April 30, 2016 at 12:40AM

Today I Learned: I've been reading Ron Milo and Rob Phillips' "Cell Biology by the Numbers", which has given me more than three TIL-worthy facts in the last twenty minutes... so today y'all get four facts! The book can be read for free here, I highly recommend picking some random chapters and reading through them if you're into cell biology stuff: http://ift.tt/1AvhcfE 1) The plasma membrane (the lipid bilayer that defines the outside of a cell) actually only contains less than 10% of the total membrane in a typical eukaryotic (i.e., human) cell. The rest is mostly bound up in mitochondria and the endoplasmid reticulum, though it varies quite a bit by cell type. 2) Mitochondria are usually pictured in textbooks as little pill-shaped organelles, roughly the size and shape of a bacteria. Which makes sense, given that mitochondria are descended from bacterial endosymbionts. However, if you look at mitochondria under a microscope, you'll see that they actually look more like a continuous net, like the outside of a morel mushroom (some example 3D reconstructions here: http://ift.tt/1Un6cPY). The reason mitochondria look like little pill-shaped things in all the EM pictures in textbooks is that those EM pictures are all 2D slices through that net. Well, today I learned that mitochondria sometimes *do* take on little separate pill-like shapes. Specifically, when yeast are grown in ethanol, their mitochondria split up and form little balls. I'm a little suspicious about this, though. Ethanol is a pretty stressful condition for yeast, and the blebbing of mitochondria reminds me (at least superficially) of what happens to cells when they apoptose. Perhaps that mitochondrial morphology is just a precursor to apoptosis? 3) Chloroplasts in a plant cell will often move away from light to avoid photodamage! How ironic, given how much effort plants put into moving towards light. 4) Speaking of chloroplasts (and mitochondria, for that matter), one of the most fascinating events in the history of life is the acquisition of cyanobacteria by some ancient Eukaryote, making the first chloroplast and the first photosynthetic Eukaryote (similar to what happened with the ancestor of all Eukaryotes when it first acquired a mitochondrion). One thing that's been seen repeatedly during endosymbiotic events like that is that the symbiont's genome slowly migrates to the nucleus, until the symbiont is left with a tiny little core genome of things that can't move for one reason or another. Here's a question -- how long should we expect that process to take? Some enterprising scientists performed an experiment to try to get at least some of the way towards an answer to that question. They engineered chloroplasts to hold a reporter gene (don't ask me how the heck they did that), then introduced those chloroplasts into hundreds of thousands of pollen grains (again, how?!?!), mated those pollens with plant eggs, and checked the resulting embryos for nuclear versions of the reporter. It turns out that the reporter was moved to the nucleus in about 1:10,000 mating events. That seems pretty darned fast! Of course, other events would also have to happen to make those genes *functional*, but it gives some hint about how long it should take to move a chloroplast's genome to the nucleus.

April 29, 2016 at 12:27AM

Tody I Learned: 1) ...a little bit about how praying mantises eat. Like a lion, it seems they like to go for the throat, ripping off their prey's head before munching down the body. Unlike a lion, they're surprisingly not that good at actually *killing* their prey. It can take quite a while for the mantis to chew all the way through. They don't seem particularly concerned about holding a creature a significant fraction of their own size still struggling to get away. They're really scarily effective at holding insects. 2) ...how to cut RNA with Cas9! It turns out Cas9 is perfectly happy cutting RNA as long as you provide a little DNA oligo containing the PAM that binds to the PAM site on the RNA. In other words, Cas9 only requires DNA at the PAM sequence, and only on one strand -- it doesn't particularly care what kind of nucleic acid it binds to and cuts past that (though to be fair, cleavage of RNA isn't nearly as efficient as cleavage of dsDNA (though it should *also* be noted that the Cas9 does *bind* quite well to RNAs -- it just doesn't cleave them quite as well once attached, it seems)). More here (Nature, sorry abuot the paywall): http://ift.tt/1rDQDZr 3) ...a circuit architecture for exact tracking of a target (up to a factor). Say you want to detect molecule A, and you want a readout that follows the concentration of A -- for instance, you might want to produce GFP proportional to the amount of some hormone in the cell. You can do that with species B and C and the following regulatory relationships: A (linearly) turns on production of B B (linearly) turns on production of C C very strongly competitively inhibits B, or otherwise stops it from activating A And that's it! For certain parameter choices, at least. It's pretty robust, but not perfectly so. Thanks to Niles Pearce on this one!

April 27, 2016 at 11:28PM

Today I Learned: 1) ...how cells maintain nonzero finite steady-state populations! Well, at least how T-cells do it, at least according to a simplified model. For all of y'all who guess they were secreting some kind of inhibiting factor into their environment... close! Feedback *is* implemented by excreting something, in this case IL-2 (interleukin-2, a protein). Critically, though, IL-2 doesn't just inhibit cell growth -- it also activates cell growth. More specifically, IL-2 triggers cell death in a linear fashion (cell death ~ IL-2 * k, k is some constant) and enhances growth cooperatively, which is a fancy way of saying that the growth rate of cells against concentration of IL-2 looks something like this: http://ift.tt/1rjuv5X. This combination of feedbacks gives the population two steady states, one at no cells and one at some finite amount of cells*. The fact that IL-2 mediates *both* cell death and cell growth is critical! If it just modified one or the other, or if it modified both in a linear fashion, you would not get a non-zero steady state (you can try something similar to what I describe below to show this). Interestingly, there's reason to expect cells to only use one growth factor instead of two. You *can* get the same behavior with two growth factors, but if you use two growth factors, the system is more sensitive to parameters like production rate of the growth factors. * To see why, draw a plot of cell death vs IL-2 concentration and cell growth vs IL-2 concentration on the same plot. Remember that cell death is linearly dependent on IL-2 concentration, while cell growth is sigmoidal with IL-2 concentration. Now pick a "current" cell concentration, which should be proxied well by IL-2 concentration. This is a location on the x-axis. If cell death is higher than cell growth, then the cells are net dying out, and the population moves left, towards zero. If cell growth is higher than cell death, then the cells are net growing, and the population moves right, towards cancer. Anywhere that cell growth and cell death are equal, you have a steady state population that doesn't move. You should find three steady states -- two that populations will move towards, and one that populations move away from. 2) It takes about 25 minutes for a relatively normal-sized ice cube to melt on a plate, outside, in shade, on a relatively warm day. That's a lot longer than I expected! Thanks to Dawna Bagherian for donating an ice cube! 3) The promoter on the ColE1 origin of replication that produces RNAII (which is what primes the plasmid for replication) can be replaced by other, standard promoters, and the plasmid replicates just fine.

April 27, 2016 at 02:28AM

Tody I Learned: 1) By far the majority of magnesium ions in a bacterial cell (and, I think, most eukaryotic cells, too) are bound to something at any given time. Only something like 10% of a cell's magnesium ions are actually floating around in solution. 2) In Arch Linux (or any other systemd-based Linux), you can see what services/units are running by simply running "systemctl". 3) ...a new debugging rule to go in my Linux-user debugging toolkit -- if a secondary package manager (in my case, pacaur, which is like pacman (which is like apt-get or homebrew but for Arch Linux) but for unofficial repositories as well as official ones) isn't downloading anything, and your database list is synchronized, then try reinstalling the package manager. You may have to do a manual install, since the package manager isn't working....

April 26, 2016 at 02:05AM

Tody I Learned: 1) ...that it's possible to debug the Windows kernel, live. This is a process known as "kernel debugging", appropriately enough, and it's done by using one computer as a debugger and one as a test machine. There's some built-in Windows software that makes it relatively easy to set up and run. Pretty cool, Microsoft! Pretty cool! 2) There's an inherent instability in maintaining cell populations that pops up above and beyond what you see in, say protein populations in a cell. See, a cell can keep a protein at a constant (steady state) concentration very easily, by producing at a constant rate and letting the protein degrade at a constant rate *per protein molecule*. Because the total amount of protein degradation increases when there's more *total* protein in the cell, degradation becomes stronger relative to production as the concentration of protein increases. As the cell produces more protein, degradation and production eventually balance out, and the cell reaches a steady-state protein concentration. This doesn't work for cell populations (i.e., the population of white blood cells in the blood), at least not naively. The problem is the way you get "constant production" of a cell population. In the protein case, absolute production is fixed beacuse there's exactly one genome (well, probably two copies of any particular chromosome, but you get the idea) and it produces at some fixed rate. For that to work with cells, they would have to come from a fixed population of progenitor ("stem") cells... but then you have to have some way to maintain *that* population at a fixed size, so that solution really begs the question of how you maintain a constant population. You could also just have each cell in the population divide at a fixed rate, but unfortunately that gives you growth that's proportional to the size of your population. If that growth is *just perfect*, you can get perfect balance and have a constant population, but that's an unstable situation. If the growth rate is just a little too low, then the cells will die off too quickly and the population will crash to zero. If the growth rate is just a little too high, then the population explodes and probably gives you a tumor. How does the body handle this dynamic instability? Tune in on Wednesday to find out. In the meantime, care to take a guess? 3) ...about "The Next Rembrandt". The Next Rembrandt (which I will abbreviate TNR) is a rather complex piece of software deisgned to produce novel paintings in the style of Rembrandt. By which I don't just mean "kind of like Rembrandt's paintings" -- TNR was built to mimic Rembrandt in exquisite detail, from his overall composition to the size and shape of his figures' eyes to the technique and order of his brush strokes. TNR's debut painting was created by instructing it to paint a white man with certain facial features, in a black outfit, facing to the right. TNR designed a painting, then 3D-printed it, presumably out of paint, to give it the full texture of a Rembrandt painting. Go Google Image search "the next rembrandt" to see the result.

April 23, 2016 at 02:28AM

Today I Learned: 1) Apparently much of Chicago was built on swampland and straight-up water. This was accomplished largely by importing vast swaths of nearby dune-sand and filling in the land until it could support buildings. Also, the swampy ground of Chicago forced architects to come up with more sturdy building methods like steel framing, which is how we got skyscrapers. Go building-in-swamps! Cred to Mengsha Gong!!!!! 2) The poodle was originally bred for helping waterfowel hunters collect their catches. The traditional poodle haircut is descended from a style designed to keep the poor dog's joints when wet. Also, the word "poodle" probably comes from a german name "Pudelhound", which translates quite directly to "splashing-about-dog". ALSO, there are some ridiculous old drawings/paintings/engravings of old poodles on the Wiki poodle page. 3) Old thermal cyclers are ridiculous. For one thing, their interfaces suck, but I knew that already. Also, they tend not to have terribly many spots for tubes, and many don't have heated lids (which normally keep water from condensing on top of the sample tube -- without it, you have to cover your sample with mineral oil to prevent evaporation). But I kind of knew those problems, too. What I was *not* expecting was a thermal cycler designed for 0.6 mL tubes instead of the usual 0.2 mL tubes (formally known as "PCR tubes", as they're SPECIFICALLY DESIGNED FOR PCR IN THERMAL CYCLERS). Also, you can get a thermal cycler for less than $150 on ebay if you don't care how much space it takes and you're willing to be limited to 24 samples at a time. For a nice (non-gradient) thermal cycler, you're looking at between $500 and $1,200, and for anything with all the bells and whistles, you're looking at dropping more than $1,000. a

April 23, 2016 at 01:18AM

Tody I Learned: 1) Apparently much of Chicago was built on swampland and straight-up water. This was accomplished largely by importing vast swaths of nearby dune-sand and filling in the land until it could support buildings. Also, the swampy ground of Chicago forced architects to come up with more sturdy building methods like steel framing, which is how we got skyscrapers. Go building-in-swamps! Cred to Mengsha Gong. 2) The poodle was originally bred for helping waterfowel hunters collect their catches. The traditional poodle haircut is descended from a style designed to keep the poor dog's joints when wet. Also, the word "poodle" probably comes from a german name "Pudelhound", which translates quite directly to "splashing-about-dog". ALSO, there are some ridiculous old drawings/paintings/engravings of old poodles on the Wiki poodle page. 3) Old thermal cyclers are ridiculous. For one thing, their interfaces suck, but I knew that already. Also, they tend not to have terribly many spots for tubes, and many don't have heated lids (which normally keep water from condensing on top of the sample tube -- without it, you have to cover your sample with mineral oil to prevent evaporation). But I kind of knew those problems, too. What I was *not* expecting was a thermal cycler designed for 0.6 mL tubes instead of the usual 0.2 mL tubes (formally known as "PCR tubes", as they're SPECIFICALLY DESIGNED FOR PCR IN THERMAL CYCLERS). Also, you can get a thermal cycler for less than $150 on ebay if you don't care how much space it takes and you're willing to be limited to 24 samples at a time. For a nice (non-gradient) thermal cycler, you're looking at between $500 and $1,200, and for anything with all the bells and whistles, you're looking at dropping more than $1,000.