• Constructed 2 anti-IL-6 CAR plasmids and an IL-6 fusion protein plasmid using modular cloning, after iterations of optimizing cis-regulatory elements
• Transfected all plasmids into HEK293 cells and observed significant expression of constructs as determined by fluorescence levels
• Co-cultured CAR-expressing cells with IL-6 and observed binding of IL-6 to anti-IL-6 CARs

Plasmid sequence design

Modular cloning and transfection

After receiving our plasmids from IDT, we cloned them into a plasmid backbone (YTK001) provided by the Yeast Toolkit (YTK). The other components of the plasmid were directly taken from the YTK. Our designs are noted on our lab notebook.

After verification of the plasmids via diagnostic digests, we cloned them again to include mammalian cis-regulatory elements and to form a complete level 1 transcriptional unit plasmid. We initially started with the SV40 promoter and SV40 3’ UTR, but saw nearly 0 transfection efficiency, meaning that extremely few cells per well in our 12 well plate successfully transfected and fluoresced.

Representative photos of our first attempts are to the right.

Figure 1: First attempt at transfecting HEK293 cells with 3xFLAG-IL6-mCherry (left) and anti-IL6 Megf10 CAR

However, we also ordered and used the CD22-Megf10 plasmid from Morrissey et al. as a positive control; we saw that the CD22-Megf10 plasmid had extremely high transfection efficiency, so we compared our existing promoter and 3’UTR to those used by Morrissey et al. Through many iterations of cloning, detailed on our engineering page, we finally ended up using the pEF1α promoter and SV40 3’UTR.

After many iterations of plasmid construction, we ended up seeing substantial fluorescence from all three of our constructs! Representative photos from transfection of our finalized designs are to the right.

Figure 2: Final transfection of HEK293 cells with 3xFLAG-IL6-mCherry (left) and anti-IL6 Megf10 CAR

Mammalian cell line choice

Stable versus transient transfection

Purification of 3xFLAG-IL6-mCherry protein

A central component of our experiment rested on the assumption that we would be able to purify our 3xFLAG-IL6-mCherry fusion protein from HEK293 cell lysate with extremely high yield. First and foremost, we needed to optimize the transfection efficiency of our fusion protein. Figure 3 shows the transfection of HEK293 cells at passage 20 with 600 ng of plasmid DNA at a 48 hour timepoint.

Figure 3: HEK293 cells transfected with our 3xFLAG-IL6-mCherry fusion protein

At this point, we began to approach protein purification from a few different perspectives.

Initially we considered biotinylating our protein to fluorescent streptavidin silica beads. This would knock out three birds with one stone:

1. We could create IL-6 clusters of variable, controlled sizes on a micron scale to surpass the size and signal threshold of macrophages.
2. We could isolate the IL-6 protein from the rest of the HEK293 cell lysate.
3. We could detect via fluorescence if the macrophages only bound to the IL-6 or actually phagocytosed the whole bead

However, we decided against the bead idea because it would require multiple rounds of purification, with unknown and likely decreasing protein yield rates. We valued the quantity of final protein above the existence of IL-6 aggregates for our proof-of-concept. Although we had hypothesized that aggregates are necessary to induce phagocytosis, for the scope of one summer and with our limited resources, we decided it would be enlightening and worthwhile to study how soluble IL-6 interacts with our CARs. Even if we have the chance to move on to macrophages in the future, studying soluble IL-6 now will still provide insight into how macrophages may endocytose or pinocytose soluble IL-6 in addition to any aggregates we are able to form in the future.

We instead turned to simply running one round of anti-FLAG tag purification. We utilized Pierce anti-FLAG magnetic agarose beads to selectively isolate the 3xFLAG-tagged IL6. We optimized the protocol by varying over three elution buffers:

1. IgG Elution Buffer
2. 0.1 M glycine
3. SDS-PAGE sample loading buffer, namely 1x Laemmli buffer

In order to give us an idea of our protein yield rate, we turned to using SDS-PAGE gel electrophoresis, the Bradford assay, and the NanoDrop spectrophotometer. At first, we had hoped that the SDS-PAGE gels would be able to verify the existence of our protein (by length) and the relative concentration (by band color intensity). After dozens of attempts to optimize the SDS-PAGE equipment available to us, we concluded that it would be an unreliable form of measurement. Then, we looked to quantify our protein concentration via Bradford assay and absorbance comparisons to bovine serum albumin standards. We were getting readings extremely close to 0 Abs and even negative absorbances, and we verified this tragic data with the NanoDrop.

Regardless of this insanity check, we proceeded with attempting to incubate the so-called “purified protein elution” with our CAR-expressing HEK293 cells. After concluding that there was no red fluorescence detected after incubation, we reconsidered our purification procedure. Noor, one of our mentors, brought up the idea of incubating the CAR-expressing HEK293 cells directly with the lysate of IL6-expressing HEK293 cells. We examined the lysis and purification steps that we had been applying—so far, we had been lysing the IL6-expressing HEK293 cells with ThermoFisher M-PER reagent. Since M-PER was a detergent that would unintentionally lyse our CAR-expressing cells and could not be extracted from the final lysate solution, we turned to another form of mammalian cell lysis to accomplish our goal: freeze-thaw lysis. The final lysate was not purified; we directly added it to wells of CAR-expressing HEK293 cells to observe the resulting behavior. This way, we knew IL-6 was in the lysate and would no longer act as a limiting factor.

We followed the 2013 Calgary iGEM team’s protocol but for a 10 cm dish of HEK293 cells. Ultimately, after viewing the fluorescence under our Keyence BZ-X810 microscope using the GFP and RFP filter cubes, we saw significantly more red fluorescence with freeze-thaw lysate than the purified protein eluted with Laemmli buffer and the purified protein eluted with IgG buffer.

Figure 4: Incubating CAR HEK293 cells with 3xFLAG-IL6-mCherry freeze-thaw lysate

Figure 5: Incubating CAR HEK293 cells with IgG-eluted 3xFLAG-IL6-mCherry

Incubating CAR-expressing (CAR+) HEK293 cells with 3xFLAG-IL6-mCherry

This phase of our project was full of the most unknowns. We would not be following any other lab’s protocols nor be able to reference multiple papers that cited running similar experiments. Ultimately, we had three types of IL-6 protein solutions we wanted to try incubating with CAR-expressing HEK293 cells:

1. Protein eluted with 1x Laemmli buffer
2. Protein eluted with IgG buffer
3. IL6-expressing HEK293 freeze-thaw lysate

These three solutions corresponded to our intention of experimenting with multiple protein concentrations to determine the best concentration for binding efficiency. We had two types of CAR+ cells: those that expressed Fcγ intracellularly and those that expressed Megf10 intracellularly. These populations would hopefully determine which protein could be attributed to better binding efficiency. Since we were using free-floating IL-6 instead of aggregates, we were unable to mimic the experimental conditions that Morrissey et al. used in their experiments.

We had additionally thought about varying incubation time in order to observe if the net amount of binding between the CAR and the IL-6 would positively correlate with time. We ended up running one experiment with a 1 hour incubation period and one experiment with a 6 hour incubation period. We saw significantly better binding with the 6 hour incubation – the result could have also been attributed to the fact that we had used the freeze-thaw lysate in this longer experiment.

We also accounted for controls in our design. As a negative control we introduced purified IL-6 to untransfected HEK293 cells. We additionally intended to introduce untransfected HEK293 whole cell lysate to CAR-expressing HEK293 cells, but we were unable to due to time restraints. Thus we are unable to document whether or not the CAR is specific to IL-6, or possibly to simply any protein it is incubated with over time.

Ultimately, we had uncertainty approaching this phase of the project, but we were able to nail down certain experimental conditions that allowed us to test for our proof-of-concept.

We viewed our cells post-incubation in two ways: first, using our Keyence BZ-X810 fluorescence microscope, and second, using flow cytometry. Figure 6 show one of our wells under the microscope, and below, you can read how we analyzed our data using flow cytometry.

Importantly, after the incubation period, the wells of cells that had been incubated with Laemmli-buffer-eluted IL-6 had fully lifted off the bottom of the wells. They were unable to be prepared for flow cytometry analysis – therefore, we were unable to quantify the frequency of double positive cells in those populations. We suspect this occurred because of the chemical nature of 1x Laemmli buffer, which is a reducing agent that may have denatured the proteins responsible for making the cells adherent.

Figure 6: Incubating CAR HEK293 cells with 3xFLAG-IL6-mCherry freeze-thaw lysate

Flow cytometry data

Ultimately, we have successfully achieved a proof-of-concept for the IL-6 binding component of our therapeutic by demonstrating that IL-6 binds to CAR-expressing cells significantly more frequently than to control cells.

It took a few attempts to run everything smoothly, but our final protocol followed a precise timeline that allowed us to see encouraging results!

Day 0: We transfected a 10cm dish of HEK293 cells with our 3xFLAG-IL6-mCherry fusion protein. mCherry control wells were transfected too.
Day 1: We transfected a 12-well plate of HEK293 cells, half of the wells with our anti-IL6 Fcγ CAR and half with our anti-IL6 Megf10 CAR. GFP control wells were transfected too.
Day 2: We extracted the cells of the IL6-transfected dish using a cell scraper and processed the cells in 2 different ways.

1. Scraped cells were lysed according to a freeze-thaw protocol (detailed on our experiments page) then co-cultured with wells containing CAR-expressing cells (detailed in our lab notebook).
2. Scraped cells were lysed according to ThermoFisher's M-PER lysis protocol (detailed on our experiments page) then purified with IgG elution buffer (detailed on our experiments page). Purified IL6-mCherry fusion protein was added to wells containing CAR-expressing cells (detailed in our lab notebook).

After a day of incubation of CAR cells with IL-6, we washed the wells with PBS to remove free-floating IL-6 unbound to CARs then prepared the cells for flow cytometry (detailed in our experiments page). We then ran flow with untransfected cells as a negative control; mCherry-expressing cells and GFP-expressing cells as fluorescent controls; and our CAR plus IL-6 experimental cells.

Due to our team's inexperience running flow cytometry, we struggled with establishing appropriate voltages and determining settings for parameters such as forward scattering and back scattering. Our post-flow analysis using the software FlowJo was challenging because of this, and we had to plot data with unconventional methods such as switching from linear to logarithmic axes to ensure all of our data points fit on graphs. However, with the help of mentors with much more flow experience than us—Hayden Sandt and Noor Radde, in particular—we were able to effectively and objectively re-format our data to better match academic standards.

We then gated our data to determine cell populations based on whether they fluoresced green (i.e., expressed CARs) and/or whether they fluoresced red (i.e., bound to IL6-mCherry). For full transparency, our backgating data and analysis is presented below. In each figure below, the left graph illustrates gating for HEK293 cells; the middle graph illustrates the subset of cells that are GFP positive (CAR+); and the right graph illustrates the subset of CAR+ cells that are mCherry positive (CAR+IL6).

Figure 7: Negative control backgating.

First, we used negative control cells to separate cell debris from actual HEK293 cells in our data. This was performed under the assumption that debris is significantly smaller in size than actual HEK293 cells. Thus, we strictly gated for the area around the hotspot and above, but discarded the area that tailed underneath. Retroactively, we applied our eGFP+ and mCherry+ gating to the negative control population to verify our gating stringency; as expected, neither eGFP+ nor mCherry+ cell populations were observed in the negative control HEK293 cells.

Figure 8: eGFP control backgating.

In Figure 8, we next used eGFP-expressing cells as a positive control. We intended to gate for green fluorescing cells (eGFP+) in this population, but due to the realization that we did not have a clear second peak representing the eGFP+ population, we decided to use this population to gate for mCherry+ instead. We knew these eGFP+ cells had no mCherry expression; therefore, our mCherry gate was decided where there was no fluorescent signal on the histogram. In the same manner, we determined our eGFP+ gate by using the mCherry-expressing population and selecting the portion with low to no mCherry signal. Retroactively, we verified that the majority of the eGFP-expressing population was correctly gated for eGFP+, and similarly for mCherry+.

Having gated for green expression, red expression, and no color, we then analyzed our CAR plus IL-6 experimental cell populations.

Figure 9: Megf10 CAR with IgG-purified IL-6 backgating.

Figure 9 corresponds to cells transfected with our Megf10 CAR and incubated with IgG-purified IL-6. We first subsetted the entire population to only HEK293 cells, removing debris. Next, we gated for cells which were eGFP+, implying CAR expression. We then further subsetted this green population to cells which also fluoresced red (doubly positive), implying that the CAR bound to our fusion IL6-mCherry protein. As desired, a significant portion of the entire population fluoresces both green and red, indicating successful binding.

Figure 10: Megf10 CAR with IL-6 freeze-thaw lysate backgating.

Figure 10 corresponds to cells transfected with our Megf10 CAR and co-cultured with IL-6 freeze-thaw lysate. We first subsetted the entire population to only HEK293 cells, removing debris. Next, we gated for cells which were eGFP+, implying CAR expression. We then further subsetted this green population to cells which also fluoresced red (doubly positive), implying that the CAR bound to our fusion IL6-mCherry protein. As desired, a significant portion of the entire population fluoresces both green and red, indicating successful binding.

Figure 11: Fcγ CAR with IgG-purified IL-6 backgating.

Figure 11 corresponds to cells transfected with our Fcγ CAR and incubated with IgG-purified IL-6. We first subsetted the entire population to only HEK293 cells, removing debris. Next, we gated for cells which were eGFP+, implying CAR expression. We then further subsetted this green population to cells which also fluoresced red (doubly positive), implying that the CAR bound to our fusion IL6-mCherry protein. As desired, a significant portion of the entire population fluoresces both green and red, indicating successful binding.

Figure 12: Fcγ CAR with IL-6 freeze-thaw lysate backgating.

Figure 12 corresponds to cells transfected with our Fcγ CAR and co-cultured with IL-6 freeze-thaw lysate. We first subsetted the entire population to only HEK293 cells, removing debris. Next, we gated for cells which were eGFP+, implying CAR expression. We then further subsetted this green population to cells which also fluoresced red (doubly positive), implying that the CAR bound to our fusion IL6-mCherry protein. As desired, a significant portion of the entire population fluoresces both green and red, indicating successful binding.

Finally, we aggregated the above data and concluded with exciting results: a significant percentage of cells expressed CARs and bound IL6-mCherry!!

Figure 13: Control and experimental cell populations positive for both green and red fluorescence. Being positive for both implies successful CAR expression and ability to bind IL-6.

From top to bottom, the populations are:

1. Negative control: as expected, cells not transfected with CARs bound no IL-6.
2. anti-IL6 Megf10 CAR (IgG method): 10.4% of cells transfected with our Megf10 CAR and incubated with IgG-purified IL-6 were double positive.
3. anti-IL6 Megf10 CAR (freeze-thaw method): 12.0% of cells transfected with our Megf10 CAR and co-cultured with IL-6 freeze-thaw lysate were double positive.
4. anti-IL6 Fcγ CAR (IgG method): 11.0% of cells transfected with our Fcγ CAR and incubated with IgG-purified IL-6 were double positive.
5. anti-IL6 Fcγ CAR (freeze-thaw method): 16.2% of cells transfected with our Fcγ CAR and co-cultured with IL-6 freeze-thaw lysate were double positive.
6. eGFP control: as expected, cells transfected only with an eGFP plasmid did not appear to fluoresce red.

Analysis of results

The precise protocols that we followed are detailed on this page and on our experiments page and lab notebook. We have made every effort to thoroughly document our experiments and ensure other researchers can reproduce our results, and we welcome questions or clarifications addressed to igem-2023-students@mit.edu

For replication of our experiments, we recommend preparing all of the possible controls to account for all variables of the project. Namely, to test for binding specificity, we suggest incubating CAR+ cells with whole cell lysate (ideally with some sort of fluorescent reporter) for 6 hours and comparing the double positive data with the data we have achieved.

Moreover, it is unclear why we observed significant loss of protein yield during the two-step M-PER lysis protocol and anti-FLAG protein purification protocol. We urge future researchers to test anti-FLAG purification of the freeze-thaw lysate, with the goal of understanding where yield is lost. It is possible that the loss of protein lies in the M-PER lysis protocol. Additionally, if it is found that the FLAG tag is the issue, we urge the creation of a similar IL-6 protein tagged with an amino acid chain that can assist in a different method of non-denaturing purification.

We are extremely proud of having achieved a proof-of-concept that HEK293 cells expressing our CARs bind IL-6 significantly more strongly than cells lacking our CARs. Due to constraints on our time and resources, we were unable to progress further than this level of achievement. However, we have planned potential next steps that can be taken to achieve the loftier goal of developing a real, practical cachexia therapy.

1. Transfecting macrophages: We wish to be able to test the transfection of our plasmids on macrophages such as RAW264.7 cells. We are curious to observe the transfection efficiency and the temperament of epitopes on the macrophages, especially as we experiment on them. We anticipate that macrophages may prove to be more difficult to engineer because, as immune cells, they have a proclivity for immune activation and changes in morphology.
2. Inducing IL-6 aggregation (if needed): We wish to consider the need for meeting the size and signal threshold for macrophages at the experimental level. This could mean engineering a different plasmid that codes for a secreted aggregation protein that is specific to IL-6.
3. Incorporating modulation via genetic circuitry: We wish to specify our therapy’s directed travel to the tumor micro-environment. This could be informed by chemotaxis, especially because the tumor micro-environment is notably characterized. Furthermore, we would ideally produce a therapy that modulates IL-6 levels rather than depleting it without a sense for if levels become dangerously low. We hope that next steps could include a sense and response genetic circuit.
4. Designing a delivery system: We wish to further consider the clinical delivery of the therapeutic, whether that involves engineering cells in vivo through lentiviral or CRISPR-based systems, or directly engineering cells after extracting them from patients, or something entirely different. Although the therapeutic is largely inspired by CAR T-cell therapy, we understand that our therapeutic would work in tandem with other cancer therapy. Therefore, we wish to answer the question of how administration of the therapy could coexist with other cancer treatment.
5. Quantifying pharmacokinetic/pharmacodynamic parameters: Due to a lack of time and resources, we were unable to thoroughly characterize many aspects of our therapeutic. We wish to undergo more stringent calculation of PK/PD parameters, such as association/dissociation constants of our CAR cells binding to IL-6. This well help us compare our therapeutic to related therapies, such as monoclonal antibodies, and enable computational modeling and simulation to advise further improvementss.
6. Testing in animal models of cachexia: Prior to even considering advancing testing to humans, we wish to utilize animal models of cachexia to get a feel for the efficacy of our therapy. Some of the experts we consulted, in particular Dr. Ishan Roy, have offered our team the resources to complete this; if we are able to progress that far, we would love to take this opportunity to begin in vivo studies.
7. Continuing to consider social and ethical implications: As mentioned in our human practices tab regarding the various conversations we held with experts, we wish to be better informed about how our therapy could affect people of different demographics. We understand that our therapy is currently limited to being informed by data that has a heavy male-bias. How can we make our therapy more inclusive, or identify the populations that would benefit the most from our therapy? Additionally, how can we make the education and treatment of cancer cachexia culturally-sensitive and knowledgeable? We wish to continue raising awareness about cancer cachexia so more patients can receive appropriate care.

1. Toolkit for Rapid Modular Construction of Biological Circuits in Mammalian Cells. Fonseca JP, Bonny AR, Kumar GR, Ng AH, Town J, Wu QC, Aslankoohi E, Chen SY, Dods G, Harrigan P, Osimiri LC, Kistler AL, El-Samad H. ACS Synth Biol. 2019 Nov 15;8(11):2593-2606. doi: 10.1021/acssynbio.9b00322. PubMed: PMID: 31686495.
2. Assembly of Genetic Circuits with the Mammalian ToolKit. Fonseca JP, Bonny AR, Town J, El-Samad H. Bio Protoc. 2020 Mar 5;10(5):e3547. doi: 10.21769/BioProtoc.3547. PubMed: PMID: 33659521.
3. Morrissey, Meghan A, et al. “Chimeric Antigen Receptors That Trigger Phagocytosis.” eLife, vol. 7, 2018, https://doi.org/10.7554/elife.36688.