models of Hh gradient formation. A: Drawings the concentration gradients by binding depicting two cells expressing and secreting Hh (blue) that interacts directly with HSPGs in to signaling proteins, co-receptors and the ECM and disperses to generate a concentration gradient across two receiving cells other proteins that are involved in (brown; Ptc, green) by surface diffusion. The “conceptual” ECM is depicted as three tiers controlling the dispersion of signals consisting of (1) glycolipids (black lines) and surface glycoproteins (tan cylinders with black lines), (2) HSPGs (tall tan cylinders with black lines) and (3) extracellular microfibils (blue), (Fig. 2A) (reviewed in [34, 35]). fibronectin (purple) and hyaluronan (beaded strand). B: This drawing depicts Hh dispersing along cytonemes (black) that extend from expressing (brown) and receiving (blue) cells. C: Mutant receiving cells (brown) that do not express HSPGs neither extend cytonemes nor are contacted by cytonemes that extend from Hh-expressing cells. TGF-b signaling involves interactions with the ECM
The extracellular matrix tions decline exponentially with dis- The process that generates TGF-b and has a critical role in signal tance across the epithelial layer in which moves it from producing to receiving dispersion they are produced and appear to cells involves multiple steps and many disperse and signal only in that epithe- proteins (reviewed in [36, 37]), and it Passive diffusion is a conceptually lial layer. These distributions suggest will be described here in outline be- simple gradient-generating mechanism, that morphogens are constrained to the cause the principles of TGF-b trafficking but it is not compatible with the plane of the epithelium that expresses may be relevant to the way other morphogen gradients that form in them, and various mechanisms of “re- paracrine signaling proteins move anatomically complex tissues such as stricted diffusion” have been proposed through extracellular space. The TGF- imaginal discs [33]. The discs are not flat to account for morphogen dispersal. One b translation product is an inactive and also have multiple deep folds, and such mechanism is based on the extra- preproprotein. Signal peptidase gener- although they have closely apposed cellular matrix (ECM), whose constitu- ates an inactive proprotein that has an epithelial layers, morphogen concentra- ent proteoglycans are proposed to sculpt N-terminal propeptide (predicted MW
Bioessays 37: 25–33, ß 2014 WILEY Periodicals, Inc. 29 S. Roy and T. B. Kornberg Insights & Perspectives.....
with the usage that encompasses all required functions. However, even with this more restrictive designation, we might argue that despite the multiple control points, the one critical step that defines the biological output of the pathway is the release of TGF-b from the producing cell. Because there is no biological activity while TGF-b is bound and no biological activity until
Think again it is released, we might consider that the TGF-b maturation process strictly maintains TGF-b in an inactive state while priming the system for signaling. TGF-b is a powerful and critically important regulator of growth and differentiation, so its temporal and spatial activity must be precisely con- trolled. The steps that prepare TGF-b for release ensure that it cannot signal until the process is complete and signaling is appropriate. By analogy, the steps that regulate neurotransmitter synthesis and seques- tration in neurosecretory vesicles are essential to endow a neuron with the capacity to signal, but they are without consequence until neurotransmitters are released from the presynaptic cell. Figure 3. Diffusion and cytoneme models of TGF-b dispersion. Drawings depict TGF-b This aspect of TGF-b signaling is simi- (purple þ orange cylinder) that is synthesized and processed into mature peptide (orange lar, but neuronal signaling at synapses circle) that is inactive while bound with latent associated peptide (LAP, purple) and latent differs from the ECM-dependent model b TGF- binding protein (LTBP, red). A: Dispersion by diffusion predicts that inactive LAP- of TGF-b signaling in a fundamental bound TGF-b that is secreted as extracellular protein binds to ECM and that release from the ECM initiates the process of activation that involves binding to integrin (red cylinder) and way. A neuron is sensitive to neuro- receptor (green cylinders). B: The cytoneme model predicts that LAP-bound TGF-b transmitter whether the signal arrives disperses as an external but not extracellular protein and that release occurs only upon from a synaptic partner or from afar, binding to integrin and receptor. because its receptors cannot distinguish the source, but the synaptic mecha- nism, which involves pre-selection of 30 kDa) joined at a furin convertase mediate placement of latent TGF-b in targets and regulation of both release site to the functional TGF-b peptide the ECM. TGF-b is inactive and cannot and uptake of neurotransmitter, imparts (MW 13 kDa). TGF-b proprotein dimer- bind surface receptors until activated tight temporal, positional and quantita- izes and is glycosylated in the endo- by an as yet uncharacterized process tive specificity. In the ECM-dependent plasmic reticulum, but the mature that dissociates it from LAP and LTBP model, however, release of TGF-b from 13 kDa TGF-b peptide that furin hydro- (Fig. 3). the producing cells is of secondary lysis subsequently generates remains The pathway that leads to paracrine importance to its liberation from the bound to the prodomain and inactive. TGF-b is regulated at its start (synthesis ECM, and binding sites in the ECM are a In different contexts, furin cleavage of the mRNA) and at its end (release of consequence of a random walk after has been observed either before or after active signaling protein), and at least at release from the producing cell. Tempo- secretion; extracellular hydrolysis is one point in between (furin cleavage). ral, positional and quantitative control regulated by a secreted glycoprotein There are clearly many more intermedi- of signaling is therefore diminished called Emelin1. Irrespective of its cleav- ate steps at which distinct functions are by the spatial spread of TGF-b in the age state, TGF-b is secreted together required, but here we apply the term ECM. Cytoneme-mediated signaling at with the prodomain (termed the laten- “regulate” strictly, referring only to “morphogenetic synapses”, by contrast, cy-associated peptide, LAP) and a those steps whose rates are controlled. endows paracrine signaling with the binding protein called latent TGF-b This definition is consistent with its specificity characteristic of neurons [38]. binding protein (LTBP). Non-covalent usage in, for instance, glycolysis, in The finding that signaling by the TGF-b interactions between fibrillins, which which the activities of only three of family member Dpp is cytoneme-depen- are important structural components of the ten enzymes vary in response to dent raises the question of the role of the the ECM, and LTBPs are thought to metabolic conditions, and it contrasts ECM in cytoneme-mediated signaling.
30 Bioessays 37: 25–33, ß 2014 WILEY Periodicals, Inc...... Insights & Perspectives S. Roy and T. B. Kornberg
Cytoneme-mediated are not directly attached to the cell. anterior side of the compartment bor- signaling is dependent on Signaling proteins such as TGF-b and der, which do not activate Hh signal FGF bind to the GAG chains of extracel- transduction normally and whose ante- hn again Think the ECM lular proteoglycans, and the journey rior neighbors also do not activate Hh Mutant Drosophila cells that cannot they take to the receptors in the lower signal transduction normally [39, 45], synthesize glycosaminoglycan (GAG)- tier is dependent on interactions with do not have cytonemes on their basal modified proteoglycans are deficient glycoproteins in the upper tiers. Esti- surface (Fig. 2B, C) [24]. This finding for Hh, Wingless, Dpp, and FGF signal- mates of the number of cell surface supports the idea that cytonemes medi- ing ([35, 39] and reviewed in [40, 41]). HSPGs (105 106 per cell) are approxi- ate Hh transport between producing Whereas these signaling proteins nor- mately an order of magnitude lower and receiving cells, and suggests that mally activate signal transduction and than the number of glycoproteins and interactions with HSPGs are essential are present in cells many cell diameters glycolipids in the first tier (J. Esko, for cytoneme growth and/or stability. away from their source, most mutant personal communication, and [42]). If a The essential role of the HSPGs for Hh cells activate signal transduction poorly 10 mm diameter “average” cell (which signaling in the wing disc is therefore to and appear to take up signaling proteins has a surface area of 3 1010 sq. A˚ ) provide a substrate for the cytonemes at reduced levels. The only mutant cells distributes the 105 106 HSPGs evenly that track over the surface of the disc that respond at high levels to the over its surface, then for purposes of a cells. signaling proteins are those that are back-of-the-envelope calculation, each juxtaposed to producing cells. More- HSPG will occupy an average area of a over, signaling proteins are also signifi- square with sides of 170 550 A˚ . GAG Discussion cantly reduced in genetically normal chains are approximately 20 100 kDa, cells that are separated from source cells and because a 100 kDa GAG chain is The evidence for cytoneme-mediated by mutant territory, and these distal predicted to have a radius of gyration of and contact-dependent signaling in cells are also signaling-deficient. These approximately 230 A˚ [43], we estimate Drosophila larval tissues is strong. observations have been interpreted as that an HSPG with multiple GAG chains Although we do not yet understand functional evidence for a GAG-depen- might occupy a square with 500 A˚ sides. how this mechanism of dispersion and dent mechanism of dispersion and for Although this calculation is imprecise, transfer of signaling proteins generates models such as the one illustrated in it yields a model that is consistent with a concentration gradients across fields of Fig. 2A. This model is based on surface surface topography that has full cover- cells, the gradients depend on it. One diffusion and the idea that heparan age by HSPGs, and therefore with a possibility takes into account the plas- sulfate proteoglycans (HSPGs) tran- sliding, surface diffusion model for ticity and transient nature of cytonemes siently bind signaling proteins, keeping dispersion. and the apparent correlation between them close to the cell surface as they A different interpretation of the number of cytonemes, cytoneme length, diffuse away from their source. Surface mutant HSPG phenotype focuses on proximity to source cells and abun- diffusion is a well-characterized pro- the mutant cells that activate signal dance of signaling protein. Shorter cess that involves particles hopping transduction at normal levels despite cytonemes are more numerous than between adjacent adsorption sites on a their lack of HSPGs. These are the cells longer ones and are more numerous surface; its rates are dependent on a that are juxtaposed to source cells. Their closest to source cells where signaling number of factors including the responsiveness suggests that direct proteins are most abundant. Concentra- strength of adsorption, the structure contact with source cells may be suffi- tion gradients may therefore be a and properties of the binding species on cient to enable binding and uptake of product of the frequency and duration the surface and the chemical potential the signaling proteins in an HSPG- of functional contacts where signaling gradients at the surface. Although we independent manner, whereas long proteins are released and taken up. The lack any measures of these parameters distance signaling, which is cytoneme- reason for proposing this model is not to for cells, we can make a rough estimate mediated and contact-dependent [22], champion it but simply to describe one of the extracellular topography with may be HSPG-dependent. Recent results that is consistent with the data and that respect to HSPGs. imaging cytonemes in the context of is not based on diffusion. We do this A low-resolution model of a typical HSPG mutant cells indicates that this in order to discount the notion that the extracellular space might have three interpretation is correct [24]. role of cytonemes in gradient formation tiers (Fig. 2A). At the cell surface are As described above, cytonemes ex- should be dismissed because our un- numerous glycoproteins and glycolipids tend along the basal surface of the derstanding is incomplete. Cytoneme- that have short, branched glycans. A columnar cells of wing imaginal discs, mediated dispersion is the only mecha- second tier that may be thought of as a and some carry Hh from posterior nism that is supported by direct obser- canopy that extends over the lower tier compartment Hh-producing cells across vation and for which functional genetic consists of proteoglycans that have the anterior/posterior compartment evidence exists. The more important linear transmembrane protein cores border to anterior compartment cells question relevant to this discussion is and extracellular domains to which [24, 44]. Wild type anterior cells at the the role of cytonemes in contexts other branched and modified GAG are at- border have many cytonemes crossing than Drosophila larval tissues, and in tached. Above these two tiers is a their basal surface. In contrast, clones of particular the implications for the role network of “matrix” glycoproteins that HSPG-deficient mutant cells on the of the ECM.
Bioessays 37: 25–33, ß 2014 WILEY Periodicals, Inc. 31 S. Roy and T. B. Kornberg Insights & Perspectives.....
We first address two issues. The term relevant to Dpp dispersion. The fact that neme-like structures that correlate with “extracellular” is frequently used to cytonemes are present in the lumen paracrine signaling in other inverte- describe secreted proteins, but without invalidates such assumptions. brate [47, 48] as well as vertebrate specifying their state. Extracellular can The issue of cell association is not systems [31, 49, 50]. refer to the external face of proteins that resolved by the presence of protein in These observations favor a mechanism are integral to the plasma membrane or medium conditioned by cells grown in of paracrine signaling that is shared by all are attached to the outside surface of a culture, by observations made of over- cells, including neurons, but they do not cell. It is also used to refer to proteins expression conditions, or by responses to eliminate the possibility of non-contact that cells secrete and release and that are exogenously supplied signaling protein dependent paracrine signaling. Certainly not attached to the cell. The imprecision (see [38]). The issue is also relevant to the there are many well-characterized signal-
Think again becomes problematic, for instance, with interactions of secreted proteins with the ing systems for which the presence and antibody staining protocols for fluores- ECM. Whereas studies of vertebrate TGF- role of cytonemes has not been explored. cence immunohistochemistry that detect b implicate proteoglycans in the upper Guidance cues that steer neuronal path- extracellular protein. Fluorescence mi- tier of the ECM as critical for signaling, finding are examples, although the actual croscopy lacks sufficient resolution to neither histochemical nor biochemical distributions have only been inferred and distinguish cell-associated protein from studies have the resolution or sensitivity have not been directly observed (reviewed protein that has been released but to establish whether TGF-b that is bound in [51]). Long distance Dpp signaling and remains in close proximity to its source. to ECM is extracellular (free) or is Dpp dispersion in the early Drosophila It is also an issue for protein that is in a external and cytoneme-associated. In- embryo is another example [52], and cytoneme and at a distance from the cell deed, it is possible that vertebrate TGF-b studies of the Xenopus embryo also reveal body if the cytoneme is not marked. Such and Drosophila Dpp move between cells long distance dispersion of signaling protein will appear to be in exovesicles by similar mechanisms, that TGF-b is proteins [53]. Unfortunately, the conclu- and not associated with a cell. Hence- cytoneme-bound when passing between sion these studies make that diffusion forth, we use the term “external” for cell- expressing and receiving cells and that disperses the signaling proteins is based associated moieties on the outside of ECM-associated TGF-b is not extracellu- solely on the observed patterns of expres- cells “they have been externalized” lar (Fig. 3). For both the vertebrate and sion and dispersion. and reserve the term “extracellular” for Drosophila systems, a better understand- The speeds at which Dpp and FGF those that are not bound to the cells that ing of the structure of the ECM, of the role spread through tissues (measured in release them. and structure of cytonemes, and of the various ways including Fluorescence Both externalized and extracellular state of in transit signaling proteins is recovery after photobleaching (FRAP), proteins are secreted, but distinguish- needed in order to know if the apparent spatial FRAP, fluorescence correlation ing between these two states is critically differences reflect different mechanisms. microscopy and pair correlation func- important for mechanism. The lumen tion microscopy [14, 54, 55] are consis- of the wing imaginal discs is a small tent with free diffusion when certain space that separates the apical faces of Conclusions values for relevant parameters are the columnar and peripodial cells by assumed, but it is important to recognize as little as 6 mm. Although Dpp is The structural and functional similari- that diffusion is but one mechanism of expressed only by a narrow stripe of ties of synaptic signaling by axons and dispersion and neither rates nor patterns cells at the A/P compartment border cytonemes reveal an ancient kinship, of distribution distinguish between in both the columnar and peripodial and such evolutionary conservation them. In the absence of evidence for layers, fluorescence immunohistochem- might favor the idea of a universal or against the presence of signal protein- istry detects Dpp across the entire A/P mechanism for paracrine signaling carrying cytonemes, conclusions should expanse of the lumen [46]. If we assume that “every cell is a neuron and a neuron reflect the uncertainties that the state of that the Dpp is free in the lumen, it is not alone” in the sense that all cells understanding demands. would be reasonable to conclude that it communicate at distance by reaching has dispersed by diffusion. However, out with extensions to make direct because apical cytonemes do not survive contacts where signals are exchanged. Acknowledgment the fixation protocols that were used for The observations we made in Drosophi- The authors acknowledge support from immunohistochemistry and because la larval tissues that are consistent with the National Institutes of Health (grants cytonemes cannot be detected if their this cytoneme model include EGF sig- GM030637 and GM105987 to T. B. K. and components are not marked, the relation- naling in the eye imaginal disc, Hh and K99HL114867 to S. R.). ship between Dpp and cytonemes in the Dpp signaling in the wing imaginal disc, lumen was not revealed in these studies and FGF and Dpp signaling in the ASP. References anditisnotclearwhethertheDppis Indeed, we have not found any signal-
external (and cytoneme-associated) or ing system that lacks cytonemes linking 1. Spemann H, Mangold H. 1924. U¨ ber Weck- extracellular (and free). The presumption signaling cells. Studies of the Drosophi- ung organisatorischer Fa¨ higkeiten durch Ver- of the “simpler” mechanism of extra- la ovary, wing disc and abdominal planzung in organisatorische Umgebung. cellular diffusion is justified in the histoblasts also report cytonemes that Roux’s Arch 109: 557 77. 2. Lander AD, Lo WC, Nie Q, Wan FY. 2009. absence of any knowledge of structural correlate with signaling [24, 26, 44], and The measure of success: Constraints, objec- elements in the lumen that might be there are numerous reports of cyto- tives, and tradeoffs in morphogen-mediated
32 Bioessays 37: 25–33, ß 2014 WILEY Periodicals, Inc...... Insights & Perspectives S. Roy and T. B. Kornberg
patterning. Cold Spring Harb Perspect Biol 1: 23. Sato M, Kornberg TB. 2002. FGF is an tumour suppressor EXT-1 and is needed for a002022. essential mitogen and chemoattractant for Hh diffusion. Nature 394:85 8. 3. Ramı´rez-Weber FA, Kornberg TB. 1999. the air sacs of the drosophila tracheal system. 40. Lin X, Perrimon N. 2000. Role of heparan
Cytonemes: Cellular processes that project to Dev Cell 3: 195 207. sulfate proteoglycans in cell cell signaling in again Think the principal signaling center in Drosophila 24. Bischoff M, Gradilla AC, Seijo I, Andres G, Drosophila. Matrix Biol 19: 303 7. imaginal discs. Cell 97: 599 607. et al. 2013. Cytonemes are required for the 41. Tabata T, Takei Y. 2004. Morphogens, their 4. Kornberg TB, Roy S. 2014. Cytonemes as establishment of a normal Hedgehog mor- identification and regulation. Development specialized signaling filopodia. Development phogen gradient in Drosophila epithelia. Nat 131: 703 12. 141: 729 36. Cell Biol 15: 1269 81. 42. Xu D, Esko JD. 2014. Demystifying heparan 5. Faix J, Breitsprecher D, Stradal TE, Rottner 25. Chen W, Kornberg TB. 2014. The Hedgehog sulfate protein interactions. Annu Rev Bio- K. 2009. Filopodia: Complex models for simple gradient of the Drosophila wing imaginal disc chem 83: 129 57. rods. Int J Biochem Cell Biol 41: 1656 64. requires cytonemes. 55th Annual Drosophila 43. Bathe M, Rutledge GC, Grodzinsky AJ, 6. Heckman CA, Plummer HK 3rd. 2013. Research Conference. San Diego, CA. Tidor B. 2005. A coarse-grained molecular Filopodia as sensors. Cell Signal 25:2298 311. 26. Rojas-Rios P, Guerrero I, Gonzalez-Reyes A. model for glycosaminoglycans: Application to 7. Hoshino D, Branch KM, Weaver AM. 2013. 2012. Cytoneme-mediated delivery of hedge- chondroitin, chondroitin sulfate, and hyalur- Signaling inputs to invadopodia and podo- hog regulates the expression of bone morpho- onic acid. Biophys J 88: 3870 87. somes. J Cell Sci 126: 2979 89. genetic proteins to maintain germline stem cells 44. Callejo A, Bilioni A, Mollica E, Gorfinkiel N, 8. Mellor H. 2010. The role of formins in in Drosophila. PLoS Biol 10: e1001298. et al. 2011. Dispatched mediates Hedgehog filopodia formation. Biochim Biophys Acta 27. Mandal L, Martinez-Agosto JA, Evans CJ, basolateral release to form the long-range 1803: 191 200. Hartenstein V,etal. 2007. A Hedgehog- and morphogenetic gradient in the Drosophila 9. Raftery LA, Umulis DM. 2012. Regulation of Antennapedia-dependent niche maintains wing disk epithelium. Proc Natl Acad Sci USA BMP activity and range in Drosophila wing Drosophila haematopoietic precursors. Na- 108: 12591 8. development. Curr Opin Cell Biol 24: 158 65. ture 446: 320 4. 45. Takei Y, Ozawa Y, Sato M, Watanabe A, 10. Entchev EV, Schwabedissen A, Gonzalez- 28. Cohen M, Georgiou M, Stevenson NL, et al. 2004. Three Drosophila EXT genes Gaitan M. 2000. Gradient formation of the Miodownik M,etal. 2010. Dynamic filopodia shape morphogen gradients through synthe- TGF-beta homolog Dpp. Cell 103: 981 91. transmit intermittent Delta-Notch signaling to sis of heparan sulfate proteoglycans. Devel- 11. Lecuit T, Cohen SM. 1998. Dpp receptor drive pattern refinement during lateral inhibi- opment 131:73 82. levels contribute to shaping the Dpp morpho- tion. Dev Cell 19:78 89. 46. Gibson MC, Lehman DA, Schubiger G. gen gradient in the Drosophila wing imaginal 29. Peng Y, Han C, Axelrod JD. 2012. Planar 2002. Lumenal transmission of decapenta- disc. Development 125: 4901 7. polarized protrusions break the symmetry of plegic in Drosophila imaginal discs. Dev Cell 12. Schwank G, Dalessi S, Yang SF, Yagi R, EGFR signaling during Drosophila bract cell 3: 451 60. et al. 2011. Formation of the long range Dpp fate induction. Dev Cell 23: 507 18. 47. Akiyama-Oda Y, Oda H. 2003. Early pattern- morphogen gradient. PLoS Biol 9: e1001111. 30. Luz M, Spannl-Muller S, Ozhan G, Kager- ing of the spider embryo: A cluster of 13. Teleman AA, Cohen SM. 2000. Dpp gradient meier-Schenk B,etal. 2014. Dynamic mesenchymal cells at the cumulus produces formation in the Drosophila wing imaginal association with donor cell filopodia and Dpp signals received by germ disc epithelial disc. Cell 103: 971 80. lipid-modification are essential features of cells. Development 130: 1735 47. 14. Zhou S, Lo WC, Suhalim JL, Digman MA, Wnt8a during patterning of the zebrafish 48. McClay DR. 1999. The role of thin filopodia in et al. 2012. Free extracellular diffusion creates neuroectoderm. PLoS One 9: e84922. motility and morphogenesis. Exp Cell Res the Dpp morphogen gradient of the Drosoph- 31. Sanders TA, Llagostera E, Barna M. 2013. 253: 296 301. ila wing disc. Curr Biol 22: 668 75. Specialized filopodia direct long-range trans- 49. Danilchik M, Williams M, Brown E. 2013. 15. Baena-Lopez LA, Nojima H, Vincent JP. port of SHH during vertebrate tissue pattern- Blastocoel-spanning filopodia in cleavage- 2012. Integration of morphogen signalling ing. Nature 497: 628 32. stage Xenopus laevis: Potential roles in within the growth regulatory network. Curr 32. Hamada H, Watanabe M, Lau HE, Nishida morphogen distribution and detection. Dev Opin Cell Biol 24: 166 72. T,etal. 2014. Involvement of Delta/Notch Biol 382:70 81. 16. Crozatier M, Glise B, Vincent A. 2004. signaling in zebrafish adult pigment stripe 50. Fierro-Gonzalez JC, White MD, Silva JC, Patterns in evolution: Veins of the Drosophila patterning. Development 141: 318 24. Plachta N. 2013. Cadherin-dependent filo- wing. Trends Genet 20: 498 505. 33. Kornberg TB, Guha A. 2007. Understanding podia control preimplantation embryo com- 17. Kornberg TB. 2012. The imperatives of morphogen gradients: A problem of disper- paction. Nat Cell Biol 15: 1424 33. context and contour for morphogen disper- sion and containment. Curr Opin Genet Dev 51. Dickson BJ, Zou Y. 2010. Navigating inter- sion. Biophys J 103: 2252 6. 17: 264 71. mediate targets: The nervous system midline. 18. Schwank G, Basler K. 2010. Regulation of 34. Restrepo S, Zartman JJ, Basler K. 2014. Cold Spring Harb Perspect Biol 2: a002055. organ growth by morphogen gradients. Cold Coordination of patterning and growth by the 52. Wang YC, Ferguson EL. 2005. Spatial Spring Harb Perspect Biol 2: e001669. morphogen DPP. Curr Biol 24: R245 55. bistability of Dpp-receptor interactions during 19. Wartlick O, Gonzalez-Gaitan M. 2011. The 35. Yan D, Lin X. 2009. Shaping morphogen Drosophila dorsal ventral patterning. Nature missing link: Implementation of morphogenetic gradients by proteoglycans. Cold Spring Harb 434: 229 34. growth control on the cellular and molecular Perspect Biol 1: a002493. 53. Plouhinec JL, Zakin L, Moriyama Y, De level. Curr Opin Genet Dev 21:690 5. 36. Munger JS, Sheppard D. 2011. Cross talk Robertis EM. 2013. Chordin forms a self- 20. Hsiung F, Ramirez-Weber FA, Iwaki DD, among TGF-beta signaling pathways, integ- organizing morphogen gradient in the extra- Kornberg TB. 2005. Dependence of Dro- rins, and the extracellular matrix. Cold Spring cellular space between ectoderm and meso- sophila wing imaginal disc cytonemes on Harb Perspect Biol 3: a005017. derm in the Xenopus embryo. Proc Natl Acad Decapentaplegic. Nature 437: 560 3. 37. Todorovic V, Rifkin DB. 2012. LTBPs, more Sci USA 110: 20372 9. 21. Roy S, Hsiung F, Kornberg TB. 2011. than just an escort service. J Cell Biochem 54. Kicheva A, Pantazis P, Bollenbach T, Specificity of Drosophila cytonemes for dis- 113: 410 8. Kalaidzidis Y,etal. 2007. Kinetics of morpho- tinct signaling pathways. Science 332: 354 8. 38. Kornberg TB, Roy S. 2014. Communicating gen gradient formation. Science 315:521 5. 22. Roy S, Huang H, Liu S, Kornberg TB. 2014. by touch—neurons are not alone. Trends Cell 55. Yu SR, Burkhardt M, Nowak M, Ries J,etal. Cytoneme-mediated contact-dependent Biol 24: 370 6. 2009. Fgf8 morphogen gradient forms by a transport of the Drosophila decapentaplegic 39. Bellaiche Y, The I, Perrimon N. 1998. Tout- source-sink mechanism with freely diffusing signaling protein. Science 343: 1244624. velu is a Drosophila homologue of the putative molecules. Nature 461: 533 6.
Bioessays 37: 25–33, ß 2014 WILEY Periodicals, Inc. 33