Rational design of memory in eukaryotic cells

Ajo-Franklin, C.M., Drubin, D., Eskin, J.A., Gee, E.P.S., Landgraf, D., Phillips, I., and Silver, P. 2007. Rational design of memory in eukaryotic cells. Genes & Development 21: 2271-2276.

Overview

The final paper that will be discussed was published in September of 2007 by the lab of Pam Silver at Harvard Medical School. It describes the design and construction of memory in yeast. Using an autoregulatory positive feedback design in conjunction with an accurate mathematical model and quantitatively characterized parts, their work produced yeast that were capable of turning on a YFP positive feedback loop in response to a galactose stimulus. The production of YFP was shown to be able to survive cell division. As was the case with the previous paper, construction of a functioning synthetic network in a eukaryotic organism marks a significant step forward in the design of novel cellular devices.

Specific Biological Design

Figure 1: Biological design of permanent memory in yeast (permission pending).

The specific biological design of this gene network is a fairly standard autoregulatory positive feedback loop. As can be seen in Figure 1 on the right, two separate plasmids were constructed that each performed separate tasks. The sensor plasmid (on the top of Figure 1) consisted of an galactose-inducible promoter (P_gal) upstream of a hybrid RFP (Red Fluorescent Protein) gene. Fused to the RFP gene was a DNA binding domain (DNA BD) that is specific to the P_CYC1 promoter, a VP64 activator region, and a nuclear localization signal (NLS). In the presence of galactose, this hybrid RFP protein would be produced and localized to the nucleus of the cell by the NLS. Once in the nucleus, the DNA BD would allow binding of the hybrid RFP to the P_CYC1 promoter, at which point the VP64 activator would turn the P_CYC1 promoter on.

The auto-feedback plasmid (on the bottom of Figure 1) consisted of a hybrid YFP gene (Yellow Fluorescent Protein) downstream of the P_CYC1 promoter. The same fusions (DNA BD, VP64 activator, and NLS) were made to the YFP gene that had been made the the RFP gene. Thus, the production of the hybrid YFP protein would create an autoregulatory positive feedback loop.

In the presence of galactose, these yeast cells should fluorescence both red and yellow because of activation of both promoters. After galactose is removed from the cells' environment, they should retain their yellow fluorescence but lose their red fluorescence.

Mathematical Modeling

The mathematical model for this design is extremely similar to the example of cooperativity given in the mathematical modeling section of this wiki paper. An activator dilution equation was set equal to an activator production equation in order to determine the stable steady states of the system. The activator production equation took into account the concentration of the auto-activator (YFP) and the concentration of the sensor (RFP), as both were capable of activating the P_CYC1 promoter. It also took into account basal levels of transcription from each promoter as well as the cooperativity of binding of the two activator proteins, a necessary component of the system functionality.

Results

Figure 2: Experimental results of the memory network in yeast (permission pending).

The results obtained from this biological design are shown in Figure 2, on the right. Cells were exposed to either galactose or raffinose (a negative control) for a short period of time. Figure 2A shows DIC (Differential Interference Contrast Microscopy) images of cells in order to show the position of all cells in a given sample. Below the DIC images, RFP and YFP fluorescence images are taken of the same samples to detect any fluorescence in the cells. As expected, raffinose produces no fluorescence while galactose produces both red and yellow fluorescence. These dual fluorescent cells are then moved into a galactose free environment, where they lose their red fluorescence but maintain yellow fluorescence. Figure 2B quantifies and confirms the fluorescence that is detected visually though flow cytometry. Note that most, but not all, of the cells maintain their yellow fluorescent phenotype after being removed from galactose. According to the paper, 90% of the cells remain in the memory state post-cell division, although the data for this claim is not shown. Regardless, these results demonstrate a "prolonged response to a transient stimulus" (Ajo-Franklin, 2007).

Conclusions

I have referred to this design as "permanent memory" because cells in the memory state remain in the memory state regardless of any further system inputs. While the previous two papers described systems toggled that from on to off and from off to on, once this system is turned on by an input, it can never be turned off by another input. In a way, this can be thought of as non-rewritable memory. In addition to the construction of such a permanent memory circuit, this paper also demonstrated the ability to accurately predict system functionality through mathematical modeling of quantitatively characterized parts. The researchers were able to correctly predict how changes in growth rate and plasmid copy number would affect the functionality of the system. This was, once again, a progressive step in the development of complex, but well understood, gene networks in eukaryotic organisms.