tim read this

tim is a bioengineering student graduate graduate student from fairfax va boston ma vancouver bc irvine ca. sometimes he reads papers.

[timreadthis]

Putting cells on things, part 2

# So what about PDMS? So PAAm gels may not be what they appear as a system for determining how cells respond to the bulk stiffness of a substrate. But what about PDMS? A team working in Japan decided to [take a closer look](http://dx.doi.org/10.1016/j.actbio.2012.10.015) at PDMS in an article due for publication in Acta Biomaterialia in March. Seo, Sakai, and Yui (SSY) made PDMS gels of different stiffnesses in the same way as Trappmann et al., by combining the PDMS precursor and cross-linker in different ratios. Instead of covalently linking collagen to their surfaces with a reactive cross-linker like Trappmann, SSY used passive adsorption to coat their PDMS with fibronectin, another important ECM protein. (This helped pique my interest, since this is a process that I use in my own research.) Then, they measured the macro- and nano-scale mechanical properties of the gel and evaluated whether the binding domain of fibronectin was visible to cells. SSY evaluated the macroscale properties of the constructs by bulk tensile and press tests. Nanoscale properties were determined by pressing a round-tip atomic force microscopy cantilever as a compressor. The macroscale and nanoscale properties showed different trends as a function of crosslinker concentration, one indication that the macro-scale stiffness may not simply represent the environment that cells encounter: Things start to go off the rails for this paper when they take a look at whether the amount of fibronectin available for cells to bind was different on the different surfaces. SSY performed a Micro-BCA assay to measure protein on the surfaces and ran ELISAs with two different monoclonal anti-fibronectin antibodies: FN-15 and HFN7.1, the latter of which specifically binds at a locus containing RGD and PHSRN cell-adhesion motifs. FN-15 is taken to bind the amount of fibronectin "at the outermost surface" (suggesting that micro-BCA reagents experience less steric hindrance than antibodies?) while HFN7.1 binding is supposed to represent the availability of the adhesive domain to cells seeded on the coated surface. (I can't find any data about what epitope the FN-15 antibody recognizes or why it's taken to be a better indicator of total protein; any clues?) The amount of fibronectin detected by binding by the FN-15 antibody varied by 50% between different substrates, while the amount of fibronectin detected by HFN7.1 remained consistent. Micro-BCA did not reveal any difference in total bound protein between the different conditions. SSY's interpretation of the data is that the different amount of fibronectin available at the "outermost surface" (measured by FN-15) is compensated almost exactly by changes in the orientation of fibronectin on the surface such that the same amount of the fibronectin binding domain (as measured by HFN7.1) is presented to cells in each case. I confess that I find this interpretation confusing, and I do not see that it is supported by the data they present. Occam's Razor might prefer the idea that only the accessibility of the FN-15 epitope, and not the accessibility of the cell-binding domain, is changing on the different surfaces. SSY press on, claiming that the changes in FN-15 binding follow roughly the same trend as repulsion force measured in the nanoscale indentation experiment. Looking at the graphs, that's a pretty ballsy claim, and certainly not quantitative: SSY also use a technique called QCM-D, which I won't pretend to understand further, that requires PDMS to be dissolved in chloroform and spin-coated onto a sensor before taking measurements. If I were reviewing this paper formally I would try to understand it and whether the differences in energy dissipation factor and "frequency of adsorbed fibronectin" might possibly be attributed directly to mechanical properties of the PDMS layer instead of changes in surface energy. Other measures of surface energy (water contact angle) showed no differences. Dissolving PDMS in organic solvents is bound to alter the PDMS in other ways, so I'm not sure I'd trust these data much even if I did understand them. SSY close by observing the effect of the different substrates on cell adhesion density. The plot of cross-linker concentration vs. cell adhesion density looks quite similar to the nanoindentation plot: Inanely, SSY proceed to compare it to their "relative exposure ratio" plot, which it does not convincingly resemble. I think I have to conclude that the authors of this work talk a much bigger game than they bring. Correlating cell adhesion density with nanoscale mechanical properties is interesting and useful, and it may even be novel. Whether these fibroblasts are representative of other cell types is something I'd like to know, and I'd also be curious to see whether the same trend exists for microcontact-printed fibronectin on oxidized PDMS. But the questionable claims they make about what their antibodies are doing by ELISA lead them to make conclusions about the orientation and folding state of fibronectin that I don't think they can support. I'd like to see the authors apply the FRET assay that Trappmann et al. employed in an aside in their PAAm paper. **Reference**: Seo, J., Sakai, K., & Yui, N. (2013). Adsorption state of fibronectin on poly(dimethylsiloxane) surfaces with varied stiffness can dominate adhesion density of fibroblasts Acta Biomaterialia, 9 (3), 5493-5501 DOI: 10.1016/j.actbio.2012.10.015

[timreadthis]

Cells getting stiffed: cellular response to material properties

Cells on gels

How cells sense their mechanical environments and how they integrate mechanical signaling cues into their behavior has become one of the fundamental questions of biomedical engineering and remains a focus of my own studies. One important mechanical property is stiffness. About fifteen years ago, Robert Pelham and Yu-li Wang described an elegant system for producing surfaces of the same material but with different stiffnesses, by combining polyacrylamide (PAAm) and its crosslinker bis-acrylamide in different ratios but at the same total concentration, and then conjugating the gels they produced to type I collagen to permit cell adhesion. They showed marked differences in the behavior of two different cell types on flexible vs. stiff materials. Their method became quite popular and inspired a number of studies considering different cell types and processes.

One such study was conducted by Adam Engler in Dennis Discher's lab at U Penn. Engler, using the PAAm system, showed that human mesenchymal stem cells (MSCs) can be directed towards neural, muscle, or bone-like lineages in the same media on gels of different stiffnesses. His results were published in Cell.

Trouble lurks

A challenge to the PAAm system was published in Nature Materials last year by a group of 13 authors sprawling across England, Germany, Switzerland, and the Netherlands. Trappmann (now a postdoc in Chris Chen's group at Penn) and colleagues, working with human epidermal stem cells, found that their cells were profoundly affected by their PAAm gels of different stiffnesses. But when they made silicone rubber (PDMS) substrates that had the same stiffnesses as their PAAm gels, their cells behaved identically independent of stiffness. This happened whether the PAAm gels had collagen or fibronectin, another adhesion molecule, tethered to the surface. Having noticed the Engler study, they decided to try with human mesenchymal stem cells. They reproduced Engler's findings on PAAm but MSCs differentiated into osteoblasts on all of the PDMS surfaces.

If materials of the same stiffness are affecting cells in two different ways, there has to be another factor at play. Trappmann & co. looked at SEM images of their gels and discovered big changes in porosity as the stiffness of polyacrylamide gels varied:

In order to assess whether porosity was a factor affecting the cells, they freeze-dried polyacrylamide gels of different stiffnesses and coated them with a thin layer of gold, so that they would each display the same rigidity but different porosities to the cells. As before, even though the surfaces now had the same stiffness, cells on "soft" gels with large pores detached; cells on "stiff" gels with small pores spread.

Anchors aweigh

So what makes porosity matter? Trappmann & colleagues decided to take a look at how the different surfaces affected how collagen was anchored to the surface. Like many polymers, collagen acts a little bit like a bicycle chain: over short distances, like the individual links in a chain, collagen is very hard to bend. The "persistence length" of a polymer describes the largest distance over which it acts like a rigid chain link. Over long distances, like a well-oiled chain of many links, collagen is quite flexible. If you tied a bike chain down to a board at every link, it would be very rigid. If you tied it down less often and left some slack in the chain, the exposed parts of the chain would be more flexible.

Which is exactly the experiment Trappmann performed: by embedding gold nanoparticles (which bind to collagen) at fixed distances in a polyethylene glycol (PEG) gel (which doesn't), they tied down collagen either every 60 nm (about once per persistence length), or every 190 nm. Then, they exposed the surfaces to epidermal stem cells. Cells that saw tightly bound collagen behaved like cells on stiff substrates, spreading without differentiating. Cells that saw the loosely bound collagen behaved like cells on very soft substrates, differentiating and detaching.

Therefore, they conclude, the difference between the PAAm gels that the MSCs were responding to wasn't stiffness of the gels at all! It was the microscopic porosity, which controlled how tightly bound collagen was to the surface. Big pores offered fewer binding sites, which meant cells found it easy to pull on the collagen they were attached to. Small pores offered more tethering sites, so the collagen was much harder to pull on. Local stiffness still matters, but not in the way we thought.

I found the Trappmann study convincing. I'm going to have to look into whether there are other experimental systems that allow the collagen tethering to be modulated independent of surface stiffness, or whether it's possible to fabricate the gold-in-PEG system in our lab. Next, I'll talk about an analysis of another system which I found a little less impressive…

Reference: Trappmann, B., Gautrot, J., Connelly, J., Strange, D., Li, Y., Oyen, M., Cohen Stuart, M., Boehm, H., Li, B., Vogel, V., Spatz, J., Watt, F., & Huck, W. (2012). Extracellular-matrix tethering regulates stem-cell fate Nature Materials, 11 (7), 642-649 DOI: 10.1038/nmat3339

[timreadthis]

Poking and prodding collagen

I'm beginning an exploration of Anne Plant's oeuvre on the advice of my advisor. I'm interested in how macrophages sense and interpret tissue architecture, including collagen structure; Dr. Plant has done a lot of work on thin films of collagen and their properties.

I'm shoveling articles into Pocket this morning but I thought I'd start with a short conference paper, Mechanical stability of collagen fibril networks (Materials Research Society Proceedings, 898: 0898-L15-02 (2005)).

I was hoping to learn something about how fibrils interact with and exert force on each other. The paper goes in a different direction, though. Gordon Shaw & colleagues use two different techniques to probe the force/displacement curves of individual collagen fibrils. Nanoindentation applies force with a tip radius of about 100nm; AFM has a much finer tip. They observed the failure load for both dry and hydrated collagen films, and make some pretty images of the dry ones using the nanoindentation probe.

Some takeaways:

NIST is much more careful than most academic labs. Their devices are narrowly calibrated, they do their work in a clean room, and they keep the temperature stable "within 0.02 °C" (yikes!).

Loading these fibers to failure by applying force through an AFM tip induces breaks in the collagen fibers.

Hydrated collagen is more resistant to cutting by nano indentation than dry collagen.

Hydrated collagen shows more hysteresis than dry collagen and may show evidence of more fiber-fiber interaction, but it's not super clear.

I haven't seen many conference papers that discuss biomaterials! This doesn't seem as common in bioengineering as it does in e.g. CS. Interesting to realize it's an option.

So that's well and good. Doesn't affect my research very much but gives a me a little bit of physical context.

[timreadthis]

Well now so today we're gonna honor my temporarily adopted nation by bringing you a paper from Calgary: the Houston of America's Frozen Hat. And it's sort of great. And really, incredibly dense. If you want to follow along with the supplemental methods,  which are very briefly summarized in the paper, there's 11 pages of 'em.

Our Cowtown friends are studying how neutrophils, a key effector cell of the innate immune response, respond to sterile inflammation, which is of course a response to sterile tissue injury (bruising, say, or a heart attack, or some kind of toxicity; a stab wound or an abscess would be non-sterile injuries). Neutrophils are pretty badass and carry a lot of weaponry around ("hydrolytic, oxidative, and pore-forming molecules"), so if they get a little too excited they can cause a lot of damage, which they tend to do, so figuring out how they're regulated could identify new pharmacological targets for treating sterile-tissue insults.

The really sort of awesome thing about this study is that they're observing these processes directly by intravital confocal microscopy. It's vivisection in the most literal sense and, sorry PETA, but it's really cool. They're straight-up anesthetizing mice with xylazine and Special K, opening them up the side, and positioning the livers under a confocal microscope. (It sounds like they're maybe very carefully removing the liver from the body while keeping it attached to the circulation!?) After inducing an experimental injury (they have a couple of different models, one of which is roughly "poke it with a hot electrocautery needle"), they superperfuse the liver with propidium iodide (a red fluorescent vital stain, rejected by living cells) to visualize the necrotic tissue they caused. They can see neutrophils, because they're using a transgenic mouse line where they express GFP. Then, they use blue fluorophores to see various other tissues.

It's audacious stuff, and they work it. They start by following the path of neutrophils arriving at the necrotic site and identifying adhesion molecules necessary for neutrophil adhesion at the site by perfusing the mice with blocking antibodies to each adhesion molecule, inducing the injury, and then seeing whether neutrophils are recruited or not. They show that different attachment molecules are required to respond to the presence of infection instead of a sterile injury.

Hypothesizing that extracellular ATP is a signal for neutrophil recruitment, they induce injury and then dump some ATPase on top, and sure enough manage to knock down the neutrophil response but not directed migration towards the injury in the few neutrophils that do show up, indicating that other chemotactic cues are also at play. They walk up the signal pathway using knockout and bone marrow chimeric mice (so you can have your cake and eat it too, here, for hemapoetic cells -- showing that neutrophils lacking a marker accumulate at the injury site at the same rate as neutrophils carrying a marker, *in the same mouse*) and illustrate that ATP danger signals act on endothelium, and not directly on neutrophils. Nice.

Then, they show that MIP-2 cytokine gradients are produced in the surrounding vasculature and that mice lacking the appropriate MIP-2 receptor fail to produce neutrophils that migrate towards the injury. But there's no MIP-2 directly at the site of injury, even though there's apparently intact endothelium, apparently because it doesn't stick to the glycocalyx. For some reason? Some consistent reason, with a consistent 150 um radius. Shady! But anyway, there's none there, but the neutrophils are still charging on right to the necrotic cells, so there's gotta be another cue that overrides MIP-2 signalling once the neutrophils get closer.

So there are other DAMPs, damage-associated molecular patterns, that get released from necrotic cells: including formylated peptides! Which apparently just happen, in mitochondria? And are released during necrosis? But knocking out FPR1 -- you guessed it, formyl peptide receptor 1 -- causes neutrophils to draw within 150 um of the injury site and then start wandering around, which is the same thing that happens if you cause injury that doesn't actually kill any cells. So: neutrophils respond directly to formylated peptides, which are released from necrotic cells, as near-range chemotactic cues.

They also demonstrated all of this over again in skin instead of liver, but because this is Science, that's all consigned to the supplementary info. Holy hell. And these suckers at UCSD should shut it, about ATP inducing chemotaxis, they suggest -- though that looks like a well-designed study (also in Science) from here, and nobody is much commenting on it; it'd be interesting to explore this discrepancy further. Maybe next time.

Well, that was a lot of work, but it probably took them years to write that paper, so I'm okay with it. Man!

[timreadthis]

You put your circuits in my cells!

Regot S, Macia J, Conde N, Furukawa K, Kjellen J, Peeters T, Hohmann S, de Nadal E, Posas F, Sole R. Distributed biological computation with multicellular engineered networks. Nature 2010 Dec;advance online publication [cited 2011 Jan 5] Available from: http://dx.doi.org/10.1038/nature09679

So synthetic biology is k-rad, right? One focus of effort in the synbio world is getting cells to perform Boolean computations on defined inputs. Boolean logic is the foundation of digital electronics, so in theory this is really powerful stuff! (Considering that transistors on IC’s are already considerably smaller than cells, cells are unlikely to become a better computing platform, qua computing platform, compared to electronic digital logic, but being able to transduce signals from and control different kinds of systems natively in wetware without an electronic interface is presumably very interesting. I can imagine some applications; does anybody have a good review article?) My understanding from reading this paper is that researchers have tended to approach this problem by trying to produce cells that act as self-contained logic devices – i.e., for a two-input function like an AND gate, the same cell would receive both inputs and itself produce an output signal. But often, this is hard!, and requires lots of genetic engineering work every time you want to produce a new logic component.

Regot, Macia et al. from Universitat Pompeu Fabra in Barcelona came up with a new solution for composing Boolean operations in a cell culture system, instead of inside individual cells, using extracellular “wires” to connect isolated, modular logic circuit components. They built a small set of different kinds of yeast cells that each perform a specific part of a computation and publish the result as a mating pheromone. Other types of cells accept the pheromone “wire”, and sometimes perform another computation based on a chemical input, and publish the result by producing GFP.

If I understand this right, this approach offers a substantially simpler way to approach assembly of a novel circuit than single-cell methods, and there’s some subtle reason here why this is better than other, previous consortia-based approaches which I’m missing right now, perhaps based on the fundamental functions selected? I think it's the choice of AND and N-IMPLIES as fundamental operations that helps reduce the number of wires required?

I admit I'm kind of lost but, look! Truth tables! Cells! FACS! :p

Having to do analysis by FACS is also really a downer for this kind of thing, though it's a good way to get solid numbers for a paper -- in practice I feel like it'd be way cooler/more useful if you could read colony outputs on an inverted microscope or a plate reader.

[timreadthis]

Hello!

Hello! Inspired by Six Impossible Things, I decided a good New Year's resolution is to read more papers. And what better way to make sure I read papers than to make sure I talk about reading papers? No? Well, it made sense to me. We'll shoot for one a day and see how we do!