Premutation allele pool in myotonic dystrophy type 2.

BACKGROUND: The myotonic dystrophies (DM1, DM2) are the most common adult muscle diseases and are characterized by multisystem involvement. DM1 has been described in diverse populations, whereas DM2 seems to occur primarily in European Caucasians. Both are caused by the expression of expanded microsatellite repeats. In DM1, there is a reservoir of premutation alleles; however, there have been no reported premutation alleles for DM2. The (CCTG)(DM2) expansion is part of a complex polymorphic repeat tract of the form (TG)(n)(TCTG)(n)(CCTG)(n)(NCTG)(n)(CCTG)(n). Expansions are as large as 40 kb, with the expanded (CCTG)(n) motif uninterrupted. Reported normal alleles have up to (CCTG)(26) with one or more interruptions. METHODS: To identify and characterize potential DM2 premutation alleles, we cloned and sequenced 43 alleles from 23 individuals. Uninterrupted alleles were identified, and their instability was confirmed by small-pool PCR. We determined the genotype of a nearby single nucleotide polymorphism (rs1871922) known to be in linkage disequilibrium with the DM2 mutation. RESULTS: We identified three classes of large non-DM2 repeat alleles: 1) up to (CCTG)(24) with two interruptions, 2) up to (CCTG)(32) with up to four interruptions, and 3) uninterrupted (CCTG)(22-33). Large non-DM2 alleles were more common in African Americans than in European Caucasians. Uninterrupted alleles were significantly more unstable than interrupted alleles (p = 10(-4) to 10(-7)). Genotypes at rs1871922 were consistent with the hypothesis that all large alleles occur on the same haplotype as the DM2 expansion. CONCLUSIONS: We conclude that unstable uninterrupted (CCTG)(22-33) alleles represent a premutation allele pool for DM2 full mutations.

Characterization of human retinal fascin gene (FSCN2) at 17q25: close physical linkage of fascin and cytoplasmic actin genes.

Retinal fascin is a newly identified photoreceptor-specific paralog of the actin-bundling protein fascin. Fascins crosslink f-actin into highly ordered bundles within dynamic cell extensions such as neuronal growth cone filopodia. We have isolated cDNA and genomic clones of human retinal fascin and characterized the structure of the human retinal fascin gene (FSCN2). The cDNA predicts a protein of 492 amino acids and molecular mass 55,057 that shows 94% identity to bovine retinal fascin and 56% identity to human fascin. Promoter analysis reveals a consensus retinoic acid response element and several potential binding sites for transcription factors Crx and Nrl, which correlates with the retina-specific expression of FSCN2 mRNA. Fluorescence in situ hybridization analysis and genomic clone sequencing indicate that the FSCN2 gene lies within 200 kb of the actin gene ACTG1 at 17q25. Database searches revealed that the human fascin gene FSCN1 and actin gene ACTB at 7p22 also coexist within a 200-kb genomic clone. The close physical linkage of these fascin/actin gene pairs suggests that they derive from a common gene duplication event and allows comparison of fascin and actin phylogenetic analyses. Finally, a possible link to the retinitis pigmentosa 17 allele (RP17) at distal 17q was excluded by demonstration of multiple independent segregation events in two RP17 kindreds. Informative FSCN2 polymorphisms were identified and will serve as useful markers in future linkage studies. The likely function of retinal fascin, in light of known fascin roles in other cell types, is to assemble actin microfilaments in support of photoreceptor disk morphogenesis.

Identification and characterization of a novel gene (C4orf5) located on human chromosome 4q with specific expression in cardiac and skeletal muscle.

The loci of several genes responsible for arrhythmogenic right ventricular dysplasia (ARVD) have been mapped. Since ARVD involves the right ventricle, we sought candidate genes preferentially expressed in the right ventricle utilizing differential display polymerase chain reaction (PCR) on mRNA from the chambers of an adult human heart. PCR products were cloned, sequenced, and used to screen an adult heart cDNA library. A novel 1.3-kb cDNA (HGMW-approved symbol C4orf5) with an open reading frame of 795 bp was identified. A probe designed from the 3' untranslated region of the 1.3-kb cDNA was hybridized to the 1.3-kb transcript and an alternatively spliced 2.5-kb transcript in the heart and skeletal muscle RNA lanes on a multitissue Northern blot. Analysis of a 39-kb partial genomic sequence identified three intronic splice sites in the 1.3-kb transcript. The gene was mapped to human chromosome 4q26-q27. Computer-based analysis indicated that this gene is novel with no known function.

hRAD17, a structural homolog of the Schizosaccharomyces pombe RAD17 cell cycle checkpoint gene, stimulates p53 accumulation.

The RAD17 gene product of S. Pombe is an essential component of the checkpoint control pathway which responds to both DNA damage and disruption of replication. We have identified a human cDNA that encodes a polypeptide which is structurally conserved with the S. Pombe Rad17 protein. The human gene, designated hRAD17, predicts an encoded protein of 590 amino acids and a molecular weight of 69 kD. Amino acid sequence alignment revealed that hRadl7 has 28.3% and 52.5% similarity with the S. Pombe Rad17 protein, and 21.8% identity and 45.8% similarity to the budding yeast cell cycle checkpoint protein, Rad 24. When introduced into the S. Pombe rad17 mutant, hRAD17 was able to partially revert its hydroxyurea and ionizing radiation hypersensitivity, but not its UV hypersensitivity. Permanent overexpression of the hRAD17 gene in human fibrosarcoma cells resulted in p53 activation and a significant reduction of S- and G2/M-phase cells accompanied by an accumulation of the G1-phase population, suggesting that hRAD17 may have a role in cell cycle checkpoint control. Immunostaining of HT-1080 cells transiently transfected with a hRAD17 construct confirmed the nuclear accumulation of p53, which mimics the induction caused by DNA damage. Using FISH analysis, we have mapped the hRAD17 locus to human chromosome 5q11.2.

Human hydroxysteroid sulfotransferase SULT2B1: two enzymes encoded by a single chromosome 19 gene.

We have cloned and characterized cDNAs that encode two human hydroxysteroid sulfotransferase (SULT) enzymes, SULT2B1a and SULT2B1b, as well as the single gene that encodes both of these enzymes. The two cDNAs differed at their 5'-termini and had 1050- and 1095-bp open reading frames that encoded 350 and 365 amino acids, respectively. The amino acid sequences encoded by these cDNAs included "signature sequences" that are conserved in all known cytosolic SULTs. Both cDNAs appeared, on the basis of amino acid sequence analysis, to be members of the hydroxysteroid SULT "family, " SULT2, but they were only 48% identical in amino acid sequence with the single known member of that family in humans, SULT2A1 (also referred to as DHEA ST). Northern blot analysis demonstrated the presence of SULT2B1 mRNA species approximately 1.4 kb in length in human placenta, prostate, and trachea and-faintly-in small intestine and lung. Expression of the two human SULT2B1 cDNAs in COS-1 cells showed that both of the encoded proteins catalyzed sulfation of the prototypic hydroxysteroid SULT substrate, dehydroepiandrosterone, but both failed to catalyze the sulfate conjugation of 4-nitrophenol or 17beta-estradiol, prototypic substrates for the phenol and estrogen SULT subfamilies. Both of these cDNAs were encoded by a single gene, SULT2B1. The locations of most exon-intron splice junctions in SULT2B1 were identical to those of the only other known human hydroxysteroid SULT gene SULT2A1 (previously STD). The divergence in 5'-terminal sequences of the two SULT2B1 cDNAs resulted from alternative transcription initiation prior to different 5' exons, combined with alternative splicing. SULT2B1 mapped to human chromosome band 19q13.3, approximately 500 kb telomeric to the location of SULT2A1.

RAD1, a human structural homolog of the Schizosaccharomyces pombe RAD1 cell cycle checkpoint gene.

Cell cycle checkpoints are gating mechanisms that govern cell cycle progression in the presence of DNA damage and incomplete DNA replication. The Schizosaccharomyces pombe Rad1 protein is an essential component of cell cycle checkpoints activated by both types of genomic stress. In this study, we report the isolation of a human homolog of the S. pombe RAD1 gene. The hRAD1 protein is also similar to the Saccharomyces cerevisiae cell cycle checkpoint protein Rad17 and the Ustilago maydis 3' --> 5' exonuclease, Rec1. We show that human RAD1 partially complements the hydroxyurea and ionizing radiation hypersensitivities of a S. pombe rad1 mutant, suggesting phylogenetic conservation of the DNA damage and replication checkpoints. The human RAD1 locus was mapped to human chromosome 5p13.2, a locus frequently altered in non-small-cell lung cancer and bladder cancer.

Philadelphia chromosome-positive myeloid cells in the peripheral blood of chronic myelogenous leukemia patients: comparison with the frequency detected in cycling cells of the bone marrow.

Monitoring the frequency of the Philadelphia (Ph) chromosome in chronic myelogenous leukemia (CML) is important in determining the effectiveness of treatment for patients during therapy. This can be done with high resolution by subjecting short-term bone marrow cultures (48 h) to 24 h of mitotic arrest before harvest and detecting Ph-positive (Ph+) metaphases by fluorescence in situ hybridization (FISH) in a procedure termed hypermetaphase FISH or HMF. Here, we compared procedures for detecting Ph+ interphase cells (interphase FISH or I-FISH) in peripheral blood polymorphonucleocytes (PMNs) with HMF results on the bone marrow of the same 26 CML patients in different stages of remission. The probes for I-FISH in these experiments were relatively large (200-300 kb) and sufficiently resolved in PMNs so that 97.5% of the cells were scorable. The correlation between the frequencies of Ph+ cells from the two different cell sources was excellent (r = 0.983, P < 0.0001); however, there was a consistently higher level of Ph+ cells observed in the cycling marrow cells than in the peripheral blood PMNs. This was discussed in terms of current theories of apoptosis in CML cells. The large number of PMNs analyzable by I-FISH (approximately 500/patient in this study) provided sufficiently narrow 99% confidence intervals to suggest the procedure as an effective and efficient method for monitoring the frequency of Ph+ cells in CML patients undergoing therapy. However, for detection and quantification of minimal residual disease, HMF is preferable to I-FISH because of the much lower frequency of false-positive readings with the former procedure.

The biological effects of antisense mos expression in fibroblasts.

In order to determine the physiological significance of c-mos RNA expression in somatic cells, we introduced antisense c-mos under the control of an inducible promoter. NIH/3T3 cells were stably transfected with antisense mos under the control of the mouse mammary tumor virus long terminal repeat (MMTV-LTR). Positive transfectants were selected under G418 conditions. Following selection, NIH/3T3 cells that received the antisense mos failed to form foci, whereas sense mos transfected cells grew normally. Moreover, v-mos-transformed cells were unaffected by antisense mos transfection. Of: interest, NIH/3T3 antisense mos transfectants that survived selection were growth-arrested. Nuclear abnormalities and the extrusion of microvesicles containing cellular material were observed in these cells. In order to rescue these cells from growth inhibition, the v-mos gene was introduced into cells by acute infection with Moloney murine sarcoma virus. Following infection, these cells resumed growth and became rapidly transformed. In other experiments, mouse C2 cells stably transfected with antisense mos showed a slower growth rate and gross morphological changes. C2 cells containing antisense mos under the control of mouse metallothionein-1 promoter had a large and flattened morphology and a relatively high percentage (30%) of binucleated cells. Our results indicate that basal level expression of antisense mos (under uninduced conditions) results in either arrested or retarded cell growth. The phenotypes exhibited in both cell lines leads us to suggest that the c-mos expression may play a role in mitotic progression in some somatic cells, in particular affecting cytokinesis.

Towards stoichiometric silver staining of proteins resolved in complex two-dimensional electrophoresis gels: real-time analysis of pattern development.

Toward the ultimate goal of deriving quantitative protein data from silver-stained proteins resolved in complex two-dimensional electrophoresis gels we have performed computerized real-time analysis during pattern protein "silver stain" development. Three points emerged from this study: (i) Development time is a far more important variable than originally envisioned. (ii) Previous studies which, on the basis of a single development time, sought to relate silver stain slope and threshold properties to amino acid composition are non-informative since it will be shown that slope values change markedly with development time. (iii) The combination of real-time analysis and total protein load consideration allows, for the first time, protein "spots" to be grouped into categories, which in turn aid in quantification.

Somatic-cell expression of the mos protooncogene is cell-cycle regulated - highest RNA expression in the g2 phase.

c-mos expression, which occurs at relatively high levels in male and female germ cells, plays an important role in oocyte meiotic maturation. The c-mos proto-oncogene product (c-Mos) is necessary and sufficient to initiate meiosis. It is also an essential component of the cytostatic factor (CSF), which is responsible for arresting vertebrate oocytes at the second meiotic metaphase possibly via stabilization of the maturation promoting factor (MPF). However, much less is understood about c-mos expression and function in somatic cells. We report here that c-mos transcripts can be detected in NIH 3T3 cells by the highly sensitive RNA-PCR method and by RNase protection assays. We found that expression of c-mos RNA is tightly controlled in a cell cycle-dependent manner with highest levels of transcripts (approximately 5 copies per cell) present in the G2 phase. The level of c-mos RNA in synchronized G0/G1 cells was undetectable, and that in S phase cells was extremely low. Similarly, only very low levels of c-mos RNA were detected in nocodazole-arrested M phase cells. The presence of contaminating G2 cells in the synchronized S phase: and M phase populations as well as unsynchronized populations' could 'account for the very low levels of c-mos transcripts detected and supports the interpretation that c-mos RNA is absent in, all phases except G2. These results establish that c-mos expression is not restricted to germ cells, but instead indicate that c-mos RNA expression occurs during the G2 stage of the cell cycle in somatic cells. As in meiosis, c-mos may have a similar function in regulating cell cycle events in somatic cells particularly in controlling entry into mitosis via activation of MPF.

The physical interactions between p37env-mos and tubulin structures.

The c-mos protein has been reported to be complexed with tubulin and to co-localize with microtubules in unfertilized Xenopus eggs as well as in NIH3T3 cells transformed by the Xenopus c-mos gene. We performed experiments to determine whether the viral mos protein, p37v-mos, also associates with tubulin. Both mouse c-mos and v-mos proteins synthesized in vitro co-polymerized with tubulin. Upon incubation at 37 degrees C, essentially all of the mos protein (both viral and cellular) co-polymerized with tubulin, while more than 50% of the tubulin remained in the depolymerized state. The mos-tubulin interaction was specific, as indicated by the insolubility of the v-mos protein following a second cycle of temperature-dependent depolymerization/polymerization. Beta-tubulin was shown to co-precipitate with p37v-mos and to be phosphorylated by the mos kinase in vitro. Although both v-mos and c-mos proteins co-polymerize with tubulin, p37v-mos behaved differently from p39c-mos on gel filtration columns under conditions that favor disassembly of microtubules. Like Xenopus c-mos, the bulk of the mouse c-mos protein synthesized in vitro appeared in structures that fractionate at about 500 kDa. In contrast to c-mos, the majority of the v-mos protein, either isolated from stably transformed NIH3T3 cells or synthesized in vitro, eluted in the 100 kDa fraction, co-fractionating with tubulin dimers. Therefore, the v-mos protein appears to have a higher affinity for unpolymerized tubulin than c-mos, under conditions that favor disassembly of microtubules.

The use of radioactive bacteriophage proteins as X-Y markers for silver stained two-dimensional electrophoresis gels and quantification of the patterns.

We have demonstrated the feasibility of using bacteriophage ghost proteins, tritiated by metabolic labeling, as a set of standard markers for two-dimensional gels in which the proteins are to be detected by silver staining. The results indicate that a 2.5 microgram load of phage proteins yields a reproducible silver pattern of 48 spots. The spots can also be readily identified by radioautography and radiofluorography, establishing their value as a standard constellation of markers. Quantification of these patterns by computerized densitometry indicates a general agreement between detection by silver staining and detection by radiofluorography.

v-mos protein produced by in vitro translation has protein kinase activity.

The v-mos protein, termed p37v-mos, has a closely associated serine/threonine protein kinase activity. To provide further information about its protein kinase activity, we tested the activity of p37v-mos produced in a cell-free translation system from transcripts generated from a cloned v-mos gene. Anti-mos(37-55) immunoprecipitates of in vitro-produced p37v-mos were found to possess serine/threonine protein kinase activity, whereas those obtained with anti-mos(260-271), known to block v-mos autophosphorylation, lacked kinase activity. The phosphorylated products were identical in size to p37v-mos and p43v-mos produced in protein kinase assays from Moloney murine sarcoma virus-infected cells expressing authentic p37v-mos. These results provide further proof that the protein kinase activity associated with p37v-mos is an intrinsic property of the v-mos gene product. This translation system also provides a useful experimental model to study the activation of the mos protein kinase. Thus, protein kinase assays performed on [35S]methionine-labeled p37v-mos produced p43v-mos at the expense of p37v-mos. Phosphatase treatment removed the p43v-mos species, resulting in increase of the p37v-mos-sized protein, confirming our previous interpretation that p43v-mos is a hyperphosphorylated form of p37v-mos.

Genetic analysis of nonhistone chromosomal protein inheritance in recombinant inbred mouse strains using two-dimensional electrophoresis.

Analysis of hepatic nonhistone chromosomal protein (NHCP) expression in male mice from progenitor strains (C3H/HeN, C57BL/6N), their F1 hybrid (B6C3), and seven recombinant inbred strains (RIs) (B6N x C3N) by high-resolution two-dimensional polyacrylamide gel electrophoresis (2D-PAGE) detected 16 NHCPs whose expression in RIs could be correlated to each other and with strain distribution patterns (SDP) of 20 genetic markers differing in the progenitors. Of the 400+ NHCP spots detected in RI 2D-PAGE maps, 172 were common to progenitors and all RIs. There was a characteristic absence of five NHCPs in one RI, Y. Ten C3H-specific and six C57-specific NHCP inherited in B6C3 also appeared in RIs. The SDP of C3H-specific NHCP 2 matched the SDP of beta-glucuronidase on chromosome 5 and carbonic anhydrase on chromosome 3, and C57-specific NHCP 5 SDP corresponded to that for nonagouti trait on chromosome 2. These 16 NHCP genetic marker inheritance differences detected in RIs add to the 23 previously established genetic marker differences between the progenitors.

Evidence for somatic cell expression of the c-mos protein [corrected].

The c-mos protein has been found to be enriched in germ cells of male mice, as described in a recent report from this laboratory (Herzog et al., Oncogene 3, 225, 1988). We report on further studies which indicate that the c-mos protein (a 41 to 43 kDa protein termed p43c-mos) is expressed in somatic tissues of mice and in cells grown in culture. In testes of mice, germ cell fractions have increased levels of p43c-mos relative to other cells of the testes. However, non-germ cells harbor significant levels of p43c-mos, as judged by comparison of testes from normal mice to those with mutations that affect the germ cell content of the testes. Thus, homozygous S1, at, and the W/Wv mutant mice are sterile due to severe deficiencies of germ cells. Such mice had only an estimated 50%-60% reduction in p43c-mos as judged by western immunoblotting using two different site-directed anti-mos antibodies. Similarly, X/X-sex reversed mice in which germ cells die after 10 days of age had only an 85% reduction of p43c-mos in mice 35 days of age. Thus, the germ cell content of testes did not correlated with p43c-mos levels in this tissue. Direct analyses of non-germ cells derived from mouse testes confirmed these findings, since Sertoli and Leydig cell lines grown in culture expressed p43c-mos. In addition, tissues such as kidney, liver, spleen and brain were found to contain p43c-mos. Surprisingly, mouse NIH3T3 cells were found to express significant levels of the c-mos protein based upon immunoblotting and one-dimensional peptide mapping experiments performed with both anti-mos antibodies. The concentration of the c-mos protein was not affected by expression of viral mos proteins. We conclude that the c-mos protein is enriched in male germ cells, but p43c-mos is also expressed in significant amounts in somatic tissues and in fibroblastic cells grown in culture.

Heat-stress proteins of rat lung endothelial and mammary adenocarcinoma cells.

A rat mammary adenocarcinoma cell clone, MTC, and a rat lung endothelial cell clone, RLE cl.4, both syngeneic to the Fisher 344 rat, were compared for proteins synthesized at 37 degrees C and after a 1-h, 42 degrees C heat dose. The heat stress-induced or -enhanced synthesis of a series of molecular mass groups and isoelectric point species (isomers) was observed in both equilibrium and nonequilibrium two-dimensional gel electrophoresis. Tumor and endothelial cell heat-stress proteins (hsp) were strikingly similar with most hsp in 11 or 13 molecular mass groups having from 1 to 12 major isomers. In comparing the two cell types, 6 of about 23 major hsp isomers appeared different in equilibrium pH gels, with tumor cells seemingly exhibiting less synthesis of these 6 isomers. Four additional endothelial cell hsp isomers were apparent in nonequilibrium pH gels. Since two of these later hsp can be found at higher heat doses in tumor cells, some of these apparent differences between tumor and endothelial cells may be attributable to different dose ranges for induction of hsp. Fluorograms and silver-stained gels showed that several hsp were being synthesized at appreciable levels in unheated cells. However, there were hsp whose synthesis appeared to be de novo rather than representing enhanced synthesis of existing proteins. These last two observations were made in both tumor and normal cells. The constitutive levels of hsp synthesis appeared to be generally similar in unheated tumor and normal cells in vitro with few exceptions. These results indicate the presence of few unique hsp in syngeneic tumor and normal cells in vitro. However, focusing subsequent studies on the few differences may lead to insights concerning hyperthermic biology of tumor and normal cells, phenotypic differences between these cells, and roles of some hsp.

Alterations in nonhistone chromatin proteins during hepatocarcinogenesis induced by diverse acting carcinogens.

Nonhistone chromosomal proteins (NHCP) from normal, regenerating rat liver, fetal liver, stages during acetylaminofluorene (AAF) and diethylnitrosamine (DEN) induced carcinogenesis, and resultant primary hepatocellular carcinomas (PHC) were analyzed by two-dimensional polyacrylamide gel electrophoresis (2D-PAGE). These studies sought to determine if changes in proteins putatively involved in catalyzing specific gene expression (NHCP) occur during liver cancer development that might be related to the malignant phenotype. NHCP extracted in high salt-urea buffers, analyzed by 2D-PAGE and silver staining were resolved into some 530-560 polypeptides. Increases in number of NHCP amounting to 8.4, 8.6 and 8.8%, respectively, were detected in AAF induced nodules (AAF-NOD), AAF-PHCs and DEN-PHCs when compared to normal chromatins. The majority of the 51 qualitative changes detected reflected cell cycling and/or reexpression of fetal-NHCP. Within the total changes, seven new NHCP were found only in AAF- and DEN-induced PHCs. Further, four NHCP with isoelectric points and relative molecular weights (pI/MW) of 5.62/19.3, 5.96/30.7, 6.25/46.6 and 8.16/53.5 occurring in both AAF- and DEN-PHC also were found in AAF-NOD, a carcinogenesis stage considered to represent premalignant nodules. Reciprocally, three NHCP of pI/mol.wt.: 6.81/34.0, 5.82/43.7 and 8.18/67.0 present in normal liver, disappeared in all carcinogen involved tissues analyzed. These findings indicate that while AAF and DEN exposure results in a number of qualitative NHCP changes specific for the particular carcinogen, a total of only ten changes, seven inductions and three losses, occurred in common during hepatocarcinogenesis induced by these diverse agents. At least four of these NHCP may prove critical inductions during malignant conversion or alternatively might serve as tumor markers since they appear first in a well characterized premalignant stage and persist in resultant tumors.

Decrease in the phosphorylation of the proteins associated with heterogeneous nuclear RNA after the application of 3'-methyl-4-dimethylaminoazobenzene.

The effect of 3'-methyl-4-dimethylaminoazobenzene (3'-MeDAB) on the phosphorylation of the proteins of the nuclear ribonucleoprotein (RNP) particles was studied in liver of rats. Forty eight hours after the application of 4 mg of the hepatocarcinogen per 100 g of body wt. by stomach intubation the particle proteins contained only 7% as much phosphate per mg of protein as the proteins of the same particles isolated from liver of control animals. Determination of the protein kinase and protein phosphatase activities in the total fraction of the non-histone nuclear proteins 48 h after the application of the carcinogen have shown an increase (200% and 159%, respectively) in both enzymatic activities. These results suggest that the hepatocarcinogen could induce the observed high turnover of the phosphates on the proteins of the liver nuclear ribonucleoprotein particles and the resulting dephosphorylation of these particles by stimulation of nuclear protein kinases and phosphatases. Qualitatively the same, but quantitatively much smaller changes were also observed 48 h after the application of the non-carcinogenic p-aminoazobenzene (AB) by stomach intubation and in regenerating liver. After the application of AB phosphorylation of the proteins of rat liver nuclear ribonucleoprotein particles decreased to 70% and in regenerating liver to 61% of the phosphorylation of particle proteins in control liver. Since it is assumed that nuclear RNP particles are involved in the processing and transport of newly synthesized premessenger RNA it is possible that the drastic dephosphorylation of the particle proteins induced by the carcinogen could be connected with the distortion of RNA processing which is observed in liver of animals treated with hepatocarcinogens.