Greater Precision

Goal: Engineer killer cells to increase their precision in detecting cancer cells.

Approach: After discussing our goals for the iGEM project, we have come up with an approach to increase precision by using:
1. CARs that recognize many types of different cancer ligands.
2. Logic gating to set higher restrictions on killer cell activation.

1. ANDN gate - we have successfully developed devices for this type of logic gate, and we have confirmed through testing data their ability to increase precision.
2. AND gate - we have constructed potential devices for this type of logic gate, and we are now working on optimizing the assay to measure their effects on precision.

Background information

In the same way that we recognize people based on their appearance, killer cells (cytotoxic T cells and NK cells) recognize target cells based on their different types of surface proteins [1]. This important ability to recognize the many different types of cells allows killers cells to eliminate unhealthy cells but avoid harming healthy cells. Unfortunately, killer cells can have trouble recognizing cancer cells among healthy cells due to complex profiles of surface markers on cancer cells. When it comes to cancer, killer cells are disadvantaged because they target foreign and dangerous organisms, but cancer cells originate from formerly healthy cells. Therefore, killer cells face difficulty in labeling cancer cells as dangerous entities because cancer cells express self antigens [2]. This fact is unsettling in that this method of differentiation is currently the only means by which killer cells can recognize cancerous cells.

We hope to introduce a logic gating system as an engineering platform to make killer cell recognition more specific and precise. We took advantage of the fact that many of the killer cells’ receptors bind to specific target cell surface proteins, much like antibodies bind to specific antigens. So why not replace the receptors’ recognition domains with different antibodies to create new receptors to recognize the many surface proteins on cells? That’s exactly what we did. These chimeric antigen receptors (CARs) are the products of an immune receptor intracellular signaling chain and an antigen binding domain, which, when put together as a single unit, can bind specifically to target antigens and trigger signaling responses [3]. Killer cells engineered with such modularly constructed synthetic receptors can overcome the restriction imposed by the presence of self-antigens on cancer cells.

Domain structure and function of a chimeric antigen receptor [4].

We also took advantage of the fact that killer cells’ receptors relay extracellular information intracellular compartments using modular signaling motifs such as ITAM, ITAM-like activation motifs, and ITIM. When ITAM and ITAM-like activation motifs become activated, they recruit kinases in the cytoplasm that initiate cell killing. ITIM motifs recruit phosphatases that cancel out the effects of ITAM activation [5].

The modularity of CARs and killer cells’ receptors makes it feasible to create a multitude of recognition systems that function as logic gates in the killer cell. The logic gates should enable engineered killer cells to recognize specific combinations of surface proteins on target cells. Each specific combination of surface proteins acts as the prerequisite to activate a logic gate in order to trigger a killer cell action, which makes recognition a highly precise process. This in turn allows killer cells to distinguish cancer cells from normal cells more effectively because the specific combinations of antigens found only on cancerous cells can be set as logic gate prerequisites to trigger cell activation, whereas the normal cells do not fulfill the prerequisites and are left unharmed.

Experimental Design and Results

For our project we have designed two main gates: ANDN and AND gate.

i. ANDN gate

In order to understand the ANDN gate, let us set up a hypothetical situation in which antigen A and antigen B are expressed on the membrane surface of healthy cells. Since cancer cells typically discard many surface proteins as a result of genetic mutation, we represent this discarded protein as antigen B in our scenario. Our ANDN gate is designed to address this issue by triggering cytotoxicity in the presence of antigen A and absence of antigen B. Therefore, cancerous cells that express antigen A and “hide” antigen B will be targeted. Healthy cells expressing both antigen A and B will not set off the activation of the ANDN gate. This concept is valuable for ensuring a level of specificity that prevents the overly indiscriminate activation of killer cells.

We tested two different ANDN gate designs to determine their effects on target recognition. To achieve the level of specificity as described by our hypothetical situation, we have set two different antigen binding domains to recognize antigen A and antigen B, respectively. Attached to the domain that recognized antigen A was an ITAM-bearing intracellular chain, from either the CD3 zeta or Fc receptor gamma, that signaled for killer cell activation. The domain that recognized antigen B, the antigen found in healthy cells, was fused to the intracellular portion of the ITIM-based receptor KIR3DL1, which inhibits killer cell activation. As a result of this combination, target cells expressing only antigen A would trigger killer cell activation, and target cells that do not express antigen A would not. Target cells that express both antigens A and B would be unharmed due to ITIM inhibitory signals, meaning that the presence of antigen B overwrites the input of antigen A. In conclusion, only a specific combination of surface antigens can set off the chain of activation, resulting in increased precision of detection and cancer killing.

Function of ANDN Gate

To evaluate our ANDN gate designs, we presented the dually transfected killer cells with target cells expressing antigen A, antigen B, both antigens A and B, or none of those antigens. The killer cells had been engineered to express the GFP reporter from the promoter of NFAT, a gene induced during killer cell activation. This reporter cell line enabled us to quantify the percentage of activated killer cells that express using FACS (fluorescence activated cell sorting). As shown in the figure below, killer cells presented with target cells expressing neither antigen or only antigen B showed basal levels of activation. Target cells expressing only antigen A, which represent cancer cells in this experiment, increased the percentage of activated killer cells. Notably, killer cells presented with target cells expressing both antigens had basal levels of activation. Living up to expectations, the ANDN gates proved to increase specificity because more killer cells became activated only in the presence of antigen A and the absence of antigen B.

Experimental Data of ANDN Gate on T-Cell Activation

ii. AND gate

Cancer cells are prone to overexpressing proteins on their cell surfaces. This fact allows us to detect the presence of cancer cells using AND gates. In our new hypothetical situation, normal cells express either antigen C or antigen D. In contrast, the overproduction characteristic of cancerous cells allows for the expression of both of these proteins [6]. AND gates are useful in this situation because they become activated only in the presence of two defined antigens. In the case of our new situation, the AND gate will trigger cytotoxicity only in the presence of antigen C and D, a condition that only applies to cancerous cells.

Applying the AND gate scenario and concept to the lab, we have explored using the activation adaptor DAP10. In normal killer cells, DAP10 recruits two different proteins to two different motifs along its main body when activated. This recruitment will trigger the killing response only when both proteins are present. In order to ensure that killer cells will only kill in the presence of two specific antigens, we used two mutant versions of DAP10, each of which has a different motif that does not allow its complementary protein to bind. Each mutant version is fused to a different extracellular part that recognizes specific antigens on cell surfaces. As a result, when such CARs only recognize one antigen on healthy cells, they will not be able to trigger activation because only one motif is activated. When both CARs bind to both antigens found on cancerous cells, both motifs are able to become activated to induce cell cytotoxicity.

AND gate design

Due to time constraints, we have not been able to evaluate our AND gate design. We could not measure the AND gate’s effect in killer cells using the NFAT promoter reporter assay. We attempted to test our AND gate using assays directly measuring the level of target cell killing. However, we faced a major technical obstacle that prevented us from obtaining informative results before the summer ended. The technical challenge was that only a small percentage of killer cells expressed the logic gate CARs after transfection. Due to this low transfection efficiency, the vast majority of killer cells used in target cell killing assays did not express our constructs. This was problematic because untransfected killer cells have an innate ability to kill, which produces a high killing background that makes it difficult to pinpoint the killing ability of transfected cells.

Future Directions

Our future goal would be to increase the efficiency for transfecting the killer cells (use viral vectors) and hopefully to obtain cells stably expressing our logic gate parts. This would allow us to test both the ANDN and AND gate designs directly based on cell killing efficiency with a significantly reduced basal killing level.


1. Formation and function of the lytic NK-cell immunological synapse.

Orange JS.

Nat Rev Immunol. 2008 Sep;8(9):713-25.

2. Learning how to discriminate between friends and enemies, a lesson from Natural Killer cells.

Bottino C, Moretta L, Pende D, Vitale M, Moretta A.

Mol Immunol. 2004 Jul;41(6-7):569-75.

3. Redirecting T-cell specificity by introducing a tumor-specific chimeric antigen receptor.

Jena B, Dotti G, Cooper LJ.

Blood. 2010 Aug 19;116(7):1035-44. Epub 2010 May 3.

4. Chimeric antigen receptor-engineered T cells for immunotherapy of cancer.

Cartellieri M, Bachmann M, Feldmann A, Bippes C, Stamova S, Wehner R, Temme A, Schmitz M.

J Biomed Biotechnol. 2010;2010:956304. Epub 2010 May 5.

5. Dissecting natural killer cell activation pathways through analysis of genetic mutations in human and mouse.

Tassi I, Klesney-Tait J, Colonna M.

Immunol Rev. 2006 Dec;214:92-105.

6. Oncogenic stress sensed by the immune system: role of natural killer cell receptors.

Raulet DH, Guerra N.

Nat Rev Immunol. 2009 Aug;9(8):568-80.