Hello My name is Randy Schekman I’m at the University of California, Berkeley, in the Department of Molecular and Cell Biology This is the third of three lectures on the subject of how cells manufacture proteins for export In the first lecture, I described the history of membrane cell biology, what we know about biological membranes, how they encapsulate proteins, and how these proteins are transferred between organelles inside of a eukaryotic cell In the second lecture, I discussed how the mechanism of this process was dissected using a combination of genetic and biochemical approaches in baker’s yeast, and how that knowledge has led to the application applications to the biotechnology industry and to an understanding of certain human diseases In my third lecture, I’m going to take a departure and describe a rather newer set of observations in a subject that is I think, has very exciting possibilities, and that is how cells manufacture vesicles that are exported outside of the cell, and how these vesicles capture small RNA molecules These extracellular vesicles have been called exosomes and there is some evidence that the exosomes may mediate the transfer of information between cells, for instance, in the human body or in metazoans, that may influence development or may be subverted in disease states, for instance, in the progression of cancer, metastatic cancer Well, let’s begin with a description of what these small extra… extracellular vesicles look like, what they contain, and some thoughts about how they’re manufactured inside of nucleated human cells Well, here’s a cartoon of a vesicle This is just like any vesicle of the sort that I’ve described before It’s a bilayer around 80 or 90 nanometers in diameter It has, in this instance, some membrane proteins I’ll tell you about a few of them, some of which are integral, that span the membrane several times There are small proteins and other small molecules in the luminal interior of the vesicle But what distinguishes these vesicles from those that are found inside of cells, and that are responsible for secretion, is the presence of small RNAs — you’ll see some examples of this over the next few minutes — microRNAs that are believed to be involved in control of gene expression, or even transfer RNAs, or other unusual RNAs are housed sometimes in high chemical concentration in the interior of these vesicles And these vesicles, as you’ll see in a moment, can be expelled outside of cells, where the vesicle may be targeted to a distal tissue, and taken up by fusion at the cell surface or by internalization by endocytosis Well, one idea about how these things are made inside of nucleated cells is shown in this cartoon This pathway, depicted here, is a pathway that is used when cells internalize receptors and deliver them, eventually, for destruction by proteolysis in an organelle called the lysosome The process often begins when a cell surface receptor is taken up into a membrane called an endosome Often, these cell surface receptors have a chemical tag, a small peptide called ubiquitin And this tag protein marks a prot… a membrane protein for destruction It is captured into a membrane that pinches into the interior of the endosome, an invagination into the interior of the endosome, that results in a small intraluminal vesicle This process can continue for some time to build up an endosomal structure that has many vesicles And this structure was known for decades, and called the multivesicular body In more recent years, it’s known that this multivesicular body targets and delivers these intraluminal vesicles to the lysosome, where the content of these vesicles may be degraded into amino acids and sugars, pumped back into the cytoplasm of the cell, to be reused in biosynthetic processes Now, it’s been realized only for the last 15 or 20 years that sometimes, and by means that are not entirely clear, these multivesicular bodies may actually travel to the cell surface and fuse, where the intraluminal vesicles, now, are expelled to the outside of the cell
as a bolus of vesicles, often called exosomes or extracellular vesicles And as such, these vesicles may then be targeted to distant tissues in the body And the molecules that are contained within them may then be internalized by a target cell to change gene expression or change signaling by that cell Alternatively, this export of vesicles may be another way that the cell has simply of disposing of these… these vesicles, not for some positive function but just to get rid of them Those two alternatives remain viable Now, I became interested in this several years ago, when a wonderful graduate student joined my lab, by the name of Matt Shurtleff, and I’m gonna tell you about his work that we recently published just this year in the journal eLife Now, many investigators who have studied this problem of how exosomes are made and how they acquire their content have relied on a crude method for their isolation, basically taking human cultured cells, often tumor cells, growing them in cell culture, removing the intact cells by low-speed sedimentation, taking the conditioned medium and centrifuging at high speed to form a mixed vesicle collection of membranes And Matt and I felt that this purification scheme was crude, and decided to try to devise a more thorough means of purification of a discrete exosome species produced by a cell line that we were growing in the lab, human embryonic kidney 293 cells, a convenient cell whose genes we can control, and which was a convenient source of material And so I’ll just summarize a three-step procedure that Matt devised for the isolation of exosomes secreted by these cells, based on their content of a membrane protein called CD63, a four-spanning integral membrane protein that is often found in cruder preparations of exosomes that have been published Here’s Matt’s procedure We start with a conditioned medium from HEK293 cells The medium is centrifuge at low speed to remove large organelles and debris Small vesicles are concentrated to form either a pellet or sedimented onto a dense shelf These small vesicles are then suspended in a high concentration of sucrose, 60% sucrose, overlaid with two lower concentrations of 40 and 20%, and then the sample is centrifuged at high speed, where membranes, being buoyant, float up We found that CD63, detected by SDS page and immunoblot, sediments to a position between the shelves of 20 and 40% sucrose Further, Matt found that vesicles at this fraction could be fractionated, enriched, by binding the membranes to large beads onto which an antibody against the CD63 protein has been fixed chemically So that this, now, affords an immunoselective purification of CD63-containing vesicles We know these vesicles also have RNA They can… we can detect chemically… chemical levels of RNA associated with CD63 vesicles And we can show that this fractionation scheme results in approximately five-fold purification over the course of the several steps Now, using this enriched material, Matt then did a thorough deep sequence microRNA sequencing reaction And evaluated some 600 different microRNA species that are found in HEK293 cells or in the exosomes that arise from those cells And a Venn diagram of that distribution is shown in the next slide Most of the RNAs, most of the microRNAs, are found either entirely inside 293 cells, or both in exosomes and in the cells Some 90% of the RNAs are not enriched in the extracellular vesicles On the other hand, about 10% of the RNAs that Matt found are actually enriched in these vesicles, thus they are somehow sorted there during their biosynthesis However, even among this 10%, only four species –three shown here that are chemically abundant, and one less abundant species — are highly enriched
This species, for instance, a microRNA called miR223, a microRNA that has been implicated in stress response and in cholesterol homeostasis in cells… this microRNA we found to be enriched 1000-fold in exosomes isolated from CD63… from HEK293 cells So much so that it’s barely detected in the intact cells And others such as miR144 enriched 2-300-fold Well, that by itself was really very surprising It says that the cell has gone to some considerable effort to sort these few RNAs There must be something special about those RNAs, either some special positive function or some special negative function, that the cell has troubled itself to sort into exosomes And so we decided we wanted to figure out how this happens What is… what does the cell do? What is the mechanism that the cell uses to sort, for example, miR223 or miR144 so abundantly into these small vesicles? Well, let’s have a look, first, at some simple characteristics of the vesicles that we purified And the first is we can look to see, in fact, how highly enriched these two microRNAs are by examining their enrichment during the course of the fractionation that Matt devised And you see on the left a progressive increase in the relative content of miR223 enriched at each step in the fractionation scheme that I described Further, the same is true of the enrichment of miR144, progressively during the fractionation, ultimately obviously contained within vesicles marked by their content of CD63 We could further show that these RNAs that are so highly enriched are inside of the vesicle There’s a simple way of showing that these things are contained inside the vesicle And that is that the RNAs in these vesicles are resistant to degradation by the addition of a potent ribonuclease added to the buffer in which the vesicles are suspended The RNase… presence of RNase barely affects the chemical detection of these two RNAs Whereas, when the membrane surrounding the vesicles is dissolved with a mild detergent, Triton X-100, both RNAs are exquisitely sensitive and quantitatively degraded So, that’s a simple test to show that the RNAs are both enriched in the vesicles and inside the vesicles Now, the question is mechanism How can one figure out how these vesicles acquire the RNA molecules? Well, if you listened to my second lecture, you’ll know that what we like to do in my laboratory is to devise a biochemical, a cell-free reaction that reproduces some significant biosynthetic process in a test tube, where, in this case, membranes and cytosolic proteins are mixed together in an attempt to recapitulate the biosynthetic process It’s a very simple biochemical reaction We start with membranes centrifuged from the cell lysate that are then mixed in buffer with cytosolic proteins obtained from the HEK293 cells, ATP, which is often used to promote biosynthetic reactions, and importantly a chemically synthetic, pure form of mature miR223 RNA The incubation is conducted at 30 degrees for 20 minutes, followed by centrifugation of the membranes to remove most of the unincorporated RNA, followed further by treatment with ribonuclease under conditions where any unincorporated remaining RNA is degraded The membranes are then collected again, ruptured with detergent to release any sequestered RNA, which is then amplified by qRT-PCR, and quantified to detect how much biogenesis has occurred Let’s look at the data that we obtained from this reaction And focus first on the left, which is a simple comparison between a reaction containing microRNA, membranes, and cytosol and ATP, incubated for 20 minutes at 30 degrees, where approximately 7% of the exogenous miR223 appears to be captured
Surprisingly, that reaction was almost completely eliminated if cytosol was not contained in the reaction or if membranes were omitted from the incubation Further, if the reaction was conducted in the presence of detergent from the outset, very little RNA is sequestered And finally, for another control, if the intact sample in the absence of detergent is held at 4 degrees, basically, on ice, very little RNA is segregated So, these really very crude but powerful controls show that this sequestering requires membranes, requires cytosol, requires intact membranes, and requires incubation at a roughly physiologic temperature Now, these biosynthetic reactions very often require ATP, hydrolysable ATP And that was true in this, as well as you’ll see in the experiment on the right If the reaction is conducted without extra ATP, the signal is reduced about twofold If the incubation is conducted in the presence of an enzyme that degrades ATP, likewise, the signal is reduced Or if the incubation is conducted in the presence of an analog of ATP that cannot be hydrolyzed, likewise, reduced Though, in all, cases these controls are not as low as a sample simply held on ice So, in summary, then, this reaction suggested that the biosynthetic event required cytosol, required intact membranes, required ATP, and required incubation at a physiologic temperature Now, the critical question is, can we show that the RNA uptake into vesicles formed in the test tube is sequence-specific? Does it depend upon some sorting event that distinguishes an exosomal RNA, such as miR223, from a purely cytoplasmic microRNA So, this was a simple experiment Matt examined, in a two… in two incubations the uptake of miR223 with the uptake of a purely cytoplasmic microRNA, one found only in the intact 293 cells, an RNA called miR190 And the data for that is shown here Again, now Matt sees, in this case, closer to 8% uptake of miR223 in an incubation at 30 degrees Very little, maybe 1% 1-2% in a control incubation held on ice A cytoplasmic RNA examined in the same reaction is barely taken up into a detergent-sensitive ribonuclease-resistant fraction of membranes Only slightly more than is taken up in an incubation held on ice So, the difference between this data and this data suggested to us that we had reproduced in the test tube an RNA sorting event reflecting something about the nor… the normal pathway that cells use Well, encouraged by this, we then asked the next obvious question for us, which was, if there is some RNA selectivity, then surely some RNA binding protein might be involved in the sorting of miR223 RNA And for this, Matt used the same experiment that I’ve described, only introducing a hook on miR223 that allowed… that would allow him to fish out miR223 after the reaction to see if anything came along into the exosomal vesicle, along with the RNA Here is the result Here is the outline of that experiment It’s the same experiment that you’ve just seen In this case, we used a biotinylated form, where biotin is attached to the 3′ hydroxyl of miR223 to produce a form of the RNA that has a hook that would allow us to fish out any proteins that adhere to the RNA in the course of this cell-free incubation The incubation is conducted, the membranes are centrifuged, they’re treated with ribonuclease, they’re then dissolved in detergent And then the proteins that bind… that bound to the RNA are captured with the biotin-RNA absorbed on beads that contain a protein called streptavidin, that binds very tightly to biotin groups
The proteins that adhere to such… to the beads are then eluted by treating the beads with a high concentration of salt buffer The eluted proteins are then evaluated by mass spec to identify proteins that came along with miR223 Matt found a number of RNA binding proteins, but one stood out because over half of its sequence was detected in peptides that turned up in the mass spec analysis And this particular protein had already been described in the literature to be secreted by cells, apparently in vesicles, exosomes, secreted outside of cells Here’s a paper that reported that discovery These investigators found this protein It’s called the Y-box 1 protein, YBX1 It was secreted by cells They imagined that it may be serving as some kind of an extracellular mitogen, contained within extracellular vesicles And importantly for us, they showed in this paper that the YBX1 protein in the medium from these cells was insensitive to a proteolytic enzyme, trypsin, unless the sample was suspended in detergent, just what you would expect if the YBX1 protein were contained within a detergent-sensitive vesicle So, we obtained a polyclonal antibody against the YBX1 protein and asked two questions The first is, can we show in our purification scheme the YBX1 protein in the vesicles immobilized on CD63 antibody-bound beads? And secondly, can we show that the YBX1 protein is packaged along with miR223 in the course of the biogenesis, cell-free biogenesis reaction, and dependent on the same conditions of incubation that we established for the packaging of miR223? These data are shown in the next slide Let’s focus, first, on the left This is an immunoblot of two samples, the sample of material bound to CD63 antibody beads and a sample that did not bind to these beads We used antibodies against several other previously characterized membrane protein constituents of exosomal vesicles and found, sure enough, that all of them can be detected by SDS-PAGE and immunoblot, associated with vesicles containing CD63 And further, we could show that the new protein that we had discovered in exosomes in our system, YBX1, sure enough, is also contained within these vesicles, CD63 Another membrane protein, called flotillin, previously characterized as a constituent of extracellular vesicles, in this case seems not to be in vesicles that contained CD63, because it was in the unbound fraction This suggests that there may be yet another population of extracellular vesicles, marked by this protein but not by CD63 Now, let’s look at the experiment on the right This is the same experiment that I’ve now shown you in several forms, the cell-free reaction that Matt devised Incubations containing biotinylated miR223 were incubated with membranes and cytosol for 20 minutes at 30 degrees What… the membranes were centrifuged and suspended in buffer with ribonuclease The ribonuclease was then inactivated and the membranes were dissolved with detergent The sample was mixed with streptavidin beads to collect the biotinylated miR223 And then the beads were incubated with sample buffer and run on an SDS gel, followed by an immunoblot And sure enough, there is YBX1 protein, coming along with the microRNA in this reaction Importantly, when the incubation was conducted without cytosol or without membranes for 20 minutes at 30 degrees, the same process, the same processing, revealed no YBX1 protein carried along for the ride Further, if the incubation was conducted with all these components but held on ice, 4 degrees, little if any YBX1 protein is detected Thus, the reaction requires some physiologic process And finally, if the biotinylated RNA is omitted from the incubation, little if any is seen,
obviously, little if any YBX1 protein is carried along for the ride Now, surprisingly, in the course of our analysis, one very important protein involved in microRNA biogenesis, a protein called Argonaute, was not found in our preparations And we were curious why this was so And let me highlight why this was such a surprise by showing you a cartoon that depicts the pathway of biogenesis of microRNAs, established by many investigators around the world A microRNA is made is a precursor, a stem-loop structure, which is trimmed in the nucleus, is exported into the cytoplasm, where it is subject… subjected to action by an enzyme called Dicer, that cleaves the loop structure to produce a double-stranded RNA with a duplex consisting of the mature microRNA strand and its complement The complement is degraded The mature microRNA strand is then captured in a complex by a struck… by a set of components called the RISC complex, including this protein, Argonaute, which binds very tightly to the mature RNA and is presented as a complex for control of messenger RNA translation, in the form of RNA stability And most previous investigations have emphasized that this Argonaute protein is an essential component of microRNAs, at least in its role in RNA… the role of microRNA in control of gene expression So, the fact that we didn’t find it was… in exosomes in our preparation was troubling So, we went back and looked very carefully to see whether we could in fact see Argonaute in the exosomal vesicles and our conclusion is it is simply not there One piece of evidence is shown here If one examines the crude initial high speed pellet fraction of exosomes, sedimented from conditioned medium from HEK293 cells, one sees the typical membrane constituents of exosomes, and one also sees the Argonaute protein, there seemingly abundant In fact, there are publications in the literature that claim that Argonaute is contained within exosomes, but unfortunately those publications are based on the evaluation of this crude high speed pellet fraction If this pellet fraction is fractionated just one step more, by sedimentation on a buoyant density gradient, the Argonaute completely disappears There is no detectable Argonaute And in other experiments we’ve confirmed it simply isn’t there, whereas the YBX1 protein is, as are the membrane proteins that characterize an exosomal vesicle So, this is a surprise and must yet be explained Now, in the last experiments, I want to delve into what role the YBX1 protein may play in this process What I’ve said thus far is that it’s there It’s in the vesicles It’s bound to miR223 It is delivered into vesicles along with miR223 But what we really want to know is, is the YBX1 protein actually required for the sorting of miR223 and of other microRNAs? And for this we turned to the creation of a null allele, a homozygous null allele of the YBX1 locus using the now classic technique of CRISPR/Cas9 knockout We created both heterozygous and homozygous null alleles of YBX1 Shown here, the heterozygote continues to produce some YBX1 protein, as detected by immunoblot, but the homozygote, which in this case is not a deletion, rather, likely a frameshift mutation, produces very little residual YBX1 protein The 293 cells missing the YBX1 gene grow, at least as far as we can tell, perfectly normally in the laboratory So, there’s no overt essential function for YBX1 in normal growth processes However, we can ask, importantly, is the YBX1 protein required for the production of exosomes? And more specifically, is it required for the sorting of microRNAs into exosomes produced by either wild type or YBX1 mutant cells?
So, I’ll show you two remaining experiments that addressed this question This is an experiment that seeks to measure two aspects of exosome biogenesis using our cell-free reaction The top reaction is another assay that measures simply the production of exosomes as detected by the formation of exosomes that enclose a soluble enzyme in the cell-free reaction And we found that cytosol obtained from the YBX1 mutant cell was perfectly active in making exosomes in our cell-free system, compared to cytosol obtained from a wild type cell No defect in exosome manufacture in the cell-free system In contrast, if we used the microRNA packaging reaction that I described, cytosol obtained from the YBX1 null is dramatically deficient in sorting miR223 into vesicles formed in vitro, whereas cytosol from control wild type cells is perfectly active, in this case, packaging about 10% of the miR223 Another control was to take the YBX1 null cell and reintroduce a fresh full-length copy of the YBX1 gene and then to produce cytosol from that cell And that cytosol is restored to near but not full activity in sorting miR223 into vesicles formed in the test tube Now, another experiment was to evaluate intact cells for their production of exosomes We compared wild-type cells with YBX1 homozygous null and measured particles in the growth medium using a particle tracking device And found that the number of particles produced by YBX1 null cells is essentially the same as wild type cells Further, we evaluated the secretion of miR223 and another microRNA, miR144, in the normal and YBX1 null cells, and then compared that to the cytoplasmic accumulation of those two RNAs in cells missing the YBX1 protein And those data are shown here If we just look at the medium fraction, the conditioned medium, we find that both miR223 and miR144 are relatively depleted from the medium when the cells are missing the YBX1 protein, whereas the cytoplasmic RNA, miR190, is not changed at all It’s still present in the cytoplasm in YBX1 null cells If we look at the cells, the cells that sediment, that are missing YBX1, we find that both miR223 and miR144 now accumulate in the cell, at the expense of being secreted in exosomes So, both of these RNAs are dependent on the YBX1 protein for their packaging into exosomes based on the cell-free reaction and on the intact cell assay Now, the last data we’ll I’m simply going to evaluate the major RNA species that are found in exosomes, that are produced from wild type cells compared to the YBX1 cells I want to share with you one more recent unpublished observation conducted in collaboration with an RNA expert at the University of Texas by the name of Alan Lambowitz Alan has devised techniques for the detection of small RNAs, even those that have extensive secondary structure And so Matt, in collaboration with the Lambowitz lab, had a look at the major chemically abundant RNA species found in normal exosomes produced by 293 cells and those produced by the YBX1 null cells So, let’s have a look In the first instance, the most abundant RNA species that the Lambowitz lab detects in our purified fraction are intact full-length transfer RNA molecules And surprisingly, these transfer RNAs, this abundant species, is quite dependent on the
YBX1 protein, about a five-fold reduction seen in exosomes produced by YBX1 null for this RNA species There are two other more unusual RNAs that are fairly abundant in these exosomes, the so called Y-RNAs and Vault RNAs And interestingly, at least the Vault RNAs are, similarly, highly dependent on the YBX1 protein for their capture into exosomes Well, let me summarize, then We are at a point where there are many questions that remain I’m very excited I’m particularly excited by the fact that we have this cell-free reaction that can be resolved, I’m confident, by biochemical fractionation to identify all of the components that are required for this sorting event And I hope, as the years go on, that we can use this reaction to understand why the cell has gone to such considerable lengths to sort RNAs so selectively into exosomes for secretion What are these RNAs doing? That remains a most important question Well, let me, then, summarize our, in this case, relative state of ignorance, but what prospects we have for discovery We believe that the sorting of microRNAs into exosomes begins in the cytosol… cytoplasm with the interaction between proteins such as YBX1 and microRNAs such as miR223 to form a complex We have some evidence for sequence selectivity I showed you that cytoplasmic RNAs are not engaged in this, and can be discriminated in our cell-free system And in more recent experiments, we’ve been able to see some particular sequences on miR223 that are important Now, we now… we want to understand how this ribonucleoprotein complex can engage the endosomal membrane Are there proteins on the surface of the endosome that interact with this RNP? That bring it in, that deliver it into the interior of this organelle? Is this the site of sorting? Or is it possibly sorting to produce vesicles that may bud directly from the cell surface? So, that is another important question What is the membrane source? But once again, and importantly, the feeling that I have is that with the availability of this cell-free system we are in a position to understand, in some depth, the mechanism of this reaction by fractionating the required proteins And we hope, then, that that will lead to a further understanding of why the cell goes to such trouble to sort RNAs into extracellular vesicles Well, let me leave you, most importantly, with the people in my lab, in recent years, who have done so much important work I’ll highlight, just at the end, Matt Shurtleff, who’s recently taken his PhD, but many other students and fellows in the lab who have made my life so wonderful at UC Berkeley Thank you