Introduction

Based on literature review and protein sequences available online, we designed and ordered plasmids from IDT containing our CARs and our mCherry-IL6-3xFLAG fusion protein. We then spent months experimenting with different ways to construct functional transcriptional units using these type 3 CAR- or IL6-coding parts with other Mammalian Toolkit and Yeast Toolkit (MTK and YTK; El-Samad et al.) parts. In general, our engineering cycle here consisted of selecting MTK/YTK parts for each component of a level 1 plasmid, constructing the plasmid via modular cloning, verifying correct construction via diagnostic digest, then transfecting HEK293 cells and visualizing fluorescence.

First iteration: learning from 0 transfection efficiency

In our first round of this cycle, we used the SV40 promoter and SV40 3' UTR as our MTK type 2 and type 4 parts, respectively, and observed minimal fluorescence. In fact, our CAR transfections—which were expected to fluoresce green due to the eGFP tag on the CARs—inexplicably showed significant red fluorescence. Brainstorming with the team and mentors led us to believe that either (1) photoconversion of GFP to RFP occurred, perhaps due to spatial issues during fusion protein folding; or (2) the cis-regulatory elements (CREs), including the SV40 promoter and 3' UTR, insufficiently stimulated translation. We were concerned about (1) being a possibility because, if the sequence itself produced faulty proteins, we would need to start from scratch and entirely redesign our parts. However, after replicating the transfection with identical conditions and comparing transfected cells with control cells, we ruled out (1) by noticing that both cell populations exhibited red fluorescence; aided by literature review, we determined that the red fluorescence was likely autofluorescence and was not cause for concern.

Second iteration: achieving significant transfection efficiency

Having ruled out photoconversion, we moved on to the second round of this cycle by researching optimal CREs for transfection of HEK293 cells. Literature review led us to select pEF1α for the promoter and retain SV40 for the 3' UTR. We again used modular cloning to construct plasmids with the same coding sequences as in our first iteration, but now with a different promoter. Upon transfection of HEK293 cells with these new plasmids, we observed significant fluorescent signal! For photos of the incredible difference in transfection efficiency between iterations 1 and 2, see our results page.

Third iteration: attempting to optimize further

While we were content with the results of our second iteration, we did make an attempt to further optimize our constructs. In particular, we tried changing the backbone in order to enable stable transfection of our plasmids. However, we experienced difficulty with the modular cloning process due to the large number of parts we were attempting to clone together. Despite multiple attempts, we were unsuccessful in constructing entirely correct plasmids. We ended up settling on the design from iteration 2.

Introduction

While designing our chimeric antigen receptors (CARs), we noticed that CARs contain an inherently modular structure: every CAR requires an extracellular domain to recognize a target, a transmembrane domain to transduce signal into the cell, and an intracellular domain to induce some reaction by the cell (e.g. phagocytosis or proliferation and cytotoxicity). This realization inspired us to design CAR-TK, a chimeric antigen receptor toolkit that facilitates modular assembly of CARs.

To validate that our design and workflow produce the expected MTK-compatible parts, we used PCR to extract components of CAR plasmids that we already ordered, and we verified results with diagnostic digests.

Design and workflow

Either by ordering linear parts to be synthesized or by using PCR to extract linear parts from plasmids, CAR-TK users must begin with linear gene blocks containing the sequence of a CAR module. These gene blocks must also be flanked by specific overhangs that enable the creation of an MTK-compatible Type 3 part via type IIS restriction enzyme digestion.

0. Assembly of pL0D (level 0 domain) entry vector

   To accommodate the fact that each gene block is a linear DNA sequence encoding a single component of a CAR, we designed a entry vector for these gene blocks to be cloned into, prior to assembly into a single level 0 type 3 part. The pL0D entry vector—which we have termed YTK000—is identical to the pL0 entry vector from the yeast toolkit (YTK001) but with the BsmBI sites replaced with BbsI sites, and with the overhangs modified to avoid similarity to the BsmBI sites present during cloning into the pL1 entry vector. We synthesized YTK000 by extracting parts of YTK001 via PCR then using BsaI golden gate to ligate them to form YTK000; the primers that we used can be found as part of our primer software, detailed below.

1. BbsI cloning into pL0D entry vector
   

   Most distal from the CAR component sequence is a BbsI site. These overhangs are common to all CAR-TK gene blocks and enable cloning of each gene block into the YTK000 plasmid for longer-term storage of individual CAR components.

   
   
   
   

   Upon BbsI digestion followed by ligation, the GFP portion of YTK000 drops out and the desired type 3D part replaces it to form a pL0D CAR component plasmid.

   
2. BsmBI cloning into pL0 entry vector

   Between each BbsI site and the CAR module sequence is a BsmBI site. These overhangs are specific to each type of CAR-TK part and enable cloning of four pL0D plasmids into the YTK001 plasmid for assembly of an MTK-compatible type 3 part containing a fully functional CAR.

   

   
   
   
   

   Upon BsmBI digestion followed by ligation, the GFP portion of YTK001 drops out and the four CAR components enter to form a pL0 CAR plasmid that is MTK-compatible as a type 3 part.

   
   
   
   
3. BsaI cloning into pL1
   

   Between certain BsmBI sites and the CAR module sequence is a BsaI site. This is necessary so that, as an MTK-compatible type 3 part, the CAR pL0 can be used in typical BsaI Golden Gate assembly with other MTK or YTK pL0 parts to create a transcriptional unit that will express the CAR. In particular, the following overhangs are required in CAR-TK gene blocks:

   

   
4. Decisions about restriction sites and overhangs
   

   Our choices for how restriction sites are ordered and for the overhangs corresponding to each CAR-TK part type have been carefully thought out to maximize the chances of successful cloning. In each phase of cloning described above, the restriction site used is cut out of the final plasmid product; thus, any restriction sites needed for downstream cloning must lie more proximal to the CAR component sequence. This philosophy motivated the BbsI-BsmBI-BsaI framework present in our design. Additionally, the CAR-TK type-specific overhangs we chose for cloning into the pL0 entry vector are not random four-base sequences but rather overhangs taken from other MTK type-specific overhangs that have already been optimized for minimal off-target ligation.

CAR-TK primer design software

To facilitate design of primers for PCRing out CAR-TK-compatible parts from plasmids containing full CARs, we have coded software that accepts annotated GenBank files and the desired CAR-TK part types corresponding to annotated regions, and returns primer sequences with appropriate overhangs and homologous regions. Visit our software page to learn more and download the software!

Our thought process behind our CAR-TK design

Initial design

At first, our toolkit was designed to accommodate three types of CAR parts: extracellular receptor, transmembrane signal transduction domain, and intracellular signaling domain. The overall workflow we proposed is cloning of linear CAR components (either ordered online or PCR'ed out of a plasmid) into YTK001, the MoClo pL1 entry vector, to create a normal type 3 part. After ordering primers and extracting our desired test parts via PCR, we constructed a pL1 CAR and verified its correct construction with a diagnostic digest gel.

However, we soon realized that multiple aspects of our design could be improved significantly. In particular, we felt it would be helpful if CAR-TK users could first clone individual CAR components into a entry vector before cloning those components into the pL1, because this would allow for creation of glycerol stocks of the components rather than storage of linear parts. Additionally, upon further literature review, we realized that enabling customizability of the spacer between the extracellular and transmembrane parts is vital. These two desired improvements motivated us to undergo another round of design to improve upon our CAR-TK.

Second iteration

For our second iteration, our toolkit was designed to accommodate four types of CAR parts: extracellular receptor, flexible spacer, transmembrane signal transduction domain, and intracellular signaling domain. The overall workflow we proposed is depicted in the diagrams above.

To assist with designing primers to PCR out the components of CAR plasmids we already have, and to help future CAR-TK users design primers, we additionally created software to automate CAR-TK primer design. To learn more and access the tool, view our software page.

Future steps

We are proud of our current CAR-TK design and believe it will serve as a useful tool to significantly simplify researchers' CAR construction workflow. Some ideal additions we would incorporate in future design iterations include the ability to customize new component types, for example an intracellular tag on the 3' end of the intracellular signaling domain. This could be useful for adding fluorescent tags, purification tags if needed for certain applications, extra signaling tags, etc. We would also like to conduct more experimental validation of the design and potentially iterate upon it further by considering other Type IIS restriction enzymes that may improve the CAR-TK assembly success rate and prevent off-target effects.