A peptide extension dictates IgM assembly.

Professional secretory cells can produce large amounts of high-quality complex molecules, including IgM antibodies. Owing to their multivalency, polymeric IgM antibodies provide an efficient first-line of defense against pathogens. To decipher the mechanisms of IgM assembly, we investigated its biosynthesis in living cells and faithfully reconstituted the underlying processes in vitro. We find that a conserved peptide extension at the C-terminal end of the IgM heavy (Ig-µ) chains, termed the tailpiece, is necessary and sufficient to establish the correct geometry. Alanine scanning revealed that hydrophobic amino acids in the first half of the tailpiece contain essential information for generating the correct topology. Assembly is triggered by the formation of a disulfide bond linking two tailpieces. This induces conformational changes in the tailpiece and the adjacent domain, which drive further polymerization. Thus, the biogenesis of large and topologically challenging IgM complexes is dictated by a local conformational switch in a peptide extension.

Surface IgM of CLL cells displays unusual glycans indicative of engagement of antigen in vivo.

Surface IgM (sIgM) has a key influence on the clinical behavior of chronic lymphocytic leukemia (CLL). We now report that it exists in 2 forms with different N-glycosylation patterns in the mu-constant region. One glycoform is similar to normal B cells in bearing mature complex glycans common to most cell-surface glycoproteins. The other is an immature mannosylated form more characteristic of mu chains in the endoplasmic reticulum. Unmutated CLL (U-CLL) expresses a higher proportion of mannosylated surface mu chains than mutated CLL. Normal B cells express only the mature glycoform but can express the immature form after persistent engagement of sIgM, suggesting that glycan modification is a consequence of antigen exposure. CLL cells express variable proportions of the mannosylated form and can revert to the mature form after incubation in vitro. Both glycoforms are able to signal after sIgM engagement in vitro, leading to enhanced tyrosine phosphorylation. These findings support the concept that CLL cells are continuously exposed to antigen in vivo, driving the N-glycosylation pattern of expressed sIgM toward a mannosylated form, especially in U-CLL. Strikingly, this is reminiscent of follicular lymphoma, where mannosylated Ig is expressed constitutively via N-glycosylation sites in the variable region, suggesting a functional asset for this glycoform.

Production and characterization of monoclonal antibodies to IgM of Pacific herring (Clupea pallasii).

Pacific herring (Clupea pallasii) have a central role in the North Pacific ecosystem as a forage fish species and are natural reservoirs of several important finfish pathogens, including Viral hemorrhagic septicemia virus (VHSV). Here, we report the identification of the gene encoding the immunoglobulin mu (IgM) heavy chain, as well as the development and characterization of monoclonal antibodies (MAbs) that specifically react with Pacific herring IgM. Pacific herring immunoglobulin was purified and consisted of heavy and light chains of approximately 80 and 25 kDa. Three hybridoma clones were initially identified by ELISA as reactive with purified immunoglobulin but only one clone was able to detect an 80 kDa protein in Pacific and Atlantic herring (Clupea harengus) whole plasma by denaturing western blot. However, all three MAbs were able to precipitate an 80 kDa protein from Pacific herring and LCMS sequencing of peptide fragments derived from this protein matched the predicted amino acid sequence of the cloned, heavy chain gene. In addition, two of the MAbs stained cells within the putative lymphocyte gates for the spleen, anterior kidney and posterior kidney but were not reactive for myeloid/granulocyte gates, which is consistent with these MAbs reacting with surface IgM⁺ B-cells. To our knowledge, this is the first report of IgM-related gene sequences and anti-IgM monoclonal antibodies from any member of the family Clupeidae. The antibodies produced in this study are critical for achieving our long-term goal of conducting serological surveillance to assess pathogen exposure in natural populations of Pacific herring.

Inhibition of intracellular transport of B cell antigen receptor complexes by Kaposi's sarcoma-associated herpesvirus K1.

The B cell antigen receptor (BCR) is a large complex that consists of a disulfide-linked tetramer of two transmembrane heavy (mu) chains and two light (lambda or kappa) chains in association with a heterodimer of Igalpha and Igbeta. Kaposi's sarcoma-associated herpesvirus (KSHV) encodes a transforming protein called K1, which has structural and functional similarity to Igalpha and Igbeta. We demonstrate that K1 downregulates the expression of BCR complexes on the surface. The NH(2)-terminal region of K1 specifically interacts with the mu chains of BCR complexes, and this interaction retains BCR complexes in the endoplasmic reticulum, preventing their intracellular transport to the cell surface. Thus, KSHV K1 resembles Igalpha and Igbeta in its ability to induce signaling and to interact with mu chains of the BCR. However, unlike Igalpha and Igbeta, which interact with mu chains to direct BCR complexes to the cell surface, K1 interacts with mu chains to block the intracellular transport of BCR complexes to the cell surface. These results demonstrate a unique feature of the K1 transforming protein, which may confer virus-infected cells with a long-term survival advantage.

Synthesis of abnormal immunoglobulins by hybridomas from autoimmune "viable motheaten" mutant mice.

Secretory defects in abnormal plasma cells, called Mott cells, that appear in lymphoid tissues of spontaneously autoimmune, "viable motheaten" (mev/mev) mice lead to deposition of immunoglobulin in RER-bound vesicles. Such vesicles have been termed Russel bodies. Cells with Russel bodies can also be observed rarely in normal animals, usually as a result of extreme antigenic loads or pathologic states. To understand why these abnormal cells appear commonly in mev/mev mice, we have established a panel of hybridomas that contain Russell bodies. Using immunochemical analysis and immunoelectron microscopy, we have characterized the secretory defects. Although these hybridoma cells synthesize a normal size heavy chain and it associates with light chain, the Russell bodies have many characteristics of inclusion bodies, which commonly appear in cells synthesizing mutant proteins and often are associated with incompletely or abnormally folded proteins. Pulse-chase experiments showed that immunoglobulins synthesized by these hybridomas accumulate rapidly into insoluble complexes and have an intracellular half life approximately 10 time greater than normal immunoglobulins. The defect affected only the immunoglobulin derived from the mev/mev mice and did not affect the secretion of normal immunoglobulin produced by an IgG1-secreting fusion partner. In addition to accumulating intracellular immunoglobulins, many mutant cell lines also secreted immunoglobulin. Endoglycosidase H digestion was used to determine the state of processing of the N-linked carbohydrates on the immunoglobulin molecules. This analysis demonstrated that the N-linked carbohydrates on the secreted immunoglobulin were resistant to endoglycosidase H digestion, indicating that they were processed normally. The insoluble IgM molecules were sensitive to endoglycosidase H, which is consistent with their localization to the RER. We propose several models by which these abnormal immunoglobulin-secreting cells commonly appear in this autoimmune mutant mouse.

Removal of the BiP-retention domain in Cmicro permits surface deposition and developmental progression without L-chain.

Nascent, full length, immunoglobulin (Ig) heavy (H)-chains are post-translationally associated with H-chain-binding protein (BiP or GRP78) in the endoplasmic reticulum (ER). The first constant (C) domain, CH1 of a C gene (Cmu, Cgamma, Calpha), is important for this interaction. The contact is released upon BiP replacement by conventional Ig light (L)-chain (kappa or lambda). Incomplete or mutated H-chains with removed variable (VH) and/or C(H)1 domain, as found in H-chain disease (HCD), can preclude stable BiP interaction. Progression in development after the preB cell stage is dependent on surface expression of IgM when association of a micro H-chain with a L-chain overcomes the retention by BiP. We show that IgM lacking the BiP-binding domain is displayed on the cell surface and elicits a signal that allows developmental progression even without the presence of L-chain. The results are reminiscent of single chain Ig secretion in camelids where developmental processes leading to the generation of fully functional H-chain-only antibodies are not understood. Furthermore, in the mouse the largest secondary lymphoid organ, the spleen, is not required for H-chain-only Ig expression and the CD5 survival signal may be obsolete for cells expressing truncated IgM.

Roles of N-glycans in the polymerization-dependent aggregation of mutant Ig-µ chains in the early secretory pathway.

The polymeric structure of secretory IgM allows efficient antigen binding and complement fixation. The available structural models place the N-glycans bound to asparagines 402 and 563 of Ig-µ chains within a densely packed core of native IgM. These glycans are found in the high mannose state also in secreted IgM, suggesting that polymerization hinders them to Golgi processing enzymes. Their absence alters polymerization. Here we investigate their role following the fate of aggregation-prone mutant µ chains lacking the Cµ1 domain (µ∆). Our data reveal that µ∆ lacking 563 glycans (µ∆5) form larger intracellular aggregates than µ∆ and are not secreted. Like µ∆, they sequester ERGIC-53, a lectin previously shown to promote polymerization. In contrast, µ∆ lacking 402 glycans (µ∆4) remain detergent soluble and accumulate in the ER, as does a double mutant devoid of both (µ∆4-5). These results suggest that the two C-terminal Ig-µ glycans shape the polymerization-dependent aggregation by engaging lectins and acting as spacers in the alignment of individual IgM subunits in native polymers.

5-iodonaphthyl-1-azide labeling of plasma membrane proteins adjacent to specific sites via energy transfer.

We have examined conditions optimal for 5-iodonaphthyl-1-azide (INA4) labeling of membrane proteins proximal to known membrane sites. Membrane-bound INA can be indirectly activated by energy transfer from visible chromophores. We demonstrate that the efficiency of this sensitized activation is enhanced by use of triplet-forming chromophores such as eosin and by deoxygenation. Variation of sensitized activation efficiency with INA concentration indicates that the critical distance for eosin-INA energy transfer in solution is 8-14 A. We suggest that photosensitization occurs through triplet exchange and present an improved labeling protocol based on these findings. This protocol was used to examine whether different accessory proteins are associated with isolated and crosslinked Type I Fc epsilon receptors on 2H3 rat basophilic leukemia cells. 2H3 cells were incubated with eosin-conjugated IgE and irradiated at 514 nm yielding [125I]INA derivatized peptides at 53, 38, 34, and 29 kDa. Crosslinking IgE with mouse anti-rat IgE prior to irradiation labeled three additional proteins at 60, 54, and 43 kDa. These results demonstrate the utility of sensitized INA labeling in characterizing protein-protein interactions in membranes of intact cells and indicate the importance of considering photophysical factors when selecting sensitizers and reaction conditions. We discuss estimation of the size of the membrane region surrounding a sensitizing chromophore within which INA labeling of membrane proteins occurs.

The involvement of immunoglobulin E isotype switch in scleroderma skin tissue.

BACKGROUND: The involvement of mast cell, which is activated by immunoglobulin E (IgE), has been reported in the formation of systemic sclerosis (SSc) abnormality. IgE is generated with isotype switch. During isotype switch, switch circles resulting from direct mu to epsilon, or from sequential mu to gamma via epsilon switching will be created. OBJECTIVE: We studied whether switching occurs in SSc. METHODS: We used nested polymerase chain reaction to analyze the S fragments from switch circles. Fifty-two patients with SSc, and 62 healthy women were studied. RESULTS: Neither of 62 normal skin tissues showed direct switch, nor sequential switch. Neither of seven normal whole blood cells showed direct switch, nor sequential switch. In 52SSc skin tissues, three (5.8%) showed direct switch, and two (3.8%) showed sequential switch. As a result, five (9.6%) of SSc skin tissue showed immunogobulin E class switch. These results were confirmed by DNA sequencing. CONCLUSION: These results demonstrated that isotype switch to the epsilon locus achieved by direct and/or sequential switch are involved in SSc skin.

Membrane immunoglobulin is characterized by distinct structural subpopulations.

Membrane immunoglobulins are integral proteins on B cell surfaces that bind foreign antigens and are critically involved in the regulation of the immune response. Based upon the model of serum IgG, it has been assumed that membrane immunoglobulins are essentially four chain disulfide-linked structures of the form H2L2, where H represents an immunoglobulin heavy chain, and L a light chain. We show here that membrane immunoglobulins of the mu and delta isotypes are present on spleen cell surfaces in a much more diverse group of disulfide linked structures. In some cases mIg is linked into structures as large as H5L5, while in other instances mu or delta chains appear to be linked by disulfide interactions to non-immunoglobulin molecules. These various structural complexes may represent distinct functional entities, as the association of mIg with the cytoskeleton after mIg cross-linking appears to depend upon its structural subgroup.

Biogenesis and function of IgM: the role of the conserved mu-chain tailpiece glycans.

The tailpiece of secretory Ig-mu-chains (mu(s)tp) is highly conserved throughout evolution: in particular, a carboxy-terminal cysteine residue (Cys575) and a glycan linked to Asn563 are found in all species sequenced so far. Here we show that the mu(s)tp oligosaccharide moieties are important for the binding of J-chains and for the process of IgM polymerization. In the absence of the mu(s)tp glycans, pentamers cannot be assembled and polymers containing six or more subunits are secreted. Despite their increased valency, these molecules have a lower association rate with antigen than wild-type polymers. Unexpectedly, the C-terminal oligosaccharides also affect kinetic parameters on unpolymerized subunits. Thus, monomers lacking the C-terminal sugars because of either site-directed mutagenesis or selective enzymatic deglycosylation with endoglycosidase H, have a lower k(on) for the antigen. Taken together, our results indicate that the C-terminal mu-chain glycans can shape the structure of mu(s2)L2 subunits and their further assembly into polymers.

The repetitive portion of the Xenopus IgH Mu switch region mediates orientation-dependent class switch recombination.

Vertebrates developed immunoglobulin heavy chain (IgH) class switch recombination (CSR) to express different IgH constant regions. Most double-strand breaks for Ig CSR occur within the repetitive portion of the switch regions located upstream of each set of constant domain exons for the Igγ, Igα or Igϵ heavy chain. Unlike mammalian switch regions, Xenopus switch regions do not have a high G-density on the non-template DNA strand. In previous studies, when Xenopus Sµ DNA was moved to the genome of mice, it is able to support substantial CSR when it is used to replace the murine Sγ1 region. Here, we tested both the 2kb repetitive portion and the 4.6 kb full-length portions of the Xenopus Sµ in both their natural (forward) orientation relative to the constant domain exons, as well as the opposite (reverse) orientation. Consistent with previous work, we find that the 4.6 kb full-length Sµ mediates similar levels of CSR in both the forward and reverse orientations. Whereas, the forward orientation of the 2kb portion can restore the majority of the CSR level of the 4.6 kb full-length Sµ, the reverse orientation poorly supports R-looping and no CSR. The forward orientation of the 2kb repetitive portion has more GG dinucleotides on the non-template strand than the reverse orientation. The correlation of R-loop formation with CSR efficiency, as demonstrated in the 2kb repetitive fragment of the Xenopus switch region, confirms a role played by R-looping in CSR that appears to be conserved through evolution.

NF-Y, a CCAAT box-binding protein, is one of the trans-acting factors necessary for the response of the murine ERp72 gene to protein traffic.

The accumulation of incompletely assembled immunoglobulin mu heavy chain in transfected COS cells stimulates the cellular response to protein traffic that results in the increased transcription and elevated synthesis of several ER chaperones, including ERP72, a member of the protein disulfide isomerase family of molecular chaperones. The ERp72 promoter contains an 82 bp ER protein traffic response element (ERPTRE) that is sufficient to mediate this response. Previously, it had been shown that the alteration of a putative AP-2 site and a CCAAT and inverted CCAAT site within the ERPTRE significantly decreased the response of ERp72 promoter to mu chain accumulation. We have extended these findings by demonstrating a role for NF-Y and a potentially novel DNA-binding protein in the regulation of transcription from the ERp72 promoter. The fact that NF-Y binding to the ERPTRE is observed in extracts from both control cells and cells in which the response to protein traffic has been activated indicates that the binding of NF-Y, while necessary, is not sufficient to account for the response. Each of the two CCAAT sites in the ERPTRE can bind NF-Y independently, but both sites must be intact for full ERPTRE function. A second protein can bind to the ERPTRE independently of NF-Y and at a site overlapping or close to the 3' end of the reverse CCAAT site. It is possible that interactions between NF-Y, this protein and perhaps other factors are responsible for the regulation of the protein traffic response.

Single-step purification of F(ab')2 mu fragments of mouse monoclonal antibodies (immunoglobulins M) by hydrophobic interaction high-performance liquid chromatography using TSKgel ether-5PW.

A procedure is described for preparation and single-step purification of F(ab')2 fragments, herein designated as F(ab')2 mu' from mouse monoclonal antibodies of the IgM class. Hydrophobic interaction high-performance liquid chromatography (HPLC) using TSKgel Ether-5PW was well applicable to the purification. The IgM was digested with pepsin at the pepsin-to-IgM ratio of 1:200 (w/w) in 100 mM citrate buffer (pH 4.2) at 37 degrees C for 2 h. The digests were applied to the gel equilibrated with the buffer containing 1 M ammonium sulfate. F(ab')2 mu fragments were adsorbed onto the gel with the same buffer, and eluted by reducing the ammonium sulfate concentration to 0 M. The fraction containing F(ab')2 mu fragments was homogeneous (purity higher than 97%) by both sodium dodecyl sulfate-polyacrylamide gel electrophoresis and gel-filtration HPLC. The recovery of the antigen-binding site was 55-72%. The cycle time of the Ether-5PW HPLC was 40 min, and up to 98 mg F(ab')2 mu fragments. The molecular mass of F(ab')2 mu was estimated to be 144-146 kDa. In comparison with IgM, F(ab')2 mu lost entirely the complement C1q binding activity, and the sugar content was greatly reduced. The binding of IgM with non-specific proteins turned to be negligible, when IgM was converted to F(ab')2 mu, suggesting that the fragments are useful for immunological application.

A monoclonal antibody distinguishes between two IgM heavy chain isotypes in Atlantic salmon and brown trout: protein characterization, 3D modeling and epitope mapping.

Atlantic salmon (Salmo salar) and brown trout (Salmo trutta) possess two distinct subpopulations of IgM which can be separated by anion exchange chromatography. Accordingly, there are two isotypic µ genes in these species, related to ancestral tetraploidy. In the present work it was verified by mass spectrometry that IgM of peak 1 (subpopulation 1) have heavy chains previously designated as µB type whereas IgM of peak 2 (subpopulation 2) have heavy chains of µA type. Two adjacent cysteine residues are present near the C-terminal part of µB, in contrast to one cysteine residue in µA. Salmon IgM of both peak 1 and peak 2 contain light chains of the two most common isotypes: IgL1 and IgL3. In contrast to salmon and brown trout, IgM of rainbow trout (Oncorhynchus mykiss) is eluted in a single peak when subjected to anion exchange chromatography. Surprisingly, a monoclonal antibody MAb4C10 against rainbow trout IgM, reacted with µA in salmon, whereas in brown trout it reacted with µB. It is plausible to assume that DNA has been exchanged between the paralogous A and B loci during evolution while maintaining the two sub-variants, with and without the extra cysteine. MAb4C10 was conjugated to magnetic beads and used to separate cells, demonstrating that µ transcripts residing from captured cells were primarily of A type in salmon and B type in brown trout. An analysis of amino acid substitutions in µA and µB of salmon and brown trout indicated that the third constant domain is essential for MAb4C10 binding. This was supported by 3D modeling and was finally verified by studies of MAb4C10 reactivity with a series of recombinant µ3 constructs.

All oligosaccharide moieties of the µ chains in the pre-BCR are of the high-mannose type.

Although it is well established that pre-BCR signaling governs proliferation and differentiation during B cell development, the components of the pre-BCR that are important for signaling are a matter of controversy. It has been suggested that signaling by the µ heavy chains of the pre-BCR induces survival and differentiation of pre-B cells, while the λ5 part of the pre-BCR is essential for proliferation and clonal expansion. However, the mechanism by which pre-BCR µ chains initiate differentiation signals is not clear. Using two variants of a murine B-lymphocyte cell line that differ only in surface expression of either BCR or pre-BCR, we demonstrated that surface µ chains in the pre-BCR are of the high-mannose type only, while those in the BCR are of the complex type. It is hypothesized that mannose-specific lectin-like molecules on accessory cells or in solution may function as the non-antigen ligand that triggers the pre-BCR.

Aptamer directly evolved from live cells recognizes membrane bound immunoglobin heavy mu chain in Burkitt's lymphoma cells.

The identification of tumor related cell membrane protein targets is important in understanding tumor progression, the development of new diagnostic tools, and potentially for identifying new therapeutic targets. Here we present a novel strategy for identifying proteins that are altered in their expression levels in a diseased cell using cell specific aptamers. Using an intact viable B-cell Burkitt's lymphoma cell line (Ramos cells) as the target, we have selected aptamers that recognize cell membrane proteins with high affinity. Among the selected aptamers that showed different recognition patterns with different cell lines of leukemia, the aptamer TD05 showed binding with Ramos cells. By chemically modifying TD05 to covalently cross-link with its target on Ramos cells to capture and to enrich the target receptors using streptavidin coated magnetic beads followed by mass spectrometry, we were able to identify membrane bound immunoglobin heavy mu chain as the target for TD05 aptamer. Immunoglobin heavy mu chain is a major component of the B-cell antigen receptor, which is expressed in Burkitt's lymphoma cells. This study demonstrates that this two step strategy, the development of high quality aptamer probes and then the identification of their target proteins, can be used to discover new disease related potential markers and thus enhance tumor diagnosis and therapy. The aptamer based strategy will enable effective molecular elucidation of disease related biomarkers and other interesting molecules.

Distinguishing between retention signals and degrons acting in ERAD.

Endoplasmic reticulum-associated degradation (ERAD) eliminates aberrant proteins from the secretory pathway. Such proteins are retained in the endoplasmic reticulum and targeted for degradation by the ubiquitin-proteasome system. Cis-acting motifs can function in ERAD as retention signals, preventing vesicular export from the endoplasmic reticulum, or as degrons, targeting proteins for degradation. Here, we show that microstp, the C-terminal 20-residue tailpiece of the secretory IgM mus heavy chain, functions both as a portable retention signal and as an ERAD degron. Retention of microstp fusions of secreted versions of thyroid peroxidase and yellow fluorescent protein in the endoplasmic reticulum requires the presence of the penultimate cysteine of microstp. In its role as a portable degron, the microstp targets the retained proteins for ERAD but does not serve as an obligatory ubiquitin-conjugation site. Abolishing microstp glycosylation accelerates the degradation of both microstpCys-fused substrates, yet absence of the N-glycan eliminates the requirement for the penultimate cysteine in the retention and degradation of the unglycosylated yellow fluorescent protein. Hence, the dual role played by the microstpCys motif as a retention signal and as a degron can be attributed to distinct elements within this sequence.

Mu-heavy chain disease: presentation as a benign monoclonal gammopathy.

mu-Heavy chain disease (HCD) is a rare monoclonal lymphoid disorder characterized by the failure to assemble a complete IgM immunoglobulin. The mu-heavy chains analyzed to date revealed absence of the variable region and a shortened constant domain. We report the first case of mu-HCD presenting as a benign monoclonal gammopathy. The literature on the 27 reported mu-HCD cases is reviewed, and important clinical and laboratory findings are discussed. The ages of the patients ranged from 15 to 80 years (median, 57.5 years). Twenty-two of 27 patients had an associated lymphoplasma cell proliferative disorder. A monoclonal spike on routine serum protein electrophoresis was found in only 8 of 19 patients. Fourteen of 22 patients had Bence Jones proteinuria, but mu-HCD protein was reported in the urine of only two patients. The survival ranged from less than 1 month to 11 years (median, 24 months).

Evolution of vertebrate IgM: complete amino acid sequence of the constant region of Ambystoma mexicanum mu chain deduced from cDNA sequence.

cDNA clones coding for the constant region of the Mexican axolotl (Ambystoma mexicanum) mu heavy immunoglobulin chain were selected from total spleen RNA, using a cDNA polymerase chain reaction technique. The specific 5'-end primer was an oligonucleotide homologous to the JH segment of Xenopus laevis mu chain. One of the clones, JHA/3, corresponded to the complete constant region of the axolotl mu chain, consisting of a 1362-nucleotide sequence coding for a polypeptide of 454 amino acids followed in 3' direction by a 179-nucleotide untranslated region and a polyA+ tail. The axolotl C mu is divided into four typical domains (C mu 1-C mu 4) and can be aligned with the Xenopus C mu with an overall identity of 56% at the nucleotide level. Percent identities were particularly high between C mu 1 (59%) and C mu 4 (71%). The C-terminal 20-amino acid segment which constitutes the secretory part of the mu chain is strongly homologous to the equivalent sequences of chondrichthyans and of other tetrapods, including a conserved N-linked oligosaccharide, the penultimate cysteine and the C-terminal lysine. The four C mu domains of 13 vertebrate species ranging from chondrichthyans to mammals were aligned and compared at the amino acid level. The significant number of mu-specific residues which are conserved into each of the four C mu domains argues for a continuous line of evolution of the vertebrate mu chain. This notion was confirmed by the ability to reconstitute a consistent vertebrate evolution tree based on the phylogenic parsimony analysis of the C mu 4 sequences.