Fig 2: CD93 directly binds to MMRN2. (a) Diagrams of GFP-tagged CTLD group 14 family members showing domain architecture and relative size with number of amino acids (aa) without the GFP tag. C-type lectin domain (CTLD), sushi, epidermal growth factor (EGF) domain, transmembrane (TM), cytoplasmic tail (cyto) and green fluorescent protein (GFP) tag are displayed. (b) HEK293T were transfected with GFP-tagged group 14 family members, lysates were separated by SDS–PAGE under non-reducing conditions and far western blotted with MMRN2 full-length (MMRN2FL), showing MMRN2FL binds CLEC14A and CD93 but not thrombomodulin or CD248, probing with anti-GFP confirmed expression of all proteins. (c) Immunoprecipitations of CD93 using monoclonal R139 antibody co-immunoprecipitates MMRN2 from HUVEC lysates. (d) Immunoprecipitations of MMRN2 using mouse polyclonal antibodies co-immunoprecipitates CD93 from HUVEC lysates. CD93 was detected using goat polyclonal antibodies in each immunoprecipitation experiment. IgG heavy chains included as loading control.

Fig 3: CLEC14A binds MMRN2 between residues 530 and 624. (a) Diagrams of MMRN2 truncation mutants, showing the elastin microfibril interface (EMI) domain, coiled-coil (CC) domain and complement related C1q domain. The minimal CLEC14A-binding fragment is highlighted in black. (b and c) HEK293T were transfected with truncation mutants, lysates were separated by SDS–PAGE under reducing conditions and far western blotted with CLEC14A-ECD-Fc and western blotted with His tag antibodies. (b) The smallest binding fragment was identified as MMRN2487–820, MMRN2FL fails to bind CLEC14A-ECD-Fc. (c) Further truncation mutants were far western blotted revealing the smallest MMRN2 fragment binding CLEC14A is MMRN2530–624. (d) Purified MMRN2495–674 and MMRN2495–603 bind to streptavidin under reducing conditions after biotinylation. (e) Flow cytometry histograms of HEK293T transfected with GFP-tagged CTLD group 14 family members and stained with biotinylated MMRN2495–674 (grey line), confirming binding to CLEC14A and CD93 but not thrombomodulin. Streptavidin-PE alone was used as a control for background binding (grey shaded).

Fig 4: CD248 binds to MMRN2 in a distinct region from CLEC14A and CD93 binding. (a) HEK293T transfected with GFP, CLEC14A-GFP, CD248-GFP, and chimeras CLEC14ACD248(sushi) and CLEC14ACD248(CTLD) were lysed and far western blotted with MMRN2FL and western blotted with anti-GFP. MMRN2FL binds to CLEC14A and both chimeras CLEC14ACD248(sushi) and CLEC14ACD248(CTLD) but not CD248-GFP or GFP alone. (b) Immunoprecipitations of GFP-tagged proteins after cell surface biotinylation. CLEC14A, THBD, CD93 and all four CLEC14A chimeras bind to streptavidin showing they are cell surface expressed. GFP alone was included to demonstrate intracellular proteins are not cell surface biotinylated. (c) MMRN2 truncation mutants were transfected into HEK293T and lysates under reducing conditions were far western blotted with mCD248-ECD-Fc, revealing binding to minimal fragment MMRN2133–486. His tag western blot confirmed expression of each MMRN2 protein fragment. (d) mCD248-ECD-Fc pull-downs from HUVEC lysates resulted in enrichment of MMRN2 compared with hFc control. (e) Immunofluorescence analysis of HUVEC stained with MMRN2 mouse polyclonal antibodies, hFc, mCD248-ECD-Fc or CLEC14A-ECD-Fc. MMRN2 antibody staining partially co-localizes with CD248-ECD-Fc and CLEC14A-ECD-Fc binding; scale bar, 40 µm. (f) ELISAs of mCD248-ECD-Fc bound to plate capturing MMRN2 FL His, (**P<0.01 Mann–Whitney test, n=5) error bars represent s.e.m. (g) ELISAs of mCD248-ECD-Fc capturing MMRN2 FL His and then binding by CLEC14A-ECD-Fc detected by anti-CLEC14A antibody C2. (**P<0.01 Mann–Whitney test, n=5) error bars, s.e.m. (h) Diagram of CLEC14A or CD93 expressed by endothelial cells binding to MMRN2 in the ECM, which in turn is bound by CD248 expressed by fibroblasts or vasculature associated pericytes. CD248, CLEC14A and CD93 binding is due to the CTLD, CD248 binds MMRN2 in the region 133–486, whereas CLEC14A binds in the region 530–624.

Fig 5: Expression of CD93 in human cells and its response to LPS. (A) The level of CD93 transcription was evaluated by qRT-PCR and presented as the relative value after normalization to GAPDH. The transcription level of sCD93 in THP-1 cells was set as 1 and compared between different types of cells (n = 4 per group). (B) The level of CD93 translation was immunoblotted with GAPDH as a loading control and (C) the value of CD93 relative to GAPDH was presented with the value of THP-1 as 1 (n = 4 per group). (D) HUVECs were harvested at the designated time point after LPS treatment and immunoblotted for CD93 expression. ß-actin was used as a loading control. (E) The quantification of (D) was presented with the expression level at baseline set as 1 (n = 4 per group). (F) The localization of CD93 in the HUVECs was visualized by immunofluorescent staining with CD93, CD31, and DAPI after 48 h of 500 ng/mL LPS stimulation. Scale bar, 50 µm. (G) The supernatant of cultured HUVECs was harvested after 8 h or 24 h of 500 ng/mL LPS stimulation and the sCD93 level was measured by ELISA. All values are fold change compared to the value of 8 h without LPS stimulation as the control. The difference was compared between the PBS and the LPS group (n = 4 per group). Data are presented as the mean ± SE. * p < 0.05.