The NF2 tumor suppressor regulates microtubule-based vesicle trafficking via a novel Rac, MLK and p38(SAPK) pathway

Neurofibromatosis Type 2 patients develop schwannomas, meningiomas and ependymomas resulting from mutations in the tumor suppressor gene, NF2, encoding a membrane-cytoskeleton adaptor protein called merlin. Merlin regulates contact inhibition of growth and controls the availability of growth factor receptors at the cell surface. We tested if microtubule-based vesicular trafficking might be a mechanism by which merlin acts. We found that schwannoma cells, containing merlin mutations and constitutive activation of the Rho/Rac family of GTPases, had decreased intracellular vesicular trafficking relative to normal human Schwann cells. In Nf2−/− mouse Schwann (SC4) cells, re-expression of merlin as well as inhibition of Rac or its effector kinases, MLK and p38^SAPK, each increased the velocity of Rab6 positive exocytic vesicles. Conversely, an activated Rac mutant decreased Rab6 vesicle velocity. Vesicle motility assays in isolated squid axoplasm further demonstrated that both mutant merlin and active Rac specifically reduce anterograde microtubule-based transport of vesicles dependent upon the activity of p38^SAPK kinase. Taken together, our data suggest loss of merlin results in the Rac dependent decrease of anterograde trafficking of exocytic vesicles, representing a possible mechanism controlling the concentration of growth factor receptors at the cell surface.

A syngeneic mouse glioma model for study of glioblastoma therapy

Animal models of human tumors serve a vital role in the development and testing of new anticancer therapies. Since the immune system is likely to play an essential role in tumor eradication, there is a particular need for modeling human disease in immunocompetent hosts. Few models of glioma have been developed in immunocompetent mice that are commercially available and none of these tumors have histological and antigenic characteristics of human gliomas. We have used a cell line, 4C8, derived from a spontaneous glioma-like tumor that arose in a transgenic mouse to develop a new glioma model. The intracranial injection of 4C8 cells into immunocompetent syngeneic B6D2F1 mice resulted in tumors that were densely cellular, developed a pseudopallisading pattern of necrosis, and expressed GFAP; all important features of human malignant gliomas. The average neurological endpoint was 51 days after intracranial injection. The 4C8 cells also grew rapidly in the flank, retaining histologic features seen in intracranial tumors. Flank tumors reached an average volume of 100 mm3, a volume ideal for therapy testing, by 34 days postinjection. These results suggest that the 4C8 mouse glioma model is an excellent system in which to test new antiglioma therapies for use in humans.

A simple, reproducible technique for establishing leptomeningeal tumors in nude rats

The incidence of leptomeningeal carcinomatosis is on the rise and current treatment modalities have limited efficacy. The development of new treatment strategies has been hampered by the lack of an appropriate animal model that accurately parallels the clinical condition. We have developed a new surgical technique for the establishment of leptomeningeal tumors in rats. Our technique is simple, reproducible and associated with low morbidity and mortality. Tumor implantation resulted in a defined neurological endpoint and a reproducible time course of disease progression using both human medulloblastoma and breast carcinoma cell lines. This animal model provides an appropriate system for testing conventional and new biologic therapies in both early and late stages of leptomeningeal disease.

Novel protein isoforms of the APC tumor suppressor in neural tissue

The conventional protein isoform of the APC tumor suppressor is 310 kD and is encoded by exons 1 - 15 of the APC gene. Other RNAs are expressed from the APC gene and include one form that contains an exon upstream of exon 1, designated BS, but this transcript does not include exon 1. This transcript recently has been shown to be enriched in non-dividing, terminally-differentiated cells (Santoro and Groden, 1997). To determine if the BS-containing transcript encoded an alternate APC protein isoform, we generated and affinity-purified a polyclonal antibody directed to protein sequence predicted by exon BS. The BS antibody labeled a band of approximately 300 kD on immunoblots of cerebral and cerebellar tissue from adult human, baboon, rat and mouse. These same tissue lysates also contained prominent BS-reactive proteins of 290 kD, 200 kD and 150 kD. Lysates from mitotically active cells did not contain these APC isoforms. To verify that BS-reactive proteins were APC isoforms, BS-immunoprecipitates were blotted and labeled with commercially available APC antibodies. All four high molecular weight BS-antibody-precipitated proteins were recognized by antibodies directed against epitopes encoded by APC exons 2 and 15. BS isoforms were not, however, labeled with antibodies to an epitope encoded by APC exon 1, consistent with the prediction that BS - APC isoforms lack the domain encoded by these sequences. Like conventional APC, at least one of the four BS-APC protein isoforms also interacts with beta-catenin. BS-APC isoforms that lack exon 1-encoded sequences are incapable of dimerization with the conventional form of APC, yet retain the ability to bind beta-catenin. Such isoforms are likely to be functionally distinct from the conventional APC protein.

A novel multiply-mutated HSV-1 strain for the treatment of human brain tumors

A promising approach for the therapeutic treatment of brain tumors utilizes replication-competent, neuroattenuated herpes simplex virus-1 (HSV-1) mutants. This approach requires mutation of HSV-1 to eliminate killing of normal, nondividing cells of the brain (e.g., neurons). We have generated a HSV-1 double-mutant, designated 3616UB, by interrupting the uracil DNA glycosylase (UNG) gene in a previously studied ICP34.5 mutant, R3616. The HSV-1-encoded UNG gene is required for efficient HSV-1 replication in nondividing cells, but is dispensable for replication in rapidly dividing cells. The specific function of the HSV-1 ICP34.5 gene is not completely clear, but it is thought to be necessary for viral replication in cells of the nervous system, because, when mutated, the resultant viral strains are fully neuroattenuated. Strain 3616UB did not replicate in primary neuronal cultures in vitro or in mouse brain, but efficiently killed six of six human tumor cell lines within 6 days in vitro and successfully infected and replicated within brain tumor xenografts. The potential safety of 3616UB for human use is enhanced by an unexpected hypersensitivity to the antiherpetic drug ganciclovir. These data suggest that 3616UB may be effective for the treatment of human brain tumors. Intratumoral injection of 3616UB into human medulloblastoma or angiosarcoma xenografts established in severe combined immunodeficient (SCID) mice produced significant growth arrest and some tumor regressions. Strain 3616UB was as effective as R3616 in this therapy study and did not cause any obvious distress in the treated animals. Together, the data show that 3616UB is a very safe alternative to other HSV-1 mutants because the presence of two mutations reduces the possibility of recombinational events in situ that could lead to the generation of virulent viral progeny during 3616UB therapy.

Peripherin assembles into homopolymers in SW13 cells

The properties of full-length and mutant peripherins were studied in intermediate filament-less SW13 cells to define regions of peripherin that are essential for initiation of filament assembly. A full-length rat peripherin gene transfected into SW13 cells resulted in filament formation, consistent with the close structural relationship of peripherin to other type III intermediate filament proteins that readily form homopolymers. Translation of full-length rat peripherin is initiated predominantly at the second of two inframe AUGs. Deletions within the amino terminus of wild-type peripherin abolished its ability to form filaments in SW13 cells. In contrast, deletion of the entire carboxyl-terminal tail of peripherin did not affect its ability to form filamentous arrays in transfected SW13 cells. These results indicate that, of the intermediate filament proteins that are expressed in mature neurons, only peripherin and alpha-internexin are capable of making homopolymer intermediate filaments. In addition, mutations of the carboxyl tail of peripherin generally do not interfere with filament network formation.

Peripherin gene is linked to keratin 18 gene on human chromosome 12

Peripherin is a neuron-specific intermediate filament (IF) protein, found primarily in phylogenetically old regions of the nervous system. Whereas other neuronal IF genes have only two to three introns and are scattered in the genome, the peripherin gene (PRPH) has a complex intron-exon structure like nonneuronal IF genes that are clustered in tandem arrays, e.g., those encoding the keratins. We used a cosmid containing the human peripherin gene (PRPH) to determine its chromosomal location in relationship to nonneuronal IF genes. Using a rodent-human mapping panel, we localized the PRPH gene to human chromosome 12. Since a cluster of keratin genes maps to 12q12-13, polymorphic markers were developed for PRPH and for one of the keratin genes presumed to be in the cluster, keratin 18 (KRT18). Both markers were typed in CEPH reference families. Pairwise and multipoint analyses of the CEPH data revealed that KRT18 is tightly linked to DNA markers D12S4, D12S22, D12S90, D12S96 and D12S103, which lie between D12S18 and D12S8, with odds greater than 1000:1. These markers are physically located at 12q11-13, thus supporting the fine localization of KRT18 in or near the group of type II keratins in this region. Furthermore, linkage analysis showed that the peripherin gene (PRPH) is tightly linked to KRT18 (Z = 15.73, theta = 0.013), and therefore appears to be in close proximity to the cluster.

Expression of the neural intermediate filament proteins peripherin and neurofilament-66/alpha-internexin in neuroblastoma

BACKGROUND

Peripherin and neurofilament (NF)-66/alpha-internexin are recently characterized, neuron-specific intermediate filament proteins that are expressed in the developing peripheral nervous system. Peripherin, in particular, is highly enriched in neuronal derivatives of the neural crest. We speculated that these intermediate filament proteins would be expressed in neuroblastoma (NB), a neural crest-derived tumor with many neuronal features.

EXPERIMENTAL DESIGN

By use of antibodies specific to peripherin and NF-66/alpha-internexin, we detected these proteins on Western blots of NB tissue extracts and in paraffin sections of NBs.

RESULTS

Western blotting indicated that NB tumor extracts contained immunoreactive proteins that co-migrated with rat peripherin and human NF-66/alpha-internexin from normal tissues, thus establishing the specificity of the antibodies for these proteins in tumors. The antibody specific for peripherin labeled all NBs, including immature NBs, composite ganglioneuroblastomas and ganglioneuromas. In contrast, the NF-66/alpha-internexin antibody labeled only 50% of NBs, and only weakly labeled most ganglioneuroblastomas and ganglioneuromas. Neither antibody labeled other small blue cell tumors such as lymphomas, rhabdomyosarcomas, Wilms' tumors, and Ewing sarcomas.

CONCLUSIONS

The specificity of the peripherin labeling of NB and the ability of the peripherin antibody to label the entire spectrum of NBs, including ganglioneuroblastomas and ganglioneuromas, indicate that this intermediate filament protein has potential as a diagnostic marker for these related neural crest neoplasms.

The structure of the human peripherin gene (PRPH) and identification of potential regulatory elements

We determined the complete nucleotide sequence of the coding region of the human peripherin gene (PRPH), as well as 742 bp 5' to the cap site and 584 bp 3' to the stop codon, and compared its structure and sequence to the rat and mouse genes. The overall structure of 9 exons separated by 8 introns is conserved among these three mammalian species. The nucleotide sequences of the human peripherin gene exons were 90% identical to the rat gene sequences, and the predicted human peripherin protein differed from rat peripherin at only 18 of 475 amino acid residues. Comparison of the 5' flanking regions of the human peripherin gene and rodent genes revealed extensive areas of high homology. Additional conserved segments were found in introns 1 and 2. Within the 5' region, potential regulatory sequences, including a nerve growth factor negative regulatory element, a Hox protein binding site, and a heat shock element, were identified in all peripherin genes. The positional conservation of each element suggests that they may be important in the tissue-specific, developmental-specific, and injury-specific expression of the peripherin gene.

In vitro evidence that metabolic cooperation is responsible for the bystander effect observed with HSV tk retroviral gene therapy

Tumor cells transduced with a retroviral vector expressing a herpes virus thymidine kinase (HSV tk) gene are rendered sensitive to the antiherpetic drug, ganciclovir. The bystander effect refers to the observation that not all cells need be transduced to eradicate the cell population by treatment with ganciclovir. We demonstrate that metabolic cooperation can account for this bystander effect. When HT1080 human fibrosarcoma cells marked with a lacZ gene (LZ+5) were cocultured with HT1080 cells transduced with a retrovirus expressing HSVtk (HT1080tk11), at a density at which the majority of cells were in contact, both HT1080tk11 and LZ+5 cells were killed by ganciclovir. When cells were cocultured at a low density where the majority of cells are not in contact with one another, however, only the HT1080tk11 cells were killed. This result suggests that cell contact with HT1080tk11 cells is necessary to render the HSVtk- LZ+5 cells sensitive to ganciclovir. Because involvement of metabolic cooperation in the killing of the LZ+5 cells would require not only contact between HT1080tk11 and LZ+5 cells but also the capacity to transfer small cytotoxic molecules from the former cell to the latter, transfer of radioactive molecules between the two cell lines was assessed by autoradiography after treatment of a coculture with [3H]ganciclovir. Isolated HT1080tk11 cells incorporated the labeled ganciclovir into their nuclei, whereas isolated LZ+5 cells did not. LZ+5 cells incorporated [3H]ganciclovir, only when in contact with HT1080tk11 cells. These findings indicate that a ganciclovir metabolic product, presumably a phosphorylated form, can pass from HSV tk+ to HSV tk- cells and mediate cytotoxicity as a consequence of direct contact.

Intragenic sequences are required for cell type-specific and injury-induced expression of the rat peripherin gene

Peripherin is a 57 kDa type III intermediate-filament protein that is thought to play a role in axonogenesis both during development and following nerve injury (Oblinger et al., 1989; Escurat et al., 1990; Gorham et al., 1990; Troy et al., 1990b). We have used transgenic mouse technology to define peripherin gene sequences that are necessary for cell type-specific expression and for the increase in peripherin that occurs in response to axonal injury. Correct temporal and nervous system-specific expression resulted when 5.8 kilobases of peripherin 5' flanking sequence were linked to a reporter gene, but precise cell type-specific expression was achieved only when intragenic sequences were included. When intragenic sequences were present, peripherin transgenes were expressed in dorsal root ganglion neurons and spinal cord motor neurons and were upregulated in these cells following nerve injury.

S-cells from a highly N-myc-amplified neuroblastoma are tumorigenic in nude mice

Some neuroblastoma tumors when cultured in vitro give rise to N (neuronal) and S (non-neuronal) cells, which differ in morphology, state of differentiation, and tumorigenicity. Previously, tumor-forming potential was shown to be characteristic of N cells but not S cells. We examined cultures of N and S cells derived from a well-characterized, N-myc-amplified human neuroblastoma cell line, NBL-W, to determine whether these N and S cells also show differential tumorigenicity in nude mice. N cells formed tumors in 94% of trials and S cells formed tumors in 56% of trials. Although S cell tumors had a longer lag phase prior to tumor development, when tumors developed, both N and S cell-derived tumors grew rapidly. These results suggest that S cells do not always represent a benign component of neuroblastomas.

Differential expression of N-myc in phenotypically distinct subclones of a human neuroblastoma cell line

Neuroblastomas are malignant childhood neoplasms that arise from derivatives of the neural crest. We report the characterization of a new neuroblastoma cell line, designated NBL-W, derived from the primary tumor of a patient with stage IVS disease (S. L. Cohn, C. V. Herst, H. S. Maurer, and S. T. Rosen, J. Clin. Oncol., 5: 1441-1444, 1987) according to the criteria of Evans [A. E. Evans, G. J. D'Angio, and J. Randolf, Cancer (Phila.), 27: 374-378, 1971]. Neurite-bearing (N) and substrate-adherent (S) cell lines have been subcloned from the parent line. N and S cells can interconvert, and both cell types label with the neural crest cell surface marker antibody, HNK-1. Cells in the subcloned lines and in the parent line have been shown by Southern blot analysis to contain approximately 100 copies of the N-myc gene. Cytogenetic analysis shows a homogeneously staining region present on chromosome 19. Although these subclones are of identical genotype, the S cells express lower amounts of N-myc mRNA and protein as compared to the N cells. N cells express several neuronal proteins including the neurotransmitter-processing enzymes tyrosine hydroxylase and dopamine beta-hydroxylase, the neuronal intermediate filament proteins peripherin and NF66/alpha-internexin, and the neural cell adhesion molecule. S cells generally lack neuronal markers but express the mesenchymal intermediate filament protein vimentin, and a small subset of the S cells express glial fibrillary acidic protein. Some S cells were labeled weakly with neural cell adhesion molecule antibody; others were negative. S cells did not express the glial marker S-100 or a melanocyte marker, tyrosinase. Thus, S cells express the neural crest marker HNK-1 but do not express a set of antigens characteristic of any known cell type derived from the neural crest. These results are consistent with the suggestion that differential N-myc expression may be involved in the interconversion of N and S cells but indicate that the S cell phenotype need not represent a highly differentiated neural crest derivative.

Some neural intermediate filaments contain both peripherin and the neurofilament proteins

Mammalian neurons and neuron-like cultured cells express the neural intermediate filament (IF) proteins neurofilament (NF)-L, NF-M, NF-H, and peripherin. To determine whether these proteins are found within the same 10-nm filament, light and electron microscope immunocytochemistry using peripherin and NF-specific antibodies was performed on PC12 cells, nervous tissue, and isolated neural filaments from the cauda equina. Double-label immunofluorescence showed that peripherin and NF-L, -M, and -H were found in identical filamentous patterns in interphase and mitotic PC12 cells. Furthermore, expression of mutant peripherin in PC12 cells disrupted not only the peripherin network but also NF-containing filaments. Immunoelectron microscopy of PC12 cell cytoskeletons showed that peripherin and NF subunit proteins were found in the same filament. In situ, in the sciatic nerve, peripherin/NF-L or peripherin/NF-M/-H double-label immunofluorescence illustrates at least three types of nerve fibers: those containing NF only, those labeled predominantly for peripherin, and fibers in which peripherin and NF subunits were colocalized. Immunoelectron microscopy of filaments isolated from nerve roots comprising the sciatic nerve also showed the same three labeling patterns seen by light microscopy. Some neural IF appear to contain predominantly NF proteins or peripherin, but in others, both proteins are found within the same IF.

Complementary immunohistochemical distribution of the neurofilament triplet and novel intermediate filament proteins in the autonomic and sensory nervous system of the guinea-pig

We have previously established that immunoreactivity for the triplet of polypeptides that comprise the class IV intermediate filament proteins (NFP-triplet) is localized in specific subpopulations of neurons in guinea-pig sensory and autonomic ganglia. Antibodies to novel neurofilament proteins, including a polyclonal antibody to a 57 kDa neuronal intermediate filament polypeptide (NIF57kD) and a monoclonal antibody (CH1) to a 150 kDa intermediate filament, or associated, protein were used in combination with antibodies to the NFP-triplet for double-labelling immunohistochemistry. The results show that different subpopulations of neurons in the guinea-pig dorsal root ganglia, coeliac ganglion and enteric ganglia can be distinguished by their complementary immunoreactivity for these proteins. In dorsal root ganglia, larger neurons are intensely immunoreactive for the NFP-triplet while immunoreactivity with CH1 and NIF57kD antibodies is restricted to the small to medium-sized neurons. In the coeliac ganglion, two regionally defined subpopulations of neurons can be distinguished by their immunoreactivity for either the NFP-triplet or NIF57kD, whereas CH1 labels all neurons with equal intensity. Three classes of morphologically distinct myenteric neuron subpopulations are also distinguished by their immunoreactivity for either the NFP-triplet, NIF57kD or CH1 antibodies. Two classes of submucous neurons are labelled both with CH1 and NIF57kD antibodies but show faint or no immunoreactivity for the NFP-triplet. It is concluded that intermediate filament protein immunoreactivity marks different subpopulations of neurons, which suggests that these proteins may have specific roles in neuronal function.

Axotomy-induced changes in the expression of a type III neuronal intermediate filament gene

The effect of axotomy on the expression of the 57 kDa neuronal intermediate filament (IF) protein in adult rat dorsal root ganglion (DRG) neurons was examined. This IF protein is known to have an exclusively neuronal localization but is considerably more limited in its distribution in the nervous system than the neurofilament (NF) triplet proteins. The 57 kDa neuronal IF protein is similar (and perhaps identical) to the protein "peripherin" and is known to be the product of a Type III IF gene. Since the down-regulated expression of NF proteins (products of type IV IF genes) has been well established, it was of interest to determine whether the novel 57 kDa IF protein was regulated in a similar or different manner from that of the NFs in axotomized neurons. In vitro pulse-labeling of DRGs with 35S-methionine: cysteine followed by 2-dimensional gel electrophoresis/fluorography revealed that the synthesis of the 57 kDa neuronal IF protein was increased 2 weeks after sciatic nerve crush. Immunocytochemical studies using a polyclonal antibody to the 57 kDa neuronal IF protein showed that the immunodetectable levels of this protein increased in DRG neurons after peripheral axotomy. In the normal DRG, staining was localized almost exclusively to small-sized neurons. At 2 weeks after axotomy, however, large- and medium-sized neurons also became immunoreactive; in addition, the overall level of staining in the DRG was greater than normal. Quantitative analysis of in situ hybridizations of DRG neurons with a 35S-labeled cDNA probe specific for the 57 kDa neuronal IF protein revealed a significant increase in the level of 57 kDa IF mRNA in the large-sized (greater than 1000 microns2) neurons 2 weeks after axotomy; the level of 57 kDa IF mRNA in the small neurons was not different from normal at that time. Finally, using a newly developed paradigm for examining the composition of regenerating axons by axonal transport, we determined that significant amounts of the 57 kDa neuronal IF protein were conveyed into the regrowing axonal sprouts of DRG neurons. When DRG neurons were conditioned by a previous axotomy (a crush axotomy of the distal sciatic nerve 2 weeks earlier) and then stimulated to regenerate axons by a second crush axotomy located very close to the DRG, the regenerating sprouts incorporated and conveyed significantly more 57 kDa IF protein by slow axonal transport than did those elaborated by unprimed DRG neurons.(ABSTRACT TRUNCATED AT 400 WORDS)

Selective distribution of the 57 kDa neural intermediate filament protein in the rat CNS

In order to determine the CNS distribution of the 57 kDa neural intermediate filament protein (NIFP), a specific antiserum was used in immunofluorescence studies on serial sections taken from each spinal cord level and at 300-microns intervals through the rat brain. The labeling pattern was recorded onto camera lucida tracings of adjacent sections stained with hematoxylin and eosin/luxol fast blue. Three major immunolocalization patterns were revealed. (1) Both large-caliber and fine-caliber axons of optic and all brain stem cranial nerves and their tracts except for the auditory portion of the VIIIth nerve. (2) An extensive array of fine-caliber fibers in the cerebellar white matter and brain stem with region-specific variation in pattern and density. Prominent among the regions with dense arrays of labeled processes were selective cerebellar afferent systems, particularly olivocerebellar fibers, visual afferents arising in the retina, and selective regions of reticular formation. In contrast to the brain stem, the telencephalon contained rare label. (3) Filamentous labeling of neuronal cytokarya in sensory ganglia and a small number of CNS nuclear groups, including all autonomic nuclei and the cholinergic pontine nuclei. On sections of selected CNS regions, distribution patterns of the 57 kDa NIFP were correlated to results obtained by in situ hybridization of a cDNA probe for the 57 kDa NIFP. The data suggest the possible existence of neuronal IF proteins specifically related to selective anatomic and/or neurotransmitter systems.

A type III intermediate filament gene is expressed in mature neurons

A cDNA (199E) specific for the 57 kd neural IF protein has been isolated from a PC12 cell lambda gt11 library. Antibody eluted from the fusion protein produced by 199E recognizes the 57 kd protein on immunoblots and, in PC12 cells, labels a pattern of fibrillar structures identical to that seen with 57 kd antiserum. In situ hybridization using antisense RNA transcripts labels areas of the nervous system known to contain the 57 kd protein. 199E hybridizes with a single mRNA species of approximately 2.0 kb from PC12 cells. A 199E-reactive message can be detected as early as E10 in rat embryos. Southern analyses suggest that there is only one gene for this protein. Amino acid sequence predicted from 199E indicates that the 57 kd protein is a type III IF protein like vimentin and desmin. Thus, expression of IF structural genes in neurons is not limited to the type IV neuronal IF triplet proteins.

Distribution of a novel 57 kDa intermediate filament (IF) protein in the nervous system

A 57 kDa protein, that is not vimentin, is the major component of intermediate filaments (IF) obtained after 2 cycles of in vitro assembly from PC12 cells (Parysek and Goldman, 1987). By use of an antiserum to the 57 kDa protein, a cross-reacting antigen (of identical molecular weight) was detected on immunoblots of IF preparations and by immunofluorescence of various rat tissues. Immunolocalization studies on 3-4 micron frozen sections of tongue, small intestine, and adrenal gland showed bright labeling of nerve bundles and fine-caliber nerve fibers. The chromaffin cells and ganglion cells of the adrenal medulla also were labeled. In the nervous system, intense labeling was seen in small-caliber nerve fibers in sciatic nerve and spinal cord dorsal roots, as well as in the dorsal white columns and dorsal root ganglia. Of the ganglion cells, preferential labeling was seen in small-sized ganglion cells, whereas a monoclonal antibody to the 150 and 200 kDa neurofilament triplet (NFT) components labeled the large-sized ganglion cells. In the areas of the brain thus far examined with 57 kDa antiserum, there was labeling of components of cranial nerves and labeling of thin fibers in several areas, including the granular layer of the cerebellum and the corticospinal tract in the brain stem. For each tissue, adjacent sections treated with vimentin or glial fibrillary acidic protein antibody revealed labeling patterns distinct from that seen with either the 57 kDa or NFT antibodies. These results indicate that the 57 kDa IF protein is a neural IF component. Furthermore, this protein is distributed in only certain neuronal elements; these elements may be unified by an as yet unrecognized pattern of function in the nervous system.

Proteolytic fragmentation of Dictyostelium myosin and localization of the in vivo heavy chain phosphorylation site

Dictyostelium myosin was associated into dimers and small oligomers at very low ionic strength, filamentous at intermediate ionic strength, and monomeric in solution conditions of high ionic strength. These different associations were probed by fragmenting myosin with chymotrypsin, trypsin, or V-8 protease. All three proteases digested monomeric myosin giving rise to multiple fragments with a wide range of molecular weights. Filamentous myosin was not digested by the V-8 protease, was preferentially cleaved at a single site in the middle of the heavy chain by chymotrypsin, and was cleaved at several sites by trypsin. If the reaction was carried out in very low ionic strength, however, two of these proteases generated stable fragments of high molecular weight. Electron microscopic analysis of these stable fragments showed that tails were shorter than in intact myosin, indicating that the cleavage sites were in the rod portion of the molecule. Under the same conditions of enzymatic digestion, myosin that had been radio labeled in vivo with 32P was analyzed by SDS-PAGE and autoradiography. By comparing the state of phosphorylation and the size of the stable fragments, it was determined that the heavy chain phosphorylation site was located between 55 and 70 kD from the tip of the myosin tail, near a region where the tail displayed sharp bends.

Effect of heavy chain phosphorylation on the polymerization and structure of Dictyostelium myosin filaments

In Dictyostelium amebas, myosin appears to be organized into filaments that relocalize during cell division and in response to stimulation by cAMP. To better understand the regulation of myosin assembly, we have studied the polymerization properties of purified Dictyostelium myosin. In 150 mM KCl, the myosin remained in the supernate following centrifugation at 100,000 g. Rotary shadowing showed that this soluble myosin was monomeric and that approximately 80% of the molecules had a single bend 98 nm from the head-tail junction. In very low concentrations of KCl (less than 10 mM) the Dictyostelium myosin was also soluble at 100,000 g. But rather than being monomeric, most of the molecules were associated into dimers or tetramers. At pH 7.5 in 50 mM KCl, dephosphorylated myosin polymerized into filaments whereas myosin phosphorylated to a level of 0.85 mol Pi/mol heavy chain failed to form filaments. The phosphorylated myosin could be induced to form filaments by lowering the pH or by increasing the magnesium concentration to 10 mM. The resulting filaments were bipolar, had blunt ends, and had a uniform length of approximately 0.43 micron. In contrast, filaments formed from fully dephosphorylated myosin were longer, had tapered ends, and aggregated to form very long, threadlike structures. The Dictyostelium myosin had a very low critical concentration for assembly of approximately 5 micrograms/ml, and this value did not appear to be affected by the level of heavy chain phosphorylation. The concentration of polymer at equilibrium, however, was significantly reduced, indicating that heavy chain phosphorylation inhibited the affinity of subunits for each other. Detailed assembly curves revealed that small changes in the concentration of KCl, magnesium, ATP, or H+ strongly influenced the degree of assembly. Thus, changes in both the intracellular milieu and the level of heavy chain phosphorylation may control the location and state of assembly of myosin in response to physiological stimuli.

Characterization of intermediate filaments in PC12 cells

A 57 kDa protein is the major polypeptide in intermediate filament (IF)-enriched cytoskeletal preparations obtained from the neuronal cell line PC12 (rat pheochromocytoma). Under the conditions used to assemble IF in vitro from other cultured cell lines, 10 nm filaments are formed after 2 cycles of disassembly-assembly from PC12 IF-enriched cytoskeletal preparations; the 57 kDa protein is the major component of the final IF pellet. The 57 kDa protein is immunologically related to the BHK-21 fibroblast 55 kDa protein (vimentin), but a comparison of the peptide maps of PC12 57 kDa and BHK 55 kDa indicates that they are different proteins. With the use of a polyclonal antiserum to the PC12 57 kDa protein, immunofluorescence observations of PC12 cells not treated with NGF reveal a juxtanuclear "knot"-like structure. After NGF treatment, the "knots" are less prominent and many IF arrays are seen coursing through the cytoplasm and extending into the neurites. These immunofluorescence observations of the distribution of IF are corroborated by fine-structural analyses. SDS-PAGE analyses indicate that IF-enriched cytoskeletons isolated from NGF-treated cells have a polypeptide composition similar to that of untreated cells, that is, the 57 kDa protein remains the major polypeptide. SDS-PAGE and immunoblotting analyses show that untreated and NGF-treated PC12 cells also contain relatively minor amounts of the 68, 150, and 200 kDa neurofilament triplet (NFT) proteins. Under immunofluorescence, only 5% of untreated PC12 cells are found to contain a juxtanuclear "knot" labeled with NFT antibodies, but with time following NGF treatment, the number of fluorescent cells increases. After about 2 weeks of NGF treatment, all of the PC12 cells appear to contain NFT antibody-positive filamentous structures. As assessed by immunofluorescence, the NFT polypeptides appear to codistribute with the 57 kDa protein in both untreated and NGF-treated PC12 cells. These data indicate that PC12 cells contain IF composed of a complex set of polypeptides, including a previously unidentified 57 kDa IF protein. While NGF may induce production of NFT polypeptides, there does not appear to be a "switch" from known mesenchymal IF polypeptide expression to NFT polypeptide expression upon stimulation of PC12 cells with NGF.

Microtubule-associated protein 4 antibody: a new marker for astroglia and oligodendroglia

An antibody to a 240,000 dalton microtubule-associated protein, microtubule-associated protein 4, was used to illustrate the distribution of this protein in semi-thin sections of the central nervous system. Immunofluorescence microscopy indicated that microtubule-associated protein 4 was restricted to non-neuronal elements of the brain and spinal cord. Astrocytes, oligodendrocytes and "specialized" glia, including tanycytes, Bergmann glia and Muller cells, contained microtubule-associated protein 4. This distribution of microtubule-associated protein 4 in neural tissue in distinct from that described for the other major brain microtubule-associated proteins, microtubule-associated protein 1 and microtubule-associated protein 2. The reactivity of MAP 4 antibody with these glia demonstrates the antigenic relatedness of these cells and further distinguishes glia from other elements of the nervous system.

MAP 4: a microtubule-associated protein specific for a subset of tissue microtubules

The cytological distribution of microtubule-associated protein 4 (MAP 4) (L. M. Parysek, C. F. Asnes, J. B. Olmsted, 1984, J. Cell Biol., 99:1309-1315) in mouse tissues has been examined. Adjacent 0.5-0.9-micron sections of polyethylene glycol-embedded tissues were incubated with affinity-purified MAP 4 or tubulin antibodies, and the immunofluorescent images were compared. Tubulin antibody labeling showed distinct microtubules in all tissues examined. MAP 4 antibody also labeled microtubule-like patterns, but the extent of MAP 4 reactivity was cell type-specific within each tissue. MAP 4 antibody labeled microtubules in vascular elements of all tissues and in other cells considered to have supportive functions, including Sertoli cells in the testis and glial elements in the nervous system. Microtubule patterns were also observed in cardiac, smooth, and skeletal (eye) muscle, podocytes in kidney, Kuppfer cells in liver, and spermatid manchettes. The only MAP 4-positive cells in which the pattern was not microtubule-like were the principal cells of the collecting ducts in kidney cortex, in which diffuse fluorescence was seen. MAP 4 antibody did not react with microtubule-rich neuronal elements of the central and peripheral nervous system, skeletal muscle from anterior thigh, liver parenchymal cells, columnar epithelial cells of the small intestine, and absorptive cells of the tubular component of the nephron. These observations indicate that MAP 4 may be associated with only certain kinds of cell functions as demonstrated by the preferential distribution with microtubules of defined cell types.

MAP 4: occurrence in mouse tissues

A polyclonal antiserum to a microtubule-associated protein (MAP) from mouse neuroblastoma cells (MAP 4) was used to examine the distribution of this protein in mouse tissues. Immunoblots of neuroblastoma cell microtubule protein preparations demonstrated that the antiserum reacted with a triplet of proteins at 215,000-240,000 mol wt. Antibodies affinity purified from any of the bands showed cross-reaction with the other bands, indicating these polypeptides were all immunologically related. Antibodies specific to MAP 4 decorated microtubules in cultured murine cells fixed with glutaraldehyde, and diffuse staining was seen following treatment of cells with nocodazole. The antiserum reacted with MAP 4 in extracts of brain, heart, liver, and lung from adult mouse; the triplet in brain was more closely spaced than in the other tissues or neuroblastoma cells. In kidney, spleen, and stomach, only a single band (band 4) was labeled; this band was immunologically related to the triplet and was also present in all tissues positive for the triplet. Skeletal muscle, sperm, and peripheral blood contained no reactive polypeptides. After taxol-induced polymerization, the MAP 4 triplet was preferentially associated with the microtubule pellet whereas band 4 remained in the supernatant. These data indicate that there is tissue specificity in the distribution of MAP 4, and that some tissues contain a polypeptide related to MAP 4 (band 4) that does not bind to microtubules in vitro.

Vimentin filaments in spreading, randomly locomoting, and f-met-leu-phe-treated neutrophils

Human neutrophils contain intermediate filaments of the vimentin type. A cytoskeletal preparation, produced by high-salt and Triton X-100 extraction of human neutrophils, reveals a major band at 57000 Mr that comigrates with 3T3 cell vimentin on one-dimensional gels. Two-dimensional gel electrophoresis of whole neutrophils illustrates the presence of vimentin but not desmin- or keratin-filament subunits. The presence of vimentin in neutrophils is also shown by its specific staining with avian vimentin antiserum by two-dimensional gel immunoautoradiography. Indirect immunofluorescence studies show that vimentin antiserum labels an area on one side of the nucleus in spreading neutrophils. This bright area appears as a loose knot of vimentin filaments; a few filaments may radiate from the knot. In contrast to spreading neutrophils, those undergoing random locomotion contain a fine network of filaments that are located in the cytoplasm between the nucleus and the trailing end of the cell. Similarly, in chemoattractant-treated neutrophils, vimentin filaments are bundled in the uropod. Transmission electron microscopy of human neutrophil monolayers confirms the intracellular distribution of intermediate filaments as shown by immunofluorescence in spreading and randomly locomoting cells.

Assembly of keratin onto PtK1 cytoskeletons: evidence for an intermediate filament organizing center

Purified keratin, solubilized in 8 M of urea, was added to Triton X-100-extracted PtK1 cells in 5 mM PIPES buffer. The buffer conditions induced assembly of keratin filaments which appear to associate with nuclei of extracted cells. These keratin fibers extend beyond the original margin of the cells and frequently form bridges between adjacent cells. Electron microscopy shows that keratin filaments associate closely with the surface of the nucleus. We suggest that the site of association between keratin and the nucleus may represent an intermediate filament organizing center.