Epigenetics/Attractors: Read Article 12 S. Huang. Reprogramming cell
fates: reconciling rarity with robustness. BioEssays 31 (5):546-560,
2009.. Study this carefully. Post a PCRC
The Systems Biology of COVID-19 and the SARS-CoV-2 virus. The class will build a foundation that includes the emergence of complexity, simple biological subsystems, their reductionist and equivalent toy and organ-chip models, and the measurements required to specify model architecture and parameters. Applications to biology, physiology, medicine, chemical and biological defense, pharmacology, drug discovery, and toxicology. UGrad: PHYS 240 01 and BME 290B; Grad: PHYS 326 and BME 395C.
Kelly McGee
ReplyDeleteArticle 11b
Reprogramming cell fates:
reconciling rarity with robustness
0: Knew: Most of the pieces of knowledge I already knew were from the first paper, 11a, regarding cell phenotype and differentiation. I also was somewhat aware of attractants.
1: Learned: I learned how the physical/mathematical concepts of attractants are useful in discussing and potentially modeling cellular differentiation and reversal back to progenitor states.
2: Questions: What kind of studies are out there that have utilized this method of visualization/analysis to discuss how reversal to progenitor status is accomplished, or in fact any form of differentiation or differentiation reversal?
3: Presentation: Cellular differentiation reversal to progenitor status, and how it has been accomplished in detail.
4: Comments: Lengthy, thick paper, but useful figures.
Chinowsky_TheorSysBio_PCRC_09129016
ReplyDelete0. Knew
I knew from previous biology classes that it had been long thought that cells “lost” their ability to act as pluripotent stems cells after they had differentiated from the germ line cells, but that this had been proven untrue. From the previous paper for this class, I felt as though I had a reasonable understanding of state vectors, and how they can be used to describe and construct biological networks.
1. Learned
I thought it was very interesting how this paper made a clear distinction between the architecture of a network and the dynamics of a network. While the architecture may just contain information on the various relationships between genes and their directionality, the dynamics of the network must seek to incorporate the “metadata”, as we discussed in class on Tuesday. The paper also argues that gene network architecture is not responsible for cell differentiation—something that makes sense, particularly given the fact that the essential “network”, aka, the human genome is the same starting point for every type of cell. Stability or instability of gene behavior is dependent on gene interaction within a given state vector—this creates a barrier that can be thought of as the reason for final differentiated cell types that don’t dynamically switch back to pluripotent behavior. Cells types can be described as attractors in an N-dimensional space.
2. Pressing
It seems as though the real challenge of system biology is not network architecture, but the network dynamics. How often are models such as the one described in this paper (that use state space, attractors, etc), used in system biology today?
3. Presentation
GEDI maps
4. Thoughts
I liked the comparison to Maxwell’s Demon. Overall, this paper was easy to comprehend, although I would have preferred if there was more mathematical notation that was separated from the body of text. LOVELY figures!
Could you give a quick explanation of Maxwell's Demon?
DeleteMaxwell's Demon was first proposed as a thought experiment to violate the second law of thermodynamics. Essentially, you start with a box filled with a gas that has two chambers separated by a door. The question is what if there was some entity (in this case a demon) that could control opening and closing the door in such a way as to move faster moving molecules to one chamber and slower ones to the other. This would then cause a decrease in entropy (heat flowing from cold to hot instead of hot to cold) and violate the second law of thermodynamics. I believe the resolution of this thought experiment is that anything that would be able to know the speeds of all the gas molecules and separate them would itself have to produce more entropy than what is lost from separating the molecules, and the second law is saved (not even demons can violate the second law).
DeleteIn the context of this paper, instead of the demon knowing all of the molecules' speeds, it knows all of the gene expression profiles of each cell type. It could then change the expressions to obtain a desired profile/cell type, just like how the previous demon could separate the molecules to obtain a desired decrease in entropy. I think the paper says the resolution to this would be the barriers in the epigenetic landscape prevent certain profiles from being accessed, but maybe I am wrong on this.
0. Knew:
ReplyDeleteI knew about attractors from the previous paper. I knew about the use of expression profiles, denoted by S, as a state vector to hold information about the level of gene expression for a gien cell type. I knew about the attributes of a GRN (or networks in general) from previously discussed papers. I knew about state space and flow fields as explained in box 2 and the section on state spaces, trajectories, and attractors.
1. Learned:
The paper begins by bringing attention to two surprises in biology: the reversibility in cell differentiation (induced pluripotent stem cells) and pluripotency being a “ground state”. This cell reprogramming is rare yet robust, meaning that under the right conditions it is inevitable. The main goal of the paper is to explain and relate these surprises through epigenetic bariers in the framework of a gene regulatory network.
I learned that the squared Euclidean distance between two state vectors actually does have biological significance: it measures how “hard” it is for a cell to go from one state to another; however, this distance measurement does not explain the stability of cell lineages. This is explained through the use of the epigenetic barriers between different gene expression profiles.
I learned that there are two major problems in labeling certain DNA modifications as “epigenetic”. First, there is a lack of stability, meaning that the modifications are reversible, dynamic, and error prone. Second, there is a lack of locus-specificity, meaning that these modifications work on any gene in the genome.
The epigenetic barriers prevent the gene expression profile from accessing states that are “forbidden”, and thus prevents certain trajectories. These barriers are encoded in the topology of the GRN, and they arise in the dynamics of the GRN.
The paper goes on to discuss how differentiated cell types are in fact attractors in the high-dimensional gene expression profile state. In this view, cell types are stable to small perturbations (since an attractor state is like the minimum of potential energy), but with large perturbations a cell type can change (much like supplying enough energy to overcome a potential barrier). In this way, reprogramming a cell type is just a transition between attractors.
I learned that the multipotent state is an attractor state due to the fact that the mutually inhibitory genes also stimulate their own expression (this goes along with figure 2D-F). In terms of a potential energy landscape, this state is still at a higher energy than the differentiation attractor states, which is why a stem cell in the multipotent state can still differentiate to other types.
Finally I learned that the rareness of cell reprogramming comes from multiple, not as stable “mini-attractors”. This forms a “wash-board potential” that is harder to traverse than a smooth landscape.
2. Pressing Questions:
On page 556 it is mentioned that cell fate decisions are the result of altering the landscape structure from changes in the network (bifurcation). I thought from previous parts in the paper (especially box1) that the network its self is not changing, just gene expression. Therefore, since the landscape is determined by the network, I thought it would not be possible to alter the landscape and remove attractor states/barriers. So I guess my question is what am I missing here?
3. Presentation Topic:
I really followed the concepts of attractor states, networks, etc. in this paper. However, I need more understanding on the biology side. A presentation over the biology briefly mentioned in this paper would be beneficial. For example, an overview of stem cells would help understand the context of this paper.
4. Thoughts:
This was my favorite paper from this class so far. It was explained very well and was very easy to grasp due to the examples and figures provided.
0. KNEW:
ReplyDeleteMost of what I knew about this paper I had learned from 11A. I knew about cell fate determination and its relationship to regulatory networks. I had previously learned about attractors, but learned more about theoretical applications of this concept with respect to stem cells.
1. LEARNED:
I learned that chromatin modification has a dynamic relationship with genes, allowing for a reversible triggering of the gene. I learned that a transcriptome is the sum of all the messenger RNA molecules that are expressed from a gene, and how this can be used to approximate the stage of development for a cell. I understand more that dendrograms are used to illustrate gene clusters often based on Euclidean distance, but need further clarification. I also learned that epistemology is the theory of knowledge, and I thought that has great significance to this class.
2. PRESSING QUESTIONS:
In the conclusion the paper refers to the term "stemness" that was originally coined by Gretchen Vogel in a paper that searches for the proof of such a gene. This paper challenges the belief that this is a gene, but rather a dynamical state. Is "stemness" then the ground state characteristic of a particular entity?
3. PRESENTATION:
A complete presentation about each part Figure 1 would be extremely useful before moving onto the finer details of the paper.
4. THOUGHTS:
The graphical explanations helped illustrate many of the more difficult concepts, allowing this paper to be well understood. I enjoyed the paper, but found myself losing interest
Ben Terrones
ReplyDeleteSysBio16 Asgn_11B_Class_11_Article_12_2016_09_29
0. KNEW:
I knew about the default way to diagram cell behavior as a pathway with arrows between individual events showing causation. I knew about the idea of attractor states from the other paper assigned.
1. LEARNED:
How much more can be learned if we look at expression profiles through the lens of dynamical systems theory. I learned, at a basic level, what an integrative reading of gene expression patterns looks like. I learned what “distances” between gene expression patterns are and how they’re calculated. I learned about epigenetic barriers, barriers that prevent inter-conversion between stable gene expression patterns. I learned what a multi-potent state is, with its expression pattern between the two attractors. The cells in attractors are able to move around in the “basin” of the attractor state but stay around it waiting for external cues to differentiate. I also learned why it is rare for reprogramming events to occur.
2. PRESSING ?:
So the cells can transition from attractor states by either being kicked out of their current states to lower states by external cues, or by the landscape itself changing? I am slightly confused about what it would mean for the landscape to change.
3. PRESENTATION:
A presentation specifically on the reprogramming as transitions between attractors.
4. THOUGHTS:
Very interesting paper, I liked how they didn’t just use mathematical equations but instead explained everything qualitatively; it was much easier to understand.
Sylvia Morrow
ReplyDeleteAsgn11B_S. Huang. Reprogramming cell fates: reconciling rarity with robustness.
0. KNEW: The basic approach of using a state vector to quantify, organize, and describe gene expression (from the previous article). As soon as I saw Fig. 2B it was clear that this was very similar to a surface potential.
1. LEARNED: I was not aware of the "the rarely articulated but deeply rooted and broadly accepted thinking in developmental biology is that the lineage of a mature cell, once established, is essentially irreversible." (546). I think overall I gained a much better understanding of what gene expression means and what it means to turn 'on/off' a gene. I found the discussion of gene regulatory networks pretty interesting. In particular that "interaction specifications of the GRN architecture are determined by the structure of proteins and target DNA sequence, the GRN architecture is ‘‘hard-wired’’ in the genome" (550). The concepts of attractors, sub-attractors, and multi-stability seemed obvious to me from the potential analog, but it was useful to get comfortable with the specific terminology and context.
2. PRESSING ?:
--Overexpression: what draws the line between expressed and overexpressed?
--Figure 1E: what are the lines on the lower half of the figure supposed to represent?
--To what extent are the variables used in the model measurable?
3. PRESENTATION:
--I've lost track of how this relates to MOA. Maybe a short MOA presentation in this context?
4. THOUGHTS: Paradigm reversals are always fun.
Natalie Hawken
ReplyDeleteAsgn 11B, Article 12: Reprogramming cell fates: reconciling rarity with robustness
0. KNEW
I knew the basic biology of cell differentiation (or at least how they teach it now). I knew that the differentiation process is often due to higher/lower concentrations of specific transcription factors or other epigenetic effects (like mechanical stress and strain on cells). I was aware of the recent technologies developed to de-differentiate somatic cells to pluripotent stem cells and I knew that it involved fluctuations in certain transcription factors. From Article 11, I learned the basics of the GRN, attractor states, high-dimensional analysis of cells, and gene expression patterns.
1. LEARNED
The brief overview of attractor states in Article 11 left me more confused than when I started, but this paper really explained the ideas of "basin of attraction," population heterogeneity, and separatrix steady-states. I understand how multiple genes and their expression patterns lead to the creation of the attractor states. Figure 3 in the paper was what really convinced me of the application of attractor states to cell differentiation. The paper explained really well how stem cells can differentiate in so many ways and how very small populations of dtem cells can de-differentiate in certain circumstances. I really liked how the paper explained the "robustness" of the cell states and how that accounted for individual cell heterogeneity, which I think has been left out in a lot of the previous papers that focus more on cell populations. When discussing the stability of the multipotent state, the authors mention how self-activation leads to the disappearance of the separatrix, replaced by a small basin of attraction. This really tied up the argument well in explaining how this cell state with very high potential energy for differentiation does not automatically differentiate.
2. MOST PRESSING QUESTIONS
These theories are super cool and ground-breaking, but there isn't a lot of real-life proof of the concept of attractor states beside mentions of previous papers. Since this paper has been published, are there any in vivo or in vitro studies that prove or disprove the theories presented here (specifically the epigenetic barriers of attractor states, the self-stabilizing effect of the "ground state", the heterogeneity of cells leading to easier changes to other attractor states) (pg 558)?
How can we use the knowledge we currently have of gene-gene expression interaction profiles to develop multi-dimensional state space graphs (like in Fig 2) for lots and lots of genes involved in changes/maintenance of attractor states?
3. PRESENTATIONS
Current protocols for de-differentiation of somatic cells to pluripotent stem cells. What is the easiest way to do it? How do the reprogrammed cells differ from naturally occurring pluripotent stem cells?
Look into a dendrogram of cell fates/attractor states?
4. THOUGHTS
Ok, so this paper kinda blew my mind a bit. I really enjoyed how this paper challenges the current beliefs of cell differentiation. The authors did well to flesh out the basic knowledge of attractor states and the high-dimension network before diving into their main points. Also, I like his ideas in theory, but I wonder if it would be difficult to study these attractor states with so many dimensions in actual cells. Just in general, I find epigenetics fascinating, so this paper was pretty awesome.
0. Knew:
ReplyDeleteI knew about stem cells and induced pluripotency. Working in a bone TE lab, we use MC3T3 cells in a lot of our work and look for their differentiation into osteoblasts. Multipotent MSCs are also used frequently in the field.
1. Learned:
I was very interested to learn some of the mechanistic background of differentiation. This is something that is easily taken advantage of without really understanding anything that is going on at the gene expression level. The simplistic explanations really helped clarify what we've discussed in the past week and made it pretty clear how difficult expanding this approach to include all 25000 genes.
2. Pressing ?:
I am confused about the stabilization of pluripotency ("Multipotent State: An Attractor with Balanced Expression Patterns" section). So each transcription factor is self-producing (at constant rates to keep the ratio balanced)? When you consider all of the genes involved, this is a lot of factors that must remain balanced..
3. Presentation:
iPSCs definitely complicate (or I guess could be a good tool for) studying gene expression networks. I think a presentation on iPSCs and how they are commonly produced in vitro would be good.
Thoughts on if they are useful or make gene expression study more complicated?
4. Thoughts:
This paper was longer, but I really enjoyed reading it. The simplistic explanations were very clear and straightforward and it helped me to see how everything we've discussed relates to my research.
Class 11. Article 12 S. Huang. Reprogramming cell fates: reconciling rarity with robustness. BioEssays 31 (5):546-560, 2009.
ReplyDelete0. KNEW:
I am familiar with the language of statistical mechanics and differential equations (including attractors, sinks, and state spaces) so I followed Huang's attempt to make use of those areas without much trouble. I know that stem cells exist and that they differentiate, but I am not sure if I have thought much about if and how they might be able to change into other cell types after differentiating the first time. I am aware, however, that this would be huge--being able to reprogram cells to stem cells, and stem cells to other cells at will would be huge in helping people fix parts of their body.
1. LEARNED:
The main thing I learned was Huang's model of cell fate determination, in which state space is conceptualized as a region full of hills and valleys, and cells are balls rolling around those hills and valleys. Cells stuck at the bottom of a deep valley are thought of as differentiated cells (blood cells, liver cells, lung cells,...), since you have to push them hard to get them into another valley (which would correspond to another type of cell).
In this model, the stem cell state is like a shallow valley at the top of a tall mountain. Stem cell differentiation corresponds to a ball rolling off a particular side of the mountain, and into one of the deep valleys below. The ability of many types of differentiated cells to be reprogrammed back into stem cells corresponds to the ability to get back up the mountain from all sides; the corresponding difficulty follows from the fact that they're trying to climb a mountain!
Randomness is incorporated by having balls jiggle around randomly; the process is not entirely deterministic, so this jiggling allows balls to differentiate or not depending on, for example, whether their jiggling has brought them close to the edge of a mountain or far away from it.
2. QUESTIONS:
I certainly appreciate the math discussed here, and the model seems reasonable. However, I do wonder whether this model can be used to obtain some legitimately new and practical insights on stem cell reprogramming, or whether it is simply an interesting but ultimately unhelpful way of thinking about stem cell differentiation.
Put another way, how is this model of use to experimentalists? Can thinking about stem cell differentiation this way help someone figure out how to reprogram a blood cell into an induced pluripotent stem cell?
Also, a minor question from the paper: it is said (pg. 549, end of first paragraph) that neural stem cells can be reprogrammed more easily than mature fibroblasts. In general, what are the molecular factors at work that make one kind of cell easier to reprogram than another? I know you can conceptualize the 'mountain' in state space as being higher, but what does that really mean biologically?
3. PRESENTATION:
The presentation should follow the logic of the paper. First, some facts about stem cell differentiation and cell reprogramming are presented, and then the model is presented (in elementary terms, with balls rolling around on hills and in valleys) and it is shown how each of those facts is reflected in the model.
4. THOUGHTS:
This is a neat paper. Is Huang a theoretical biologist? Or more generally, is that even a thing? I thought there weren't dedicated theoretical biologists like there are in physics.
0.Knew: I knew that cell behavior is commonly explained by causal relationship and diagrams and that these diagrams do not accurately describe the principles of the dynamic system it truly is. I knew from another article assigned (I read these “out of order”) that attractor states referred to the equilibrium states of gene regulation networks.
ReplyDelete1.Learned: I did not know before reading this that murine adult dermal fibroblasts had been reprogrammed into pluripotent stem cells that resembled embryonic stem cells by the overexpression of only four transcription factors. I learned how to describe a research surprise. I learned that the pluripotent and self-renewing states of embryonic stem cells were default states that required no maintenance. I learned that chromatin-modification is usually dynamic so that each gene can reversibly be switched on and off. I learned that the term “gene expression profile” referred specifically to how active proteins encoded by the genome are expressed. I learned that constraints are imposed on gene behavior by the regulatory network and that certain cell types correspond to attractor states.
2.Pressing Questions: “Why can cells not simply alter their expression profile to morph from one cell type to another?” Wouldn’t this be energetically unfavorable? Also, is it likely that the needs of the body would demand such constant changing of cell types after differentiation? (p.547)
“…a specific, biologically meaningful pattern of gene expression requires that the expression of an individual gene ‘considers’ the expression status of other genes.” Is there a way to measure or study how genes “consider” other genes? Or is this referring to such things as cell signaling? (p.549)
3.Presentation: A review/brief explanation of cell potency (pluripotent, multipotent, etc.) and an explanation of the figures shown would be helpful.
4.Thoughts: I appreciate how thought the topics discussed were a little challenging, the author did not use much technical vocabulary in his explanations. I always think figures and included pictures are a plus. I did get lost a few times, but overall, this was an interesting read.
0. KNEW
ReplyDeleteI was aware of attempts to reprogram differentiated cells into pluripotent stem cells in a variety of methods.
1. LEARNED
The difficulty in reprogramming cells is explained by epigenetics. Recent reprogramming involves only overexpression of four key transcription factors, and its been shown that the pluripotent and self-renewing state of embryonic stem cell is a ground state.
I learned that 'ad libitum' equates to 'ad lib.'
Epigenetic barriers are a naturally emerging property of a gene regulatory network.
State vector S(t): gene expression profile of a particular combination of 25k genes with a particular expression status. State vectors are unique for each cell type m. Transcriptomes are a good approximation of Sm. Majority of possible S are forbidden by gene regulatory networks; act as 'barriers'. S moves until a stable state that satisfies network interactions are found. Attractor states: stable equilibrium states, 'sinks' in 'flow field'.
Work to quantify how different each Sm is from each other: high-dimensional distance, related to sum of Euclidian distance (sum(xim-xin)^2 for all i).
Hierarchical clustering based on Euclidian distances produces dendrograms. Transdifferentiation between neighboring branches is readily achieved by fewer overexpression of fate-determining TF.
Epigenetic markers: gene silencing via DNA methylation and covalent histone modifications. Nonspecific. Who writes the 'histone-code'?
2. QUESTIONS
Are there ways to take into account post-transcriptional cellular processes when using transcript profiling techniques?
3. PRESENTATION
The math and logic involved in this paper is really interesting and worth talking about.
4. THOUGHTS
A well written and eloquent paper. I never really considered the question, "why can cells not simply alter their expression profile to morph from one cell type to another?"
Class 11, Assignment 11a
ReplyDeleteReprogramming cell fates: reconciling rarity with robustness
(0) Knew:
I knew about the process used to generate induced pluripotency, that stemness represents a ground state in the context of cell fate regulation, and that modulating gene expression is closely related to cell fate determination. From the previous paper, I was familiar with the idea of cell fates as attractors.
(1) Learned:
I learned about how the pluripotent state can be reprogrammed in spite of covalent chromatin marks on genomic DNA, and how these marks can be thought of as dynamic rather than immutable. I learned about the simplistic notion of Euclidean distance as a measure of expression alteration between cell states, and the limitations inherent in such a calculation for cells that develop discretely rather than linearly. I learned about the interpretation of attractor states as sinks in a flow field, and that any developmental trajectories are beholden to this topology.
(2) Pressing Questions:
Is multipotency an attractor state but not a ground state? It is certainly interesting that when perturbed, multipotent cells will stochastically differentiate into any of their succeeding cell types, while variations in gene expression levels are much better tolerated in pluripotent cells. What heterogeneity is the paper referring to in describing a “wash-board potential”?
(3) Presentation Topic:
I could do with a run-through of the figures in this paper.
(4) Thoughts:
This was an interesting paper. Looking forward to seeing how this discussion proceeds in class.
0. KNEW:
ReplyDeleteI knew about differentiation. I knew about stem cells and the induction of pluripotent stem cells. I knew about state vectors from the previous paper. I understood the idea of distance between cell states before reading this paper. I knew about attractor states but I didn’t realize that they could be thought of as ‘sinks’ in a topology. This clarifies the importance of them as you can reach the same point from an opposite trajectory.
1. LEARNED:
I learned about the concept of an epigenetic landscape. I learned exactly what a transcriptome is. I learned how methylation is dynamic and how epigenetic markers are not the best model for explaining cell regulation.I learned the term ‘epigenetic barriers’ and how this reduces the amount of configurations the state of a cell can be in. I learned that multiple attractors can be present in a system and a way to think of this intuitively.
2. PRESSING ?:
How can the ease of moving up the epigenetic landscape be modeled through just statistics?
3. PRESENTATION:
Definitely get topographical pictures. The whole presentation could one use those.
4. THOUGHTS:
Very nice paper for giving a mental model for gene expression states. Would have been nice to read before the previous paper
0 Knew
ReplyDeleteAs a BME I am familiar with the basics of cell and tissue engineering. I am also familiar with epigenetics as it relates to cancer biology.
1 Learned
It was interesting to consider pluripotency as the cell's "ground state," because as Huang noted, this is contrary to the robustness and complexity of cell fate regulation. The semantics of the "epigenetic landscape" and use of transcriptomes to approximate S(m) was also interesting.
2 Pressing
Doesn't considering pluripotency a ground state imply low epigenetic barriers? At t=0 maybe, I think the time component of this analogy should be addressed (as it is with the discussion on epigenetics).
3 Presentation
Mathematical explanation of PCA
Transcriptome analysis/RNAseq
4 Thoughts
I appreciated the figures and thought this might have been helpful before the WGCNA talks last week.