- Top of page
- Materials and methods
Previous studies demonstrated that 20% of haemoglobin is lost from circulating erythrocytes during their total lifespan by vesiculation. To study whether removal molecules other than membrane-bound haemoglobin were present in erythrocyte-derived vesicles, flow cytometry and immunoblot analysis were employed to examine the presence of phosphatidylserine (PS) and IgG, and senescent cell antigens respectively. It was demonstrated that 67% of glycophorin A-positive vesicles exposed PS, and that half of these vesicles also contained IgG. Immunoblot analysis revealed the presence of a breakdown product of band 3 that reacted with antibodies directed against senescent erythrocyte antigen-associated band 3 sequences. In contrast, only the oldest erythrocytes contained senescent cell antigens and IgG, and only 0·1% of erythrocytes, of all ages, exposed PS. It was concluded that vesiculation constitutes a mechanism for the removal of erythrocyte membrane patches containing removal molecules, thereby postponing the untimely elimination of otherwise healthy erythrocytes. Consequently, these same removal molecules mediate the rapid removal of erythrocyte-derived vesicles from the circulation.
Erythrocytes lose a substantial amount of haemoglobin during their lifespan. We have previously described the fractionation of erythrocytes using a method in which the first separation, based on volume, was combined with a separation using a density gradient, deploying the percentage HbA1c as marker of erythrocyte age (Bosch et al, 1992). Using this method it was demonstrated that in splenectomized subjects, erythrocytes lose 15% haemoglobin gradually during their lifespan in a linear fashion. In healthy subjects, an additional 5% of haemoglobin is lost during the second half of the erythrocyte lifespan (Willekens et al, 2003). This is in accordance with the studies using 14C/15N-glycine labelling of erythrocytes, that demonstrated a gradual and substantial haemoglobin loss of circulating erythrocytes as well as the continuous appearance of the label in the haemoglobin degradation products stercobilin and urobilin (London et al, 1950; Berlin et al, 1954). The erythrocyte-derived vesicles isolated from the plasma of freshly drawn blood contained all haemoglobin components in a pattern similar to that of old erythrocytes, in accordance with the increased haemoglobin loss in the second half of the erythrocyte lifespan (Willekens et al, 2003). Vesicles were also observed in erythrocytes of all ages and were shown to accumulate after splenectomy (Reinhart & Chien, 1988). In asplenia, vesicles or vacuoles that are larger than 300 nm can be observed as ‘pits’ in the smear. After splenectomy, these pits originate equally in young and old cells (De Haan et al, 1988), whereas the mean corpuscular haemoglobin of old erythrocytes remains relatively higher because of the increased absolute amounts of HbA1c and HbA1e2 (Willekens et al, 2003). On the basis of these findings we concluded that the removal of haemoglobin by vesiculation, including irreversibly modified haemoglobin, occurred during the total erythrocyte lifespan.
It has been calculated that there is not enough membrane in the erythrocyte as a whole to package 20% of the haemoglobin in the form of vesicles (V.L. Lew, personal communication). It must therefore be assumed that either the haemoglobin loss was somehow overestimated, or that the extra loss of surface area in comparison with the haemoglobin loss is compensated for by incorporation of lipids, probably originating from plasma lipoproteins (Reed, 1968; van Deenen & De Gier, 1974; Verkleij et al, 1976; Brossard et al, 1997). As we are not aware of an alternative route for haemoglobin to leave the circulating erythrocyte, we think the second explanation to be the most plausible.
In a rat model, almost 50% of erythrocyte-derived vesicles were rapidly removed from the circulation by Kupffer cells through a phosphatidylserine (PS)-dependent mechanism (Willekens et al, 2005). The same PS-dependent mechanism was responsible for the elimination of oxidatively damaged erythrocytes by Kupffer cells in a murine model (Terpstra & van Berkel, 2000). PS is considered to be an early sign of apoptosis in nucleated cells, proffering a binding site for macrophages. Indeed, PS exposure is enhanced in a substantial proportion of sickle red blood cells, and may contribute to the phagocytosis of these cells in vitro (Kuypers & de Jong, 2004). Also, the erythrocytes of thalassaemia patients show increased PS exposure (Kuypers & de Jong, 2004).
In addition, it is generally accepted that age-dependent removal of erythrocytes from the circulation is mediated by the binding of physiological autoantibodies (Kay, 1975, 1978), which form the trigger for the recognition and removal of senescent erythrocytes by macrophages. The senescent cell-specific autoantigens (SCA) originate on band 3, the anion exchanger and the major membrane protein of the erythrocyte (Kay, 1984a, 2004). It is speculated that SCA generation is triggered by the binding of denatured haemoglobin to the cytoplasmic domain of band 3, thereby inducing clustering and/or degradation of band 3. Both aggregation and degradation may lead to neoantigen formation (Kay, 1984b; Low et al, 1985; Kay et al, 1986; Schluter & Drenckhahn, 1986). It is uncertain whether, in normal erythrocyte ageing, activated complement or complement-regulating proteins play a role comparable with that in sickle cell anaemia and paroxysmal nocturnal haemoglobinuria (Lutz, 2004).
To be able to prevent premature loss of erythrocytes, either ex vivo during processing in the blood bank or in vivo in patients, it is important to elucidate the mechanisms that have an effect on the relationship between the generation of erythrocyte-derived vesicles and the process of erythrocyte ageing at the molecular level. In the course of our investigations in this area, we developed the hypothesis that generation of vesicles may not only serve to remove modified haemoglobin and its toxic derivatives, but may also help to eliminate premature removal molecules from otherwise functional erythrocytes. The effective disposal of such molecules would preserve the integrity of the erythrocyte and may expand its lifespan. To test this hypothesis, we explored the presence of putative removal molecules, such as band 3-related senescent cell antigens and PS, and the presence of IgG and complement-regulating proteins on erythrocyte-derived vesicles and on the membranes of erythrocytes of various ages.
- Top of page
- Materials and methods
In this study, the erythrocyte-derived vesicle concentration was found to be approximately 170 vesicles per microlitre, comparable with the number reported by others (Berckmans et al, 2001; Shet et al, 2003; Hron et al, 2007). The mean diameter of these vesicles was 0·5 μm (Willekens et al, 2003). These data enabled us to calculate roughly the dynamics of vesiculation, when the loss of 15–20% (15 μm3) of the erythrocyte volume occurred through vesiculation. Assuming a spherical shape, the vesicle volume was approximately 0·065 μm3. To lose 15 μm3, the erythrocyte has to produce, on average, 231 (15 μm3/0·065 μm3) vesicles over 120 d, or 2/d. A total erythrocyte mass of 25·1012, as can be found in a healthy male, produces 580·106 vesicles per second, which have to be removed at the same time, assuming an equilibrium condition. The total amount of erythrocyte-derived vesicles measured in the blood is 850·106 (170 × 5·106), indicating that it takes only 0·7 s to clear half of all the vesicles from the circulation. This is at least one order of magnitude smaller than one circuit around the circulation, and in large contrast with the calculated half-life of 25 s of 61%, and a half-life of 6·6 min of 30% of 51Cr-labeled erythrocyte-derived vesicles in a rat model (Willekens et al, 2005). The same experiments showed that 70% of the vesicles were cleared within 2 min, while at the same time 30% of the radioactivity was taken up by the liver, indicating that 40% was taken up elsewhere. Accordingly, after 30 min the distribution of radioactivity was 45% in the liver and 46% in other organs, namely, bone, skin, muscle, spleen, kidney and lung (23%, 10%, 6%, 3%, 2·5% and 1·5% respectively) (Willekens et al, 2005). This indicated that the liver and other organs were equally effective in removing erythrocyte-derived vesicles from the circulation. Taken together, these data suggest that most vesicles are taken up almost directly by the macrophages of the organ in which they originate before they can reach the venous circulation and be counted. Apparently, the body has developed an efficient mechanism to remove these vesicles that could be harmful, because of the coagulation-promoting activity of the PS-exposing vesicle membrane. The rate of clearance was substantially inhibited by a prior intravenous injection with PS-exposing liposomes in the liver, but not in the other organs, indicating a saturation of scavenger receptors by PS in Kupffer cells(Willekens et al, 2005). In the present study of human vesicles, a two-colour flow cytometric analysis employing anti-glycophorin A and annexin-A5 showed that the erythrocyte vesicles that expose PS all showed a high expression of glycophorin A (Fig 1). As a considerable proportion of these PS-exposing vesicles also contain IgG, it seems reasonable to assume a role for an immune-mediated mechanism that helps to maintain the overall quality of the erythrocyte population. As has been postulated before, clearance may be facilitated by PS-mediated phagocytosis (Wu et al, 2006), but IgG is also recognized by specific Fcγ-receptors on macrophages (Nimmerjahn & Ravetch, 2006), leading to phagocytosis.
Almost all vesicles carry the complement-regulating proteins CD55 and CD59. This observation strongly suggests the presence of a complement protection mechanism, and indicates that, under physiological circumstances, complement activation does not play an important role in vesiculation or vesicle clearance.
Vesicle formation appears to be accompanied by the breakdown of band 3. The present study demonstrated the presence of band 3-related proteins in vesicles that are formed in vivo. These band 3-related proteins were visualized with antibodies directed against membrane areas that are involved in senescent cell antigen activity (Table II, Fig 4). In addition, virtually no intact band 3 was found in the vesicles; proteins with a molecular weight of approximately 70 kDa or less dominated the immunoblot picture. In earlier studies, we observed that antibodies with a similar specificity only bind to their epitopes after these had been made accessible by proteolysis of band 3 (Renkawek & Bosman, 1995). Taken together, these data support the hypothesis that vesicle formation in vivo is associated with changes in the structure and/or breakdown of band 3, that probably result in an increased exposure of senescent cell antigen-related epitopes on vesicles.
Using a combined volume/density separation procedure that results in well-defined fractions (Bosch et al, 1992), we found that only a very small proportion of cells exposed PS. The numbers of these cells were slightly increased only in the fraction containing the youngest red blood cells, presumably as a consequence of the presence of reticulocytes that shed RNA and other cell components through exocytosis. The other fractions showed a remarkably equal percentage of erythrocytes exposing PS. This suggests the presence of a process independent of cell age. As vesiculation itself is independent of cell age and linked to the development of PS exposure, it is tempting to speculate that the PS-exposing erythrocytes are involved in the process of vesiculation. On average, 0·1% of erythrocytes are PS exposing. Assuming that the PS-exposure is a sign of vesiculation, one in every thousand erythrocytes is releasing a vesicle. As 1000 erythrocytes release 1340 (2000 × 0·67) PS-exposing vesicles per day or 0·93/min, this leads to the conclusion that the formation of a PS-exposing vesicle takes approximately 1 min.
Although others reported an increase in the proportion of PS-exposing erythrocytes in the most dense fractions (Shukla & Hanahan, 1982; Connor et al, 1994), it has to be emphasized that the density separation techniques used to obtain the fractions enriched with erythrocytes of old age yield only a heterogeneous fraction (Bosch et al, 1992). Our combination separation method proved superior in producing erythrocyte fractions of well-defined cell age (Bosch et al, 1992), which enabled us to conclude that the increase of PS exposure in the most dense fraction obtained by density separation alone is not because of the old cells in that fraction and that there are no data to support the hypothesis that PS exposure constitutes a causal factor in the physiological removal of old cells in vivo.
Theoretically, the ageing-related decrease in the amount of CD55 and CD59 on the oldest erythrocytes could foster an enhanced susceptibility to complement binding and activation. However, the net outcome of this ageing process does not result in an ageing-related decrease of the density of CD55 and CD59, as the 14% and 11% decrease in the concentrations of CD55 and CD59 molecules per red cell respectively, was accompanied by a simultaneous decrease in the mean surface area of about 16% (Bosch et al, 1994).
Starting from the same well-defined fractions as described before, we could confirm previous reports on the occurrence of structural and immunological changes in band 3 during erythrocyte ageing in vivo (Kay, 2005). These changes were especially apparent with antibodies reactive with epitopes in the N-terminal part of the membrane domain. These antibodies show an age-related increase in immunoreactivity, which leads us to conclude that the age-related changes in band 3 occur principally in regions that participate in the generation of senescent cell antigens. Subtle changes occurring in intact band 3 proteins may generate activity of senescent cell antigen; breakdown of band 3 may not be necessary for this process (Kay, 2005).
Taken together, the available data indicate that vesiculation is not only associated with the removal of membrane-bound haemoglobin, but also associated with generation of senescent cell antigen, a neoantigen that originates from band 3 after breakdown of band 3 in senescent red blood cells. The high percentage of erythrocyte vesicles that expose PS suggests that, similar to that found in rats, exposure of this molecule is an additional important factor for their rapid removal in humans. Binding of IgG, another removal molecule, is putatively associated with the presence of senescent cell antigens. The capability to remove damaged cell components is crucial for the viability of the erythrocyte, as this cell contains a restricted mechanism for self-repair and lacks the capacity of repair by renewal. It is concluded that vesiculation constitutes a mechanism for the removal of erythrocyte membrane patches associated with removal molecules, thereby postponing the untimely elimination of otherwise healthy erythrocytes. Consequently, these same molecules play a causative role in the rapid removal of erythrocyte-derived vesicles from the circulation. However, the present study has not determined why the protective mechanism eventually fails to prevent the death of the erythrocyte.