Dynamics and Structure-Function Relationships of the Lamin B Receptor (LBR)

The lamin B receptor (LBR) is a multi-spanning membrane protein of the inner nuclear membrane that is often employed as a “reporter” of nuclear envelope dynamics. We show here that the diffusional mobility of full-length LBR exhibits significant regional variation along the nuclear envelope, consistent with the existence of discrete LBR microdomains and the occurrence of multiple, asymmetrically-spaced anastomoses along the nuclear envelope-endoplasmic reticulum interface. Interestingly, a commonly used fusion protein that contains the amino-terminal region and the first transmembrane domain of LBR exhibits reduced mobility at the nuclear envelope, but behaves similarly to full-length LBR in the endoplasmic reticulum. On the other hand, carboxy-terminally truncated mutants that retain the first four transmembrane domains and a part or the whole of the amino-terminal region of LBR are generally hyper-mobile. These results suggest that LBR dynamics is structure and compartment specific. They also indicate that native LBR is probably “configured” by long-range interactions that involve the loops between adjacent transmembrane domains and parts of the amino-terminal region.


Introduction
The lamin B receptor (LBR) is an integral protein of the inner nuclear membrane originally identified by in vitro binding to lamin B [1][2][3]. The LBR molecule contains eight (predicted) transmembrane domains. These domains exhibit high sequence homology to fungal and animal sterol reductases [4] and are flanked by two hydrophilic end-pieces at the amino-and carboxy-terminal ends [5][6][7]. The carboxy-terminal "tail" is rather short (32 residues) and does not contain recognizable structural features or motifs. However, the amino-terminal "head" region is quite large (> 200 residues) and consists of three distinct domains: the Tudor domain (Td); the serine-arginine (SR)-rich "hinge"; and the globular domain (GD) [8,9].
As all membrane proteins, LBR is synthesized on the surface of the endoplasmic reticulum (ER). After biosynthesis, the protein diffuses laterally across the ER, reaches the outer nuclear mebrane, passes through the nuclear pore complexes (NPCs) and arrives at the inner nuclear membrane, where is "trapped" by binding to nuclear lamina and peripheral heterochromatin a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 [10][11][12][13]. The amino-terminal region of LBR, which contains specific binding sites for a variety of nuclear proteins and self-associates in vitro [3,6,8,9,[14][15][16][17][18][19], is generally believed to contribute in nuclear retention. However, several studies have shown that also critical for correct targeting of LBR is the presence of at least one transmembrane domain -and preferably transmembrane I [5,[20][21][22][23].
Based on these data, a GFP-fusion protein representing the amino-terminal region and the transmembrane domain I of LBR has been widely used as a "reporter" of nuclear envelope (NE) dynamics [12,13,17,[24][25][26][27]. Such a "minimal construct" is often preferred, because elimination of (all but one of) the transmembrane domains seems to prevent the formation of inclusion bodies, which occurs with variable frequency when cultured cells are transfected with fulllength LBR [24]. However, there is a flip-side to this "advantage": it is now known that several mutations that map downstream to transmembrane domain I affect dramatically the function and cellular partitioning of LBR [28][29][30][31][32][33][34]; therefore, it is unlikely that carboxy-terminally truncated constructs, such as the one mentioned above, could faithfully report the dynamics of the native protein.
To investigate the structural determinants that are involved in LBR dynamics, we assessed the mobility of full-length and partially truncated proteins employing an array of quantitative and morphological approaches. The data reveal that LBR is part of multiple dynamic ensembles, which possess distinct diffusional properties and are distributed non-uniformly along the inner nuclear membrane. Despite this regional variability, comparison of ensemble-averaged data allows us to conclude that carboxy-terminal LBR mutants have distinct properties in relation to full-length LBR. Furthermore, the results suggest that oligomerization of LBR in situ and formation of "local structures" via intra-molecular interactions are probably critical for the correct partitioning of the protein along the inner nuclear membrane.  ) supplemented with 2 mM L-glutamine, 2 mM penicillin/streptomycin and 10% fetal calf serum (FCS) (all from PAA) in a humidified incubator at 5% CO 2 and 37˚C. The MDCK cell line (a generous gift from Prof. Savvas Christoforidis, ATCC No CRL-2936) was cultured in DMEM supplemented with 2 mM L-glutamine, 2 mM penicillin/streptomycin and 10% FCS at 10% CO 2 and 37˚C. BHK cell line was cultured in DMEM supplemented with 2mM Lglutamine, 2mM penicillin/ streptomycin, 5% FCS and 10% tryptose phosphate (Gibco, Life Technologies).

Constructs and transfection
The coding sequence of full-length mouse LBR was amplified using specific primers and mouse liver RNA as template. PCR fragments were then digested with EcoRI/BamHI and ligated in pEGFP-N2 (Clontech, Mountain View, CA). The rest of the constructs were generated by standard cloning procedures using as a template the FL-LBR and appropriate primers (see S1 Table). The Sec61β-eGFP construct was a generous gift of Dr. Carol Murphy (Univeristy of Birmingham, UK). All constructs were verified by sequence analysis. Transfection was carried out by PEI (Sigma Aldrich) according to manufacturer's instructions, or by electroporation using the ECM630 apparatus (BTX) operated at 260 V, 85 μF, 725 O.
Mouse liver RNA was isolated with Trizol (1ml/50-100mg of tissue). The tissue was first frozen with liquid nitrogen and ground. Trizol was added, the specimen was incubated for 20 min at room temperature and centrifuged at 4˚C. Then, the standard procedure of chloroform/isopropyl alcohol extraction was followed and the RNA pellet was washed with 75% EtOH. Finally, RNA was dissolved in DECP-H 2 O. The mouse used for the liver RNA isolation was not specifically breed for this study. Ethics Committee approval was not required per our Institution's research regulations.

Imaging
Cells grown on coverslips were washed, fixed in 4% formaldehyde in phosphate buffered saline, permeabilized with 0.2% Triton X-100 and blocked with 0.5% fish skin gelatin. The specimens were visualized in a Leica TCS SP5II confocal microscope. For multi-color analysis sequential image acquisition was applied and emission detection ranges were adjusted to minimize crosstalk between the different signals. The following antibodies were used: anti-Nup84/ 107, anti-BiP (obtained from Developmental Studies Hybridoma Bank); anti-lamin-B [2] (prepared and purified in the EMBL animal facility from rabbits not particularly breed for this study). Ethics Committee approval was not required per our Institution's research regulations. For STED microscopy, Hela cells transfected with LBR-YFP were fixed in 4% formaldehyde in phosphate-buffered saline, permeabilized with 0.2% Triton X-100 and blocked with 0.5% fish skin gelatin. Specimens were examined under a 100x objective in a Leica SP5II STED microscope.
Live imaging and FRAP/FLIP assays FRAP experiments were performed in a Leica SP5 confocal microscope using a 63X /1.4 NA oil immersion objective and a 135 mW Argon/Neon laser. Cells were imaged in glass bottom dishes (MatTeK Corporation) kept at 37˚C using an air-stream stage incubator. For membrane-spanning constructs, recovery data were acquired with the argon laser operating at 10% laser power and 15% transmission. The confocal pinhole was set to 3.99 Airy units. The zoom factor ranged from 7 to 10. Images were 512 X 512 pixels, line average was set at 2 and scan speed at 400 Hz. Prebleach images were collected at 6.5 s time intervals for 32 s. Photobleaching in the NE was performed using either a circle of 1.0 or 1.5 μm in diameter, or a strip 1.7 μm wide and 10-70 μm in contour length. Photobleaching in the ER was performed using a 3 μm-circle ROI. The bleach duration was 2.6 s, with laser power at maximum. Postbleach images were collected for~307 s (6 images at 3.5 s time intervals, followed by 8 images at 35 s intervals). For "soluble" constructs, recovery data were acquired with the argon laser operating at 10% laser power and 15% transmission. The confocal pinhole was set to 6.6 Airy units. The zoom factor ranged from 10 to 13. Images were 128 X 128 pixels. Scanning was bidirectional with scan speed at 800 Hz. Ten prebleach images were acquired at 0.38 s time intervals, while postbleach images were collected at 0.18 s time intervals for~12.6 s. The photobleach was performed in a circle ROI 3.0 μm in diameter (unless otherwise specified) with bleach duration 0.36 s and laser power at maximum. For FRAP analysis, data were corrected for fluorescence quench and recovery observed in the entire cell and in the background, using the following formulas: I=I 0 ¼ ðI t À I first postbleach Þ=ðI prebleach average À I first postbleach Þ and I t ¼ ðI t frap À I t backround Þ=ðI t unfrap À I t backround Þ; where the t subscript denotes the specific time point for which the intensity (I) is calculated and unfrap represents an area of the same size with the ROI positioned outside of the bleached region. We confirmed that multiple consecutive FRAP experiments in the same cell and ROI (pulse-FRAP) give similar results and that cell viability is not affected. For FLIP assays, the samples were bleached repetitively and fluorescence loss in the non-bleached region was monitored. FLIP experiments were performed with the same settings as that for FRAP.

Statistical and quantitative analysis
The coefficient of variation (CV) was calculated as the ratio of the standard deviation to the mean in each dataset. The Kolmogorov-Smirnov test was applied using the BioStat 2009 software (AnalystSoft).

Distribution and dynamic properties of full-length LBR
To analyze the dynamic properties of LBR, we engineered 16 eGFP-fusion proteins (all carboxy-terminally tagged) that correspond to full-length LBR (FL-LBR) or LBR mutants (S1A Fig). Experiments with several cell lines showed that over-expression of FL-LBR does not generally perturb cellular architecture (S1B Fig). However, in one particular line (osteosarcoma U2OS cells) accumulation of FL-LBR caused detachment of the outer from the inner nuclear membrane in~30% of the cells, as previously reported by Zwerger and co-workers [34]. For this reason, we continued our experimental work using Hela cells as a model system.
Upon transient transfection of Hela cells, FL-LBR fluorescence was detected in both the ER and the NE (Fig 1A). Based on intensity per unit area measurements, the fusion protein was 3-fold more concentrated at the nuclear periphery than in the surrounding cytoplasm, consistent with retention in the inner nuclear membrane. FL-LBR co-localized to a large extent with nuclear lamin B, but its distribution pattern of was distinct from that of BiP (a luminal protein of the ER) and Nup84/107 (a marker of the NPC) ( Fig 1B). In addition, the eGFP fusion protein formed closely spaced microdomains, similar to those identified in previous studies using specific antibodies [16,35]. These raft-like structures, which had contour lengths from 300-600 nm, were readily detectable by both conventional and super-resolution (STED) light microscopy (Fig 1A, detail).
To measure the diffusional mobility of FL-LBR, we took a FRAP/FLIP approach. In a typical FRAP experiment, fluorescence recovery at the NE reached a plateau 250-300 s after the end of the bleach pulse and remained steady for at least 500 s ( Fig 1C). After normalization (see Materials and Methods), the ratio of the post-bleach to pre-bleach fluorescence intensity was consistently <1.0, indicating that a proportion of LBR is immobilized, presumably by binding to nuclear sub-structures. FLIP assays showed that non-bound FL-LBR molecules exchange throughout the ER-NE continuum: when a part of the NE was repeatedly bleached, a significant loss of fluorescence was detected in both the peripheral ER and the non-bleached sector of the NE (Fig 1D, image series). The loss of fluorescence at the NE was always less than the loss of fluorescence in the ER (Fig 1D, histograms), consistent with a higher mobility in the latter compartment.
To study LBR dynamics in more detail, we experimented with spot-like (circular) and striplike (curvilinear) regions-of-interest (ROIs). The circular ROIs were 1.0 or 1.5 μm in diameter, while the curvilinear ROIs (referred to as "arc", "half-rim" and "whole-rim") were 1.7 μm in width and had contour lengths from 10-70 μm (Fig 2A). Photobleaching of fixed cells showed that the bleach profile across the ROIs was roughly Gaussian, with a lateral halo <1 μm ( Fig  2B). However, with the optical setup employed (see Experimental Procedures), the bleached region had a substantial axial half-depth, extending about 2.6 μm proximally and 1.2 μm distally to the incoming beam ( Fig 2C). Therefore, the effective ROI extended beyond the confines of the NE (which has a thickness of~70 nm) and evidently included a part of the perinuclear ER. Irrespective of ROI type used in the FRAP assays, the recovery curves of FL-LBR differed significantly from one experiment to the next, yielding a range of mobile fraction (Mf) and recovery halftime (t 1/2 ) values ( Fig 3A). To distinguish between genuine differences in protein mobility and fluctuations due to physical perturbation, we performed a series of control experiments. Potential flaws arising from photodamage or shifting of the cells during data acquisition were assessed recording fluorescence recovery after consecutive bleaching of the same cell and ROI (pulse-FRAP assays). As could be seen in S2A Fig, the coefficients of variation (CVs) of Mf and t 1/2 in pulse-FRAP assays were significantly lower than that recorded by sampling different cells and ROIs. Furthermore, a comparison of the corresponding cumulative distribution functions (CDFs) by the Kolmogorov-Smirnov (K-S) test showed that the dynamic parameters of FL-LBR did not vary in a systematic fashion in relation to the pulse sequence. Finally, in cells whose FL-LBR content differed more than 3-fold, the corresponding CVs were almost the same (S2B Fig). Taken together, these data suggest that LBR dynamics exhibits genuine variability.
When we used circular ROIs, the ensemble-average Mf of FL-LBR, as well as the CV Mf , inversely correlated with ROI size (Fig 3B). Examining the frequency distributions of Mf, we noticed that assays with 1-μm ROIs yielded a major peak at low Mf values and a less prominent peak at higher Mf values (Fig 3C, upper graph). However, assays with 1.5-μm ROIs yielded a unimodal Mf distribution, with a broad peak at low Mf values (Fig 3C, middle graph). This explains the lower Mf and CV Mf obtained in the latter case and strongly suggests that LBR displays length scale-dependent heterogeneity. Interestingly, the distribution of t 1/2 was rather insensitive to changes of ROI size (Fig 3D; upper and middle graphs), which implies that LBR binding to underlying sub-structures is dominant over diffusion [for more details on "reaction-dominant" and "diffusion-dominant" processes see ref. [36]].
Experiments with arc-shaped ROIs gave a bimodal Mf distribution (as observed previously with small circular ROIs). However, in this case the high Mf peak was more pronounced than the low Mf peak (Fig 3C, lower graph), resulting in an increase of the average Mf (Fig 3B). Expanding the contour length of the arc-shaped ROI resulted in a gradual decrease of Mf (S3 Fig). However, this should be interpreted with caution, because the visual inspection of these samples revealed extensive bleaching of the ER throughout the cell, probably due to the higher angle of the incoming beam (check Fig 2A, whole-rim configuration), which apparently results in the depletion of the unbleached LBR pool. Irrespective of the technical limitations encountered with very large ROIs, the distinct shifts in Mf distribution when the change of ROI size (or shape) was within limits suggested a regional variability.
To examine more directly whether LBR dynamics exhibits such a regional variability, we performed experiments at the level of single cells. On a first approach, we asked whether FL-LBR recovers in a uniform fashion across the same ROI. In most of the samples, the mobility of FL-LBR at the edges and the central region of an arc-shaped ROI did not differ significantly ( Fig 4A). However, in a few cases, the mobility of FL-LBR across the same ROI did differ, reaching the levels of variation that we had previously observed by assaying different cells and ROIs. To confirm that this variation reflected molecular events, we inspected frameby-frame the recovery series in the corresponding samples and examined visually the re-distribution of fluorescence along the bleached zone. From this survey it became clear that FL-LBR recovery across a given ROI was asynchronous and non-uniform (Fig 4B; image series).
More detailed profile FRAP experiments showed that the non-uniformity of the recovery process was not due to uneven photobleaching across the ROI (Fig 4B; upper graph). Furthermore, when we measured the local rates of fluorescence recovery (i.e., the ratio of fluorescence intensity at a given time point divided by the fluorescence intensity immediately after bleaching), we noticed that the areas flanking the bleached region were gradually losing fluorescence as a function of time; conversely the central area of the ROI was gaining fluorescence (Fig 4B,  lower graph). The rate of fluorescence recovery in the central zone of the ROI was not steady, displaying distinct "peaks" and "valleys". Furthermore, the fluorescence intensity in different sub-regions did not always increase as a function of time, but sometimes regressed to a lower level. As a rule, this was paralleled by fluorescence gain in neighboring areas, revealing that segments of the NE that had transiently accumulated fluorescent molecules could act as a "source" for the flanking regions.
When we probed non-contiguous sectors of the same NE, regional differences in the diffusional mobility of FL-LBR became even more apparent (Fig 4C). Based on these data, we would argue that variation of the dynamic parameters arises primarily from the non-uniform distribution of LBR along the inner nuclear membrane (i.e. the existence of discrete microdomains that contain predominantly immobile protein and stretches that contain mostly "free" LBR). Since ROI selection in the FRAP experiments is entirely random, it is likely that the area probed each time contains a variable number of LBR microdomains. Obviously, the more ROI size increases, the more probable it becomes that the bleached region will span both microdomain-rich and microdomain-poor regions of the NE. This explains the "homogenization effect" (lower variability) and the trend towards a lower Mf observed in the FRAP assays, as is graphically depicted in the "scanning model" of Fig 5. We have shown above (Fig 2C), that cisternae of perinuclear ER are inescapably included in the bleach volume when the laser beam is focused on the surface of the nucleus, due to light diffraction. Knowing that ER resident proteins often exhibit variable dynamic behavior due to the spatial complexity of the endomembranes [37], we found it reasonable to compare the variation of LBR dynamics in the NE and the ER. As will be shown in the next section, the mobility of FL-LBR in the ER varied noticeably less than in the NE. However, in both compartments the dynamic parameters of FL-LBR fluctuated more than one would expect for a monodispersed protein species diffusing across the lipid bilayer. Therefore, at least a part of the variation observed in the FRAP assays should be attributed to the fact that LBR exists in many molecular forms and oligomeric states. This point should be taken seriously into account when pursuing a more definitive interpretation than that proposed in

Dynamics of membrane-spanning and "soluble" LBR mutants
Along with FL-LBR, we also examined several forms of the protein that lack parts of the carboxy-terminal region (see S1A Fig). Among these were: Gr (or ΔTMVIII,CT), which lacks transmembrane domain VIII and resembles a pathogenic mutant associated with Greenberg's dysplasia; Icj (or ΔTMV-VIII,CT), which lacks transmembrane domains V-VIII and resembles a mutant that causes mouse ichthyosis; and Mod (or ΔTMII-VIII,CT), which contains the entire amino-terminal and transmembrane domain I, resembling the "minimal construct" used in previous studies (see Introduction). A provisional model explaining regional variation of LBR mobility. The "scanning probe" cartoon explains how LBR mobility could vary from one region of the NE to the next, depending on the relative abundance of immobile (dark green) or mobile (light green) LBR molecules. The basic assumptions in this model are that: a) ROI selection is entirely random and therefore, the probability of including or not including LBR microdomains is proportional to the abundance of the latter; b) LBR microdomains contain many more molecules than "free" LBR; c) the areas of the INM neighbouring to the NPCs, where the lamina meshwork is interrupted, contain exclusively free LBR. The same applies for the outer nuclear membrane and the surface of the ER. The histogram shown is based on an entirely hypothetical example, where the ROIs include from 0 to 2 microdomains each, depending on size. Arbitrarily, we have assumed that regions containing free LBR regain 80% of the initial fluorescence, whereas areas containing 0.5 to 2 microdomains regain from 20% to 70% of the fluorescence, respectively. Notice that increase of ROI size results in the broadening of the Mf frequency distribution ("homogenization") and The overall distribution of the three truncated proteins was similar to that of FL-LBR ( Fig  6A). However, in about 20% of transfected cells, the Gr mutant over-accumulated in the cytoplasm and formed large clusters. Some of these clusters stained for lamin B, suggesting trapping of nuclear lamins by the ectopically localized protein (Fig 6B).
As shown in the previous experiments, FRAP assays with FL-LBR yielded variable readouts. The same was observed when we tested the three membrane-spanning LBR mutants (Fig 6C and 6D box plots). To find out whether the fluctuations of the dynamic parameters were protein or compartment-specific, we measured the mobility of FL-LBR and LBR mutants, together with Sec61β in the NE and the bulk ER. Sec61β is an integral membrane protein of the ER that also partitions (without being retained) with the NE [38]. As demonstrated in Fig 6C and 6D (tables), the Mf values of Sec61β exhibited variation close to the levels of experimental error in both compartments. On the other hand, whereas significant variation was detected at the NE with all LBR proteins (especially Mod), this did not happen in the ER. Variability in t 1/2 was in all cases above the threshold of experimental error and most noticeable with the Mod and Icj mutants. From these results we infer that mobility Despite the variability, a comparison of the FRAP data in cells expressing intact or truncated forms of LBR could still be made comparing the corresponding Mf values by the K-S test (Fig 6E). Using this approach, we found that Icj and Mod were significantly more mobile in the ER than in the NE, as was FL-LBR. Gr behaved in roughly the same way, but the mobility differences in the two compartments were less clear-cut, consistent with the fact that this mutant over-accumulates in cytoplasmic membranes in a sub-population of the cells. By K-S criteria, the differences in exchange rate (t 1/2 ) between NE-associated and ER-localized proteins were not significant. Taken together, these data suggest that the three LBR mutants are retained in the inner nuclear membrane by binding to underlying sub-structures, as does the native protein.
That said, it is important to note that Gr exhibited a higher mobility than FL-LBR at the NE, whereas Mod displayed lower mobility than the intact protein (Fig 6C). Icj also showed a higher mobility than FL-LBR, but this difference was not as significant by statististical criteria. The lower mobility of the Mod mutant and the increased mobility of Gr and Icj at the NE did not appear to be the consequences of structural aberrations that affect the diffusional properties of the protein, because none of the three mutants showed significant differences from FL-LBR in the bulk ER. Furthermore, the divergent properties of Mod and Gr rule out other effects resulting from the flipping of the hydrophilic carboxy-terminal tail of LBR from the nucleoplasm to the ER lumen. (According to the predicted topology of the LBR protein, in both of these mutants the tailpiece should be facing the lumenal side of the inner nuclear membrane, due to the fact that the truncated polypeptides contain an odd number of transmembrane domains; check S1A Fig) To find out whether the lower mobility of Mod at the NE was due to the fact that the eGFP moiety is fused near transmembrane domain I (which might possess some unique structural properties), we did additional experiments. Another LBR mutant, bearing exactly the same amino-terminal part with Mod, but possessing transmembrane domain VI instead of transmembrane domain I at its carboxy-terminal region, was constructed and assyed by FRAP. As shown in S4B Fig, this mutant, termed NtTM4 (or ΔTMI-III,V-VIII,CT), exhibited similar properties to Mod at the NE, suggesting that the differences between FL-LBR and Mod are not due to the peculiar features of the first transmembrane domain. However, in the bulk ER, the NtTM4 mutant showed a faster exchange rate by comparison to FL-LBR. Taken together, these data suggest that carboxy-terminal truncations affect the dynamic properties of LBR in different ways, depending on the membrane compartment.
To get more insight, we also studied the properties of TL (or ΔCt), which lacks the entire hydrophilic tailpiece, and three more LBR mutants that contain transmembrane domain I (as the Mod protein), but miss various parts of the amino-terminal region (S1A Fig). No differences between any of these mutants and FL-LBR could be distinguished at the level of the confocal microscope (S4A Fig). However, when we measured their diffusional mobility by FRAP and analyzed the data as outlined above (S4B and S4C Fig and S4 Supplementary Text), we arrived at some interesting conclusions. First, whereas TL behaved similarly to FL-LBR, the mobility of some amino-terminally truncated mutants in the NE and the bulk ER did not differ. Thus, unlike the carboxy-terminal truncations described above, amino-terminal deletions could affect LBR retention in the inner nuclear membrane, presumably by interfering with binding to the nuclear lamina and the peripheral heterochromatin network. However, it should be taken into account that amino-terminally truncated mutants often behaved differently from FL-LBR in the ER, suggesting a more "global" effect on protein structure.
One point that should not be missed here is that the deletion of large segments from the LBR molecule did not render the corresponding mutants invariably hyper-mobile and faster exchanging. This is evident when one compares the properties of Mod to that of FL-LBR. On the other hand, it should be pointed out that the differences identified between FL-LBR and LBR mutants by applying statistical methods do not necessarily describe the properties of these proteins at the level of single molecules and single cells. As we have shown, individual assays sometimes yield large mobility differences, irrespective of which LBR form is expressed, due to the regional fluctuations of LBR dynamics.
Proceeding further, we examined seven "soluble" mutants that do not contain any transmembrane domain. As shown in Fig 7A and S5A Fig, the soluble mutants did not accumulate at the nuclear periphery, confirming that the presence of a transmembrane domain is necessary for retention in this territory [for relevant observations see [5,[20][21][22][23]]. Nevertheless, from these experiments it became clear that all mutants containing the RS domain accumulate in the nucleoplasm, while those lacking RS exhibit a more or less pan-cellular distribution, resembling that of free eGFP. This means that the RS motif is essential for translocation and retention of LBR into the cell nucleus.
Special care was taken to compare the properties of the soluble mutant Nt (or TDRSGD), which represents the entire amino-terminal region, to the properties of FL-LBR. When assessed at different ROI sizes, this mutant yielded much lower variation in Mf and t 1/2 ( Fig  7A), confirming that differences in dynamic parameters are specific to membrane-associated LBR.
Finally, FLIP assays with Nt revealed another aspect of LBR dynamics, which adds complexity to the diffusion-retention mechanism utilized to target the protein to the inner nuclear membrane. As shown in Fig 7B, the nucleoplasmically disposed Nt exchanged to a greater extent with nucleoplasmic than with cytoplasmic molecules. This indicates that NPCs impose a significant barrier that must be overcome before LBR molecules enter the cell nucleus [for further comments see [39]].

Discussion
The introduction of "F-techniques" has revolutionized Cell Biology, allowing study of cellular events in real time. However, the widespread application of these methods has largely neglected the fact that, unlike model planar or spherical surfaces, real biological interfaces are rather anisotropic or have a complex, asymmetric architecture. The effect of surface geometry on the measurement of dynamic parameters cannot be overemphasized [40][41][42]. As it turns out, diffusion of integral membrane proteins is significantly different from diffusion of luminal or cytosolic components, with t 1/2 varying by a factor of~4, even when the membrane shapes are the same [37]. Furthermore, membrane geometry-induced variation in t 1/2 can be as much as 1.8-fold [43].
The NE not only is a highly curved membrane, but its curvature changes from one region to the next. In addition, the number of ER "connections" along the outer nuclear membrane is highly variable, affecting the influx of membrane proteins from the ER to the NE at a local scale. Finally, the NE membranes are tightly associated with macromolecular assemblies (e.g., the nuclear lamina meshwork and peripheral heterochromatin), the spatial distribution of which is far from homogeneous. Considering these non-regularities, it is surprising that most of the studies conducted so far with NE proteins have treated this assembly as an "ideal" 2-D membrane surface and have not stumbled on quantitative differences that cannot be explained by mere experimental error. To assess LBR dynamics, in this study we have taken an unbiased FRAP/FLIP approach. The choice of this approach was dictated by several factors. First, FRAP was precisely the method utilized in all previous studies on LBR dynamics, allowing us to compare our results with those reported by other authors. Second, FRAP permitted quantitative assessment of molecular mobility at good spatial resolution, which was a critical factor when assessing the properties of "fine" interfaces, such as the NE. Finally, FRAP assays were not particularly time consuming and could be used as a screening means for a large number of samples, as required.
FRAP does have several limitations and is not immune to artifacts [44,45]. Knowing that, we have made no presumptions on bleaching format and tried to be as consistent in sampling as we technically could. To supplement the data obtained by FRAP, we also considered other techniques, such as FCS. However, this technique proved rather problematic, because the very low fluorescence intensities required when applying it did not allow sufficiently good imaging and proper selection of ROI. Given the non-homogeneous distribution of LBR, future experiments based on single-particle tracking may provide more clues as to whether or not LBR microdomains represent "obstacles" that create a regime of anomalous diffusion at the NE and thus affect the dynamics of both LBR and other integral proteins.
Under our experimental conditions, significant variation was observed when LBR dynamics was assessed by conventional FRAP assays. In this context, the Mf of LBR in single nuclei varied up to 9-fold, while the t 1/2 differed up to 40-fold, depending on which region and subregion of the NE was probed. The mere magnitude of these differences strongly suggests that, beyond the known asymmetries, LBR itself probably exists in various binding states, transition intermediates and aggregation forms. Deriving specific thermodynamic and kinetic parameters from such a complex mode of dynamics is not realistic at the present time, because crucial details remain to be investigated. For instance, LBR has been reported to bind multiple proteins [1,6,14,46], is post-translationally modified at several sites [47] and follows a rather convoluted route from the peripheral ER to the inner nuclear membrane, after crossing the straights of the NPCs [39]. Without knowing which of these interactions and modifications are mutually exclusive and which prevail in a given micro-environment, it is simply not feasible to fit the experimental data to any of the current reaction-diffusion schemes. However, until we arrive at that point, useful information can be obtained by taking the FRAP data for what they worth. For example, evaluating the CDFs of the Mf and t 1/2 , one can compare the ensemble-average properties of different LBR forms and obtain valuable information as to whether or not specific mutations affect LBR dynamics and result in functional defects. A demonstration of this treatment is provided by the comparison we have made assessing the relative mobility of FL-LBR and Mod. From this assessment, it is clear that the mobility of the mutant protein in the NE is markedly lower than that of wild type LBR.
The lower mobility of Mod is contrasted by the apparent hyper-mobility of less radically truncated mutants, such as Gr and Icj, which are much shorter in amino acid sequence. This probably means that there are extensive intra-molecular interactions among the loop segments that connect the transmembrane domains to each other. Analogous interactions, which confer structural stability and ligand specificity in multi-spanning membrane proteins, have been identified over the years in several receptors and ion channels. With that as a given, it would not be wise to extrapolate directly from the linear sequence and the domain structure of LBR Notice that variation is much lower with Nt than with FL-LBR. (B) A typical FLIP experiment with Nt (10 successive bleaching-recovery cycles). The histograms show the changes in fluorescence intensity (I final /I initial ) in the nucleoplasm (NUC) and the cytoplasm (CYT). Bars in all panels, 6 μm. to the properties of truncated LBR, in an attempt to interpret their in vivo behaviour by a "lossof-function" approach logic.

internally deleted and truncated LBR mutants
To better understand the structure function relationships of LBR, apart from the mutants presented in the main text, we also examined the following mutants (S1A Fig): TL (missing the hydrophilic tailpiece); NtTM4 (missing the tailpiece and transmembrane domains I-III, V-VIII); ΔTd (missing the amino-terminal Tudor domain); TdRSTM1 (missing the tailpiece, the amino-terminal GD domain and transmembrane domains II-VIII) and TdTM1 (missing the amino-terminal RS and GD domains, as well as all sequences downstream to the transmembrane domain I). The subcellular distribution of these mutants was similar to that of FL-LBR (S4A Fig). However, when we assayed their diffusional mobility by FRAP and analyzed the data by the K-S test (S4D Fig), we arrived at some interesting conclusions. First, no statistically significant differences were found when we compared the mobility and diffusion rate of ΔTd and TdRSTMI in the NE and the bulk ER. This suggested that, unlike carboxy-terminal truncations, amino-terminal truncations affect severely the ability of LBR to bind to underlying sub-structure. Second, the TdRSTMI and the TdTMI mutants appeared to be more mobile and faster exchanging than FL-LBR at the NE. However, this was less apparent with the ΔTd mutant, although this (minimally truncated) protein was generally as mobile at the NE as was in the bulk ER (see above). When we attempted a similar comparison with the ER-distributed mutants, we did not find differences between TdRSTMI and ΔTD, but TdTMI and ΔCt behaved differently in terms of mobility and exchange rate. From these data we infer that although amino-terminal truncations affect primarily LBR binding to the nuclear acceptor sites, the different mutant proteins do not diffuse with the same ease along the membrane.