We are interested in understanding the developmental logic for making the larval and adult nervous systems. What is the code for establishing neuronal diversity within each lineage? How do the phenotypes of neurons made during the embryonic phase of neurongenesis [for the larval CNS] compare with those made during the postembryonic phase [for the adult]? What is the relationship of the various lineages to the circuit architecture of both the larval and adult CNS? Can one attribute discrete behavioral roles to the neurons of a given lineage and how have these roles changed through insect evolution?
Throughout my career, I have worked at the intersection of development, endocrinology, and behavior. My group at Janelia has a similar focus, using developmental and endocrinological approaches to understand nervous system organization and function.
The development of the nervous system in higher insects such as Drosophila is especially intriguing because these animals have two active life stages separated by a profound metamorphosis.Through their life history, they produce two sequential nervous systems that are adapted for the functioning of a larval and an adult stage, respectively. Though being dramatically different in size and complexity, the larval and adult nervous systems are nevertheless interlinked. For the central brain and ventral CNS, the neurons for both stages are generated from the same set of neuronal stem cells, the neuroblasts [NBs]. These NBs segregate from the neuroectoderm during early embryogenesis and constitute a stereotyped set of neuronal precursor cells. There are a little over 100 pairs of NBs that make the neurons of the brain and a repeating set of about 30 pairs that produce the neurons of the segmental ganglia. Each NB undergoes repeated asymmetric divisions. The smaller product of each division is the ganglion mother cell (GMC), which typically divides once to produce two daughters, "A" and "B." During embryogenesis, the NBs produce an initial, small set of diverse neurons that make the CNS of the larva. The NBs then go dormant but reactivate during larval growth to make a much larger set of neurons that are used in the adult. The postembryonic lineage from each NB is typically composed of two neuronal classes, the “A” and “B” hemilineages that correspond to the two daughters from the division of the GMC. We find that the neurons in a hemilineage typically share common molecular expression, path-finding choices, initial targets, and, likely, function.
Our research is focused on the neurons of the larval CNS
The larval brain has about 1,000 neurons per hemisphere, as compared to the approximately 20,000 neurons per hemisphere for the central brain of the adult. With the support of the FlyLight project, we have imaged the larval CNS expression pattern of the 7,200 lines of the Rubin Gal4 enhancer collection. This collection has provided us with a detailed catalog of the neuron types that make up the larval brain and ventral CNS and also provides experimental access to the vast majority of larval neurons.
On the anatomical side, we are interested in how the circuit architecture of the larval brain compares with that of the adult. Detailed comparison of the larval and adult olfactory circuit by other labs [e.g., Ramaekers et al., 2005 Current Biology 15, 982-992] have noted that while differing greatly in numbers of neurons and in the extent of neural convergence and divergence, the larval and adult olfactory systems contain similar neuronal classes but with the larva having only a few cells in a given class while the adult has many. Thus, the larval olfactory system possesses an “elemental circuit” for the wiring of an olfactory system, which then becomes more elaborate in the adult. We think that this concept of elemental circuits extends well beyond the olfactory circuit and that the simplified larval brain will provide insights into the circuit motifs that characterize other regions of the brain and ventral CNS. This will hopefully provide some the circuit logic that is used for forming a functioning brain and also provide simple models for understanding the more complex wiring for regions of the adult brain.
A long-standing interest has been in understanding how the neurons of the CNS deal with the profound change in body form that occurs at metamorphosis. Almost 40 years ago, we showed that many larval neurons live through this metamorphic upheaval, undergoing extreme pruning of their axonal and dendritic branches before sprouting new branches to make their adult connections. In Drosophila, though, the extent of pruning and regrowth has been documented for only a few central neurons, such as the TV neurosecretory cells and the gamma neurons of the mushroom bodies. We want to know if different groups of interneurons undergo varying degrees of remodeling through metamorphosis and to what extent are circuit connections preserved from larva to the adult? Using various flip-out strategies and the sparse lines of the Rubin collection, we are systematically following larval interneurons through metamorphosis so that we can eventually compare the larval and adult form and function of the individual neurons and circuits.
Besides being a basis for anatomical and developmental studies, a database of larval neurons provides the anatomical underpinnings for the behavioral studies of the groups of Marta Zlatic and of Lynn Riddiford, and for the nascent Larval Olympiad Project. My own group is specifically interested in the circuits that control larval feeding behavior (Feng Li) and ecdysis behaviors.
In Drosophila, as in other insects, a neuron’s identity is based on its lineage of origin. One of our primary interests is in understanding the rules that control this identity. Besides being units of development, though, we think that the lineages serve as units of function and of evolution. However, to test these latter ideas we need tools that allow us to manipulate specific lineages and hemilineages so that we can examine the functional roles of the neurons that they contain.
Development of Lineage-Based Tools for Studying CNS Function
Our hemilineage-based approach requires driver lines that are capable of targeting gene expression to a single, defined hemilineage. With this goal in mind, we are collaborating with Barret Pfeiffer and Gerald Rubin at Janelia in constructing a library of such lines. Our starting material is the 7,200 lines of the Rubin enhancer collection. However, while many of these lines show expression in specific lineages or hemilineages, there are few that confine their expression to a single lineage or hemilineage. Moreover, most enhancers are quite dynamic in their expression and drive expression in different sets of neurons in the larva versus the adult. Working with the Rubin group, we are devising intersectional strategies to restrict expression to a single lineage or hemilineage in the larval CNS. Larval expression is dependent on the larva feeding on RU486, a progesterone mimic, and results in active recombinases in the cells of the selected hemilineage. The recombinases then remove a “stop” cassette from an actin>stop>LexA construct, thereby allowing the constitutive expression of LexA in the hemilineage. The actin-driven expression is then maintained through the life of the fly while the original enhancers become nonfunctional because RU486 is removed from the food. Our goal is to generate a library of such lines that cover all of the hemilineages of the central brain and ventral CNS. This library will be a platform for the studies on the developmental, functional, and evolutionary aspects of neuronal lineages and hemilineages.
Generation of Neuronal Phenotypes
We want to understand the molecular code that determines the identity of neurons that are produced by each neuroblast. For the ~140 unique neuroblasts located in the central brain and segmental ganglia, are there overarching rules for neuronal phenotypes that apply across the entire set, or are there 140 ad hoc directions that establish the identity for each lineage? We are focusing on the transcription factors that provide molecular signatures for specific neuroblasts, their GMCs, or their daughters. One approach has been in using antibodies to define patterns of transcription in the cells of the lineages. We have identified core GMC genes that are expressed in all of the GMCs of a given lineage, core lineage genes that are expressed in all of the daughter neurons, and core hemilineage genes that are expressed in only the A or the B daughters. We are using both loss- and gain-of-function approaches to determine the role of these genes in controlling neuronal identity and development. Thus far, we find that core GMC genes are typically involved in establishing the identity of one or both of the sibs, and loss of these genes results in a major shift in sib identity. For the core hemilineage genes, by contrast, loss of function does not cause an obvious change in identity, but rather selected aspects of the neuron's phenotype are altered; for example, the mutated neurons navigate to novel initial targets but maintain the other properties characteristic of that type of neuron.
Construction of adult neuropils
We are interested in understanding how the early molecular expression in the neurons of a lineage relates to the circuits that they construct. To begin, however, we need to know how the lineages and hemilineages interact to form the adult neuropils. We have begun with the lineages that make the neuropils of the ventral CNS. The major neuropils in the thoracic ganglia are the lateral leg neuropils and the dorsal flight neuropil. Seventeen hemilineages are dedicated to making each leg neuropil, while an additional 10 hemilineages provide the hemisegmental contribution to the flight neuropil. The leg neuropil begins as a partitioned neuropil, with the endings from each hemilineage occupying an exclusive domain within the forming structure. Boundaries between hemilineages are maintained until the start of metamorphosis, when steroid signals cause exuberant sprouting of the immature neurons and the partitions disappear. An important question is how the neighbors at partition boundaries relate to the final synaptic partners. Do the initial contacts in the early developing neuropil prefigure the major synaptic partners seen in the mature CNS?
Using the specific lines that we are developing, we can drive different fluorescent proteins in hemilineages that potentially interact in building the leg or flight neuropils. In collaboration with Philipp Keller’s group, we will be using cultured nervous systems and live imaging techniques to follow the behavior of different hemilineages as they start to “wire-up”.
Behavioral functions of the lineages
Since many functional classes of interneurons are based on hemilineages, we think that having a library of hemilineage-based driver lines will allow a systematic manipulation of the neuronal classes of the CNS and an assessment of the role of each in behavior. As a proof of principle, we are starting with the lineages that make the thoracic neuropils because these can be directly associated with the patterning and coordination of walking and flight behavior. The initial stages of this work, carried out by Robin Harris, involves the video and high-speed video analysis of flies in which specific hemilineages have been either silenced or activated using UAS-shibirits or UAS-trpA1, respectively. Subsequent experiments will then move into optical techniques to activate the neurons of specific lineages or to monitor their activity.
Macro- and microevolution of the lineages
We are focusing on the hemilineages to assess variation in the neuronal composition of the CNS and how this variation relates to behavior. Since we think that the secondary neurons within a given class have broadly overlapping functions, alterations in the numbers of these neurons would likely increase or decrease the temporal and spatial computation within the network to which they belong. As we complete the collection of molecular markers for each hemilineage, we can ask how constant is the number of neurons generated by a given stem cell? Do numbers change with different fly populations or in response to environmental factors during the larval growth? How does this variation relate to function? For example, if flies have been selected for a reduced capacity for flight, do we see corresponding reductions in the number of interneurons in flight-related hemilineages but no change in leg-related hemilineages?
We are also interested in broader issues relating to the evolution and diversification of the CNS. After we characterize the neurons made in each of the hemilineages in Drosophila melanogaster, we can then explore how stable these phenotypic classes have been through evolution. Do the neurons that are made by the homologous stem cells in other insects, such as moths, beetles, and bees, have the same pattern of initial projections and the same transmitter phenotypes as seen in the fly? For the lineages that differ significantly, we would be interested in how this difference is reflected in molecular expression, especially of core GMC, lineage, and hemilineage genes. If there are molecular differences (e.g., the moth lineage has a transcription factor that is not expressed in the homologous fly lineage, or vice versa), can we then go to Drosophila to make the appropriate molecular changes and produce moth-type neurons in the fly CNS?
It is intriguing that most or all of the members of some hemilineages die soon after their birth and do not contribute to a functioning adult nervous system. We suspect that these neuronal classes survive and are functional in more basal insects but are lost in the highly derived CNS of Drosophila. Can we find these neurons in other insects, and what are their phenotypes and function? Although they are not normally present in flies, we can study their function in Drosophila by genetically blocking their death. Do they then become incorporated into the adult circuitry, and, if so, how does their presence impact the development and functioning of other lineages and of the CNS as a whole?