Blood was collected into heparin from the jugular vein of cattle (n=2) in accordance with the protocol approved by the IACUC of the University of Massachusetts Amherst. Both animals were vaccinated against Leptospira serovar hardjo with the commercial whole cell inactivated vaccine Spirovac (Pfizer). PBMC were isolated from blood by centrifugation over ficoll-hypaque and suspended at 2.5x10⁶ cells/ml in complete RPMI (RPMI-1640 with 10% heat-inactivated fetal bovine serum (FBS, Hyclone), 5x10^-5M 2-mercaptoethanol, 200 mM L-glutamine (Sigma), and 10mg/ml gentamycin (Invitrogen)). Where indicated PBMC were cultured with 0.08 ug/ml of sonicated Leptospira or whole Leptospira for 1 hr to 7 days as indicated.

Cells were stained by indirect immunofluorescence with primary mAb GB21A (anti-TCRδ), CC15 (anti-pan-WC1), CACTB32A (anti-WC1.2), BAG25A (anti-WC1.1), CACT21A (anti-WC1-8, marking WC1.3⁺ cells), MM1A (anti-CD3), GC42A (anti-CD45), IL-A29 (anti-pan-WC1), ILA-12 (anti-CD4), and ILA-51 (anti-CD8) or with cholera toxin subunit B (to mark lipid rafts). Appropriate secondary goat anti-mouse isotype-specific antibodies were used conjugated with one of the following fluorophores: Alexafluor647 (AF647), Alexafluor488 (AF488), or pycoerythrin (PE) as indicated. All commercially available isotype-specific secondary antibodies (Southern Biotech) have been cross-absorbed against all other mouse Ig isotypes and tested for specificity in our hands. Product numbers of secondary antibodies conjugated with PE are 1080-09, 1070-09, 1020-09; with AF647 are 1080-31, 1070-31, 1090-31; and with AF488 are 1091-30, 1070-30. Controls included secondary antibodies alone with cells. To assess cell proliferation, PBMC were loaded with efluor670 (Invitrogen) in accordance with the manufacturer’s protocol prior to culture. For AMNIS experiments, all primary and secondary antibodies were titrated to avoid background fluorescence.

Cells were fixed with 4% paraformaldehyde following immunofluorescence staining before being analyzed by imaging flow cytometry using AMNIS Imagestream Mark II. Results were analyzed using the AMNIS IDEAS software. FRET was assessed using the donor and acceptor fluorophore combination of PE (Blue laser 488) and AF647 (Red laser 642), respectively. Compensation matrices for FRET experiments were obtained with all lasers on (i.e., Red 642, Blue 488, Violet 405, SSC 785) using a sample with no fluorescence in the acceptor channel as well as samples with fluorescence in the acceptor channel. Juxtaposition of cholera toxin B with cell surface proteins was assessed with the colocalization Wizard within the IDEAS software package. Aspect ratio displayed in figures is calculated by the AMNIS software as the ratio of the length of an individual cell’s major axis and minor axis.

Following staining of cells by immunofluorescence, cells were placed onto glass coverslips coated with poly-L-lysine for 1-2 hours before or after fixing with 4% paraformaldehyde followed by washing. Imaging buffer (690 µL Buffer B containing 50mM Tris at pH 8.0, 10mM NaCl, and 10% glucose) with 7 µL 2-mercaptoethanol and 7 µL GLOX solution [14 mg glucose oxidase, 50 µL catalase, and 200 µL Buffer A (10mM Tris, and 50mM NaCl)] was placed directly over adhered cells. Images were acquired in a Nikon N-STORM microscope using a Nikon PlanApo 100x NA 1.36 objective. To achieve super-resolution, a total of 20,000 images were collected in a sCMOS camera at a rate of 99 frames/sec. Exposure time was 10 millisec. Single molecule localization, reconstruction and cross-correlation for drift correction were performed using the FIJI image J ThunderSTORM plugin (26). The localization precision was 92 ± 22.4 nm for ex vivo cells and 90 ± 22.0 nm for cultured cells. Pearson’s coefficient of colocalization was used as recommended (27) to assess relative fluorescence from two channels and analyzed using the corr2 MATLAB function with 8-bit super-resolution reconstructions from ROIs of each cell from each channel to achieve a range of values from non-correlated (value = zero) up to perfectly correlated (value = one).

Leptospira interrogans serovar hardjo strains 1343 and 818 spirochetes were cultured in enriched Ellinghausen McCullough Johnson Harris medium (Sigma) at 30°C for 2 months, splitting cells to a concentration of 5x10⁷/ml every 2 wks. Bacterial concentration was determined by OD 600 (1 OD 600 = 8x10⁸ bacteria/ml). Spirochetes were fixed with 8% paraformaldehyde for 2 hrs and washed with PBS before biotinylation using the EZ-link sulfo-NHS-LC-biotin (Thermofisher) in accordance with the manufacturer’s protocol for biotinylating cell surface proteins (200 µL of 10 mM biotin per 2.5x10⁷ cells for 30 min). Unbound biotin was quenched with 100mM glycine solution in PBS. Biotinylated bacteria were then pre-stained with either Streptavidin-PE or Streptavidin-AF488 as indicated before use in binding experiments.

For evaluating binding of bacteria to lymphocytes, the labeled bacteria were incubated with bovine PBMC at a concentration of 2 to 2.5 x10⁶ bacteria per 5x10⁶ cells in a volume of 1 ml for 2-3 hr at 37°C with agitation every 30 min for flow cytometric analysis [FACS ARIA (BD)] to assess binding of bacteria to the cells. For flow cytometry sorting to enrich for WC1⁺/Leptospira ⁺ cells for STORM analysis the volume was reduced to 0.25 ml with the same number of bacteria and lymphocytes and PBMC were stained by indirect immunofluorescence with mAb CC15 (anti-pan-WC1) after incubation with the bacteria. The CC15⁺ PBMC with bound Leptospira were sorted to a purity of 59.9% (the low percentage is a result of bacteria being released from the lymphocytes during the sorting process). Assessing blocking of Leptospira binding by antibodies was performed by staining of lymphocytes with the indicated mAb and isotype specific secondary Ab conjugated with fluorochrome either before or after a 3 hr incubation with pre-stained bacteria with several washes between steps. Flow cytometry was then used to determine the percentage of mAb-stained cells binding fluorescent bacteria. The results were expressed as the ratio of cells binding bacteria with and without mAb blocking.

Bovine PBMC were stained with mAb GB21A (anti-TCRδ, IgG2b) and IL-A29 (anti-pan-WC1, IgG1) and secondary antibodies goat anti-mouse IgG, goat anti-mouse IgG2b, and/or goat anti-mouse IgG1 as indicated. They were cultured in 96-well plates with 1.25 x 10⁵ cells/well in a total volume of 200 µl. Cultures were established in triplicate for each condition and incubated for 4 days that included an overnight incubation with ³H thymidine (New England Nuclear) with 1µCi/well added on day 3 of culture. Cells were harvested with a cell harvester, incorporation of ³H thymidine determined by liquid scintillation and results expressed as counts per minutes (CPM).

For all FRET comparisons a 2-way ANOVA was performed followed by 1-way Student’s t-test for those showing differences. For co-localization Pearson’s correlation coefficient was employed and Student’s t-test of cells that were analyzed. For blocking of Leptospira binding to cells by antibodies, a Mann-Whitney rank sum test was used since the data were not normally distributed as a result of differences in bacterial binding efficiency among experiments. Significance is indicated as *p≤ 0.05 **p ≤ 0.01. ³H-thymidine results are shown as mean and standard error of the mean.

When co-cross-linked with the γδ TCR the γδ T cell co-receptor WC1 becomes phosphorylated on key tyrosines and augments T cell responses (16). When those tyrosines are replaced through point mutations or WC1 is silenced the cells fail to respond (19, 20). As shown here, when wild-type WC1 was not silenced but rather separated from the TCR by using cross-linking with isotype-specific secondary Abs the cells also could not respond to stimulation through the TCR, while co-cross-linking them together using a general anti-IgG secondary Ab they were activated to proliferate ( Figure S2 ). Thus, we hypothesized that WC1 and γδ TCR colocalize for cell activation to occur. To assess this, imaging flow cytometry (AMNIS) was used to measure FRET that occurs when molecules are ≤ 9 nm apart. As a pilot study to establish a baseline of FRET levels, we used two secondary Abs that react with different epitopes of an anti-WC1 mAb ( Figure S1A ). As a positive control using cells, we first evaluated CD3 and TCR interaction and found that after culture with Leptospira there was a visible shift in fluorescence indicating FRET had occurred ( Figure 1A ). In contrast, this did not occur with ex vivo quiescent cells. As a negative control CD45 and γδ TCR interaction was chosen for evaluation since those proteins are known to be in different protein islands both before and during activation of αβ T cells (28). Also, while CD45 is known to move into lipid rafts following activation of αβ T cells (29, 30) we found here that only about 20% of WC1 molecules in bovine γδ T cells were within lipid rafts and that none of the γδ TCR was ( Figure S2 ). These latter results for WC1, γδ TCR and lipid rafts are in agreement with a previous study (18). As predicted, on δ TCR⁺ cells that constituted about ~8% of quiescent ex vivo cells and ~17% of cells from cultures stimulated with Leptospira, we found no FRET between CD45 and γδ TCR ( Figure 1B ). Finally for the principal experiment, when FRET between WC1 and γδ TCR on ex vivo resting, i.e., quiescent cells, was assessed none occurred ( Figure 1C ), while some lymphocytes cultured with Leptospira showed fluorescence sensitized emission of the acceptor fluorophore indicating FRET ( Figure 1C ). The cells that were positive for FRET had a clear shift in fluorescence and a demarcation was evident between stimulated and quiescent cells ( Figure 1D ). In contrast, with no primary antibody for the acceptor channel present, designed to measure the contribution of background autofluorescence, the two types of cells (ex vivo and cultured) had essentially complete overlap in fluorescence levels ( Figure S1B ). Statistical analysis of the 3 independent experiments showed significantly more FRET⁺ cells after Leptospira culture when CD3-TCR or WC1-TCR interactions were assessed but not between CD45 and TCR ( Figure 1E ). As an additional control, we found that secondary Ab alone showed few cells with background fluorescence that fell within the gates used to signify FRET⁺ cells ( Figure S1C ).

We then examined the cells using STORM to obtain super-resolution fluorescence data on WC1 and γδ TCR interactions. On quiescent cells, there was a clear separation of WC1 and γδ TCR protein islands ( Figure 2A ). This was consistent with the literature regarding protein islands on resting lymphocytes in general (31). After culture with Leptospira, the WC1 and TCR protein islands became juxtaposed ( Figure 2B ) to various degrees ( Figure 2C ). Quantitative measurements of the colocalization were obtained through the corr2 MATLAB function to compute Pearson’s correlation coefficient. There was a significant difference between the ex vivo and Leptospira cultured groups (p=0.046). We found ~40% of the cells cultured with Leptospira had a significantly higher WC1 and TCR localization than quiescent ex vivo cells ( Figure 2D ). This agreed with the proportion of WC1⁺ γδ T cells from vaccinated animals that are known to undergo cell division following culture of PBMC with Leptospira as shown in previous studies (25).

Unlike CD4 and CD8 coreceptors of αβ T cells, there are variants of WC1 molecules arising from the bovine WC1 multigenic array of 13 genes. There are multiple stable subpopulations of γδ T cells each expressing 1 to 6 different variants (32). These WC1 variants may differ in the number of extracellular SRCR domains as well as their endodomains (24). γδ T cells designated as WC1.1⁺ and WC1.2⁺ differ in the WC1 genes they express and in their responses to pathogens (22, 25, 32). For example, cells within the WC1.1⁺ γδ T cell subpopulation proliferate and produce IFN- γ in response to Leptospira, while many fewer WC1.2⁺ cells do (33). The WC1 variants designated as WC1.1-types (4) bind Leptospira, while the WC1.2-types do not (14), suggesting a reason for the difference in cellular response. Thus, we hypothesized that following stimulation with Leptospira the WC1 molecules on some cells within the WC1.1⁺ subpopulation would colocalize with the γδTCR, while WC1 molecules on fewer cells within the WC1.2⁺ population would do so. We found FRET occurred between the WC1 molecules that reacted with the anti-WC1.1 mAb BAG25A and the γδ TCR if cells had been stimulated with Leptospira in in vitro recall cultures ( Figure 3A ). In contrast, for WC1.2⁺ cells (identified by mAb CACTB32A) very few showed FRET between their WC1.2 molecules and the γδ TCR for either quiescent cells or those cultured with Leptospira ( Figure 3B ).

We next utilized a mAb against a single WC1 variant, WC1-8 (recognized by mAb CACT21A). WC1-8 is a leptospira-binding variant of WC1 expressed by cells within the WC1.1⁺ subpopulation. These specific cells are known as WC1.3⁺ and express the WC1-3 variant in addition to WC1-8. FRET occurred between WC1-8 and γδ TCR on cells cultured with Leptospira ( Figure 3C ) although it was less than occurred when the anti-WC1.1 mAb BAG25A (which reacts with WC1-3) or anti-pan-WC1 mAb CC15 (that is a pan-anti-WC1 mAb) was used. Interestingly, because mAb CACT21A was against a single WC1 variant (WC1-8), small microclusters of γδ TCR and WC1 were more distinct than WC1 clusters in experiments using the anti-pan-WC1 mAb CC15 (see Figure 1C ). Quantitation showed that in both experiments performed there was consistent FRET between γδ TCR and WC1 on cells within the WC1.1⁺ and WC1.3⁺ subpopulations but not in the WC1.2⁺ subpopulation ( Figure 3D ).

We then evaluated whether the different WC1 variants act similarly to CD4 and cluster together despite their amino acid sequence and structural differences. To determine this, we evaluated FRET on a population of cells that are known to express just two specific variants of WC1 molecules. These cells are known as WC1.3⁺ γδ T cells, expressing WC1-8, recognized by mAb CACTB21A, and WC1-3, recognized by mAb BAG25A. There was no FRET between WC1 variants on the WC1.3⁺ cells when evaluated in their quiescent state ( Figure 4A ). Separation of these different WC1 variants on quiescent cells was affirmed with STORM using WC1.3⁺ flow cytometrically sorted cells ( Figure S4 ). However, their pre-clustered protein islands containing either WC1-8 or WC1-3 become juxtaposed with one another after Leptospira stimulation ( Figure 4A ). FRET occurred between these WC1 variants in both of the experiments performed ( Figure 4B ).

Cell division of γδ T cells stimulated with Leptospira antigen occurs by day 5 of culture (34) ( Figure 5A ). Using this as a guideline, we evaluated the temporal clustering of cell surface molecules following culture with Leptospira. There was no profound FRET above baseline observed between CD45 and γδ TCR throughout any of the timepoints ( Figure 5B ), while γδ T cells showed significant clustering of γδ TCR-CD3 complexes with FRET by day 3 ( Figure 5B ) and significant FRET between WC1 and the γδ TCR was measurable on day 7 when either a pan-anti-WC1 mAb ( Figure 5B ) or WC1 variant-specific mAbs ( Figure 5C ) were used. While we measured some cell division by day 5 of culture, additional cell proliferation had occurred by day 7 ( Figure 5A ) which agreed with the results showing WC1 and γδ TCR had increased FRET at day 7 ( Figures 5B, C ). Overall clustering of γδ TCR/CD3 complexes (as measurable by AMNIS) occurred earliest (day 3) before clustering of WC1 with the γδ TCR (day 7) in all the replicate independent experiments performed.

We next evaluated direct interaction of Leptospira with WC1 molecules using STORM. Since most studies have employed sonicated bacteria, we wanted to ensure γδ T cells cultured with whole Leptospira proliferated in recall responses; we found they did so with responses being even greater than with sonicated bacteria ( Figure S5A ). To determine whether leptospires bound to WC1 on γδ T cells we used biotinylated Leptospira ( Figure 6A ). When incubated with PBMC we found that the WC1⁺ cells bound Leptospira ( Figure 6B ) with one or more leptospires as shown by AMNIS imaging flow cytometry ( Figure 6C ). To determine whether WC1 and Leptospira were juxtaposed on the cell surface STORM imaging of flow cytometrically sorted WC1⁺ cells with bound bacteria was done ( Figure 6D ). There was some colocalization between WC1 and the bacteria evident as indicated by yellow overlay ( Figure 6E ). The bacteria were more difficult to image as shown by their discontinuous appearance (see Figure S5B , AMNIS images of bacteria) due to their size which was considerably larger than the protein islands in quiescent cells.

Leptospira is known to bind to a variety of cells as well as to extracellular matrix through adhesins (35). To ensure that binding to γδ T cells was due to the interaction with WC1 the bacteria were incubated with lymphocytes before and after the lymphocytes were coated with mAb to block cell surface molecules including several anti-WC1 mAbs ( Figure 7A ). We found that mAb against WC1, when used before incubating the cells with the Leptospira, blocked binding of the bacteria to the cell ( Figure 7B ). This varied from experiment to experiment (4 independent experiments performed) due to variation in the level of bacterial binding. However, while blocking with anti-WC1 mAbs was consistent in all 12 evaluations no consistent blocking by mAbs that bind to TCRδ, CD4, or CD8 occurred ( Figure 7C ). The anti-pan-WC1 mAb CC15 had the best blocking ability, blocking nearly 70% of Leptospira binding and the blocking was statistically significant.