One of the factors which saw the successful assimilation of the yeast-derived GAL4-UAS system as a tool in D. melanogaster-centric studies is in the fact that the fly does not carry the 17bp-stretch of sequence identical or similar to UAS on its own genome (Brand & Perrimon, 1993). This means that even if GAL4 is constantly produced within a cell, the protein should have no phenotypic consequences, as its DNA-binding dependent functionality is rendered inconsequential (Duffy, 2002). In spite of the clear logic behind this assumption, several studies have shown that it is not always true. The expression of GAL4 under the neural-specific driver GMR, for example, alters the compound eye by inducing apoptosis of its imaginal discs during the fly’s larval stages. This phenotypic consequence was apparent even in heterozygotes for GMR, showing the potency of GAL4 proteins in truly causing cell death (Kramer & Staveley, 2003). More recently, accumulation of insoluble GAL4 in parts of the brain was shown to correlate to severe neuronal apoptosis, leading to behavioural and locomotive defects in emergent adult flies (Rezaval, Werbajh, & Ceriani, 2007). Detrimental side-effects of GAL4 is however not limited to D. melanogaster. Its cytotoxic nature was recently documented in another insect model organism, Bombyx mori (Hara et al., 2017).
These were some of the reasons why we endeavoured to thoroughly establish driver-specific reference phenotypes prior to screening. Furthermore, the experimental design of our miRNA-overexpression screen relies solely on changes to cytoophidia. The pseudo-organelle are in themselves sensitive towards cellular changes. They are known to multiply in numbers and appear denser in dying cells (Liu, 2011). This step was therefore crucial, as it allowed us to ensure that any differences in cytoophidia morphology and behaviour were attributable to the candidate miRNA, and not a consequence of the purported cell-death inducing capability of GAL4. Our results subsequently demonstrated that the protein does alter cytoophidia lengths. This was especially true under the influence of the follicle cell driver C323a. The severe shortening of cytoophidia was prominent even when only one copy of it was present i.e. in heterozygous C323a-GAL4>Oregon-R flies. In fact, of the four types of drivers utilized throughout this screening, only nos-GAL4>Oregon-R did not bear visibly truncated cytoophidia. The conundrum now is in the why and how: could GAL4 proteins bind non-specifically to regions on the Drosophila genome after all? Or are the observed effects tissue promoter-specific? If it is the latter, then shouldn’t C323a and T155 confer the exact same effects, given that both are drivers in follicle cells?It is thusly important to understand that although both C323a and T155 are indeed active in follicle cells, their expressional signatures at other points of D. melanogaster development are vastly different. As aforementioned, the cytoophidia-shortening effect of ectopic GAL4-expression was most severe under C323a followed by Act5c and T155. Like Act5c, C323a is expressed early on in nascent gut and salivary gland, becoming highly active from the final stages of the fly’s embryogenesis (Manseau et al., 1997). T155 is only activated in the same tissues at the third instar larval stage, and may in fact be lethal if expressed in embryos (Hrdlicka et al., 2002; Perrimon, 1997). Meanwhile, cytoophidia of NGT-nos-expressing ovaries were virtually unaffected, despite the promoter being active in late-stage embryos too. However, its expression is in this case restricted to germline cells (Jenkins et al., 2003). All in all, these observations have led us to deduce that ovarian cytoophidia alteration is likely caused by the accumulation of GAL4 proteins over time, but the phenotype is also dependent upon where the tissue-specific promoter drivers were initially made active.
Non-cell Autonomous Cytoophidia Regulation in Nurse Cells as a Result of miRNA-overexpression in Follicle Cells
Throughout the primary stages of screening involving follicle cell drivers, we consistently observed that though a candidate miRNA should only be overexpressed in those somatic cells, germline nurse cell cytoophidia were morphologically affected as well. More often than not – and especially if a CytSh or CytEl-miRNA bears strong effects – the shortening or elongation of cytoophidia within sheathing follicle cells was similarly seen in macro-cytoophidia of encapsulated nurse cells. This seemingly non-autonomous nature of nurse cell macro-cytoophidium formation was also apparent within controls i.e. C323a-GAL4>Oregon-R and T155-GAL4>Oregon-R where, as discussed in the previous section, the egg chamber-wide truncation of these filamentous structures were typical. Though this proved particularly useful for shortlisting candidate miRNAs, the greater significance of this outcome lies within the new aspects in cell-to-cell communication (CCC) it may have demonstrated.
The importance of effective oocyte-centric CCC during embryogenesis is well-understood (Mahowald & Hardy, 1985). The oocyte relies on the systematic deposition of maternal factors from nurse cells to develop, whereas follicle cells provides it with the nutrients it requires to grow. Mass ‘material dumping’ into this young egg, as such processes are called, occur through specialized ring canals or gap junctions from nurse or follicle cells, respectively (Cabej, 2012). In the same manner, follicle cells are connected to nurse cells; however, little is known of what may be communicated between these cell populations, or if any sort of communication occurs at all.
As effects of miRNA-overexpression within follicle cells were explicitly mirrored in nurse cells, our results therefore implies that follicle-to-nurse CCC does indeed take place. This claim is further supported by two factoids: (A) excessive levels of CytEl-miRNAs in follicle cells universally reversed cells C323a-conferred truncation of cytoophidia, and (b) despite proven non-functionality of UASt-miRNA constructs in germline cells, effects of its expression were nonetheless evident in nurse cells. We did not, however, witness signs that this is a two-way relationship. When a candidate miRNA was driven by germline-cell specific NGT-nos-GAL4, resultant changes to nurse cell cytoophidia were never paralleled by follicle cell cytoophidia. An example in this case is with prime CytSh-miRNA, miR-932. Eventhough its overexpression appeared to shear macro-cytoophidia of nurse cells into much smaller filaments, no such changes were detectable in surrounding follicle cells. This observation additionally shows that whilst nurse cell cytoophidia could be non-autonomously controlled, their formation in follicle cells are autonomous and unaffected by external stimuli.
Many questions remain, the most pressing of which is in the component(s) relayed throughout this instance of follicle-to-nurse CCC. In plants, the plasticity in movement of cellular products including miRNAs and proteins between neighbouring cells is well-studied (Marin-Gonzalez & Suarez-Lopez, 2012; Zambryski, 2004). Intercellular trafficking is made possible through plasmodesmata, channels which are formed through the connective tissue connecting one plant cell to another (Brunkard, Runkel, & Zambryski, 2015). By comparison, channel-mediated transfer of macromolecules rarely occur in animal systems. ‘Mobile’ RNAs and proteins are instead often relayed to a nearby cell within phospholipid-enclosed vesicles (Jose, 2015). Furthermore, although follicle cells are physically connected to outer nurse cells in the Drosophila egg chamber, ring canals are not known to form in between them (Airoldi et al., 2011). Evidence of transmigration of small components from follicle to nurse cells do exist nonetheless: the somatic to germline stem cell transmission of the bacteria Wolbacchia in early-staged egg chambers of numerous Drosophila species has been documented, although the means through which this happens is yet to be explained (Toomey et al., 2013).
As such passive mechanisms of CCC appear size-restricted, we believe two scenarios to be most likely causal towards the non-cell autonomous regulation of nurse cell cytoophidia observed here. One is due to the transfer of excess candidate miRNAs from follicle cells. Their concurrent action upon the same mRNA targets in both cell populations would predictably result in the same phenotypic outcomes. The second scenario implicates CTP molecules. Certain nucleotides have long been acknowledged for their integral function in CCC as signalling molecules (Chen, Levy, & Lightman, 1995; Gründling & Lee, 2016). Much of their function in such cases is related to maintenance of homeostasis and coordinating metabolic processes (Mediero & Cronstein, 2013; Meshkini, 2014). The CTPsyn enzyme itself is allosterically regulated by CTP. A long-standing argument in metabolomics portends that nucleotides, but not macromolecules, are transportable through ring canals and gap junctions (Pitts & Simms, 1977; Subak-Sharpe, Burk, & Pitts, 1969). We therefore reason that as miRNA overexpression increases or decreases CTP production in follicle cells, the amount seen in the passive movement of these fine molecules into nurse cells dictates the activity of CTPsyn, thus reducing or inducing cytoophidia filamentation events.
The Utility of CTPsyn-cytoophidia as a Biomarker of Cellular Homeostasis
One of the important takeaways from screening is how dynamic the CTPsyn-cytoophidium can be. Its ability to respond to changes in cellular compositions by freely associating into or dissociating from its filamentous form shows the potential use of this structure as a biomarker of homeostatic maintenance in cells. MiRNA overexpression is one such process expected to disrupt this cellular ‘normal’: any instance whereby cytoophidia morphology is compromised thusly signifies the dysregulation brought about by that particular miRNA.
Apart from changes in length, we have also highlighted polarity distortion and CytPol-miRNAs. This group of candidates was ultimately not investigated further as cytoophidia polarity is at best indeterminate. As aforementioned, a loss of uniformity and misalignment are indicative of affected cytoophidia polarity in follicle cells. It is difficult to ascertain, however, if observed changes under the microscope is a true consequence of miRNA-overexpression, or simply an artefact of mounting technique. We are also of the opinion that the control of polarity is much more complex than that of cytoophidia length. Like most organelles, CTPsyn-cytoophidia positioning is reliant on microtubules and actin cytoskeletal elements within the cell (Ingerson-Mahar et al., 2010; Stebbings, 1995). Overexpression conferred changes to cytoophidia polarity could therefore be apparent due to the candidate miRNA’s involvement in regulating those components instead, and could have actually had very little to do with CTPsyn itself.
Nonetheless, this projected relationship between cytoophidia and such important cytoskeletal elements might have added value to CTPsyn-cytoophidium as a biomarker. The crucial roles microtubules and actin filaments play within the cell are indisputable. Though not typically causal to diseases, their disruption could foreshadow that something has indeed gone wrong (Bossing et al., 2012; Korb et al., 2004). However, these filaments are very fine; aside from specialized immunostaining techniques, they require high-resolution, sophisticated tools to be satisfactorily visualized. As cytoophidia are much more easily discernible, they can therefore be utilized as a surrogate observation tool in ascertaining whether the integrity of the meshed networks of microtubule and actin filaments is preserved, and by extension, if all is well within the cell.