BIOLOGICAL OVERVIEW

Border cells are a small group of specialized follicle cells in the ovary that undergo a dramatic cell migration during stage 9 of oogenesis. During stage 9 there is a posteriorward movement of the outer layer of follicle cells; eventually 95% of the follicle cells stack up in the posteior half of the egg chamber, in contact with the oocyte. The remaining cells stretch to cover the nurse cell cluster, anterior of the oocyte.

During this rearrangement of cells, six to ten follicle cells, referred to as border cells, remain rounded at the anterior tip of the egg chamber. One or two border cells extend cytoplasmic processes in between the anterior nurse cells, initiating their own unique migratory journey. Over the course of several hours, they move posteriorly, traversing a distance of over 15 cell diameters, passing several nurse cell junctions, making correct choices at each juncture along the way. The migration terminates once they reach the oocyte-nurse cell boundary. At this time, the anteriormost oocyte-associated follice cells begin moving inward to cover the anterior end of the oocyte, eventually joining the border cells at the anterior end.

The slbo gene was discovered in a brute force experiment. 7800 lines of flies, made mutant by the random insertion of a P element vector in the fly genome, were examined for border cell defects. One strain was found, and as luck would have it, the P element, designed to drive ß-galactosidase expression when inserted into an expressed gene, was capable of driving ß-galactosidase in border cells, beginning just prior to their migration. The P element was used to clone the gene into which it had inserted. In this manner the slbo gene was characterized. Slow border cells also express the novel protein torso-like, a maternal gene involved in determination of terminal pattern elements (Montell, 1992).

SLBO is the Drosophila homolog of vertebrate C/EBP, the CCAAT/enhancer-binding protein, a transcriptional activator of a group of adipose-specific genes. There is evidence that a basic-leucine zipper protein has a similar function in Drosophila: a bZIP site is involved in the regulation of yolk protein (Yp) genes at a transcriptional enhancer that regulates sex- and tissue-specific transcription. Such regulation is complex, involving interaction between a bZIP protein and Doublesex, a sex specific transcription factor. Refusing any neat or simple characterization, SLBO does not appear to be the bZIP protein that activates Yp genes (An, 1995).

However, SLBO does play a role in regulation of transcriptional control of breathless, coding for an FGF receptor homolog involved in the developmental control of cell migration. During embryonic development btl expression registers in several locations: in cells of the tracheal system, in a subset of glial cells and in salivary duct cells. Mutations in the btl locus cause defects in the migration in a pair of midline glia and in the migration of tracheal cells. SLBO binds to eight sites in the btl regulatory sequence, suggesting that its regulation of btl is direct. Most likely additional SLBO target genes contribute to efficient and complete border cell migration: a null allele in btl does not eliminate border cell migration, as does the loss or even a reduction in SLBO expression (Murphy, 1995).

SLBO is unlikely to regulate breathless expression during embryogenesis because SLBO expression in the tracheal system does not begin until long after breathless expression (Rorth, 1992). Recently however, a POU domain transcription factor, Drifter, has been described which may enhance btl expression in tracheal cells. Drifter protein is expressed in tracheal cells near the time that btl expression initiates: the dfr mutant phenotype is similar to btl; and dfr expression is not altered in btl mutants (Anderson, 1995). Thus it is possible, even likely, that dfr regulates btl expression. Preliminary experments suggest that dfr is not expressed in the border cells. One interpretation then, is that DFR may regulate btl in the embryo in much the same way that SLBO does in the ovary (Murphy, 1995).

A regulatory circuit between STAT, APT, and SLBO functions to convert an initially graded signal into an all-or-nothing activation of JAK/STAT and thus to proper cell specification and migration

In both normal development and in a variety of pathological conditions, epithelial cells can acquire migratory and invasive properties. Border cells in the Drosophila ovary provide a genetically tractable model for elucidating the mechanisms controlling such behaviors. An apontic (apt) mutant has been identified in which the migratory population expands and separation from the epithelium is impeded. This phenotype resembles gain-of-function of JAK/STAT activity. Gain-of-function of APT also mimics loss of function of STAT and its key downstream target, SLBO. APT expression is induced by STAT, which binds directly to sites in the apt gene. The data suggest that a regulatory circuit between STAT, APT, and SLBO functions to convert an initially graded signal into an all-or-nothing activation of JAK/STAT and thus to proper cell specification and migration. These findings are supported by a mathematical model, which accurately simulates wild-type and mutant phenotypes (Starz-Gaiano, 2008).

In many migratory cell types, including metastatic carcinomas, motile cells must detach from an epithelium to move to their final location. However the precise mechanisms by which cells disengage from their neighbors remain poorly understood, and in most cases it is not possible to view the process directly in vivo. Border cells in the Drosophila ovary represent a model for studying epithelial cell migration in vivo that is amenable both to genetic approaches and live imaging. This study reports the identification of a mutant, apt, in which the distinction between invasive and noninvasive cells was compromised. In apt mutants, as the border cell cluster moved away from the epithelium, additional migratory cells -- the stretched border cells -- ingressed in between the nurse cells. These stretched border cells maintained connections with both the main cluster of border cells, and the outer epithelial cell layer, resulting in a defect in detachment (Starz-Gaiano, 2008).

Recent technological advances have enabled analysis of border cells throughout their six hour migration by live imaging. Time-lapse movies of wild-type egg chambers revealed that the process of border cell detachment is surprisingly slow. This indicates that the ability to extend and retract protrusions is not sufficient for the border cells to exit the epithelium, and that there is sufficient time for transcriptional events to contribute to the process. In apt mutants, the border cells rounded up and advanced in between the nurse cells normally, but cells with an apparently intermediate identity were frequently trapped in between the border cell cluster and the follicle cell epithelium, unable to detach from either one. Thus, the two cell types must be clearly distinguished in order for them to be able to disconnect from one another (Starz-Gaiano, 2008).

In a variety of contexts throughout development, a graded distribution of a signaling molecule in a field of cells can elicit discrete cellular responses. Such threshold-like behavior can be achieved by positive autoregulation. Therefore, prior to the current work, it would have been reasonable to propose that STAT autoregulation could convert initially graded activity in the follicular epithelium to 'on' and 'off' states. In wild-type, the migrating border cell cluster takes the source of JAK/STAT activation (UPD expressed by the polar cells) with it, reinforcing SLBO expression in the migratory cells and removing the source of JAK/STAT activation from the anterior follicle cells. So, one could have postulated that the physical separation of the JAK/STAT signaling center from the anterior follicle cells was sufficient to create a significant difference in levels of STAT activity between the migrating cells and those left behind, and thus to distinguish the two cell types and behaviors. However, unexpectedly this study showed that neither STAT autoregulation nor the movement of the signaling center is sufficient to convert the gradient into a step function in the absence of APT (Starz-Gaiano, 2008).

It is proposed instead that feedback inhibition of JAK/STAT combined with the mutual repression of APT and SLBO is responsible for generating the stepwise activation pattern. When two genes mutually repress each other, a slight increase in the activation of one leads to a stronger repression of the second, which, in turn, leads to a further increase of the first. Thus, together these two genes behave as an autocatalytic system. Since apt and slbo are both targets of STAT activity, a three-component regulatory circuit is proposed. The mathematical model demonstrates that this circuit is sufficient to explain what is observed in vivo. In the absence of APT, JAK/STAT activation takes place in an enlarged region and, remarkably, the 'on-off' character of the JAK/STAT activation is lost. This suggests that the threshold behavior of the system does not result from JAK/STAT autoregulation but from the mutual repression of APT and SLBO (Starz-Gaiano, 2008).

The model that most accurately simulates the wild-type and mutant phenotypes is one in which SLBO antagonizes APT activity more strongly than its expression. This is consistent with experimental observation that overexpression of SLBO completely mimics the apt loss-of-function phenotype, but only reduces and does not eliminate APT expression (Starz-Gaiano, 2008).

It is striking that different patterns of SLBO and APT are induced by the same gradient of JAK/STAT activity. An important consequence is that, at high concentrations of active STAT, more SLBO is produced than APT. One way this could be explained is through the observation that STAT binds four different sites in the slbo enhancer with differing affinities. In cells with high concentrations of STAT, more sites, including low affinity sites, should be occupied and thus a higher level of slbo expression generated than in cells with lower STAT levels. In contrast, the apt gene contains only two detectable STAT binding sites, to which STAT can bind nearly as well as it binds the optimal STAT consensus sequence. Thus, apt expression should turn on in response to lower levels of active STAT than slbo and also should saturate at relatively low concentrations of active STAT, yielding a broad and shallow expression gradient across the anterior field of follicle cells. These are precisely the expression patterns observed. Therefore, in cells adjacent to the polar cells, SLBO wins the competition whereas further away from the source of UPD, APT wins the APT/SLBO competition. Higher levels of SLBO block the repression of JAK/STAT by APT in the cells next to the polar cells, causing an even stronger JAK/STAT activation and so on (Starz-Gaiano, 2008).

In addition, evidence was found for a low level of JAK/STAT-independent APT expression, which was also incorporated into the model. This baseline APT expression depended on the transcription factor known as Eyes absent, and based on the model it is proposed that its function could be to prevent any possibility of a renewed trigger of the JAK/STAT pathway in the cells that remain in the anterior epithelium (Starz-Gaiano, 2008).

The JAK/STAT pathway is highly conserved from insects to mammals and is critically important in development, immunity, and inflammation. Intriguingly, Drosophila APT is expressed in many domains where JAK/STAT signaling occurs, including embryonic trachea and the hub of the testes. In addition, apt has been uncovered as a downstream target of STAT in microarray analysis of testis and border cells. Therefore, apt may be a downstream target of STAT signaling in a variety of cell types (Starz-Gaiano, 2008).

It is also possible that this relationship is conserved in other animals, since genes highly related to apt are found in all sequenced insect genomes. In humans, the closest gene to apt is fibrinogen silencer-binding protein (FSBP). Interestingly, two strong loss-of-function alleles of apt contain missense mutations in well-conserved residues, demonstrating the functional significance of this region. Although FSBP has not been extensively characterized, it has been reported to be a negative regulator of the gamma chain of fibrinogen transcription. Fibrinogen is highly expressed in hepatocytes in response to inflammatory cytokine-mediated activation of the JAK/STAT pathway, and there are at least three STAT3 binding sites on the human gamma-fibrinogen promoter. This suggests that APT and FSBP could fulfill similar functions as negative regulators of STAT-responsive genes (Starz-Gaiano, 2008).

All of the major growth factor and cytokine signaling pathways are subject to extensive positive and negative feedback regulation, which is crucial to generate appropriate physiological responses. The work presented here establishes APT as a feedback inhibitor of JAK/STAT signaling and cell invasion (Starz-Gaiano, 2008).

GENE STRUCTURE

There are two major transcription start sites of SLBO mRNA: P1 (distal) and P2 (proximal) (Rorth, 1992).

Exons - 1

PROTEIN STRUCTURE

Amino Acids - 444

Structural Domains

The C-terminal region of SLBO contains a basic region followed by a leucine zipper domain. The leucine zipper domains of the mammalians C/EBPs are less conserved than are the basic regions. 19 of 21 amino acids of the basic region are identical. The leucine zipper domain shows only 26% identity to mammalian C/EBP alpha. The leucine zipper functions to stabilize C/EBP dimerization. The only N-terminal region with an identifiable sequence is a stretch of 23 glutamines encoded for by an opa (CAG/A) repeat (Montell, 1992).

date revised: 25 July 2022

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.