I Grew Real Spider Silk Using Yeast

this video was made possible by brilliant it might seem strange but this little tube of yeast is probably the most valuable thing i own just to make it took hundreds upon hundreds of hours of design time lab work and countless failures and not only that it cost thousands of dollars in gene synthesis equipment reagents and so much more but after more than a decade of dreaming about this moment and frankly years of work i've finally done it the yeast in this tube produce real spider silk okay let's rewind a little bit long time viewers will know that this isn't the first time i've talked about this project in fact there are at least three other videos about some of the early work and an overview of the project on the channel already but so much has changed since then when i first started this project the channel had just barely passed a hundred thousand subscribers but fast forward to now and it's gained more than half a million more and even though i had several years of biowork under my belt at the start i've grown so much since then and have so many more completed projects under my belt now which ultimately led to the eventual success we'll be talking about in this video the original goal of the project was straightforward use yeast to produce spider silk hence spider beer even though you almost definitely wouldn't want to drink the result of this fermentation as these things always go the original road map to success was nice and simple get some spiders isolate their dna isolate the silk gene stick that into some yeast dna and then use that to modify some yeast and we're done but things didn't work out exactly like that like at all the spider i chose was the southern black widow for a few reasons first its silk gene wasn't protected by any patents so i could open source this when it's done and make it available for people to play with second its silk is just as strong as the golden orb weavers that are normally used for this sort of thing topping out similar to kevlar in theory southern black widows are also very easy to buy on ebay and i was able to purchase a whole vial of nightmare fuel for a grand total of 15 and thematically let's be honest black widows look cool as hell and make for some awesome merch designs and this is where things went spiraling into the failure inferno or as we like to call it an average tuesday in a bio lab that vial of nightmare fuel preserve the spiders in 70 ethanol this is plenty concentrated if you want a physical specimen to look at but is utterly useless at preserving dna as the dna degrading enzymes in the spiders remains active so after several months of attempting to isolate high quality dna and perform pcr reactions on it we realized that this avenue was completely useless and had to look somewhere else i reached out to some black widow breeders online and bought a few dead dry samples from them but at this point the idea of isolating the gene directly from the spiders was seeming pretty pointless and frankly exhausting so we cut our losses and pivoted to a different angle i'm also leaving out a ton of other small experiments that we tried that led nowhere as well as progressively upgrading most of my equipment and fine-tuning various protocols by running them dozens and dozens of times for anyone who's tuned into the who's gene is in any way live streams you'll know that i've become quite competent at designing dna from scratch but it was this project that really drilled those skills into my head and is the reason i can even do those streams so with the help of my fantastic lab partner vesta who i've been working on this project with we sat down and designed a new version of the silk project from the ground up with the goal being to skip most of that tedious lab work and just have a company print the dna and assemble the final plasmid for us then all we'd need to do is load it into some yeast saving us a ton of time and expensive resources to understand our design let's first take a quick look at the molecular structure of spider silk like most things in the world of biology spider silk is a protein which means it's a long string of amino acids and the exact order and composition of that chain is what gives silk it's fantastic properties in fact just saying spider silk is actually a bit of a misnomer most spiders actually produce anywhere from three to ten different varieties of silk and can mix and match them to smoothly alter the properties of their web as they build it the main type of silk synthetic biologists like myself are usually interested in is dragline silk this is the type of silk used in the main construction of the web for things that need to be tough enough to withstand an insect thrashing around without getting torn up this type of silk comes from the major ampulet gland in the spider's abdomen and depending on the species is usually called masp1 or major ampulet spideroin protein 1. it's quite firm and not overly stretchy but very very strong masp2 another related form of silk comes from the same gland but is more elastic and has a bit more give puriform silk comes from the piriform gland and is the type used to cement the web to various surfaces it's a little bit sticky but importantly it also has to be incredibly strong as it's the only thing stopping the web from coming loose when something hits it and finally flagella form silk or flag silk for short is what's used to make the rungs of the net of a spider web it's the most stretchy by far as it needs to bend a lot or it would just sheer and snap when things hit it all of these different types of silk can be broken down into monomers these are short stretches of amino acids that within a margin of error just repeat over and over again throughout the length of the gene sequence or at the very least they recur in the gene many times with some random spacers between them when biologists make various organisms like bacteria or yeast produce silk artificially they typically take one monomer from one type of silk and have that short piece of dna synthesized then in a lab they basically just copy and paste it end over end until they have a stretch of dna that'll give them a protein of the length that they want i didn't go this route for a few reasons primarily that process is extremely tedious and expensive and requires a lot of reagents and time also the chance of failure compounds every time you do it so you could end up redoing various steps over and over and over for months and each failure costs money potentially lots of it but there is a pretty big reason that that's normally the method used the way we synthesize dna is very sensitive to repeating sequences so normally things like silk can't actually be printed as it confuses the machine so you're stuck doing it by hand the hard way we tend to think of dna like it's computer code but it's not it's a physical molecule taking up physical space dna is normally printed in short 150 letter sections which have to be stuck together into one big sequence with repeating sections like this it's very easy to get pieces mixed up and have things go together in the wrong order so it takes more manual effort to account for this so most companies will just either refuse the order or charge enormous sums to manufacture something complicated so i'm left with a bit of a catch-22 i want to chemically synthesize a big silk protein gene to save myself time and effort but that process is sensitive to repeats and silk is all repeat all the time to get around it i had to do something very weird and in my opinion a little clever i needed to introduce randomness to the gene while trying to maintain the protein structure best i can to do this i first broke down different kinds of silk into their monomers and divided those into subunits masp1 for example has two subunits in its main monomer which can be thought of as the crystalline and elastic subunits the crystalline region forms what are called beta sheets and the elastic region forms alpha helices and normally they come one after the other over and over and over again the crystalline regions are what give silk it's strength and the elastic regions are what makes it stretchy what makes silk strong is that the crystalline regions from various individual proteins can all stick together like lego forming massive strong crystalline regions these come together into one cohesive strand held together by hundreds of thousands of hydrogen bonds but rather than just being a solid crystal those lego-like regions are strung together by the highly elastic regions allowing the protein to bend back on itself and not together into a solid fiber composed of many thousands of individual strands and because each individual fiber is quite long and has many crystalline domains it can connect with dozens of different proteins in multiple spots tying them all together tightly the other subunits that i used are the beta spiral region from flag silk and two regions from piriform silk which i noted as the proline rich and glutamine rich subunits the flag sections should add more elasticity and the piriform regions could make the fibers slightly sticky though it's unclear from the literature if these subunits are what actually makes those proteins sticky instead one of the reasons i chose puriform silk is that fiber for fiber it's almost as strong as major ampulet silk but can take more strain so i can get the randomness i want without compromising its material properties too much by mixing and matching these i should be left with a protein which mostly maintains the strength of masp1 while actually having a little bit more elasticity when we talk about the strength of a material there's two things that we normally measure stress and strain stress is how much load the material can actually take and strain is how far the material will deform before it breaks toughness is the combination of the two masp1 can take more stress that is more load but less strain because it's not very stretchy whereas the other silks can take more strain but less stress that is to say that they'll stretch more before they break but it takes less force to do so it's kind of the difference between a steel cable and a rubber band one can hold a lot of weight while one can stretch a lot silk is like what happens when you mix the two and make stretchy steel and that's not really a figure of speech spider silk can take both more stress and strain than most steels and outperform most common materials like cellulose nylon and other plastics so considering that i'm actually quite comfortable trading a bit of stress tolerance for more strain tolerance and should still end up with a very strong fiber and some of the other features i added that we'll talk about in a bit should actually compensate and raise the stress tolerance above even natural silk if i'm lucky with these five subunits in hand i had to make multiple versions of each so that when i string them together into a gene it still doesn't get too repetitive one of the nice things that help with this is that dna is broken into codons these are three letter codes that tell the ribosomes which amino acid to stick onto the protein next as it's being synthesized and there can be multiple codons for each amino acid for example in the gene if there's a gct gcc or gca all of these get read and an alanine amino acid is added regardless of which one of these is present so after going through all the subunits to figure out which amino acids i'd need for each i made a chart of all the different codons that are possible for each amino acid and i removed any which aren't used in the species of yeast that i plan on using while a codon table like this makes it seem like any codon can be used for any amino acid that's not accurate and not all organisms use the same codons and each tend to prefer specific ones over others so using the alanine example from before some organisms may vastly prefer gct over gcc so it wouldn't make sense to use a codon that an organism lacks the machinery to use properly those preferences are hard coded into the chemistry of the organism so trying to use a codon that the organism lacks the machinery for could ruin the protein and have it eject before it's finished being synthesized with that done i had a table that i could start building with for each subunit i figured out how many of each amino acid i would need and then used a random number generator to produce strings of numbers from 1 to however many codons i had available for each instance of the amino acid present in each subunit i realize that sounds convoluted so let's see one example i'm going to use the glutamine rich subunit because it's nice and short the amino acid sequence is qqssva where q stands for glutamine s is serine v is valene and a is alanine so i need two glutamine codons and i have two to choose from so i had the random number generator spit out 20 pairs of ones and twos so either a 1 1 a 1 2 2 1 or 2 2 randomly then i did the same thing for the other aminos and strung them together in the proper order but to make it clearer when i moved on to the next type of amino acid i'd shift the numbers so for serines i'd use three fours and fives the valence sixes and sevens and alanines eights nines and zeros if this is my newly randomized sequence i would then take that string of numbers and go back in and replace them with the appropriate codons for my list so a 1 1 would become c-a-g-c-a-g 1-2 would become c-a-g-c-a-a and so on this was repeated until i had 20 versions of a dna sequence that would all code for that same qq amino acid sequence once i had repeated this for all five subunits i had a list i could work with the only one i didn't make 20 versions of was the prolene rich region because it was just so repetitive but now the big question is how do i mix and match these do i try and keep the same order as the source silk or do i do it totally randomly or do i try and eyeball it one of the saving graces at this point is that when you look at the various types of silk there's actually a fair bit of variability in how the subunits go together for example there's a lot more elastic regions between the usual crystalline regions in masp2 yet the protein remains quite strong so as long as i have large quantities of the crystalline region with stretchy stuff in between it should be fine so i started generating randomized variants until i found one that looked right there's a story about how the sr-71 was designed where one of the lead designers brought the designs home to his wife and she would say if it looked fast enough that was basically what i did with this i just kept messing with it until i found spacings of things that looked good to my eye and it looked silky enough and honestly it doesn't actually matter that much this is just a starting point if the end fibers have garbage properties i can always generate a new version that more closely resembles the source protein and have that printed maybe using less of the piriform or flag subunits and more of the major ampuled ones but this will give something as a baseline to work with and should have most of the properties i want as i did try really hard to have it resemble the real thing and this costs so much less to do so generating a new version while a bit tedious is fairly easy every time i generated a string of numbers like this i would go back in and replace each one with the next random sequence in my subunit list accordingly so the first time i hit a one i'd use the first sequence from my crystalline list and the next time i'd use the second one and so on and once all the numbers were replaced with dna sequences i checked the resulting dna and protein using some online tools to see how the protein itself looked visually and see if it seemed synthesizable i use a company called gene universal for all my gene synthesis needs because they're fantastic and they have a few different dna backbones in stock to choose from that work in yeast you can think of the dna backbone as sort of a blank cd which can be loaded with whatever file you want which in my case is the new silk gene i ended up picking one called ppic 9k because it had a bunch of features that i wanted first it obviously works in yeast which is what i wanted not just because saying spider beer is fun but because yeast are much better at producing big proteins like this in large quantities and so are better for large scale stuff next it uses g418 to select for the modified yeast g418 is an antibiotic and is cheap easy to use and frankly i already have lots of it so it makes isolating yeast cells which have taken in the new dna after we modify them very easy next it's an integrating plasmin which means that when i put the dna into the yeast it'll work only if the dna integrates itself into the yeast core genome this makes the mod permanent and means that once i've got the finished yeast i can grow them without needing to keep adding g418 to the growth media making it much cheaper to grow them in bulk think of this like updating the yeast operating system rather than installing a self-contained program and finally it contains a special tag which is added to the front of the silk gene that causes any silk produced to be excreted by the yeast into the media they're growing in this means i won't need to worry about collecting the yeast and popping them open to get the silk i can just harvest the growth media to collect it also this reduces the burden on the yeast as without it they would literally fill with silk until they clog up and explode so each individual yeast cell can actually produce more silk protein before they eventually die but before we sent off the finished construct to the company to be printed there was one last thing i wanted to add because screw it project creep for days a field of study i'm fascinated by is biomineralization biomineralization at its simplest is the idea that various biological proteins can pull minerals metals and more out of a solution and solidify them into particles this sounds strange but if you think about it is exactly what happens when bones grow or creatures grow a shell various researchers have narrowed down which proteins and importantly which very short pieces of said proteins are actually responsible for this mechanism for a variety of materials these short protein fragments are called biomineralization peptides and literally hundreds have been identified for a huge variety of minerals metals and other materials i chose four peptides that should each bond to one of four different materials and strung them together with an inert spacer between them this will give them a little bit of wiggle room so as to not damage their individual functions this small chunk of bio-mineralization peptides was added to the very end of the silk gene and then the whole plasmid was sent off to the company to be made the reason i did this was because i wanted to make sure the silk had extra properties for example the first peptide taken from oysters should collect calcium onto the silk if carbonate ions are also present this will turn into calcium carbonate which is basically how shell forms in fact the same peptide also binds to a molecule called cytosine which is one of my all-time favorite materials in theory this should make the silk mineralize into a hard tough material very similar to mother-of-pearl if exposed to the right chemicals but even without mineralizing this should actually make the individual protein filaments stick together at their ends when calcium is present this sort of artificially extends the length of an individual fiber by linking two together one of the reasons natural silk is so strong is because the protein itself is actually really long about 3000 amino acids total my protein is only about a third of that so these mineralization peptides are one way of sort of cheating a longer protein into existence by binding two shorter ones together end to end the other three peptides i added stick to silica silver and graphitic carbon materials like nanotubes and graphene the silica peptide comes from diatoms which are microscopic organisms that actually build a glass shell made of silica around themselves this beautiful footage actually comes from the fantastic james over on journey to the microcosmos and he's got an amazing video about diatoms which i would highly recommend checking out as they are truly fascinating little creatures in the context of the silk though adding this peptide would allow it to be part of the mineralization process so if i had a big glob of silk goo and stuck it into a solution containing silica it should all solidify into a sort of glass or it could allow the silk to bind to things like glass fiber to make interesting composite materials and if there was calcium and carbonate ions in there too i imagine this would sort of form a biological concrete the silver one would allow me to stick silver nanoparticles to the silk fibers which would make them antimicrobial and prevent them from rotting since unlike plastic these are biological materials and should compost easily without this addition and finally the carbon binding peptide would allow you to make fibers with things like graphene or carbon nanotubes mixed in making the fibers way stronger this was the extra feature i mentioned before graphene and nanotubes are some of the strongest materials known so a composite of silk and graphene should be ridiculously strong and there's already a lot of literature that has attempted this with amazing results those ridiculously strong fibers would have some very specific uses i feel and you can mix and match these in theory a graphene silica plate held together by silk and dotted with silver could be a ridiculously tough material while being extremely lightweight and its properties can be varied seamlessly using chemical gradients while growing the plates and growth is the best word for it unlike petroleum products all of these materials could be grown in vats making them very environmentally friendly however this was one of the first dna constructs i ever designed so as i was writing the script for this video i actually found a few errors it seems like i'd copied the peptide sequences incorrectly out of the various papers that i found them in so some amino acids were either wrong or missing for the mineralization peptides so one of the first new orders i'll be putting in is to have this section remade and fixed luckily when we designed the plasmid we added restriction sites between every section of the construct so we could cut out and swap pieces if we found errors because its bio and errors happen all the time so luckily this is a really easy fix and it's such a short piece of dna that it won't cost that much to have it fixed also after reviewing the literature most of these should still work just maybe at a lower efficiency in some cases as most of the errors weren't fatal to the peptide's functions so i'll start by testing the current version while i wait for the fix to be synthesized and if we find out that the mineralization doesn't work for some materials we know why the silk sequence itself though looked fine after submitting the original design all we had to do was wait and four months later a package arrived containing 10 vials of more dna than i could ever really use all i had to do now was take that dna and get it into a specific species of yeast called piscia pastoris but that was actually much easier said than done pishi itself is actually a really interesting species it can eat methanol and the plasmid actually has a methanol trigger built in so we'll only produce silk when given that trigger it can also survive in quite high salinity well past ocean salinity and it's also fantastic at producing huge quantities of proteins and so is actually used industrially to produce things like insulin we're talking gram per liter quantities here which might not sound like a lot but keep in mind that at scale growing thousands of liters of yeast is really quite easy just look at the beer industry vegemite is only a thing because of the hundreds of kilograms of yeast left over from brewing beer however it turns out that pistia are kind of divas and so refuse to take in dna if you so much is look at them the wrong way but what saved the day was actually some of you in my last video about genetic engineering i modified some brewer's yeast to produce beta-carotene and in that video i asked for anyone who works with yeast to write in with their protocols to improve my own and sure enough more than a dozen researchers did and using those improvements i managed to modify the pitia without issue in fact that was actually why i made that video i knew that the hive mind of the internet would have the answers i needed to finally crack this project this video was already pretty long and that video already contained a super detailed protocol for modifying yeast so if you'd like to see how that works i'd recommend watching that video what i will say are that some of the changes i made that fix the protocol and makes it vastly more efficient thing number one be gentle with the yeast in that video for various steps of the protocol i used a vortexer to mix my solutions but apparently this stresses the yeast out a lot and can actually pop them so instead simply gently pipetting up and down to mix things worked way better in the last video for the washing step where you prepare the yeast after growing it overnight instead of using lithium acetate solution to wash the yeast i just use plain distilled water and this actually seemed to work a lot better also unlike the last video ppi c9k has to be linearized using a restriction enzyme before you do the transformation protocol basically this means no rings of dna allowed lines only this is because of the integration functionality i mentioned before it uses something called homologous recombination to insert itself into the genome of the yeast the short version of how this works is we're essentially hijacking the yeast dna repair mechanism to trick it into putting our custom dna into its genome and finally i changed how i plate the result after heat shocking my yeast rather than plating a small amount of the end result and then spreading it around i concentrate all the yeast down into a little pellet and then plated all of them as super concentrated little drops then after a few days spread the result around to get individual colonies for use in other experiments this is a little bit weird but was very very effective okay so i've got some yeast that seem like they're modified because now they grow on antibiotic plates like they're supposed to be now what well first let's do some tests to make sure this is real and that silk is actually being produced first up let's do a side by side growth of unmodified pistia and the silk carrying variety in a liquid broth already the difference is striking pishia are normally a dull pink color but the silk yeast are bone white and when i tip the tube the silk modified ones are goopy stringy i don't really know how to describe it they behave very differently than what i've come to expect from yeast grown in solution like this but just to make sure these visual clues are the real deal i ran some more tests to make sure this wasn't an issue with the antibiotics in my plates i made some fresh plates with fresh antibiotics and then streaked out the silk yeast on them as well as a little area for some unmodified pistia to act as a control sure enough the silk yeast grew but the control unmodified ones didn't just as expected so the yeast must at least be carrying the antibiotic resistance gene then to definitively prove that the yeast actually carried the silk gene i designed a set of pcr primers we've talked about pcr many times on this channel so if you want to learn how it works there's some links to earlier videos below i design primers that stick to part of the backbone and then part way through the silk gene so if i get a product of the right size i know the silk gene must be present i took a sample of four different colonies of modified yeast off of one of the antibiotic plates and then used that to run a pcr reaction on each sample this is called colony pcr and is actually a really standard technique for checking to make sure mods have worked in modified organisms and sure enough when i ran it and then ran the result on a gel i got a perfect size band for every single yeast sample i tested which means that the yeast really do carry the silk gene and that weird behavior is most likely due to silk being produced further when i look through the literature on the topic silk is noted as being dramatically insoluble so all that white goop in the tube that collects at the bottom is most likely silk-like proteins as there's nothing else that makes sense to be there all told this is a ton of evidence that things are working properly one last test was to directly see if the silk is working as intended i grew two tubes of yeast one of the original pistia and one theoretically expressing the silk proteins i collected the goop at the bottom of each tube and resuspended it in two milliliters of plain water then i made a solution of graphene oxide in water normally graphene oxide is stable in solution and with a quick mix you can see that it disperses very easily now because of the graphene binding peptide on the end of the silk gene when i add the silk solution to it it should all clump and crash out whereas with the plain yeast it should leave the graphene in solution i added one milliliter of each yeast sample to their respective graphene solutions and then gave them a mix it was impossible to see on camera but the silk solution had an immediate reaction and it was clear that tiny clumps were forming i left these to sit for a few minutes and when i came back the silk tube had coagulated nicely and after about an hour the coagulated mass of silky graphene had mostly sunk in the tube leaving a water clear graphene free solution above it while the plain yeast looked exactly the same as it did at the start of the test this is a huge result for two reasons first it means the silk must be produced in full because the graphene binding peptide is on the very end and second it means that the secretion function is also working very well i think that's about as clear of a result as i could really ask for so where do we go from here i've ordered more materials to run more tests like a protein gel and lots of materials for scaling now that this seems to be working the focus is on the future and actually doing the fun part of growing silk making fibers testing the bio-mineralization peptides and so much more the first thing i'm going to do after this video goes out is grow a huge batch of the yeast i'm going to start with 400 milliliters and if the results are good i've got a one gallon carboy i've had tucked away specifically for this project i've also got a little bioreactor on the way for continuous silk production one of the first things i'll do before attempting to make fibers from my silk is to test the process out on these silk cocoons that i bought online from the common silkworm the composition of their silk is very similar to that of spider silk so if i can successfully dissolve and extrude fibers from these cocoons it should be very easy to do the same thing with my silk and of course there'll be lots of videos about all of this and from there the sky's the limit spider bricks a silk sweater web shooters you name it once i get a bioreactor online i plan to move on to a wet spinning apparatus and then start messing with the composition i've also ordered a modified version of the plasmid which contains a colorful protein added to the end this should make the fibers purple which will be definitive proof that everything's working properly and would mean that i can make purple silk fibers and if that works there's nothing stopping me from doing that in essentially every color the plasmid has been posted on my github repo for those who want to tinker with it and i'm looking into setting up an online store to sell all the bits of interesting dna i've been designing like this silk plasmid or the deer milk or egg yeast plasmids i'm working on as you saw at the beginning of the video this video was sponsored by brilliant brilliant is this amazing online learning platform which has a ton of really diverse courses created by a diverse collection of experts covering all sorts of different fields in science and mathematics what i really love is that the courses are set up in this series of word puzzles that help teach you to think outside the box to solve the problems while giving you the freedom to make mistakes and learn at your own pace they have a great course on computational biology that i think is an absolute perfect fit if you found the way that i designed the silk protein interesting designing proteins is a really tricky thing to do as you saw and the course helps show some of the intricacies that i couldn't cover here like the mathematics of protein folding and the ways that proteins interact with themselves and each other i also love their scientific thinking course as it teaches you to break down big problems like this into little pieces and i think is very applicable to developing the skills needed to tackle a big project that's as complex as this one all big problems can be broken down into much more digestible pieces and it's this idea of breaking things down into manageable pieces that science is all about and it's all done in a way that puts curiosity first just like my own work the first 200 people to go to brilliant.org slash the thought emporium or use the link in the description will get 20 off the annual subscription speaking of online store i've actually got a whole bunch of merch which you may have noticed throughout the video everything from these awesome widows wine and widows draft posters to these widows brew t-shirts which i absolutely love and even a pipette cry repeat shirt so if you'd like to help support the show and projects like this there's some links below where you can pick one up and as always i need to say a massive thank you to my amazing patrons channel members and supporters on kofi and streamlabs like i said at the beginning this project has been a massive investment of both my time and resources and it's been your amazing support over the years that has allowed me to keep bashing my head against it until it finally worked and that'll continue to be true as i keep working on it and pushing it further and further patrons and members also get access to the new patron-only discord where i post very frequent updates and there's some chances to collaborate on projects with me so thanks again and if you'd like to keep the flow of science videos coming there's some links below if you enjoyed you know what to do hit that like button subscribe and ring the bell to see when i post new videos and be sure to follow me on my other social media pages like instagram and twitter to see these projects long before they end up on videos links to everything can be found in the description below as well as links to the previous videos in the series that's all for now and i'll see you next time

Channel: The Thought Emporium

Views: 2,693,610

Keywords: biology, genetic engineering, science, learning, spider silk, material science, yeast, spider beer, genetics, protein, protein engineering, engineering, chemistry, amino acid, dna, rna, gene synthesis, pcr, tutorial, how to, spider man, black widow, spider, web, spider web, widow, black widow spider, graphene, carbon nanotubes, fibers, thread

Id: 2hf9yN-oBV4

Length: 32min 46sec (1966 seconds)

Published: Tue Sep 22 2020