Endothelial colony-forming cells from patients with paroxysmal nocturnal haemoglobinuria are not PIGA mutated


Correspondence: Dominique Helley, M D, Ph D., Service d'Hématologie Biologique, Hôpital Européen Georges Pompidou, 20-40 Rue Leblanc, 75015 Paris, France. E-mail: dominique.helley@egp.aphp.fr

Paroxysmal nocturnal haemoglobinuria (PNH) is a rare acquired disease due to somatic PIGA mutations of the haematopoietic stem cells, which result in the absence of glycosyl phosphatidyl inositol (GPI)-anchored proteins and lead to haemolytic anaemia. Splanchnic vein thrombosis (more than 45% of thrombotic episodes) is a frequent complication and accounts for the majority of deaths (Ziakas et al, 2007; Peffault de Latour et al, 2008). The underlying mechanisms of these thromboses remain unknown. However, we have previously shown that endothelial cells are activated in PNH (Helley et al, 2010), but we could not exclude that this was due to the PIGA mutation. Interestingly, it was reported that progenitor endothelial cells (endothelial colony-forming cells, ECFC) from patients with myeloproliferative disorders and a history of thrombotic events bore the JAK2V617F mutation, which suggested a common precursor of the endothelial and haematopoietic stem cell (Teofili et al, 2011), although others studies did not find it (Yoder et al, 2007; Piaggio et al, 2009).

To examine the possibility that PIGA mutations responsible for PNH may affect a common endothelial and haematopoietic progenitor cell and thus induce endothelial dysfunction, we tested ECFC obtained from peripheral blood, as surrogate of endothelial cell genetic status.

Blood samples from 22 patients with PNH were collected with the patients' written informed consent. All experimental procedures conformed to the Helsinki Declaration of 1975, as revised in 2000, and the protocol was approved by the local ethics committee.

Peripheral blood mononuclear cells (PBMC) and polymorphonuclear neutrophils (PMNL) were fractionated on Pancoll and then dextran density gradients. PMNL were used to isolate DNA, while PBMC were placed on gelatine-coated plates for culture until ECFC appeared (usually 15–20 d) and then cultured to perform a study of GPI-anchored proteins (CD55, CD58 and CD59) by flow cytometry, and to isolate DNA. PIGA sequencing was restricted to PBMC DNA of subjects whose mononuclear cells culture gave rise to ECFC. Mutations were identified by sequencing the coding exons (2–6) of PIGA and their flanking regions by using cycling sequencing.

The mononuclear cells culture gave rise to ECFC in 11 patients out of the 22 PNH patients enrolled (50%), but were obtained in sufficient amount to perform both DNA isolation and GPI-anchored protein study in six patients only, and DNA isolation only in two others. The characteristics of the patients tested for PIGA sequencing are depicted in Table 1. This frequency of 50% was similar to that reported by others using a similar method in studying healthy subjects (Teofili et al, 2011). No particular haematological or clinical features, including a history of thrombosis, were associated with successful ECFC culture from PNH patients. The mean age was 47 (range 24–64) years. Mean white blood cell count, platelet counts and haemoglobin were 4·65 × 109/l, 148 × 109/l and 104 g/l, respectively. Mean lactate dehydrogenase level was 490 U/ml.

Table 1. Characteristics of PNH patients, and results of GPI-anchored proteins expression on ECFC and PIGA sequencing from PMNL and ECFC
PatientGenderAge (years)PNH granulocytes clone size (%)History of thrombosisGPI-anchored proteins expression on ECFCPIGA mutation
  1. PNH, paroxysmal nocturnal haemoglobinuria; GPI, glycosol phosphatidyl inositol; PMNL, peripheral blood neutrophils; ECFC, endothelial colony-forming cells; BCS, Budd-Chiari syndrome; DVT, deep vein thrombosis, ND, not determined.



p.48 Gly>Asp

2M4093BCS + DVTnormal expression of CD55, CD58 and CD59

Del g.3976-3978GTT

Del p.118Val

3F4091Nonormal expression of CD55, CD58 and CD59Del g.10480, 10481 or 10482C p.311Thr>Leu, Stop at 329none
4F2499Nonormal expression of CD55, and partial or total lack of CD59 and CD58 expression



5F5898Nonormal expression of CD55, CD58 and CD59Del g.4158, 4159 or 4160A p.179Asn>Thr, Stop at 194none
6F5672BCSnormal expression of CD55, and partial or total lack of CD59 and CD58 expressiong.3562 g>c (ivs1)none


p.X Gly>Glu

8F6494Strokenormal expression of CD55, and partial or total lack of CD59 and CD58 expressiong.10891 g>a (ivs5)none

Endothelial profile of the cultured ECFC defined as CD146+ /CD45- cells was confirmed as previously described (Helley et al, 2010). For the GPI-anchored protein study, 104 CD146+ /CD45- events (Fig. 1A) per sample were analysed on a Navios flow cytometer and using Kaluza software (Beckman Coulter). All ECFC from three of six patients (Patients 2, 3 and 7) showed a normal expression pattern of CD55, CD58 and CD59 (Fig. 1B). All ECFC from the other three patients (Patients 4, 6 and 8) showed normal CD55 expression, while some ECFC had a partial or total lack of CD59 and CD58 expression (Fig. 1C). This suggested that GPI-anchored protein deficiency might also affect endothelial cells in PNH. Overestimation of protein expression due to the presence of cell doublets was ruled out by eliminating such doublets on the Forward Scatter (FSC) Int/FSC peak dot-plot during analysis. Besides PIGA mutations, the different profiles of GPI-anchored protein expression could have been due to ECFC being at different stages of maturation. Indeed, Hernandez-Campo et al (2007) showed that the expression of most GPI-anchored proteins varies during normal haematopoietic differentiation.

Figure 1.

Examples of flow cytometric analysis of ECFC. (A) ECFC were characterized in dot plots (SSC/CD146) by their CD146 positivity (I) and their CD45 negativity (CD146/CD45) (II), after exclusion of cell doublets. GPI-anchored proteins CD59, CD58 and CD55 expression by CD146 + /CD45- cells from PNH patients with normal (B) (Patient 2) and abnormal (C) (Patient 8) expression profiles. Shaded peaks represent cells stained with antibodies against specific proteins; non-shaded peaks represent corresponding isotype controls.

We identified a single mutation in the DNA issued from PMNL of the eight patients (Table 1). All had a potentially deleterious effect, as they are predicted to result in a shortened (and therefore non functional) protein, or affected amino acids perfectly conserved among 30 species. None of these mutations were found in the ECFCs of the corresponding patients (Table 1). These results suggest that ECFC do not carry PIGA mutations in patients with PNH, although we cannot exclude the possibility that a colony of normal ECFC was selected in culture. Indeed, some GPI-anchored proteins, such as UPAR (urokinase-like plasminogen activator receptor) (Basire et al, 2006) or CD177 (Dillon et al, 2008), have been shown to favour the proliferation of ECFC, and thus, their lack on PNH cell membranes could impair cell growth and expansion.

In conclusion, we showed that PIGA mutations are not present in ECFC from patients with PNH, suggesting that the mutation occurs in haematopoietic stem cells and not in a putative common precursor. Thus, the possible role of endothelial cells in PNH-associated thrombosis could be due rather to their activation than to their PIGA mutation itself. Further studies are needed to identify the local factors responsible for splanchnic venous thrombosis in PNH.


The authors reported no potential direct conflicts of interest.


This work was supported by an unrestricted grant from France HPN, a non-profit association. S. Gandrille was supported by a grant (Contrat d'Interface Inserm/AP-HP) from Institut National de la Santé et de la Recherche Médicale (INSERM) and Assistance Publique Hôpitaux de Paris (AP-HP). The authors thank Angéline Courquin, Noël Lucas (Clinical Research Unit, Georges Pompidou European Hospital) and Véronique Remones (Department of Biological Haematology, Georges Pompidou European Hospital) for their helpful assistance.