Engineering Cellular Memory

By: Una Choi


We often think of memory as tied innately to the brain. Humans can perceive, encode, and consolidate an event through activation of brain components like the hippocampus (1). Memory allows us to record finite events into lasting impressions, and past memories can affect our future perceptions and reactions.

Just as organs like the brain can permit the formation of memory, individual cells also possess the capacity to remember. The phage lambda (λ) system is one striking example of naturally occurring cellular memory (2). Phage λ is a bacteriophage, a virus that targets bacteria. Phage λ injects its genetic information into the target bacterium, and, in a stage known as lysogeny, will integrate this genetic material into the host genome.

Lysogeny is often referred to as the dormant stage, as the viral genetic material is passively replicated along with the host genome as the bacterium splits. The shift from the lysogenic stage to the following stage of viral infection, lytic growth, is due to a genetic switch composed of the cl gene and gene cro located in the viral genome (3). cl encodes for a repressor protein, which can halt transcription and thus expression of the targeted viral gene, while cro encodes for Cro, which can inhibit the repressor. During the lysogenic stage, repressor concentration is high, which inhibits transcription of cro and thus keeps Cro levels low. During the lytic stage, Cro concentration is high, which in turn represses transcription of cl.

This shift to increased Cro and decreased repressor is due to cellular stimuli; the viral genome can register an external event and can thus trigger a stable response. These stimuli often involve cellular stress, as phage λ can only ensure the continued replication of its viral genome if its host bacterium is healthy. In response to starvation or DNA damage, the bacterial enzyme RecA cuts the viral repressors and thus render them inactive. Without repressors to inhibit transcription of the cro gene, Cro is produced, which further acts to inhibit the viral repressors. High Cro levels, in turn, are responsible for triggering the lytic stage.

Phage λ exemplifies only one of the many naturally-found cellular instances of memory, as the genetic switch is used to induce the lytic stage in response to external stimuli. Cellular memory also plays a pivotal role in cell differentiation and cell division (4).


Using the same principles of stimulus response, synthetic biologists have begun engineering DNA memory circuits. These circuits either achieve volatile memory, which requires activated processes to function in a sustained fashion, or non-volatile memory, which does not require the continued activation of these cellular processes. While both volatile and non-volatile memory systems can switch between their states, volatile memory circuits are bistable; because volatile processes function in a sustained manner, switching between states is rare. As volatile memory circuits encompass the same “switch” function as the aforementioned natural phage λ system, this article will focus primarily on examining recent developments in creating volatile memory circuits.


Scientists can use the gene-editing device, CRISPR/Cas9, to manipulate DNA. Because DNA sequencing is becoming easier and cheaper, more and more host genome sequences are known. The elucidation of these genomes permits the creation of guide RNA strands, which are complementary to the target sequence in the host genome. The specific localization of the guide RNA to the target sequence can guide Cas9, a DNA-cutting enzyme (nuclease), to cut the desired sequence (5). After Cas9 generates these double strand breaks, the host cell can repair these breaks using nonhomologous recombination, which crudely re-ligates the broken DNA strands. This inaccurate mechanism often leads to the incorporation or deletion of several random nucleotides, thus leading to high incidences of mutation at the cut regions.

Scientists at MIT capitalized on these resulting mutations in their creation of a gene circuit that expressed Cas9 only in response to TNF-alpha, a tumor necrosis factor involved in systemic inflammation. Because the degree of increase in Cas9-mediated mutations in the guide sequence was positively correlated with concentrations of TNF-alpha, researchers could determine the concentration of TNF-alpha and the length of exposure to TNF-alpha from the number of mutations accumulated in the mammalian DNA sequence.


Gardner, et al (2000) constructed a double-negative feedback system in the prokaryotic Escherichia coli.6 In the event of chemical or thermal stimuli, the genetic system is flipped between two stable states. The device is bistable and once in one of the two steady states, the cell remains in that state even without the continued stimulus of the original signal. This sustained state is due to cooperativity; the binding of repressors onto the prokaryotic DNA circuit reinforces further binding of these repressors. This stability in the absence of continuous stimulus holds broad implications for further research as researchers can examine the cells without the constraint of continually stimulating the cells.


The ability to synthesize stable memory circuits in vitro and in vivo holds broad implications for the field of health diagnostics. The successful construction of memory devices in vivo can allow for the accurate categorization of cells; in other words, one can identify those cells that respond differently to stimuli by sorting the cell populations on the basis of the expected response to stimulus. Burrill, et al (2012) constructed three synthetic circuits to track cellular response to doxycycline, an antibiotic used to treat bacterial infections, hypoxia (oxygen deficiency), and DNA-damaging agents.7 After receiving the stimulus, the memory device activated, ultimately causing altered patterns of gene expression, growth rates, and cell viability. These altered patterns suggest memory devices are heritable, further demonstrating the potential benefits of using memory devices for diagnostics; the “switch” can be preserved through future generations of cells.

Burrill et al. also coupled the memory device with the sequence encoding red fluorescent protein (RFP), placing this tag downstream of the targeted gene. Doxycycline inhibits TetR, which normally represses the CMVtetO2 reporter. Hence, in the absence of doxycycline, no RFP is expressed because the repressor is bound to the CMVtetO2 operator. The addition of the stimulus (doxycycline) would remove the repressor and thereby activate transcription of the target gene and the downstream RFP tag. The transcription of ZF, another downstream element, could then activate transcription of another circuit, which includes a downstream element for yellow fluorescent protein (YFP). The YFP can then bind to that same promoter, creating a positive feedback loop.

By measuring the time elapsed between the addition of stimulus and cell fluorescence stemming from the continued production of YFP, Burrill et al. identified those cells most susceptible to doxycycline, hypoxia, and DNA-damaging agents; the cells most susceptible to the added stimuli exhibited more rapid onset of continued promoter activation. This also permitted the mapping of cell progeny to study more clearly the temporal stability of these memory-activated changes.


In addition to the diagnostic benefits of stably engineered memory devices, the ability to generate sustained cellular responses to a controllable stimulus is invaluable in the production of antibiotics and other cellular products.4 A common setback in industry is the high cost of continually inducing large cell cultures to express a certain gene or series of genes, which often requires multiple stimuli. The advent of customizable memory circuits can thus decrease production costs. Indeed, sustained expression of a targeted gene can also be beneficial in terms of increasing protein production.

Further study, however, is needed to elucidate the effects of sustained expression and the inheritance patterns of these memory-activated changes in future progeny.

Una Choi ’19 is a sophomore concentrating in Molecular and Cellular Biology.


[1] Nadel, L., et al. Neuroscience and Biobehavioral Reviews. 2012, 36, 1640-1645.

[2] Ptashne, M. Nature Chemical Biology. [Online] 2011. 7. http://www.nature. com/nchembio/journal/v7/n8/abs/ nchembio.611.html (accessed Oct. 1, 2016).

[3] Ptashne, M. A Genetic Switch (Cold Spring Harbor Laboratory Pr). 2004.

[4] Inniss, M. C.; Silver, P. A. Curr. Biol. 2013, 23, 1-10.

[5] Trafton, A. Recording analog memories in human cells. MIT News. 2016.

[6] Burrill, D. R.; Silver, P. A. Leading Edge. 2009.

[7] Burrill, D. R., et al. Genes & Dev. 2012, 26, 1486-1497.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: