More details to come.

A recurring theme in evolution is how existing components in a biological system can be modified or reused for another context or function. Take, for example, signaling pathways. In complex organisms, such as humans, there are thousands of processes that require coordination via signaling, such as developmental programs, immune processes, homeostatic regulation, etc. Yet, signaling in these organisms has evolved so that remarkably few of the highly conserved signaling pathways govern multitudes of aforementioned processes. Part of the ongoing question in biology involves how a single pathway is able to support such a variety of processes.

One could imagine a number of ways this could be made possible: contexts outside the signaling pathway, such as cell state or epigenetic modification of the genome, could direct signaling to take effect in a specified manner. Another way is through introducing layers of complexity in the signaling pathway itself. Adding complexity to how a signal is transmitted or transduced can broaden the information capacity and enable the signaling to be used in more contexts. For example, signaling systems often have multiple ligands and receptors, some of which are only used for specific purposes or create a unique signaling response. In our study, we discovered how the canonical Wnt pathway has implemented a layer of complexity through the use of two distinct but opposing types of transcriptional elements that respond to a single signal, allowing rich variation of transcriptional output from a monotonic input (Kim, Cho, Hilzinger, et al., 2017).

The canonical Wnt pathway works through a messenger protein called β-catenin, which is in constant flux of synthesize and rapidly degradation in a cell. Signal activation by Wnt proteins inhibits degradation of β-catenin, which allows the protein to accumulate and localize to the nucleus with its binding partner, Tcf. Traditionally, it is understood that the Tcf/β-catenin complex binds to a specific DNA element- aptly called the Wnt-Responsive Element (WRE), recruits transcription factors, and activates gene expression.  Barring crosstalk and feedback mechanisms, this signal transduction only allows monotonic response to signaling by Wnt proteins: β-catenin increase leads to activation of a target gene.

However, we’ve discovered another DNA element that recruits β-catenin/Tcf  to suppress gene activation, named 11-bp Negative Regulatory Element (11-bp NRE), in a recently published work that was began with Lea’s groundwork in Marc Kirschner’s lab, and completed with major efforts from Jae Cho, Tom Hilzinger, and several others from our own lab,. We have identified these 11-bp NREs in Wnt target genes from humans to mice to frogs, and demonstrated that they play important roles in regulation of these genes.

The fascinating part about the discovery of the 11-bp NREs is that they are located closely to the traditional WREs in a gene’s promoter, and work in tandem in regulating gene expression. Here, we have two elements that both recruit β-catenin, one activating a gene and the other suppressing it.

How does this paradoxical interaction work to regulate genes? This type of regulation is actually found frequently and well characterized as a network motif in biology, identified as Incoherent Feedforward Loop (IFFL). Modeling simulations by Harry Nunns in our lab demonstrate that the IFFL composed of the two elements in the Wnt pathway can generate a diverse array of transcriptional responses, such as pulsing, oscillations, bistable response, or fold change detection.  Thus, presence of two distinct elements to transduce information by b-catenin allows versatility in how the Wnt signaling pathway controls the response of its target genes, which suggests a layer of mechanism for how the Wnt signaling pathway can function in various contexts to produce a rich variety of dynamic responses.

— Kibeom Kim

Reference: Two element pathway in the Canonical Wnt Pathway. (Current Biology, Volume 27, Issue 15, p2357–2364.e5, 7 August 2017)

Cover photo designed by Andrew Liu

In this study, we asked a simple question: how effectively does a single cell regulate its messenger protein in response to signal? We looked specifically at a messenger protein in the Tgf-β/Smad pathway known as Smad3. When a cell encounters Tgf-β protein (the signal), Smad3 (the signal-messenger) within that cell shuttles into the nucleus and alters expression of genes. Strikingly, when stimulated with Tgf-β signal, cells did not adjust the level of Smad3 to be equal in all cells. In fact, they don’t even come close, with some responses being up to 40 times stronger than neighboring cells (even though these cells were exposed to the same strength of signal).

But all this means is that cells do not effectively regulate their messenger protein in the way we thought they might. We next wondered whether cells might regulate their messenger protein (Smad3) in a different way, a way we did not expect? We addressed this question by making time-lapse movies of the signaling process, using fluorescently-labeled Smad3. The movie data showed that while cells increase Smad3 to varying final levels after signal, the ratio of Smad3 level after signal divided by the level before signal is precise across cells. This can be seen in the video below, where the four cells increase Smad3 by ~3 fold. The before/after ratio of Smad3, or fold-change, is roughly 4 times more precise than the final level of Smad3, and increases the cells’ ability to accurately measure the strength of signal in their environment by about 5-fold.

Movie of mouse myoblast cells expressing fluorescent Smad3 responding to Tgf-β stimulation show precise fold-change. Upon exposure to Tgf-β, Smad3 is shuttled to the nucleus (circled in red) in each cell. Real-time quantification of Smad3 abundance and Smad3 fold-change in the nucleus of the four tracked cells is shown on the right hand side.  Frames were taken every 4 minutes.

This result led us to ask a significant, specific question: do cells actually compute this ratio, dividing the level after signal by level before, and use that to control gene expression? We tested this directly in individual cells by measuring fold-change (using movies) and then measuring gene expression in cells (using a technique with a really long name: single molecule RNA fluorescent in situ hybridization). Testing for correlation, we found gene expression showed no correlation with the abundance of Smad3, and showed clear dependence on the fold-change of Smad3.

Understanding how signaling pathways function is a central goal in understanding the biology of humans and animals. We have shown in this work that Tgf-β signal is transmitted in Smad3 fold-change. Perhaps one of the most important implications of this work is that in order to understand signaling in a tissue (say a tumor, for example) it is insufficient to simply measure the total level of Smad3, one instead needs to measure the Smad3 response over time. Further, this study also adds to the growing list of pathways where fold-change detection has been proposed, suggesting that fold-change detection may be a widely implemented strategy in animal signaling pathways.

— Christopher Frick, 2017

Reference: Sensing relative signal in the Tgf-β/Smad pathway, PNAS 114 (14): E2975–E2982, 2017

We sought to study a particular instance of elegant design in biology, that of logarithmic sensing. This phenomenon, where a cell’s response is proportional to the ratio of change in signal as opposed to the absolute difference, is observed across a wide spectrum of biological contexts. This is most apparent in human sensory systems, where we measure our perceived intensity of sound in the logarithmic unit of decibels and our visual system is more sensitive to contrast than to absolute changes in brightness. Essentially we wanted to understand how this sort of behavior, observed in both large-scale cognitive system and in the response of single cells, is encoded at the molecular level.

We found that a large class of proteins, those that are allosterically regulated, are particularly well suited to perform logarithmic sensing. Allosteric proteins are biological molecules that have two or more different conformational states, each of which have a different biological function. A classical example is hemoglobin, the oxygen transport molecule found in almost all vertebrate life. In 1904, Christian Bohr (father of Niels Bohr) observed that hemoglobin would change its affinity for oxygen when he varied the amount of carbon dioxide in the medium. It turns out that high levels of CO2 would bias the hemoglobin towards a low affinity state, and low levels would bias it towards a high affinity state.

It turns out that the way in which the binding kinetics of allosteric proteins shifts in response to effectors, like CO2 for hemoglobin, is naturally logarithmic. That is to say the activation curve has a regime that looks linear when plotted on a logarithmic scale and can be shifted without changing its overall shape. We show that this behavior fits nicely into a known architecture for Fold-Change Detection, where a logarithmic transformation is coupled with negative feedback.

More generally, this work shows how we can think about the biophysical properties of proteins as functional components of sophisticated information processing systems. This moves towards a goal of unifying the low-level properties of molecules with the high-level properties of living systems.

— Noah Olsman, 2017

Reference: Allosteric proteins as logarithmic sensors, PNAS 113 (30): E4423-4430, 2016

Moon jellyfish. Photo credit: Ty Basinger and Michael Abrams.

Some facts about jellyfish according to National Geographic.

1. Jellyfish are favorite meal of sea turtles.
2. In 1991, moon jellies flew aboard the space shuttle Columbia for a study on weightlessness and development of juvenile jellyfish.
3. Although they have no brain, jellyfish have somehow been smart enough to survive for more than 500 million years.

Almost by accident, we began experimenting on the moon jelly Aurelia aurita in the lab.  To our delight, we discovered that jellyfish ephyra, the juvenile jellyfish, displays a unique mode of self-repair.  Check out our manuscript here.

Click image to see movie of a swimming ephyra.

Upon amputation, we observed that instead of regenerating the lost parts, jellyfish ephyra reorganized existing body parts and arms, and regained radial symmetry – all completed within 12 hours to 4 days.

We  call this process symmetrization, to denote the recovery of functional symmetry, rather than regrowing precise parts.

Click image to see movie of a symmetrized tetramer swimming.

About 90% of the ephyrae amputated symmetrized.  Symmetrization facilitates further growth and development, whereas those ephyrae that remained asymmetrical grew abnormally and could not swim very well.

Fascinatingly, we found that symmetrization does not require cell proliferation or cell death.  Rather, it is predominantly driven by mechanical forces generated by its own muscle-based propulsion machinery.  To see how muscle contraction could drive symmetry recovery, we teamed up with Chin-Lin Guo from Academia Sinica.  We put the compressive forces from muscle contraction in the context of the ephyra geometry and viscoelastic response from the mesoglea.  And indeed, by considering these forces generated during propulsion, we can mathematically recapitulate the recovery of global symmetry.

Click image to see mathematical simulation of an ephyra regaining radial symmetry.

Therefore we have here a self-repair mechanism driven by a constitutive physiological machinery.  Forces generated by the propulsion machinery sense the imbalance and drives its own repair.  In addition to Aurelia, we also observed symmetrization in the sea nettle Chrysaora pacifica, the lagoon jellyfish Mastigias sp., and the Mediterranean jellyfish Cotylorhiza tuberculata.

And this is not all!  We are being surprised more and more by incredible phenomena we are seeing in jellyfish.  Stay tuned for more updates soon.

Our friends in this project:

1. The Dabiri Lab at Caltech
2. Kiersten Darrow and her crew at the Cabrillo Marine Aquarium, in San Pedro, California
3. Wyatt Patry at Monterey Bay Aquarium, Monterey, California
4. Steve Spina at New England Aquarium, Boston, Massachusetts

In multicellular animals, cells constantly signal with one another to coordinate function.  About twenty or so signaling pathways have been identified so far. Each signaling pathway consists of a network of dozens of interacting proteins. How do these pathways transduce signal from outside into specific responses inside the cells?  How is information propagated?  When we set out to ask these questions, we found that another question had to be asked first:  What is the information that the cells sense?

Our work in the Wnt signaling pathway led to a tantalizing suggestion: the cells sense signal in a relative manner.  The information that the cells sense in the Wnt pathway is not the absolute level of the signal, but instead the cells sense changes in the level of the signal.  Specifically, it is the fold-change in the signal that appears to be important in the Wnt pathway (Goentoro and Kirschner 2009; Goentoro et al., 2009; Shoval et al., 2010).

Why does it matter whether signaling is relative or absolute?  It leads to different models for how signaling works.  Relative sensing means that the cells are continually adjusting signal detection to background or basal signal level.  This may help explain how signaling works in the face of variation from cell to cell.  Relative sensing may help explain how the Wnt pathway can function in many contexts, in spite of variation across tissues.  Relative sensing helps maintain sensitivity during repeated signal, when cells may not have enough time to return to the basal state.

Weber’s Law in cell signaling.  Relative sensing in cell signaling brings to mind the way our sensory system works.  We perceive our world in a relative manner.  Our sensory system continually adjusts signal detection to background.  To understand this, imagine how we can easily talk to each other in a quiet room, but we have to shout a bit in a lively market.  The relative nature of our sensory system was described about 150 years ago in what is now know as the Weber’s Law:

Detectable Signal / Background = Constant

Weber’s Law has also been observed in bacterial chemotaxis and phototropism in phycomyces (Delbruck and Reichardt, 1956).  We may be seeing convergence between information processing at the organism, tissue, and cell level.

Ernst H. Weber. Image source: http://heilpraktiker-lexikon.com/Weber.html

Might Weber’s Law be a more general principle in cell signaling?  Work by Celina Cohen-Saidon in Uri Alon’s lab suggests that signaling in the EGF/MAPK pathway may also follow Weber’s Law.  Moreover, we recently found evidence for Weber’s Law in another major signaling pathway.  Stay tuned as we follow up on this exciting lead!

Mechanism of Weber’s Law in cell signaling.  What is the mechanism for Weber’s Law detection in the Wnt pathway? Sensing fold-changes in signal level means that cells remember the level of signal from the past, and take the ratio between the present and past level of signal.  We are now deep in biochemical purification to identify the molecules that serve as the memory in the Wnt pathway and mediate fold-change computation.

Kardiermaschine. Photograph by massenpunkt.

When we think about how many uncertainties a growing embryo is facing, it is remarkable that embryos do not grow into unrecognizable monsters.  Protein levels are different from cell to cell, environmental parameters fluctuate, genetic variations abound, size and geometry vary.  But development of embryos relies on a large network of biochemical reactions and molecular transport, both of which are sensitive to variation in concentration, temperature, rate constant, geometry, among many other parameters.

It is easy in the lab to make this tail-less, two-headed frog tadpole. How do embryos manage to develop normally? Photograph by A. Baetica.

C.H. Waddington called it “the constancy of wild type” in his 1942 writing:  “…or perhaps it would be better to call it the buffering… is evidenced most clearly by the constancy of the wild type.  …the wild type of an organism, that is to say, the form which occurs in Nature under the influence of natural selection, is much less variable in appearance than the majority of mutant races.  In Drosophila the phenomenon is extremely obvious; there is scarcely a mutant which is comparable in constancy with the wild type, and there are very large numbers whose variability, either in frequency with which the gene becomes expressed or in the grade of expression, is so great that it presents a considerable technical difficulty.”

The fine-tuned-ness of the underlying physical and chemical processes, and the robustness of the overall system, suggest that buffering comes from the way the processes are designed and connected to one another.  We have now deciphered the mechanism of buffering at the level of the working of an enzyme (e.g., see Shinar et al., 2009), biochemical circuits (e.g., see Barkai and Leibler, 1997, Alon et al., 1999), signaling pathways (e.g., see Batchelor and Goulian, 2002, Goentoro and Kirschner, 2009), and molecular networks (e.g., see von Dassow et al., 2000, Eldar et al., 2002, Ben-Tsvi et al., 2008).

In the lab we ask, how do we integrate what we know about robustness at the molecular level and truly make sense of the remarkable robustness at the organismal level?