Top

iBiology

Bringing the World's Best Biology to You

Computational Modeling of DNA Communication in the Nucleus

Part 1: Wang Xi, PhD interview

00:00:05:20 - Hi I'm team wing and I'm with the science communication
00:00:08:06 lab.
00:00:09:02 We just got done filming our SYR series
00:00:11:20 or share your research.
00:00:12:18 And we're here today to talk to Wang. Wang
00:00:15:01 could you introduce yourself?
00:00:17:06 - Hi I'm Wang.
00:00:18:18 I'm a 50 year PhD student in John Hopkins University.
00:00:23:16 Um I come from China and now my research is
00:00:27:19 about computational regulatory economics
00:00:30:14 which is applying mathematic
00:00:32:17 and statistical tools to understand biological data
00:00:36:18 and build models to understand the mechanisms in biology.
00:00:42:03 - How did you get interested
00:00:43:08 in this intersection of math and biology?
00:00:46:12 - So I've always been interested in quantitative science
00:00:50:17 and especially mathematics and but then I figure
00:00:55:06 out I'm not smart enough to be a mathematician.
00:00:58:22 So I got into interested
00:01:00:17 into um things like physics chemistry including biology
00:01:05:23 and there's just there are just so many unanswered question
00:01:09:16 in biology right now.
00:01:11:14 So it became very natural
00:01:13:15 for me to combine uh those two things together
00:01:16:18 to make new discoveries.
00:01:18:15 - What kinds of questions are you curious about?
00:01:21:01 - So I'm particularly fascinated by DNA.
00:01:25:12 So DNA as you all know is the most essential
00:01:28:22 genetic material.
00:01:30:10 And there's just so rich information in our DNA.
00:01:34:22 So for example for us as human there are 3 billion bases
00:01:40:00 in each of us in each of ourselves and we have
00:01:45:20 over 20 K genes.
00:01:49:10 So there must be a complicated machinery that ensures
00:01:53:16 everything is going smoothly inside this tiny needle
00:01:57:17 nucleus to make sure all the genes are being properly turn
00:02:02:08 on or turn off at the right timing.
00:02:06:01 And but how it is done right?
00:02:08:21 This is a super complicated process.
00:02:11:05 And we're in the middle of the process to figuring out
00:02:15:20 - You were talking a little bit
00:02:17:09 um about the different types of DNA.
00:02:20:01 And you mentioned before when we were talking
00:02:22:21 that you also think that there's an interesting difference
00:02:25:07 about the genetic information in different cells.
00:02:28:11 What's curious to you about that?
00:02:30:12 - So this has been a longstanding question in biology
00:02:34:15 as you all know we are multi cell organisms.
00:02:39:03 There are not not millions but over billions
00:02:42:23 of cells in each of in each of our human body
00:02:46:08 but all of those cells are coming from one single zygote
00:02:52:00 which means that each of those cells
00:02:54:07 in our body has nearly identical genetic information
00:02:58:20 but there are over 200 cell types.
00:03:01:06 I mean if you look at a skin cell
00:03:03:05 it looks just so different
00:03:04:18 from a blood cell or a muscle cell or maybe a bone cell
00:03:09:16 in terms of either their shape their function
00:03:13:12 or the genes pro or the gene expression profile of them.
00:03:18:00 So how did the question I'm interested
00:03:19:22 in is how did the same genetic information
00:03:23:14 in each cell generates all those different cell types?
00:03:28:17 And the the answer we think is called epigenetics.
00:03:34:10 This is something going on on top of DNA.
00:03:37:05 So the DNA are composed of those all those spaces
00:03:40:00 A D C G and those information are consistent across
00:03:43:06 cells
00:03:44:06 but there could be all those different types
00:03:46:18 of modifications on the chromatin that determines
00:03:50:18 cell type specific gene expression.
00:03:53:20 For example there could be methylation on those spaces.
00:03:58:11 There could be modifications on the protein
00:04:02:08 which DNA wraps around
00:04:04:21 Um...
00:04:05:16 and there could also be the 3D interaction um
00:04:09:01 between this toe between distant DNA elements
00:04:14:05 and which is what I talked about in my SYR video.
00:04:17:12 - Yeah. And what motivated you to get
00:04:19:23 into science and work on DNA?
00:04:22:18 - I think it's just such a fascinating field that to figure
00:04:27:12 out how DNA information is being encoded and decoded right?
00:04:33:09 Because ultimately all of our behavior
00:04:36:12 all of our appearance what we're doing right now
00:04:40:04 to this world is all determined
00:04:42:05 or partially determined by our DNA.
00:04:45:19 Right. And and of course
00:04:48:00 there's definitely interaction between the genetic material
00:04:51:17 between the environment um but genetic material definitely
00:04:55:23 without any doubt plays a major role in it.
00:04:58:19 So um the dream I have is like
00:05:01:22 if you really wanna understand human beings
00:05:04:15 you have to study their DNA.
00:05:06:19 And that is my initial motivation to study how DNA works
00:05:10:21 and how RNAs are being transcribed and being regulated.
00:05:15:23 - So getting back to basics yeah.
00:05:19:09 And interpreting
00:05:20:06 - Yeah absolutely.
00:05:21:10 I'm definitely yeah.
00:05:23:10 I'm definitely a reductionism support
00:05:27:02 of reductionism to get things to the very basic
00:05:31:07 and to make them simpler to understand them.
00:05:34:04 - What is one thing that you want other people to know
00:05:36:21 about being a scientist?
00:05:39:01 - I think being a scientist means
00:05:41:03 that you are really really lucky.
00:05:43:19 That's just my own opinion because I think as a scientist
00:05:47:20 you're given the you're given the chance or you're
00:05:50:12 given the privilege to do things that can look less useful.
00:05:55:07 You don't have think about um like how can I apply this
00:05:59:20 to the real world in the very beginning of your research?
00:06:02:22 For example for me
00:06:03:23 I'm just interested in the information in DNA.
00:06:06:13 I think that's the very very
00:06:07:17 most basic thing in my opinion.
00:06:10:00 So I got interested in it is just
00:06:12:18 just like how Einstein got interested in time and light.
00:06:17:09 And that's how he dis- how he discovered the
00:06:20:07 the the general relativity.
00:06:22:12 Right? Those are the things that we see every day
00:06:26:08 but just because we have seen them so many times
00:06:29:17 sometimes we just get used to it and ignore it and forgot
00:06:32:15 that we actually don't understand them you know as we're
00:06:36:22 we're all living organisms
00:06:38:08 but we don't actually don't understand there.
00:06:41:22 99% of ourselves are the things we don't understand.
00:06:46:15 - Yeah.
00:06:47:11 So yeah those are really really basic questions.
00:06:48:22 And when you ask me how is
00:06:51:04 is it useful to understand those questions?
00:06:53:12 I mean I don't know.
00:06:54:18 And I mean if you if I'm asked to write a grant
00:06:58:07 write a proposal about it maybe I can definitely think
00:07:00:19 of one or two real world applications.
00:07:03:17 There could be millions of applications
00:07:05:19 if those problems are being solved but what's
00:07:07:23 really motivating me at the very beginning to
00:07:10:04 solve them is not the real world application is just
00:07:12:23 because I'm interested in those um mysteries of nature.
00:07:18:10 And I think
00:07:19:16 as scientists um we're we're given the chance to
00:07:24:05 make discoveries that have uh more long lasting impacts.
00:07:29:15 And I think that's worth
00:07:31:08 that's what worth doing for my life.
00:07:34:21 - Thank you so much.
00:07:35:22 It's so been so nice to talk to you
00:07:37:20 and hear your stories and the mysteries that excite you.
00:07:40:17 I hope you all go and check out.
00:07:42:10 Wang's talk.

Part 2: Computational Modeling of DNA Communication in the Nucleus

00:00:00:06 (light music)
00:00:12:16 - Hi my name is Wang.
00:00:14:16 I'm a PhD student from Mike Beer's group
00:00:16:23 from Johns Hopkins University.
00:00:19:13 Today I'm going to talk about
00:00:21:03 DNA communication in the nucleus.
00:00:23:21 So let's first talk about DNA
00:00:25:21 which is the essential genetic material
00:00:28:19 that determines our appearance
00:00:30:15 and behaviors as living organisms.
00:00:34:05 So what does DNA look like?
00:00:36:07 You might think of it as a strand of double helix
00:00:39:11 as Watson and Crick are also thinking very hard about here
00:00:43:13 or you might also think of
00:00:45:05 the famous x-ray diagram from Rosalind Franklin.
00:00:49:08 But if you ever looked at the nucleus
00:00:51:20 of an interphase cell under a microscope
00:00:55:08 the chromatin of it looks just like a hairball.
00:00:58:16 And in fact in each human cell
00:01:01:04 a total of two meter long DNA which is the length of a bed
00:01:05:10 is packed inside this one-micron tiny little nucleus.
00:01:10:20 So it is super crowded there
00:01:13:12 and instead of acting independently
00:01:16:02 there are always extensive interactions
00:01:18:19 and communications across different parts of the chromatin
00:01:22:22 and this complex 3D structure
00:01:25:11 is also known as chromatin organization.
00:01:28:21 So chromatin organization has
00:01:30:15 important implications in common diseases.
00:01:34:08 For example when normal cell becomes a malignant tumor cell
00:01:39:07 not only its shape become irregular
00:01:42:03 but also the chromatin organization
00:01:44:07 gets loose and disrupted.
00:01:47:01 And my research interest
00:01:48:12 is to find out the pattern and mechanism
00:01:51:15 behind the seemingly random chromatin organization.
00:01:55:03 And personally I'm interested in figuring out
00:01:58:01 the rules in biology.
00:02:00:14 In other fields for example
00:02:02:20 mathematicians have their formulas
00:02:05:07 and physicists have their theories.
00:02:08:01 The formulas and theories could be very simple and powerful
00:02:12:07 and in biology things are usually a bit more complicated.
00:02:16:22 But in this case I do believe
00:02:18:23 that there must exist some rules
00:02:21:18 because the chromatin organization
00:02:24:04 are tightly associated with regulation of gene expression.
00:02:28:20 Recalling from the central dogma
00:02:31:00 which is probably the most famous rule in molecular biology
00:02:35:16 it says DNA is transcribed into RNA
00:02:39:11 and then translated into proteins
00:02:42:05 but none of them are simple one-step processes.
00:02:46:05 There are lots of regulations
00:02:48:04 to ensure that the right amount of RNA
00:02:51:02 and proteins are being generated at the correct time.
00:02:55:14 The expression level of the same gene
00:02:57:16 across cell types can be dramatically different
00:03:01:02 and in our genome
00:03:02:11 the coding region of genes only constitutes 2% of it.
00:03:06:16 The remaining 98% of our genome
00:03:09:15 contains a lot of regulatory elements.
00:03:12:16 For example here for RNA transcription
00:03:16:01 when the gene needs to activated during development
00:03:19:05 in a specific cell type
00:03:21:03 it is going to spatially interact with a distal DNA element
00:03:25:13 called enhancer that controls its expression.
00:03:29:07 The activity of enhances are highly cell type specific
00:03:33:04 and to help enhancers
00:03:34:20 find their target genes faster or easier
00:03:38:05 a pair of protein called CTCF form a loop with each other
00:03:42:21 to bring the enhancer and gene inside them into proximity
00:03:46:22 and therefore activating gene expression.
00:03:50:13 So CTCF is the type of barrier protein
00:03:53:18 that acts as loops anchors
00:03:55:19 at the boundaries of active transcription domains.
00:03:59:12 As you can see
00:04:00:17 CTCF loop is a critical structure unit
00:04:03:15 of mammalian chromatin organization.
00:04:06:16 And what I'm trying to understand here
00:04:08:22 is how these CTCF loops are formed.
00:04:13:16 This is a critical problem
00:04:15:13 because not all the CTCFs
00:04:17:13 bound to the chromatin can interact.
00:04:20:18 There are around 50K CTCF binding sites in the genome
00:04:24:21 which corresponds to over one million CTCF pairs.
00:04:28:17 However only less than 5% of them
00:04:31:15 are identified as interacting
00:04:33:21 or looping by experiment.
00:04:36:12 So is there a generic rule for CTCF interaction?
00:04:41:08 Well we're not completely sure about that
00:04:44:19 but we do have
00:04:45:15 a compelling hypothesis called loop extrusion.
00:04:50:00 It says that a ring-shaped protein cohesin
00:04:52:21 is loaded onto the genome
00:04:54:16 in between a pair of CTCF binding sites
00:04:57:02 and due to its motor activity
00:04:59:14 cohesin translocates along the chromatin fiber
00:05:02:13 and extrudes DNA out of it
00:05:05:11 and cohesin gets blocked when it meets CTCF at both ends.
00:05:11:22 So there has been in vitro experimental evidences
00:05:15:15 that validates the loop extrusion model.
00:05:18:22 In this example both ends of a piece of DNA
00:05:22:08 is tethered on the surface.
00:05:24:19 And when cohesin and cofactors
00:05:27:00 are put together into the medium with the DNA
00:05:30:05 single-molecule imaging shows that a loop is generated
00:05:33:22 and enlarged in real time from this U-shape DNA
00:05:37:21 as the white arrow is pointing at here.
00:05:41:08 This direct visualization provides very strong evidence
00:05:44:21 of the loop extrusion model
00:05:47:11 but there are two remaining questions.
00:05:51:03 The first one is how do we validate
00:05:53:21 this loop extrusion model for all the CTCF loops?
00:05:57:19 It is extremely difficult to do so experimentally
00:06:01:16 simply because there's just so many CTCF loops
00:06:05:08 and we're not able to trace them
00:06:07:04 all at once at the same time.
00:06:10:04 And the second is can we use the loop extrusion model
00:06:13:15 to explain why some CTCFs are interacting
00:06:17:08 and the others are not
00:06:19:01 or the CTCF interaction specificity?
00:06:22:09 So the way I come up with to answer these two questions
00:06:26:00 is to develop a mathematic model of loop extrusion
00:06:29:15 and then apply it to predict CTCF interaction specificity.
00:06:34:09 If the predictions from the loop extrusion model
00:06:37:08 are consistent with experimentally detected CTCF loops
00:06:41:11 then the model itself also gets validated.
00:06:45:03 Understanding the formation of CTCF loops
00:06:47:21 again will add to our knowledge
00:06:50:11 of a mammalian chromatin organization
00:06:52:21 as well as transcriptional regulation
00:06:55:14 and of course with applications
00:06:57:20 to developing treatments of genetic diseases.
00:07:02:07 So to do that let's first revisit the model.
00:07:05:13 A couple of steps needs to be done
00:07:07:12 before a loop can be formed.
00:07:09:17 First we have two CTCF binding motifs
00:07:12:19 that are in convergence orientation towards each other
00:07:15:22 because this is a preferred configuration.
00:07:18:20 And second CTCF protein dynamically binds
00:07:22:07 on those two binding sites.
00:07:24:12 And third cohesin is loaded
00:07:27:04 and extrudes the DNA for a long distance.
00:07:30:15 And last but not least
00:07:32:06 an existing loop can block the formation
00:07:34:19 of a neighboring loop in 3D space.
00:07:38:03 After breaking this model down
00:07:40:00 we can use the language of math
00:07:42:00 to describe each of these processes.
00:07:44:20 And what we're doing here
00:07:46:03 is to predict interaction probability
00:07:48:13 between a pair of CTCF binding sites.
00:07:52:00 First a scalar W is used as the contribution
00:07:55:19 of motif orientation to loop stability.
00:07:59:04 It represents how much
00:08:00:12 this convergent orientation is preferred
00:08:03:07 and a bigger W means that this orientation
00:08:05:23 is more preferred than tandem or divergent orientation.
00:08:12:08 The CTCF binding intensity is described
00:08:15:05 by the binding chemical equilibrium equation.
00:08:18:21 It depends on local CTCF concentration
00:08:21:21 which can be different among CTCF binding sites
00:08:25:01 as well as the dissociation constant Kd.
00:08:30:00 The activity of cohesin also depends on distance
00:08:33:09 as cohesin can stochastically fall off when translocating
00:08:37:01 along the chromatin fiber as distance gets larger
00:08:40:19 so we used an exponential decay term to describe it.
00:08:45:20 And we also considered
00:08:47:02 all neighboring CTCF binding sites for loop competition.
00:08:52:01 So our final probability of any CTCF pair
00:08:55:21 interacting with each other is given by the product
00:08:59:09 of those four components.
00:09:01:07 Orientation binding intensity distance
00:09:04:11 as well as loop competition.
00:09:06:13 And among those CTCF pairs
00:09:09:01 all four components can be different
00:09:11:02 and that can result in the difference
00:09:13:00 in loop stability or interaction probability.
00:09:17:07 And CTCF pairs with high interaction probability
00:09:20:23 are predicted to be CTCF loops.
00:09:23:23 We then compare those predicted CTCF loops
00:09:26:18 with those actual loops that are experimentally detected
00:09:30:02 to see how accurate our predictions are.
00:09:35:10 And the experimental data
00:09:36:19 are coming from a technique called ChIA-PET
00:09:40:05 or Chromatin Interaction Analysis of Paired-End Tag.
00:09:44:23 It belongs to a class
00:09:46:02 of chromatin conformation capture assay
00:09:49:07 and in brief the chromatin is cross-linked
00:09:52:00 using paraformaldehyde and sonicated
00:09:55:09 and then chromatin complexes are put down
00:09:58:10 using a specific antibody
00:10:00:13 against a particular protein factor.
00:10:03:07 For example CTCF here.
00:10:06:03 Then DNA from distinct genomic loci
00:10:09:09 are ligated into the same fragment.
00:10:14:20 And it allows us to use sequencing to detect DNA fragments
00:10:18:08 that are linearly apart but spatially close
00:10:21:15 and each of those arcs is an output of ChIA-PET
00:10:24:22 that represents a spatial interaction or a loop.
00:10:29:10 So here's an example
00:10:30:14 of relatively complicated chromatin organizations.
00:10:34:12 Each of those peaks on the first track
00:10:37:00 is a CTCF binding event
00:10:39:14 and each of those arcs from the second track
00:10:42:06 is a loop detected by ChIA-PET.
00:10:45:07 The distance component of our model
00:10:47:20 is simply linear genomic distance
00:10:50:13 between each of those peaks
00:10:52:10 and any overlapping loops
00:10:53:23 are in loop competition with each other.
00:10:57:01 As you can see there are many overlapping and nested loops
00:11:01:02 so this is quite complicated.
00:11:03:05 However most of those loops
00:11:05:01 are correctly predicted by our computational model
00:11:08:09 as indicated on the third track.
00:11:12:07 We then applied the comparison to the whole genome
00:11:15:00 and looked at each individual features of our model
00:11:18:12 as well as combinations of them.
00:11:20:23 And the evaluation metric we used here
00:11:24:01 is an accuracy score called AUPRC
00:11:28:03 which is between zero and one.
00:11:30:22 As we can see none of the four features
00:11:33:11 alone can accurately predict CTCF interaction specificity.
00:11:38:22 But when we combine them with each other
00:11:41:15 the accuracy increases
00:11:43:22 and the best model is given by the one
00:11:46:06 composed of CTCF binding intensity orientation
00:11:50:11 as well as loop competition.
00:11:53:08 The best model reached AUPRC about .6
00:11:58:22 which is quite good considering
00:12:01:04 it is a very difficult task
00:12:03:01 to distinguish 100K CTCF loops
00:12:06:05 from over one million non-interacting CTCF pairs.
00:12:11:03 And the result means that loop extrusion model
00:12:14:04 can indeed accurately predict CTCF interaction.
00:12:18:03 So this model itself
00:12:19:16 is also likely to be the true underlying mechanism
00:12:22:22 for CTCF loop formation.
00:12:25:18 And among those four features
00:12:27:19 loop competition and distance
00:12:30:00 are highly correlated with each other.
00:12:32:15 This is very natural to understand
00:12:34:20 because as distance gets larger
00:12:36:23 the number of competing loop also increases.
00:12:40:06 But to further understand this model
00:12:42:11 we wanna know which one of them is the causal factor
00:12:45:19 that really determines loop formation probability.
00:12:51:14 And the way we do it
00:12:53:00 is like whenever we have two free variables
00:12:56:02 that we wanna study
00:12:57:22 we fix one of them and then study the other.
00:13:01:16 So for each interacting CTCF loop we first sampled
00:13:06:10 a distance-matched non-interacting CTCF pair.
00:13:10:13 In other words in between those two datasets
00:13:14:02 binding intensity orientation and distance
00:13:17:02 are all being matched.
00:13:19:01 As you would imagine
00:13:20:14 it would be much more difficult to do classification here
00:13:24:04 because loop competition is the only feature
00:13:26:20 that you could use.
00:13:28:19 But with loop competition alone
00:13:30:20 we still reached accuracy about .75
00:13:34:11 meaning that about 75% of the cases
00:13:38:04 are being correctly classified
00:13:40:20 and that indicates there are
00:13:42:16 some unique information to loop competition
00:13:46:00 that determines CTCF interactions specificity.
00:13:51:00 However when we do the same thing
00:13:53:12 to control loop competition
00:13:55:16 we found that distance itself is not predictive at all.
00:13:59:23 Therefore we conclude that loop competition
00:14:03:01 is the more important component than distance in this model.
00:14:07:15 And finally we use the predicted CTCF loop
00:14:10:16 to show that enhancers and promoters
00:14:13:07 are more likely to contact with each other
00:14:16:02 when CTCF interaction is strong.
00:14:19:04 On the figure on the right
00:14:21:00 on the x-axis are the contact frequency
00:14:23:13 of CTCF loops from weak to strong
00:14:26:14 and on the y-axis are the enrichment of interactions
00:14:30:03 between enhancers and promoters inside those CTCF loops.
00:14:34:17 And this result implies that gene expression
00:14:37:16 is indeed regulated by chromatin organization
00:14:41:04 and facilitated inside CTCF loops.
00:14:45:04 So to summarize we developed a mathematic model
00:14:48:20 that is able to accurately predict CTCF interaction
00:14:52:21 based on the loop extrusion process.
00:14:56:00 We found that loop competition is a more important component
00:14:59:14 in this model than distance
00:15:01:21 and we also validated the role of CTCF loops
00:15:05:15 in constraining active transcription domains.
00:15:09:10 This study shows that mathematic modeling
00:15:12:08 can be really powerful
00:15:13:14 in understanding biological mechanisms.
00:15:17:05 We expect our loop extrusion computational model
00:15:20:10 to be useful for further explorations of both the features
00:15:24:23 and mechanisms of chromatin packing
00:15:27:14 as well as its impact on gene regulation.
00:15:31:06 I wanna thank everyone from the Mike Beer group
00:15:34:00 for their support of my work
00:15:35:17 as well as my funding resources
00:15:37:23 includes Johns Hopkins University
00:15:40:02 as well as the Encode Consortium thanks for listening.
00:15:55:00 (light music)

Videos in this Talk
  • Part 1: Wang Xi, PhD interview
    Part 1: Wang Xi, PhD interview
    Audience:
    • Student
    • Researcher
    • Educators
    • Educators of Adv. Undergrad / Grad
  • Part 2: Computational Modeling of DNA Communication in the Nucleus
    Part 2: Computational Modeling of DNA Communication in the Nucleus
    Audience:
    • Student
    • Researcher
    • Educators
    • Educators of Adv. Undergrad / Grad
Speaker: Wang Xi
Total Duration: 1:24:22
Recorded: May 2022
All Talks in Genetics and Gene Regulation

Talk Overview

What do chromosomes look like in our cells? A total of two meters of DNA is packed inside a 1 micrometer tiny nucleus of each cell. How is this condensation achieved? In this 2022 Share Your Research Talk, Wang describes his research to build a mathematical model for a process called Loop Extrusion, which describes how DNA folds and forms loops at kilobase scale. This model helps us understand complex interactions between DNA fragments as well as their 3D structure. The model can be used to predict how small changes in DNA sequence disrupts higher order chromatin architecture and has implications in common genetic diseases like cancer and Alzheimer’s disease.

Speaker Bio

Wang Xi

Wang Xi

Wang Xi, PhD completed his doctoral program at Johns Hopkins University. At the time of filming, Wang was a PhD candidate and his thesis research focused on chromatin interactions and their implications on gene regulation through computational modeling. Wang seeks to continue doing interdisciplinary research as a postdoc at Harvard University. Continue Reading

Playlist: Share Your Research (formerly YSS)

Related Resources

  • Xi, W., Beer, M.A. Loop competition and extrusion model predicts CTCF interaction specificity. Nat Commun 12, 1046 (2021). https://doi.org/10.1038/s41467-021-21368-0
  • Xi, W., Beer, M.A. Local epigenomic state cannot discriminate interacting and non-interacting enhancer–promoter pairs with high accuracy. PLoS Comp Bio (2018). https://doi.org/10.1371/journal.pcbi.1006625

Credits

Director: Thi Nguyen
Producer: Shelley Carter
Videographer: Eric Kornblum
Editor: Rebecca Ellsworth, Jiefei Yuan
Original Music: Marcus Bagala
Graphics: Chris George, Maggie Hubbard

Reader Interactions

Leave a Reply

Your email address will not be published. Required fields are marked *