Tissue factor purified from different cellular sources and non-glycosylated tissue factor exhibit similar procoagulant activity

Tissue factor is a transmembrane glycoprotein which contains four potential N-linked glycosylation sites. Of these, three sites located in the TF protein’s extracellular domain (amino acid residues 11, 124 and 137) possess carbohydrate residues. There are conflicting reports on whether TF glycosylation regulates TF coagulant activity (see rev [1]). The observation that both E.coli and 293 cell produced rTF exhibit similar functional activity suggested that glycosylation is not required for TF procoagulant activity [2]. In contrast to these data, a recent study by Krudysz-Amblo et al. [3] showed that activities of the natural placental TF (pTF), recombinant TF (rTF) from Sf9 insect cells and rTF from E. coli were different and that placental TF was more active than others. They also reported that in vitro enzymatic removal of the carbohydrates from pTF under native conditions led to a pronounced reduction in the catalytic efficiency of pTF-FVIIa activation of FX [3]. However, our recent studies involving cell model systems showed that glycosylation is not required for either TF expression or functional activity at the cell surface [4]. At present, the basis for the contrasting data in the above studies is unknown. It is possible that TF incorporated into liposomes may behave differently from that of TF embedded in the cell membrane in terms of the requirement of carbohydrates for its optimal coagulant activity. Alternatively, the carbohydrate composition of rTF expressed in mammalian cells could be different from that of carbohydrates on natural placental TF. Therefore, the present study was carried out to investigate the procoagulant activity of wild-type (WT) TF and TF mutant lacking carbohydrates in a purified, reconstituted system. Additionally, we also analyzed the procoagulant activity of natural TF purified from human placenta and MDA-MB-231 breast cancer cells.

WT-TF and TF glycosylation mutant protein (TF_N11/124/137A) devoid of glycosylation at all three sites were over expressed by adenoviral infection of human umbilical vein endothelial cells (HUVEC) using the pacAd5 CMVK-NpA Shuttle vector (Cell Biolabs, San Diego, CA, USA) containing the WT-TF or TF_N11/124/137A cDNA. WT-TF and TF_N11/124/137A over expressed in HUVEC and TF from MDA-MB-231 cells were purified by affinity chromatography using anti-human TF antibody coupled to Affi-Gel-10 agarose [5]. Placental TF was extracted from acetone powder of human placenta and purified in a similar manner as mentioned above. TF protein concentration was estimated in an ELISA [4]. Purified TF protein was reconstituted in PC/PS (75%/25%) liposomes as described earlier [6]. Immunoblot analysis and molecular weight determination based on pre-stained Precision Plus Protein standards (Bio-Rad) (Fig. 1A) indicated that TF from MDA 231 cells, placenta, and WT-TF from HUVEC migrated as ~48 kDa monomeric protein under both reduced and non-reduced conditions. The glycosylation mutant protein showed increased mobility and had a molecular weight of ~ 38 kDa, which is in close agreement to the previously described molecular weight for the deglycosylated TF protein [3]. As reported earlier [4], TF glycosylation mutant showed reduced reactivity to TF antibody upon SDS denaturation. However, it is important to note here that the same TF antibodies recognize TF glycosylation mutants and WT-TF with equal avidity at the cell surface and in non-denaturing detergents, so could be used for determining the levels of wild-type and mutant TF at the cell surface and in ELISA [4]. A minor lower molecular weight band (~41 kDa) seen in both wild-type and the glycosylation mutant TF purified from HUVEC most likely represents partial post-translational modification of TF other than the glycosylation.

Fig. 1.

Glycosylation does not affect TF procoagulant function. Purified TF (5 nM) was incubated with 50 µM PC:PS (75% PC; 25% PS) in Hepes buffer (10 mM N-2-hydroxyethylpiperazine-N’-2-ethanesulfonic acid and 150 mM NaCl, pH 7.5) containing octylglucoside (33 mM) at 37°C for 45 min followed by extensive dialysis in HEPES buffer at 4°C to remove detergent. TF antigen in the liposomes was re-estimated by ELISA to correct for any minor changes in protein concentration associated with slight volume changes upon dialysis. (A) Immunoblot analysis of affinity purified naturally occurring TF protein from placenta, MDA-MB-231 breast cancer cells, and wild-type and TF glycosylation deficient mutant (TF_N11/124/137A) over expressed in HUVEC. 0.25 ng of relipidated TF protein was separated on 12% reducing or non-reducing SDS-PAGE gel and the blot was probed with TF polyclonal antibody (3 µg/ml). TF-FVIIa activation of FX was determined by either incubating 10 pM relipidated TF with varying concentrations of FVIIa (1– 100 pM) (B) or incubating 10 pM FVIIa with varying concentrations of TF (1–100 pM) (C) for 5 min, followed by FX (175 nM) for 10 min. At the end of 10 min, the reaction was terminated by adding Tris buffered saline containing 10 mM EDTA. FXa generated was measured in a chromogenic assay using S-2765 chromozyme substrate. (□) MDA TF; (▪) Placental TF; (●) Wild-type TF from HUVEC; (○) TF_N11/124/137A from HUVEC. Data are mean ± SD (n = 3).

Next, we analyzed TF-FVIIa activation of factor X (FX) of the reconstituted TF under limiting TF concentration (10 pM). As demonstrated in Fig. 1B, analysis of FVIIa interaction with TF using varying concentrations of FVIIa showed no significant differences in the rates of FX activation supported by TF purified from various sources. The TF glycosylation mutant showed slightly lower TF activity compared to its wild-type counterpart, but its activity was very similar to that of pTF. Analysis of the data by two-way ANOVA showed that the row means of the amount of FXa generated at individual FVIIa concentrations did not differ significantly among TF proteins from different sources. All TF preparations bound FVIIa with very high affinity (K_D, ~2 to 3 pM range).

Next, we performed FX activation using limiting FVIIa (10 pM) and increasing concentrations of TF protein. As shown in Fig. 1C, FVIIa bound to various TF preparations produced near identical rates of FX activation. Notably, the minor difference seen in FX activation between WT-TF and TF_N11/124/137A under the limiting TF condition was not evident when the limited concentration of FVIIa was saturated with increasing concentrations of relipidated TF. Overall, our data indicate that TF lacking carbohydrates is as efficient as WT-TF in activating FX. A minor difference seen in FX activation by WT-TF and TF_N11/124/137A in Fig. 1B was negligible, and probably may reflect minor variations in the effective concentration of fully folded TF. Here, it is pertinent to note that we obtained very similar specific activities of the wild-type TF and TF_N11/124/137A in the total cell lysates used for purification. This, combined with the above data showing similar activation of FX by purified and reconstituted WT-TF and TF_N11/124/137A preparations rules out any possibility that alterations in the integrity of TF protein, either WT-TF o r TF_N11/124/137A, during the purification procedure is responsible for the lack of difference between the wild-type and mutant TF in their coagulant activities. In additional studies, we attempted to evaluate the role of carbohydrates on TF activity by enzymatic removal of carbohydrate. Our multiple attempts of enzymatic removal of carbohydrates from placental TF under native conditions using PNGaseF from different sources were unsuccessful. Although we could deglycosylate TF under denaturing conditions with PNGaseF, the data obtained from these studies was inconsistent and unreliable as the denaturing conditions used for removal of carbohydrates inactivated TF even in the absence of PNGaseF, and prolonged incubation with PNGaseF appeared to result in degradation of TF protein.

Earlier studies [7,8] that showed importance of carbohydrates for TF procoagulant activity using chemical inhibitors should be interpreted cautiously. Concanavalin A, a lectin that binds to carbohydrates, was shown to inhibit TF activity [7]. However, concanavalin A, being a multivalent molecule, can form large aggregates and intra-molecular bridges that can sterically interfere with FVIIa binding to TF [7]. Similarly, tunicamycin, an inhibitor of N-linked glycosyl reactions, which was shown to inhibit TF activity in a cell system [8] also inhibits the synthesis of TF protein [4].

In summary, the present data confirm our earlier study using the cell systems [4] that showed carbohydrates are not essential for TF procoagulant function. The present study also indicates that the reported difference on the importance of carbohydrates for TF coagulant activity in earlier studies [3,4] could not have come from potential differences between cell and reconstituted systems. It is possible that failure to properly insert in the lipid bilayer of some of TF molecules lacking carbohydrates and/or other unintended modifications in TF by PNGaseF could have been responsible for the observed lower procoagulant activity for TF lacking carbohydrates in the earlier study [3]. Our present data showing similar coagulant potential of TF from varied sources, ranging from endothelial cells to cancer cells, known to have altered/aberrant glycomic phenotype[9]. and natural TF from placenta further strengthens our conclusion that carbohydrates do not influence TF activity. It is important to note that the observation that TF purified from various tissues/cell systems exhibits a similar coagulant activity in the reconstituted system does not mean that TF expressed on various cell surfaces also exhibits a similar specific activity. TF activity at the cell surface is highly dependent on the phospholipid content and composition of the cell membrane, which could vary in various cell types, and therefore may exhibit different specific activities.