Measuring Transcellular Interactions through Protein Aggregation in a Heterologous Cell System

Synaptic interactions facilitated by synaptic adhesion molecules are foundational for the development, organization, specification, maintenance and function of synapses and the generation of neural networks. The identification of these transsynaptic cell adhesion molecules is rapidly increasing; thus, it is fundamentally important to identify binding partners and understand how these new adhesion molecules interact with each other. Additionally, genome sequencing has identified mutations in many of these adhesion molecules that are commonly linked to a multitude of neurodevelopmental, neuropsychiatric, and addiction disorders¹. Mutations in genes that code for synaptic cell-adhesion molecules may detrimentally alter trans interactions and may contribute to pathophysiological alterations in synapse formation and or maintenance.

Multiple assays exist to quantitatively assess protein-protein interactions such as isothermal calorimetry, circular dichroism, surface plasmon resonance² and although quantitative in nature, they have several limitations. First, they require recombinant protein, sometimes demanding lengthy and tedious purification steps. Second, they require sophisticated specialized equipment and technical expertise. Third, they can overestimate the extent of binding as they allow for both cis and trans interactions between proteins that are naturally tethered to a membrane in vivo. Here we propose a simple and relatively rapid assay that exclusively tests trans interactions.

To circumvent many of the complications associated with purified protein assays, we have optimized a cell-based protein interaction assay that recapitulates trans interactions in a reduced heterologous cell system. This assay has been previously used in various forms to study transcellular interactions. In this approach, candidate cell adhesion molecules are transfected into HEK293T cells. At physiological conditions, HEK293T cells do not exhibit self-aggregation, making them exemplary models for this assay. However, when individual populations of HEK cells expressing receptor and ligand are combined, the binding of the receptor and the ligand forces aggregation of HEK cells to occur. This aggregation is mediated exclusively by trans interactions and is usually observable in tens of minutes. No protein purification steps are required in this method, and the efficiency of the method relies on the paradigm that populations of HEK cells expressing cognate adhesion molecules are being combined and then imaged only tens of minutes later. Additionally, this method is relatively inexpensive, as neither antibodies nor costly equipment are required. The only equipment required for the acquisition of data is a standard fluorescent microscope. An additional advantage to this cell-based assay is the ability to quickly screen the effect of disease relevant point mutations on trans interactions. This can be performed by transfecting HEK cells with cDNAs of the mutant variants of the protein of interest.

In this protocol, we present an example in which we investigate whether a missense mutation in Neurexin3α (Neurexin3α^A687T), identified in a patient diagnosed with profound intellectual disability and epilepsy, alters interactions in trans with leucine-rich repeat transmembrane protein 2 (LRRTM2). Neurexin3α is a member of the evolutionarily conserved family of presynaptic cell-adhesion molecules and while recent work has identified multiple roles at the synapse^3,4,5,6,7, our synaptic understanding of this molecule and all members of the neurexin family remains incomplete. LRRTM2 is an excitatory postsynaptic cell adhesion protein that participates in synapse formation and maintenance^8,9,10. Importantly, LRRTM2 exclusively interacts with neurexin isoforms that lack the splice site 4 alternative exon (SS4-) but not with neurexin isoforms containing the splice site 4 alternative exon (SS4+). The human missense mutation (A687T) identified in Neurexin3α is located in an unstudied extracellular region that is evolutionarily conserved and is conserved between all alpha neurexins⁷. As the interaction between these two molecules has been established^8,9,11, we posed the question: is the binding capability of Neurexin3α SS4- to LRRTM2 altered by an A687T point mutation? This assay revealed that the A687T point mutation unexpectedly enhanced the aggregation of Neurexin3α to LRRTM2 suggesting that the extracellular region in which the point mutation is located, plays a role in mediating transsynaptic interactions.

1. Cell culture and transfection

1. Make HEK cell media with DMEM, 1x (Dulbecco's Modification of Eagle's Medium) supplemented with 4.5 g/L glucose, L-glutamine & sodium pyruvate and 10% FBS. Sterile filter.
2. Predetermine suitable ligands and receptors for aggregation assay.
   NOTE: Neurexin3α SS4+/- and one of its known ligands, LRRTM2, were used in this study. Ligands and receptors of interest were expressed from cDNAs in pcDNA3.1. A Gibson assembly was used to insert Neurexin3α into pcDNA3.1¹². Neurexin3α F/R: TTTAAACTTAAGCTTGGTACCGAGCTCGGATCCGCCACCATGAGCTTTACCCTCCACTC/
   GAGCGGCCGCCACTGTGCTGGATATCTGCAGAATTCTTACACATAATACTCCTTGTCCTT.
3. Prepare HEK293T cells.
   1. Grow HEK293T cells to confluency in one T-75 flask.
   2. Once confluent, use 2 mL of trypsin and place in 37 °C incubator for 2 min. Add 6 mL of HEK media to the flask to resuspend cells and transfer all 8 mL to a 15 mL conical tube.
   3. Pellet at 500 x g for 5 min and resuspend in HEK cell media for a total of 8 mL.
   4. Count cells and add 735,000 cells into each well of a 6-well plate. Adjust final volume to 2 mL for each well using HEK cell media.
   5. Place in 37 °C incubator and allow cells to grow overnight or until they reach 50-60% confluency.
4. Transfect HEK293T cells using the calcium phosphate method¹³.
   1. Transfect well-1 with 3 µg of the protein of interest and co-transfect with 1 µg of fluorescent protein (3 µg of pcDNA3.1-Neurexin3α^WT SS4- and 1 µg of mCherry).
   2. Transfect well-2 as in step 1.4.1. but with the mutated protein of interest (pcDNA3.1-Neurexin3α^A687T SS4-).
   3. Transfect well-3 with 3 µg of the ligand of interest and co-transfect with 1 µg of another fluorescent protein (3 µg of pcDNA3.1 LRRTM2 and 1 µg of GFP).
   4. Transfect well-4 and well-5 to serve as negative controls: well-4 with 1 µg of GFP and well-5 with 1 µg of mCherry.
   5. Prepare another plate (as in step 1.4.1-1.4.4) if requiring additional conditions or controls (Neurexin3α^WT/A687T SS4+).
      NOTE: Transfection efficiency is analyzed 24 h after transfection under an epifluorescence microscope and quantified as the number of cells expressing the fluorescent protein they were transfected with. A more streamlined approach would include the transfection of HEK cells with a bicistronic vector coding for a fluorescent protein and the ligand of interest and is highly recommended above co-transfection. In the case of this study, alpha Neurexins are ~4.3 kb and low fluorescence intensity was observed using a bicistronic system necessitating co-transfection.
5. 48 hours after transfection, harvest cells for aggregation.
   1. Wash each well twice with PBS.
   2. Add 1 mL of 10 mM EDTA in PBS into each well to gently dissociate cell-to-cell interactions and incubate plate at 37 °C for 5 min.
      NOTE: Trypsin is not recommended for step 1.5.2 due to potential proteolytic cleavage of adhesion molecules in study. Additionally, after EDTA addition the protocol may not be stopped until completion as cells will now be exposed to ambient conditions.
   3. Gently tap plate to detach the cells, and harvest each well into separate 15 mL conical tubes.
   4. Centrifuge conical tubes at 500 x g and room temperature for 5 min.
6. While cells are pelleting, prepare 6 incubation tubes by labeling the top of each microcentrifuge tube with each condition.
   NOTE: Each permutation of GFP and mCherry conditions should be used to encompass all experimental conditions and proper controls. For example: 1. GFP/mCherry, 2. mCherry/LRRTM2-GFP 3. GFP/Neurexin3α^WT SS4-—mCherry, 4. GFP/Neurexin3α^A687T SS4- –mCherry, 5. Neurexin3α^WT SS4-—mCherry/LRRTM2—GFP, 6. Neurexin3α^A687T SS4- –mCherry/LRRTM2—GFP. Make additional tubes to accommodate further conditions and controls.
7. Remove the supernatant and resuspend cells in 500 µL of HEK media with 10 mM CaCl₂ and 10 mM MgCl₂ warmed to 37 °C.
   NOTE: The addition of CaCl₂ and MgCl₂ allows adhesion molecules to reestablish binding and is only required if the transcellular interaction partners in question require divalent cations for adhesion.
8. Count the cells in each 15 mL conical tube using a hemocytometer and aliquot 200,000 cells of each condition into appropriate tube from step 1.6.1 for a 1:1 mix in a total volume of 500 µL.
   NOTE: It should only take 5 min per condition to count and aliquot amounts.
9. Incubate tubes at room temperature in a slow tube rotator.

2. Image acquisition

1. Optimize microscope acquisition parameters for specific samples. In this example, images were taken on a wide-field microscope. Use a 5x air objective (NA: 0.15; WD: 20000 μm) to get a large enough field for analysis.
2. Assess baseline aggregation immediately after mixing the two conditions of HEK cells in step 1.8. These are now the ‘time zero’ images.
   1. Pipette 40 µL of each sample mixture onto a charged microscope slide and image under fluorescence in both the 488 and 561 channels.
   2. Acquire three different fields of view at one focus plane per sample drop.
3. Acquire final images at 60 min as the ‘time 60’ image.
   1. To obtain the ‘time 60’ image of the mixture after a 60 min incubation, take another 40 µL sample of each condition from rotating tubes and pipette each sample onto a charged slide. Image as in step 2.2.2.
      NOTE: Cell aggregation should be checked every 15 min until saturation occurs. Timing of aggregation will depend on the proteins being tested.

3. ImageJ/Fiji Analysis

1. To quantify the extent of aggregation using Fiji/ImageJ, save analysis files.
   1. Save the provided Supplemental coding files into the imageJ macros folder on the computer.
   2. Install the aggregation macro provided (Plugins, Macros, Install, and select the “AggregationAssay.txt” file).
2. Determine thresholds.
   1. Load a ‘time zero’ .tif file into imageJ and split the channels (Image | Color | Split Channels).
      NOTE: The ‘time zero’ image is used to determine the thresholding and smallest puncta size for the whole experiment.
   2. Mask each channel (Plugins | Macros | AggregationAssay_MakeMask). Make Mask From Image window will appear. Check boxes next to Determine Threshold for Image and Determine Cluster Params from Histogram and click OK.
   3. Determine the threshold of the image using the slide bar, record the number to the right of the slide bar and click OK.
   4. A Histogram of Cluster Size will appear. Select a cluster size from the histogram that suits the experiment, type this number in the Min Cluster Size: box, and click OK. Clusters below this size will not be analyzed.
3. Run the analysis.
   1. Open the ‘time 60’ image of condition 1 in imageJ and split the channels as in step 3.2.1.
   2. Mask each channel (Plugins, Macros, AggregationAssay_MakeMask). Use the same threshold and size determined in step 3.2.3 and step 3.2.4. Unselect the boxes next to Determine Threshold for Image and Determine Cluster Params from Histogram and manually type the size and thresholds into the appropriate fields then click OK.
   3. Calculate the aggregation index (Plugins | Macros | AggregationAssay_CalculateOverlap). Select the masked channels to be compared and directory into which the resulting files will save.
   4. Repeat steps 3.3.1–3.3.3 for every ‘time 60’ image in every condition.
      NOTE: The aggregation index is defined as the total overlap area divided by the sum of the two channel areas minus the overlap area multiplied by 100 (Aggregation index = overlap area/[area of channel 1 + area of channel 2 – overlap area] x 100). This normalization is an ‘OR’ operation between the two masked channels representing the total pixels in either mask.

The A687T mutation increases Neurexin3α SS4- binding to LRRTM2⁷
To investigate how intercellular interactions of two known synaptic proteins are affected by the introduction of a point mutation found in a patient with intellectual disability and epilepsy, we used the above HEK cell aggregation assay (Figure 1). Cells were transfected according to section 1 and prepared for imaging according to sections 1 and 2 of the protocol. Cells were imaged at baseline where no aggregation was observed as expected (not shown). Images acquired at 60 minutes were analyzed as in section 3 of the protocol. To minimize selection bias, conditions were randomized to blind the experimenter. For similar reasons the whole field of view of every image was selected as an ROI.

Conditions in which cells were not expressing any synaptic ligands (GFP/mCherry) showed minimal aggregation after a 60-minute incubation (Figure 2). Equally, conditions in which only one of the two populations of cells were expressing synaptic ligands (mCherry/LRRTM2-GFP or GFP/Neurexin3α^WT/A687T SS4- —mCherry) exhibited minimal aggregation after a 60 min incubation (Figure 2). As expected, conditions that contained two populations of cells with incompatible adhesions molecules (Neurexin3α^WT/A687T SS4+—mCherry/LRRTM2-GFP) exhibited no aggregation at 60 minutes (Figure 2B). These conditions served as critical negative controls as LRRTM2 is known to bind exclusively to SS4 lacking isoforms (SS4-) of Neurexins and highlights the specificity of this aggregation assay.

Conditions with compatible adhesion molecules^8,9,10 (Neurexin3α^WT SS4- —mCherry/LRRTM2-GFP) exhibited significant aggregation after a 60 min incubation (Figure 2C). Aggregation can be visualized as yellow and is present in the overlap between cells (Figure 1 and Figure 2). Surprisingly, the condition where Neurexin3α A687T SS4- was co-incubated with LRRTM2 (Neurexin3α^A687T SS4- —mCherry/LRRTM2-GFP) exhibited significantly more aggregation as compared to its wildtype counterpart (Neurexin3α^WT SS4- —mCherry/LRRTM2-GFP; positive control). This suggests that the A687T point mutation in Neurexin3α enhances the binding capabilities of Neurexin3α SS4- to LRRTM2.

Figure 1. Workflow of aggregation assay. (A) HEK293T cells are transfected using a protein-1/mCherry, or a protein-2/GFP. (B) Expression of Neurexin3α takes 48 h. (C) Cells are harvested using 10 mM EDTA and then pelleted (D). (E) Cells are mixed in a 1:1 ratio and resuspended in 500 µL of HEK media containing 10 mM CaCl₂ and MgCl₂. (F) Cells are incubated at room temperature until aggregation occurs and a final image is acquired at ‘time 60’. Aggregation is visualized as the number of yellow puncta visible in the field of view. No aggregation is observed when cells are not expressing the correct receptor ligand pairs and are thus seen as individual red or green puncta. Please click here to view a larger version of this figure.

Figure 2. Neurexin3α^A687T SS4- has enhanced aggregation with excitatory ligand LRRTM2 compared to Neurexin3α^WT SS4-. (A) Representative images of negative controls after a 60 min incubation of mCherry with GFP, mCherry with LRRTM2, GFP with Neurexin3α^WT SS4+/-, and GFP with Neurexin3α^A687T SS4+/-. Aggregation was observed after 60 minutes in both LRRTM2 with Neurexin3α^WT SS4-, and LRRTM2 with Neurexin3α^A687T SS4-. (B) Quantification of the aggregation index in all SS4+ conditions after 60 minutes. Aggregation index = overlap area/(area of channel 1 + area of channel 2 – overlap area) x 100. (C) Same as (B) but SS4-. Data shown represent the mean ± SEM of the number of experiments (SEM: GFP/mCherry ± 0.0245, mCherry/LRRTM2 ± 0.02465, GFP/Neurexin3α^WT SS4- ± 0.02453, GFP/Neurexin3α^A687T SS4- ± 0.0109, Neurexin3α^WT SS4-/LRRTM2 ± 0.0538, Neurexin3α^A687T SS4-/LRRTM2 ± 0.0174; p= 0.0136). Numbers presented in bars represent the number of independent experiments carried out. Dotted lines represent baseline or average minimal aggregation of control conditions. Statistical significance was determined by one-way ANOVA, multiple comparisons; *p<0.05; ****p<0.0001.This figure has been modified from Restrepo et al.⁷. Please click here to view a larger version of this figure.

Supplemental Coding Files. Please click here to download these files.

Dissecting the protein-protein interactions that occur in trans during cell adhesion can lead to a better understanding of the molecular mechanisms underlying basic cellular processes including the formation, function and maintenance of synapses during maturation and remodeling. The implications of cell-to-cell interactions expand beyond neurobiology and have broader roles in signal transduction, cell migration and tissue development¹⁴. Aberrations in cell adhesion can disrupt cellular processes imperative for proper cell function and can underly a variety of etiologies such as cancer, arthritis, addiction, autism and schizophrenia^1,15,16. Here, we provide an optimized detailed protocol involving HEK cell aggregation that allows for testing cell-adhesion interactions in trans.

With this HEK cell aggregation protocol, we can dissect differences in aggregation to approximate binding capacity between a presynaptic protein and its postsynaptic ligand. As such, we can ask questions such as: is the interaction between two essential synaptic proteins affected by a point mutation? More specifically, is the trans interaction of Neurexin3α with LRRTM2 affected by A687T? The A687T missense mutation studied here is located in a previously unstudied linker region of the extracellular domain of Neurexin3α⁷. The results illustrate that the A687T mutation enhances the aggregation of cells containing Neurexin3α^A687T to cells containing LRRTM2 (Figure 2C). This finding is significant because it was previously shown that LRRTMs only bind to Neurexins via a Neurexin domain called LNS6¹⁷, and further suggests that sequences upstream of LNS6 can exert an independent effect on the trans interactions between Neurexin3α SS4- and LRRTM2⁷.

The results in Figure 2 exhibit the specificity of this assay as all negative controls (GFP/mCherry; Neurexin3α^WT/A687T SS4- —mCherry/GFP or LRRTM-GFP/mCherry) had no observable aggregation after 60 minutes. A critical control used in this study, Neurexin3α^WT/A687T SS4+ —mCherry/LRRTM-GFP, also exhibited no aggregation after 60 min because LRRTM2 is known to bind exclusively to SS4 lacking (SS4-) isoforms of Neurexin. This control demonstrates that simple overexpression of membrane-bound molecules is insufficient to force aggregation in this system, which further illustrates the specificity of this assay.

The cell adhesion assay described here has been widely used to test the interactions of transsynaptic cell-adhesion molecules^8,18,19; however, it may be used to test adhesive cell-to-cell interactions of membrane tethered proteins. This protocol, which has evolved over time, was optimized from the original published protocol and subsequent variations of the protocol by changing three experimental parameters. One, the incubation temperature for the cells was changed. The original protocol calls for the incubation of cells in step 1.9 at 4 °C. This low temperature can act as an environmental stressor and can decrease cell viability leading to cell death and lysis. During cell lysis, the cells secrete genomic DNA and cell debris into media that causes cells to clump in solution leading to false positives during imaging. Two, the number of cells per condition were increased to 200,000 per population and the objective size was changed to a 5x; this allows the researcher to image a greater number of interactions per field of view in order to increase statistical power within each n. Three, the aggregation index was measured differently. Previous aggregation indices were limited by the selection of cluster size by the experimenter leading to the exclusion of “subthreshold” clusters. By contrast thresholds are now set for individual cell size at ‘time zero’, and the aggregation is now calculated as the overlapping area divided by the sum of the two channel areas minus the overlap area multiplied by 100 at ‘time 60’ which allows for the inclusion of smaller positive clusters making the assay more sensitive to other protein-ligand pairs (Figure 2B,C).

Troubleshooting and optimization may be needed depending on the interacting proteins tested. Although the incubation period until final image acquisition will differ depending on the tested proteins, it is critical that no aggregation is observed at ‘time zero’. If aggregation is seen at time zero, it may imply that the cells were not healthy to begin with. This may be due to several factors including sudden exposure to temperatures below 37 °C or slow experimental procedure. Importantly, the resuspension media for step 1.7 should be at 37 °C, which will allow the cells to gradually reach room temperature without a sudden harsh drop in temperature. One way to have optimal cell health throughout the experiment is to ensure that steps 1.7 through 1.9 are completed within a 25 min timeframe with no breaks in between. It is recommended that the microscope is setup and ready to use before cell harvest cells in step 1.5; this ensures that a true ‘time zero’ image can be taken immediately in step 2.2.

If aggregation is not observed after 60 minutes of incubation, several factors may be contributing to this lack of adhesion. First, the proteins in question may not be binding partners or may engage in low affinity interactions not detectable by this assay. In this case, a more sensitive assay may be required.  Second, the protein(s) of interest may not localize to the membrane when expressed alone in HEK cells. In this case, confirm that the protein is membrane localized by using biochemical techniques (surface biotinylation) or directly tag the protein of interest with a fluorescent tag and image HEK293T cells at 40x or higher magnification to assess surface expression. However, after membrane localization is confirmed, we recommend using untagged variants for this HEK cell aggregation assay in order to more accurately assess the binding capacity of the native protein. Third, in this protocol we allowed Neurexin3α to express for 48 h; however, protein expression time may vary depending on the proteins tested. Fourth, it is critical to supplement the media in step 1.7 with MgCl₂ and CaCl₂ if the proteins in question are dependent on divalent cations as is the case with the example of Neurexins and LRRTM2¹¹. If it is unknown whether the interaction partners require divalent cations for adhesion, add EDTA into one condition. The EDTA should chelate remining cations naturally present in DMEM ensuring a calcium and magnesium free solution. If adhesion is observed after EDTA addition, the proteins in question do not require divalent cations for adhesion.

Many methods exist to test protein interactions; however, most are not specific for testing only trans interactions that occur from cell to cell and require tedious protein purification steps. Although the HEK cell aggregation assay is not a direct measure of affinity and should not be used to replace a more quantitative approach to recover dissociation constants, to the best of our knowledge, this assay represents an approach to test true trans interactions in a semi-quantitative and relatively efficient manner. Moreover, due to the relative ease of this assay, the HEK cell aggregation assay can be used in conjunction with these more quantitative approaches to obtain a broader and more complete characterization of the interactions taking place.