To the Editor:
The recent article by Perrault et al1 calls into question the entire series of observations that have been made by us2-5 and by others6 since 1988 relating to the endothelial glycoprotein (Gp) Ib complex. It is an irony of timing that their manuscript was accepted for publication 2 weeks before our article appeared in Blood,7 demonstrating the involvement of GpIbα in endothelial cell (EC) attachment to von Willebrand factor (vWF), particularly after exposure to the combination of tumor necrosis factor-α (TNF-α) and interferon-γ (IFN-γ), and was published only 2 weeks before our subsequent publication, again in Blood, demonstrating the presence of all four components of the GpIb complex on the surface of cultured EC and in vessels in vivo.8
Perrault et al state that they were unable to identify either GpIbα message or surface protein in EC, using Northern blotting of poly A+ mRNA (even after TNF-α treatment) and flow cytometry with several monoclonal antibodies (MoAbs) to GpIbα, respectively. They furthermore state that adhesion to vWF could not be inhibited by MoAbs to GpIbα. The reason for the discrepancy between their results and ours is difficult to ascertain. As already stated, the presence of GpIbα in/on EC, and its involvement in EC adhesion to vWF, has been reported by more than one laboratory. Moreover, the GpIb complex has been identified in EC by a large number of techniques—Northern blotting; cloning of EC GpIbα, GpIbβ, and GpIX, with complete sequencing of the latter two; surface labeling of HUVEC and adult human EC by MoAbs to GpIbα, GpIX, and GpV and by polyclonal antibodies to GpIbα and GpIbβ; and Western blotting of EC extracts with MoAbs and/or polyclonal antibodies to all four chains. One possible explanation for the negative results of Perrault et al with antibody probes is our observation that, even though several GpIbα MoAbs possess attachment-inhibiting ability, very few react strongly with EC by flow cytometry. With regard to their negative Northern blotting data, it appears that insufficient polyA+ mRNA may have been used to detect the admittedly weak GpIbα signal in untreated human umbilical vein endothelial cells. The lack of reaction after cytokine treatment is more difficult to explain, but it would have been helpful to see the degree of upregulation of ICAM-1 mRNA in their cultures.
The involvement of the EC GpIb complex in adhesion to vWF, both in the absence of and, to a greater extent, in the presence of cytokines, has been demonstrated by inhibition studies with two different MoAbs to GpIbα and with the vWF-A1 fragment, containing the GpIbα binding domain.7 The lack of inhibitory effect on EC attachment of MoAbs to GpIbα observed by Perrault et al may perhaps be related to the relatively low attachment they observed (40%, v 65%-70% which we and most others obtain) and/or to the apparent absence of protease inhibitors and the low EDTA concentration of their lifting buffer. Finally, Perrault et al state that conflicting data concerning the expression of GpIbα mRNA in endothelial cells after stimulation with TNF-α have been reported, and refer to three of our publications. However, there is no disagreement in the results presented in these three publications, all of which report upregulation of GpIbα by TNF-α. Furthermore, they misquote us7and DeFilippi et al9 when they state that TNF-α alone downregulates αvβ3—both we and DeFilippi clearly indicate that only the combination of TNF-α and IFN-γ downregulates αvβ3.
With regard to our divergent observations concerning the GpIb complex in EC, we look forward to a planned exchange of reagents with Perrault et al.
We thank Dr Shapiro and his colleagues for their interest in our study reporting on EC adhesion to vWF under static conditions.1-1Our purpose for many years has been to address the issue of a vWF interaction with the endothelial GPIb-V-IX complex from a vWF-oriented point of view. In a prior study, we used vWF proteolytic fragments and MoAbs against functional domains of vWF,1-2 whereas more recently we used mutated recombinant vWF (r-vWF) and took into consideration the importance of cell stimulation by TNF-α.1-1 Our results unequivocally show that, under our experimental conditions, cultured EC adhere to vWF in an αvβ3-dependent manner and suggest the absence of cooperativity between vWF RGD-containing region and an additional site of the A1 domain. We found quite interesting that Beacham et al agree with us upon the lack of inhibitory effect on EC adhesion to vWF by the anti-GPIbα MoAb 6D1.1-3,1-4 However, we find more puzzling the discrepancy concerning the effect of MoAb AS-7, kindly donated to us by Dr Jonathan Miller,1-5 which has been reported as an inhibitor of EC adhesion to vWF or its RGD mutant,1-3,1-4 but which we found devoid of inhibitory effect on EC adhesion to WT-rvWF or to RGGS-rvWF. Dr Shapiro and his colleagues state that a possible explanation for this discrepancy is to be found in the “relatively low attachment” that we obtain. However, we disagree that our values can be directly compared with their data since we express adhesion as a percentage of total cell number added to the well, and they report adhesion as percent of a reference value arbitrarily set to 100%.
Given the results, the level of expression of endothelial GPIbα protein and mRNA emerged as a necessary question to be assessed in our experimental conditions. We reported that these levels remained undetectable by a variety of techniques (flow cytometry, immunoprecipitation, and Northern blot) despite the use of stimulation conditions supposed to result in a significant increased level of GPIbα, which indeed were efficient as indicated by an increase in ICAM-1 expression. We acknowledge that flow cytometry has a limit of detection of ∼500 molecules per cell which could explain our failure to detect low amounts of a weakly expressed receptor. However, we also used indirect immunofluorescence without further success (personal unpublished data, March 1997). This is even more puzzling when considering the recently reported high number of GPIbα copies per EC (>300,000 per apparently unactivated cell), which could not have escaped our range of sensitivity.1-6 As outlined by Dr Shapiro and his colleagues in the above letter, there is a general agreement that untreated EC express a weak mRNA signal for GPIbα. Since expression of the endothelial GPIb-V-IX complex was not the main focus of our study, we decided not to continue along this line, and stated that GPIbα is not expressed to detectable levels in our EC cultures, whether activated or not. Clearly, our present findings do not call into question the observations of Shapiro and colleagues concerning the identification of endothelial GPIbβ, GPV, and GPIX.1-6,1-7
We are aware that Dr Shapiro and his coworkers have actually raised most observations reporting on the presence of the GPIb-V-IX complex on EC, with the exception of a report that included observations on smooth muscle cells.1-8 Based on their more recent results (that were indeed published 2 weeks after acceptance of our manuscript, and which we could thus not have misquoted1-4) it appears somewhat surprising that Dr Shapiro and his colleagues express their apparent disagreement with our findings. Indeed, they clearly agree with us that the function of GPIbα as a vWF receptor can be best evidenced when αvβ3 expression is downregulated, and is strongly dependent on a balance between both receptors.1-4 Therefore, the most likely explanation for the discrepancy between both functional studies is that our results experimentally favor the αvβ3-dependent pathway, whereas their results do not. An interesting objective will be to find physiopathological conditions resulting both in a downregulation of αvβ3 and an increased GPIb-dependent adhesion to vWF.