Build
Engineering Fusion Protein Plasmids
Our team engineered fusion proteins composed of two parts: the HIRMAb and the neuroactive protein (either FGF-2 or NT-3). HIRMAb is a recombinant monoclonal antibody (a form of IgG), as such it is composed of two separate subunits expressed as independent proteins that will combine to form the mature HIRMAb after they have each been translated inside the cell. The two parts are the HIRMAb Heavy Chain (HIRMAb-HC) and the HIRMAb Light Chain (HIRMAb-LC). The neuroactive proteins are fused to the C-terminus of HIRMAb-HC forming HIRMAb-HC-FGF2 and HIRMAB-HC-NT3.
HIRMAb-LC (BBa_K4482007), HIRMAb-HC (BBa_K4482006), HIRMAb-HC-FGF2 (BBa_K4482011), HIRMAb-HC-NT3 (BBa_K4482012)
Using HIR-eGFP (Addgene, Cat# 22286) as a backbone, we cloned HIRMAb-HC and HIRMAb-LC, the plasmids expressing the heavy and light chains of the HIRMAb antibody. We also cloned HIRMAb-HC-FGF2 and HIRMAb-HC-NT3, the plasmid expressing the heavy chain of the HIRMAb antibody with FGF2 or NT3 fused to the C-terminus.
After running a PCR and validating our plasmids through gel electrophoresis, we ran a Gibson cloning and plated E. coli bacteria with our plasmid. The next day, we picked colonies, allowed them to multiply in LB broth, and then ran midi-prep to produce our final DNA constructs.
HIRMAb, HIRMAb-FGF2, HIRMAb-NT3 (BBa_K4482014, BBa_K4482015, BBa_K4482016)
We used PCR and Gibson cloning to engineer our final Trojan horse plasmids HIRMAb, HIRMAb-FGF2, and HIRMAb-NT3. To create HIRMAb, we used HIRMAb-HC as the PCR template to generate a backbone fragment with forward and reverse primers. Through Gibson cloning, we combined two inserts with HIRMAb-HC to create our final recombinant HIRMAb plasmid expressing both subunits of the HIRMAb.
We engineered HIRMAb-FGF2 and HIRMAb-NT3 through a similar method. To clone both plasmids, we used HIRMAb-HC-FGF and HIRMAb-HC-NT3 as templates with forward and reverse primers. We also combined the backbone with two inserts that were generated by PCR.
After running our PCR/Gibson cloning results on a gel with restriction enzymes, we saw results for our Trojan horse plasmids that were consistent with our Benchling models.
Expressing Fusion Proteins
After designing our recombinant plasmids, we transfected them into CHO-K1 cells. CHO-K1 cells are a mammalian cell line that are popular for manufacturing proteins. We chose to transfect the CHO cells with the jetOPTIMUS transfection reagent. After the Trojan horses were transfected into the CHO-K1 cells, they secreted the fusion proteins into the media. Due to the fact that the rate of proteins produced by the jetOPTIMUS reagent is highest 24 hours after the initial transfection, we had to time this reaction correctly with our transwell and interaction assays.
CHO-K1 cells growing in media.
Testing
Computational Modeling
Prior to designing our fusion proteins, we also wanted to determine how the structures of NT-3 and FGF-2 would be affected. For our in-silico design, we performed 3-dimensional modeling of wild-type FGF-2 and NT-3 and compared their structures to their structures when fused to HIRMAb. In the case of wild-type FGF-2, it would normally bind heparin and possess broad mitogenic and angiogenic activities. However, the attachment of HIRMAb with FGF2 causes changes in the secondary structure of the protein, which could change some of these functions. In the case of wild-type NT-3, the resulting fusion protein HIRMAb-NT3's structure possessed both alpha helices and beta sheets, as opposed to wild-type NT-3's structure, which only had the anti-parallel beta sheets. This demonstrated that the structure of HIRMAb-NT3 was actually more stable than that of wild-type NT-3.
Wild-type FGF-2 (left) compared to HIRMAb-FGF2 (right, FGF-2 in red)
Wild-type NT-3 (left) compared to HIRMAb-NT3 (right, NT-3 in red)
For our full report, click here
Transwell Assay
To test that our Trojan horse can cross the Blood-Brain Barrier, we designed a Transwell assay. We seeded HCMEC/D3 cells in 10 transwells and added enough LiCi to the growth media to a concentration of 10 mM to stimulate the Wnt pathway and encourage the formation of tight junctions between cells, accurately modelling the BBB. After a day, we refreshed the transwells with new media containing the Trojan horses. We ran 5 samples with replicates and kept 0.5 mL of each Trojan horse's media in a 1.5 mL tube to compare in a final analysis. The 5 samples were an untreated sample, HIRMAb-HC, HIRMAb, HIRMAb-FGF2, and HIRMAb-NT3. After time points of 30, 60, and 120 minutes, we collected 0.5 mL of media from the bottom chamber of the transwell and stored it in -20 degrees Celsius until analysis. After completing the assay, we performed a Western Blot to detect human IgG.
Model of our transwell assay.
After running a Western Blot on the media in the lower compartment of each transwell, we did not see any bands that would indicate the presence of our fusion protein. At first, this would seem to indicate that our Trojan horse was unable to cross the Blood-Brain Barrier, but through further testing mentioned below, we were able to create a new idea for why this assay may have failed.
Interaction Assay
The interaction assay determined if the components of our fusion protein would still be able to interact with their cell receptors. When HIRMAb is fused to NT-3 or FGF-2, the 3-dimensional structure of each protein is altered, something that we proved in our computational modeling. Due to the fact that protein form affects function, the changes that occur in HIRMAb's, NT-3's, or FGF-2's tertiary structure might affect its ability to bind to its respective cell receptor. Thus, our fusion protein may lose its ability to cross the BBB and promote neuroprotective benefits. Thus, we wanted to test if our fusion protein would still be able to bind to HIR (human insulin receptor), FGFR2, and NTRK3, the respective cell receptors for HIRMAb, FGF-2, and NT-3.
Example of interaction assay with HIRMAb and NT-3:
NT-3's receptor fused to eGFP is transfected into CHO-K1 cells.
Media containing HIRMAb-NT3 is supplemented to the cells.
Anti-human IgG antibody conjugated with red fluorescent dye is added into the media, binding to HIRMAb.
HIR-eGFP (BBa_K4482013)
We ordered the plasmid for HIR-eGFP from AddGene (Cat# 22286). We mini-prepped the bacteria containing the plasmid, and after fusing eGFP to the C-terminus of FGF-2 and NT-3, we midi-prepped HIR-eGFP along with FGFR2-eGFP and NTRK3-eGFP.
FGFR2-eGFP (BBa_K4482004), NTRK3-eGFP (BBa_K4482005)
We ordered the sequences for FGFR2 and NTRK3 from Twist Bioscience as gene fragments. We generated the backbone fragments formed of the plasmid containing eGFP through PCR. We then purified the PCR amplicons with gel electrophoresis and extracted the correct band from the gel. Using the Gibson assembly mix, we combined the backbone and insert fragments. After that, we used the Gibson assembly mix for bacterial transformation, and plated the bacteria on selective LB-agar plates. Our team picked out and amplified single colonies of bacteria by mini-prep, which we then analyzed by restriction digest and agarose gel electrophoresis to determine that we we had the correct plasmid. We then midi-prepped our plasmids for FGFR2-eGFP and NTRK3-eGFP.
After running our PCR/Gibson cloning results on a gel with restriction enzymes, we saw results for our receptor plasmids that were consistent with our Benchling models.
After midi-prepping the plasmids for HIR-eGFP, FGFR2-EGFP, and NTRK3-eGFP, our team transfected these plasmids into CHO-K1 cells. We prepared three 6-well plates and transfected every receptor plasmid into three of the wells in one of the 6-wells, except for HIR-eGFP which was transfected into 4 wells. This meant that we were using a total of 10 wells. We waited for 24 hours, and were able to use fluorescent microscopy to visualize the eGFP in each of our plasmids validate that our transfections were successful. We added media containing our fusion proteins from the "Expressing Fusion Proteins" section. We tested different combinations of receptors and Trojan horses. HIRMAb, HIRMAb-FGF2, and HIRMAb-NT3 were separately added into each of the 6-wells for a total of 9 combinations. We also added media containing HIRMAb's heavy chain into one of the wells containing cells transfected with HIR-eGFP as a negative control.
Finally, after an incubation time of two hours, we crosslinked the Trojan horses to the respective receptors using 1% paraformaldehyde (PFA) and after washing the PFA away with PBS, we added 0.5 mL of anti-human IgG solution to all 10 of our wells. The antibody binds to HIRMAb if it successfully binds to or is part of a fusion protein that binds to the CHO-K1 cell receptor. After 30 minutes of incubation, we viewed the cells under a fluorescent microscope. The expected results of the visualizations are shown below.
|
HIRMAb |
HIRMAb-FGF2 |
HIRMAb-NT3 |
HIRMAb-HC |
| HIR-eGFP |
This combination should show red signal (from the secondary
antibody binding the HIRMAb) on the surface of cells expressing GFP
(by the transfected insulin receptor). |
This combination should show red signal (from the secondary
antibody binding the HIRMAb) on the surface of cells expressing GFP
(by the transfected insulin receptor). |
This combination should show red signal (from the secondary
antibody binding the HIRMAb) on the surface of cells expressing GFP
(by the transfected insulin receptor). |
There should be no signal on the surface
of cells since the heavy chain alone of HIRMAb is not supposed to be
able to bind to the insulin receptor. |
| FGFR2-eGFP |
This is a negative control and should not show double signal
(green + red) on the same cell. There should be no signal on the surface
of cells since the HIRMAb alone is not supposed to be able to bind to the
FGFR2. |
FGF-2 in the HIRMAb-FGF2 should be able to bind to FGFR2 in
the transfected cells. Therefore, this combination should show red signal
(from the secondary antibody binding the HIRMAb in the HIRMAb-FGF2)
on the surface of cells expressing GFP (by the transfected FGFR2). |
This is a negative control and should not show double signal
(green + red) on the same cell. There should be no signal on the surface
of cells since the HIRMAb-NT3 is not supposed to be able to bind to the
FGFR2. |
There should be no signal on the surface
of cells since the heavy chain alone of HIRMAb is not supposed to be
able to bind to the insulin receptor. |
| NTRK3-eGFP |
This is a negative control and should not show double signal
(green + red) on the same cell. There should be no signal on the surface
of cells since the HIRMAb alone is not supposed to be able to bind to the
NTRK3. |
This is a negative control and should not show double signal
(green + red) on the same cell. There should be no signal on the surface
of cells since the HIRMAb-FGF2 is not supposed to be able to bind to
the NTRK3. |
NT-3 in the HIRMAb-NT3 should be able to bind to NTRK3 in the
transfected cells. Therefore, this combination should show red signal
(from the secondary antibody binding the HIRMAb in the HIRMAb-NT3)
on the surface of cells expressing GFP (by the transfected NTRK3). |
There should be no signal on the surface
of cells since the heavy chain alone of HIRMAb is not supposed to be
able to bind to the insulin receptor. |
Co-transfection of HIR-eGFP with HIRMAb under green light.
Co-transfection of FGFR2-eGFP with HIRMAb-FGF2 under green light.
Co-transfection of NTRK3-eGFP with HIRMAb-NT3 under green light.
The transfections for our receptors were consistent with our expectations for the experiment. The eGFP for HIR, FGFR2, and NTRK3 exhibited bright green illuminescence in their wells.
Co-transfection of HIR-eGFP with HIRMAb under red light.
Co-transfection of FGFR2-eGFP with HIRMAb-FGF2 under red light.
Co-transfection of NTRK3-eGFP with HIRMAb-NT3 under red light.
We expected the transfections with our Trojan horses to co-localize with the receptor transfections and produce a green-red illuminescence in some of our wells. However, none of our wells displayed a greeen and red double signal. We ran the experiment again with a higher concentration of our secondary antibody from 1:1000 to 1:200, but still received the same results. At this point, we considered the possibility that our Trojan horses might not have ever been there in the first place, that our CHO-K1 cells were unable to produce our fusion proteins and secrete them into media. Thus, we ran a Western Blot on our Trojan horses. We found that there were no bands produced, similar to the Western Blot from our transwell assay. This suggested that our CHO-K1 cells had difficulties expressing our fusion protein after transfection with our recombinant plasmids. While all of our plasmids were confirmed with restriction digests, none of them were sequenced. Thus, it is possible that our plasmids acquired a mutation early on, like one that introduced a stop codon into the protein's reading frame.
Learn
Our wet lab studies could be improved in a few ways in order to obtain better, more informative data. For instance, our project likely should have chosen to investigate another neurotrophin in place of FGF-2, as this is a protein that is usually not associated with neurons like NT-3 is. Furthermore, our team could improve the interaction assay by using a neuronal cell line model of stroke instead of CHO cells. SH-SY5Y cells are often used with oxygen-glucose deprivation (OGD) to model stroke. Immortalised cells do have some limitations, however, in that they contain oncogenes that distinguish them from actual neurons by increased proliferation, cell adhesion, and different cell morphology. Induced pluripotent stem cells (iPSCs) from stroke patients would be harder to acquire but could provide a better understanding of neuron response. Our Transwell assay could also be improved to model the Blood-Brain Barrier more accurately. In the brain, the BBB and neurons are separated by a layer of astrocytes. While it is more difficult, we could design a co-culture of HCMEC/D3 cells with an immortalised cell line for astrocytes. The HCMEC/D3 cells would be plated above the transwell membrane, while the astrocytes would be plated on the basolateral side of the transwell membrane. Finally, our study could investigate the downstream genetic effects of treatment with our fusion protein with Next-Gen Sequencing (NGS). This, combined with computational modeling, could provide a wider picture of how NeuroTrojan affects neuron health.
After receiving the results from fluorescence microscopy, we were confused by the lack of a strong double signal in some of the wells of our interaction assay. At that point, we considered the possibly of using a higher concentration of secondary antibody, or using a more sensitive method like flow cytometry to detect the binding of our Trojan horses to their cell receptors. After receiving the results from our Western Blot, we were able to connect the dots for why our Transwell assay did not have our fusion protein in the lower compartment. Considering these results, our main takeaway was that our protocol should have been more airtight to consider the possibility of this kind of event occurring. We did not think that there would be any problems from our transfection due to the relatively low chance of a plasmid mutation, and spent multiple weeks in the lab trying to run our Transwell and interaction assays without success. Given the literature on HIRMAb's efficiency, had we sequenced our fusion protein, we likely would have saved ourselves weeks of effort and likely even proved our experimental model.
For future research in this area (although outside of our capabilities as high school students), it is also important to run in-vivo studies on animal models, something we learned from Mr. O'Neill at Xylyx Bio. These studies would provide rich data on how much of the therapy is able to pass the Blood-Brain Barrier and how effective it is in restoring neuronal function after a traumatic incident like stroke.
