Gene name - 14-3-3zeta
Synonyms - leonardo, Par-5
Cytological map position - 46E1--46E9
Function - phosphoserine/threonine interacting protein
Symbol - 14-3-3zeta
Genetic map position - 2-
Classification - 14-3-3-protein
Cellular location - cytoplasmic
|Recent literature||Le, T.P., Vuong, L.T., Kim, A.R., Hsu, Y.C. and Choi, K.W. (2016). 14-3-3 proteins regulate Tctp-Rheb interaction for organ growth in Drosophila. Nat Commun 7: 11501. PubMed ID: 27151460
14-3-3 family proteins regulate multiple signalling pathways. Understanding biological functions of 14-3-3 proteins has been limited by the functional redundancy of conserved isotypes. This study provides evidence that 14-3-3 proteins regulate two interacting components of Tor signalling in Drosophila, translationally controlled tumour protein (Tctp) and Rheb GTPase. Single knockdown of 14-3-3ɛ or 14-3-3ζ isoform does not show obvious defects in organ development but causes synergistic genetic interaction with Tctp and Rheb to impair tissue growth. 14-3-3 proteins physically interact with Tctp and Rheb. Knockdown of both 14-3-3 isoforms abolishes the binding between Tctp and Rheb, disrupting organ development. Depletion of 14-3-3s also reduces the level of phosphorylated S6 kinase, phosphorylated Thor/4E-BP and cyclin E (CycE). Growth defects from knockdown of 14-3-3 and Tctp are suppressed by CycE overexpression. This study suggests a novel mechanism of Tor regulation mediated by 14-3-3 interaction with Tctp and Rheb.
14-3-3 proteins carry out numerous physiological functions in a variety of species. Leonardo is one of two known Drosophila isoforms. In vitro the 14-3-3 proteins activate tyrosine and tryptophan hydroxylases, the rate-limiting enzymes in the biosynthesis of catecholamines and serotonin respectively. They function in Ca2+ regulated exocytosis, cell cycle control and in protein kinase C (PKC) regulation. They associate physically with PI-3 kinase, CDC25 phosphatase (see Drosophila String), and RAF-1 kinase. The association with this apparently diverse group of proteins may be due to the intriguing ability of 14-3-3 to bind to residues sites phosphorylated on serine or threonine residues. Their interaction with diverse signaling proteins suggests that these proteins function as modulators of activity or specificity of various kinases, or as coordinators for the assembly of signaling complexes for different cascades (Sokoulakis, 1996).
The two Drosophila 14-3-3 isoforms, Leonardo (14-3-3zeta) and 14-3-3eta, have been shown to positively regulate Ras-mediated signaling in the development of the compound eye. 14-3-3eta was identified as a suppressor of activated Ras1 in eye development. Suppressor of Ras1 3-9 (SR3-9) alleles act as dominant suppressors of sev-Ras1, that is, of Ras1 expressed ectopically in the eye disc using a sevenless promoter, a treatment that produces a rough eye phenotype. The eyes of flies carrying SR3-9 are less rough than those of flies expressing sev-Ras1. It is known that SR3-9 functions either downstream of or in parallel to Raf kinase, since SR3-9 dominantly suppresses the rough eye defect caused by an activated raf expression construct (Chang, 1997).
Leonardo exhibits similar involvement in the Raf/Ras pathway. Clones of mutant leonardo show a loss of photoreceptors. Ommatidia lack outer as well as inner photoreceptors. This phenotype is reminiscent of clones homozygous for Drosophila ras or raf hypomorphic alleles. When Leonardo antisense RNA is expressed in eye imaginal discs in postmitotic photoreceptors, a weakly penetrant but reproducible loss-of-photoreceptor phenotype results. Since the artificial activation of Raf rescues the nonviability caused by leonardo mutation and permits photoreceptor development, it has been concluded that leonardo acts downstream of Ras and upstream of Raf in the signaling pathway that controls cell proliferation in the Drosophila eye imaginal disc (Kockel, 1997).
Of particular interest is the involvement of leonardo in facilitating olfactory learning in the fly. leonardo was identified in an enhancer detector screen for genes preferentially expressed in mushroom bodies of the brain: a lacZ reporter construct was found to have been inserted in a gene expressed in mushroom bodies, the seat of olfactory learning in insects. Isolation of multiple insertion strains was carried out to obtain viable leonardo mutant alleles. All the viable alleles were found to be confined to genomic excisions of insertions in the first intron. Since introns do not code for the primary sequence of proteins but instead are involved in splicing or carry regulatory sequences, this suggests that these excisions disrupt regulatory elements necessary for mushroom body expression or RNA splicing (Skoulakis, 1996).
To examine associative learning in these flies, a Pavlovian olfactory conditioning assay was used (see Dunce for a more complete description of this procedure). Briefly, flies are trained by exposure to electroshock paired with one odor (octanol or methylcyclohexanol) and subsequently exposed to a second odor without electroshock. Immediately after training, learning is measured by forcing flies to choose between the two odors used during training. No preference between odors results in a performance index of zero (no learning), as is the case for naive flies. Avoidance of the odor previously paired with electroshock, however, yields a performance index greater than O. Alleles of leonardo classed as modest by immunohistochemical criteria (meaning that there is a decrease in Leonardo protein relative to the controls, but that some remains) exhibit a 20-25% decrease in 3 minute memory (as compared to non-mutant controls), whereas alleles nearly devoid of Leo in mushroom bodies exhibit a increased decrement of 30-35% relative to the controls. It is concluded that Leo is involved in olfactory learning in flies (Skoulakis, 1996).
Longer-term memory of the conditioned association was assessed for a selected set of mutants at 45, 90 and 240 min after training. One excision mutant exhibited a 30% reduction in 3 minute memory and retained this highly significant difference over time through four hours. Other leo insertion mutants exhibit significant differences for initial memory and at 45 min, although their performance at 90 and 240 min was not significantly different from that of controls. As reduction in leonardo expression does not precipitate developmental abnormalites in the brain anatomy, it is concluded that the behavioral deficits observed do not have an anatomical basis, but instead involve deficits in signal processing (Skoulakis, 1996).
Since 14-3-3 proteins are not known to participate in the cAMP cascade, the results suggest that Leo protein is a member of an additional signaling cascade that mediates learning and memory. This function of Leonardo is consistent with its role in protein kinase C-mediated processes or alternatively in an interaction with Raf-1 in the mitogen-activated protein kinase (MAPK) signal transduction cascade (Skoulakis, 1996). How does it happen that the reduction of the leonardo gene product does not cause proliferation or differentiation defects in the mushroom bodies? leonardo codes for an adult-specific 2.9-kb splicing variant that is strongly enriched or even exclusively expressed in the head. Thus, it might be possible that the different splice variants operate in different signal transductoin pathways not necessarily linked to Raf function. Alternatively, the strong reduction in leonardo expression may affect only acquisition of memory, whereas low residual levels of expression are sufficient to mediate the cellular aspects of differentiation and proliferation, processes that are severely affected in the lack-of-function mutations (Kockel ,1997).
The functional specialization or redundancy of the ubiquitous 14-3-3 proteins constitutes a fundamental question in their biology and stems from their highly conserved structure and multiplicity of coexpressed isotypes. This question was addressed in vivo using mutations in the two Drosophila 14-3-3 genes, leonardo (14-3-3ζ) and D14-3-3ε. 14-3-3ε is essential for embryonic hatching. Nevertheless, D14-3-3ε null homozygotes survive because they upregulate transcripts encoding the LeonardoII isoform at the time of hatching, compensating D14-3-3ε loss. This novel homeostatic response explains the reported functional redundancy of the Drosophila 14-3-3 isotypes and survival of D14-3-3ε mutants. The response appears unidirectional, since D14-3-3ε elevation upon Leo loss was not observed and elevation of leo transcripts was stage and tissue specific. In contrast, Leo levels are not changed in the wing discs, resulting in the aberrant wing veins characterizing D14-3-3ε mutants. Nevertheless, conditional overexpression of LeoI, but not of LeoII, in the wing disk can partially rescue the venation deficits. Thus, excess of a particular Leo isoform can functionally compensate for D14-3-3ε loss in a cellular-context-specific manner. These results demonstrate functional differences both among Drosophila 14-3-3 proteins and between the two Leo isoforms in vivo, which likely underlie differential dimer affinities toward 14-3-3 targets (Acevedo, 2007).
A fundamental issue concerning members of highly conserved protein families is the extent to which they are functionally redundant or exhibit specialized biological functions. The 14-3-3 proteins compose a highly conserved family of acidic molecules present in all eukaryotes. 14-3-3's share a common structure composed of nine antiparallel α-helices forming a horseshoe shape with a negatively charged interior surface. Interactions among particular amino acids in the first helix, with ones in helix 2 and helix 3 of another monomer, promote dimerization. Dimerization generates a tandem binding surface, which can simultaneously bind to one or two sites on one target protein or to sites on two different client molecules. The dimers bind clients containing phosphoserine- or phosphothreonine-containing motifs via highly conserved amino acids within the groove. 14-3-3 proteins can also bind targets with surfaces outside the conserved phosphopeptide-binding cleft. 14-3-3 binding may allosterically stabilize conformational changes, leading to activation or deactivation of the target or to interaction between two proteins. Furthermore, 14-3-3 binding may mask or expose interaction sites, often leading to changes in the subcellular localization of client proteins (Acevedo, 2007 and references therein).
An extraordinary feature of this protein family is the high sequence conservation among isotypes, characterized by long stretches of invariant amino acids, suggesting functional redundancy. However, despite this extensive sequence identity, multiple 14-3-3 proteins exist in metazoans, indicating at least some functional specificity. Vertebrates contain seven distinct protein isotypes, β, ε, ζ, γ, η, θ, and σ. In vertebrate brains where these proteins are highly abundant, there is some specificity in isotype distribution, but generally 14-3-3's are expressed in complex overlapping patterns. In addition, multiple heterodimers are possible in tissues that contain more than one isotype. It is unclear whether the presence of multiple highly similar proteins with overlapping distribution reflects functional differences among them or represents a mechanism to ensure that ample functionally redundant 14-3-3's are available to mediate the multiple essential cellular functions that require them. Thus, the question of 14-3-3 functional specificity in vivo is fundamental in understanding their biology. The highly overlapping isotype distribution in vertebrate models hinders systematic investigation of this question (Acevedo, 2007).
To address the issue of functional specificity in vivo, Drosophila, which offers a simple, but representative, genetically tractable metazoan system, was used. It is simple because it contains only two 14-3-3 genes, an ortholog of the mammalian 14-3-3ζ (88% identity) leonardo and an ortholog of the ε isotype, D14-3-3ε. It is representative because the two fly genes belong to the two different 14-3-3 conservation groups. leonardo encodes two nearly identical protein isoforms (Leo I and Leo II) via alternative splicing of the primary transcript, with modest tissue specificity. In contrast, D14-3-3ε encodes a single protein, apparently present in all developmental stages and tissues examined with only slight enrichment in the adult brain (Acevedo, 2007 and references therein).
Maternal Leo is required for normal chromosome separation during syncytial mitoses, whereas D14-3-3ε appears required to time them, suggesting distinct functions for the two 14-3-3's in the single-celled syncytial embryo. Maternal Leo is also essential for early Raf-dependent decisions that pattern the embryo. Zygotic leo loss-of-function mutants exhibit functional impairments of their embryonic and adult nervous system. D14-3-3ε functions in photoreceptor formation and appears involved in development of the wing, but whether it is important for the function of the nervous system is unknown. Leo and D14-3-3ε appear at least partially redundant for photoreceptor formation. Furthermore, Leo and D14-3-3ε have been reported to function redundantly in anterior–posterior axis formation of the developing oocyte (Benton, 2002) and follicle cell polarity (Acevedo, 2007 and references therein).
Nevertheless, three reasons motivated a systematic investigation of potential functional specificity of the two Drosophila 14-3-3 isotypes by searching for isotype-specific phenotypes. First, studies to date used a transposon allele of D14-3-3ε (j2B10), which may not be a null allele. In fact, although D14-3-3ε has been reported dispensable for viability, a lethal deficiency uncovering this gene was used to show its involvement in Raf-mediated developmental processes in the embryo. Second, leo mutations are homozygous lethal, suggesting that D14-3-3ε cannot functionally compensate for its loss, although Leo was suggested to at least partially compensate for the lack of D14-3-3ε in embryonic development. Third, the dynamic expression pattern of 14-3-3's during embryonic development and larval and adult nervous systems suggested involvement in additional processes other than photoreceptor and oocyte development, which may specifically require one but not the other. The results demonstrate 14-3-3-isotype-specific functions and a tissue- and temporal-specific transcriptional mechanism to compensate for loss of D14-3-3ε and suggest dynamic temporal and spatial interactions of the two 14-3-3 isotypes (Acevedo, 2007).
The results utilizing null alleles indicate that D14-3-3ε is not dispensable for viability, but its loss is partially compensated by elevation of endogenous leo levels. Consequently, homozygotes survive to adulthood, whose number is higher when the hypomorphic allele D14-3-3εl(3)j2B10 is used. This is the likely reason for the suggestion of previous reports that the gene is not essential. This interaction is uncovered genetically by the inability to obtain D14-3-3εex4 and D14-3-3εl(3)j2B10 homozygotes when one copy of leo is mutated (i.e., leoP1188/+; D14-3-3εl(3)j2B10/D14-3-3εl(3)j2B10 animals). Embryos homozygous for mutant alleles do not exhibit obvious morphological defects because maternally provided D14-3-3ε is likely sufficient to fulfill its requirement in syncytial cellular blastoderm and gastrulating animals). D14-3-3ε mutant homozygotes die ostensibly because lack of zygotic protein from the nervous system renders them unable to hatch. Similarly, Leo accumulates in embryonic motor neurons innervating the body-wall musculature and its loss in leo mutants is the likely reason for their failure to hatch despite their apparently normal progression through development (Acevedo, 2007).
The results demonstrate that LeoII overaccumulates in late D14-3-3ε null embryos and that this elevation allows a fraction of them to hatch and survive. The conclusion is supported by the striking increase in the number of D14-3-3ε mutant homozygotes that survive upon expression of leoII transgenes in the nervous system. Because endogenous LeoII accumulates preferentially in the CNS, the data suggest that its elevation in this tissue leads to successful hatching and survival of D14-3-3ε mutant homozygotes. This 'homeostatic' response in D14-3-3ε mutants is specific to late embryogenesis after the maternally supplied D14-3-3ε, which perdures almost until stage 8, has decayed. Therefore, the response appears specific to a period when the overall level of either 14-3-3's or D14-3-3ε, specifically, is critically important for survival (Acevedo, 2007).
It appears that a mechanism sensing the absence of D14-3-3ε operates in embryos and responds by increasing the level of LeoII. Congruent with this, Leo elevation was not observed in embryos homozygous for the dominant-negative allele D14-3-3εE183K, which compromises D14-3-3ε functionally, but does not change its overall levels in the embryo. It is possible that this is the reason that D14-3-3εE183K homozygotes are never recovered. Furthermore, this response appears specific to the loss of D14-3-3ε, because levels of this protein remained normal in homozygous leo mutant embryos and adults, consistent with their strong lethal phenotype. Clearly, this sensing mechanism responds by increased accumulation of leoII transcripts by preferential utilization of one of two possible promoters and splicing of the primary transcript to include the leoII-specific exon 6′. How is lack of D14-3-3ε sensed and how could leoII transcription be increased? 14-3-3's have been reported to participate in nuclear/cytoplasmic trafficking of transcription factors. Therefore, it is possible that loss of D14-3-3ε enhances transcription from the proximal promoter of the leo gene by not mediating nuclear export of a factor that binds that site. Alternatively, D14-3-3ε may be part of a repressing complex and, upon its loss, transcription from this site is enhanced. In contrast, excessive transgenic elevation in the amount of D14-3-3ε results in recovery of few adults (<10% of expected) homozygous for strong hypomorphic leo mutations. This suggests that although D14-3-3ε can at least partially compensate for the loss of Leo in high concentrations, an endogenous molecular mechanism to elevate it in leo homozygotes does not appear to exist (Acevedo, 2007).
Although leoI transcripts accumulate in the wing disc, Leo does not appear to play a role in wing-vein formation because animals that develop with as low as 10% of normal Leo do not exhibit wing aberrations. Therefore, the venation deficits are a phenotype specific to D14-3-3ε mutant homozygotes. Interestingly, in congruence with the mechanism proposed above, the leoI transcripts normally expressed in that tissue were not upregulated and leoII transcripts were not ectopically transcribed in D14-3-3ε mutant homozygote wing disks. This is because the proposed D14-3-3ε-interacting factor(s) required for exon 1′-containing leoII transcription are likely absent from the wing disk where these transcripts do not normally accumulate. Exon 1-containing leoII transcripts do not appear to require such D14-3-3ε-interacting factor(s), since these transcripts were not upregulated in embryos. Therefore, loss of D14-3-3ε does not alter the tissue specificity of leo transcriptional regulation and specific isoform accumulation (Acevedo, 2007).
It is presently unclear whether this compensatory mechanism is operant in other systems where mutant analyses of 14-3-3's have been initiated. Interestingly, single nulls of either 14-3-3-encoding gene in Saccharomyces cerevisiae are viable, while the double mutant is lethal and similar results were obtained for the two Schizosaccharomyces pombe genes. These observations may reflect similar 14-3-3 'homeostatic' mechanisms in these species. Directed reduction of specific 14-3-3 protein levels during Xenopus laevis development yielded gastrulation and patterning defects for all proteins tested except for 14-3-3ζ. Unlike Drosophila, Xenopus 14-3-3ζ may not be essential for development, but it is also possible that loss of this isotype is specifically compensated for by elevation of the remaining 14-3-3's. Such mechanisms, if extant in mammals, are likely to hinder genetic analysis of 14-3-3 function, especially in the brain where all family members are expressed. Interestingly, mice mutant for 14-3-3ε exhibit severe brain abnormalities and die perinatally, yet a small fraction survive to adulthood appearing smaller, but otherwise normal, much like the Drosophila mutants. It is unknown whether 14-3-3ζ or other isotypes are elevated in these animals as predicted by the current results (Acevedo, 2007 and references therein).
Functional specificity of 14-3-3 family members may be the result of tissue or temporal-specific gene expression and regulation or of isotype-specific ligand selectivity. Isotypes may have redundant functions within a cell if they are able to interact with the same targets. Even then, affinity differences toward common ligands predicted by their amino-acid sequence and tertiary structure may functionally differentiate coexpressed 14-3-3's (Acevedo, 2007).
Although both leoI and leoII transgenes rescued the lethality of D14-3-3ε mutants, they clearly exhibited different efficiency. Rescue was invariably higher upon accumulation of LeoII either ubiquitously or specifically in the nervous system. However, rescue required excessive accumulation of LeoII to overcome loss of D14-3-3ε. In fact, the two- to threefold Leo elevation could be as much as a 50%-60% underestimate of the level of this protein in D14-3-3ε mutant embryos that hatch. Therefore, a large excess of Leo appears to be necessary to functionally substitute D14-3-3ε in the embryonic nervous system, which may be attained only in a small number of mutant homozygotes. This probably reflects the affinity differences that Leo dimers exhibit toward client proteins normally bound either by D14-3-3ε homodimers or by D14-3-3ε/Leo heterodimers. If so, then even a small amount of D14-3-3ε would increase the number of mutant homozygotes obtained. In agreement with this, more homozygotes were recovered from the transposon allele D14-3-3εl(3)j2B10, which likely contains residual D14-3-3 ε. Differences in ligand binding between LeoI and LeoII are likely reflected in the large difference with which the two isoforms rescue the lethality of D14-3-3ε mutants. This is the first unequivocal demonstration of functional differences between LeoI and LeoII. These differences must reside in the five unique amino acids of helix 6 that distinguish the two isoforms. It is unknown whether LeoI, LeoII, or both contribute to the reported redundancy with D14-3-3ε in photoreceptor development and oocyte polarity (Acevedo, 2007).
Interestingly, the functional redundancy of Leo isoforms with D14-3-3ε is tissue specific. In contrast to the embryonic nervous system, enhanced accumulation of LeoI, and not of LeoII, was able to compensate for anterior cross-vein deficits and partially for the posterior cross vein. Again, this suggests that D14-3-3ε ligands in the wing disk necessary for cross-vein formation can be targeted by excess LeoI (and not LeoII). Hence, LeoII can be redundant with D14-3-3ε specifically in the embryonic nervous system where it is presumed to accumulate preferentially, while in the wing disk LeoI, which is normally found in this tissue, is the potential compensating isoform. Similarly, although the two S. cerevisiae 14-3-3 genes are functionally redundant for viability, only one, Bmh1p, is required for efficient forward transport to the endoplasmic reticulum. Thus, redundancy of 14-3-3's largely depends on the specific function and interacting proteins that they engage within a particular tissue or developmental context (Acevedo, 2007).
Collectively, these data show a tissue- and temporal-specific upregulation of leo transcription that can account for the apparent functional redundancy between Leo and D14-3-3ε with respect to the lethality of D14-3-3ε mutants and possibly other processes requiring these proteins. In addition, this analysis for the first time demonstrates tissue and temporal functional differences between the two Leo isoforms. Whether these functional differences and the functional redundancy among the Drosophila 14-3-3's will also occur in the adult nervous system where they are all most abundant is currently unknown. A previous study failed to uncover differences between LeoI and LeoII with respect to learning and memory). Nevertheless, the data strongly support the notion that the existence of multiple 14-3-3 isotypes in metazoans reflects a combination of tissue- and temporal-specific isotype expression, localization, and functional specialization. Importantly, this analysis indicates that understanding the biological roles of 14-3-3's will require identification of proteins engaged by homo- and heterodimers of particular composition in a tissue- and temporal-specific manner (Acevedo, 2007).
Five different types of Leonardo cDNAs have been isolated. The transcripts differ by the alternative use of exons I or I' and VI or VI' as well as by use of three different alternative sites of addition of poly(A) onto the 3' terminus of the mRNA. Switching of exons VI and VII' affects the open reading frame and causes sequence differences in the respective translation products (Kockel, 1997).
Bases in 3' UTR - 309 (1.3 kb transcript) (Swanson, 1992)
A cloned 1.3-kb cDNA hybridizes to genomic clone 549, containing genes predominantly expressed in the head of Drosophila melanogaster. DNA sequencing shows that the cDNA-encoded protein is similar to a family of mammalian proteins, called 14-3-3, which activate tyrosine hydroxylase (TyrOHase) and tryptophan hydroxylase (TrpOHase), the two key enzymes regulating biosynthesis of biogenic monoamine neurotransmitters, such as dopamine and serotonin, in the brain. The putative D. melanogaster 14-3-3 protein (D14-3-3) shares 72.4, 74.3 and 78.3% amino acid (aa) sequence identity and 83.5, 87.7 and 85.9% aa sequence similarity with the beta, gamma and eta forms of bovine 14-3-3 protein, respectively. A lower (71%), but significant level of aa sequence identity was also found between D14-3-3 and sheep brain protein kinase C inhibitor protein (KCIP) (Swanson, 1992).
The variant amino acids caused by the switching of exons VI and VI' are predicted to lie in alpha-helix 6 on the outside of the groove-shaped 14-3-3 dimer. Interestingly, helix 6 is composed of the sequences that are least conserved throughout the 14-3-3 protein family. This suggests that helix 6 might confer specificity to 14-3-3 interactions with target proteins. Both exons VI and VI' encode a potential phosphorylation site characterized previously in mammalian 14-3-3 beta and zeta (Kockel, 1997).
Multiple alignments were constructed from forty-six 14-3-3 sequences retrieved from the GenBank and SwissProt databases. The alignments constructed reveal five highly conserved sequence blocks. Blocks 2-5 correlate well with the alpha helices 3, 5, 7, and 9, which form the proposed internal binding domain in the three-dimensional structure model of the functioning dimer. Amino acid differences within the functional and structural domains of plant and animal 14-3-3 proteins were identified, which may account for the functional diversity among isoforms. Epsilon isoforms from the animal lineage form a distinct grouping, which suggests an early divergence from the other animal isoforms. Epsilons were found to be more similar to yeast and plant isoforms than other animal isoforms at numerous amino acid positions, and thus epsilon may have retained functional characteristics of the ancestral protein. The known invertebrate proteins are most properly grouped with the nonepsilon mammalian isoforms. Most of the current 14-3-3 isoform diversity probably arose through independent duplication events after the divergence of the major eukaryotic kingdoms. Divergence of the seven mammalian isoforms beta, zeta, gamma, eta, epsilon, tau, and sigma occurred before the divergence of mammalian and perhaps before the divergence of vertebrate species. A possible ancestral 14-3-3 sequence has been proposed (W. Wang, 1996).
date revised: 4 June 97
Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.
The Interactive Fly resides on the
Society for Developmental Biology's Web server.