David enjoys walking, but his balance and coordination are not good. He has taken many falls in his lifetime. Sometimes he hits his head. He has had numerous stitches over his eye, on his forehead and has injured his nose more than once. I have always worried that these falls might have a cumulative effect on his brain.
In 2019, just before COVID-19 started to spread, a group of researchers in Shandong, China asked the question: what does the gene CELSR1 do? CELSR1 is on my list of high importance genes of PMS (see Which PMS genes are most important?) One copy of CELSR1 is missing in about half of all people with Phelan-McDermid syndrome who have a terminal deletion of chromosome 22. Only a few months ago I wrote about the relationship between CELSR1 and lymphedema (see Have we found the gene that causes Lymphedema?), based on research done elsewhere in China. As I describe in that blog, research on CELSR1 has become truly international.
It has been known for years that the CELSR1 gene is especially active (producing the celsr1 protein) following a brain injury. The celsr1 protein is associated with the creation of new neurons and enhancing the blood flow in the brain. Until now, it was not clear if the increase in celsr1 protein following injury protects the brain. Now we know a lot more. I will describe the research (see the scientific paper here).
The researchers used a very modern technique. They used a “short hairpin RNA” (shRNA) to reduce protein production by the CELSR1 gene in rodents. This method cut the level of the celsr1 protein production by 50%. Humans missing one copy of the CELSR1 gene should also have about a 50% loss in protein production. In other words, this is an animal model of PMS!
Using a standardized procedure to invoke a brain bleed in rats, the researchers looked to see if rats lacking half of the normal celsr1 sustained more brain damage than expected. Both the amount of brain damage and the ability to walk were significantly affected by the reduced celsr1. The researchers then refined their methods to analyze exactly how celsr1 was protecting the brain. Celsr1 is an amazing protein. It protects injured neurons from dying, it speeds up the production of new neurons and it promotes the formation of new blood vessels to provide oxygen to the injured region. The researchers were also able to pinpoint what part of the brain jumps into action when the celsr1 is needed.
For years, scientists have been looking for natural substances that might protect the brain after a bleed. These “neuroprotective” substances have been studied and they include familiar chemicals, like IGF-1. Thus far, none of the known chemicals or their derivatives have provided useful treatments for brain injury, intellectual disability or autism. The question becomes, is celsr1 nature’s way of protecting the brain? Perhaps we should be focused on restoring the normal level in people missing the CELSR1 gene?
We know that, in general, people with PMS who have larger than average deletions (average size is about 4.5 Mb) have more deficits/problems than people with smaller deletions. This has been shown repeatedly. What we do not know is whether people with deletions greater that 4.5 Mb accumulate brain damage over time. The brain bleeds (strokes) in the rat study are severe enough that a few of the animals die from the bleed. But, are smaller events — perhaps events that David experienced — causing serious long term problems because of the missing CELSR1? If these problems accumulate, perhaps the deletion size will be more important in older people with PMS.
This new information about CELSR1 reminds us that with each important gene comes an opportunity to help people with PMS. Some of the important genes remain poorly characterized. As we are seeing with CELSR1, the more we study these less-well-known genes, the more opportunities we have to develop new treatments.
Originally posted 4 July 2021. Updated 28 May 2023.
David loves to walk. He has a strange cycle of activity and there are periods where he just cannot sit still. So, he walks and walks to burn off the need to have activity. Getting him to walk was a major undertaking. It took 6 years of hard work with the physical therapist and his family at home until walking emerged. His gait is clunky. He walks as if he has weights in his shoes. But, he walks and walks when he needs to. Walking is so important for David in so many ways.
David might not be able to walk if he had serious lymphedema. David has peers with 22q13.3 deletion syndrome (Phelan-McDermid syndrome, PMS) who suffer from lymphedema. Lymphedema is seen much more often in cases of large terminal deletions. Undoubtedly, there is a gene somewhere in the list of PMS genes that causes lymphedema (see How do I know which genes are missing?). A group of medical researchers in China who study lymphatic surgery and pediatrics teamed up to treat a 20-year-old woman with lymphedema due to PMS. After careful examination of the woman, her genetics and the pathological mechanisms that can lead to lymphedema, the research group has come up with a very plausible explanation. Subsequent work has verified the gene involved.
The key to this investigation is that the young woman suffers not only from lymphedema, but also from protein-losing enteropathy (PLE). PLE is a condition where the body has an insufficient amount of protein. PLE is related to lymphedema when problems with the lymph vessels supplying the lining of the intestine cause poor absorption of nutrients. When improperly formed intestinal lymph vessels or blockage of lymph flow from the intestines (called lymphangiectasia) occurs, you can have insufficient protein absorption. In 2013 it was shown that loss of the gene celsr1 in mice leads to improperly patterned lymph vessels and lymph valves. In 2016 it was shown that an error in the CELSR1 gene in several generations of a family caused inherited lymphedema in their legs (Gonzalez-Garay et al 2016). Through this understanding, Xia and colleagues have provided a strong case for human CELSR1 as the causal gene for lymphedema in PMS (Xia et al 2021 Lymphedema complicated by protein-losing enteropathy with a 22q13.3 deletion and the potential role of CELSR1).
The researchers were not able to specifically demonstrate GI lymph flow was disrupted in their clinical case, partly because the young woman was not a candidate for lymphography and there was no clinical justification for sampling with endoscopy. Still, the lymphedema was quite clear in this patient and likely the cause of PLE. More importantly, the patterning gene CELSR1 can explain both lymphedema and PLE.
So far as seen the in literature, PLE seems to be very uncommon in PMS. Maybe it showed up in this case because of a diet less protein-rich than in many Western countries. Would we have discovered this sooner if we had more low-income or otherwise disadvantaged people in our community? Diversity and inclusion is not just an ethical concern. It is at the heart of scientific understanding.
A new study was published in May 2023 confirming CELSR1 as the central gene for lymphedema (https://pubmed.ncbi.nlm.nih.gov/37232218/). In this study, the authors reviewed 404 cases from the PMS Foundation Data Hub and showed that the likelihood of lymphedema is greatest when CELSR1 is missing.
Perhaps it is time to look more closely at protein absorption in people with PMS. Are there subclinical cases of malabsorption? Since CELSR1 is the central gene for lymphedema, people with PMS terminal deletions of sizes greater than 4.3 Mb are at risk for lymphedema and (perhaps) PLE. In both cases, the penetrance is incomplete: the deletion of one copy of CELSR1 makes someone more prone to these problems. We should not dismiss the importance of this work even if the problems are not expressed in every case.
I see an additional take home message from this research. It is clear that helping people with PMS requires understanding and considering all the genes lost when a deletion occurs. This requires input from many different medical and research fields. As parents of children with PMS, it is our job to encourage and reward novel and valuable contributions. In science, one way to acknowledge the contribution of a fellow scientist is to cite his/her work. This blog serves as a thank-you to Song Xia, Wenbin Shen and colleagues. This blog is not the first time I have acknowledged the work of Gonzalez-Garay et al in the Sevick-Muraca laboratory. The mouse work on the Celsr1 protein was done in the Makinen laboratory in the UK and the mouse model was developed by Fadel Tissir in Belgium. Fadel Tissir has studied Celsr1 over many years. All of these people deserve recognition for their contributions. I have personally emailed Dr. Tissir to thank him for his contributions to PMS. Letters of encouragement are a powerful and largely untapped tool we parents have to accelerate the discovery of new treatments for our children. I think we miss an opportunity each time a relevant paper gets published without thank-you notes from our community.
Phelan McDermid syndrome (22q13.3 deletion syndrome, PMS) comes in three flavors. By far, the most common variety of PMS that has been observed is a terminal microdeletion. A terminal deletion occurs when from three to 108 genes are chopped off the end of chromosome 22. There is a growing group of identified patients who have structural changes restricted to the SHANK3 gene, a gene very near the end of the chromosome. These changes can be minimal (e.g. single nucleotide polymorphism (SNP)) or substantial, with large parts of the gene duplicated or missing. The least studied form of PMS is found among those deletions that damage the same general region of chromosome 22 without affecting the very end of the chromosome where SHANK3 resides. These are often called interstitial deletions.
Genes of PMS
The SHANK3 gene is an elephant. It is a large, complicated gene and its disruption often (but not always) has severe consequences for its owner (the person harboring the mutation). Numerous studies show that the many genes of 22q13.3 contribute to intellectual disability, muscle tone and movement disorders, hearing problems, lymphedema, autism, schizophrenia and other problems. But, the penetrance of SHANK3 mutations, that is the likelihood that a pathological variant of the gene affects the owner, is high. So, it garners a lot of attention. Especially for people with the second flavor of PMS, fixing SHANK3 could ameliorate many of the symptoms associated with its disruption. For terminal deletions, the most common form of PMS, fixing SHANK3 should have at least a very helpful impact.
There are at least 18 genes that may need repair to fully restore normal function in a child with a large terminal deletion. Average size deletions (about 4.5 Mbase) would benefit from the repair of 10 genes. This blog discusses what we hope will be the future of PMS treatment: genetic repair. While regular medical treatments have been helpful for some problems, no experimental drugs have had an impact any better than physical therapy, speech therapy, ABA or other standard treatments for children with developmental disabilities. Although drug companies are sponsoring testing on children with PMS, these have traditionally been drugs looking for an application rather than targeted treatment for PMS.
We are in the early days
Excitement over gene editing, especially in stem cells (cells typical of a very early embryo) has led to giant research efforts and early attempts to fix diseases by editing. The same excitement has led to the promotion of unproven and occasionally dangerous therapies by entrepreneurs who provide services like stem cell infusion. So, let’s be clear. Gene therapy for PMS is well off into the future. There are many reasons for the delay. Some will be discussed below.
Obviously, we are at the earliest stages of gene therapy. Research and initial attempts at gene therapy target diseases that are the safest and most likely to work. This means that the disease most likely to be cured first may be a rather obscure one. It may also be a lethal disease, since the risks of side effects are no worse than letting the disease take its course.
One relatively safe way to test gene therapy is to modify blood products and then infuse them back into the body. This has safety advantages in a few different ways. First, the genetic modification occurs outside the body. The target cells can be modified without risking other cells of the body. Second, the target cells can be tested and evaluated for successful gene targeting (editing the right spot and not accidentally damaging another gene) outside the body. Third, the product can be infused a little at a time to test the benefits/impact gradually. Fourth, blood products usually have a limited lifetime in the body, so if the test begins having problems, the problems will likely go away with time. Thus, gene editing for diseases of the blood are good early candidates.
Another target for early genetics testing is the eye. It is a squeamish thought, but making injections into the eye are relatively easy. (I have had eye injections, so no need to educate me on the downsides.) The inner parts of the eye are well isolated from the rest of the body, which affords safety. Also, only very small injections are needed to bathe the retina. There is little likelihood of impacting the rest of the body. The amazing thing about eye therapy is that modern cameras and computers can take images of the eye with spectacular resolution. These images can track changes. Combined with vision testing, any progress associated with the therapy is easy to assess. As gross as it sounds, if something goes wrong with gene editing of the eye, the eye can be removed. Gene therapy for the eye would be to save vision. If vision will be lost with no intervention, the risks of trying a new therapy may be acceptable. Thus, genetic defects of the retina are high on the list of early therapeutic trials.
The human brain is not well isolated from the rest of the body and is not easily observed from the outside. There is a barrier between the bloodstream and the brain. The so called blood-brain-barrier can make treatment more difficult. Drugs injected into the bloodstream (systemic injections) may or may not reach the brain. Direct brain injection of genetic altering “vectors” may be an approach used in the future, but it is not ideal. Systemic injections would be a simpler way to reach all parts of the brain and to allow repeat injections to gradually reach the desired effect while watching for side effects. So, genetic editing in the brain is not going to be as simple as other targets.
Once we start thinking about genetic manipulations in the body, we need to remember that nearly every cell in the body has the same 20,000 genes. Given that all cells in the body could be the same, what makes heart and lungs different from kidney and brain? The answer is gene expression. Heart cells know they are heart cells because only heart cell genes are turned on, making the proper heart cell proteins in the proper proportions. We hope that injecting a vector into the body will improve brain function, but we must worry about what else may change. We worry in two ways. First, we could modify our target gene in such a way that the brain improves, but the heart (or some other organ) suffers. Second, we could accidentally modify other genes (so called, off target effects) that have no adverse impact on brain cells, but might be bad for the heart. I used the heart as an example, but there are so many different tissues in the body we cannot know which might be adversely affected.
One more consideration is important in choosing a target for the early tests of genetic therapy. The complexity of the organ and the importance of early development in creating its detailed structure influence the age at which therapy may be effective. The brain is complex almost beyond comprehension and its early patterning during development is critical to adult function. Repair of a gene after early development may have limited benefit, and except for the most deadly of disorders, we dare not test interventions on fetuses or babies until we are confident of our methodology. Additionally, the complexity of the brain poses another risk. Increasing the production of a gene may be beneficial for one part of the brain, but could be detrimental elsewhere in the brain.
The challenges of PMS
The opening of this blog hinted at some of the challenges of PMS. Together with the subsequent discussion, we can now list the major challenges of applying early genetic methods to PMS: 1) the overwhelming majority of people diagnosed with PMS are missing many genes and modern methods are still struggling for success with single genes, 2) the gene of greatest interest is large and complex, neither of which is good for gene editing, 3) the brain is a challenging target for genetic manipulation, and 4) PMS is a developmental disorder and we have insufficient information on whether treatment in older children or adults will be beneficial.
Given the many hurdles, it may be a very long time before a comprehensive treatment for PMS emerges. To be sure, there are people working on strategies to manage the disadvantages. One approach is to target only one gene in the hopes that at least some subset of patients might benefit. Likewise, there are model (rodent) experiments that suggest some of the deficits of PMS can be mitigated by treating adults. Genetic treatments for other diseases are addressing the problem of reaching the brain without adversely impacting other organs. Both animal and early clinical trials are creating improved toolboxes for genetic therapies. We hope these technologies converge on better opportunities for PMS.
Progress in the field
Techniques like CRISPR/Cas9 grew out of studies in bacteria and other science far afield from human clinical work. Recent work in E. Coli and other bacteria, as well as work on yeasts and viruses have moved genetic editing towards more precise targeting, the possibility of replacing whole deleted sections of chromosome, and the ability to enhance the activity of the remaining gene after one has been deleted. There are new methods of turning on and off a newly inserted gene, so the proper dose of a gene can be safely titrated to avoid overdose. This blog has not addressed the different methods of gene editing and which ones might apply to PMS. Rather, it provides a framework to understand why one approach or another might be more suitable. Perhaps a future blog will cover the different tools and which ones are likely to provide the first clinical tests.
A strategy for PMS
After reading the scientific literature and talking with scientists, I have a policy suggestion for how we should approach PMS. Currently, the only genetic intervention being pursued for PMS is targeted at the SHANK3 gene. There are historical and logical reasons for targeting this gene. It is highly associated with several developmental disorders. That work should continue, but from a technical perspective, SHANK3 is not an easy gene to treat. The SHANK3 gene can produce from 20 to 100 isoforms (versions) of the shank3 protein, up to 1731 amino acids in size. (By comparison, the UBE3A gene associated with Angelman syndrome produces a maximum of 12 ube3a isoforms, with 875 amino acids being the largest. The MECP2 gene associated with Rett Syndrome, has two isoforms, with 486 amino acids being the largest.) Manipulating a large, complex gene like SHANK3 has consequences, at least some of which will likely be negative consequences. Like all early genetic clinical studies, initial trials will be on a few hand-picked cases. Unfortunately, even among people with SHANK3 mutations, there are many flavors and early treatment methods might not generalize to many patients.
If we are serious about a long-term solution for PMS we need to work on a smaller and more simple gene. Solving problems caused by SHANK3 is going to be a very slow process. We should choose a gene much easier to fix, and yet still has a serious impact in our children with PMS. The idea is to bring relief sooner, and to begin the important task of treating the full spectrum of PMS. There are 17 possible genes, three or four of which are among the best candidates. One possible target is BRD1. This is a regulatory gene that has been very well characterized. It is smaller than SHANK3 and one copy is missing in about 90% of the PMS population. Evidence so far supports the importance of missing only one copy of BRD1.
Titrating expectations, but not hope
There is always a balance between the enthusiasm for research into effective treatments and maintaining a realistic view of scientific progress. Young parents have so many compromises to make in their hope for the future when they receive a diagnosis of PMS. Placing the disorder into perspective takes years. We grow into a life consumed by a child with a major developmental disability. Hope requires embracing the possibilities, maintaining enthusiasm about the future, yet titrating our expectations to match today’s realities. We are a long way off. Our own children will likely be adults, perhaps old ones, before effective genetic treatments are ready for widespread application. We need to be realistic about our expectations without diminishing our hopes. Indeed, this is what raising a child with PMS is all about.
While SHANK3 has been identified as a major contributor to intellectual disability and autism spectrum disorder, there has always been a lingering question of why people with Phelan McDermid syndrome (PMS) have very similar symptoms (phenotype) whether or not SHANK3 is impacted. That is, people with terminal deletions that remove SHANK3 are not very different from people with so-called “interstitial deletions”. This observation is the main reason why the current definition of PMS says that SHANK3 is usually, but not always, impacted (see The 22q13.3 Deletion Syndrome).
Genes with a high “pLI” (genes that are very rarely mutated in the general population) are associated with intellectual disability. For 22q13 they are: SHANK3, MAPK8IP2, PLXNB2, TRABD, PIM3, ZBED4, BRD1, TBC1D22A, GRAMD4, CELSR1, SMC1B, PHF21B, PRR5, SULT4A1, SCUBE1, TCF20, SREBF2, and XRCC6. Most, if not all of these genes contribute to intellectual disability, although some are more certain. While 18 genes may seem like a lot of genes, remember that PMS can delete up to 108 genes. Still, it is impressive that nearly 17% of the genes of 22q13 potentially contribute to intellectual disability. The high number of these high pLI genes explains why intellectual disability is generally more severe with larger deletions, and it explains why interstitial deletions can also produce moderate to severe intellectual disability. But, what about autism spectrum disorder (ASD)? The frequency of ASD in PMS is an area of debate, various studies report anywhere from under 30% to over 70% incidence of ASD. The natural history study of PMS should narrow this range. Regardless, ASD is a component of the PMS population.
So far, three genes of 22q13 have been clearly associated with ASD: SHANK3, MAPK8IP2 and SULT4A1. It is probably no accident that these three genes also show up as high pLI genes.
What is new
Two new lines of research have identified new candidate ASD genes on 22q13. One candidate gene has arisen from studies of epigenetics, the regulation of genes. Genes occupy only about 1% of the DNA. Much of the remaining DNA is dedicated to controlling when and how often a specific gene is put to work. Gene regulation is essential for successful development of the brain (and all other body parts) and for responding to changes in the environment. By environment we can refer to the external environment (e.g., adapting to colder weather) or we can be concerned with the internal environment of the body (e.g., building muscle with increased exercise). There is earlier data suggesting a weak contribution of the gene BRD1 to ASD (see Which genes cause brain abnormalities in Phelan McDermid syndrome?). However, epigenetic studies have reinforced the connection between BRD1 and ASD (see Next-gen sequencing identifies non-coding variation disrupting miRNA-binding sites in neurological disorders). So, BRD1 can be considered an autism-associated gene.
Perhaps more exciting is the latest work on “second hits”. Second hits are when one gene is lost or damaged, and then the remaining copy of the gene is also affected. Second hits matter because some genes can operate just fine with only one copy. We have two copies of most genes in our DNA, one from each parent. When a person has a chromosomal deletion, which is what happens in over 95% of the cases of PMS, that leaves the person with only one copy of those genes. A second hit is when the remaining copy has a mutation, or even just an dysfunctional variant inherited from one parent. What do we know about second hits? A new paper has identified up to 4 genes of 22q13 that may contribute to ASD, given a second hit (see Recessive gene disruptions in autism spectrum disorder). One PMS gene stands out, KIAA0930 and three others are additional candidates, CHKB, ARFGAP3, and ZBED4.
This rounds-out the current full list of 8 ASD-associated genes: SHANK3, MAPK8IP2, CHKB, ZBED4, BRD1, KIAA0930, SULT4A1, and ARFGAP3.
PMS is most often a terminal deletion of 22q13. A deletion of the chromosome that includes SHANK3 is the most common deletion, but sometimes a deletion does not disrupt that gene. Interstitial deletions can disrupt up to 17 other genes likely to contribute to the moderate to severe intellectual disability that characterizes PMS. Infrequently, a second hit on one or more of 7 genes in the interstitial region of 22q13 may also contribute to autism spectrum disorder. In the overwhelming cases of PMS (called terminal deletions), many genes contribute to intellectual disability and, if autism spectrum disorder is part of the clinical picture, SHANK3 and one or two other genes may also be contributing to that aspect of the disorder.
Phelan McDermid syndrome (PMS) is an intellectual disability developmental disorder. The most common reported form is a “terminal deletion” of the q end of chromosome 22. A terminal deletion occurs when a continuous segment of the chromosome is broken off at the end. Terminal deletions lead to intellectual disability (ID), language problems and coordination problems. Most people with PMS have additional problems (sleep disorder, feeding disorder, seizures, etc.), and these problems can often be severe. Many, perhaps most, have autism spectrum characteristics. I would argue that the hallmark trait of PMS is intellectual disability, because that manifestation of the disorder is the only one that occurs in 100% of people with a terminal deletion. By consensus, cases of unusual variants of the gene SHANK3 are also called PMS. These pathological variants of SHANK3 can have many of the same manifestations as seen with terminal deletions, although not all.
There are two related conditions that have lead to disagreement regarding what is and is not PMS. First, are deletions of 22q13.3 that do not disrupt SHANK3 (often called “interstitial deletions”) still called PMS? Second, does a person with a SHANK3 disruption, but without the hallmark manifestation (ID), have PMS? Out of consistency, I would argue that a 22q13.3 deletions that causes ID should be called PMS, and SHANK3 variants that do not cause ID should not be called PMS. To go one step further, given two people with the same SHANK3 variant, one person might have ID (and therefor, PMS) and another person might not have ID. This unusual combination of circumstances has been observed. Likewise, two people with the same 22q13.3 interstitial deletion might or might not have PMS. This sort of definition is common with neurodevelopmental genetic syndromes. The genetics (called “genotype”) must cause a matching set of manifestations (called “phenotype”), to meet criteria for a named disorder. Some disorders have subtypes, but PMS has no official subtypes yet.
The rest of this discussion does not depend whether you call interstitial deletions PMS, treat interstitial deletions as a future subtype of PMS, or just avoid giving them a name. Understanding interstitial deletions is crucial to understanding PMS, because PMS can include the deletion of up to 108 different genes. Some of these genes are very important. If we want to end the suffering of PMS, we need to know which genes are important and in what way. Interstitial deletions can help teach us. Let me explain how, starting with old radios.
When I was a child I took apart radios to understand how they worked. This was a dangerous undertaking for a young boy. Radios were high voltage affairs in the old days. If the power mains didn’t kill you, burned fingers from hot vacuum tubes or a hot soldering iron left painful reminders of what not to touch. My logic in those days was to remove parts until the radio stopped working. The obviously necessary part was then soldered back into place and the hunt for more nonessential parts continued. When done, I still had a working radio, plus a collection of spare parts. “Working” did not always mean “working perfectly”.
Fifty years later, teams of scientists have used this same logic to grade the importance of each gene in the human genome. One such measure is the pLI score. Think of all people who are healthy enough to have children. Analyze every gene in every one of these people. Make a note of which genes in these healthy people are missing or incomplete in some way. These are the nonessential parts. A gene that is almost never missing or incomplete gets a pLI score of 1. It must be important. A gene that is often missing or incomplete gets a pLI score of zero. You can sound technical by calling the score a measure of reproductive fitness, but the theory is no more complicated than a 10-year-old with a soldering iron. Essential parts are almost never missing from people or radios. The pLI score is the measure of gene importance.
In 2018, a team of scientists studied all deletions greater than 50 Kb in groups of people with ID (Huguet et al 2018). Basically, they asked the question, “Can ID be explained by looking at the deletion size or (similarly) counting the number of genes deleted?” They came up with a formula: add up the pLI scores of all the deleted genes, multiple by about 2.6, then add the impact of known ID genes. That gives you the number of IQ points lost due to the deletion. (Technical note: I have averaged performance IQ and verbal IQ together).
Everyone has heard of IQ to measure intellectual ability. The IQ measure was designed so the median score on an IQ test is 100 across a large population. The work of Huguet et al, including subsequent work shows that you can predict the IQ loss caused by a deletion. A deletion removing genes with a cumulative pLI of 10 will reduce a persons IQ score by about 26 IQ points. The expected IQ of a person with such a deletion would be 100-26=74. This is not a good way to predict a child’s future IQ, since we don’t know if the child’s IQ would have been 75 or 125 without a deletion. But, if the prediction is a loss of 26 IQ points and the person has mild ID, it is likely that the genetic result essentially explains that person’s intellectual disability. There is an on-line tool to help do the calculation.
The next part is a little complicated, but PMS deletions are complicated. I hope everyone can understand at least the main points.
When we apply this IQ calculation to PMS, lots of strange things about PMS start to make sense. I will use a graph to explain. The graph below shows the number of IQ points that are lost when each part of chromosome 22 is deleted. It is a prediction based on some reasonable assumptions (which will not be discussed here). Read the graph from upper left to bottom right. The graph tracks how much the IQ falls as deletion sizes get larger and larger. I have an explanation below for each numbered circle on the graph.
Circle 1: Loss of SHANK3 at the very end of the chromosome (top left corner) has a major impact on intellectual function (IQ). See how the curve drops from 0 to -30 IQ points next to circle 1? I have assumed a SHANK3 deletion costs 30 IQ points, which is a big drop even for an identified intellectual disability gene.
Circle 2: Deletion of the next 1 Mb of the chromosome has a cost of another 20 IQ points. Already, we see that deletion of SHANK3 is not necessary to reduce ID. We also see that even relatively small 22q13.3 deletions (e.g. 1 Mb) can have a large impact over-and-above the loss SHANK3.
Circle 3: See how flat the curve is at circle 3? Additional deletion of the chromosome between 1.1 Mb and 4.1 Mb has virtually no impact. For those people who say that deletion size does not matter, that is why there are so many examples. The curve is flat and, indeed, in that region increased deletion size does not influence IQ.
Circle 5: An important intellectual disability gene shows up about 8.4 Mb from the end of the chromosome. This causes another steep drop in the curve comparable to (perhaps larger than) SHANK3. See my earlier blog about this gene (TCF20).
Circle 6: I have created a hypothetical example of a 2 Mb interstitial deletion. A 2 Mb deletion is about half the size of an average 22q13.3 deletion. This deletion causes a drop in IQ (27 IQ points) that is roughly equivalent to a SHANK3 deletion. Thus, from an intellectual disability perspective, interstitial deletions can easily be equivalent to other, more common cases of PMS.
This method of studying IQ impact of chromosome deletions was not created specifically for PMS, but it seems to apply very nicely. To accurately apply this method, we need to accurately measure the IQ cost of a complete SHANK3 deletion without including the effects of other genes. Calibrating the IQ cost of deletions, including SHANK3 loss, can be done by carefully studying cases of interstitial deletions. Current data from the longitudinal PMS studies may be sufficient if more effort is put into the analysis of interstitial deletions. Finally, the method can be used to identify who might need further testing. If the estimated IQ loss does not agree with the deletion size, the person could have other genetic issues worth exploring: a mismatch could be used to justify more detailed testing (e.g., whole exome sequencing).
The research in IQ loss associated with chromosome deletion shows that, for most people with 22q13.3 deletion syndrome, fixing SHANK3 is likely to be beneficial, but not a cure. SHANK3 accounts for less than half the intellectual disability for an average size deletion (4.5 Mb) and less than 25% of a large deletion. Finally, we need to take interstitial deletions more seriously. From a scientific perspective they are hugely informative. From a PMS perspective, they are part of the same disorder.
Interstitial deletions matter because 1) they can be used to help calibrate IQ loss as a function of deletion size, 2) they can help identify which genes cause some of the manifestations of PMS not caused by SHANK3, and 3) effective treatments of PMS will depend upon how well we can treat all the important genes of 22q13.3.
David, like most people with Phelan-McDermid syndrome (PMS), is missing one of his two SHANK3 genes. One copy of SHANK3 is gone, the other copy is intact. Some people with PMS have both copies fully intact (often called interstitial deletions) and some people have small or large disruptions of SHANK3 without affecting other genes (often called variants or mutations). For the vast majority of people with PMS, loss of SHANK3 contributes to the problems of PMS. Fixing SHANK3 has been a priority in the PMS world, although fixing SHANK3 won’t necessarily fix PMS (see Which PMS genes are most important?). I am an advocate for studying (and fixing) the other critical genes of PMS (see Looking for opportunities and Why don’t we have better drugs for 22q13 deletion syndrome?). Too much focus on SHANK3 has been an impediment to progress. Regardless, this blog is about SHANK3 research.
Science is slow
The autism world is hot on SHANK3, so research on this gene has moved forward relatively quickly. That sounds like good news, and it is, as long as we recognize that “quickly” is measured in decades. Research on SHANK3 started when my son, David, was 12 years-old. He is now 32. Over the past 20 years there have been many papers heralding the “rescue” of autism-behaviors in SHANK3 mice. Parents need to understand that “rescue” is mouse research jargon for “our genetically modified mouse is different from normal mice, and the drug we tested made them a little more like normal mice”. It does not mean “a treatment is almost here”.
The reality of research progress is less rosy than the headlines. Most research studies are done these days on rodents. A mouse with a modified SHANK3 gene is nothing like a human with PMS. Most mouse models are SHANK3 mutations, not deletions. Most people with PMS have deletions (see Gene deletion versus mutation: sometimes missing a gene is better), and human deletions affect other critical genes (see How do I know which genes are missing? and Which PMS genes are most important?). Most “SHANK3 mice” have mutations in both copies of the gene. Mutation of just one copy often results in no detectable effect. Contrast that with humans, where it is unlikely that humans can even survive without at least one working copy of SHANK3. Further, although this should be obvious, only rodent researchers talk about “autism-like behavior” in rodents. It is a rather strange concept. Measuring autism in PMS patients is difficult and sometimes controversial. There is no measure of autism in rodents, just tests to see if a mouse prefers exploring another mouse over an inanimate object. Why a mouse might do that is anyone’s guess. Certainly, you won’t have much luck asking the mouse.
It is not simply bad luck that multiple drug rescues have been reported in mice without the development of any PMS or even autism drugs that work on the core symptoms in people. The reality is that not enough is understood about the relationship among genes, drugs and behaviors. To date, there are no PMS-specific drugs currently available for testing on people. If you get invited into a drug study, that drug was invented for some other purpose. Most often, it was a drug that failed its original purpose and researchers or drug companies are looking for a different disease group to test it on.
Despite the limitations of mouse research for testing drugs, mouse research does help us learn about the proteins used in the brain and what category of drugs might someday be useful for treatment.
Twenty years of looking at SHANK3 has laid important groundwork. More recent studies have benefited from these earlier studies and from development of new research tools. Researchers are beginning to address the complexity of SHANK3 regulation. SHANK3, like most genes, is simply a recipe for making its protein, shank3 (note lower case spelling, no italics). The recipe is copied onto a template called mRNA, and then many copies of shank3 are manufactured using the mRNA template. The shank3 protein is then delivered to the synapses that need more. Like any manufactured commodity, you need to manage the supply to meet the demand. The manufacture takes place at the factory of the cell (soma) and shank3 is used in 10,000 to 100,000 or more synaptic sites elsewhere in the cell. Thus, there is an ordering and distribution network throughout the cell. Orders for more shank3 come from thousands of different sites. SHANK3 regulation is the complex process of making sure the right amount of shank3 is available at each synapse at any given time.
Synapses are the communications connection between neurons. Each synapse has a different strength of connection, and together, turn the protoplasm of the brain into an amazing computer-like machine. The machine is constantly adjusting itself to perform the tasks we call learning, memory, skill acquisition, and decision-making. Like a muscle, when a synapse is used more, it tends to get stronger and larger. That is how the computer adjusts itself. The increase and decrease of shank3 is important for these adjustments. Thus, processes like learning and memory rely on having the right amount of shank3 at the synapse. Regulation of shank3 production, distribution and utilization starts with the amount of synaptic activity at each of thousands of synapses. Complex processes connect synaptic activity with every step of shank3 production and distribution.
New research on SHANK3 regulation
There are two new papers on SHANK3 regulation that represent the next steps in understanding how the cell manages the amount of shank3 protein at each synapse. One paper forces us to rethink about what a shank3 deficiency really means.
Campbell and Sheng just published a paper on DUB enzymes and the regulation of shank3 at the synapse (USP8 Deubiquitinates SHANK3 to Control Synapse Density and SHANK3 Activity-dependent Protein Levels). DUB enzymes prevent the destruction of a protein. Normally, shank3 is destroyed after use by the USP system. These authors have identified an enzyme “USP8”. When the synapse gets very active, USP8 finds shank3 (and shank1) molecules already tagged for destruction by the USP system, and untags them. It is part of the cell’s natural system to retain extra shank3 in anticipation of needing to build up the synapse for future increases in activity. The authors point out that drugs might someday be found to mimic USP8 and help increase the amount of shank3 at the synapse of those people who have insufficient shank3 production.
In the other recent study, work by Yan and her colleagues has teased-apart some of the communications between the synapse and the nucleus used to regulate shank3 production (Social deficits in Shank3-deficient mouse models of autism are rescued by histone deacetylase (HDAC) inhibition). They show that too little shank3 at the synapse can trigger movement of β-catenin to the nucleus. The surprising part is that when β-catenin reaches the nucleus, it triggers a series of genetic effects that lead to unusual mouse social behavior. The unusual behaviors can be turned on and off independently of shank3. Thus, the strange social behaviors of these mice are not caused by shank3 levels directly. Rather, the reduced shank3 simply releases β-catenin. It is a case of synapse regulation system gone awry. These results suggest that shank3 is not the most important protein for poor social behavior. Rather, the reduced level of shank3 triggers a series of events that improperly regulate other brain genes. Curiously, this agrees with another recent study of proteins in humans with autism (see Which PMS genes are most associated with Autism?).
This is the end of the blog except for a few final thoughts, below. For those who wish to dig a little deeper into the science, I have provided additional background material on genes, proteins and regulation called “A primer on the lifespan of a protein“.
A primer on the lifespan of a protein
This is a primer about the lifespan of proteins. Shank3 is a protein (note lower case spelling for the protein). Like nearly all proteins, it is born, gets called-upon to work, does its job, then is dissolved and discarded. The most important point is, the amount of shank3 in the cell is highly regulated. At any given moment, there are complicated processes deciding how much is needed and where it is needed. Usually, some parts of a cell may be building up shank3 supplies and incorporating them into synapses, while other parts of the cell are breaking down shank scaffolding (structures built from shank molecules). I sometimes think of shank3 as a construction material, like plywood. Lots of wood scaffolding may be used to frame a concrete pillar. After the pillar is in place, the wood may be torn down and discarded. Some wood may be discarded at the same time other is being nailed in place for the next pillar, wall or sidewalk. You always want a supply of wood around, but that supply should never be more than the anticipated need. Construction management involves reading blueprints, anticipating needs, ordering the manufacture of materials, and delivering what is needed, on time. Timely and organized manufacture, distribution, utilization and disposal of shank3 very important for the cell.
The SHANK3 gene is simply a blueprint for shank3 protein. The blueprint is converted into a template for stamping out the protein (call mRNA). Gene regulation (via “promoters”) control how many copies of mRNA are created. Each mRNA is degraded and discarded after a certain number of copies of shank3 have been made. While functioning, mRNA is used to stamp out copies of shank3. Shank3 is then transported and collected in a pool of proteins near the synapse. The synapse is a very active site, like the beehive of activity at a busy construction site. Shank3 molecules near the synapse gets incorporated into the “post-synaptic density” as needed. The buzz of activity includes constantly building up and breaking down of shank3 molecules. Old shank3, whether it be after a piece of scaffolding is no longer needed, or if the molecule has been sitting around unused, is degraded and discarded.
Breakdown and disposal of shank3 protein
Shank3 is tagged and trashed by the USP system (ubiquitin-proteasome system). Ubiquitin tags it and the proteasome dissolves it. That is the last step in the life cycle. There is a way of untagging, with a deubiquitinating enzyme (DUB), which I would never even mention except that a recent study looks at untagging as a way to increase shank3 levels at the synapse.
Regulation of shank3 occurs at every step
Many different processes regulate transcription (making the mRNA template), synthesis (stamping out shank3 molecules), incorporation (using the shank3), and degradation (USP system). Each of these processes is a potential therapeutic target for increasing shank3 protein in people who have only one working copy of the SHANK3 gene.
Early work on shank proteins (shank1, shank2 and shank3) focused on how each protein is used at the synapse. Increases in shanks are associated with establishing and maintaining strong synaptic connections, whereas decreases can inhibit the formation of new connections. Too much or too little shank3 affects the sizes or number of synapses. Synapse size is related to information transmission in the brain. Too much information from the wrong channel creates noise. Too little information interferes with learning, memory, skill acquisition or other function. Thus, the early research looked closely at shank3 levels at the synapse.
Shank3 production is largely regulated in the cell nucleus where DNA is found. Each brain cell has only one nucleus, yet it regulates shank3 production for thousands to over 100,000 synapses in that cell. So, the signal to increase and decrease Shank3 production comes from a potentially huge number of synapses and converges at the one and only nucleus. Transcription (making the mRNA template) is regulated by many mechanisms. Most mechanisms influence the “promoter” region of a gene. That is the region where the hardware for reading DNA and making the mRNA template is assembled to do that task. A dizzying array of molecules influence the promoter region. To complicate things even more, shank3 has not one, but 7 total promoter regions that regulate not only when to transcribe, but also which of the 20 to 100 various versions of Shank3 to produce. Yes, there are at minimum 20 versions (isoforms) of shank3 produced during a person’s lifetime. “Turning on and off” a gene is another way of saying the gene is set to transcribe or not. In actuality, neurons that use shank3 don’t turn the genes on or off. Rather, they increase and decrease transcription rate of one, two, up to 20 isoforms of shank3. If it sounds complex, it is. Most papers focus on one or two isoforms. They overlook the rest to make the research manageable. For the moment, it is enough that we recognize that transcription of the SHANK3 gene for making mRNA and many copies of the shank3 protein is regulated at each step. The gene is regulated largely at the promoters, controlling how many mRNA molecules are created for which isoforms. Recent studies have looked for ways to modify the regulation of transcription to compensate for a missing copy of the SHANK3 gene.
Shank3 mRNA is used to synthesize shank3 protein. How rapidly shank3 protein is produced is, like the other steps, under careful regulation. For example, each mRNA does not last forever. At some point it is disassembled (degraded) and can no longer produce protein. Its regulation is yet another possible way to manage shank3 protein production.
As mentioned earlier, shank3 protein is used (or just sits around waiting), and is then broken down by the USP system. Unlike signaling in the nucleus, the USP system is operating at or near the synapse. It can influence shank3 availability on a fine grained scale.
Increasing or decreasing shank3 with blunt instruments
Each step along the life cycle of shank3 is an opportunity to increase the amount of shank3 in the cell. One might be eager to try one or many of the steps on mice or people. This eagerness should be tempered by the potential pitfalls of circumventing the normal regulatory process of the cell. Let’s remember why shank3 is so highly regulated. Shank3 levels must be adjusted at each synapse on an individual basis. Large cells can have up to 200,000 synapses. If we choose to increase shank3 transcription in the nucleus, we may start to force some, perhaps too many, synapses to overproduce shank3. The cell needs to remove unnecessary and unwanted synapses (called “pruning”). Pruning is one of the most important processes in brain development. Another related process is called synaptic homeostasis. One theory about why sleep is important is that it gives the brain a chance to readjust all the synapses to consolidate learning and properly reset the brain for new learning the following day. Both pruning and homeostasis are likely affected by changes in overall shank3 levels.
A drug that simply increases the shank3 in a cell could be beneficial, but we must be wary of blunt instruments. We must be concerned that increasing shank3 by short-cutting the built-in regulation of shank3 may worsen PMS, or perhaps replacing one problem with a new one. This is why a potential drug treatment requires thorough testing in animals to develop a deep understanding of the mechanism of action. Improving one behavior in a mouse is not enough. At a minimum, we need to carefully explore what other behaviors might be affected in the model animal. In this blog, I have not discussed that SHANK3 is used in different ways in different parts of the brain, and regulation is likely to vary for each brain region. Methods to deliver different amounts of a drug to different brain regions are complex and experimental.
Mice are handy experimental animals, but their brain function is not at all like human brain function. In mice, you can remove 100% of all shank3 protein and the mice, although behaviorally unusual, are able to eat, drink, run, play and procreate much like normal mice. Humans missing both copies of SHANK3 are unheard of, most likely because they don’t survive in the womb.
The take-home messages from the new research are simple enough. First, there is much hype in the online press reports and magazines about scientific progress. The fact is, science is slow and we have a long way to go. The details are sometimes complicated, but the basic principles are not. Second, studies of shank3 have not reached the point where we truly understand the relationship between SHANK3 gene loss and the many problems that result. Progress is being made, but, as often happens in science, the latest results tell us what we thought we understood was not exactly correct.
What the new study shows is, regardless how a person gets autism or schizophrenia, the same networks of genes become dysregulated. Let’s first discuss what gene regulation means. DNA is like a well-stocked bakery. A good cook can prepare many different kinds of breads or desserts by choosing how much of each ingredient to use, and when. Just about every cell in the body has the same DNA. What makes one part of the body different from another is how much, and when, each gene is used. DNA cooking is called gene regulation. In autism and schizophrenia, the proportions of ingredients have gone awry.
The green diagram at the top of this blog maps the results of the new study. The researchers found certain critical “modules” (functional groups) of genes that are dysregulated in the brains of individuals with these two disorders. Once, again, these genes are dysregulated regardless of how one acquires autism or schizophrenia. The map identifies the 20 most dysregulated genes in each module (140 total) and how they interact in the brain.
What does this diagram tell us? It says some things we already knew. Autism (and schizophrenia) cause problems in neurons, the brain cells responsible for sensation, thinking and action. Less obvious, autism seems to be related to two other cell types, astrocytes and microglia. Astrocytes nourish neurons. Microglia, which also come in contact with neurons, are known to regulate the formation and removal of synapses. There are other important cell types, as well.
What is the news for PMS? We learn that two PMS genes are core genes of the dysregulated neuron networks. I have circled these genes in RED. There are about 20,000 genes in the human genome. The paper identifies the top 140 dysregulated genes. Obviously, they are quite important for psychiatric disorders. The two PMS genes are MAPK8IP2 and SULT4A1. Not surprisingly, MAPK8IP2 and SULT4A1 have already been identified as two of the 18 most important genes of PMS (see Which PMS genes are most important?).
Which individuals with PMS are missing these genes? Nearly all (over 95%) of people with PMS are missing MAPK8IP2. About 30% of people with PMS are missing both MAPK8IP2 and SULT4A1. If your child has a typical (terminal) deletion, you can look up which important PMS genes are missing in this blog: Which PMS genes are most important?
At this point, it seems pretty likely that deletions of 22q13.3 do more than raise the risk of autism. Deletions can directly impact MAPK8IP2 and SULT4A1, two core genes dysregulated in autism, schizophrenia and other neuropsychiatric disorders. Perhaps the good news is that people who study autism and schizophrenia have a new impetus to study MAPK8IP2 and SULT4A1. It is up to PMS parents to lobby, cajole and otherwise let everyone know that studying these genes is very important to us.
Two very recent studies of Phelan McDermid syndrome (PMS) drew exactly the same conclusion: We need to recruit and study more PMS patients with interstitial deletions if we are going to understand the syndrome (see references 1 and 2, below). This blog explains why that is a critical need. In some ways, this blog is an update to an earlier blog (Why don’t we have better drugs for 22q13 deletion syndrome?).
PMS can be broken down into a few obvious classes. The original disorder, 22q13.3 deletion syndrome, has terminal deletions and interstitial deletions. Later, SHANK3 variants (often called “mutations”) were added. As I have discussed before (Gene deletion versus mutation: sometimes missing a gene is better), mutations are a mixed bag. Some mutations produce symptoms like 22q13.3 deletion syndrome, but other mutations produce other disorders (like ASD or Aspergers), or no disorder at all.
PMS research started out with SHANK3, but somehow it got stuck there. Being stuck has led to some serious deficiencies in our understanding of PMS. First, very little is being done for the future of children with interstitial deletions. Their SHANK3 gene is intact, so SHANK3 research does them no good. Second, drug studies that use PMS patients to study SHANK3 are likely to fail without accounting for the important genes in each PMS patient. This was discussed in the recent paper on PMS genes (reference 2). PMS patients have such a mix of deleted genes that the benefits of a drug for SHANK3 loss might not be detectable. Third, certain serious problems seen in PMS are unlikely a result of SHANK3. These issues, like poor thermoregulation (body temperature control), lymphedema, cerebellar malformation, mitochondrial problems, and certain developmental problems, impact a large proportion of children with PMS. Every year children and adults with PMS die. We need to know which genes are associated with lethality. These issues will remain serious problems for people with PMS as long as SHANK3 remains the narrow focus of PMS research. Even our understanding of SHANK3, itself, is incomplete without a much better understanding of the other important genes of PMS.
The best way to understand the many genes of PMS is to study people with interstitial deletions. They are the only PMS patients where we can safely say that SHANK3 deletion does not play a role. My last two blogs show that we actually know a lot about PMS genes that are most likely to cause problems. However, we need to know much more about how each of these genes affect people. That requires people with different size interstitial deletions.
There was one research study of people with interstitial deletions published in 2014 (Disciglio et al.). It covered 12 patients. Since that paper, there has been only one additional (single) case study of an interstitial deletion. By comparison, PubMed shows 164 papers with SHANK3 in the title. Most PMS families are probably not aware that the current major studies of PMS specifically exclude interstitial patients: Natural History of Phelan McDermid Syndrome and the Electrophysiological Biomarkers of Phelan-McDermid Syndrome. Some of the sites in these multisite studies have not excluded participants with interstitial deletions, recognizing the scientific importance of these cases. Scientifically, excluding interstitial deletion patients makes no sense. We should be seeking them out, recruiting them. As a parent, excluding interstitial deletions seems unfair to both those families, and to the rest of us. We need to get unstuck. We need the best science possible to help our children.
If you ask someone familiar with 22q13 deletion syndrome (Phelan McDermid Syndrome, PMS), “which genes are most important?”, they will start with SHANK3. Even though some people with PMS have no problem with SHANK3, it is still the most important gene for two reasons. First, deletions of SHANK3 and many uncommon variants of SHANK3 can have a strong negative impact on individuals. Second, a large percentage of the PMS population are missing SHANK3. Thus, SHANK3 meets the criteria of 1) potentially large impact on an individual and 2) a large percentage of the population is missing SHANK3. This blog is a closer look at all the genes that meet these two criteria.
In a recent study, a group of researchers looked at genes that are highly likely to contribute to PMS and are missing in most people with PMS (Identification of 22q13 genes most likely to contribute to Phelan McDermid syndrome [full disclosure: this blog was written by an author on the paper]). That is, genes that appear to meet the two criteria listed above. What makes this study important is that it does not differentiate between genes that have been carefully studied and genes that have never been studied. From my perspective as a parent of a child with PMS, we parents are not interested in gene popularity contests. We are interested in learning what is making our children sick.
I read that SHANK3 was the only important gene
Until recently, nearly all PMS research was focused on one well-known gene, SHANK3. Some scientists have promoted the false narrative that only the SHANK3 matters. But, PMS is a polygenetic disorder. That is, many genes are involved. For individuals who have a pathogenic variant of SHANK3 and no other genetic issues (PMS Type 3), it is true that only SHANK3 matters. But, the vast majority of people identified with PMS have chromosomal deletions (PMS Type 1). Some 97% of people identified with PMS are missing up to 108 genes! (See Understanding deletion size and How do I know which genes are missing?). Consider, as well, that some people have PMS even with intact SHANK3 (PMS Type 2). SHANK3 is clearly not the only important gene. One major goal of PMS research it to identify which genes are most important.
How can we find out which genes are most important?
There are 108 PMS genes and only 44 have been well studied. If there was no independent way to identify the important genes, we would be in serious trouble. Fortunately, there is a way to predict the importance of a gene without knowing what the gene does. A large group of scientist compiled a database of over 140,000 genomes including those of “normal” individuals (Genome Aggregation Database). Normal in this case means no developmental or neurological disorder. This huge database lets you predict which genes can cause trouble. Let me explain.
There is a trick to finding a likely troublemaker gene. Pick a gene. Look at all the copies of that gene in the database. (Each human has 2 copies of each gene.) The same genes in different people are not always the same. Different people can have slightly different versions, called “variants”. It is possible to estimate (with biology and statistics) how many different variants you would expect to occur by chance in a population. For example, for a given size gene you might expect to discover 40 different variants among 60,000 people. What if you find only 5 variants? Something’s fishy if you find only 5. The best explanation is — here is the trick — that the other 35 possible variants of that gene cause serious problems. For one reason or another, those 35 variants remove a person from the category of “normal individuals”. There is a name for this: natural selection. It is a principle discovered by Darwin and it works for genes.
Those missing 35 variants are pathogenic. The variation causes a loss of normal function, and the body cannot operate normally without the gene. Genes that are highly sensitive to variation are called “loss-of-function (LoF) intolerant”. Genes that are LoF intolerant in this way are the ones most likely to cause major health problems. Only a minority of genes are LoF intolerant. Most genes tolerate all kinds of variation. Either they are less important genes, or the body has found ways to work around variations.
Wow! Which genes are most important?
So, which PMS genes are very LoF intolerant? That is an easy question to answer. You can go to the gnomAD browser and look up any gene. Look for the row with pLoF and get the “pLI” value. A value between 0.9 and 1.0 is a bad news gene. SHANK3 is 1.0 — no surprise there, but what about other genes? Let me save you some time. Below is a list of PMS genes that have a pLI value above 0.9.
Genes in this list are in the order of their position on the chromosome. The ones at the top of the list are more frequently lost in the population. If your child has a terminal deletion, look at all the genes with a Kb value smaller than your child’s deletion size. Those are the genes that most likely contribute to his/her disorder.
The first thing to notice is that what started out as 108 genes is now reduced to 18 genes. There are a few other genes with pLI below 0.9, but not far below 0.9. These may also be important. Regardless, the number of PMS genes has gone from intractable (108) to something much more manageable (18 or maybe 20). If your child has an average size deletion (around 4,500 kb), then there are 10 relevant genes. Note that some children are missing only SHANK3. We will not forget them. But, in the meantime we need a major push to fully understand the other important genes of PMS.
In future blogs I will discuss what some of these genes do and how they might impact your child.
If we want to find treatments for Phelan McDermid syndrome (PMS), first we need to figure out what is PMS. That was spelled out in my blog: Why don’t we have better drugs for 22q13 deletion syndrome? My next blog addressed how to organize all the different genetic deletions and mutations so that we can define PMS (Defining Phelan McDermid syndrome). Today’s blog addresses ways we can define different types of PMS. If we don’t define different types, we are wasting our time experimenting with treatments. For instance, some PMS kids talk fluently, some talk in short sentences, some can only say single words and many, like David, do not talk at all. These and many other difference warrant different groups of kids when we test treatments.
Just as there is huge variation in abilities and behavioral characteristics, our kids have very diverse genetics. Recent studies of rodents show that not all with Shank3 mutations are alike. In fact, drugs may work very differently on different Shank3 mutations. Anyone who has kept up with my blogs knows that deletions of different genes are likely to have very different effects on our children. These difference are very important.
Useful drug testing is stuck right now until we develop a way to categorize people with PMS based on both phenotypic characteristics (symptoms and manifestations) and genotypes (deletions versus mutations and which genes are affected).
I have heard scientists who study Shank3 mice talk about “splitting” and “lumping”. Splitting is breaking groups into subgroups. Lumping is putting everyone/everything together into a single group. Lumping has not worked and the growing consensus is that lumping will never work in our population. Splitting based on just one characteristic (e.g., deletion size) probably won’t work, either. We need a more refined approach. What we need is “clustering”. Clustering is what mathematicians and scientists do when categorizing requires using many different characteristics at once.
Here is an example. Let’s say you want to buy a car. You might look at various cars and think about both price and gas milage. You could make a graph something like this:
Similar types of cars have similar prices versus gas milage tradeoffs. Race cars are more expensive, but get poor gas milage. Clustering is when you identify meaningful subgroups on a graph because the individual points are close together. Each group is a cluster. Even if not every car fits neatly into a cluster, you still have an organizational scheme that can be very helpful.
PMS needs meaningful groups. Clustering can get complicated when there are more and more features that divide up the population. However, computer programs can take care of the complexities. What we need first is to identify which characteristics are important for grouping. As a practical matter, researchers go back and forth. They consider characteristics, run a program that automatically clusters the data based on those characteristics, and then look to see if the clusters make sense. That is what we need to do.
When we took David to the PMS Foundation Family Conferences in 2008 and 2010, we met a handful of kids that were remarkably like David (see photo of David, above). What was it about those kids? As I recall, they walked the same way, loved watching music videos, asked for help the same way, were nonverbal and all have relatively larger deletions. Are those meaningful characteristics? Will they help us divide PMS into different groups for meaningful drug studies? We need to find out.
One nice thing about writing a blog is getting feedback. While I embrace and benefit from all kinds of feedback, I admit being partial to positive feedback. I get some nice comments from friend on my Facebook page. I love the encouraging comments that get posted here. There is another kind of feedback that is neither positive nor negative, but very informative. It is the kind of feedback that we should all pay attention to. Visitors vote with their mouse (or touchscreen). Every time someone clicks on a blog link, WordPress adds one to a blog page counter. Now that it is 2017, let’s see what the numbers recorded in 2016 tell us.
The most requested page is at the bottom of the graph: How do I know which genes are missing? That is the number one question on parents’ minds. It makes sense. If your car breaks down, you want to know what caused it, even if you don’t know much about cars. At the very least, you have some idea of what needs fixing.
Three more questions are virtually tied for 2nd place. Have you ever met a child like mine? and How to fix SHANK3 are discussions of SHANK3 and its relationship to the other genes of 22q13 deletion syndrome. Together, with the most requested blog page, over one-third (35%) of mouse clicks on this blog are from people who want to understand how all the genes of 22q13 deletion syndrome operate together to produce the disorder. The other blog page that is tied for 2nd place addresses the same topic from the opposite direction: How can the same deletion have such different consequences?
Taken together, nearly half (46%) of the information people want from this blog is to understand what genes are missing and why those genes matter.
Most visitors in 2016 already knew about 22q13 deletion syndrome. Only about 4% of all clicks went to 22q13 deletion syndrome – an introduction. Fewer clicks went to learning about the author. (I can live with that!) But, I think there are a other links that deserve attention. Here is a suggestion for the new year:
Gene deletion versus mutation: sometimes missing a gene is better. SHANK3 is not the only gene of 22q13 that can have serious consequences when mutated (modified, but not lost altogether). Much is said about SHANK3 mutations, but 98% of people with 22q13 deletion syndrome are missing SHANK3 altogether. Understanding the difference may be crucial to finding cures.
I have dedicated the past few months to formal writing about 22q13 genes aimed at the scientific community. That work has taken me away from this blog, but, hopefully, taken us all closer to effective treatments for our children. That work is done for the moment and I hope to get back to this blog on a more regular basis.
That’s my dad. No, not the dapper gentleman standing in the back. He is the diapered baby sitting in front. One interesting thing about this circa 1924 photograph, I can tell you unequivocally, is that both males in the photograph are carriers of 22q13 deletion syndrome. I can say this even though neither man was ever genetically tested and neither ever had children with a known diagnosis of 22q13 deletion syndrome. How can I be so sure? I spent several years contacting relative and having them tested. It was an interesting and often challenging undertaking. My motivation was to warn family members and at least prepare them for the possibility of having to raise a son or daughter with 22q13 deletion syndrome. Along the way I received “thank you” from some family members and angry words from others. Some family members simply did not want to know. It is a story for some future blog, I suppose.
From this family research I was able to deduce that Joe, my dapper grandfather, was a carrier of 22q13 deletion syndrome with a “balanced translocation” of genetic material between chromosome 22 and chromosome 19. Some of my grandfather’s siblings were carriers and at least two of Joe’s sons were carriers. My dad was one of the carriers. He had four sons, including me, and I am a carrier. My dad did not have children with 22q13 deletion syndrome, but I had two.
The power of genetic principles
I know my dad was the carrier (not my mom) because a few of dad’s relatives are carriers. (See the line chart on my page “Who is arm 22q13?“.) I was able show that my grandfather was a carrier using similar family evidence. Genetics and inheritance follow certain rules and those rules can be used to peer into the past. Genetics and evolution are two different aspects of the same rules, and understanding them can be very powerful tools for understanding where we come from and where we might be going.
Somewhere between 15% and 24% of all children with terminal deletions inherit that deletion from a carrier parent. If your family has carriers, nature has provided a curious way to remove carriers from future generations: have small families. This graph shows why.
(right click on graph to enlarge in a new window)
The main graph has three colored lines. (Ignore the small “inset” graph with bars; it provides details some researchers might want to see.) The green line on the main graph represents what happens when people in the extended family have relatively large families (4.4 children, on average). The black line shows the same process when the average family size is less (3.6 children per family). The red line shows the impact of small families (1.5 children per family, on average). What impact are we talking about? The beginning of the graph starts today. The end of the graph shows what happens after 10 to 50 generations from today. Since most people assume 25 years for each generation to pass, the first 10 generations will take 250 years. Here is the point. If people have only small families, we can expect carriers to disappear (reach 0.0 on the scale) from the population in fewer than 10 generations. However, if people choose to have large families (green line), carriers are unlikely to ever disappear (green line never reaches zero).
Let me be clear. I am not advocating for any specific choice. This is not about ethics. In a sense, these are God’s rules. They are inferred from the statistics of inheritance in the same way quantum tunneling is inferred from the statistics of nuclear emission. I worked with a member of my family to generate this graph using a mathematical simulation. I wanted to know how long 22q13 deletion syndrome has been in our family. The answer comes from the green line. Historically, my European ancestors had large families. My great-grandfather had six children. His children had an average of 4.8 children each. These numbers suggest that the translocation could have existed in our family for tens of generations.
In my prior posting (“Understanding deletion size“) I promised to discuss a brain gene that is missing in 100% of the cases of terminal deletions. I realize that explaining its importance will first require explaining a bit about evolution. So, the rest of this blog will set the stage for judging the importance of a brain gene.
~~~~~ INTERMISSION ~~~~~ There is a lot of material here, so you are welcome to take a break before reading the second part.
Evolution: There ain’t no missing link
Earlier I noted that genetics and evolution are closely related. Describing evolution is simple in the same way that describing police work is simple. The task seems like it should be easy to explain, but the devil is in the details. Many of the principles are not obvious at first, and both requires a lot of study.
Consider this make-believe story. A farmer has two children. The son grows up to be a christian missionary in Africa and the daughter becomes an international arms dealer. Their divergent lives lead to divergent branches of the family. Years after dad passes away, two great-grandchildren meet. One lives in a hut, is very religious and dresses modestly. The other shows up on a yacht. They are very different, but connected through a common ancestor (the farmer). Evolution works the same way. The Chimpanzee is our closest living relative species. However, there was never a species halfway between Chimpanzee and Homo sapiens. We share a common ancestor. Some primate, extinct now, had members that experienced very different genetic and environmental events and each evolved into a different species. These two offshoot species each underwent their own evolutionary history.
Primates (e.g., monkeys, apes, chimpanzees, humans) are special for a lot of reasons, but most notably for the development of higher brain function through the evolution of a new type of prefrontal cortex. The new cortical areas help manage uncertainty, understand complexity and better imagine the future (Wise, 2008). Rodents do not have an equivalent to the granular prefrontal cortex of primates. Importantly, this area has undergone its greatest expansion in humans. Two genetic features drive this type of dramatic specialization of brain function in humans: changes in the genes (either new ones or altered ones) and changes in when, where and how the genes are expressed.
Paralogs and gene expression
Here is a hypothetical example. Let’s say a very early microorganism has a gene that we will call gene L. Gene L is required for movement through its water environment. Gene L is needed for “swimming”. It is so important that the organism cannot survive any mutation of the gene. However, one day there is a genetic error during cell division and an offspring ends up with 2 copies of gene L (duplication event). The new copy of gene L is somewhat “liberated”. It can mutate and change without interfering with swimming, since the old copy of gene L is still available to do its job. We name the two genes L1 and L2. They are “paralogs” of the original gene L. In our hypothetical case, gene L1 allows the organism to swim, and L2 is “free” to mutate and change. One thousand years later, an L2 mutation event allows the organism to detect light in the environment. The evolution of vision has just begun! Thus, “duplication events” are crucial to evolution. They copy important genes so that one copy can continue its original job and the other can do something new. Sometimes, the two paralogs are very similar to each other, but are used differently in some crucial way. L1 and L2 don’t have to be very different as long as having two different versions opens the door to new evolutionary opportunities.
The gene I will discuss next time is a paralog that only exists in ourselves and our very closest primate relatives. That is, you and I carry a pair of genes that were duplicated and then evolved for specialized use in only the largest and most developed brains. Moreover, humans have the most specialized use of the gene, and its specialization takes place in our brain. From an evolutionary point of view, this is a very special gene. This gene is missing from every child with a terminal deletion, 98% of all known cases of 22q13 deletion syndrome. What critical functional role does it play in the human brain and how does that impact our children?