Skip Navigation


Human Molecular Genetics Advance Access originally published online on July 6, 2004
Human Molecular Genetics 2004 13(17):1857-1871; doi:10.1093/hmg/ddh205
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
13/17/1857    most recent
ddh205v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (31)
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Masny, P. S.
Right arrow Articles by Winokur, S. T.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Masny, P. S.
Right arrow Articles by Winokur, S. T.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

Human Molecular Genetics, Vol. 13, No. 17 © Oxford University Press 2004; all rights reserved

Localization of 4q35.2 to the nuclear periphery: is FSHD a nuclear envelope disease?

Peter S. Masny1, Ulla Bengtsson1, Seung-Ah Chung1, Jorge H. Martin1, Baziel van Engelen2, Silvere M. van der Maarel3 and Sara T. Winokur1,*

1Department of Biological Chemistry, University of California, Irvine, CA, USA, 2Department of Neurology, Nijmegen University, The Netherlands and 3Department of Human Genetics, Leiden University Medical Center, The Netherlands

Received May 18, 2004; Accepted June 29, 2004


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 REFERENCES
 
Facioscapulohumeral muscular dystrophy (FSHD) may be a new member of the class of neuromuscular diseases (NMD) due to defects in the nuclear envelope. Unlike other NMDs with primary defects in nuclear envelope proteins, however, FSHD may result from inappropriate chromatin interactions at the envelope. 3D Immuno-FISH and a novel method of 3D by 2D analysis using NucProfile were developed to examine nuclear organization of the FSHD genomic region. In contrast to most other telomeres, the FSHD region at 4q35.2 localizes to the nuclear periphery. This localization is consistent in normal myoblasts, myotubes, fibroblasts and lymphoblasts, does not vary significantly throughout the cell cycle, and is independent of chromosome territory effects. The nuclear lamina protein lamin A/C is required for FSHD region chromatin localization to the nuclear envelope, as the association is lost in lamin A/C null fibroblasts. As both normal and affected alleles (deleted for the subtelomeric repeat D4Z4) localize to the nuclear periphery, FSHD likely arises instead from improper interactions with transcription factors or chromatin modifiers at the nuclear envelope. Interestingly, it is not D4Z4 itself that mediates interaction with the envelope, as sequences proximal to D4Z4 (i.e. D4S139) localize closer to the nuclear periphery, perhaps accounting for the chromosome 4 specificity of the disease.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 REFERENCES
 
Facioscapulohumeral muscular dystrophy (FSHD, MIM 158900) remains one of the few neuromuscular disorders for which the causative pathophysiologic pathway is unknown (1). Clinically, the disease often manifests as facial weakness during adolescence with progression to the upper and then lower extremities later in adulthood (2–4). The primary tissue affected is skeletal muscle, although other symptoms may include sensorineural hearing loss, retinal telangiectasias, and epilepsy and mental retardation in the most severe cases (2,5,6). Asymmetric involvement of affected musculature is highly characteristic of FSHD. Complete absence of the pectoralis muscle and the condition pectus excavatum in some patients hint at a developmental nature of the disease (7,8).

The molecular genetics of FSHD reveals the simultaneous occurrence of two disease-causing polymorphisms on chromosome 4: a reduction in the copy number of a subtelomeric repeat, termed D4Z4, to 1–11 units, and the presence of cis elements distal to the D4Z4 repeat block (9,10). The chromosome 4q telomeric region distal to D4Z4 exists as two polymorphic variants, 4qA and 4qB (11). 4qA alleles are homologous to the chromosome 10q telomeric region, whereas 4qB alleles lack beta-satellite and other cis element sequences found distal to the D4Z4-like array on 10q. FSHD occurs as an autosomal dominant disease only in the presence of 4qA alleles with integral deletions of D4Z4 (10). In general, the size of the residual repeat array (i.e. the shorter the D4Z4 repeat block) is inversely correlated to the severity of the disease (12). FSHD alleles (deleted D4Z4, presence of beta-satellite) exhibit marked hypomethylation relative to normal alleles (13). The likely involvement of altered chromatin structure in the genesis of this disorder is further supported by rare cases of FSHD with no D4Z4 deletion, yet consistent hypomethylation of the 4q D4Z4 repeat block (13).

The prevailing mechanistic hypothesis for FSHD until a year ago was that of a position effect, with altered chromatin structure in the subtelomeric region of 4q influencing the expression of genes within or just proximal to D4Z4, i.e. those in the ‘FSHD gene region’ at 4q35.2 (14–16). Expression analysis of 4q35 genes remains a controversial area of scientific investigation, with the majority of laboratories finding no support for a position effect on gene expression (17–19). None of the unique genes within the FSHD gene region exhibit an altered expression pattern upon oligonucleotide and cDNA microarray analysis of FSHD muscle (17,19, Y.-W. Chen, personal communication). This is in contrast to a study in which several genes (FRG1, FRG2 and ANT1) were reported to be upregulated, although non-quantitative RT–PCR was utilized (16). More recently, real-time quantitative RT–PCR of FSHD muscle did not detect altered expression of these genes, supporting the microarray data and challenging the position effect hypothesis (18).

In considering other models for this disease, we examined the nuclear positioning of the FSHD region within the interphase nucleus in light of potential epigenetic influences on gene expression in this disorder. Epigenetic mechanisms of gene regulation can include DNA methylation, histone modifications, chromatin boundaries and insulator elements (20,21). A higher level of gene regulation, the functional architecture of the three-dimensional nucleus, has been more recently recognized (22–25). The mammalian nucleus is a highly compartmentalized structure with specialized domains carrying out distinct functions such as transcription, RNA processing and replication (26). Individual chromosomes occupy distinct and well-defined territories, and their specific intranuclear positions seem to be related to their gene density, transcriptional activity, replication timing and chromosome size (27,28). The location of a gene relative to specific nuclear domains can therefore dictate access to transcription and splicing factors, in turn influencing its pattern of expression (29).

One discrete nuclear domain is the nuclear envelope and associated lamina, which plays a role in gene expression, chromatin organization and differentiation (25,30). Several neuromuscular disorders result from deficiencies in components of the nuclear envelope, including Emery–Dreifuss muscular dystrophy (EDMD), Limb Girdle muscular dystrophy 1B (LGMD1B) and dilated cardiomyopathy (DCM) (31–33). EDMD, LGMD1B and DCM arise from a spectrum of mutations within the genes encoding emerin and lamin A/C. Muscular dystrophies involving the nuclear envelope are thought to result from altered patterns of tissue-specific differentiation and gene regulation (31–33), or from decreased nuclear envelope integrity with resulting fragility specifically of muscle cells. Support for the former hypothesis comes from several disorders with defects in lamin A/C with no muscle phenotype, but with deficiencies in the differentiation of other tissues (34,35).

The localization of telomeres in relation to the nuclear envelope has been studied in cultured cells and tissues from various species. In yeast and several plant species, telomeres are tethered to the nuclear envelope (36,37). In Drosophila, chromosomes are anchored to the nuclear envelope at multiple sites along their length, including the telomeres (38). In mammalian germ cells, telomeres attach to the nuclear envelope during early meiosis, forming a bouquet structure instrumental in chromosome alignment during homologous recombination (39). In contrast, in mammalian somatic cells, telomeres are largely confined to the nuclear interior and are evenly distributed throughout the nucleus in both quiescent and cycling cells (40–44). In support of this observation, biochemical and ultrastructural data demonstrate that telomeric DNA and the telomeric repeat binding factor (TRF) co-localize in individual, matrix-associated structures throughout the nuclear volume (40,45,46).

Although the majority of human telomeres are dispersed throughout the nucleus, we report here that the FSHD region near the chromosome 4q telomere is closely associated with the nuclear periphery. The localization of the 4q telomere near the periphery is consistent and non-random across several cell types and throughout the cell cycle, indicating that the nuclear positioning of the FSHD region has biological significance. Indeed, 4q35.2 localizes to the nuclear rim, an area rich in nuclear envelope proteins. This localization of 4q35.2 at the nuclear rim is not simply a reflection of chromosome territory (CT). Importantly, the positioning of the FSHD region at the nuclear periphery depends on the presence of lamin A/C, a major component of the nuclear lamina underlying the nuclear envelope. We propose that FSHD may be a new member of the class of nuclear envelope disorders. Rather than a primary defect in nuclear envelope proteins, FSHD may arise through aberrant interactions of distal 4q chromatin with the nuclear envelope.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
DISCUSSION
 MATERIALS AND METHODS
 REFERENCES
 
3D by 2D distribution analysis of genomic loci within the nucleus
Three-dimensional (3D) fluorescence in situ hybridization (FISH) studies of telomere localization in lymphoblast nuclei were performed using a PNA all-telomere probe. A consistent pattern emerged: few telomeres were seen near the nuclear envelope, whereas the majority were seen distributed throughout the nucleoplasm, with some degree of clustering (Fig. 1A). The localization of most telomeres to the nuclear interior is a commonly described finding in human somatic cells (40–43).



View larger version (43K):
[in this window]
[in a new window]
  		Figure 1. (A) Montage image of all telomere immuno-FISH in a 3D analyzed lymphoblast. The majority of telomeres are distributed throughout the nucleoplasm. Few telomeres are seen at the nuclear envelope. Anti-nuclear pore IF (blue), PNA all telomere probe (red). Image deconvoluted and sections shown at 0.7 µm intervals. (B) Stereopair image and corresponding montage. 4q35.2 localizes to the nuclear envelope in 3D lymphoblasts. 100% of 4q telomeres in all 31 nuclei analyzed were at the nuclear envelope. Anti-nuclear pore IF (stereopair digitally pseudocolored: green on top half of nucleus, blue on bottom half of nucleus; montage: blue), 4q35.2 probe (red). Image deconvoluted and viewing angles offset at 4°. Montage sections at 2 µm intervals. (C) Stereopair image and corresponding montage. The 10q telomere does not localize to the nuclear envelope. In 39 nuclei analyzed, only 12% of telomeres were seen at the nuclear envelope. Anti-nuclear pore IF (stereopair digitally pseudocolored: green on top half of nucleus, blue on bottom half of nucleus; montage: blue), 10q probe (red). Image deconvoluted and viewing angles offset at 4°. Montage sections at 2 µm intervals.

 
In contrast, the chromosome 4q telomere is localized at the nuclear envelope in lymphoblast 3D FISH studies (Fig. 1B). All 4q35.2 signals on all 31 nuclei scored were at the nuclear envelope (<0.2 µm from the nuclear envelope), with 46 of 62 signals intermixed with anti-nuclear pore IF signals (74% of total signals). As these lymphoblasts have both 4qA and 4qB alleles, there is no difference in localization of the normal polymorphic 4q alleles to the nuclear periphery. In stark contrast to the consistent localization of 4q telomere to the nuclear envelope, the 10q telomere, although virtually identical to the 4q telomeric region by sequence analysis, showed essentially no association with the nuclear envelope (Fig. 1C). Of 10q probe signals studied in 39 nuclei, only nine of 78 signals (12%) were <0.2 µm from the nuclear envelope (mean 1.1 µm from envelope). In addition, only two of the 10q signals seen at the envelope were intermixed with anti-nuclear pore IF signals (3% of total signals). Although this shows that 4q specifically associates with the nuclear envelope in lymphoblasts, localization studies for this disease should be carried out in cells from the affected tissue, i.e. myoblasts, as FSHD is primarily a disease of adult skeletal muscle. However, myoblasts have flattened nuclei often only a few microns deep (47). Although conventional 3D microscopy is well suited for studying localization within the spherical lymphoblast nucleus, it does not provide enough resolution to determine the position of a probe relative to the nuclear envelope in myoblast nuclei.

In order to examine the association of the FSHD region with the nuclear envelope in myoblast nuclei, we designed a novel ‘3D by 2D’ method of analysis, utilizing the geometric characteristics of the nuclear ‘rim’. The nuclear rim is the region closest to the edge of the nucleus in a 2D projection and represents a region of low nuclear volume and high nuclear surface area. A probe distributed randomly within the nucleoplasm would rarely be seen within the nuclear rim (Fig. 2A). On the other hand, a probe distributed evenly along the nuclear envelope would be well represented in the nuclear rim (Fig. 2B). Actual measurement of the distribution of DAPI-stained chromatin and IF nuclear pore signals confirms these distribution patterns (Fig. 2A and B).



View larger version (13K):
[in this window]
[in a new window]
  		Figure 2. (A) 3D Sphere projection onto a 2D circle. When all values of equal distance from the center of the circle are grouped, that group represents a cylindrical shell within the original sphere (purple). The 2D distance measurements from the center of the circle to the edge correspond to concentric shells in the 3D sphere. A distribution plot of the volume of these cylindrical shells shows the distribution of volume of a 3D sphere in terms of their 2D distance from center. The circumference of the shells grows larger as they expand from the center, whereas their height decreases as they approach the edge of the 2D projection. This creates a characteristic distribution plot, steadily rising with a drop-off near the edge. The left of the distribution plot is the center of the nucleus (r=0%). The right of the distribution plot is the nuclear edge (r=100%). This also represents the distribution of volume of an ellipsoid projected onto an ellipse. The region from 94 to 100% out from the center of the 2D projection of a nucleus, here called the nuclear ‘rim’, is represented in gray. It contains 4% of the nuclear volume. A distribution graph of DAPI signal intensity from seven DAPI stained nuclear 2D projections is shown (rim represented in red). The observed distribution closely matches the expected theoretical distribution. (B) The surface area of a sphere, projected onto a 2D circle, also follows a specific pattern. In this case, each distance from the center to the edge of the 2D projection corresponds to the area of the top of the shell for that position (in yellow above). The surface area distribution plot increases as the circumference of these cylinders increase, and also increases greatly as the cylinder tops steepen near the edge of the 2D circle. This creates a characteristic distribution pattern with a steep increase near the edge, which also applies for ellipsoids projected onto ellipses. The nuclear rim, represented in gray, contains 34% of the nuclear surface area. A distribution graph of nuclear pore signal intensity from nine IF nuclear pore stained nuclear 2D projections is shown (rim as defined using DAPI stain is represented in red, and nuclear pore signal measured beyond 100% is represented in pink). The observed distribution matches the expected theoretical distribution, with a steep increase near the edge. Note that a large amount of the edge ‘spike’ is beyond 100%, as expected because the nuclear pores span the nuclear envelope, extending beyond the inner nuclear border.

 
The FSHD region at 4q35.2 localizes to the nuclear rim
The nuclear position of 4q35.2 was measured in 2D projections of interphase nuclei of myoblasts, myotubes and fibroblasts. ‘Pies’ were generated which depict measured locations of the 4q subtelomeric FISH signals combined into one virtual nuclear quarter (Fig. 3). The probe used was LPT04Q, which hybridizes 275–500 kb from the telomere of chromosome 4q in close proximity to the D4Z4 repeat deleted in FSHD. The location of 4q35.2 across the different cell types examined is consistent. Especially of note is the number of signals located close to the nuclear ‘rim,’ the outer ring of the nuclear 2D projection, here defined as the region 94–100% of the distance from the center to the edge. A high percentage of 4q35.2 signals were found within this nuclear envelope rich region: 36% in myoblasts, 43% in myotubes and 34% in fibroblasts (Fig. 3), suggesting an association with the envelope. The remainder of the signals may very well be adjacent to the envelope on the upper or lower surface of the nucleus. However, it is not possible to measure this using 2D projections of the nuclei. In 2D, only signals located at the nuclear rim can show that a locus is positioned near the envelope. The conservation of the association of 4q35.2 with the nuclear rim across cell types implies that localization of the 4q subtelomeric region near the envelope has biological significance.



View larger version (34K):
[in this window]
[in a new window]
  		Figure 3. Virtual ‘pies’ combining the observed locations of 4q35.2 and distribution histograms summarizing the results. The label in the center corresponds to the cell type and chromosome represented. The percentage of signals seen in the nuclear ‘rim’ (region from 94% to the edge) is listed below the label (represented in gray in the pies and histograms). In all cell types, 4q35.2 is well represented in the nuclear rim. The size of the virtual nucleus is the average size of nuclei in the three data sets. Myoblast (n=538, median=90); myotube (n=238, median=92); fibroblast (n=522, median=88).

 
4q35.2 localization is not dependent on differentiation state or cell cycle phase
Myoblasts grown in differentiation media assume various forms. Some begin to assume the shape of elongated myotubes, with one or two nuclei (here referred to as pretube or paired nuclei). Others form multinucleated myotubes, with either linearly aligned or clustered nuclei. Finally, some solitary myoblasts do not differentiate, a developmental path parallel to quiescent satellite cells muscle tissue (48). Using the myotube results from Figure 3, we analyzed the position of 4q35.2 within these subgroups (Table 1). The position of 4q35.2 within the myotube subgroups demonstrates a consistent association with the nuclear rim, further supporting the integral association between the FSHD region at 4q35.2 and the nuclear envelope rich region. The slight variability seen with paired and pre-tube nuclei is likely due to the small number of nuclei observed with these patterns.


View this table:
[in this window]
[in a new window]
  		Table 1. Myotube 4q35.2 data classified according to the association of nuclei with neighboring nuclei
 
Chromosome localization can vary throughout different phases of the cell cycle (44). Myoblasts cannot be synchronized through standard methods, such as serum starvation or high density as this will cause them to differentiate. Two experimental approaches were used to determine whether 4q35.2 localization is cell cycle dependent. In one approach, cells were grown in the presence of both BrdU and Nocodazole for 6 h before harvesting, and subsequently detected with FITC conjugated anti-BrdU. The G2 cells were arrested in mitosis, and all unlabeled interphase cells were thus considered to be in G0 or G1 (Table 2). In a second approach, BrdU was added to the cultures 20 min prior to fixation in order to identify cells in S-phase. BrdU positive nuclei were categorized on the basis of their staining pattern and replication status (Fig. 4, Table 3). The 4q telomere is again associated with the nuclear rim across subgroups, although less consistently than in the previous analyses. This may reflect some temporary repositioning of the telomere during replication. Interestingly, replicated signals (doublets) were seen predominantly in the last (heterochromatic) stage of S phase, and 4q35.2 replication was asynchronous in 64% of all S-phase replicated nuclei. Asynchronous replication of this region has also been observed in other FSHD studies (49,50).


View this table:
[in this window]
[in a new window]
  		Table 2. Undifferentiated myoblast 4q35.2 data classified according to the staining pattern of a nucleus after 6 h BrdU staining
 


View larger version (83K):
[in this window]
[in a new window]
  		Figure 4. Images of nuclei after 20 min BrdU incorporation immediately before harvest. Stages of S-phase progression have a characteristic BrdU pattern. Localization of 4q35.2 does not vary throughout the cell cycle. Blue: DAPI, green: BrdU and red: 4q35.2 FISH. (A) ‘Intermediate’ characterized by light amounts of centrally located BrdU incorporation. (B) ‘Distributed’ characterized by even BrdU incorporation throughout the nucleus. (C) ‘Heterochromatic’ characterized by strong BrdU incorporation in nuclear regions containing heterochromatin. (D) None (G2) characterized by absence of BrdU incorporation and two pairs of replicated probe signals. S-phase progression patterns based on patterns described for human fibroblasts by Kennedy (83).

 

View this table:
[in this window]
[in a new window]
  		Table 3. Undifferentiated myoblast 4q35.2 data classified according to the staining pattern of a nucleus after 20 min BrdU staining
 
4q35.2 localization to the nuclear rim is independent of chromosome territory
The DNA of a single chromosome is not distributed throughout the nucleus, but rather collects in a contiguous region called the chromosome territory (CT) (24). However, the nuclear volume distribution used in the 3D by 2D analysis presumes that chromosomal DNA sequences are evenly distributed throughout the nucleus, and not limited to a territory. The 3D by 2D method of analysis was adapted to this limitation by using the distribution of the chromosome contents instead of the nuclear volume. Once measured, the distribution of a CT's chromatin should reflect the distribution of a sequence randomly positioned within the territory. The distribution of chromosome 4 was measured, as well as the large chromosomes 1 and 5, chosen because they have been found to have CT distributions similar to 4, i.e. near the nuclear edge (47). We also included chromosome 18 because its territory had been observed to be peripherally located (51). Finally, chromosome 10 was included because it has a subtelomeric D4Z4 sequence nearly identical to the one in the FSHD region on 4q. The chromosome territories are distributed differently within the nucleus; however, they share one feature: as predicted by the 3D by 2D method of analysis, only a small amount of chromatin should be found in the nuclear rim. The outer region of the 2D nuclear projection represents a shallow region unable to contain much chromosome ‘mass,’ when compared with the larger central areas. By summarizing the intensity of pixels within a territory when using a chromosome paint probe, distribution plots representing the individual chromosomes were made (Fig. 5). As expected, the amount of chromatin for all chromosomes drops off considerably at the nuclear rim, despite the density of heterochromatin near the nuclear envelope (30). Specifically, the peripherally located chromosome 4 has about 9% of its chromosome mass located within the nuclear rim. Despite this, 34–43% of the 4q35.2 signals are found in this region. Thus, the incorporation of the chromosome distribution data reiterates that the FSHD region is localized in a distinctive non-random pattern.



View larger version (31K):
[in this window]
[in a new window]
  		Figure 5. Double histograms of observed probe signal distributions and CT position in 2D projections. Probe signal distributions are shown with hollow bars. CT signal distributions are shown with solid areas. The left edge represents the center of the 2D nuclear projection. The area in red represents probe signal in the nuclear rim, whereas pink represents probe signal located just beyond the nuclear border. The double histograms are annotated with the probe used. All results are in myoblasts unless listed as ‘Fib’ (normal fibroblasts) or ‘LACF –/–’ (LACF –/– fibroblasts). The percentage represents the percentage of probe signals found in the nuclear rim. (94%+). Midpoint is the point at which the distribution area of a CT is equal on each side and parallels the median of probe signals. (A) Other telomeres do not show the same level of localization to the nuclear rim as 4q. Myoblast 1q (n=190, median=83), 1CT (n=52, midpoint=73); myoblast 10q (n=104, median=81), 10CT (n=50, midpoint=63); myoblast 5q (n=98, median=71), 5CT (n=52, midpoint=58); myoblast 18q (n=78, median=57). (B) The chromosome 4 centromere and 4p telomere do not show the same level of localization to the nuclear rim as 4q. Myoblast 4q (n=538, median=90), 4CT (n=50, midpoint=72); myoblast 4p (n=184, median=64); myoblast 4cent (n=136, median=63); fibroblast 4cent (n=178, median=57), 4CT (n=68, midpoint=74). (C) Telomere position in lamin A/C deficient fibroblasts compared with normal fibroblasts. The 1q control only shows a slight change in the position of the telomere after the loss of lamin A/C in its nuclear envelope (NE). However, the position of 4q is greatly changed: after the disruption of the NE, 4q no longer displays its predilection for the nuclear rim. Normal fibroblast 1q (n=386, median=84), 1CT (n=68, midpoint=77); normal fibroblast 4q (n=552, median=88); LACF –/– fibroblast 1q (n=158, median=79), 1CT (n=50, midpoint=63); LACF –/– fibroblast 4q (n=218, median=67), 4CT (n=50, midpoint=64).

 
Other telomeres do not similarly localize to the nuclear rim
As shown earlier, the chromatin of all chromosomes examined is nearly absent from the nuclear rim, even for those chromosomes with territories near the nuclear periphery in 2D projections. Thus, they are all equal candidates for analysis using the 3D by 2D method. The location of probes for the telomere regions of 1q, 5q, 10q and 18q were measured. None of the other telomeric regions display the predilection for the nuclear rim displayed by 4q35.2 (Fig. 5A). Although 1q and 10q show distributions somewhat higher than expected at the nuclear rim on the basis of territory distribution, 19 and 17% respectively, it is nowhere near the high number of 4q subtelomeric signals seen in this region. The 5q and 18q telomeres are essentially absent from the nuclear rim altogether. Indeed, our lymphoblast results indicated that most telomeres are not associated with the nuclear envelope (Fig. 1). To reiterate, a signal seen in central area of the 2D projection is not necessarily distant from the envelope, as it could be adjacent to the upper or lower nuclear envelope. A signal seen in the nuclear rim, however, must be adjacent to the nuclear envelope.

The centromere of chromosome 4 and the 4p telomere do not localize to the nuclear rim
In order to completely avoid the potential confounder of chromosome positioning within the nucleus, we compared the distribution of 4q telomere with chromosome 4 centromere and 4p telomere probes. Because these DNA sequences reside on same chromosome, they share the same chromosome distribution and thus should be similarly located if they were randomly placed within the territory. This was not the case, however (Fig. 5B). Neither the 4 centromere nor the 4p telomere shows the same association with the nuclear envelope that 4q35.2 does. The difference in position is striking and highly significant (P<2.2E–16, by Wilcoxon Rank Sum Test for both 4 centromeres in myoblasts and fibroblasts and 4p telomeres in myoblasts). This, interestingly, contrasts with findings by others that centromeres are localized more peripherally than both p and q telomeres in G0 lymphocytes (41,42). This may reflect the influence of cell type and cell cycle stage on nuclear organization, and highlights the importance of using the appropriate tissue for localization studies, such as myoblasts in FSHD.

Lamin A/C is required for localization of 4q35.2 to the nuclear rim
As mentioned previously, the nuclear rim, the outer circumference of a 2D projection of a nucleus, is a region characterized by a lack of nuclear volume and an abundance of nuclear envelope. 4q35.2 clearly demonstrates localization to this region, inferring an association between the nuclear envelope and the FSHD region. In order to more directly evaluate the possibility of an interaction between the FSHD region and the envelope, we studied 4q35.2 localization in lamin A/C null primary fibroblasts in which the lack of lamin A/C disrupts the nuclear envelope structure (52). We limited our study to nuclei that did not show morphological abnormalities such as herniations or folding, so that any differences noted would solely be the result of changes in the nuclear envelope at a sub-microscopic, molecular level. We included a 1q telomere probe as a control (Fig. 5C). While the 1q telomere was present in the nuclear rim of the normal fibroblast (25% compared with 19% in the myoblast), its localization to that region decreased only slightly when the nuclear envelope was disrupted (25% in normal fibroblast versus 20% in lamin deficient fibroblast). In contrast, the otherwise consistent presence of 4q35.2 in the nuclear rim (34% for normal fibroblasts), dramatically decreased, falling to 11% in the lamin A/C null fibroblasts, showing that its localization to the envelope-rich area is dependent on normal envelope integrity. The difference in localization change between 1q and 4q is visually obvious and highly significant (P<2.2E–16, see Materials and Methods).

Long D4Z4 arrays do not direct localization to the nuclear rim
In order to study the localization of 4q in FSHD, 3D by 2D analysis was performed on patient myoblasts. Because FSHD may involve aberrant muscle differentiation (17,19,53,54), the study was further refined by examining only myotubes (differentiated myoblasts). Furthermore, in order to determine whether there is differential localization of the two alleles (normal and affected/deleted) in FSHD myoblasts, a co-hybridization experiment was performed. The 4q35.2 specific probe, cosmid c88F8 (containing the locus D4S139), was co-hybridized with the D4Z4 probe, plasmid pBS-K3.3. These probes are 215 kb apart and will produce two closely spaced signals in most interphase nuclei (55). Whereas D4Z4 hybridizes to multiple loci within the human genome (14,56,57), D4S139 is a chromosome 4 specific probe that is present on both homologs. The intensity of the pBS-K3.3 (D4Z4) signal directly correlates with the size of the D4Z4 repeat array, so that the signal intensity from the affected (deleted) allele is weak or altogether absent as compared with the robust signal from the normal allele (which has >11 D4Z4 units). Thus, a c88F8 (D4S139) signal without an adjacent D4Z4 signal identified the deleted FSHD allele, and a D4S139 signal with an adjacent D4Z4 signal identified the normal homolog (Fig. 6). A 3-fold increase in intensity was assumed to be sufficient for differentiating the two D4Z4 alleles, although nearly all signal pairs far surpassed this cutoff revealing only one effectively measurable D4Z4 signal. Only 2% of nuclei images captured were unable to be scored, because the D4Z4 signals were <3-fold different. Three normal and two FSHD myoblast lines were analyzed by D4S139/D4Z4 co-hybridization (Table 4). There is no discernible difference in 4q35.2 localization between the normal and FSHD myotubes, nor is there any discernible difference in localization between the normal and affected alleles in FSHD nuclei. This suggests that any putative defective interaction with the nuclear envelope occurs at a higher level of molecular resolution than can be detected with fluorescence microscopy. Interestingly, the relative localization of the D4Z4 repeat may provide an explanation for the chromosome 4 specificity of the disease. In all five myoblast lines examined, the D4Z4 repeat is more centrally located in the nucleus than the D4S139 locus, with a combined P-value<1.3E–9 (by Wilcoxon Rank Sum Test). This implies that it is not the D4Z4 sequence itself that directs the localization to the nuclear rim, but rather a sequence closer to the chromosome 4 specific locus D4S139 (proximal to the D4Z4 repeat). The consistency of this finding suggests that the region of 4q35.2 responsible for localization at the nuclear rim is very close to, or within, the D4S139 locus. The D4S139 locus is in a chromosome 4 specific region, which can explain the differential localization of the 4q and 10q chromosome ends.



View larger version (28K):
[in this window]
[in a new window]
  		Figure 6. (A) FISH image of FSHD myotube nuclei labeled with 4q35.2 (green) and D4Z4 (red) probes. Only one of the two 4q35.2 signals has an adjacent D4Z4 signal. The D4Z4 signal was often seen internal to the 4q35.2 signal. (B) FISH image of normal myotube nuclei labeled with 4q35.2 (green) and D4Z4 (red) probes. Both 4q35.2 signals have a corresponding D4Z4 signal. Again, the D4Z4 signal was often seen internal to the 4q35.2 signal.

 

View this table:
[in this window]
[in a new window]
  		Table 4. Co-hybridization experiments with D4S139 and D4Z4
 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
MATERIALS AND METHODS
 REFERENCES
 
FSH, presents a challenge to molecular geneticists 12 years after the initial DNA mutation was identified. Testing pathogenic models such as a position effect, or deletion of a putative D4Z4 gene have yielded little insight into the disease. Presently, epigenetic models for FSHD appear to hold the most promise in our understanding of this complex disorder (13,21). One epigenetic modification relevant to FSHD is differential DNA methylation, as FSHD alleles have been shown to be hypomethylated (13). Epigenetic influences on gene expression can also result from differential nuclear localization of genomic loci (23,58–60). In light of altered gene regulation and myogenic differentiation in FSHD (17,19,53,54), we examined the localization of the 4q35.2 genomic region within the interphase nucleus. We demonstrate here that the FSHD region localizes consistently to the nuclear periphery, that lamin A/C is necessary for positioning of 4q35.2 at the nuclear envelope, and that the D4Z4 repeat itself is not responsible for this localization.

Our 3D immuno-FISH study of lymphoblasts reveals that telomeres in general are distributed throughout the nucleoplasm, with only a small fraction of telomeres located at the nuclear rim. This is in accordance with other studies (40–43). When specifically examining the 4q and 10q telomeres, only the 4q telomere is uniquely associated with the nuclear periphery. Although only 12% of the chromosome 10q telomeres were seen at the nuclear envelope, a striking 100% of 4q telomeres localized at the nuclear envelope.

Our 2D studies of myoblasts and myotubes correlate well with 3D analysis of 4q and 10q in lymphoblasts. The FSHD genomic region at 4q35.2 localizes to the nuclear envelope in all cell types examined, including proliferating myoblasts, fibroblasts and lymphoblasts, and differentiated myotubes. This nuclear organization does not vary significantly throughout different stages of the cell cycle, as determined by BrdU incorporation. Furthermore, the positioning of 4q35.2 at the nuclear rim is not simply a reflection of CT, as supported by several lines of evidence. First, chromosome 4 as a whole has little chromatin within the nuclear rim. Second, neither the chromosome 4 centromere nor the 4p telomere localizes to the nuclear rim, even though they reside in the same CT as the 4q telomere. Third, although other chromosome territories occupy the outer regions of the nucleus similar to chromosome 4, only the 4q telomere shows preferential association with the nuclear rim.

Interestingly, it is not the deleted D4Z4 repeat itself that mediates interaction with the nuclear envelope. Instead, chromosome 4 genomic regions just proximal to the D4Z4 repeat (D4S139) are more tightly associated with the nuclear periphery. This finding may account for the chromosome 4 specificity of the disease, as D4Z4 sequences are dispersed throughout the genome, yet only the chromosome 4 D4Z4 units are associated with FSHD (14,56,57). Our 3D interphase FISH results demonstrate that most D4Z4 sequences are distributed throughout the nucleoplasm (data not shown). In addition, the subtelomere of chromosome 10, on which D4Z4 is also present, does not localize to the nuclear periphery in lymphoblasts, and the p-arms of the acrocentric chromosomes, containing large blocks of D4Z4 sequences, associate with the nucleolus in the nuclear interior (61,62).

The localization of the 4q telomere to the nuclear envelope is consistent and non-random, suggesting that the positioning of the FSHD region at the nuclear periphery has biological significance. Indeed, FSHD chromatin may be directly associated with the nuclear envelope, as we demonstrate that lamin A/C is required for 4q35.2 localization to the nuclear periphery. In examining lamin A/C null fibroblasts from a LGMD1B pedigree (52), we found that the FSHD region at 4q35.2 no longer localized to the vicinity of the nuclear envelope. This dependence on lamin A/C for localization to the nuclear envelope is specific to the FSHD region at 4q35, as the whole chromosome 4 territory distribution does not change in lamin A/C null fibroblasts (Fig. 5C). Similar distributions of whole chromosome territories were also found in emerin mutated lymphoblasts and fibroblasts (63). Examination of locus-specific positioning, such as the FSHD region at 4q35.2, in these cells may provide further insight into neuromuscular disorders associated with the nuclear envelope.

At the nuclear periphery, chromatin is anchored to the inner nuclear envelope via lamin A/C, LAP2ß and BAF (31–33). Disruption of this peripheral chromatin organization is seen in neuromuscular disorders such as EDMD and in lamin A deficient, dystrophic mice (64,65). In these diseases, mutations of nuclear envelope proteins such as lamin A/C and emerin result in alterations in gene expression patterns and differentiation (30–32,57–59). Chromatin modifiers and transcription factors complexed with lamina proteins at the nuclear periphery likely mediate the influence of the nuclear envelope on gene expression. Most of these transcription factors are repressors, such as HP1, Oct-1, Rb associated E2F, HDAC and GCL, although transcription activators such as the insulin activator IPF-1 are tethered to the nuclear periphery as well (30,69,70). Thus, the nuclear periphery provides a scaffold for protein–DNA interactions key to the modulation of gene expression patterns in the cell. A complex binding to D4Z4, containing the repressor/activator YY1, HMG2B and nucleolin (16) may contribute to transcriptional regulation near the nuclear envelope.

Aberrant interaction of FSHD region chromatin with the nuclear envelope may account for a generalized disruption in myogenesis seen in FSHD. Gene expression profiling of FSHD muscle and cell culture assays of FSHD myoblasts indicate a defect in myogenic differentiation (17,19,53,54). Indeed, disease myoblasts in culture fuse at a faster rate than controls, suggesting that the muscle differentiation program in FSHD has been turned on at an earlier, perhaps premature timepoint (53,54). However, these alterations in myogenesis do not appear to arise from an increase or decrease in gene expression at 4q35, as has been suggested by the position effect hypothesis (17–19). Rather, epigenetic factors such as hypomethylation and nuclear localization may contribute to a more global effect on gene expression (13,21,71). Alterations in D4Z4 copy number affect myogenic differentiation in C2C12 myoblasts, with deformed myotube morphology and a reduced myotube fusion index resulting from increasing numbers of D4Z4 (72). Although the mechanism for this effect is difficult to reconcile with D4Z4 deletions in FSHD, the demonstration that D4Z4 can alter myogenesis in trans supports the potential influence of epigenetics on gene expression in this disorder.

The finding that FSHD region chromatin localizes to the nuclear periphery provides an alternate model for pathogenesis in this disease. In other characterized nuclear envelope disorders, it is suggested that a primary protein defect results in a secondary chromatin defect, causing dysregulation of expression (31–33,66,73). FSHD may be the first in the class of nuclear envelope disorders to be primarily due to chromatin defect. Such a model ties the interaction of 4q with the nuclear envelope together with many seemingly disparate findings in FSHD: alterations in differentiation, the lack of convincing evidence for a nearby cis effect on gene expression and hypomethylation of the affected allele. Although both normal and FSHD alleles localize to the nuclear periphery, gene expression is likely to be affected by differential interaction of transcription factors and chromatin modifiers, perhaps including DNA methyltransferases and demethylases. This localized domain at the nuclear periphery could induce aberrant expression not restricted to genes in cis on 4q35.2. Other genomic regions may associate with this domain at the nuclear periphery, leading to alterations in expression of genes elsewhere in the genome. As the specific genes adjacent to and affected by nuclear envelope defects in EDMD and LGMD1B have not been identified, these would be prime candidates for genes affected in FSHD as well.

In our model, short D4Z4 arrays and the presence of ß-satellite may act synergistically to alter recruitment of transcription factors and chromatin modifiers to FSHD region chromatin at the nuclear envelope. Many of the genes dysregulated in FSHD are direct targets of MyoD or involved in myogenesis, cellular differentiation and cell cycle control (17,19). The repression of MyoD prior to differentiation is mediated in part by histone deacetylase 1 (HDAC1) (74,75). Upon initiation of myogenesis, the retinoblastoma (Rb) protein is tethered to the nuclear lamina (76–78) and recruits HDAC1 from MyoD (74,75). This releases repression of MyoD, allowing for its acetylation and consequent activation of skeletal-muscle-specific genes. One possible mechanism for FSHD may be that disrupted chromatin associated with nuclear envelope and nuclear lamina proteins may fail to bind Rb, leading to persistent hypoacetylation of MyoD and alterations in myogenesis (79). This model provides an example of how changes in DNA–protein interactions at the nuclear envelope could affect signaling pathways leading to pathology in FSHD muscle.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
REFERENCES
 
Cells and tissue culture
The primary FSHD myoblast cell line F23 was generously provided by Denise Figlewicz (University of Michigan). The FSHD primary myoblast line F30 was developed in our laboratory from a muscle biopsy provided by Melanie Ehrlich (Tulane University) and Veda Vedanarayanan (University of Mississippi). Lamin A/C null primary fibroblasts (LACF.1) are described in the literature (52). Normal myoblasts (N23, N24 and N25) were procured from BD BioWhitaker. Normal lymphoblasts (N201) Rf.201 were transformed at Leiden University Medical Center. Normal fibroblasts (GM99) were generously provided by Leslie Redpath, Irvine, CA, USA. Myoblasts and fibroblasts were grown on laminin or poly-L-lysine (Sigma) coated chamber slides (Nunc) or secureSlip coverslips (Grace Bio Labs) in SkGM (Cambrex) and Dulbeco's modified Eagle's medium (DMEM)/15% fetal bovine serum (FBS) media, respectively. Myoblast cultures were differentiated into myotubes in DMEM/2% horse serum. For cell cycle experiments, cells were grown either with BrdU (10 µM) and Nocodazole (100 µg/ml) for 6 h, or BrdU (10 µM) 20 min before harvest.

Probes
The DNA probes utilized for FISH included specific telomere probes for chromosome 1q, 4p, 4q, 5q, 10q and 18q (Cytocell Ltd, UK distributed by Rainbow Scientific). Cytocell has reported the maximal distance from the telomere sequence (TTAGGG)n for the six telomere probes: 80 kb for chromosome 1q, 73 kb for the 4p telomere, between 275 and 500 kb for 4q, 245 kb for 5q, 270 kb for 10q and 290 kb for 18q. An alpha satellite DNA probe for all human centromeres (Oncor), an all telomeres specific probe (Oncor) and an all telomeres specific CY3 labeled PNA probe (Applied Biosystem) (80) were used; as well as whole chromosome painting probes from chromosome 1 (Oncor), chromosome 10 (Vysis) and chromosomes 4, 5 and 18 (Cambio, distributed by Vysis) were used. Cosmid c188F8 containing the locus D4S139, (215 kb proximal to the D4Z4 tandem repeat) and pBS-K3.3 (with a single D4Z4 unit) were isolated in our laboratory. Cosmid c88F8 and pBS-K3.3 were labeled with either biotin or digoxigenin using Biotin or Dig-Nick translation Mix (Roche).

3D- and immuno-FISH
Attached cells (myoblasts and fibroblasts) were fixed for ‘Three-dimensional Preserved Nuclei’ as described previously by Solovei (81). Lymphoblasts were grown in a T75 flask until they were fixed for 3D-FISH according to ‘Preparation and Fixation of Cells growing in Suspension’ by Solovei (82). Experiments were performed on multiple slides and slides were stored in 50% formamide/2xSSC for later FISH analysis. The FISH probes were diluted in hybridization solution according to manufacturers' protocol. One slide, at a time, was retrieved from storage in formamide/2xSSC. The slide was drained, and after very careful dabbing with bibulous filter paper to remove the remaining formamide without letting the cells dry, the probe mixture was as quickly as possible applied to the chosen spot. A coverslip was placed on the slide and sealed with rubber cement. Cellular DNA and probe DNA were denatured simultaneously by placing the slide on a hot block at 75°C for 2 min. Hybridization took place in a humid atmosphere at 37°C for 1–2 days. Post-hybridization washes in 0.4% SSC at 72°C for 2 min and 2xSSC/0.05% Tween at 37°C for 30 s were followed by detection of the biotin or digoxigenin labeled probes with various fluorochromes. The following fluorochromes were utilized: Alexa 488 streptavidin (Molecular Probes) and FITC or Rhodamine labeled anti-digoxigenin (Roche). The slides pre-incubated in either 4xSSC/1% bovine serum albumine (BSA) or ‘blocking solution–MB-1220’ (Vector) for 15 min, before 30 min detection with the fluorochromes diluted in the same blocking solution. Nuclei were either counterstained with 4',6-diamidino-2-phenylindole (DAPI) or, after post-hybridization washes, incubated with a monoclonal nuclear pore antibody (Covance) followed by detection of the nuclear pore antibody with AMCA conjugated anti-mouse secondary antibody (Vector) or CY3 conjugated anti-mouse secondary antibody (Jackson ImmunoResearch) combined with the required fluorochromes for detection of the FISH probes. In order to detect the incorporated BrdU, the slides were incubated in FITC anti-BrdU (Roche) immediately after the post-hybridization washes in 0.4% SSC at 72°C. Immunofluorescence with a monoclonal anti-skeletal myosin, fast antibody (Sigma) detected by a FITC conjugated anti-mouse secondary antibody (Molecular Probes) was performed in order to identify myotubes (data not shown). Lamin deficiency in LGMD1B cells was confirmed by immunostaining LGMD1B cells as well as normal fibroblast with a polyclonal anti-lamin A/C antibody (Santa Cruz) followed by a rhodamine conjugated anti-goat secondary antibody (Santa Cruz) (data not shown). Desmin immunostaining with a polyclonal anti-desmin Y-20 (Santa Cruz) followed by a rhodamine conjugated anti-goat secondary antibody (Santa Cruz) confirmed the myogenic phenotype of all myoblast cell lines (data not shown). The slides were finally mounted in an antifade (p-phenylenediamine) and viewed using a fluorescence microscope.

Fluorescence microscopy and deconvolution
The fluorescent signals were captured on an inverted widefield microscope (Nikon Eclipse TE2000-U) with an automated stage, using a Nikon Plan-Apo 63x/1.4 oil objective, a Nikon multiband filter set with separate excitation filters, and a Photometrics Coolsnap HQ CCD camera. The focal plane of the microscope was passed through the whole of the nucleus and digital image series were made at 0.1 or 0.4 µm z-steps. An image series was created for each filter's image set and saved. An ImageJ (a public domain open-source java-based image processing and analysis program, by Wayne Rasband of the National Institute of Mental Health, http://rsb.info.nih.gov/ij) plugin combined these image series into a single RGB composite image containing maximum projections of the image series in the xy, xz and yz planes.

Lymphoblast analysis
For lymphoblast analysis, the collection of 3D image stacks for 4q and 10q were combined and presented randomly by computer for evaluation. For each signal, a line was drawn on a signal-containing xy plane through the signal to the nearest edge of the nucleus as determined by the inner edge of the nuclear pore signals, interpolated between nuclear pores. A vertical slice was made along this line and presented to the observer. On this vertical plane, the distance from the signal center to the nearest nuclear edge was measured. Some signals were found beyond the nuclear edge defined by the innermost surface of the nuclear pores, and were recorded as being intermixed with nuclear pore signals. Image stacks presented were deconvoluted using Autodeblur (AutoQuant) and 3D reconstructions created with ImageJ.

Design of ‘3D by 2D’
‘3D by 2D’ analysis takes advantage of the geometric characteristics of a sphere or ellipsoid. In order to make inferences about the 3D nature of a nucleus given a 2D projection of its contents, we considered the distribution the nuclear contents would have in its 2D projection. If many loci were randomly distributed throughout the nucleoplasm, they would be more likely to locate to the central region of the 2D projection, as the nucleus is taller in the middle. At the very rim of the 2D projection, the corresponding nuclear height drastically falls off, and internally distributed probes would not be in this region (Fig. 1A). On the other hand, the distribution pattern of probes randomly distributed along the nuclear envelope shows a different pattern. The envelope associated probes could exist on the upper or lower edge of the nucleus, near its central axis, and thus be near the center of the nuclear projection. However, at the rim of the nucleus, the nuclear envelope steepens as it rounds the edge of the cell, increasing the likelihood that the envelope associated probes will be located in this region of the 2D projection (Fig. 1B). In short, the distribution pattern of the nucleoplasm in a 2D projection decreases near the rim of the projection, whereas the distribution pattern of the nuclear envelope increases near the rim of the projection. In fact, the difference in patterns of distribution becomes very drastic at the very edge of the nucleus. At 94% of the distance from the center of the nucleus to the edge, the distribution level of the envelope is three times as great as the distribution level of the nucleoplasm, and this increases exponentially approaching the nuclear edge. A locus evenly distributed within the nucleoplasm would be seen in the nuclear rim 4% of the time; however, a locus evenly distributed along the nuclear envelope would be found in the rim 34% of the time.

The region from 94% out from the center to the edge of the 2D projection is the region of interest in our 3D by 2D method of analysis. We refer to this region as the nuclear rim. If a locus preferentially is found at this region, a relationship between the locus and the envelope is strongly suggested. Note, however, that a locus that is preferentially seen in the center of a 2D projection can still be associated with the nuclear envelope, as it can be near the upper or lower edge of the nucleus. Thus, preferential positioning at the nuclear rim of a 2D projection indicates localization adjacent to the nuclear envelope, whereas preferential positioning in the center of a 2D projection does not provide information regarding an association with the envelope.

Analysis with NucProfile
We created an ImageJ plugin suite, NucProfile (www.fsh.biochem.uci.edu/suppdata/nuclearlocalization), to measure data and record annotations for our 2D images of nuclear projections. The process by which NucProfile did this is as follows:

Nucleus.
The DAPI channel (blue) for each RGB nuclear image was isolated and presented. An approximate circumference was interactively drawn around the nucleus. The SplineSnake (an ImageJ plugin for image segmentation using closed snake (active contour), by Mathews Jacob of the Swiss Federal Institute of Technology. http://bigwww.epfl.ch/jacob/software/SplineSnake) plugin was used to accurately define the border of the nucleus, using a closed snake active contour algorithm, and the border was recorded. An ellipse was then fitted to the nuclear border. The center of this ellipse was recorded as well as the absolute length of its major and minor axis. The RGB image was then restored. The nucleus' position in relation to nearby nuclei was annotated.

Locus signal.
After interactively placing a marker on a signal, the color channel of the signal was isolated and a region was automatically selected around the mark using a wand algorithm. The center of the region was recorded as the position of the signal. A theoretical line was then drawn from the center of the nucleus, through the signal, to the edge of the nucleus. The position of the signal along this line was recorded as the percentage distance from center. This measuring algorithm is uniquely sensitive to the closeness of a signal to the nuclear envelope, despite irregularities in the nuclear border. The angle of the theoretical line from the major axis was also recorded. The replication status of each signal was visually evaluated (one spot versus an adjacent pair) and recorded as well as any co-localization, if applicable.

BrdU.
The level of BrdU incorporation was subjectively evaluated as described by Kennedy (83) and annotated.

CT signal.
Each CT was roughly circumscribed manually, including large areas of nucleus not containing any territory staining, as well as regions beyond the nuclear edge. The color channel of the territory signal was isolated and the manually demarked region was limited to include only areas within the previously defined nuclear border. An automated threshold was applied using a modified Otzu algorithm plugin (an ImageJ plugin for automated thresholding, by Christopher Mei of INRIA Sophia. http://rsb.info.nih.gov/ij/plugins/otsu-thresholding.html) to calculate an appropriate threshold to differentiate the territory from background. Subsequently, the threshold value was applied to the entire originally demarked region, allowing for territory signal just outside the nuclear edge to be included in the territory. The percent distance (as described for loci) was measured for each pixel in the territory and added to a normalized distribution plot, weighted by the pixel intensity minus the median of the non-territory nuclear pixels. When performed on DAPI and nuclear pore stains, the entire nucleus and a surrounding band were included, and the intensity of background adjacent to the nucleus was subtracted from each pixel. If both territories were confluent, then they were measured as one nucleus, but weighted as two in the final analyses. Note that a position measurement of exactly 100% is unlikely owing to the pixel nature of the images, causing the dip and spike seen at 100% on the CT distribution plots.

During analysis, the different color channels were separated when possible. All datasets were computationally scanned and reviewed for similar results to ensure that no nucleus was analyzed more than once. All data sets were blindly re-evaluated to ensure accuracy. Nuclei were excluded from study if: (1) the nucleus was misshapen (not round or oval, especially relevant for LACF1 data); (2) there were more or less than exactly two spots (signals indicating replication were included and reported) for each probe (except D4Z4); or (3) no discernible condensed CT stain was seen; (4) normal and affected chromosome 4 homologs could not be distinguished with certainty. The resultant data was then imported into R (The R Project for Statistical Computing: a language and environment for statistical computing and graphics, by John Chambers et al. of Bell Laboratories, http://www.r-project.org), and grouped as applicable. For signal positions, the median value of the distance from the nucleus center was calculated, as well as the percentage of signals seen in the nuclear rim. ‘Virtual nucleus’ pie graphs were created with R. For CT distributions, the midpoint of the distributions (50% of distribution area on each side, which corresponds to the median of the spot positions), as well at the percentage of signal in the nuclear rim were calculated.

In some experiments, the nuclear periphery was identified by immunostaining with a nuclear pore antibody instead of by DAPI staining. The nuclear edge is 3% further out when measured at the outer edge of the nuclear pore staining compared with the outer edge of DAPI/chromatin staining (calculated comparing the number of pixels contained in 20 nuclei stained by both methods, median factor was 1.06). Percentage values of nuclear pore slides were scaled by 1.03 to compensate for this. The following slides were scored in this manner: 1 of 4 1q myoblast slides (signal n=52), 1 of 3 1q LACF –/– slides (signal n=70), 1 of 6 4q LACF –/– slides (signal n=70), the 1CT LACF –/– slide, the 4CT LACF –/– slide, the 4CT myoblast slide, and the 10CT myoblast slide.

Statistical analysis of 1q and 4q fibroblast and LACF –/– data was performed using a modified bootstrapping technique. Bins of the distributions, separated at 43, 52, 60, 66, 72, 82, 85, 88, 91 and 94%, were set, such that a probe distributed evenly within the nuclear volume would be expected to have at least five samples in each bin with n=158, using the mathematical theoretical expected distribution. A random sample of equal size to the 1q fibroblast dataset was taken from the dataset. A chi-square statistic was calculated with this data sample against the mathematical expected distribution and subsequently divided by the sample size, giving a chi-square per observation, which was used as an estimator of the sample ‘abnormality’. The 1q fibroblast chi-square per observation was subtracted from a similarly made 1q LACF –/– random sample chi-square per observation, giving a single ‘abnormality change’ value for the two random samples. This was repeated for 1000 iterations total, and again repeated with the 4q datasets 1000 times. The 1000 1q abnormality change measures were compared to the 1000 4q abnormality change measures using the Student's t test, giving P<2.2E–16, which is the lowest P-value the default R statistical package will return.


   ACKNOWLEDGEMENTS
 
We are grateful to our esteemed colleagues Drs Melanie Ehrlich, Denise Figlewicz and Veda Vedanarayanan for providing FSHD myoblasts, and Pamela Flodman for helpful suggestions regarding the statistical analysis. Generous support of the Shaw-Fisher families is acknowledged. This work was generously supported by grants from the Muscular Dystrophy Association (S.T.W., S.v.d.M.), the FSH Society, Inc. (U.B.) and NIAMS/NIH (S.T.W.).


   FOOTNOTES
 
* To whom correspondence should be addressed at: Department of Biological Chemistry, 202 Sprague Hall, University of California, Irvine, Irvine, CA 92697, USA. Tel: +1 9498242750; Fax: +1 9498249547; Email: stwinoku{at}uci.edu


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 REFERENCES
 

  1. Tawil, R. (2004) Facioscapulohumeral muscular dystrophy. Curr. Neurol. Neurosci. Rep., 4, 51–54.[Medline]

  2. Padberg, G.W. and Adams, C. (2000) Facioscapulohumeral muscular dystrophy. In Pulst, S.-M. (ed.), Neurogenetics. Oxford University Press, Oxford, UK, pp. 105–116.

  3. Lunt, P.W., Compston, D.A. and Harper, P.S. (1989) Estimation of age dependent penetrance in facioscapulohumeral muscular dystrophy by minimising ascertainment bias. J. Med. Genet., 26, 755–760.[Abstract]

  4. Tawil, R., Figlewicz, D.A., Griggs, R.C. and Weiffenbach, B. (1998) Facioscapulohumeral dystrophy: a distinct regional myopathy with a novel molecular pathogenesis. FSH Consortium. Ann. Neurol., 43, 279–282.[CrossRef][ISI][Medline]

  5. Padberg, G.W., Brouwer, O.F., de Keizer, R.J.W., Gruter, A.M., Wijmenga, C., Grote, J.J. and Frants, R.R. (1992) Retinal vascular disease and sensorineural deafness are part of facioscapulohumeral muscular dystrophy. Am. J. Hum. Genet., 51, A104.

  6. Funakoshi, M., Goto, K. and Arahata, K. (1998) Epilepsy and mental retardation in a subset of early onset 4q35-facioscapulohumeral muscular dystrophy. Neurology, 50, 1791–1794.[Abstract]

  7. Padberg, G. (1982) Facioscapulohumeral disease. Doctoral Thesis, University of Leiden, Intercontinental Graphics.

  8. Tyler, F.H. and Stephens, F.E. (1950) Studies in disorders of muscle: II: clinical manifestations and inheritance of facioscapulohumeral dystrophy in a large family. Ann. Int. Med., 32, 640–646.[ISI][Medline]

  9. van Deutekom, J.T., Wijmenga, C., van Tienhoven, E.A.E., Gruter, A.-M., Hewitt, J.E., Padberg, G.W., van Ommen, G.-J.B., Hofker, M.H. and Frants, R.R. (1993) FSHD associated DNA rearrangements are due to deletions of integral copies of a 3.2 kb tandemly repeated unit. Hum. Mol. Genet., 2, 2037–2042.[Abstract/Free Full Text]

  10. Lemmers, R.J.L.F., de Kievit, P., Sandkuijl, L., Padberg, G.W., van Ommen, G.J.B., Frants, R.R. and van der Maarel, S.M. (2002) Facioscapulohumeral muscular dystrophy is uniquely associated with one of the two variants of the 4q subtelomere. Nat. Genet., 32, 235–236.[CrossRef][ISI][Medline]

  11. van Geel, M., Dickson, M.C., Beck, A.F., Bolland, D.J., Frants, R.R., van der Maarel, S.M., de Jong, P.J. and Hewitt, J.E. (2002) Genomic analysis of human chromosome 10q and 4q telomeres suggests a common origin. Genomics, 79, 210–217.[CrossRef][ISI][Medline]

  12. Kissel, J.T. (1999) Facioscapulohumeral dystrophy. Semin Neurol., 19, 35–43.[ISI][Medline]

  13. van Overveld, P.G., Lemmers, R.J., Sandkuijl, L.A., Enthoven, L., Winokur, S.T., Bakels, F., Padberg, G.W., van Ommen, G.J., Frants, R.R. and van der Maarel, S.M. (2003) Hypomethylation of D4Z4 in 4q-linked and non-4q-linked facioscapulohumeral muscular dystrophy. Nat. Genet., 35, 315–317.[CrossRef][ISI][Medline]

  14. Winokur, S.T., Bengtsson, U., Feddersen, J., Mathews, K.D., Weiffenbach, B., Bailey, H., Markovich, R.P., Murray, J.C., Wasmuth, J.J., Altherr, M.R. and Schutte, B.C. (1994) The DNA rearrangement associated with facioscapulohumeral muscular dystrophy involves a heterochromatin-associated repetitive element: implications for a role of chromatin structure in the pathogenesis of the disease. Chromosome Res., 2, 225–234.[CrossRef][Medline]

  15. Hewitt, J.E., Lyle, R., Clark, L.N., Valleley, E.M., Wright, T.J., Wijmenga, C., van Deutekom, J.C., Francis, F., Sharpe, P.T., Hofker, M. et al. (1994) Analysis of the tandem repeat locus D4Z4 associated with facioscapulohumeral muscular dystrophy. Hum. Mol. Genet., 3, 1287–1295.[Abstract/Free Full Text]

  16. Gabellini, D., Green, M.R. and Tupler, R. (2002) Inappropriate gene activation in FSHD: a repressor complex binds a chromosomal repeat deleted in dystrophic muscle. Cell, 10, 339–348.

  17. Winokur, S.T., Chen, Y.-W., Masny, P.S., Martin, J.H., Ehmsen, J.T., Tapscott, S.J., van der Maarel, S.M., Hayashi, Y. and Flanigan, K.M. (2003) Expression profiling of FSHD muscle supports a defect in specific stages of myogenic differentiation. Hum. Mol. Genet., 12, 2895–2907.[Abstract/Free Full Text]

  18. Jiang, G., Yang, F., van Overveld, P.G., Vedanarayanan, V., van der Maarel, S.M. and Ehrlich, M. (2003) Testing the position-effect variegation hypothesis for facioscapulohumeral muscular dystrophy by analysis of histone modification and gene expression in subtelomeric 4q. Hum. Mol. Genet., 12, 2909–2921.[Abstract/Free Full Text]

  19. Venance, S.L., Henderson, D., Sowden, J., Welle, S., Thornton, C.A. and Tawil, R. (2003) Expression profiling: is there a role for vascular smooth muscle dysfunction in facioscapulohumeral muscular dystrophy (FSHD)? FSHD International Research Consortium Workshop, November 4, Los Angeles, CA.

  20. Turner, B.M. (2001) Chromatin and Gene Regulation: Mechanisms in Epigenetics. Blackwell Science, Malden, MA.

  21. Bickmore, W.A. and van der Maarel, S.M. (2003) Perturbations of chromatin structure in human genetic disease: recent advances. Hum. Mol. Genet., 12, R207–213.[Abstract/Free Full Text]

  22. Jackson, D.A. (2003) The principles of nuclear structure. Chromosome Res., 11, 387–401.[CrossRef][ISI][Medline]

  23. Francastel, C., Schubeler, D., Martin, D.I. and Groudine, M. (2000) Nuclear compartmentalization and gene activity. Nat. Rev. Mol. Cell. Biol., 1, 137–143.[CrossRef][ISI][Medline]

  24. Cremer, T. and Cremer, C. (2001) Chromosome territories, nuclear architecture and gene regulation in mammalian cells. Nat. Rev. Genet., 2, 292–301.[CrossRef][ISI][Medline]

  25. Marshall, W.F. (2003) Gene expression and nuclear architecture during development and differentiation. Mech. Dev., 120, 1217–1230.[CrossRef][ISI][Medline]

  26. Spector, D.L. (2001) Nuclear domains. J. Cell Sci., 114, 2891–2893.[ISI][Medline]

  27. Sun, H.B., Shen, J. and Yokota, H. (2000) Size-dependent positioning of human chromosomes in interphase nuclei. Biophys. J., 79, 184–190.[Medline]