The function of tubular epithelial organs like the kidney and lung is critically dependent on the length and diameter of their constituting branches. Genetic analysis of tube size control during Drosophila tracheal development has revealed that epithelial septate junction (SJ) components and the dynamic chitinous luminal matrix coordinate tube growth. However, to date the underlying molecular mechanisms controlling tube expansion have remained elusive. Two luminal chitin binding proteins, Serpentine and Vermiform, possessing predicted polysaccharide deacetylase activities (ChLDs) are required to assemble the cable-like extracellular matrix (ECM) and restrict tracheal tube elongation. Overexpression of native, but not of mutated, ChLD versions also interferes with the structural integrity of the intraluminal ECM and causes aberrant tube elongation. Whereas ChLD mutants have normal SJ structure and function, the luminal deposition of the ChLD requires intact cellular SJs. This identifies a new molecular function for SJs in the apical secretion of ChLD and positions ChLD downstream of the SJs in tube length control. serp and verm are not required for chitin synthesis or secretion but rather for its normal fibrillar structure. They are responsible for the N-deacetylation of chitin and conversion of chitin to chitosan, a polysaccharide composed of repeating glucosamine units; obtained by de-acetylation of chitin. The mutations also affect structural properties of another chitinous matrix, epidermal cuticle. In tracheal development, deposition of the chitin luminal matrix first promotes and coordinates radial tube expansion. It is proposed that the subsequent structural modification of chitin by chitin binding deacetylases selectively instructs the termination of tube elongation to the underlying epithelium (Wang, 2006; Luschnig, 2006).

An important late step in the morphogenesis of tubular organs is the final acquisition of distinct branch sizes. The fixed length and diameter of tubes dictate the flow rates of the transported gas or liquid and are therefore major determinants of optimal organ function. Several of the genes controlling tube size during organogenesis encode potential regulators of apical membrane and cytoskeletal organization, but the cellular mechanisms measuring and restricting lumen growth are not well understood. In the Drosophila airways, a dynamic chitinous extracellular matrix forms inside the developing branches and is required for uniform tube expansion. Loss of the transient chitin cable results in cystic and winding tubes and defects in apical cell shapes and cytoskeletal organization, suggesting that the expanding polysaccharide cable controls tube extension. How does the luminal chitin matrix signal its structural integrity to the epithelium? The lamellar structure and stiffness of arthropod cuticles depends on the arrangement of chitin polymers [β(1-4)-linked N-acetylglucosamine, GlcNAc] in planes of parallel or antiparrallel fibrils through the formation of hydrogen bonds between the chains (Merzendorfer, 2003). Moreover, the rigidity of the Saccharomyces cerevisiae spore cell wall requires chitin deacetylases that convert chitin to chitosan, a β(1-4) N-Glucosamine (GlcN) polymer with distinct physical properties. It is hypothesized that similar luminal chitin modifications resulting in structural changes of the matrix may control tube length or diameter growth. The fly genome encodes more than 100 putative chitin binding proteins (ChtB, IPR002557; Magkrioti, 2004), and three of them, encoded by CG8756, CG32209, and CG17905, contain additional Low-Density Lipoprotein Receptor (LDLR, IPR002172) and polysaccharide deacetylase (Polysacc Deac, IPR002509) domains, suggesting that they may be involved in the assembly and maturation of chitin polymers. This protein family has been named ChLDs because of the predicted structure of the encoded proteins, and the CG8756 and CG32209 members were named vermiform (verm) and serpentine (serp), respectively, because of the tracheal phenotypes of the mutants. The expression patterns of verm, serp, and ChLD3 were examined in wild-type embryos by in situ hybridizations. verm and serp are expressed in all tracheal cells from stage 12 until the end of embryogenesis, and both genes are also expressed in the epidermis from stage 15 onward. ChLD3 expression was not detected in the trachea, and it was therefore not analyzed further. To investigate the localization of the Verm protein, antibodies were raised against its C terminus and wild-type embryos were stained. Verm was detected inside all tracheal cells from stage 12. Verm deposition into the lumen of the dorsal trunk (DT) commences abruptly at stage 14, followed by all other branches thereafter. By stage 15, the Verm labeling was luminal and lining the apical cell surface of the DT. This staining was severely reduced to background levels in verm mutants. btl-GAL4 was used to express Serp-GFP fusion protein and it was found to be secreted into the lumen similarly to endogenous verm (Wang, 2006).

The dynamic localization of ChLD proteins as well as their predicted structure suggests a role in the assembly of the intraluminal chitinous filament. To determine the function of Verm and Serp, strong loss-of-function mutations were obtained for both genes. Stage-16 verm mutant embryos stained for the luminal 2A12 antigen display excessive elongation and convolutions of the DTs, which become gradually more pronounced in later stages. This defect becomes largely ameliorated in mutants expressing UAS-verm in the trachea, suggesting that verm restricts tube elongation. Similarly elongated DT tubes were detected in serp mutant embryos. In addition, Df(3L)Exel6135 embryos, lacking ten genes including verm and serp, show more severe tube-overextension phenotypes than either single mutant. These mutants also displayed a slight diametric overgrowth in the DT, suggesting that the removal of both ChLDs, or other genes in their neighborhood, may also influence the restriction of radial tube growth. The analysis of verm and serp embryos labeled with anti-DE-Cadherin antibodies also revealed overstretched cellular profiles. This indicates that both verm and serp are required to halt tube elongation and co-ordinate apical cell extensions at late stage 15. The aggravated phenotype of embryos deficient for both proteins suggests that they may cooperate in that task (Wang, 2006).

Do the tube-length defects in the mutants correlate with phenotypes in the assembly and structure of the luminal matrix? Luminal chitin assembly was analyzed by a fluorescent chitin binding protein (ChtB) and a fluorescent chitin binding lectin (WGA) (Tonning, 2005). Both reagents revealed a cylindrical structure of tightly packed luminal chitin polymers in the wild-type. In homozygous verm and hemizygous serp embryos, this structure appeared diffuse and expanded radially already at stage 15. By stage 16, its labeling intensity was severely reduced in the DTs compared to the wild-type. Analysis of transverse transmission electron microscopy (TEM) sections of wild-type DTs at stage 16 reveals a homogenously packed luminal matrix of parallel fibrils. The orientation and assembly of the fibrils is distorted in verm mutants, indicating that ChLDs organize intraluminal chitin packing. The formation of the luminal matrix depends on the chitin synthetase encoded by krotzkopf verkehrt (kkv) (Ostrowski, 2002). In kkv mutants, the chitinous cable is absent and the tubes first fail to expand their diameter and later become over elongated and convoluted (Tonning, 2005). The tube phenotypes of kkv verm double mutants were identical to the ones of kkv, indicating that chitin synthesis is a prerequisite for ChLD function. In addition, the luminal Verm staining was severely disrupted in kkv mutants. Whereas early events of tracheal cuticle assembly, like the apical deposition of the epicuticle layers at early stage 16, were unaffected in verm mutants, cuticular abnormalities were detected in DT cross-sections at late stage 16. These include irregular taenidia, reduced procuticle deposition, and aberrant apical membranes, revealing an additional, later role of Verm in tracheal cuticle assembly. The above data argue that ChLDs are primarily required to form or preserve the tight texture of the chitinous luminal cable and that their function and localization require kkv and its product chitin (Wang, 2006).

To further analyze the functional potential of ChLD proteins, native and GFP-tagged versions of Verm and Serp were overexpressed in the trachea and the phenotypes were assessed in embryos stained for the ChtB reagent or 2A12. Expression of either transgene caused distortion of the filamentous texture and a reduction in the staining intensity of the chitin cable. Whereas native Verm causes lesser effects, Serp and both GFP-tagged versions induce more-pronounced phenotypes. Moreover, the tube-overgrowth and -winding phenotypes in the tracheal branches were proportional to the severity of the intraluminal matrix defect. Thus, ChLD-induced changes in filament structure play an instructive role in tube elongation (Wang, 2006).

To dissect the function of the three predicted domains of Serp in this assay, transgenic embryos were generated expressing GFP-tagged versions lacking each of the motifs. Tracheal expression of Serp-ΔChtB-GFP or Serp-ΔLDLR-GFP did not cause any detectable luminal or tube-extension phenotype, indicating that both domains are necessary for the function of Serp in tube length control. Expression of Serp-ΔDeac-GFP did not affect tube elongation either, but, surprisingly, it was found that the protein does not become secreted but accumulates inside the tracheal cells. Thus, the deacetylase domain harbors motifs required for Serp luminal secretion. The failure of the deletion constructs to cause phenotypes argues against antimorphic effects of the ChLD-GFP proteins and indicates that all three predicted domains are required for tracheal Serp function. Overexpression of the ChLDs leads to similar luminal and tube-elongation phenotypes as their loss of function, suggesting that both lack of as well as increased levels of chitin modifications result in similar effects (Wang, 2006).

The analysis of several Drosophila mutants with overgrown tracheal tubes similar to verm and serp embryos demonstrated that paracellular septate junction (SJ) components are required for halting lumen overgrowth. Two different molecular mechanisms have been proposed to account for the tube-overextension phenotypes in SJ mutants: (1) SJ-associated proteins like Scribble and Discs Large may restrict the activity of apical-membrane growth-promoting factors like Crb and DaPKC during tube growth, and (2), SJ functions may be required in luminal matrix assembly and thereby indirectly regulate tube expansion (Tonning, 2005). Is there a mechanistic link between the function of SJs and ChLDs in tube length control? Whether ChLDs are required for the assembly of SJs was examined. Df(3L)Exel6135 and verm mutant embryos were initially examined in three different assays. (1) Staining was performed using antibodies against the SJ markers Neurexin and FasIII; (2) DT crossections of verm mutants were analyzed for the ladder-like SJ structure by TEM; (3) the function of the paracellular barrier was tested by Dextran injections. ChLD mutants were indistinguishable from wild-type in all three experiments, arguing that ChLD does not affect SJ structure or function. Is Verm affected in SJ mutants? The localization of Verm was analyzed in mutants for the α subunit of the Na/K-ATPase (Atpα). In these embryos, Verm is retained inside the tracheal cells at stage 15, whereas its luminal abundance is strongly reduced and becomes gradually undetectable at stage 16. This defect in Verm luminal secretion is accompanied by only a subtle decrease in the luminal 2A12 staining. The same phenotypes were detected of aberrant Verm intracellular accumulation and strongly reduced luminal levels in sinu (Claudin) and lac (bulb) mutants, suggesting that Verm luminal secretion depends on the assembly or maintenance of the SJ structure. The defect of Verm intracellular retention and luminal stabilization in SJ mutants is unlikely to reflect a general block of apical secretion because neither the 2A12 marker nor the luminal ZP-protein Pio showed defects in luminal accumulation at stage 15. The distinct luminal-deposition phenotypes demonstrate a new function for the transepithelial barrier junctions in the apical secretion of Verm. Moreover, these apical secretion defects provide molecular evidence that ChLDs act downstream of the SJs to regulate tube size. The sorting signal required for apical targeting via SJs is presently unknown, and its identification may be aided by the dissection of the role of sequences in the Deacetylase domain in luminal secretion (Wang, 2006).

The phenotypic analysis of known tracheal-tube-size mutants unveiled an instrumental role of the chitinous extracellular matrix structure in orchestrating branch dimensions. The assembly and growth of the chitin cable first coordinates the radial expansion of the tubes. Subsequent modifications in chitin fibril structure by secreted ChLDs instruct the underlying epithelium to terminate tube elongation. Thus, dynamic structural changes of the luminal matrix may be sensed independently to determine diametric growth and tube elongation. Given the predicted structure of ChLDs, such changes may involve chitin deacetylation and consequently the extent of hydrogen bonding between chitin fibers and their packing. The similar loss- and gain-of-function ChLD phenotypes suggest that the distinct level of chitin modification and the expected changes in matrix rigidity monitor tube elongation. Luminal fibrillar ECM structures have been described in a vertebrate angiogenesis model, and this genetic analysis provides a first mechanistic view on distinct functions of the luminal ECM in tube size control (Wang, 2006).

The two genes define a new molecular class of tube morphogenesis genes encoding apical extracellular matrix proteins that modify the chitin matrix. Both proteins contain chitin binding and deacetylase domains, and both are secreted into the apical tracheal matrix, where they associate with, and modify the structure of, the chitin cylinder that fills most of the luminal space. In serp verm double mutants, the chitin cylinder still forms, but it lacks its normal fibrillar appearance. The chitin cylinder, along with the surrounding lumen and apical tracheal surface, becomes excessively long and convoluted, a process that normally occurs gradually over the next hours and days of development as the tubes expand to their mature sizes and acquire their characteristic shapes (Luschnig, 2006).

The results, along with the recent identification of chitin biosynthesis genes in tracheal-tube morphogenesis, demonstrate the dual and genetically separable functions of chitin in tracheal tube diameter and length control. The chitin synthesis genes cystic, knk/gnarled, and kkv chitin synthase are required to synthesize the expanding chitin cylinder in the tracheal lumen, which is proposed to promote and coordinate the dilation of the surrounding epithelium so that tubes reach their proper diameter. By contrast, the chitin modification genes serp and verm are not required for synthesis of chitin and have little effect on dilation and tube diameter. Rather, they influence the structure of the chitin cylinder and the length and curvature of the tubes. The chitin cylinder may therefore function as an internal template that plays a critical role in defining the diameter, length, and shape of the tube that surrounds it. A molecular model is presented for the role of Serp and Verm proteins and the chitin cylinder in tube lengthening and it is suggested how this mechanism could be regulated to control the longitudinal growth of tracheal tubes during development (Luschnig, 2006).

There are four postulates of the model. (1) It is proposed that the Serp and Verm proteins bind and modify chitin; (2) this modification alters the physical properties of the chitin cylinder, keeping it rigid and short; (3) these changes in the chitin cylinder are sensed by the surrounding tracheal cells, perhaps through a direct link between an apical-membrane component and a constituent of the chitin cylinder; (4) this signal restricts apical-membrane biogenesis, in a manner that limits polarized growth of the cell membrane specifically along the longitudinal axis of the tube (Luschnig, 2006).

Much of the data supports the first two postulates. Serp and Verm proteins are secreted into the tracheal lumen, where they associate with the chitin cylinder, an association that is likely mediated by the chitin binding domain. The chitin modification is most probably deacetylation of N-acetylglucosamine residues by the chitin deacetylase domain of the proteins. This is a well-known structural modification of chitin in yeast and fungi, and the enzymatic catalysis of this reaction has received much attention because of the commercial use of deacetylated chitin (chitosan) in water treatment, in the food industry, and in medical applications such as fabrication of artificial skin. Deacetylation increases the solubility and decreases the density of chitin fibrils in vitro, and it may influence the structure and orientation of chitin fibrils in arthropod cuticle. This apparently increases the rigidity of the chitin matrix, as implied by the defect in cell wall rigidity in a yeast mutant lacking chitin deacetylase activity and by the lax epidermal cuticle in serp verm double mutants (Luschnig, 2006).

The last two postulates of the model are more speculative. It is not known if or how the chitin cylinder is attached to the apical cell surface. There is a characteristic ~0.5 μm gap that is between the cell surface and the chitin cylinder and is visible in fixed specimens and could contain an anchoring complex. The gap is eliminated in serp verm double mutants. How the proposed link between chitin and the apical cell surface, or a signal generated by this complex, limits apical-membrane biogenesis along the longitudinal axis of the tube is even more obscure. Perhaps it exerts a mechanical effect on the apical cell surface, physically constraining apical membrane elongation, or maybe it influences the distribution or activity of apical-basal cell-polarity regulators such as Crumbs. Whatever the mechanism, the genetic results make clear that absence of Serp and Verm proteins results in a dramatic expansion of the lumen and apical tracheal surface -- but only along the longitudinal axis of the tube (Luschnig, 2006).

Longitudinal growth of tracheal tubes normally occurs gradually and continuously during development, beginning soon after the tubes form during embryogenesis and continuing throughout larval life. This allows the size and transport capability of the tracheal network to keep pace with the increasing oxygen demand of the growing larva. Because this growth occurs in the absence of tracheal cell division and is only periodically interrupted by a burst of radial growth, it must involve the gradual and continuous expansion of tracheal cells specifically in the longitudinal axis of the tube. This is reminiscent of the effects of serp and verm mutations in the embryo, except that in the mutants, tube elongation occurs more precipitately than normal. Because the serp and verm genes are expressed early and broadly in the developing tracheal system, under control of Ribbon and possibly Trachealess, they could act as governors on tube growth from the onset, keeping tube elongation in check. Controlled downregulation of Serp and Verm expression or activity during development could gradually release this constraint and give rise to the controlled longitudinal tube growth that is normally observed. Septate-junction mutants have a tube-elongation phenotype similar to that of serp and verm mutants, so septate junctions could function to promote Serp and Verm expression or activity or to antagonize the negative regulatory pathway (Luschnig, 2006).

GENE STRUCTURE

cDNA clone length - 3535bp for serp

Bases in 5' UTR - 535 for serp

Exons - 4 for serp; 6 for verm

Bases in 3' UTR - 374 for serp

PROTEIN STRUCTURE

Amino Acids - 541 for Serp; 525 for Verm

Structural Domains

serp and verm encode similar proteins. Both have a predicted N-terminal signal peptide, a peritrophin-A-like chitin binding domain (CBD), a single type-A LDL-receptor ligand binding domain repeat (LDLa), and a polysaccharide deacetylase domain. The deacetylase domains show similarity to the NodB domain, which is shared by a group of bacterial and fungal enzymes with chitin deacetylase (CDA) activity. CDAs modulate the physical and chemical properties of chitin by deacetylation of the β,1-4 N-acetyl-D-glucosamine polymer, which converts chitin into chitosan (Tsigos, 2000). Biological functions of CDAs and chitosan (quantitatively deacetylated chitin) have been characterized in bacteria and fungi but have not been previously described in animals. The Serp and Verm protein family was called ChLD, on the basis of the predicted domain structure (chitin binding, LDL receptor ligand binding, chitin deacetylase). The same domain organization, except for a predicted signal peptide, is shared by the product of one other D. melanogaster gene, which is called Chld3 (CG17905, located at cytological position 36A13-14. Chld3 mRNA was not detected in the embryonic tracheal system, but was detected in the epidermis at stage 16. Related proteins are found in other insects and in C. elegans (Luschnig, 2006).

date revised: 30 May 2006

Home page: The Interactive Fly © 2006 Thomas Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.