© 1987 Nature Publishing Group _N_AT_u_R_E_v_o_L_. _32_6_12_M_A_R_c_H_1_9s_7 _________ NEWS ANDVIEWS------------------•_ 33 reversals have VGP paths confined to the same band oflorigitude. In one of the best- documented cases', three reversals with related paths took place over a period of about 1.4 million years and then there was an abrupt.change. It is now possible to test Gubbins's model' in the face of the available data. The time taken for the reversal is so short that models relying on the decay of the dipole field must be discounted. The mod- el passes the test; it invokes the generation of negative flux, which eventually over- comes the initial field. It is also consistent with a relatively long period of low field intensity before and after the field polarity actually switches. While the negative flux is generated, the dipole field intensity should decrease without major increases in worldwide secular variation. Indeed, this is just what is happening at the mo- ment. But what is seen is not necessarily the onset of a reversal - on a statistical basis the best bet is that the dipole field intensity will again recover, as it has fre- quently in the past, without reversing. Gubbins's model also fits nicely into the Poisson framework of reversal epoch lengths. As he notes, the dipole field will often have experienced reduction in inten- sity through development of reverse flux, but only on rare occasions will the field actually reverse. Changes in reversal rates are easily accommodated by changes in mantle convection patterns on the appropriate timescale of 10 8 years. The most critical test of the model comes from its prediction that all rever- sals should be similar within the lifetime of mantle convection patterns. Thus, the transition fields observed at particular sites should be similar. It is already known that reversals observed at the same local- ity are sometimes similar, but it is not clear whether they remain similar for the period required by the model. Preliminary in- dications are that the changes are too rapid to be consistent with it. The constant location of the region of negative flux generation specified by Gub- bins requires that reversals are all initiated in the same low-latitude region of the Southern Hemisphere. This is equivalent to a common starting point for a reversal model of the type suggested" by Hoffman, based on the Busse dynamo". Gubbins and Bloxham note 10 that the core-mantle fields they calculate are similar to those predicted by Busse, although there are far more convection rolls in that model than predicted by the calculated fields. To the degree that Hoffman's model is successful in its predictions for the last reversal, Gubbins's model is also successful for that reversal, and it produces a mechanism to explain Hoffman's geometrical model. Gubbins's model is not so obviously in conflict with available palaeomagnetic data on field reversals that it should be discounted. Unfortunately this, as is so often the case in the Earth sciences, gives a sufficient, but not necessary, condition. Some might argue that the model must be rejected because it is not consistent with the frozen-flux assumption. Gubbins, on the other hand, holds that the data show the frozen - flux assumption to be incor- rect on the reversal timescale. Whether the model turns out to be suc- cessful or not, it has already stimulated renewed interest in the reversal process. It is rooted in observations made possible by new techniques of analysis of the present and recent geomagnetic field and chal- lenges palaeomagnetists to come up with the necessary data for a critical test. Such a test requires new emphasis on obtaining Cell biology records of different reversals at individual sites, for which the best sources may well be ocean cores and rapidly deposited sedi- mentary sections. D I. Bloxham. J. & Gubbins. D. Nature 317. 777-781 (1985). 2. Gubbins, D. Nature 326, 167-169 (1987). 3. Courtillot. V. £OS 67, 809-812 (1986). 4. McFadden, P.L. & Merrill, R.T. J. geophys. Res. 80, 3354- 3362 (1984). 5. Hoffman, K.A. Science 196, 1329-1333 (1977). 6. Williams, I. & Fuller. M. J. geophys. Res. 86, 11657-11665 (1981). 7. Valet, J.-P. & Laj, C. Nature 311, 532-555 ( 1984). 8. Hoffman, K.A. Earth planet. Sci. Lett. 44. 7-17 (1979). 9. Busse, F. Geophys. J. R. astr. Soc. 42, 437-459 (1975). 10. Gubbins, D. & Bloxham, J. Nature 325, 509-511 (1987). Mike Fuller is in the Department of Geological Sciences, University of California, Santa Bar- bara, California 93106, USA. The molecular basis for clathrin light-chain diversity Dudy Bar-Zvi COATED vesicles are involved in the trans- port of proteins between the organelles of eukaryotic cells and in receptor-mediated endocytosis, a specialized pathway by which cells take up specific molecules from the extracellular fluid. The coat of these vesicles has a characteristic poly- gonal structure of two protein complexes: assembly factor, composed of polypep- tides of relative molecular mass 100,000- 115 ,000 (100- llSK) and SOK, and clath- rin, composed of three 180K heavy chains and three 30-60K light chains, organized as a triskelion ( refs 1,2; see cover picture). In two papers on pages 154 and 203 of this issue',., Peter Parham and colleagues present data that help to elucidate clathrin light-chain structure and function. There are several unexplained observa- tions and unanswered questions about the function of clathrin light chains. Because the heavy chains alone can form tri- skelions that can self-assemble, it is not obvious that light chains are needed to generate a coated vesicle. They are required for the activity of the uncoating A TPase, a 70K heat-shock protein that catalyses the ATP-dependent removal of clathrin from coated vesicles in vitro'. Light chains can also stimulate the autophosphorylation of the SOK assembly factor polypeptide in vitro'. The significance of either of these activities in vivo is unknown. With the possible exception of yeast, clathrin from other eukaryotic cells contains two distinct light chains' (called a and /3, a and b, or sometimes 1 and 2). Moreover, clathrin light chains from brain are about 3K heavier than the correspond- ing light chains from other tissues'. What do these differences in mass signify? Because light chains from brain and from other tissues compete for binding to the same sites on the heavy chain near the centre of the triskelion 8 '; because either the a- or the /3-light chains of brain can satisfy the requirements of the uncoating ATPase 5 ; and because both bind calmodu- lin and calcium, the significance of the differences in mass is unclear. Indeed, all the light chains have common features, for example, they are all heat-resistant poly- peptides and unusually sensitive to protease'. Triskelions isolated from various tissues show variability in the molar ratio of the two light chains, ranging from 1 a-chain: 2 /3-chains in brain to almost exclusively a-chain in reticulocytes. This variability, together with the observed difference in the mass of brain-specific light chains and those from other tissues, has prompted the suggestion that different specific light chains function in determining the specific properties or function of different types of coated vesicles. But because the polypep- tide pattern and light-chain stoichiometry of endocytotic and exocytotic coated vesi- cles from liver seem to be identical 10 , there is no clear indication of why multiple but apparently functionally interchangeable light chains have evolved in most organ- isms. The only functional difference be- tween the a- and /3-light chains known so far is the differential phosphorylation of the /3-chain by a coated vesicle-associated casein kinase II (ref. 11). Knowledge of the molecular and struc- tural details of clathrin light chains should help to elucidate their role and distinguish the functions of the different light-chain types. Jackson et al. in this issue' report the cloning of clathrin a- and /3-light chains from bovine brain and from lym- phocytes. The a- and /3-light chains are encoded by two closely related genes, so