Retinas, retinas, and technology

Recently, Apple and LG premiered a new technology called the retina display, which features an ultra-high pixel density [1]. For people with 20/20 vision, the pixel density of the display is actually higher than the sampling density of the viewer's retina. It is a principle similar to fast-flicker fusion, or the perception of coherent motion from a sequence of still frames presented at high-frequency. But what about people with degenerating retinas [2]? Fortunately, there are emerging technologies that can improve their viewing experiences as well (see Figure 1). These innovations are not yet ready for market, but are based on recent advances in BioMEMS and cell therapy.

Figure 1. LEFT: An image of the retina display from a next-generation iPhone. COURTESY: [1]. RIGHT: picture of the highly-complex architecture of the retina, in relation to the rest of the eye. COURTESY: [3].

According to two recent papers [4,5], there are two routes to repairing macular degeneration: stimulating existing cells in a way that re-activates them, or introducing precursor cells that can integrate into the retinal architecture. Using photovoltaic implants (Figure 2) made primarily from silicon [6], Mathieson et.al [4] 
were able to recover vision in the rat eye. There approach relies on the observation that loss of vision in degenerative diseases is primarily due to loss of cells in the outer layer (cells they characterize as "image capturing" photoreceptors), while cells in the inner layer (cells they characterize as "image processing" units) remain well- preserved. Loss of function due to degeneration is thus a blockage of this feed-forward component (e.g. from outer layer to inner layer). Using this model, the inner layer of cells can be stimulated 
in a way that mimics the effects of ambient light being processed by the outer layer cells.

This was done using the system shown in Figure 2. A camera was used to capture images in the environment. This was then converted into pulsed NIR (near-infrared) illumination, projected onto the retina using a pair of goggles (see Figure 2). Each pixel on the micropatterned array converted this signal into stimulation currents, which were delivered locally to inner layer neurons. For a rat retina, an entire micropatterned array is 0.8 x 1.2mm in size. A single pixel on this array is about 70um in size, and can elicit a neuronal response (in this case, neural ganglion cell action potentials fired as spike trains). The number of these activity bursts could be modulated by manpulating properties of the NIR stimulus such as irradiance and pulse width. Overall, among six healthy and five degenerate rats, the prosthetic seemed to recover visual function measured as bursts of retinal ganglion cell spikes. However, there are several technical challenges for implementing such a system in the eye for the long-term. One of these is maintaining a normal physiological temperature during pulsed light stimulation. A more fundamental limit involves the curvature of the eye cup (see Figure 1, right) limiting the maximum size of a single array, as graphene is not a highly compliant material. 

Figure 2: RIGHT: Histology of bionic retina demonstrating the size and placement of the implanted device. Notice the implant geometry with respect to the cell populations of interest. COURTESY: Image at left is from Figure 1 in [4], image at right is from Figure 6 in [4].

Pearson et.al [5] decided to take the cell therapy route to solving the same problem. Cell therapy has had many technical challenges in the course of its development [7], but in the past few years a number of promising studies have been published [8]. The cell therapy approach operates from the premise that a general loss of photoreceptors leads to the degradation of sight. There are no assumptions made about how the architecture functions, there simply needs to be an existing architecture in place in which transplanted cells may take root. In the attempt to regenerate this retinal architecture, 200,000 rod precursor cells were transplanted into knockout mice [9]. Of this number, up to 16% of cells are able to successfully (e.g,. functionally) integrate into target area. Some of these cells include a transgene which allows identifying high expression for the gene Nrl [10] using a GFP reporter. When the entire cell population is sorted for the Nrl+ criterion, the efficiency of cellular intergration is improved 20- to 30-fold. 

But what does it mean to functionally integrate? From a purely descriptive standpoint, integration is the ability of cells to migrate to specific target tissues and fully differentiate into mature cells. To fully assess their functionality, a number of assays are performed. Immunohistochemistry and ultrastructural analysis used to find hallmarks of fully-functioning neural cells. Notably, the transplanted rod precursors form triad synapses with existing bipolar and horizontal cells. Transplanted cells are also light repsonsive and exhibit dim-flash kinetics [11], which suggests normal function (see Figure 3). Finally, a number of behavioral tests for visually-guided behaviors such as spatial navigation and visual tracking [12], suggests that integrated cells drive these behaviors in a manner similar to what is seen in healthy, wild-type mice.

Figure 3. TOP: Measurement of dim-flash kinetics in mice using voltage-sensitive dye imaging techniques. BOTTOM: Results of the dye imaging for several different experimental conditions. For the combined panels (far right), the black patches represent mature rods, while red patches represent regenerated rods, and blue patches represent rods derived from Nrl/GFP+ (transplanted) cells. 
COURTESY: Figure 2 from [5].

While the level of development for retinal-related bionic technologies [13] is not as advanced as for visual display technologies, the promise is much greater. It is hard to say which approach is better or worse. Rather, these approaches might be used in combination, and one approach might be superior to another in select cases. Regardless, this is a relatively new and fast-growing area with much potential for new developments and great innovations. Stay tuned.

References:
[1] For more information on the retina display technology, please see the following links: http://en.wikipedia.org/wiki/Retina_display,    
http://www.apple.com/ipodtouch/features/retina-display.html


[2] Degenerating retinal function can either result from a retinopathy (due to diabetes, inflammation, or hypertension) or normal aging. For more information, please see: http://en.wikipedia.org/wiki/Retinopathy

[3] For more information on function and structure of the retina and vertebrate eye, please see: http://webvision.med.utah.edu/book/part-i-foundations/simple-anatomy-of-the-retina/

[4] Mathieson, K. et.al   Photovoltaic retinal prosthesis with high pixel density. Nature Photonics, 6, 391-397 (2012).

[5] Pearson et.al   Restoration of vision after transplantation of photoreceptors. Nature, 485, 99-103 (2012).

[6] Silicon is not the only material being used for implantation. A range of polymers can be used, provided they have the proper characteristics. For example, graphene is fast becoming a primary candidate material for regenerative medicine applications. Graphene is both biocompatible and conductive. Graphene can also be fabricated at nanoscale dimensions using layered deposition techniques. While modern graphene transistor arrays are primarily used as a sensing technology, technical limitations may be overcome that will enable greater computational ability (which will improve their usefulness as retinal prosthetics).

For more information, please see: Hess et.al   Biocompatible graphene transistor array reads cellular signals.
Advanced Materials, 23, 5045-5049 (2011) AND Schmidt, C.   The Bionic Material. Nature, 483, S37 (2012).

[7] Daley, G.Q.   The Promise and Perils of Stem Cell Therapeutics. Cell Stem Cell, 10(6), 740-749 (2012).

[8] Kim, S.U. and de Vellis, J.   Stem Cell-based Cell Therapy in Neurological Diseases: a review. Journal of Neuroscience Research, 87, 2183-2200 (2009).

[9] The knockout phenotype is Gnat double negative (-/-). This phenotype is naturally degenerate for the formation of a normal retinal phenotype. For more information on the function of Gnat (a family of acetyltransferases), please see: Vetting, M.W. Structure and functions of the GNAT superfamily of acetyltransferases. Archives of Biochemistry and Biophysics, 433, 212-226 (2005).

[10] Nrl is the gene that codes for the neural retina-specific leucine zipper protein. For more information, please see: http://ghr.nlm.nih.gov/gene/NRL.

[11] Dim-flash kinetics should be observed in rods that contain active photopigment, and is assayed in the context of light adaptation. For an example (in Salamanders), please see: Sakurai, K. et.al   Variation in rhodopsin kinase expression alters the dim flash response shut off and the light adaptation in rod photoreceptors. Investigations in Ophthalmology and Visual Science, 52(9), 6793-6800 (2011).

[12] tests include a version of the Morris water maze. For more information, please see Methods of [5].

[13] For more information on the development of prosthetic technologies (articles in Nature-related journals), please see: http://go.nature.com/hxbyrm.

Turing Machine Doodle

Today's Google doodle is a short Turing machine demo, in honor of Alan Turing's posthumous 100th birthday*. Nice*.

BONUS: here is a link to a short (5 minute) film called "Codebreaker" on Alan Turing's lifetime achievements.

* although the ultimate tribute would have been to build a Turing machine using sprites.

* culmination of the Turing Centenary year.

UPDATE (9/8/2012): I found a tribute to Turing from a computational neuroscience perspective in the latest issue of Trends in Cognitive Science, (Volume 16, Issue 9, 447–448). Featuring: Christof Koch, James McClelland, and Terry Sejnowski among others.

 

Promoting the Carnival (of Evolution)

The administrator of the Carnival of Evolution (CoE) blogroll (Bjorn Ostman) has been looking for ways to promote the cause to a wider audience. CoE is a "traveling" show (featuring the previous month's blog posts, news, and interesting publications) that is hosted by a different blog every month. This blog hosted CoE (the 46th iteration) in April of this year.

A version of this poster (this is the latest draft) will be featured in the BEACON booth at Evolution 2012 (being held in Ottawa). The graph at lower right shows that the number of posts featured on CoE has increased over time, which varies by host and what is going on that month. 

Bjorn would like feedback on the poster design and informational layout. If you have any ideas for how to publicize CoE, please let either Bjorn or myself know. 

In addition, the ievoBio conference is being held in parallel with Evolution 2012. They are sponsoring a challenge that involves finding new ways to synthesize phylogenetic data that looks like it could yield interesting outcomes.

V-poles: for a more shampoo-able planet

I just discovered this invention by Douglas Coupland (yes, the author). It's called the V-pole. I assume the V stands for both "Vancouver" and "volt", as it is intended to replace utility poles in his home city of Vancouver.

Yet the V-pole is intended for worldwide consumption. Here is a quote from the promotional materials:

“You would never think of building a house or office tower without electricity — in the same way, you would never think of developing future cities without V-Poles"

What are V-poles? V-poles are an alternative to power lines, cell phone towers, street lights, and many other nodes of our current utility infrastructure, all bundled into a visually attractive pole (as he is also an artist). Here are a few pictures of the V-pole taken from the project website.

Notes:

* a more shampoo-able planet is a reference to his book "Shampoo Planet", published in 1992.

* each V-pole is enabled with lightRadio (wireless broadband) connectivity. lightRadio is a compact, scalable, and high-bandwidth WiFi transmission technology developed by Alcatel-Lucent.

* if you were wondering, the quote is a product of a novelist writing his own public relations materials. 

Does arbitrage exist in biological systems?

In this post, I will be discussing a concept from economics called arbitrage in the context of biological and ecological systems. In economics, arbitrage is the act of exploiting so-called inefficiencies in the marketplace, usually by finding the maximum payoff in one market and the minimum cost in another [1]. This is similar to a so-called minimax strategy commonly used in behavioral ecology [2] and evolutionary optimization [3]. This also relates to the maximum power principle in the systems ecology literature [4]. However, biological arbitrage may allow for evolutionary tradeoffs that can be balanced, providing an optimum that is highly beneficial to the species in question.

Figure 1. A schematic that defines biological arbitrage in an ecological food web. The arbitrer in this example is the prey species.

An example of arbitrage from biology (ecological) is shown in Figure 1. In this case, the prey species conducts a mimimax-like search of foraging costs with regard to the current selection pressures introduced by predation. This sounds like what happens in every predator-prey interaction. However, the difference here is a single species (or organism) is a member of two markets simultaneously: a market of consumers, and a market of producers. These consumers and producers can be any set of biological agents, and usually results in an energetic hierarchy (e.g. set of trophic levels, as in Figure 1). In this case, a market is a specific set of interaction strategies. This market is defined by transactions and transaction costs [5]. Aside from transaction costs, every member of the market must pay a barrier to entry. If the market is nocturnal foraging, there are costs associated with this, and not every organism or species can pay.

Figure 2. A consumer-producer relationship with regard to energetic abundance provided by the producer.

In the case of Figure 2, the producer makes both a high nutrient content per berry and a large number of berries. Because of this, the consumer is able to do more with less. Specifically, the selection pressure on the consumer population is relaxed that allows even consumers with highly inefficient metabolism to flourish. This might provide a mechanism for sub-optimal biological traits to flourish and even become predominant [6]. Another possibility is that cultural behaviors may have emerged as a form of biological arbitrage. Cultural behaviors allow humans to adapt to and survive in a wide variety of ecosystems [7] by both developing new resources (e.g. innovative means of extraction or technology) and reducing the cost associated with resource extraction.

Figure 3. A schematic that defines biological arbitrage in an animal tissue. In this example, the arbitrer [8] is the tissue.

The self-assembly and maintenance of tissue microenvironments may also be better understood by using the concept of arbitrage. Figure 3 shows this trophic, hierarchical relationship and the proposed arbitrage that may exist between levels of organization. In this case, the “profit” is not made via predator-prey relationships, but rather through engaging in coordinated and other collective behaviors which minimize energy expenditure and maximize the information available to individual cells. The collective behavior of individual cells in a physiological system can be extended to individual organisms, or even social groups and populations.

Figure 4 shows a hypothetical distribution of costs and payoffs associated with a single individual or hierarchical level. These costs and payoffs are embedded in a hierarchy, and can involve both energy and information [9]. While in many theories it is the mean value that is of interest, in this case we care about the shared extreme value (e.g. overlapping tails of the distribution) [10]. In addition, these distributions do not have to be uniformly distributed (as shown in Figure 4). Such non-uniform distributions are expected for processes that involve patchy resources or asymmetrical hierarchies.

Figure 4. Hypothetical distributions of costs and payoffs for each hierarchical level. Blue distribution represents the cost function and the red distribution represents the payoff function. In the inset at right, the black lines demonstrate the optimal point which maximizes both. Overlap (region where red AND blue parts of the distribution exist) allows for the biological unit of interest to engage in arbitrage. Of course, this is a highly conceptual and idealized model (axes are in arbitrary units).

Why is arbitrage such a potentially powerful mechanism for enforcing biological organization? Perhaps by serving the same function as it does in economic markets. In economics, arbitrage allows for sellers to recoup costs associated with entering a new marketplace [11]. To see the biological analogy, imagine an organism that engages in an alternate foraging strategy that has the potential to unlock new resources but does not guarantee a high rate of success. The relatively high cost of this strategy can be offset by evolving adaptations that reduce the likelihood of being predated upon, such as a toxic defense mechanism. 

Hopefully, this post provides a synthesis of ideas from disparate fields of study. And while the analogies are sometimes incomplete, there are supporting concepts from the areas of economics, applied mathematics, and statistics which might be able to fill in the gaps. What I have covered here is just a quick, first attempt at understanding this idea. As always, comments and feedback are welcome and appreciated.

References:

[1] Bjork, T. (2004). Arbitrage Theory in Continuous Time. Oxford University Press, Oxford, UK.

An introduction to arbitrage can be found here, and from a financial standpoint will be able to explain it better than I will attempt here.

[2] Stewart-Oaten, A. (1982). Minimax strategies for a predator-prey game. Theoretical Population Biology, 22, 410-424.

* for information on optimal foraging (a related approach), please see: Charnov, E.L. 1976. Optimal foraging: the marginal value theorem. Theoretical Population Biology, 9, 129-136. In this paper, the author examines what determines the optimal length of time for exploration of resource patches.

* for information on the limits of optimal foraging, please see: Guyader, S. and Burch, C.L. (1976). Optimal foraging predicts the ecology but not the evolution of host specialization in bacteriophages. PLoS One, 3(4), e1946.

[3] Parpas, P. and Rustem, B. (2001). Algorithms for minimax and expected value optimization. Handbook of Computational Econometrics, Chapter 4. D.A. Belsley and E.J. Kontoghiorghes eds. Wiley, New York.

For an explanation of minimax theory as originally developed by John von Newmann in the context of game theory, read this link.

For the relationship between the Nash equilibrium, minimax, and game theory, please see: Hofbauer, J. and Sigmund, K. (2003). Evolutionary Game Dynamics. Bulletin of the American Mathematical Society, 40(4), 479-519.

This may not be made clear by this post, but approximation of the minimax strategy or return by a biological agent may oftentimes be non-convex. For a high-level treatment of minimax theory with reference to non-convex problems, please see: Du, D-Z. and Pardalos, P.M. (1995). Minimax and Applications. Kluwer, New York.

[4] Odum, H.T. and Brown, M.T. (2007). Environment, Power and Society for the Twenty-First Century: The Hierarchy of Energy. Columbia University Press.

[5] I have likely not done the idea of biological markets justice. For more information, please see the following references:

* Noe, R. and Hammerstein, P. (1995). Biological markets. Trends in Evolution and Ecology, 10(8), 336-339.

* Norscia, I., Antonacci, D., and Palagi, E. (2009). Mating first, mating more: biological market fluctuation in a wild prosimian. PLoS One, 4(3), e4679.

[6] To develop a point of view on this, I took loose inspiration from research demonstrating the limits of arbitrage with respect to market efficiency. Please see the following citations from the finance literature:

* Shleifer, A. and Vishny, R.W. (1997). The Limits of Arbitrage. Journal of Finance, 52(1), 35-55.

* Stein, J.C. (2005). Why are most funds open-end? Competition and the limits of arbitrage. Quarterly Journal of Economics, 120(1), 247-272.

[7] Boyd, R. and Richerson, P.J. (2005). The Origin and Evolution of Cultures. Oxford University Press, Oxford, UK.

[8] I am defining an arbitrer in a biological system as a constituent of an intermediate hierarchical level. In an ecological food web, a top predator or primary producer could not engage in biological arbitrage, because an agent in either of these roles cannot maintain simultaneous relations with producers and consumers. In this example, we can see that arbitrage may require a trophic middleman (although this requirement may not be absolute).

[9] Alicea, B. (2008). Hierarchies of Biocomplexity: modeling life’s energetic complexity. arXiv:0810.4547.

[10] In mathematics and finance, this is also known as a copula. For more information on copulas, please see: Nelsen, R.B. (2006). An Introduction to Copulas. Springer, New York.

[11] This is a common theme in the economics and finance literature. Companies will often engage in arbitrage to recoup losses incurred from entering a new market. In this case, the recovery is purely monetary. In a biological context, the losses and recovery could be either energetic, informational, or both.

* To date, I have not been able to find any specific references to “arbitrage” in the biological literature. Academically speaking, the idea comes mainly from finance and economics, so the ideonational translation may be a bit rough.

BONUS: you may or may not be surprised, but the legal literature surrounding the term arbitrage can also involve predators (of the unethical, human kind).