Christina Machado, Claudio E. Sunkel, and Deborah J. Andrew
The Journal of Cell Biology, Volume 141, Number 2, April 20, 1998 321-333
reviewed by Will White
Individuals suffering from autoimmune diseases produce autoantibodies, which bind to epitopes in their own tissues. These autoantibodies are often used to identify new proteins, especially nuclear proteins. The authors used autoimmune sera from patients with scleroderma, an autoimmune connective tissue disease. Scleroderma autoantibodies often bind to antigens on metaphase chromosomes.
Titin is a long, elastic protein that occurs in the sarcomeres of striated
muscle. It is a large protein (2993-3700 kD) that extends from the
Z-disk to the M-line of a relaxed sarcomere. It is thought to provide
a žmolecular scaffoldÓ for the assembly of sarcomeres into myofibrils.
Chromosomal staining with human scleroderma serum
The autoimmune sera of 40 human scleroderma patients was screened and one serum was identified that stained the mitotic chromosomes of human epithelial cells (HEp-2) and Drosophilaembryos. Figure 1 A shows this chromosomal staining. Chromosomes from both species are shown in all stages of mitosis and stained with scleroderma autoimmune serum (green), propidium iodide (red) to stain DNA, and both (yellow in areas of overlap). In both species, both types of staining yield images typical of chromosomes in the appropriate phases of mitosis. When merged, both stains overlap perfectly to form a yellow image. This indicates that the scleroderma serum is binding to an antigen that is closely associated with the entire length of mitotic chromosomes. During interphase, the antigen seems to be distributed evenly throughout the nucleus. Because the interphase DNA is uncondensed and also distributed throughout the nucleus, it is difficult to tell if the antigen is still closely associated with the DNA during this phase.
Cloning the Drosophila chromosomal antigen
When the same human scleroderma autoimmune serum was used to screen
a Drosophila genomic expression library, a gDNA (designated
LG) coding for a 71 amino acid repeat rich in P, V, E, and K was isolated.
Several cDNAs (designated KZ, NB, and JT) were also isolated from a cDNA
Figure 2 A shows a restriction map of the relevant region of the Drosophila chromosome with the locations of the various genomic phage clones, cDNAs, and gDNA indicated. KZ is thought to be an NH2 terminus because it contains a methionine start codon and an 882 nt ORF. NB contains a 1kb ORF.
Polyclonal antisera were raised against the protein products of LG and KZ. These antibodies were then used to stain HEp-2 and Drosophila embryo chromosomes in an experiment similar to that shown in figure 1 A. a-LG (figure 1 B) anda-KZ (figure 1 C) stained mitotic chromosomes in both species, and the green protein stain overlapped perfectly with the red propidium iodide DNA stain, as indicated by the uniformly yellow merged image. These results confirm that the LG and KZ sequences both code for peptides that are localized in mitotic chromosomes. As expected, these cDNAs translate into polypeptides that are very likely fragments of the chromosomal antigen recognized by the scleroderma autoimmune serum.
Comparison of Drosophila and vertebrate titin
The KZ cDNA was translated and used in a BLAST search to find similar protein sequences. Figure 2 B shows the alignment between KZ, chicken skeletal titin, and human cardiac titin. The asterisks and periods below the sequences indicate identical and conserved residues, of which there are many: the KZ ORF shows 28.6% identity/58.3% similarity to chicken skeletal titin and 27.4% identity/56.8% similarity to human cardiac titin. The NB ORF has slightly lower similarity to vertebrate titins (18.4%/48.9%; 17.2%/47.7%, respectively). These similarities are not immense, but are significant considering the phylogenetic distance between vertebrates and insects.
The JT cDNA and LG gDNA are not highly conserved with respect to vertebrate titin, but they contain PEVK-rich sequences (shown in figure 3) and they authors suggest that they correspond to the elastic PEVK-rich region of the vertebrate protein.
Because of these sequence similarities, the authors declare that their Drosophila gene encodes a nuclear protein homologous to vertebrate titin.
Expression of Drosophila titin in striated muscles
To determine the relationship between nuclear and muscular titin, the authors looked at expression of nuclear titin mRNA and protein in Drosophila embryos. Figure 3 shows the results of in situ RNA hybridization (left panels) and immunostaining (right panels) in embryos of various ages. The authors do not indicate which antisense RNA or antibodies they used. D-Titin mRNA and protein begin to accumulate in the relatively undifferentiated visceral and somatic mesoderm (A/AŪ) and are expressed up to at least embryonic stage 17 (D/DŪ) when they are present throughout the pharyngeal, somatic, and visceral musculature (Drosophila visceral muscle is striated and would contain titin, unlike the visceral žsmoothÓ muscle of vertebrates). The authors claim that this pattern of expression indicates that the chromosomal and muscular titins are the same. If we assume they used the cDNAs and antibodies created above to stain the tissue, this conclusion seems reasonable. Of course, if titin is indeed present in all mitotic chromosomes, one would expect it to have extremely widespread distribution in a rapidly dividing embryo regardless of its participation in muscular development. The next experiment, however, will confirm that the chromosomal titin also occurs in the sarcomere.
Figure 4 shows immunostains of adult Drosophila thoracic muscle sarcomeres using a-KZ and a LG antibodies and the scleroderma autoimmune serum.
Panel A shows that a-KZ, which should recognize the NH2 terminus of titin, binds only to the Z disks of the sarcomeres. This is expected, as the NH2terminus of vertebrate titin binds to a-actinin in the Z-disk.
In panel B, both a-KZ and Texas red-phalloidin are used. The latter stains the I-band actin. The stains do not overlap, indicating that a-KZ is binding only to a Z-disk localized antigen and not to any other antigen in the I-band.
Both a-KZ and the scleroderma autoimmune serum were used in panel C. Again, the a-KZ stains only the Z-disk; the autoimmune serum also stains at the Z-disk as indicated by faint yellow color there. However, the autoimmune serum also binds to antigen along the entire sarcomere, including an intense stain at what is likely the M-line. The authors explain the staining outside the Z-disk as being the result of non-titin antibodies in the serum. Although they seem to find this an undesirable result, it seems reasonable to expect that the autoimmune serum would bind epitopes along the entire length of titin, which in vertebrates extends from the Z-disk to the M-line.
Similar results are seen in panel D, where a-KZ and a-LG antibodies are both used. a-LG has heavy binding at the Z-disk, and yellow areas of overlap with a-KZ are seen there, but a-LG also appears to have light staining throughout the sarcomere. The fainter a-LG staining could simply be background: the control preimmune a-LG serum did show accumulation on myofibrils. The authors claim that a-LG is binding exclusively to the Z-disk and the M-line, and that the latter staining is a result of cross-reactivity to non-titin antigens. However, in vertebrates, the PEVK rich elastic region of titin occurs in the I-band. If the LG sequence is indeed homologous to the PEVK-rich region, as the authors claim, then a-LG should bind not to the Z-disk but to the I-band in a pattern similar to the Texas red-phalloidin stain in panel B.
Panel E shows a double stain with a-KZ, and serum from a human myasthenia gravis patient, which should bind to the major immunogenic region (MIR) of titin near the I/A band junction. As in previous panels, a-KZ binds exclusively to the Z-disk. The MIR serum shows almost complete overlap with the a-KZ stain, which is unexpected: the Z-disk and I/A junction should be quite distinct. The authors explain this result by suggesting that confocal microscopy has insufficient resolving power to distinguish the two bands. However, the fact that the human MIR antibodies even bind to Drosophila myofibrils suggests that Drosophila has a homolog to vertebrate titin.
Panel F shows double staining with a-KZ and anti-Zr5/Zr6 polyclonal antibodies, which recognizes the Z-repeats that bind to a-actinin in vertebrate titin. Both sets of antibodies bind at the Z-disk, with near perfect overlap. This provides further evidence of the homology between the Drosophila protein and vertebrate titin.
Panel G shows a-KZ staining the Z-disks of larval visceral muscle. A light micrograph not shown confirms that the fluorescent bands overlap the phase-dark Z-disks.
The authors find the evidence from figures 3 and 4 sufficient to show that the protein identified by the autoimmune serum in figure 1 is a homolog to vertebrate titin that is both nuclear and sarcomeral. They name the new gene D-Titin and the protein product D-TITIN.
Molecular Weight of D-TITIN
If D-TITIN is the homologue to vertebrate titin, it must be large enough to span half a sarcomere. This corresponds to a molecular weight of 2993 - 3700 kD. Figure 5 is a series of Western blots using a-KZ and a-LG antibodies. Panel A shows that in post-myogenesis Drosophila embryos, titin migrates in the expected megadalton range. A megadalton-size titin band is also seen in panels B and C, which are pre-myogenesis Drosophila embryos and non-muscular HeLa cells, respectively. The titin in these cells must be chromosomal, since figure 1 showed only chromosomal binding for the two antibodies used. Thus both chromosomal and muscular D TITIN have the size expected for a titin protein.
Staining of human chromosomes with anti-titin antibodies
Figure 6 shows a screening of vertebrate anti-titin antibodies in human HEp-2 cells. The five antibodies shown all bind strongly to the mitotic chromosomes and have good overlap with a propidium iodide DNA stain. Antibodies with epitopes on the A-band and M-line bound especially well; and the MIR polyclonal antibodies also stained non-chromosomal material which the authors claim is part of the mitotic apparatus. Not shown were antibodies that bind to the NH2terminus of titin, which did not bind to HEp-2 chromosomes, and antibodies that bind to the I-band region, which bound to the chromosomes very weakly. Figure 6 B is a map of the various binding locations of the a-titin antibodies.
The failure of the NH2 terminus-directed antibodies to bind to chromosomal titin suggests that the termini of the muscular and chromosomal isoforms of D-TITIN may undergo different splicing after transcription. Unpublished results of in situ hybridization to polytene chromosomes indicate that there is a single D-Titin gene, not two distinct muscular and chromosomal titin genes.
Additional unpublished data using polytene chromosomes indicate that D-TITIN remains bound to uncondensed chromosomes during interphase. Future publication of these data would be useful in establishing the exact nature of the interaction between D-TITIN and the chromosome.
Function of chromosomal titin
The authors suggest that chromosomal titin is part of the molecular scaffold that determines the shape of the condensed mitotic chromosomes, much as muscular titin aids in the assembly of sarcomeres into myofibrils.
The results of this paper provide convincing evidence for the existence of a Drosophila homologue to vertebrate titin that occurs in both chromosomal and sarcomeral isoforms. The only real weaknesses in this paper are the unexpected results, noted above, of the double immunostaining of the sarcomeres in figure 4 C, D, and E. These results merely suggest that muscular D-TITIN is not a perfect structural homologue of vertebrate titin.
The authors suggest two possible avenues of future research using the results of this paper: use of titin as an autoantigen in diagnosis of human scleroderma, and the use of the Drosophila homologue of vertebrate titin to study titin function in a simpler genetic system.
Additional research questions that could be asked to follow up this investigation are:
What is the sequence of the D-Titin gene? Because of the size of the mRNA, it is difficult to isolate entire cDNA sequences. However, the cDNAs already isolated could be used as probes to sequence the entire gene using chromosome walking. One complication with this strategy is the probable existence of alternate splicings of the D-TITIN NH2 terminus. The KZ cDNA could be used as a probe to sequence the chromosomal NH2 terminus, but to sequence the alternate muscular terminus a different probe must be used. Possibly a portion of the NH2 terminus of the vertebrate titin gene could be used as a probe if there is sufficient homology between the two genes. Hopefully the complete D-titin sequence would contain a consensus splice site for the two alternate NH2 termini. Once the D-Titinsequence is obtained, a more complete comparison can be made between the Drosophilaand vertebrate versions of the gene.
To what other chromosomal scaffold proteins does D-TITIN bind?
If D-TITIN is indeed part of the chromosomal scaffold, it likely
associates with other scaffold proteins, much as the SMC and condensin
proteins form a scaffold complex. A two-hybrid experiment could identify
proteins in an expression library that interact with D-TITIN. Of
course, since titin is a huge protein, it would be difficult to choose
a single domain to incorporate into the žbaitÓ fusion protein. Since
the NH2 terminus is known to differ between muscular and chromosomal
isoforms of D-TITIN, it is possible that region of the protein is responsible
for interaction with other chromosomal proteins. Using portions of
the NH2 terminus as the bait would be a good starting point
in a two-hybrid experiment. Such an experiment would probably identify
a protein like histone or topoisomerase II as positively binding to D-TITIN.