HIV (red arrows; tiny circles) occupies vacuoles that are continuous with but separated from the extracellular space.

HIV assembles in a compartment that lies deep within macrophages yet is continuous with the extracellular environment, report Deneka et al. (page 329).

HIV forms membrane-encapsulated particles that assemble at, and bud off from, the surface of infected T cells. In macrophages, however, the majority of virus particles are found intracellularly and have been reported to associate with structures containing an endosomal marker called CD63.

Deneka and colleagues' data, however, indicate that HIV does not enter macrophages by endocytosis. Upon HIV infection, the virus did not colocalize with any other endosome markers besides CD63. It instead associated with a newly identified vesicular structure marked by the transmembrane receptors CD81, CD9, and CD53.

The CD81/9/53 structures were shown by electron microscopy to be connected to the outside of the cell by narrow channels that are too small for HIV to pass through. By being essentially outside and inside at the same time, HIV would be simultaneously protected from the acidic digestive environment of endosomes and lysosomes and from the humoral immune system.

The authors propose that these structures might also allow for a rapid release of virus particles for their transmission to other cells. They are currently trying to determine the function of the CD81/9/53 compartments in uninfected macrophages.

In uninfected macrophages, endosomal CD63 did not colocalize with these CD81/9/53 compartments. So why does it accumulate there upon HIV infection? CD63 can be incorporated into the membrane capsule of HIV. The authors suggest that CD63 might normally traffic in small amounts throughout internal membranes and get swept up by HIV into CD81/9/53 structures.

Two other recent reports agree that HIV particles bud off from plasma membrane in macrophages rather than being associated with endosomes (Jouvenet et al. 2006. PLoS Biol. 4:e435; Welsch et al. 2007. PLoS Pathog. 3:e36). Now it is shown that this plasma membrane region is both intracellular and extracellular at the same time.