Lamins are intermediate filament proteins that form a network lining the inner nuclear membrane. They provide mechanical strength to the nuclear envelope, but also appear to have many other functions as reflected in the array of diseases caused by lamin mutations. Unlike other intermediate filament proteins, they do not self-assemble into 10 nm filaments in vitro and their in vivo organization is uncertain. We have recently re-examined the organization of a simple B-type lamina in Xenopus oocytes [Goldberg, Huttenlauch, Hutchison and Stick (2008) J. Cell Sci. 121, 215–225] and shown that it consists of tightly packed 8–10 nm filaments with regular cross-connections, tightly opposed to the membrane. When lamin A is expressed in oocytes, it forms organized bundles on top of the B lamina. This has led to a new model for lamina organization which is discussed in the present paper.
IF (intermediate filament) proteins form filaments in the cytoplasm, or, in the case of lamins, the nucleus. All IF proteins consist of a central α-helical rod domain flanked by head and tail domains. The rod contains four α-helices, 1a, 1b, 2a and 2b, separated by linkers, L1, L12 and L2 . IF proteins form coiled-coil dimers, but how these are assembled into ∼10 nm diameter filaments is uncertain. In vitro studies have shown that cytoplasmic IF dimers associate laterally into antiparallel staggered tetramers, which associate into ULFs (‘unit-length’ filaments). ULFs associate longitudinally into the ∼10 nm filament . Lamins appear to be different, and generally such filaments have not been assembled in vitro [3–5], although Caenorhabditis elegans lamin has . Likewise, in vivo, there is no evidence that lamins form simple 10 nm filaments, but rather more complex networks .
There are three major somatically expressed lamin genes, one encoding A-type lamins, lamin A and lamin C, which are splice variants from the same gene, and the others encoding lamins B1 and B2, as well as germ-cell-specific lamins [7–9]. B-type lamins are expressed in all cells and appear to be essential, whereas A-type lamins are expressed during differentiation, and, although not essential for viability, mutations can cause a variety of diseases called laminopathies .
A- and B-type lamins are different
Lamins are located exclusively in the nucleus and possess an NLS (nuclear localization sequence) within the C-terminal tail . B-type lamins are permanently associated with the inner nuclear membrane, probably due to its isoprenylation [12,13]. This lipid modification is mediated by the ‘CaaX box’ (where C is cysteine, a is an aliphatic amino acid, and X is any amino acid) at the C-terminus . In humans, lamin B1 and B2 sequences are CAIM and CYVM respectively, which are specifically farnesylated by farnesyltransferase  at the cysteine residue. Isoprenylation increases the affinity of proteins to membranes, and, without it, B-type lamins do not accumulate at the NE (nuclear envelope) [16,17]. However, isoprenylation of B-type lamins is not solely responsible for membrane location. For instance, during meiotic metaphase, isoprenylated lamins detach from the membranes . Generally, however, they remain membrane-bound throughout cell division , probably due to specific interactions with membrane proteins such as the LBR (lamin B receptor) .
Although lamin A is modified in a similar way, the C-terminus is subsequently cleaved, allowing lamin A to be located away from the membrane [20,21], sometimes deep within the nucleus . Lamin A is synthesized as pre-lamin A, which contains the CaaX box. The cysteine residue is isoprenylated by farnesyltransferase, followed by cleavage of the last three amino acids by Zmpste24 and then carboxymethylated by isoprenyl carboxymethyltransferase. Finally, the last 18 amino acids, which include the isoprene, are cleaved, releasing mature lamin A . The importance of this cleavage is demonstrated by a splice site mutation that results in deletion of the Zmpste24 cleavage site, resulting in the accumulation of a permanently isoprenylated lamin A, causing progeria . Therefore release of mature lamin A from its lipid modification is necessary. Lamin C lacks the CaaX box and is never isoprenylated, but can be still be found at the NE, showing that isoprenylation is not essential for NE localization, which may be mediated by other proteins.
B-type lamins are thought to be the ancestral lamin [24,25]. Invertebrates have only B-type lamins. Vertebrate lamins contain an additional acidic cluster of amino acids at the C-terminus which is not present in invertebrates . During evolution, A-type lamins are thought to have arisen from a gene duplication of lamin B and acquired an extra exon, resulting in an additional ∼90 residues just preceding the CaaX box. Therefore, in contrast with lamin B1/B2, mature lamin A is larger and only temporarily farnesylated. The rod domains of A- and B-type lamins are identical in length, but their sequences are variable. Such variation could be important to ensure homodimerization [24–26]. Although A- and B-type lamins appear to interact in vitro [27–29], during mitosis, A-type lamins are solubilized, whereas B-type lamins remain membrane-associated, suggesting that heterodimers are not formed in vivo. Another similarity between A- and B-type lamins is the Ig-fold, a 105-amino-acid globular domain between the rod and the C-terminus, whose atomic structure has been solved . The Ig-fold appears to be involved in interactions with DNA and other proteins , such as the structural protein titin . The N-terminal head domain on the other hand is essential for filament assembly. In vimentin, it was shown to be involved in lateral association of dimers into tetramers and further into ULFs , probably by its interaction with the rod domain of an adjacent dimer. The atomic structure of lamin coil 2b looks similar to that of vimentin, but has a different charge distribution  which could result in different interactions and contribute to their different assembly properties.
Plants and fungi may also have a ‘lamina’
Although it can be seen how the different lamins have evolved from invertebrates to mammals , it remains unclear whether similar proteins and structures exist in fungi and plants, and it is often assumed that such organisms do not need a lamina because the cell wall affords structural protection. Moreover, no homologues of lamin genes have been found in plant/yeast sequence databases, while the supportive biochemistry and electron microscopy [34,35] on their own are not sufficient to gain the final evidence for the existence of yeast or plant laminae. Considering the multiple functions the lamina fulfils in animals, the complete lack of a similar structure in complex eukaryotic cells such as plants seems unlikely. Recently, the effort to identify protein candidates for plant lamins led to promising results when NMCP1 (nuclear matrix constituent protein 1), alias LINC (LITTLE NUCLEI), was identified [36,37]. The protein contains coiled-coil domains essential for forming dimers, localizes on the nuclear rim and, intriguingly, when lacking, nuclear organization is affected. However, ultrastructural data would be necessary to reveal the structure of the potential plant/yeast ‘lamina’ and to link candidate proteins to this nuclear compartment.
FeSEM (field emission scanning electron microscopy) has proved to be a useful tool for studying animal NE structure. We have therefore used feSEM to investigate the structure of the NE in plant and yeast cells. Our results show, for the first time, the filamentous network closely attached to the inner nuclear membrane of plant/yeast nuclei that is attached to NPCs (nuclear pore complexes) and resembles metazoan laminae. Our feSEM data provide evidence that supports the existence of the lamina-like structure in both plants and fungi. Which proteins participate in the composition or reorganization of this structure and what the exact roles of this compartment in plant/yeast biogenesis are are emerging questions to be solved.
Organization of endogenous and exogenously expressed lamina in Xenopus oocytes
Although it is generally considered that vertebrate lamin assembly cannot be recapitulated in vitro, lamins do form higher-order structures, such as sheet-like paracrystals that could be relevant to in vivo organization. However understanding of in vivo organization is poor, so it is difficult to judge the relevance of in vitro structures. Simple B-type lamina found in amphibian oocytes have been studied [3,38–40], and an accurate picture of this organization is emerging. More complex laminae in somatic cells are difficult to access because of more complex interactions with intranuclear structures and because they appear less ordered.
We have taken an intermediate approach to studying lamin structure using Xenopus oocytes, and we have added somatic B- and A-type lamins by ectopic expression to construct more complex laminae. First, we re-examined the organization of the endogenous lamina which predominantly consists of a single embryonic-specific B-type lamin, LIII . This material has provided a unique view of lamina organization which led to a model that has persisted for 20 years, where the lamina is described as two sets of parallel 10 nm filaments at right angles to each other which form an orthogonal network attached to the inner nuclear membrane (see Figure 2), with a repeat distance of approx. 50 nm . This was based on transmission electron microscope images of detergent-extracted metal-shadowed NEs. However, such a high degree of order after detergent extraction is difficult to achieve [38,39,42,43].
Using feSEM, it was possible to visualize the lamina without detergent extraction (Figure 1). This revealed the LIII lamina to be a single set of tightly packed 8–10 nm filaments running parallel to each other, rather than two sets at right angles. One striking difference was that the repeat distance was only ∼15 nm, not 50 nm (Figure 2). Because lamin LIII contains the hydrophobic isoprene and methylated C-terminus, it is possible that upon removal of the membrane, there is a lateral aggregation of the filaments, resulting in opening up of the structure.
FeSEM image of the inner surface of Xenopus oocyte NEs at medium (A) and high (B) magnification
Comparison of a new model based on our observations with the old model
There is an orthogonality to the structure, which is due to cross-connections between the parallel filaments. Because the cross-connections are often precisely aligned and have a repeat distance also of ∼15 nm, they give the appearance of a second set of filaments. The cross-connections are much thinner at ∼5 nm and do not appear to be continuous. This has led to the model shown in Figure 2, which proposes a much tighter structure than thought previously. It simply consists of 8–10 nm filaments, just like other IFs, anchored firmly to the membrane by the hydrophobic C-terminal modifications, with regularly spaced cross-connections. The identity of the cross-connections is not known. They could be protruding tail domains, as seen in electron micrographs of isolated dimers, or they could be extra cross-linking proteins. As the Ig-fold is a major protein–protein interaction domain , it would be a prime candidate for mediating the cross-connections.
Xenopus oocytes are also a powerful ectopic expression system. When lamin LIII is exogenously overexpressed, the protein is not incorporated into the NE, but instead induces the formation of extra membrane structures attached to the NE . A lamina forms on the surface of these membranes, which is essentially identical with the endogenous lamina . This is then an ideal system to test which domains of LIII are involved in cross-connections, as mutant proteins can be expressed and analysed.
Interestingly, as well as parallel filaments on the extra structures, short stubby filaments similar to ULFs are observed (Figure 1B), suggesting that lamin assembly could proceed through these intermediates like other IFs. Somatic lamins B1 and B2 were also expressed in this system, forming similar filaments to LIII, but, interestingly, without such obvious cross-connections and with a less-ordered organization. Whether this is an intrinsic property or to do with different interactions with other proteins is unknown.
Lamin A, on the other hand, is different in several important ways, when expressed in oocytes. Extra membranes are not formed, and, instead, lamin A filaments attach to and pile up on top of the endogenous lamina, forming three-dimensional bundles of filaments rather than two-dimensional sheets. The filaments themselves are thicker than B-type filaments at approx. 15 nm and cross-connections are not observed (Figure 3). The organization of the filaments within the bundles does not appear to be random, and they often seem to be running parallel to each other, suggesting a higher-order lateral association between filaments, possibly related to paracrystals observed during in vitro assembly experiments. As with B-type lamins, structures consistent with ULFs are observed . The organization is most likely to be mediated by other non-membrane-bound proteins. One candidate for this could be the lamin A/C-binding protein Lap2α (lamina-associated polypeptide 2α), with a similar function being performed in the B-lamina by the related transmembrane lamin B-binding protein, Lap2β (see ), but this remains to be tested. B-type lamins are probably restricted to the membrane-surface-associated two-dimensional arrays owing to their lipid modification, whereas lamin A can form these three-dimensional arrays because they have cleaved off the modification during processing and they interact with non-membrane proteins such as Lap2α . Figure 3(B) is our model for the organization of A- and B-type lamins at the NE. Insertion of NPCs into this model awaits further analysis.
Lamin A structure
Nuclear Envelope Diseases and Chromatin Organization: Independent meeting held at New Hunt's House, King's College, Guy's Campus, London, U.K., 23–24 April 2008. Organized and Edited by Juliet Ellis (King's College London, U.K.).
This work was supported by grants from the Deutsche Forschungsgemeinschaft to R.S. (Sti 98/7-1), the Wellcome Trust (065860) and BBSRC (Biotechnology and Biological Sciences Research Council) (BBE0157351) to M.W.G.