HEADLlNES -

with nocodazole, and DHC2 appeared to redistribute with a-mannosidase II to the endoplasmic reticulum (ER) after treatment with brefeldin A. DHC2 is, therefore, a strong candidate for a factor involved in Colgi dynamics.

It is known that, during mitosis, membrane traffic is halted. It has been proposed that membrane binding bydynein may be cell-cycle regulated such that dynein does not bind mem- branes during mitosis. The report from Niclas et al. shows that this is the case and demonstrates the mechanism by which this occurs. Membranes isolated from metaphase- arrested Xenopus cells were found to have sevenfold less dy- nein associated, whereas total dynein in crude extracts re- mained unchanged. Furthermore, phosphorylation of the DLIC was increased elevenfold in metaphase extracts, whereas very little phosphorylation of the DHC or DIC was detected. Phosphorylation of the DLIC is clearly implicated, therefore, in the regulation of the association of dynein with membranes.

Thus, these two reports provide new insights into the regu- lation of dynein and its role in membrane traffic. Cell-cycle- dependent membrane association allows dynein to fulfil roles in both membrane traffic and mitosis, and different DHC isoforms may provide the necessary kinetic diversityforvesi- cle and organelle movement in membrane traffic pathways.

‘( 4= The septins that squeeze us

FIELD, C. M., AL-AWAR, O., ROSENBLATT, j., WONC, M. L., ALBERTS, B. and MITCHISON, T. J. (1996)

A purified Drosophila septin complex forms filaments and exhibits CTPase activity

/. Cell Biol. 133, 605-616

Cytokinesis, the final landmark in the cell cycle requires the action not only of actin but also of a group of proteins called septins. The first members of this family to be characterized were the neck filament proteins, discovered in budding yeast and so named for the IO-nm filaments visible by elec- tron microscopy at the bud neck (the site at which the septins are found). These filaments disappear when a gene encoding any one of the four yeast septins is mutated. Are the septins components of the filaments or, rather, do they regulate their existence?

Field et al. use an antibody affinity column to isolate a complex of three previously characterized Drosophilaseptins that spontaneously, and in the absence of any other proteins, assembles into filaments of 7-9 nm in diameter. Data from sucrose-gradient chromatography and gel-filtration chroma- tography are consistent with the septins associating as a heterotrimer of dimers. Judging from the length distribution of the filaments, each subunit is -26 nm long, which is ap- proximately the length that would be expected if all three septins lined up head to tail, based on the length of coiled- coil segments of proteins. The length distribution also sug- gests that subunits associate with an association constant that is independent of polymer length, but lateral associations between filaments (seen here at higher concentrations) could result in the whole complex showing nucleated polym- erization behaviour, which would help localize the complex.

All three septins have GTPase motifs, and Field et al. show that each septin polypeptide in the complex binds a single

molecule of guanine nucleotide that (for two of the three polypeptides) exchanges slowly and is hydrolysed rapidly. The nucleotide state may regulate interaction with plasma membrane receptors or it could control the polymerization state. Alternatively, other proteins or modifications of the sep- tins may be responsible for this latter property, and indeed one of the septins appears to have several forms that are modified differentially. This work identifies septins as the molecules to watch in cytokinesis, but it leaves many questions about mechanism and regulation waiting to be answered.

- r Development survives

integrin knockout

GARDNER, H., KREIDBERG, J., KOTELIANSKY, V. and JAENISCH, R. (1996)

Deletion of integrin al by homologous recombination permits normal murine development but gives rise

to a specific deficit in cell adhesion Dev. Ho/. 175, 301-313

lntegrins are receptors for extracellular matrix components such as fibronectin, collagens and laminins. In the case of laminin, a prominent non-integrin receptor named the dystroglycan complex has been described recently. The known receptors for fibronectin all belong to the integrin family. Two integrins, a4 and a5, have previously been in- activated in mice through homologous recombination, and, in both cases, this resulted in embryonic lethality. a4a5 double homozygous mutant mouse embryos have a phenotype that is additive with respect to knockout of a4 and a5 alone.

The al integrin forms a collagen and laminin receptor and displays dynamic regulation during embryonic develop- ment. Despite function and distribution data indicating that the al integrin is a promising target for gene inactivation, the resulting phenotype in al-null embryos described by Gardner et al. has no detectable defects. The lack of al in the examined embryonic locations is not compensated for by increased levels of the a2 integrin. Despite a lack of in vivo phenotype, cells isolated from al -deficient embryos show defects in cell adhesion and cell migration to col- lagen IV and laminin-1, but not to collagen type I.

The authors’ interpretation of the cell-attachment data is complicated somewhat by the different affinity of integrins in different cellular backgrounds. The authors have, for example, not tested the collagen-binding properties of al Bl -deficient hepatocytes (a cell type known to be criti- cally dependent on al81 for cell adhesion to collagen). From the distribution patterns of al 81 and a8Bl integrins, the authors suggest that cell adhesion to collagen is medi- ated via a8B1, which would bind fibronectin that, in turn, would act as a bridge to collagen. This assumption seems unlikely since the binding of fibronectin to native collagen is questionable. It will be informative to investigate poss- ible compensatory mechanisms in the mice deficient in al integrin. It will also be interesting to determine whether integrins that bind basement membranes, such as ~181, act in concert with non-integrins such as dystroglycans.

trench in CELL BIOLOGY (Vol. 6) August 1996