However, our data demonstrated that none of the epidi-dymosomes major membrane-bound proteins was present in vivo on cauda epididymal or ejaculated sperm extracts. Thus, if these vesicles are involved in sperm membrane transformations, it is via very subtle exchange mechanisms that remain to be explained but not by a vesicle-fusion mechanism. Moreover, the epididymal exososomes represent only a small percent of the total cauda protein, and the most hydrophobic proteins were mainly present in the caudal fluid bulk phase under a soluble form.
Indeed, these vesicles are rich in GPI anchored proteins such as CDw52, CD55, CD59, prion protein, and p25b, and it has been shown that, in vitro, the vesicles can fuse with sperm. Immuno-blotting revealed that several new hydrophobic and lipophilic compounds were bound with the epididymal vesicles. We observed the presence of the very hydrophobic protein 17 kDa secreted by the cauda epididymis and of PGDS secreted in the caput epididymis. One form of E-RABP is also associated with the vesicles, although several different isoforms are secreted in the epididymis.
Destruction of these large vesicles could give rise to smaller vesicles similar to exosomes. Other possible origins include the confluence of smaller vesicles coming from the upper tract (which may explain the actin results we have obtained). However, our findings that these vesicles have a specific protein composition, which is different from the mature sperm and to a large extent from cytoplasmic droplets, and contain proteins that are only present in the cauda region strongly suggest that they are mainly the result of a specific secretion from this region.
All this suggests that epididymal and seminal plasma vesicles may be members of an exosome-vesicle family. This hypothesis is also supported by the fact that the different proteome of the exosomes published to date also contain similar proteins to epididymal and seminal plasma vesicles. It has been suggested that exo-somal vesicles are secreted from cells upon fusion of mul-tivesicular endosomes with the cell membrane by an unknown mechanism.
Moreover, almost all proteins identified in ram epididymal vesicles have also been found in a recent pro-teomic profiling study of human seminal plasma vesicles, and one protein, aldose reductase, has also been described in bull epididymal vesicles (see also Table 2). This suggests that, as in the ram, a large proportion of the vesicles found in seminal plasma in humans (and certainly in bulls) are derived from the epididymis. Alternatively, these different types of vesicles may be shed from the cell membrane by a very similar mechanism, including the same membrane domains containing the same proteins.
Our results demonstrated clearly that membranous vesicles are normal constituents of the cauda epididymal fluid and that, although a small number of vesicles originate from the other accessory glands, this fluid is the main source of the vesicles retrieved in the ram seminal plasma. This is demonstrated by the very similar one-dimensional and two-dimensional SDS-Page protein profiles and the protein compositions found for the epididymal and seminal plasma vesicles of normal rams and the absence of these vesicles in the seminal plasma of vasectomized rams.
The 43-kDa band showed varying intensity depending on the samples, and a less reactive band at 55 kDa could also be observed.
The 35-kDa band was decreased on the percoll-washed sperm extract but was highly enriched in cytoplasmic droplets while the other bands remained at the same level (Fig. 7). This 35-kDa band was also found in the CEF and in the vesicles.
The previous samples and sperm extracts from zones 2 and 6 were probed with the different antibodies reacting with the vesicle proteins (Fig. 8). The anti-phosphodiesterase and anti-neprilysin antibodies showed reactive bands in the cauda fluid containing the vesicles, but no reaction was obtained with the cauda sperm extract or cytoplasmic droplet. The neprilysin antibody reacted with the caput sperm extract but not with corpus and cauda sperm extracts, but the neprilysin form observed in the fluid showed a lower molecular weight (less than 120 kDa; see Fig. 8, compare lanes Z2 and CEF).
As a control, an equal volume of fluid from the cauda containing both the sperm and the droplets was centrifuged and washed only with PBS.
The PBS- and percoll-washed cauda sperm, the cytoplasmic droplets, the caudal fluid and the vesicles pellet were run on SDS-PAGE (Fig 7A). Almost no difference was observed in the protein pattern between the two caudal sperm preparations (first and second lanes) and the sperm extracts from zones 2 and 6 (not shown).
We therefore analyzed the behavior of other vesicle proteins. We observed that the amounts of E-NNP3, neprilysin, lactadherin-PAS 6/7, 35-kDa G-beta in zone 7 were strongly decreased after high-speed centrifugation (Fig. 6C), confirming that these proteins are associated with the vesicles as soon as they are present in the fluid.