Why we pivoted to testing in cell culture:

Mammalian cells such as human embryonic kidney 293 or HEK293 cells are widely used as hosts to express recombinant proteins due to their ability to reconstitute key post-translational modifications, high transfection efficiency, faithful translation of protein. They have been used in the past to express receptor proteins and ion channels and have been described an extremely valuable host to conduct studies of membrane proteins.¹ However, high level expression of complex membrane proteins is a challenging task.1 NPC1L1 poses unique difficulties due to the multiple hydrophobic transmembrane regions and low expression rates. Additionally, membrane proteins are sometimes co-translationally inserted directly into the lipid membrane during synthesis which enables its proper folding.² Because the structure and function of MPs are so dependent on their interactions with the lipid bilayer, removing them usually has detrimental effects on their biological activity.²

Figure 1: Transfected HEK293 cells being grown up in a 6 well plate for a stable cell line.

Since NPC1L1 is a complex human protein, it was unlikely that the folding and posttranslational modifications such as glycosylation needed could have been expressed in bacteria. So the team devised a proposal for cell free expression which involved nanodiscs to solubilize the hydrophobic protein. To isolate a full-length membrane protein synthesised within a cell, the lipid bilayer must first be disrupted while avoiding damaging effects to the MP, then the MP needs to be reconstituted into a membrane environment (nanodiscs), typically using detergents.³ However, because no universal detergents can reconstitute every MP, achieving successful solubilization from the membrane becomes a time-consuming process as one tries to find the delicate balance between the solubilization efficiency of the detergent and the stability of the protein.³

The complexities of nanodiscs and the recommendation from HP contacts to use mammalian cells led the team to pivot toward expressing NPC1L1 and testing it while it was expressed in HEK293 cells.

Figure 2: Regular HEK293 cells 24 hours after passaging.

Figure 3: An image of the T25 flask used to contain untransfected HEK293 cells (taken at passage 24).

How were the mammalian cells used?

The first step in making this a reality was generating a plasmid that contained NPC1L1, a GFP sequence so that transfection could be identified, and a puromycin resistance gene. The backbone used was the PB-TAG-ERP2 plasmid backbone from Addgene, and Gateway cloning was utilised to add the NPC1L1 sequence in between the attb sites on the plasmid.⁴

Figure 4: The plasmid map for the PB-TAG-ERP2 plasmid backbone containing the NPC1L1 sequence and GFP that HEK293 cells were transfected with.

After training with the mammalian cells was complete, there was a lot of troubleshooting done as transfected HEK293 cells kept lifting and dying. This troubleshooting included both lipofectamine and doxycycline. It was hypothesised during lab meetings that the lipofectamine concentration could be too high and damaging or that the doxycycline could be over-expressing a membrane protein to the point of cytotoxicity. Therefore, a lipofectamine optimization test was done, trying to find an ideal ratio of how much lipofectamine 2000 to use compared to DNA.

Qualitatively, the cells that transfected and survived the best either had a 0.5uL lipofectamine to 1ug DNA ratio or a 1uL lipofectamine for 1ug DNA ratio. Higher concentrations of lipofectamine, such as using ratios that were 1ug DNA: 2ul lipofectamine up to 4 resulted in necrosis and cell lifting.

The second optimization tests conducted were doxycycline tests, which tested a range of concentrations of doxycycline fold dilutions, going from 8000x up to 500x. The cells exposed to these differing dox concentrations were then stained with propidium iodide and Hoechst staining. The propidium iodide stained dead cells, while Hoechst stained cells in general with the intention to merge images and see if greater expression of the membrane protein translated to higher levels of cell death. There were two ways this was observed, by merging images of GFP fluorescence (which indicates transfected cells) with fluorescence under Texas Red (causes fluorescence in propidium iodide) to see if transfection coincided more heavily with cell death. The second method that the dox test was quantified was using a microplate reader to get a more objective observation of the different fluorescence in the wells.

The merged image was taken at 10x of the 8000x Dox dilution. The green indicates transfected cells expressing GFP, while the red is propidium iodide staining indicating death. There was very little overlap overall, indicating that the dox-induced over-expression of plasmid likely doesn’t decrease cell viability. The graph represents the microplate reader data gathered from the doxycycline testing.

Both the merged images and the data gathered from the microplate reader indicated that the levels of cell death were more or less the same across the different doxycycline concentrations.

The project used a 1000x dilution on the recommendation of one of our TAs who had experience using dox-inducible systems, and the results demonstrated that there was fluorescence and limited cell death, so the 1000x dilution continued to be used. The doxycycline testing and results made it seem less likely that doxycycline was responsible for the cell death and lifting observed early on in the process of transfecting mammalian cells.

By early July, the team had replaced the old HEK293 cells, which had reached passage 50 with a newly thawed batch. The lipofectamine tests provided a ratio and volume that led to less cell death during and after transfections, and the doxycycline concentration was kept constant. At this point, further transfections were successfully done, however, puromycin exposure kept leading to cell death. When the wet lab checked our plasmid backbone and puromycin resistance gene in Benchling it was determined that the plasmid lacked the sequences necessary to be integrated into the genome of the HEK293s as hoped and that the puromycin resistance gene was likely not being expressed for more than a week as cells lost the plasmid.

These transiently transfected cells were still utilised in our preliminary cholesterol assay to functionally test the cholesterol absorbance of transfected versus untransfected cells. The Results indicated that transfected cells that were induced with dox did absorb more cholesterol than transfected cells that were not induced and untransfected cells.

Linear transfections were conducted to integrate the plasmid into the genome of the HEK293 cells. Our plasmid DNA could be linearized using the restriction enzymes Sap1 and Bmt1, and both of these were separately used to get two sets of stably transfected HEK293 cells that were expressing NPC1L1. These cut sites were chosen as they would linearize the plasmid, Bmt1 only cut once, and they wouldn’t interfere with any important coding regions. Two attempts were made to use serial dilutions to isolate for a single colony in order to develop a stable cell line. Once the dilution is completed and the cells are grown sufficiently, we will be able to establish a stable cell line that produces NPC1L1 and can be shared with other teams.

Figure 6: The annotated plasmid map, listing all the possible restriction sites available on our plasmid.

Figure 7: The restriction enzymes that were used in the linear transfection SAP1 and Bmt1. These were used in separate reactions to cut the plasmid and linearize it without cutting any vital parts of the plasmid.

The next steps from here with the mammalian cells involve verification of the presence of NPC1L1 and running a secondary cholesterol assay that determines the impact that including hedgehog has on cholesterol absorption. This will serve as the proof of concept for our proposed fusion protein medication by testing the binding of the key element of it with the receptor we are interested in.

Important further future directions:

Inhibition Testing:

1. Action duration - ways to extend the duration of inhibition of NPC1L1
2. Estimating the dose-response relationship
3. Compare the performance of CholesterLock to other small molecule inhibitors
4. General cholesterol uptake assays:
   • Comparing how CholesterLock’s performance changes if the NPC1L1 cells produce more membrane protein (increased doxycycline)
   • Comparing CholesterLock’s performance and binding affinity with NPC1L1 in the presence of more or less cholesterol

Stability Testing:

1. Testing the protein’s survival and stability in the conditions of the small intestine and digestive tract
2. Testing the protein’s survival and features after heating, freezing, thawing, drying
3. Testing the protein’s durability against proteolytic enzymes