Maternal Messages to Live by: A personal historical perspective

SUMMARY

In the 1980’s, the study of localized maternal mRNAs was just emerging as a new research area. Classic embryological studies had linked the inheritance of cytoplasmic domains with specific cell lineages, but the underlying molecular nature of these putative determinants remained a mystery. The model system Xenopus would play a pivotal role in the progress of this new field. In fact, the first localized maternal mRNA to be identified and cloned from any organism was Xenopus vg1, a TGF-beta family member. This seminal finding opened the door to many subsequent studies focused on how RNAs are localized and what functions they had in development. Xenopus remains the model of choice for studies on dissecting RNA/protein transport particles and RNA-sequencing analyses that are moving the field into the future.

Conceptual Background: Cytoplasmic domains, cell lineages and nuclear transplantation

There is a long and rich history of early embryologists documenting the relationship between cell fate and the inheritance of particular regions of the egg cytoplasm (Wilson, 1925). One of the most striking examples is that of the ascidian Styela that Conklin described in 1905. The Styela egg displayed a yellow crescent shaped region rich in mitochondria that was asymmetrically inherited by blastomeres that gave rise to the muscle cell lineage (Conklin 1905). Perhaps the first example in a vertebrate system linking a cytoplasmic domain to a specific lineage was the description by Blackler (1958) of germ plasm in the frog Rana pipiens. UV ablation of germ plasm led to the loss of the germline, a condition that could only be rescued by injecting cytoplasm containing germ plasm back into the vegetal pole (Smith, 1966). These experiments linked a visible structure, germ plasm, to an essential cell lineage and function (Williams and Smith, 1971). The biochemical nature of cytoplasmic determinants and their mode of action remained a mystery, although histological staining had suggested the presence of RNA (Bounoure, 1934).

Coincidentally, developmental biologists were asking whether the process of differentiation was accompanied by the loss of genetic information. Put another way, were the nuclei of differentiated cells intrinsically equivalent to the egg genome and could differentiation be reversed. To address this question, Briggs and King (1952; 1960) had carried out a series of nuclear transplantation experiments injecting nuclei from advanced embryonic stages back into enucleated eggs of Rana pipiens. Their results had mixed success in promoting normal development, most likely because the chromosomes did not have sufficient time within the ooplasm to prepare for the rapid divisions imposed during cleavage stages after fertilization.

At this time, an important event occurred that would ultimately change the direction of this inquiry. Dr. Michail Fischberg asked his new student, John Gurdon, to repeat the Briggs and King nuclear transplantation experiments, but to use a different amphibian, Xenopus laevis. In 1966, John Gurdon showed beyond any reasonable doubt that indeed, nuclei from differentiated cells could completely support normal development when introduced into the enucleated egg (Gurdon and Uehlinger, 1966). These elegant experiments earned Sir John Gurdon the Nobel Prize in 2012 and showed that nuclei could be reprogrammed to a totipotent state. Similar to the cytoplasmic localization studies, they underscored the dominance of cytoplasmic information over nuclear expression. Sir Gurdon had also published papers showing that RNAs encoding proteins restricted to differentiated cells could be translated after injection into Xenopus oocytes (Gurdon et al., 1974; Woodland et al., 1974)! These experiments catapulted Xenopus onto the world stage as a model embryological system for studying gene expression during development. I was a student in Robert Briggs’s lab in the 1970’s working with Rana pipiens as well as axolotls to address this question of developmental plasticity. I remember well meeting Sir John Gurdon during his visit to Brigg’s lab where they discussed the different outcomes of their nuclear transplantation results. In hindsight, I realized I learned valuable lessons from this part of my professional history: It is critical to choose the right model system to answer the question being asked and important to remain working in your lab as a PI. I have always admired John Gurdon for doing that and setting an example.

The Xenopus Oocyte: Identifying localized maternal RNAs

In 1976, I headed to MIT and Harvey Lodish’s lab to learn molecular cloning, a new technology at the time. After working with Rana pipiens and axolotls, I was looking for a simpler model system in which to understand gene regulation during development. Dictyostelium seemed a good choice at the time as there were only two cell types: stalk versus spore. Since these two presumptive cell types were physically separated from each other in the migrating slug, my thought was to cut thousands of these slime mold slugs, isolate RNA from the different regions and screen for differences. Wrestling with exactly how to do this, I realized how much easier the task would be using the large and visually polarized Xenopus oocyte instead.

In truth, I missed the beauty and embryological history of frog development: “How the embryonic body axis emerges from a seemingly symmetrical egg and is subsequently patterned during embryogenesis is one of the most fundamental questions in developmental biology” (Scott Gilbert). I was not going to answer that question working on Dictyostelium. After one year, I asked Harvey if I could bring Xenopus into the lab as a model system and work on translational control with a focus on maternal RNAs. He had a well-established reputation in the field of translation, and he agreed. I went on to show that different RNAs were translated during the progression of Xenopus oogenesis while some remained translationally silent. During this time, I learned Xenopus husbandry, the use of in vitro translation systems and protein analyses by 2-dimentional gel electrophoresis, all of which would serve me well in my own lab. Most importantly, I spent an intense time in the library going through the literature on the maternal contribution to early development and prepared to write my first grant to NSF. Little did I know that across the Charles River at Harvard a former graduate student of John Gurdon, Doug Melton, was thinking the same thoughts as I was: there must be vegetally localized maternal mRNAs that drove early patterning of the Xenopus embryo.

Looking back, the time I spent in the library reading and thinking about what problems I wanted to tackle was worth every minute! What I decided to write my grant about during this time would end up being a new field that would consume the rest of my professional career.

The rationale driving the search for localized maternal mRNAs was straightforward. Zygotic transcription did not begin until the embryo was at the mid-blastula stage and 4,000 cells (Newport and Kirschner, 1982). Yet three basic developmental decisions had been made that were known to initiate at the vegetal pole: the dorsal/ventral (reviewed in Weaver and Kimelman, 2004; Houston, 2012) and primary germ layer identities (Nieuwkoop, 1977) as well as the germ cell determinants in the form of germ plasm (Smith, 1966). Therefore, maternal transcripts must be involved and regionally localized within the egg. Historically, a major obstacle to examining the spatial distribution of individual mRNAs had been the inability to prepare specific cytoplasmic regions of eggs in quantities suitable for biochemical analysis. The fully grown Xenopus oocyte is visibly polarized along the important animal/vegetal axis with cortical melanosomes at the animal hemisphere and relatively few vegetally. A single oocyte at 1.3 mm contains ~4ug of total RNA. The first set of experiments I did after starting my own lab exploited the clear advantages of Xenopus. Within cryostat embedding medium, we aligned thousands of oocytes along their A/V axis and cut frozen sections essentially dividing a single cell into three regions. RNA was isolated from these frozen sections.

Previous to our experiments, Joel Richter and colleagues had shown that as much as 70% of the oocyte’s cytoplasmic poly(A)+RNA was untranslatable. “The experiments we have presented provide no support for the proposal that the interspersed poly(A) RNAs are true maternal messages, meaning molecules capable of being directly loaded on polysomes and translated during maturation or development.” (from Richter et al., 1984) From this statement, we worried about selecting non-translatable RNAs. To avoid this potential problem, we decided to translate the mRNA isolated from sectioned oocytes using in vitro translation systems, and analyze the protein products on 2-D gels looking for differences along the A/V axis. We found these differences and published the results in Developmental Biology (King and Barklis, 1985). The big disadvantage of our approach was that we did not know the identity of the proteins differentially translated.

Meanwhile, Doug Melton was cutting tips off oocytes with a scalpel that were frozen in their jelly coats. In 1984 he had published a transformative methods paper on how to transcribe mRNAs in vitro (Krieg and Melton, 1984). Doug screened a cDNA oocyte library made from mRNA transcripts for clones highly enriched at either the animal or vegetal pole. Four localized mRNAs were discovered, three at the animal pole and one at the vegetal pole (Rebagliati et al., 1985). He named the vegetal pole mRNA vg1 (gdf1). Importantly, they showed that vg1 (gdf1) mRNA remained localized into cleavage stages and thus could potentially influence cell fate. Having cloned vg1(gdf1) they were now in position to sequence it and determine its identity (Weeks and Melton, 1987). Excitedly, Vg1 (Gdf1) proved to be a TGF-beta member and, as a known signaling molecule, it’s likely function in patterning the early embryo seemed obvious. However, it would take another eighteen years before the endogenous function would be assigned to Vg1 (Gdf1) (Birsoy et al., 2005). Doug had anticipated the need to visualize the localization process and developed in situ hybridization in Xenopus. Using this new technique, vg1 (gdf1) was shown to be tightly associated with the oocyte vegetal cortex and released after fertilization where it was inherited by blastomeres comprising the endoderm (Weeks and Melton, 1987). Doug was able to published two papers in Cell and I published one in Developmental Biology. In hindsight, I should have just gone straight to cloning and not worried about translation! The image of localized vg1 (gdf1) made the cover of Cell in 1987 and laid the foundation for answering the next set of questions: what other mRNAs were localized, how were they localized, and what functions did they have?

Germ Plasm RNAs: The search for more localized maternal mRNAs

The tight cortical localization of vg1 (gdf1) mRNA pointed to the cytoskeleton as the likely agent for retaining RNAs at one pole. To test this idea, my lab extracted oocytes in a high salt detergent solution that left basically only cytokeratins in an insoluble pellet. We were surprised to find that the pellet contained virtually all the oocyte vg1 (gdf1) mRNA (Pondel and King, 1988). In the Weeks and Melton paper they had shown vg1 (gdf1) mRNA was no longer retained within the cortex, but released to occupy the vegetal blastomeres. Consistent with that observation, we found that after fertilization, vg1 (gdf1) mRNA was no longer recovered from the cytoskeletal fraction but was now found in the soluble fraction. This result was foreshadowed by Jeffery and Meier (1983) who showed that actin mRNA was retained in a detergent extracted Styela embryo at the posterior pole of the 2-cell embryo. Our discovery was important for two reasons: it focused attention on the intermediate filaments as part of the anchoring complex for mRNA; and 2) it pointed the way to biochemically isolating additional localized maternal mRNAs. We reasoned that if the oocyte cytoskeletal matrix serves to anchor mRNAs, then other localized maternal mRNAs should also be highly enriched in the intermediate filament fraction (IFF).

We generated a lambda cDNA library from the IFF sample and probed this library with mRNA highly enriched in the IFF. Clones that hybridized only to the IFF mRNA but not soluble mRNAs were isolated. This strategy was successful and we named the clones Xenopus cytoskeletal associated transcripts or xcats. Our first clone, xcat2, was identified as being the Xenopus homolog to the Drosophila nanos. As nanos was essential in the germline, these results pointed to conserved functions between invertebrate and vertebrate germlines. I presented these findings in a poster at the Developmental Biology meetings and I remember Ruth Lehmann being very happy that we had established this link. xcat3 was also found in the germ plasm and identified as one of the germline helicases. We named it deadsouth (ddx25) (Mosquera et al., 1993; MacArthur et al., 2000). The deadsouth (ddx25) 3’UTR is currently used to target transcripts to the germ line.

Rick Elinson had been working on the relationship between microtubule arrays, cortical rotation, and dorsal axis specification in Xenopus (Elinson and Rowning, 1988). He had developed a method for isolating structurally intact cortices from either the animal or vegetal pole of oocytes. Ultrastructural and immunocytochemical analyses revealed that the 10-µm-thick isolates included a cytokeratin network, mitochondria, and other elements, found in the oocyte cortex. He suggested we determine if vg1(gdf1), xcat2 (nanos1) and xcat3 (ddx25) co-localize with the vegetal cortices. All three vegetally localized mRNAs were found to be specifically retained within the isolated vegetal cortices, in sharp contrast to histone RNA. These findings supported the view that the vegetal cortex represented a unique cytoskeletal domain in which a class of maternal mRNAs became localized (Elinson et al., 1993). In that same year, Larry Etkin’s group published a paper in Science describing xlsirts, or interspersed repetitive transcripts that localized to the mitochondrial cloud. Janet Heasman and Chris Wylie had shown in 1984 that the germ plasm formed within the mitochondrial cloud (MC) also called Balbiani body. xlsirts were the first RNAs to be clearly shown by in situ hybridization to be part of the germ plasm (Kloc et al., 1993). xlsirts represented hundreds of RNAs, believed to be non-coding RNAs related to the X-chromosome inactivation xist gene. The role of non-coding RNAs within the germ plasm generated a lot of speculation.

xwnt11 was also found to localize within the MC (Ku and Melton, 1993), a surprise as xwnt11 was thought to play a role in axial patterning based on its ability to induce a secondary axis (McMahon and Moon, 1989; Smith and Harland, 1991; Sokol et al., 1991). Meanwhile, John Gerhart, Randy Moon and Carolyn Larabell were investigating the molecular basis for cortical rotation and dorsal specification, that occurs before first cleavage. They showed that vesicles were moved via microtubules during cortical rotation, an event required for the localization of β-catenin protein to the future dorsal side of the Xenopus egg (Rowning et al., 1993) and critical to dorsal specification (Heasman et al., 1994). The role of xwnt11 in dorsal specification would not be confirmed until 2005 (Tao et al., 2005). The role of xwnt11 in the germline, if any, has never been determined.

We were excited to determined the spatial pattern of expression for nanos1 (xcat2) in Xenopus because it had been shown to be exclusive to the Drosophila germline. Would the localization pattern be different from that of vg1(gdf1) mRNA? Indeed, it was, localizing very early during oogenesis to the MC, and well before vg1 (gdf1) mRNA would begin its microtubule based transport to the vegetal cortex. Our work defined two localization pathways: an Early germline pathway followed by nanos1 and xlsirts, and a Late pathway defined by vg1 (gdf1) (Forristall et al., 1995). A month later, Kloc and Etkin published similar findings, but with a much better title, clearly stating the result (Kloc and Etkin, 1995). The valuable lesson I learned was that titles matter as they do impact the number of times your work is referenced. I also learned that no one owns a field just because you were the first one on the scene. Etkin’s lab came to refer to the Early MC pathway as METRO for Message Transcript Organizer.

We and the Etkin lab mapped the 3’UTR sequences required for mRNAs to localize within the MC and discovered that the localization signal was made up of two elements. One element was required for nanos1 mRNA to concentrate within the MC while the second element was required for its germinal granule localization (Zhou and King, 1996a,b; Kloc et al., 2000). Over the next few years, investigators in the field focused on both identifying more localized mRNAs and the mechanism of mRNA localization. Hugh Woodland screened a gastrula vegetal pole cDNA library and discovered xpat (pgat), one of the few mRNAs to be exclusively expressed in PGCs. xpat (pgat) has been an invaluable marker for primordial germ cells (PGCs) (Hudson and Woodland, 1998). Xpat (Pgat) is also a major protein component of the germ plasm and provided the first glimpse into the nature of the germ plasm matrix (Machado et al., 2005). Hugh Woodland’s lab went on to discover more mRNAs expressed in PGCs using this screening approach (reviewed in King, 2014; Houston, 2013).

The Etkin lab isolated mitochondrial clouds using gradient methods. Libraries made from MC mRNA yielded additional germ plasm mRNAs including fatvg (plin2) and hermes (rbpms) (Chan et al., 1999; Zearfoss et al., 2004). Interestingly, fatvg (plin2) and hermes (rbpms) used both the early (METRO) and Late pathways suggesting overlapping or redundant elements. During this time, several labs realized that mRNAs normally localized to the MC, could also use the Late pathway if injected at later stages. The same localization signals appeared to function early or late. Confusion ruled as to just what a localization signal was and how different pathways were chosen endogenously.

Mechanism of RNA localization

When Doug Melton accepted his academic position at Harvard, he insisted on having his own lab space within his office, all set up to do experiments. He wanted to protect his time at the lab bench. The fruit of that decision was clear in his next work published in Nature as sole author (Melton 1987). Here he showed for the first time that mRNA localization occurred progressively during oogenesis using a mechanism that did not require prior translation or mRNA degradation/protection. vg1 (gdf1) RNA injected into cultured oocytes recapitulated normal localization to the vegetal cortex, opening up an avenue for studying the mechanism involved (Yisraeli and Melton, 1988). It was quickly established that vg1 (gdf1) localization required intact microtubules and actin filaments for cortical retention (Yisraeli et al., 1990).

Could an mRNA localization signal be defined? Kim Mowry joined the Melton lab and identified by deletion mapping the required sequences for localization to occur. A 340-nt piece in the 3’UTR was required and sufficient for normal localization (Mowry and Melton, 1992). She was now in a position to find the proteins that interacted with this sequence and define the localization machinery. Further refining of the localization signal into elements recognized a repeated cluster of the pentanucleotide UUCAC that seemed to be important (Deshler et al., 1997; Bubunenko et al., 2002). The Mowry lab went on to define microtubule motors involved and associated mRNA binding proteins that comprised the vg1 (gdf1) transport particle. Key findings during this time were 1) that mRNA binding proteins normally thought to be strictly nuclear (like hnRNPI) were part of mRNA transport particles, remaining with mRNAs as they were localized (Cote et al., 1999); and 2) the steps that occur as mRNAs are identified in the nucleus for transport and the remodeling that occurs as mRNAs enter the cytoplasm and associate with microtubules (Lewis et al., 2008; Kress et al., 2004), 3) the microtubule motors involved in mRNA transport (Gagnon et al., 2013; Messitt et al., 2008). The Thomas Pieler lab, using an affinity purification approach, also contributed to identifying RNP components and localization signals required for vegetal localization. Importantly, his lab identified Elr-type proteins that associate with mRNA localization elements, helping to stabilize mRNAs within the germline from somatic clearance by miR-18. Their work provided a functional link of vegetal mRNA localization and the protection (Koebernick et al., 2010).

In 1996, Larabell and Etkin’s labs moved towards a mechanistic understanding of mRNA localization into the MC taking full advantage of the complete transparency of stage I oocytes. They synthesized fluoroscein- or Texas red- labeled RNAs (xlsirt, xcat2 (nanos), and xwnt11) using derivatized nucleotides and injected these synthetic mRNAs into stage I transparent oocytes. They found that the translocation of RNAs through the early or METRO pathway, unlike that of the late pathway, occurred in the absence of intact microtubule and actin microfilament cytoskeletal elements. Inspired by this work, we collaborated with Evelyn Houliston to make movies of nanos1 (xcat2) MC mRNA localization. We showed this event occurred by diffusion through the ooplasm and entrapment on the very dense endoplasmic reticulum that occupied the MC (Chang et al., 2004). The connection between mRNA localization and the endoplasmic reticulum has never been resolved (Kloc et al., 1998).

Function of localized maternal messages

The ability to actually peel the cortex from Xenopus oocytes offered a unique advantage for isolating and identifying additional localized mRNAs. Probing the IFF cDNA library with mRNAs isolated from either animal or vegetal hand-isolated cortices, Jian Zhang in my lab identified five novel vegetal pole mRNAs. Excitingly, the screen identified the first localized transcription factor, vegt, in the T-box family. Its expression pattern was entirely consistent with a regulatory role in endoderm specification. We were anxious to move forward and determined its function as no less than three other labs had also isolated the same gene, all naming it differently: the Gurdon lab called it antipodean (Stennard et al., 1996); the Thomsen lab called it brat (Horb and Thomsen, 1997) and the Kirschner lab called it xombi (Lustig et al., 1996).

At that time, we had poor tools for determining the function of these localized maternal mRNAs. The most useful approach took advantage of the RNAse H activity in oocytes that degrades RNA-DNA hybrids. mRNAs could be targeted for degradation after injection of specific antisense cDNA oligodeoxynucleotides, or ODN (oligos). The other main antisense reagent for loss-of-function experiments came in 1997 with the advent of morpholinos oligos. These do not degrade the message but have the potential to form stable hybrids that block translation or splicing of the transcript (Summerton et al., 1997). Janet Heasman and Chris Wylie had developed a procedure commonly referred to as the Host Transfer Technique, that would allow a direct interrogation of function of maternal components. They had successfully depleted maternal cytokeratin mRNA from oocytes and generated a loss-of-function phenotype in resulting embryos (Torpey et al., 1992). Host transfer remains the only way to deplete maternal information to determine function. Fortuitously, Janet and I were both at the 1996 Xenopus International Conference at Estes Park, Colorado. I approached her to collaborate on a vegt project and she enthusiastically agreed. As it turned out, John Gurdon also asked her to collaborate shortly thereafter. The findings that resulted from our collaboration were quite stunning and this time period was one of the most exciting in my career. Our work defined vegt as the endoderm determinant and showed that it was required for mesoderm induction (Zhang et al., 1998). Other mRNAs that were discovered in Jian Zhang’s screen besides vegt included: xdazl, a long chain acyl-CoA synthetase the depletion of which causes meiotic arrest; arh for autosomal recessive hypercholesterolemia; and fingers, a zinc finger repressor (Zhang and King, 1996; Houston et al., 1998; Houston and King, 2000; Zhou et al., 2003; Wang et al., 2012; Venkataraman et al., 2004). The lesson was clear: Do not hesitate to collaborate with those that can quickly move a project forward.

Janet Heasman went on to solve other mysteries surrounding localized maternal mRNAs using the Host Transfer methods she developed. I remember the long running attempts of Doug Melton and colleagues to show the endogenous function for the protein product of vg1 RNA. They were only able to show what it was capable of doing when it’s expression was forced. However, it turned out they had been using a vg1 (gdf1) allele carrying a mutation that failed to be efficiently cleaved which was required for its activation. Janet Heasman and colleagues isolated the active vg1 (gdf1) allele and went on to show that Vg1 was required for mesoderm induction and for the expression of several key BMP antagonists. The lessons learned from the vg1 (gdf1) saga are: to make sure you have the correct allele, and continual failure may be trying to tell you something.

Anchoring localized mRNAs to the cortex: The concept of binary functions for mRNAs

Larry Etkin had previously shown that disruption of non-coding xlsirt RNAs caused other localized mRNAs to lose their cortical localization (Kloc et al., 2005). In 2001 a former postdoctoral student of Doug Melton, Dan Kessler, published a paper that rocked our world (Heasman et al., 2001). He had depleted vegt mRNA and found that it caused the release of vg1 (gdf1) mRNA from the vegetal cortex. The total amount of vg1 (gdf1) mRNA was not affected; however, a reduction of Vg1 protein was observed. Other mRNAs were affected in a similar fashion (Bicaudal-C and Wnt11) while others were not (nanos1 (Xcat2), xotx1). Were our results defining the function of vegt incorrect, and depletion was really dislodging vg1 mRNA and that explained the loss of endoderm/mesoderm fate? Fortunately, there was a way to tell. Dislodging vg1 (gdf1) depended on vegt mRNA and not protein as a morpholino that blocks translation while leaving the mRNA intact, had no effect. These results indicated a novel structural function for maternal vegt mRNA separate from its protein coding function. Lessons learned: mRNAs may not just code for proteins! Be careful in your interpretations of anti-sense RNA destruction experiments.

Earlier observations had hinted at a role for cytokeratins in mRNA cortical anchoring, but little progress had been made since then (Pondel and King, 1988; Alacron and Elinson, 2001). But in 2005, the Etkin lab published an important paper showing that the organization of the cytokeratin filaments, but not the actin cytoskeleton, depended on vegt RNA and xlsirts (Kloc et al., 2005; 2007). Very sadly, in March of 2006, Larry Etkin died at the age of 61. His lab was dissolved, but Malgorzata Kloc found a new position and carried on with these studies at an ultrastructural level (Kloc et al., 2011a; reviewed in Kloc et al., 2011b).

Current search for maternal localized RNAs and their functions

Most recently, various labs have revisited the search for novel localized mRNAs, now bringing next generation sequencing to the task. RNA-seq analysis has the distinct advantage of cataloging all RNA sequences whether coding or non-coding. It provides a data base that can be mined again and again as the annotation of the Xenopus genome improves. These efforts have identified hundreds of new mRNAs enriched along the animal/vegetal axis, with only about half annotated, leaving much work to be done (De Domenico et al., 2015; Claussen et al., 2015; Owens et al., 2017; reviewed in Houston 2013). Interestingly, no biases in mRNAs could be detected along the dorsal/ventral or right/left axis at the 8-cell stage (De Domenico et al., 2015). General themes that are emerging from this line of investigation emphasize the vegetal pole as a major hub for signaling and directing early patterning in the embryo. Comparisons between the two poles show much less complexity and enrichment at the animal pole. Interestingly, many vegetally localized mRNA were also expressed in neural ectoderm derivatives (Owens et al., 2017). Our efforts have focused on the germ cell lineage and developing a catalog of all mRNAs and proteins found in the germ plasm beginning with mitochondrial clouds, vegetal cortical germ plasm, and germ plasm in PGCs after they have segregated from the endodermal lineage. Xenopus offers unique advantages in developing such an inventory. Mitochondrial clouds from stage I oocytes are very large structures, 30 microns in diameter, and can be hand-isolated in large enough numbers that make mass spectrometry and RNA-seq analyes practical. Vegetal cortices from fully grown oocytes and pure populations of PGCs also are possible to isolate in quantities for biochemical analyses. Soon we will have a comprehensive list of RNAs and proteins operating within the germ plasm from formation to lineage separation during early development. We look forward to more surprises as to how the germ cell lineage maintains full potential within the developing embryo.

FUTURE DIRECTIONS

Of course, there are many questions that remain to be answered regarding the regulatory networks operating at the vegetal pole as well as the mechanisms of mRNA localization. Now that more sophisticated methods and techniques are available for Xenopus and are being continuously developed, the rate of progress will undoubtedly accelerate. Functional studies can now take advantage of CRISPR/Cas9 (Blitz et al., 2013; Nakayama et a., 2013), transgenic lines expressing tagged maternal mRNAs, improved sequencing and annotation, time released and spatially targeted inhibitors (detailed in Hoppler and Vize (eds.), 2012). To aid the search for RNA binding proteins-RNA complexes there is CLIP technology combined with high throughput sequencing (Ray et al., 2013). We now have significantly better structure prediction tools for RNA (Ahmad et al., 2013) and the ability to determine protein binding within the context of relevant RNA 3-D structures (Schroder et al., 2010).

Here is my personal list of areas that I believe would be exciting to pursue.

Vegetal Pole as an Organizing Center for establishing cell fates

1. Evidence from RNA-seq analyses suggest that the canonical Wnt pathway is not the only Wnt pathway operating at the vegetal pole. Components for the PCP pathway are also present suggesting a more integrated, convergent Wnt pathway functioning at the vegetal pole (Owens et al., 2017). xwnt11 mRNA is found in the germ plasm and may play as yet an undescribed role during PGC and/or neural crest migration. Perhaps the integration of these Wnt pathways would explain why several genes when knocked down, affect both dorsal specification and PGC development (Colozza and De Robertis, 2014; Chan et al., 1999).

2. The identity of the dorsal determinant remains unknown.

3. Microtubule array formation critical for cortical rotation and dorsal specification is regulated specifically within the vegetal hemisphere. What are the molecular components that regulate microtubule dynamics spatially?

4. What are the maternal gene networks inherited by the PGCs that allow full potential to be retained within the context of other signaling pathways? We have begun this analysis by comparing RNA-seq and mass spectrometry data from isolated PGCs to determine the maternal and zygotic contributions to this lineage. How are PGC mRNAs controlled at the translation level?

5. A recurrent theme is how common vegetally localized mRNAs also play a role in neurogenic fates or functions (Owens et al., 2017). In fact, the most abundant mRNA within the vegetal cortex is grip2 (glutamate receptor interacting protein), a scaffold multiple PDZ domain protein that organizes signaling complexes in neurons (Kaneshiro et al., 2007; Ge et al., 2014; Claussen et al., 2011; Tarbashevich et al., 2007; Tan et al., 2015). Grip2 protein is present within vegetal cortices and may well serve as a scaffold for signaling at the vegetal pole providing the link between Wnt and PGC migration (unpublished observations, Newman and King). We suggest that the nervous system uses many of the same components when a signaling scaffold and cell migration are required.

6. Are there LncRNAs at the vegetal pole and if so, what roles do they play? xlsirts were discovered back in the early 90’s, but further studies have waited for better methods of sequence and annotation. Now that the technology has improved significantly, real progress can be made answering these questions about non-coding RNAs.

7. Are there DNA modifying enzymes that could spatially regulate epigenetic changes in the early embryo?

Mechanisms for RNA Localization

mRNA localization is a fundamental cellular process with profound effects for function in a wide variety of cells including the oocyte. The Xenopus oocyte has provided a powerful model system to study both the early and late pathways, but major gaps still exist in our understanding.

1. What defines an mRNA localization signal? We still do not know the underlying principles that regulate RNA-protein interactions. Deletion/functional mapping of mRNA localization signals has not really defined a consensus sequence for localization although certain motifs seem to be important, but not all the time. Secondary RNA structure must be playing an important role.

2. What are the nuclear events that initiate mRNA localization? What determines localization specificity?

3. How to localization signals recruit microtubule motors? Do different mRNAs form their own particles for transport? Are non-coding RNAs involved?

4. What is the role of the ER and other organelles (mitochondria, lipid droplets) in mRNA localization? Do they provide a scaffold for mRNAs, but how?

5. How does RNA act as a structural element with the cytoskeleton?

The field can anticipate surprises ahead as these questions and more are tackled.

Figure 1.