Circulating tumor cell clusters: the ‘bad actors’ of cancer metastasis

Top

00:00:12;14 Hi, everyone.
00:00:13;22 My name is Mohit Kumar Jolly
00:00:15;02 and I'm a graduate student at Rice University.
00:00:17;13 Today, I'm going to take you on a journey
00:00:19;18 of how mathematics, or theoretical physics,
00:00:22;23 and experimental biology
00:00:24;22 can come together and produce novel insights
00:00:27;10 into biological systems
00:00:29;07 that none of those disciplines would have otherwise done alone.
00:00:32;00 And today, I'm going to talk about
00:00:33;23 how math and biology, together,
00:00:36;00 can give a better understanding of how
00:00:38;22 cancer cells communicate and cooperate
00:00:40;16 and drive tumor progression.
00:00:43;04 Math and biology...
00:00:44;17 sounds like two different extremes...
00:00:46;09 how did I get interested in coupling them?
00:00:48;26 Well, I was always interested,
00:00:50;26 but the point of no return
00:00:53;17 came during my Masters at IIT Kanpur,
00:00:56;07 when I was working on a mathematical model
00:00:58;21 of cell polarity.
00:01:00;06 I showed this model to a group of experimental biologists
00:01:03;23 and they shot at me,
00:01:06;03 Well, this is just a model
00:01:08;04 what does it have to do with reality?
00:01:09;26 And I shot back at them,
00:01:11;18 Well, what do you study?
00:01:12;29 Drosophila... that's also a model.
00:01:14;19 What does that have to do with reality?
00:01:16;27 So, both of them are models.
00:01:18;17 One of them is a mathematical model
00:01:20;04 and one is a biological model.
00:01:21;21 Both of them have their own assumptions,
00:01:24;17 strengths,
00:01:25;28 and limitations.
00:01:27;08 Can we not actually couple their strengths together
00:01:29;13 so that we can produce some new insights
00:01:31;28 into the system,
00:01:33;16 because, eventually, isn't that the goal?
00:01:35;25 Alright, coming back to cancer...
00:01:38;18 cancer...
00:01:40;17 this word sends a shiver down our spines.
00:01:42;22 Each of us, directly or indirectly,
00:01:46;04 have been affected by it.
00:01:47;12 Well, what is cancer?
00:01:48;24 So, let's take an example.
00:01:50;12 Here is a normal liver,
00:01:52;05 where all the cells are functioning properly.
00:01:54;13 But, when due to some aberrations
00:01:56;18 -- genetic, environmental, or some other reason --
00:01:59;26 some cells start dividing uncontrollably
00:02:03;12 in a manner that disrupts
00:02:05;18 the normal physiological functioning of the organ. That is when it is cancer.
00:02:10;20 As you can see in the image,
00:02:12;18 some white speckles which is cancerous tissue
00:02:15;14 which is taking over the normal tissue function.
00:02:18;19 But this is not the most destructive part about cancer,
00:02:23;06 this is not why most people die of cancer.
00:02:27;09 More than 90% of cancer-related deaths happen
00:02:31;25 because cancer spreads to different parts of the body.
00:02:35;23 How does it spread?
00:02:37;17 It spreads through the freeway in our body,
00:02:39;19 which is the blood vessels and lymph vessels.
00:02:41;26 So, imagine a tumor in a lung.
00:02:44;08 It will get on the freeway
00:02:46;21 and then travel all through the body,
00:02:48;01 and take an exit at multiple distant organs
00:02:50;06 to form metastases, to form secondary tumors.
00:02:53;06 Metastasis is the entire process of forming tumors
00:02:55;26 from one organ to another.
00:02:58;07 And this is what claims more than 90% of all deaths.
00:03:02;14 The war on cancer was declared in 1971.
00:03:05;22 What have we done since?
00:03:07;18 We have made huge process
00:03:09;26 in diagnosing cancer earlier,
00:03:11;14 such as with these mammograms.
00:03:13;24 We have made huge progress in identifying
00:03:16;03 what risk factors contribute to cancer,
00:03:18;06 such as smoking causes lung cancer.
00:03:20;17 And we have made some huge progress
00:03:22;24 in identifying what genes and what mutations
00:03:25;15 can drive cancer.
00:03:27;18 For instance, mutation in a gene called BRCA1
00:03:30;01 is the reason for many women getting
00:03:33;26 ovarian and breast cancer,
00:03:35;20 and that was the reason why Angelina Jolie
00:03:37;20 had to undergo a double mastectomy
00:03:39;26 in order to reduce her risk of getting breast cancer.
00:03:43;15 But despite this progress,
00:03:46;08 metastasis, the spread of cancer cells
00:03:49;01 from one organ to another,
00:03:51;03 remains clinically insurmountable.
00:03:53;22 We are still trying to scratch the surface
00:03:56;02 of how exactly this process happens.
00:03:58;05 This diagram here just shows where...
00:04:00;20 to which different organs
00:04:03;03 does the breast cancer metastasize?
00:04:05;25 And the most intriguing thing about metastasis
00:04:08;18 is that more than 80% of cancers, again,
00:04:15;03 in organs where cells do not have the innate property
00:04:18;00 to move or migrate.
00:04:20;03 Let's talk about breast cancer, prostate cancer, lung cancer
00:04:23;07 #NAME?
00:04:25;11 to migrate on their own,
00:04:26;26 so cancer switches on a distinct program
00:04:29;15 which enables them to get on the freeway
00:04:32;08 and survive through the harsh conditions
00:04:35;15 in the bloodstream
00:04:37;05 and then get off the freeway
00:04:38;22 and again adapt to
00:04:41;06 a whole new foreign environment
00:04:43;10 of a distant organ
00:04:45;00 and colonize it eventually,
00:04:46;19 therefore disrupting its physiological function.
00:04:49;20 So, when cancer colonizes multiple different organs of the body,
00:04:53;17 the probability of the survival of the patient
00:04:56;17 goes down further,
00:04:58;02 and that is why metastasis is most dangerous.
00:05:01;02 Well, if we have the list of genes
00:05:05;28 or proteins that control metastasis,
00:05:08;02 why don't we completely understand the process?
00:05:11;07 Let me give you an analogy.
00:05:12;21 If you have all the different car parts,
00:05:15;17 do you understand how a car moves?
00:05:17;28 Not necessarily.
00:05:19;19 You actually have to put those parts together
00:05:23;21 in an integrated, coherent manner,
00:05:26;02 then only do you understand how a car moves.
00:05:29;04 Similarly, you have to put together all different genes
00:05:32;25 in the way that they are actually connected
00:05:34;18 within a cell
00:05:36;18 to understand how exactly a cell moves,
00:05:38;17 a cancer cell moves,
00:05:40;02 a cancer cell metastasizes.
00:05:42;17 So, in other words, a car, as a system, as a whole,
00:05:46;11 is much greater than the sum of its parts,
00:05:50;07 and that is what we call an emergent system,
00:05:52;28 and that is what cancer and cancer metastasis is,
00:05:56;03 an emergent system.
00:05:58;07 And this approach of putting things together
00:06:01;01 is one which is very natural
00:06:04;04 to physicists and engineers,
00:06:06;04 because engineers are the one who put together this car.
00:06:09;11 So, can we put together different genes or proteins
00:06:12;15 in the way that they interact in a cell
00:06:14;16 in order to get a better understanding of metastasis
00:06:17;07 as a system, as an emergent, complex system?
00:06:21;19 Let me give you an example: weather forecast.
00:06:24;18 Today, we can predict weather very accurately...
00:06:27;21 well, what is going to happen in the next 7 days
00:06:30;07 or even further?
00:06:32;17 When are we going to get thunderstorms?
00:06:34;01 When are we going to get good sunlight?
00:06:35;20 And based on that we actually make our travel plans,
00:06:38;09 which means that we trust it so much.
00:06:42;22 Well, it took at least a century to get here.
00:06:46;12 Climate scientists, physicists, and engineers,
00:06:49;19 they collected all different factors that contributed
00:06:52;09 to the weather of a particular place,
00:06:53;23 the climate of a particular place,
00:06:55;19 and then put them together in the way
00:06:57;28 that they actually interact,
00:06:59;22 including how strongly they interact with one another.
00:07:02;14 And this is just a
00:07:05;08 'primitive' set of equations
00:07:06;20 that actually controls this,
00:07:08;01 there's a lot of math that went behind it,
00:07:10;02 and that is where we are today,
00:07:11;15 after having undergone this
00:07:13;10 validation and calculation of the model.
00:07:15;24 So, can we, today, begin this journey of
00:07:19;13 putting together various genes and proteins
00:07:22;03 that interact with each other to control metastasis
00:07:24;20 and, not necessarily today,
00:07:26;09 but some years down the line,
00:07:28;13 can we build a model which can actually
00:07:31;10 predict metastasis?
00:07:32;19 Can we have that amount of trust in it,
00:07:34;04 just as we have in the weather forecast today?
00:07:37;00 So, today I'm going to take you on
00:07:40;07 the first few steps of that journey
00:07:41;28 that I have latched onto.
00:07:43;19 Alright... so, metastasis...
00:07:45;12 when cancer cells leave the primary tumor,
00:07:48;00 they can do it in two different ways:
00:07:49;29 they can either leave as single cells
00:07:51;21 or they can form a cluster
00:07:53;26 before the enter the freeway
00:07:55;27 and then leave together.
00:07:57;08 It is like whether...
00:07:59;18 is there a set of cars leaving at different times
00:08:02;19 or is there a set of cars that are together on a ramp
00:08:06;04 just waiting to get on the freeway?
00:08:09;03 Experiments have shown that there are
00:08:11;04 30 times more single cells
00:08:13;18 that are shed off the tumor
00:08:15;05 as compared to only clusters.
00:08:17;22 But the clusters form
00:08:20;06 50 times more secondary tumors, or metastases,
00:08:24;02 than single cells.
00:08:25;20 And these clusters are not too large,
00:08:27;21 they are 2-7 cells large,
00:08:30;07 so you can do the math and realize that
00:08:33;12 there is indeed something
00:08:35;26 to cancer cells staying together
00:08:37;23 and that is what leads to much more metastasis.
00:08:40;21 There is a value in staying together;
00:08:42;15 united they stand, divided they fall.
00:08:44;24 So, what exactly is this value
00:08:47;01 and how do cancer cells utilize it?
00:08:49;06 Let's go through that.
00:08:51;06 So, if these are the real villains of metastasis,
00:08:54;09 these clusters, then let's try to understand,
00:08:56;28 how do cancer cells form these clusters?
00:08:58;22 How can we break these clusters,
00:09:00;15 what target can we use?
00:09:01;28 And, eventually, why do these clusters form
00:09:04;23 much more metastases?
00:09:06;09 Let's go to the first part:
00:09:08;01 ow do cancer cells form these clusters?
00:09:09;24 Before we go into the details,
00:09:11;18 let me give you a relatively detailed understanding
00:09:14;03 of how metastasis actually happens.
00:09:16;16 So, let's consider that
00:09:18;28 these primary lung cancer cells,
00:09:22;00 now, they do not have the innate property to move,
00:09:24;08 as we discussed earlier,
00:09:25;25 so in order to move, in order to get on the freeway,
00:09:28;11 they need to do two things:
00:09:30;02 they need to break their bonds with their neighbors,
00:09:32;17 the tight cell-cell adhesion that they have;
00:09:35;04 and at the same time,
00:09:37;04 they need to gain the property to migrate and invade.
00:09:40;04 Now, this property of staying together
00:09:42;22 and not moving
00:09:44;17 is the property of epithelial cells in our body,
00:09:46;23 such as the skin,
00:09:48;17 and the cells that line multiple organs,
00:09:50;19 such as breast, prostrate, and lung.
00:09:52;22 On the other hand,
00:09:54;20 the property of not adhering and migrating
00:09:58;12 is the property of mesenchymal cells in our body,
00:10:01;02 such as red blood cells.
00:10:02;18 So, when cancer cells have to get on the freeway,
00:10:05;03 they need to take on these properties
00:10:07;26 of mesenchymal cells.
00:10:09;12 I'm not saying that they have to
00:10:11;12 convert to mesenchymal cells,
00:10:12;22 they don't have to become red blood cells,
00:10:14;07 but they have to take on these properties.
00:10:15;28 And this transition is what is called as
00:10:18;21 epithelial-to-mesenchymal transition,
00:10:21;02 which is usually the first step in the metastatic cascade.
00:10:25;19 Now, once they on the freeway,
00:10:27;11 once they're in the blood circulation
00:10:29;22 and they survive and they get out at a different organ,
00:10:32;05 take an exit,
00:10:34;12 now what they need to do is that...
00:10:37;12 grow an entire new tumor,
00:10:39;19 for which they need to shed their migration
00:10:42;00 and invasion properties
00:10:43;29 and regain their property
00:10:46;08 of adhering to their neighbors.
00:10:48;14 And this is the exact opposite transition
00:10:50;16 that we just talked about,
00:10:52;09 mesenchymal-to-epithelial transition,
00:10:54;10 which is usually the last step in this metastatic cycle,
00:10:57;25 and the cycle can continue, go on and on,
00:11:00;13 forming multiple secondary or tertiary tumors
00:11:03;02 at multiple different organs in the body.
00:11:05;28 Now, EMT or MET,
00:11:07;23 as I'll refer to them throughout the talk,
00:11:10;07 are not some new processes
00:11:13;18 invented by cancer.
00:11:15;03 These are normal physiological processes
00:11:17;14 that happen during embryonic development
00:11:19;29 and even wound healing
00:11:21;23 and these processes are co-opted by cancer
00:11:24;25 to metastasize.
00:11:26;06 Let's talk about wound healing for a bit.
00:11:28;04 So, when you get a wound on your skin,
00:11:31;24 you see the cells move together
00:11:33;27 and actually close that wound,
00:11:35;22 and once the wound is filled,
00:11:37;10 they stop moving and settle down.
00:11:38;26 So, this is exactly what EMT and MET is going on.
00:11:43;28 But those cells typically move collectively, so...
00:11:49;09 which is not really epithelial and not really mesenchymal.
00:11:52;06 So, is there a state somewhere in between
00:11:54;22 where cells can both adhere and migrate,
00:11:57;09 which can close the wound and,
00:11:59;14 in the case of cancer,
00:12:00;28 it can form much more clusters...
00:12:03;03 is there a stable state, a hybrid state, in between?
00:12:06;02 So, recent studies have observed that
00:12:09;08 in circulating tumor cells in patients,
00:12:11;06 those cells can co-express epithelial and mesenchymal proteins,
00:12:16;04 suggesting that they're somewhere hybrid in nature.
00:12:19;12 More importantly, what they observed
00:12:21;21 based on the biopsies of primary tumors,
00:12:24;18 is that the more aggressive the cancer is,
00:12:27;23 the more the number of such cells.
00:12:30;17 Of course, this is just a correlation,
00:12:32;13 not a causal effect,
00:12:34;00 but it definitely points to some correlation...
00:12:37;05 it definitely points to some
00:12:39;19 connection between how hybrid cells...
00:12:41;27 how many hybrid cells are out there
00:12:43;21 and how aggressive the tumor is.
00:12:45;20 But despite this,
00:12:47;24 the feeling in the field has been that,
00:12:49;29 well, this is just a transient nature;
00:12:51;25 cells cannot actually maintain this state
00:12:54;19 for a very long time
00:12:56;12 and this is just a snapshot that got captured
00:12:58;26 when cells were in route at EMT.
00:13:01;28 So, that is the question that got me most excited:
00:13:04;10 is this a stable state
00:13:07;01 or is it just a transient, intermediate,
00:13:09;09 unstable state?
00:13:11;05 So, here are the three states we have been talking about:
00:13:13;19 epithelial, to the very left,
00:13:15;12 where cells adhere very strongly and do not move;
00:13:18;20 mesenchymal to the extreme right,
00:13:20;25 where cells migrate but they do not adhere;
00:13:24;09 and in between is this hybrid state where cells
00:13:27;29 both adhere and migrate,
00:13:29;10 therefore leading to this collective cell migration
00:13:31;14 or clustered cell migration.
00:13:33;15 Now, how do we understand
00:13:35;15 how cells can attain this state,
00:13:37;16 whether hybrid is a stable state at all or not?
00:13:40;22 Again, going back to the approach of putting the car parts together...
00:13:45;02 so, what I did was scan through literature of
00:13:48;22 multiple different cell lines belonging
00:13:50;20 to multiple different cancers
00:13:52;13 and try to identify, well,
00:13:53;27 is there some core unit,
00:13:55;13 is there some core set of proteins, genes, microRNAs,
00:13:58;15 that interact with each other
00:14:01;10 and control EMT in multiple different contexts?
00:14:04;07 And this is the network I came up with,
00:14:06;02 where there are four different players,
00:14:08;20 which are connected to each other
00:14:10;22 in a relatively complicated manner,
00:14:12;13 the details of which I am not going into.
00:14:14;11 And there... two of which are transcription factors
00:14:16;17 and two of which are microRNAs.
00:14:18;12 The numbers you see here represent
00:14:20;01 some kind of a quantitative relationship between them,
00:14:22;17 but, again, I'll not bore you with the details.
00:14:24;22 The only thing I want you to remember is that
00:14:27;10 what we know experimentally is that the amount...
00:14:29;22 if the amount of SNAIL is ZEB are high,
00:14:32;00 then cells are mesenchymal,
00:14:33;18 and if the amount of SNAIL and ZEB is low,
00:14:35;18 then cells are epithelial.
00:14:37;25 Alright.
00:14:39;14 So then, after finding these interactions,
00:14:41;27 I wrote down mathematical equations
00:14:43;28 which actually represent these interactions
00:14:45;21 and, after doing calculations with these equations,
00:14:48;15 let's see what exactly the model predicts.
00:14:53;06 So, here is the prediction of the model,
00:14:55;03 where there are three solid lines,
00:14:57;06 each of which represents a stable state,
00:14:59;10 which is a phenotype that cells can attain.
00:15:01;06 The dotted red lines represent unable states.
00:15:05;07 So, going back to standard high school physics,
00:15:08;03 we know that if the unstable state is perturbed,
00:15:10;12 it exits that state,
00:15:12;08 so cells cannot actually stay in that state,
00:15:13;25 so therefore we are going to focus only on the unstable states
00:15:16;25 in this context.
00:15:17;29 Now, let me break down this graph for you.
00:15:19;16 So, out of the four players that we've discussed,
00:15:21;08 I've here taken only two players,
00:15:22;20 just for the purpose of representation,
00:15:25;02 to see what exactly the model says.
00:15:27;29 So, at very low levels of ZEB,
00:15:32;00 which is the y axis,
00:15:33;12 you see there is only one state,
00:15:34;28 which is epithelial,
00:15:37;07 and that is what we know from experiments.
00:15:39;09 At very high levels of ZEB,
00:15:41;16 again, there is only one state,
00:15:42;27 which is mesenchymal.
00:15:45;05 And in between, you see this...
00:15:47;25 another solid blue line,
00:15:49;15 which corresponds to intermediate ZEB levels,
00:15:52;01 ZEB levels which are higher than those
00:15:54;28 seen in epithelial
00:15:56;19 and lower than those seen in mesenchymal,
00:15:58;19 and this intermediate state corresponds to
00:16:01;11 a stable hybrid state.
00:16:03;01 So, a prediction of the model is that,
00:16:04;13 yes, there is stable hybrid state that cells can achieve.
00:16:07;19 Alright.
00:16:09;06 Let me walk you through another facet of this graph.
00:16:10;27 So, if you see, the very left part of the graph...
00:16:13;06 now let's look...
00:16:14;29 we have already looked horizontally,
00:16:16;09 now let's look vertical...
00:16:17;17 so here, what you see is that
00:16:19;04 if the amount of SNAIL protein
00:16:21;08 is less than 180,000 molecules
00:16:23;05 and if you walk vertically upwards,
00:16:25;01 there is only one blue solid line that you cross,
00:16:28;11 which means that there is only one state that cells
00:16:31;04 in this region of parameters,
00:16:32;29 or in this region of conditions, can attain,
00:16:34;24 and that state, as you can see from ZEB levels,
00:16:37;05 is the epithelial state.
00:16:39;14 Going to the very right,
00:16:41;18 there is another region in which,
00:16:43;29 again, there is only one state,
00:16:45;12 but that particular state is not epithelial,
00:16:47;09 but the state with highest ZEB levels,
00:16:49;12 which is mesenchymal, actually.
00:16:51;27 And in between we see this interesting region.
00:16:54;18 For example, let's draw
00:16:57;02 a vertical line from 200,000 molecules of SNAIL,
00:17:00;07 and you see it will cross three distinct blue curves,
00:17:04;08 and each of those blue curves corresponds to a stable state,
00:17:07;12 as I've already mentioned,
00:17:08;25 and these three stable states are
00:17:11;00 epithelial, mesenchymal, and hybrid.
00:17:12;21 So, again, what the model predicts is that
00:17:15;17 there can be some conditions under which these three states can
00:17:19;10 coexist with each other,
00:17:20;29 so if... there might be
00:17:23;24 experimental cases, some cell lines,
00:17:25;29 in which you can see that all these different,
00:17:28;14 three kinds of cell or phenotypes
00:17:30;23 are present in the cells with the same genetic background,
00:17:33;08 which is what a cell line is.
00:17:36;04 So, once we made this model,
00:17:37;25 we wrote to a couple of experimental biologists
00:17:39;15 working on EMT,
00:17:41;08 trying to see, well, how do they react to this?
00:17:43;03 Do they react in a very similar way
00:17:44;25 that my master's experimental biologists did,
00:17:46;26 or they are actually supportive of this?
00:17:48;29 So, one group is Seattle actually wrote back,
00:17:51;17 among other groups as well,
00:17:53;11 and said, well,
00:17:55;00 we have two pieces of good news for you.
00:17:56;15 One is that we have been looking at
00:17:58;12 a cohort of 40 different lung cancer cell lines
00:18:01;11 and we actually see that many of them
00:18:03;21 lie somewhere in between
00:18:06;04 epithelial and mesenchymal.
00:18:08;04 And the other good news is that we are moving to Houston,
00:18:09;29 right across the street from Rice University
00:18:12;06 at MD Anderson Cancer Center,
00:18:13;24 so why don't you come over and we can discuss.
00:18:15;29 And this was indeed the turning point.
00:18:18;25 So, once we discussed with them,
00:18:20;29 they had these intermediate cell lines
00:18:22;24 which were lying somewhere in between epithelial and mesenchymal,
00:18:25;08 and how did they identify that these were intermediate?
00:18:27;21 Because they took measurements
00:18:29;27 of several epithelial and mesenchymal proteins
00:18:32;06 and their mRNA levels.
00:18:34;00 But beware, all these measurements
00:18:36;15 were taken on a population level:
00:18:38;01 they took 1000 or more cells of this
00:18:40;09 and then crushed them and see,
00:18:42;03 what was the amount of RNA present in them?
00:18:45;14 And this is the problem with taking
00:18:48;11 population-level measurements,
00:18:50;03 that you actually do not see
00:18:52;05 what is going on at the single cell level.
00:18:53;19 So, when we looked at those cell lines,
00:18:56;04 what we observed was that
00:18:58;19 there were some cell lines in which
00:19:00;24 there were some epithelial cells
00:19:03;04 and some mesenchymal cells,
00:19:05;03 but not hybrid cells.
00:19:06;23 So, here is a cell line, as shown here,
00:19:08;23 where red is marking an epithelial protein
00:19:12;11 and green is marking a mesenchymal protein
00:19:14;27 in these immunofluorescence images,
00:19:16;15 so you see cells have only red or only green,
00:19:19;16 which means they're only epithelial or only mesenchymal.
00:19:23;03 Another cell line where you see that
00:19:25;09 cells actually co-express red and green
00:19:27;15 at a single cell level,
00:19:28;28 so these are indeed hybrid cells,
00:19:31;00 which is different from a mixture.
00:19:32;18 So, a mixture is different than a hybrid cell.
00:19:34;10 Its characterization is different,
00:19:36;04 its migration traits are expected to be different,
00:19:38;13 and so on and so forth.
00:19:40;01 Again, this is a snapshot.
00:19:42;05 What is a stable state?
00:19:43;25 If it is perturbed, it comes back,
00:19:45;18 or, in other words,
00:19:47;08 it retains itself if not perturbed.
00:19:49;04 So, in order to see whether this hybrid cell line
00:19:52;24 which got us so excited
00:19:54;21 was indeed a stable state, we...
00:19:57;18 our experimental collaborators,
00:19:59;23 they actually cultured this cell line for multiple passages
00:20:02;06 and saw that, even after two months,
00:20:03;29 they were actually coexpressing red and green,
00:20:06;19 which is what is suggesting that,
00:20:09;16 yes, can they be indeed stably present
00:20:11;15 in that hybrid state.
00:20:13;16 Now, another thing that I've been talking about is that
00:20:16;24 these hybrid cells are likely to move collectively.
00:20:20;15 Now, let's see if that is shown in the experiments, also,
00:20:24;02 because eventually the proof of the pudding
00:20:26;01 is in the eating, right?
00:20:27;29 So, here we take a mesenchymal cell line
00:20:30;04 and a hybrid cell line
00:20:32;06 and what you do...
00:20:33;23 our experimental collaborators did is a scratch assay,
00:20:36;07 where you actually grow cells
00:20:38;15 until they are confluent on a plate
00:20:40;17 and then you just mark a scratch using a needle,
00:20:43;20 and the way cells move to fill in that gap
00:20:47;02 is what is used to characterize their migration properties.
00:20:50;19 Now, let's see how mesenchymal cells moved.
00:20:53;12 So, the rightmost column
00:20:56;07 is just showing a zoomed in image of the middle column,
00:20:58;05 as you can see these cells are
00:21:01;01 flying out as single cells
00:21:02;11 -- they just don't have any direct physical connection to each other.
00:21:05;27 However, the hybrid cells are moving more collectively
00:21:10;24 -- they have not yet broken the bonds with their neighbors,
00:21:13;07 they're actually moving collectively --
00:21:14;29 therefore leading to these finger-like projections
00:21:16;24 as pointed out by these white arrows.
00:21:19;00 So, indeed, these cells are hybrid,
00:21:22;02 because it's a stable state
00:21:24;10 and of course they move collectively.
00:21:26;20 Now, we were very intrigued with this hybrid cell line,
00:21:29;22 because in our diagram, in our model,
00:21:32;17 we did not have a parameter region
00:21:35;20 in which the only state was hybrid,
00:21:37;25 so that's what got us thinking that,
00:21:39;29 well, what...
00:21:41;28 how can we tweak the model,
00:21:43;15 how can we add or delete components from it
00:21:46;08 so that we can get a region in which the only state
00:21:48;14 is hybrid?
00:21:49;28 And also, at the same time, we wanted to identify,
00:21:52;04 well, what targets can break these clusters?
00:21:54;15 So, if we can identify
00:21:57;02 what can give us a stable hybrid state,
00:22:00;11 only a stable hybrid state,
00:22:02;00 deleting that can destabilize the hybrid
00:22:04;10 and potentially lead to
00:22:07;14 disrupted collective cell migration.
00:22:09;10 Alright, so let's go back to the math.
00:22:11;09 So, this is the diagram that I showed you earlier
00:22:13;05 and I would, again, like to highlight here that
00:22:15;29 there is no parameter region
00:22:17;20 in which the only state is hybrid.
00:22:19;02 There is a parameter region in which the only state is epithelial,
00:22:21;16 there is a parameter region in which the only state is mesenchymal,
00:22:24;08 but not a region in which the only state is hybrid.
00:22:27;15 So then I went back to
00:22:30;07 the wound healing and developmental EMT literature
00:22:32;06 to see if they are talking about some players
00:22:35;00 which can either disrupt or maintain
00:22:38;01 collective cell migration
00:22:39;26 and, once I had a list of players from that literature,
00:22:43;01 I looked at, well, do we know how exactly
00:22:45;01 they couple to the players that we have here?
00:22:47;14 And once we had a list of players
00:22:49;23 which were implicated
00:22:52;02 in collective cell migration
00:22:53;15 as well as we had enough quantitative information
00:22:55;17 about that to couple to this model,
00:22:57;20 I actually incorporated that into our model and,
00:23:00;23 as you will see, the diagram changes.
00:23:02;23 So, this is just one example that I'm showing you, here,
00:23:05;03 this particular protein called GRHL2,
00:23:07;25 which connects with ZEB
00:23:09;21 -- it inhibits ZEB and is inhibited by ZEB --
00:23:12;00 and you see the diagram changes
00:23:14;00 and if you look at approximately
00:23:16;05 400,000 molecules of SNAIL
00:23:17;25 and go vertically upwards,
00:23:19;23 as shown in this region highlighted by this dotted rectangle,
00:23:22;08 you see the only state is hybrid.
00:23:26;05 So, adding this single factor
00:23:28;16 stabilized the hybrid state.
00:23:31;01 Again, let's go to the inverse
00:23:33;27 #NAME?
00:23:36;29 Again, let's go back to the experiments
00:23:38;26 and see if it actually holds true.
00:23:40;25 So, we went back to the same cell line,
00:23:42;18 which was stably hybrid
00:23:44;19 and, as you can see
00:23:46;25 again, they're moving collectively, as you can see in the left column.
00:23:49;17 On the right column, what we did was we
00:23:51;25 deleted that particular protein from the cell line,
00:23:53;22 and you see that their migration
00:23:56;18 is not completely collective now
00:23:58;16 #NAME?
00:24:01;16 moving more as single cells.
00:24:03;23 Also, when we looked for the presence of red and green,
00:24:07;26 or epithelial and mesenchymal proteins
00:24:10;00 that we have been talking about,
00:24:11;12 this intermediate cell line,
00:24:13;00 as I showed you earlier, this hybrid cell line,
00:24:15;06 co-expresses red and green in the same cells,
00:24:18;02 but with the knockdown of this single protein,
00:24:21;06 as predicted by the mathematical model,
00:24:23;07 you see that the red just goes away
00:24:25;12 and the cells are only mesenchymal.
00:24:27;28 And that is consistent with the idea that
00:24:30;06 they're moving more as single cells
00:24:32;03 rather than as collective cells.
00:24:34;05 So we predicted this from the model
00:24:36;00 and then we got it eventually validated,
00:24:38;13 but I should remind you that
00:24:40;17 the whole idea about this project came from...
00:24:43;04 that there is a stable hybrid cell line,
00:24:44;27 so this is a wonderful example of
00:24:46;29 how biology can inform math and math can inform biology,
00:24:50;11 and we can iteratively go forward
00:24:52;09 to have a better understanding of the system.
00:24:54;11 Now, once we had done this in vitro validation,
00:24:56;27 we wanted to go to some clinical data
00:24:59;06 and see if it supports our hypothesis.
00:25:01;02 So, a high amount of this protein, GRHL2,
00:25:04;19 will potentially lead to more collective migration,
00:25:07;03 more clusters,
00:25:08;12 and more clusters will decrease the survival period.
00:25:11;13 So, what I did was look at a cohort of
00:25:14;01 approximately 1000 lung cancer patients,
00:25:16;11 which were divided into high and how amounts of this protein,
00:25:19;13 and then we looked at their 5-year and 10-year survival probabilities.
00:25:23;10 As you can see, the patients that have
00:25:26;25 high amounts of this protein
00:25:28;17 have a lower probability of progression-free survival,
00:25:30;17 which means that the disease has actually progressed,
00:25:33;21 in more cases,
00:25:35;15 as compared to the percentage of cases
00:25:37;17 in which there is a lower amount of this particular protein,
00:25:40;20 and that is what is shown in the graph here.
00:25:42;28 The red line shows the cases where the protein...
00:25:46;02 where the patients have high amounts of this protein
00:25:48;07 and the black graph shows that...
00:25:50;14 the black curve shows the cases where
00:25:53;04 the patients have low amounts of this protein,
00:25:54;17 and there is a difference, a stark difference
00:25:56;18 in both the 5-year and 10-year survival probabilities,
00:25:58;27 which is consistently with our hypothesis.
00:26:01;23 Alright, so we have talked about these two aspects,
00:26:04;02 both of which are connected to
00:26:07;01 how cancer cells get on the freeway.
00:26:09;04 Now, let's talk about the last aspect,
00:26:11;25 that is, once they have survived the circulation
00:26:14;29 and gotten off the freeway to a secondary organ,
00:26:18;19 to a distant organ,
00:26:20;14 how exactly do they form secondary tumors.
00:26:22;07 So, why do clusters form more metastases?
00:26:24;27 Again, a very similar approach to
00:26:27;23 what we have been talking about,
00:26:29;10 putting together parts of a car.
00:26:31;04 Take the EMT network that we have been studying
00:26:32;26 and take the tumor initiation network,
00:26:35;01 as reported in published experimental data,
00:26:37;16 and see how they're connected to each other,
00:26:39;29 again, as published in the data.
00:26:42;15 And then simulate this and
00:26:45;07 see what the model predicts.
00:26:46;06 Now, why this project was most exciting to me was
00:26:48;24 because there were three different papers
00:26:51;00 comparing the tumor initiation abilities
00:26:53;10 of epithelial and mesenchymal cells
00:26:55;08 #NAME?
00:26:57;04 and the cells that move as single cells, individually --
00:26:59;06 and they said three completely different things.
00:27:01;22 One... the first paper said that
00:27:03;13 when cells undergo EMT to become mesenchymal,
00:27:06;10 they form much more tumors.
00:27:09;04 A couple of years later, another paper saying that,
00:27:11;00 well, when cells become mesenchymal,
00:27:13;02 their tumor initiation ability goes down
00:27:15;26 and the cells that are epithelial
00:27:16;29 are the ones that form more tumors.
00:27:20;20 And the last one saying that,
00:27:22;12 well, they are almost comparable
00:27:23;27 -- epithelial and mesenchymal can initiate almost equal numbers of tumors,
00:27:26;21 there is not a stark difference in that ability.
00:27:28;28 Now, this is conflicting, hugely confounding,
00:27:31;23 and one of the reason why I thought that
00:27:35;10 this paradox is being reported is that...
00:27:38;15 because all these studies actually categorized cells
00:27:42;18 into two categories,
00:27:44;15 either epithelial or mesenchymal.
00:27:46;06 So they completely missed out on these hybrid, stable cells,
00:27:49;26 which were miscategorized
00:27:52;12 as either epithelial or mesenchymal,
00:27:54;10 and it can be different in different studies,
00:27:56;24 therefore leading to these potentially different results.
00:27:59;29 So, then when I simulated this model
00:28:03;17 of EMT and the tumor initiation network together,
00:28:06;10 the model predicted that the cells in the hybrid...
00:28:09;19 the cells that are hybrid actually form
00:28:12;06 much more tumors than cells that are
00:28:15;00 only epithelial or only mesenchymal.
00:28:16;08 Now, one year later,
00:28:18;26 after we published this, there was an experimental study,
00:28:21;04 which, to the best of my knowledge,
00:28:22;25 is the first study that categorizes cells
00:28:24;18 into three different categories
00:28:26;03 -- these are also breast cancer cells --
00:28:27;22 epithelial, hybrid, and mesenchymal,
00:28:30;13 segregates them,
00:28:32;05 and then looks at their tumor initiation ability
00:28:33;27 and, as you can see,
00:28:36;21 tumor initiation ability of the hybrid cells
00:28:38;15 is ten times more
00:28:41;07 than the tumor initiation ability of
00:28:42;18 only epithelial or only mesenchymal,
00:28:44;10 which is, again, very consistent with what
00:28:46;19 the model predicted.
00:28:47;27 Of course, we didn't have this ten times number,
00:28:49;17 but it's qualitatively the same result.
00:28:52;04 So, just summarizing...
00:28:53;25 so, I've taken you through a journey of
00:28:55;28 how theoretical physics or mathematics
00:28:58;04 can be coupled with experimental biology
00:29:00;06 to get insights into this entire process of
00:29:02;28 tumor progression and metastasis.
00:29:04;17 So, the first part was,
00:29:06;28 well, how do cancer cells form these clusters,
00:29:09;05 where we made a network of EMT and MET,
00:29:12;06 and predicted that, indeed,
00:29:14;22 the hybrid state is a stable state,
00:29:16;21 which is what is in contradiction to the current existing framework,
00:29:20;27 where people believe it's just a transient state.
00:29:23;28 Then the next part was, well,
00:29:25;13 how can we break these clusters?
00:29:27;07 Where, again, based on mathematical modeling,
00:29:29;09 then in vitro experimental validation,
00:29:31;08 and then based on some clinical data,
00:29:32;23 we concluded that GRHL2
00:29:35;24 might be an important player to
00:29:38;07 break that clustered migration.
00:29:41;02 And then in the last part,
00:29:43;03 we tried to offer an explanation
00:29:45;03 of why clusters might be forming more metastasis,
00:29:47;06 because clusters are the ones that are
00:29:49;22 usually corresponding to a hybrid state,
00:29:52;05 and the hybrid state is the one that
00:29:54;15 has much more tumor initiation ability
00:29:56;25 than either mesenchymal or epithelial.
00:30:00;15 So, just concluding, as I already indicated,
00:30:03;04 the existing framework has been that
00:30:05;02 this hybrid state is just transient,
00:30:07;02 cells can not maintain it for a very long time,
00:30:08;27 but what we showed,
00:30:10;08 first based on mathematical modeling
00:30:11;24 and then based on experimental validation,
00:30:13;14 that, indeed, it can be a stable state,
00:30:15;21 in some scenarios,
00:30:17;15 and it can move collectively.
00:30:19;07 Therefore, this state might be the reason behind
00:30:22;22 why clusters form much more metastases,
00:30:24;26 because it enables the formation of clusters
00:30:27;15 of circulating tumor cells, or CTCs.
00:30:31;16 Now, what are the clinical implications of this work?
00:30:34;10 So, these days, there's a lot of talk about liquid biopsy.
00:30:37;21 Now, what is liquid biopsy?
00:30:39;02 How is it different from solid biopsy?
00:30:40;25 So, solid biopsy is when you take
00:30:42;19 a chunk of mass from a tumor, from a patient,
00:30:44;27 and then try to characterize it.
00:30:46;11 Liquid biopsy is that,
00:30:48;05 instead of taking that chunk of solid tumor mass,
00:30:50;25 you take the blood of a patient, which...
00:30:53;11 and this process is much quicker
00:30:55;10 than taking an entire solid mass,
00:30:56;24 it is much less invasive,
00:30:58;22 it is much less painful,
00:31:00;03 so there are multiple advantages to this.
00:31:02;04 That's why people are very excited about its use
00:31:06;18 in the clinic.
00:31:08;01 So, once we have access to blood,
00:31:09;24 we can actually focus on circulating tumor cells,
00:31:12;22 and once we have circulating tumor cells, currently,
00:31:15;22 what people are doing are they are
00:31:17;28 not necessarily distinguishing between
00:31:20;18 single circulating tumor cells
00:31:22;03 versus clusters of circulating tumor cells.
00:31:23;21 All of them are being counted
00:31:25;24 just as total number of CTCs
00:31:29;01 in the blood,
00:31:30;28 which we suggest is not necessarily the entire picture
00:31:33;24 #NAME?
00:31:36;04 or they are clustered.
00:31:37;13 And if there are more clustered...
00:31:39;17 if there are...
00:31:41;02 if a patient has more clustered cells,
00:31:42;28 potentially that patient
00:31:45;12 is going to have a much worse outcome,
00:31:46;29 so therefore you need to have a much
00:31:49;09 more aggressive therapy going on
00:31:51;21 to attack that aggressive tumor.
00:31:53;09 And also, if we can take these measurements
00:31:55;03 at multiple time points from the same patient,
00:31:57;01 it can also be used to
00:31:59;03 monitor the disease progression
00:32:00;23 and see how well the patient is doing.
00:32:02;16 If there are more clusters,
00:32:04;10 the therapies are not having that good of an effect,
00:32:06;04 but if there are less clusters then,
00:32:07;28 potentially, they are having a relatively stronger effect
00:32:09;28 in this particular case.
00:32:11;26 Alright.
00:32:13;09 And I would just like to end with
00:32:16;02 the last message from the documentary
00:32:18;08 that was made based on
00:32:20;07 cancer, the emperor of all maladies,
00:32:21;18 which is that if the cancer cell is evolving,
00:32:23;18 so are we,
00:32:25;03 and I strongly believe that
00:32:27;06 a part of this evolution
00:32:29;09 is bringing together math and biology,
00:32:31;03 and therefore understanding what design principles
00:32:33;13 does cancer operate on
00:32:35;03 and can we actually be smarter than that.
00:32:38;02 And with that I would like to
00:32:39;24 acknowledge all the different mentors and colleagues
00:32:42;06 I've had on this journey of integrating math and biology together.
00:32:44;27 My advisors, Herbert Levine, Eshel Ben-Jacob
00:32:48;10 -- Eshel unfortunately passed away a year ago --
00:32:50;18 and my other mentors at
00:32:52;29 the Center for Theoretical Biological Physics,
00:32:54;24 Jose Onuchic and Cindy Farach-Carson
00:32:56;26 #NAME?
00:32:59;01 who has been extremely, extremely supportive
00:33:01;08 of using theory to understand more about cancer --
00:33:04;06 and my experimental collaborators from MD Anderson,
00:33:07;03 Duke,
00:33:08;03 MIT,
00:33:09;09 Johns Hopkins,
00:33:10;26 and my theoretical gang at the Center for Theoretical Biological Physics
00:33:12;28 at Rice University.
00:33:14;15 Thank you so much for listening.

This Talk

Speaker: Mohit Kumar Jolly

Audience:

Educators of H. School / Intro Undergrad
Student
Educators of Adv. Undergrad / Grad
Researcher
Educators

Recorded: May 2016

Talk Overview

Despite major advances in cancer biology, we do not fully understand the molecular cues that trigger metastasis. Cancer cells need to leave behind their organ of origin, travel through the blood or lymph node system, and re-establish themselves in a new tissue. This requires cancer cells to break their adhesion with their neighbors and gain the ability to migrate. Bioengineer Mohit Kumar Jolly, along with his experimental collaborators, developed a model combining theoretical physics and experimental biology to identify cancer cells that can both adhere and migrate, thereby allowing cancer cells to travel in clusters and achieve higher chances of forming a secondary tumor. This work highlights the value of integrating theory and experiments to provide new insights into the underlying principles of cancer metastasis and progression.

This talk is part of the Young Scientist Seminars, a video series produced that features young scientists giving talks about their research and discoveries.

Speaker Bio

Mohit Kumar Jolly

Dr. Mohit Kumar Jolly earned his Bachelors and Masters degrees at the Indian Institute of Technology Kanpur (IITK), India. There he began to recognize the promise of mathematics or theoretical physics to understand complex biological systems. He joined Dr. Herbert Levine at Rice University in Texas for his PhD to study the intrinsic characteristics of… Continue Reading