Results

Annexin A7 expression in red blood cells, red blood cell derived exovesicles and in platelets
Salzer et al. recently reported the presence of the 47 kDa isoform of annexin A7 and its partner sorcin in micro- and nanovesicles derived from red blood cells where they are located in the lumen and are also enriched in membrane rafts [7]. We have now extended these findings and show here that the 51 kDa isoform is present as well. In western blots of human red blood cells the 47 and 51 kDa isoforms were detected using mAb 203–217 (Fig. 1). The 47 kDa isoform was detected in silver stained gels and its identity with annexin A7 confirmed by peptide mass fingerprinting (data not shown). The 51 kDa isoform was only detected in western blots.
Figure 1  Expression of annexin A7 in human red blood cells (1) and human platelets (2). Protein homogenates were separated by SDS-PAGE (12 % acrylamide). The resulting western blot was probed with mAb 203–217, visualization was by a secondary peroxidase coupled antibody followed by ECL. Both isoforms are detected in red blood cells, in platelets only the small isoform is present. Its presence in vesicles led to the suggestion, that, along with raft domains, annexin A7 plays a role in membrane organization and the vesiculation process. To examine this, we analysed the ability of red blood cells derived from the annexin A7 knock out mice (anxA7-/-) to form exovesicles. Because of better accessibility and larger available amounts that led to clear results, we included our data obtained with human blood. Independent of the presence of annexin A7 both types of exovesicles, micro- and nanovesicles, were released after Ca2+/ionophore treatment as determined by acetylcholine esterase activity and microscopic examination. Annexin A7 is present in both vesicle types where it is more enriched in nanovesicles (Fig. 2). The 51 kDa isoform is only detectable in nanovesicles. The quantity of both vesicle types did not differ between wild type and knock out red blood cell vesicles as determined by the AChE-values (A405 nanovesicles: ~0.18; A405 microvesicles: ~0.9; both, for wt and ko). It appears that annexin A7, although it has been described to fuse membranes, is not a key component in the process of the formation of red blood cell exovesicles. When we probed for the presence of sorcin in wild type and mutant vesicles we found that the sorcin levels were reduced in the mutant nanovesicles (Fig. 2).
Figure 2  Enrichment of annexin A7 and sorcin in exovesicles derived from red blood cells. Vesicles from wild type (wt) and anxA7-/- mutant (ko) were generated with Ca2+/ionophore treatment, isolated by differential centrifugation and analysed by immunoblotting with mAb 203–217 and a sorcin polyclonal antibody. In general, annexin A7 and sorcin are more abundant in nanovesicles. The 51 kDa isoform of annexin A7 (arrows) is only observed in nanovesicles. Samples of both vesicles types were normalized according to their acetylcholine esterase activity. Human red blood cells (co) were used for control and normalized independently. To study the distribution of annexin A7 in red blood cells we used self forming iodixanol density gradients. Both isoforms were present in the soluble fraction as well as in the gradient fractions where they are assumed to be associated with membranes. These membranes are exclusively plasma membranes as red blood cells are free of any organelles. The distribution was however not homogeneous throughout the gradient. This could reflect the binding of annexin A7 to membrane subdomains that have different lipid or lipid/protein composition as discussed by Salzer et al. [7]. Likewise, sorcin does not exhibit a homogeneous distribution in the gradient. Moreover, it segregates into vesicles which are not associated with annexin A7 (Fig. 3). We also tested platelets for annexin A7 expression and detected the 47 kDa isoform (Fig. 1).
Figure 3  Association of annexin A7 and sorcin with distinct red blood cell plasma membrane fractions. Lysed human red blood cells were added to a self forming iodixanol density gradient. The density of the gradient increases from left (1.06 g/ml) to right (1.20 g/ml). Fractions were analysed by immunoblotting using mAb 203–217 and the sorcin polyclonal antibody.

The lack of annexin A7 changes red blood cell morphology and osmotic resistance
As annexin A7 is a component of red blood cell vesicles we studied the consequences of the loss of the protein in the anxA7-/-mouse for red blood cell morphology and osmotic resistance. In a standard 'blood smear' we did not note a deformation of the cells, however dark field microscopy revealed changes in shape and diameter of the anxA7-/-red blood cells. They had a statistically significant larger cell diameter of 6.0 μm compared to wild type with 5.7 μm (p = 0.015, n = 40), a remarkably lower emphasized central impression and a more flat shape (Fig. 4). No significant change in the mean corpuscular volume (MCV) was measured with the ADVIA 120 cell counter.
Figure 4  Dark field microscopy of red blood cells from anxA7-/- mutant (B) and wild type mice (A). The mean cellular diameters are 6.0 μm and 5.7 μm, respectively. The size differences are statistically significant (p = 0.015; n = 40). Bar, 3 μm. Shape and MCV of a red blood cell essentially influence its function and capillary passage. The typical biconcave form depends on various influences including the membrane lipid composition and the submembranous cytoskeleton [32] where annexin A7 might play a role. The osmotic resistance, which is the resistance towards changes in the extracellular ionic strength, is a convenient assay for analysis of the red blood cell integrity. It is measured as the sodium chloride concentration at which cellular lysis starts (minimal resistance) up to a complete lysis (maximal resistance) of the cells [33]. Osmotic resistance increases with higher MCV, larger surface area and a higher degree of cellular metabolism stabilizing the intracellular ion levels. Aged red blood cells show a lowered MCV and lower osmotic resistance [33]. Similar properties can be observed in red blood cells with lowered membrane permeability. The osmotic resistance of annexin A7 deficient red blood cells is significantly increased compared to wild type cells. The minimal resistance value of knock out cells is observed in a 0.60 % NaCl solution versus 0.65 % for wild type cells. 50 % haemolysis is achieved at 0.516 % NaCl solution for knock out versus 0.564 % NaCl for wild type cells (mean values, p = 0.00066, n = 8; Fig. 5). The resistance width of the knock out cells is slightly lower (0.20 % instead of 0.25 % NaCl solution). We also performed direct measurements of the red blood cell membrane deformability, to further characterize a possible role of annexin A7 on the membrane stability. With micropipette experiments we tried to correlate mechanical characteristics of the red blood cells with their morphology. We could not observe any statistically significant difference of values describing membrane rigidity and lysis force in these experiments (M. Heil, B. Hoffmann and R. Merkel, unpublished results). However, the distinction observed in the osmotic resistance experiments was highly significant and reflects a mean value over a high number of different red blood cells.
Figure 5  Osmotic resistance curves of red blood cells from wild type and anxA7-/- mutant mice. The osmotic resistance towards changes in the extracellular ionic strength is a convenient assay for analysis of the red blood cell integrity. It is measured as a sodium chloride concentration in which cellular lysis starts (minimal resistance) up to a complete lysis (maximal resistance) of a red blood cell. The osmotic resistance of annexin A7 deficient red blood cells is significantly increased compared to the one of wild type (p = 0.00066; n = 8). These data suggest that annexin A7 contributes to the shape of red blood cells and the osmotic resistance. As we have not observed a binding of annexin A7 to F-actin (data not shown) it could do so either by alteration of the membrane rigidity and/or by affecting the ion homeostasis.

The lack of annexin A7 affects primary haemostasis ex vivo
The use of an advanced electronic cell counter/flow cytometer (ADVIA 120) allowed us to screen other parameters as well despite of the limited murine blood volume. We analysed leukocytes, red blood cells, haemoglobin content, haematocrit, mean cellular volume, mean cellular haemoglobin, mean cellular haemoglobin concentration, platelets, neutrophiles, lymphocytes, monocytes, eosinophiles and basophiles. In these tests of whole blood we found only a change in the platelet numbers, whereas all other haematological parameters were not affected. The mean platelet counts from wild type and anxA7-/- mutant mice are determined as 674 × 103/μl and 774 × 103/μl, respectively. The platelet counts in knock out mice are significantly higher (p = 0.0275; n knock out = 11, n wild type = 14). An increase in platelet counts is a rather uncommon disorder. In humans, platelet counts normally increase only transiently as in reactive thrombocytosis and under postoperative conditions or are largely increased in neoplastic diseases. By contrast, the anxA7-/-mice are healthy and viable. Therefore we tested the platelet function and performed aggregometry measurements with platelet rich plasma. Platelet activation is observed as a morphological change from the resting discoid state to activated spherical cells with pseudopods. The morphological changes are due to cytoskeletal rearrangements. In vitro the activated cells form aggregates recruiting additional cells in the solution thereby reducing its cloudiness. Analysis of the transmission values of aggregation curves after platelet activation by ristocetin addition (von Willebrand cofactor) showed that both curves differed significantly at the time point of seven seconds after platelet initiation (p = 0.0287, n knock out = 16, n wild type = 15; Fig. 6). Annexin A7 deficient platelets showed a slightly lowered aggregation velocity. When we compared the aggregation curves of human platelets from a healthy donor with the ones obtained from an individual with a von Willebrand factor type 1 defect, we found that the difference in the curves was much more pronounced as observed in our studies of healthy mouse platelets and anxA7-/-platelets. Furthermore we found that murine platelets responded immediately to the initiating chemical whereas human platelets have a lag phase [23].
Figure 6  Platelet aggregation curves from platelet rich plasma of annexin A7 knock out and wild type mice. The transmission values were measured with an APACT photometer and aggregation was initiated by adding ristocetin. The first thirty seconds are given. Constant platelet counts were used throughout all experiments. The aggregation curve data were analysed for slope values, the maximal aggregation amplitude of every single curve was set to 100 %. Both murine curves differ significantly at a time point of seven seconds after platelet initiation, the mutant shows a slightly lower aggregation velocity (p = 0.0287, n knock out = 16, n wild type = 15, mean knock out = 67.7 % transmission, mean wild type = 77.1 % transmission, standard error knock out = 8.3 % transmission, standard error wild type = 13.8 % transmission). Murine platelets react immediately to the initiating chemical and show no lag phase like the normal human platelets. For comparison of the range of the platelet impairment slightly affected human platelets are shown as well (von Willebrand syndrome).