The reader may not be a biologist, or a chemist or a science teacher but I am certain that you are a human being. As such, everyone is touched by cancer or is aware of cancer, either themselves as a survivor or via a family member or an acquaintance. This will be a story about cancer but it will not be a story about cures, it will be a story about something unusual that I discovered in my attempt to try to understand the origins of cancer. The logic of my approach is that until you know all the ways that you can get cancer, you will not be able to cure all cancers. My topic today is, ‘what is a tumor suppressor gene doing in a neuron’? If some of those words are not quite clear, then be patient as I am going to begin at the beginning. You will read lots of definitions at the start and then I will bring the story back around to the picture at the top.

Cancer is uncontrolled cell division

I begin with the simplest of definitions. What is cancer? The many diseases grouped under the name cancer are all associated with a specific symptom – tumors. Tumors result from uncontrolled cell division. There are more than 100 types of cancer. In cancer tumors are usually named after the organ or tissue where they originate. Every cell in the body, with notable exceptions, is capable of dividing and thus becoming a tumor.

Normally one cell divides into two cells and those cells divide to become four cells in a highly regulated process called cell division or mitosis. The first level of mitotic regulation is at the initiation step. In an adult, cell division is typically needed in a few well-established circumstances such as to replace a worn out cell or to generate new tissue during wound healing. Cell division is preceded by DNA replication so that both daughter cells inherit the identical information contained in the parent cell. If an error occurs during DNA replication, it usually does not prevent mitosis and the error (called a mutation) is passed on to one of the daughter cells. That cell may then become dysfunctional. Typically the body knows how to take care of injured cells. Mechanisms such as programmed cell death, endocytosis by neighboring cells or tagging and removal by the immune system prevent a damaged cell from causing problems.

Unfortunately, for the organism (but not for the cell) certain mutations do not result in a nonfunctional cell. In these cases, the mutation leads to the cell having an advantage over its normal neighbors. This advantage is that DNA replication and cell division are out of control. So, while the progression is similar, one to two to four cells, there is no regulation. Mutant cells go through mitosis without the cues from worn out cells or the signals from a wound. Thus in a short period of time the mutant cells out number their neighbors. The mutant cell’s many descendants stick together (at first) to form a tumor. With all of their energy going to mitosis, these mutant cells no longer function properly yet they cannot be easily removed by normal means due to their rapid proliferation. Eventually the dysfunctional mutant cells cause the whole organ to become dysfunctional. That’s bad news for the organism. Unregulated cell division as described here, is the essence of the many diseases collectively known as cancer.

There are two types of mutations that can lead to out of control cell division. The first is the one described above that gives a cell an advantage over its neighbors. These are “gain of function” mutations in genes that are called oncogenes. An oncogene is a mutant gene that

encodes a protein that positively causes cancer. The oncogenic protein’s unexpected presence or abnormal activity leads to excessive mitosis. The mutant oncogenic protein must be present in the cells of the tumor for their continued rapid proliferation. One famous oncogene is called Ras. Activated Ras protein is found in many tumors. For example, nearly 100% of pancreatic tumors express activated Ras.

The other type of mutation leading to uncontrolled cell division is a “loss of function” mutation in a gene called a tumor suppressor. A tumor suppressor gene encodes a protein whose normal function is to prevent unregulated mitosis. Tumor suppressors cause tumors when a mutation cripples their protein product. Thus, tumor suppressor proteins are never found in tumors. One famous tumor suppressor gene is called RB. Loss of this gene was first discovered in a tumor of immature cells in the eye called retinoblastoma. This is a fundamentally different mechanism than the one employed by oncogenes that cause cancer by their presence.

This basic dichotomy in tumorigenesis, loss of function versus gain of function requires distinct therapeutic approaches. The dichotomy also undermines the popular notion that cancer is a single disease and somewhere waiting to be discovered there is a single “cure” or a “vaccine”.

Intercellular signaling regulates cell division

One way that Nature fights cancer is by tightly regulating cell division. Regulation happens both within the cell via intrinsic checkpoints in the cell’s life cycle as well as via communication between cells. While checkpoints serve to block unnecessary progress into mitosis, communication between a signaling cell and a responsive cell can be either pro- or antimitotic. Intercellular communication occurs via a signaling pathway. One cell sends a message in the form of a protein across the intercellular space. A responsive cell will have receptors on the surface that recognize that protein. These receptors have two parts that do distinct jobs. Outside the cell is the part of the receptor that recognizes the signal. The part of the receptor inside the cell transmits the information in the signal to effector proteins called signal transducers. Once the external part of the receptor has bound to the signal, the receptor as a

whole changes in a way that sends the information in the signal through the cell membrane to the internal part of the receptor. Inside the cell, the signal transducers recognize the change in the internal side of the receptor. They in turn experience changes that create an information-bearing cascade. This cascade, in mitotic regulation, ends in the nucleus where the final set of signal transducers turn target genes on and off.

An example, of an antimitotic signal is a protein called Transforming Growth Factor-b (TGF-b). Cancer researchers discovered this signaling protein in the early 1980’s as result of its potent antimitotic activity in adult epithelial cells such as those that form the colon, lung or skin. Where TGF-b signals are missing, there is eventually tumor formation.

When TGF-b is constantly bathing colon cells for example, they do not divide. They are focused on their job of absorbing nutrients and passing them into the blood stream to nourish the organism. When a mutation renders a cell unable to interpret the TGF-b signal, for example via a mutation in the TGF-b receptor, then that cell and its descendent cells grow out of control into a polyp (a pre-tumor stage). If the polyp cells are not recognized and removed by colonoscopy, then they continue to grow. Eventually, and inevitably without treatment, a polyp becomes an adenoma (tumor). The hypermitotic state allows additional mutations that result in adenoma cells changing their focus from nutrition to division. Eventually, there are enough cells in the adenoma to protrude into the adjacent blood vessel. When adenoma cells are washed off by blood flow, these tumor cells are able to attach to the blood vessel at another location. Then it is only a matter of time before the tumor cell figures out how to get through the blood vessel and into the underlying tissue. There it will continue to divide and in time a single adenoma cell can form a tumor in the new tissue. The dispersal of tumor cells is called metastasis and it is the most dangerous aspect of cancer.

Does TGF-b signaling do anything else? The answer is yes. TGF-b signaling is also important during embryonic and post-natal development to coordinate mitosis across the many cells of a growing organ or tissue. The process of development from a single fertilized egg into a human being, or any other multicellular organism, is highly complicated, tightly regulated and incompletely understood. Cells descendent from that fertilized egg must become all the cells you were born with. Thus, the process must be coordinated in some way. It would not be compatible with life if a subset of heart cells showed up in your neck while others were in your leg. In the human body, all of your heart cells should be present slightly left of center in your chest so that they can assemble into a beating organ to keep you alive. How do cells in the embryo know where they are supposed to end up?

Intercellular signaling in the fruit fly model organism

The process of understanding what goes right during normal human embryonic development serves as a model for understanding what goes wrong in human adults that leads to cancer. While this is a nice hypothetical, people take offense if you want to conduct medical experiments on unborn babies. Thus, the research community employs model organisms to teach us what they can about developmental biology. There are several well-studied model organisms and different models have different strengths.

One of the strengths of the fruit fly model organism is that it naturally has its own version of TGF-b. For the past 25 years my lab has utilized both embryonic and post-natal development in the fruit fly, in conjunction with the powerful genetic tools available in this model organism, to understand how TGF-b signaling works, how it is regulated and how it coordinates mitosis in responsive cells. What follows is a brief introduction to the fruit fly model organism.

For biologists, especially biology teachers, this topic needs little introduction. Scientists, starting with Thomas Hunt Morgan have been studying fruit flies as a model for understanding biology for well over a century. Roughly 10 Nobel prizes have been awarded in that time to scientists who study fruit flies. One example is Professor Edward Lewis of the California Institute of Technology who shared the Nobel Prize in Physiology/Medicine in 1996 for his pioneering breakthroughs in understanding the embryological processes of limb placement and formation. More recently, three fruit fly scientists including Professor Michael Rosbash of Brandeis University, shared the Nobel Prize in 2018 for their discovery of how the biological clock works. Everyone knows about jet lag – when you fly to a different time zone your body takes a while to catch up. Ever wonder why that is? Scientists studying fruit flies discovered the time-keeping proteins and then showed that the clock governing daily rhythms such as the sleep/wake cycle is the same in flies and humans.

Aside from the sharing new details about TGF-b signaling and cancer, one of the primary points that I hope to make in this article is the ability of model organisms to inform us about human biology and human disease. It’s an old saying among model organism biologists, “if we could cure all diseases just by studying humans and their cells in a dish, then they would all be cured by now”. Studying tumors or tumor cells alone cannot reveal all the causes of cancer. By the time a cell is a tumor it’s too late to fully understand how it got there.

Mutations facilitate studies of intercellular signaling in fruit flies

While focusing on the model organism Drosophila melanogaster (the species of fruit fly studied most commonly in biomedical research), I employ a genetics approach. The field of genetics is even older than the study of fruit flies, and originated in the 1860’s with the Austrian monk Gregor Mendel. His studies with peas in a monastery garden over many years identified principles of genetics are known as “Mendel’s Laws of Inheritance”. They are among the cornerstones of modern medicine.

The core of the genetic approach is to exploit mutations, both natural and induced for experimental purposes. A mutation is a change in the DNA that leads to a change in function. For example, a change in the DNA that encodes a protein usually leads to that protein becoming crippled. This type of mutation in a tumor suppressor gene can lead to cancer. A change in the DNA that regulates the expression of an oncogene can result in the oncogene being expressed at the wrong time or place. This type mutation also causes cancer. Geneticists break genes via mutation to see what goes wrong and then to intuit the normal function of the broken gene from the mutant phenotype. My lab utilizes mutations to understand TGF-b signaling in fruit flies.

One of the TGF-b family members we studied for many years is called dpp (short for decapentaplegic). Note that by convention a gene name is written in italics with all lower case and the name of the protein that the gene encodes is written in regular font with the first letter capitalized. This gene is not the focus of this article but it serves as a nice example of the genetic approach in fruit flies. Specifically, different mutations in the dpp gene (a change in a gene such as a mutation is called the genotype) lead to different visible effects on a fly (the visible effect of a mutation is called the phenotype). Here are three examples of dpp mutant phenotypes and their causative mutations. Mutations in a specific regulatory region of dpp turn it off in a several tissues leading to adults with small, deformed limbs (antennae, wings, legs etc.). Mutations in the region that codes for the Dpp protein turn it off completely leading to embryos with all fronts and no backs (imagine multiple belly buttons going all around your waist along with the organs that go with it and no spinal cord). These embryos die very early in development and never hatch out of the egg. Lastly, mutations in a different regulatory region of dpp turn it off in one specific tissue leading to larvae with no stored fat. As a result, when they form their cocoon and attempt metamorphosis (a life stage with no eating), they cannot survive and never emerge as adults.

Clearly, in every case something goes wrong as a result of the mutation in dpp, but what the breakdown is, is not obvious from the phenotype. For example, if all mutations led to small phenotypes, then a reasonable hypothesis would be that dpp is a growth stimulator. Or if all mutations lead to lack of a front to back distinction, then a reasonable hypothesis would be that dpp is a polarity creator. Or if all mutations lead to metabolic defects like loss of fat, then a reasonable hypothesis would be dpp encodes a digestive enzyme. Before dpp was shown to encode a signaling molecule, how it functioned eluded investigators. Once dpp was shown to be a TGF-b related signaling molecule, testable hypotheses for dpp’s involvement in each of its distinct mutant phenotypes were rigorously evaluated.

TGF-b signaling proteins are present in flies and humans

Stepping back, it is well known that dpp has many relatives. These relatives have wide species distribution with TGF-b family members found in all multicellular animals including humans. TGF-b proteins are highly efficient in transmitting information and there are typically multiple family members in each species. For example, humans have 33 TGF-b family members. Given that all animals inherit their genetic information from their parents via DNA, that

DNA encodes for proteins and that proteins are composed of only 20 building blocks called amino acids, the degree of similarity between a pair of TGF-b proteins can be calculated like the relationship of an individual to their parents versus the relationship to their cousins. The math is simple in theory. With only 20 amino acids, at the same position in the protein (say amino acid number 27) the chance that two family members have the same amino acid is 5% (1/20). If in that position the two family members have the same amino acid, their similarity is 100%. The average of the similarities, over the roughly 500 amino acids in each of the two family members, will be a number between 5% and 100%. In pairwise comparisons of two family members to a third family member the pair with the larger fraction of similar amino acids is considered more closely related. Just like a person tends to look more like their parents than their cousins.

Expanding these pairwise comparisons to the whole family of TGF-b proteins generates a family tree of relationships. The seven family members in fruit flies are a small fraction of the family. One notable point from the first TGF-b family trees was that on a subset of branches a fly and a human protein were more closely related to each other than the fly protein was to any other fly proteins and the human protein was to any other human protein. This very high cross- species similarity is the reason an approach to understanding cancer using fruit fly genetics is relevant to human health. This scenario applies to dpp and two human TGF-b family members Bone Morphogenetic Protein2 and Bone Morphogenetic Protein4 (BMP2 and BMP4).

In pairwise comparisons BMP2 and BMP4 are 90% similar while Dpp from flies is 75% similar to each of the human genes. No other fly or human gene is more closely related to BMP2/4 then Dpp. The relationship between BMP2, BMP4 and Dpp is analogous that of three brothers with two of them twins. The strong similarity led scientists to see if the proteins could fulfill each other’s function in cross-species experiments. These studies showed that they could. The similarity is biologically meaningful, rather than purely mathematical as described above.

In one experiment, BMP2 and BMP4 from humans were expressed in fruit fly embryos where they rescued the dpp mutant polarity defect allowing the embryos to hatch. Of course scientists cannot do the reverse experiment with humans, but scientists can test Dpp functionality in human cells grown in a culture dish. The investigators chose an experiment with immature bone cells. In adults these cells remain quiescent unless triggered by a signal from an injury to become mature bone during the healing process. In this experiment BMP2 and BMP4 were found to be potent signals for inducing the formation of mature bone cells. Fruit fly Dpp is also a potent signal in this experiment. Scientists call this pair of complementary results (two human proteins function in flies and a fly protein functions in humans) evidence that these three proteins perform the same function in their respective organisms. The results serve as prima facie evidence that if you study Dpp in flies, then you are going to learn about human BMP2/4.

Smad proteins transduce TGF-b signals

As a secreted protein, a TGF-b signal exits the cell where it was produced and navigates the extracellular space. The most obvious question is – how do TGF-b signals know which cells will respond to them? The next question is – how does the information in a TGF-b signal move from the exterior cell surface to the nucleus of a responsive cell. The process that begins with attracting the informational signal to a responsive cell and ends with changes in the expression of target genes in the nucleus is called signal transduction. Information transfer starts with a membrane spanning receptor that recognizes the signal with its extracellular part and transmits the information through the membrane to its intracellular part. Movement of information downstream of the receptor involves a cascade of interacting proteins. The last of these signal transduction proteins acts as a transcription factor that turns target genes on and off.

If a specific TGF-b signal is telling a receptor expressing cell to stop mitosis or do not begin mitosis, then it is serving as an antimitotic signal. Mutation and loss of function for the transducers of an antimitotic signal leads the cell to unregulated mitosis. A signal transducer for an antimitotic signal is thus a tumor suppressor. The primary signal transducers for TGF-b family members are called Smad proteins. Twenty-five years ago several colleagues and I discovered the first of these proteins during a genetic analysis of dpp mutations in flies. We named it Mad. Soon after other colleagues began finding similar proteins in their species. The name became Smads by agreement of the discovering scientists.

Smad proteins also belong to a large family that shares key features with the TGF-b family. For example, there are human and fly Smad proteins that are extremely similar like Dpp with BMP2/4. In my lab and others it was shown that human Smad proteins such as Smad2, when expressed in flies could fulfill the function of the fly gene dSmad2. Thus studies of dSmad2 in flies are directly relevant to understanding the functions of the human tumor suppressor Smad2. This is another example of the fruit fly aiding our understanding of cancer.

Stated formally, Smad proteins are both necessary and sufficient to transduce TGF-b antimitotic signals. The signal transduction process downstream of TGF-b proteins is extremely simple. There is a pair of transmembrane receptors that form a complex, one member recognizing the TGF-b signal outside the cell and the other activating a Smad protein via an enzymatic reaction inside the cell. The activated Smad then finds a partner Smad to form a dimer. The pair of Smads then relocates to the nucleus to turn genes on and off. Often the Smad multimer is joined in the nucleus by a cell-type specific protein to direct them to the correct target genes. The signal transduction cascade is simply a single Smad protein (with partners) that transmits the information from cell membrane to the nucleus to initiate an appropriate response.

The simplicity of information transfer has encouraged Nature to employ TGF-b signaling proteins over and over to tell cells do not divide. TGF-b signaling is very successful across the animal kingdom with humans having 33 TGF-b family members. Most of these are capable of eliciting responses in multiple tissues. As a result, human Smads proteins are potent tumor suppressors and Smad mutations are a hallmark of many tumors. Thus while TGF-b signaling itself is very simple, the regulation of its transducing Smad proteins is extraordinarily complicated to avoid unintended consequences.

dCORL regulation of dSmad2

The Sno protein was originally thought to regulate TGF-b signaling strictly as an antagonist. This was because Sno and its sister protein Ski, in gain of function experiments in human cells, block TGF-b antimitotic signals by binding to Smad2. This prevents Smad2 from forming a transcription factor complex, allowing the cell to initiate mitosis. We showed with genetic loss of function experiments in flies that dSno actually functions as a pathway switch. Sno family members bind to Smad4 where they antagonize signaling by one set of TGF-b family members while facilitating signaling by others.

During our studies of dSno and the Sno/Ski family we discovered that these genes have close relatives called CORL proteins. They were first identified and named CORL by scientists studying mouse embryology. I was intrigued, since at the time this gene was completely unstudied in flies. We initiated an analysis of the fly gene that we named dCORL. As noted for BMP2/4 and Sno/Ski there are two human CORL genes for each fly gene. Technically the human genes are named SKOR1 and SKOR2, but for simplicity I will refer to the human, mouse and fly genes collectively as CORL genes. Like BMP2/4 and Dpp, the fly and mammalian CORL proteins have strong amino acid similarity and show functional conservation. In cross- species experiments we showed that both mouse CORL proteins could replace dCORL function in dCORL mutant flies. We then tested the hypothesis that dCORL functions, like its relative dSno, as a TGF-b pathway switch.

We showed that the expression of dCORL is nervous system specific at all ages. The picture at the beginning is dCORL expression in the brain of a one-day old unmated adult female. The cells in green show dCORL, the “goalpost” in blue are axons of a marker we use to determine the orientation of the brain (dorsal toward the top and anterior toward the reader), and red is the expression of a gene with widespread expression in neural cells nuclei.

Not only is dCORL neural specific, but it is specific to the Central Nervous System consisting of the brain and ventral nerve cord (analogous to the spinal cord in humans but not shown in the image). We do not see dCORL in the peripheral nervous system or in any other tissue at any time. The two mouse CORL proteins are also only expressed in the brain, specifically in Purkinje cells and dorsal interneurons of the cerebellum.

Our colleagues studying mouse CORL proteins told us that they were unsure of CORL function in brain cells. They hoped we could shed some light on this topic with genetic studies of a dCORL mutation in flies. We generated that mutation and noted that a subset of dCORL mutant flies live to adulthood. Then we identified defects in their brains. One defect is in the mushroom body that is involved in learning and memory. Normally shaped like a “goalpost” this region is smaller and misshapen in dCORL mutant adults. We demonstrated that this defect leads dCORL mutant adults to fail a climbing test due to an “attention deficit”. Their limbs work fine but they quite the test in the middle and begin to wander aimlessly.

We then showed that dCORL is a regulator of Smads by studying it in the same biochemical assay employed to study dSno. In that assay we found similarities and differences between the two proteins. While dSno was able to act as a switch by binding to Medea (Smad4 in human), dCORL bound to dSmad2 (Smad2 in human). Further while dSno binds and switches Smad4 function, dCORL binds and facilitates Smad2 function.

Both dCORL features identified in these experiments, binding to Smad2 and strict facilitation of Smad signaling, are incompatible with a switch function. Thus my initial hypothesis was wrong. As a brief aside, there is nothing wrong with being wrong. Hypothesis testing is how science makes progress. No one can always guess correctly how Nature actually works, and so understanding how to make the best of a wrong hypothesis is an important skill. What we did learn is that dCORL’s similarity to dSno was enough to allow it to bind to Smads.

The paradox: a tumor suppressor and its regulator in neurons

All of the above data was coherent and thus satisfying, but there was a nagging problem. What are dCORL and the tumor suppressor dSmad2 doing in the adult fly brain? Clearly they are not regulating mitosis because mature neurons, in the brain or anywhere else, do not divide. The process that generates all of the neurons in the body relies on stem cells that divide asymmetrically. Instead of producing two equal daughter cells that are identical to the parental cell, as in normal mitosis, stem cells produce unequal daughter cells. One daughter is unlike the parental cell. This daughter divides again to become two neurons. The other daughter cell is a neural stem cell like the parent. Thus the supply of neuron-producing stem cells remains unduplicated but also undepleted. The process continues until all neurons that are needed by the organism are in place. Then the stem cells stop dividing and eventually die. This is why neural injuries like those to the spinal cord are permanent and unrepairable. A recent discovery in the mammalian brain is a small group of neural stem cells that divide during adulthood, providing some hope for future neural repair, but these have not been found in fly brains.

That mature neurons do not divide is quite logical. Think about a sensory neuron in your own body. This neuron could have an axon in the tip of your pinky finger, run up your arm, across your shoulder, has its nucleus someplace along the way, and with one of its dendrites interacting with a spinal cord neuron. Thus if you stick your little finger in the car door, you can feel the pain in your brain. Then your brain sends an electrical impulse back down an equally long motor neuron running from the spinal cord to your pinky finger so you pull it out. How can either of those neurons, perhaps 3 feet long in an average person, shut down normal function, duplicate its DNA and then divide in two? If you stick your finger in the car door during this time, you would be in big trouble.

To summarize, mature neurons do not divide. If one dies then it is replaced by progeny from the division of neuroblasts. Thus we wondered what a tumor suppressor (dSmad2) and its regulator (dCORL) are doing in mature neurons?

The solution: they regulate gene expression

Here I summarize the outcome of two papers we published that shed light on dCORL’s role in mature neurons. In 2012 we studied dCORL in larvae whose neurons all die and are replaced by adult neurons during metamorphosis. There we saw the first hint of dSmad2 and dCORL neural function. Then in 2018 we showed conclusively what dCORL (and dSmad2 in unpublished data) do in adult brain neurons. In short, dCORL and dSmad2 regulate gene expression and not mitosis in adult brain neurons. Evidence for this conclusion is that in dCORL mutants the expression of the Drosophila insulin-like peptide-2 (dILP2) is missing.

As a reminder a dissected brain from a one-day-old unmated adult female fly is shown in the opening image. To be clear, dCORL brain expression does not show any qualitative sexual dimorphism. We have done the same experiments on males (more on quantitative differences between dCORL mutant males and females below). The brain has been treated to reveal the locations of three different proteins. We are detecting these proteins via specific antibodies that recognize them and are tagged with fluorescent molecules. The fluorescent molecules glow in different colors when excited by a laser of a certain wavelength. Thus the antibody to FasII expressed on the surface of axons from neurons in the goalpost shaped mushroom body glows blue (an orientation marker). The widely expressed neighboring protein called Twin of Eyeless or Toy is expressed in the nuclei of many neurons. Thus Toy gives the appearance of glowing red dots rather than a continuous shape like FasII. dCORL is shown in green.

The green glow is not actually from the dCORL protein detected with an antibody, but it’s from a genetic technique that allows a generic protein with a very cheap commercial antibody to depict dCORL expression (making an antibody is costly and time consuming, this shortcut is widely utilized). We see green in the nuclei of a very small group of cells (roughly 14) located in the dorsal anterior region where the two brain hemispheres meet (in the center of the blue goalpost). There is so much green in the nuclei of these neurons that green overflows them and extends down their axons, stretching in unison past the horizontal part of the goalpost toward the ventral side of the brain. There are also scattered green cells that remain unidentified.

The small group of dorsal cells expressing dCORL is widely studied and so we knew something interesting was going on. Based on the location it was likely these cells are insulin- producing neurons. We subsequently proved that the dCORL expressing cells are insulin- producing cells by demonstrating universal co-expression (every cell expresses both) with an antibody to dILP2. We also showed that another protein expressed in insulin-producing cells called Drifter is expressed in a subset of the cells that express dCORL and dILP2. Thus in this region of the brain, known as the Pars Intercerebralis, there are two types of cells. There are those that express dCORL, dILP2 and Drifter (roughly 60%) as well as those with just dCORL and dILP2 (roughly 40%).

To understand what role dCORL is playing in insulin-producing neurons we examined adult brain from dCORL mutant flies. We initially focused on one-day-old unmated females for consistency with the image at the beginning, but expanded the analysis soon thereafter. When we looked carefully at the Pars Intercerebralis, we saw a disruption of the previous pattern for dILP2 and Drifter expressing cells. In the mutant there are only the dILP2 and Drifter cells but not cells expressing dILP2 alone. The cells expressing only dCORL and dILP2 in the normal female brain are not visible when dCORL is absent in the mutant. Thus we conclude that dCORL (and dSmad2 in unpublished data) regulates gene expression and not mitosis in adult insulin- producing neurons.

Loss of dCORL reduces lifespan

We wondered, what is the consequence for the individual of losing these specific 40% of cells in the Pars Intercerebralis. As I mentioned, this gene expression phenotype is common to adult dCORL mutant males and females. We studied general lifecycle characteristics of these adult mutants such as fertility, fecundity and lifespan. We saw no difference in the first two characteristics between normal and mutant flies but dCORL mutants had a significant reduction in lifespan. The table at the end of the article has this data color-coded.

In blue is the evidence for a reduction in lifespan for dCORL mutant males and females compared to normal. The top blue row shows that dCORL mutant female virgins (unmated like the donor of the brain shown at the outset) live for 12 days with a standard deviation of plus or minus 11.5 days. Normal (called wild type in genetics) unmated females live more than twice as long (27 days) with essentially the same standard deviation. This is mathematically speaking, a statistically significant reduction in lifespan for dCORL mutant virgin females. The bottom blue row shows that dCORL mutant virgin males display the same pattern. dCORL mutant virgin males live 21 days with a standard deviation larger than females while wild type virgin males live 30 days. Wild type virgin males live 41% longer and that is also a statistically significant reduction in lifespan for dCORL mutant virgin males. Recall that significant is defined by mathematicians as a probability that the difference between two trials is due to chance is less than 5% or 0.05.

With this data as a baseline, we then compared the dCORL mutant and wild type lifespan in females and males after mating. In the experiment, after single pair mating the male and female are separated and each fly placed with three other single-mated flies of the same sex. In the top red row, we show that dCORL mutant female lifespan more than doubles after mating. This is reflected in the 108% increase from mutant virgin females living 12 days to mutant mated females living 25 days with a slightly larger standard deviation. dCORL mutant mated females live statistically significantly longer than mutant virgin females. A small, statistically insignificant increase in lifespan is seen in the parallel experiment with wild type females (top black row). In fact, dCORL mutant mated females live almost as long (25 days) as wild type mated females (29 days; second black row). This small difference in mated female lifespan between mutant and wild type lifespan is insignificant. These comparisons show that the virgin female lifespan reduction resulting from the dCORL mutation is fully rescued by mating.

In the bottom red row, dCORL mutant mated males live 76% longer than mutant virgin males. The male lifespan increase due to mating is longer in an absolute sense (an extension of 16 days versus an extension of 13 days for females) but shorter in relative terms (76% increase versus 108% increase). Nevertheless, the extension due to mating is statistically significant for dCORL mutant males too. A small and statistically insignificant decrease in lifespan is seen in the parallel experiment with wild type males (third black row). In fact, mutant mated male lifespan is 144% longer than wild type mated males (green row). These comparisons reiterate the findings from females. To summarize, dCORL mutant virgin males and females do not live as long as wild type, but after mating mutants of both sexes live as long or longer than wild type.

Overall a very interesting network of relationships between aging, mating, insulin and dCORL is revealed by this analysis.

Implications for human health

Returning the question, “what is a tumor suppressor doing in a neuron”? For dCORL (and dSmad2 in unpublished data), we found that in insulin-producing neurons of the adult fly brain they regulate the expression of dILP2. Further, we showed that the loss of dCORL and thus dILP2 expression leads to a reduction in adult lifespan. The logic for extending this finding to mammals is that the expression of mouse CORL proteins in the developing brain parallels that of dCORL and both mouse CORL genes function in flies. Perhaps mutations in human CORL or Smad2 are part of the relationship between insulin and longevity shown by the reduced lifespan of patients with insulin resistance syndrome. Taken together the data leads to the very interesting new question of how this cancer mutation and aging network is wired? This is what we are trying to figure out now.

Acknowledgements

I thank the members of my lab who have contributed to the analysis of dCORL: Beth Cornell, Sam Goldsmith, Janine Quijano, Mike Stinchfield, Norma Takaesu and Nancy Tran. I have also benefitted from the contributions of external collaborators: Christos Consoulas, Pascal Kahlem, Mike O’Connor, Keiji Miyazawa and Kohei Miyazono. Our studies of dCORL have been funded the National Institutes of Health.

Stuart Newfeld, Ph.D. Professor of Life Sciences

Arizona State University, Tempe AZ 85287-4501

Studies of cancer mutations lead to a new discovery about aging