Significance analysis of microarrays applied to the ionizing radiation response

Microarrays can measure the expression of thousands of genes to identify changes in expression between different biological states. Methods are needed to determine the significance of these changes while accounting for the enormous number of genes. We describe a method, Significance Analysis of Microarrays (SAM), that assigns a score to each gene on the basis of change in gene expression relative to the standard deviation of repeated measurements. For genes with scores greater than an adjustable threshold, SAM uses permutations of the repeated measurements to estimate the percentage of genes identified by chance, the false discovery rate (FDR). When the transcriptional response of human cells to ionizing radiation was measured by microarrays, SAM identified 34 genes that changed at least 1.5-fold with an estimated FDR of 12%, compared with FDRs of 60 and 84% by using conventional methods of analysis. Of the 34 genes, 19 were involved in cell cycle regulation and 3 in apoptosis. Surprisingly, four nucleotide excision repair genes were induced, suggesting that this repair pathway for UV-damaged DNA might play a previously unrecognized role in repairing DNA damaged by ionizing radiation.

UV-induced Hyperphosphorylation of Replication Protein A Depends on DNA Replication and Expression of ATM Protein

Exposure to DNA-damaging agents triggers signal transduction pathways that are thought to play a role in maintenance of genomic stability. A key protein in the cellular processes of nucleotide excision repair, DNA recombination, and DNA double-strand break repair is the single-stranded DNA binding protein, RPA. We showed previously that the p34 subunit of RPA becomes hyperphosphorylated as a delayed response (4–8 h) to UV radiation (10–30 J/m2). Here we show that UV-induced RPA-p34 hyperphosphorylation depends on expression of ATM, the product of the gene mutated in the human genetic disorder ataxia telangiectasia (A-T). UV-induced RPA-p34 hyperphosphorylation was not observed in A-T cells, but this response was restored by ATM expression. Furthermore, purified ATM kinase phosphorylates the p34 subunit of RPA complex in vitro at many of the same sites that are phosphorylated in vivo after UV radiation. Induction of this DNA damage response was also dependent on DNA replication; inhibition of DNA replication by aphidicolin prevented induction of RPA-p34 hyperphosphorylation by UV radiation. We postulate that this pathway is triggered by the accumulation of aberrant DNA replication intermediates, resulting from DNA replication fork blockage by UV photoproducts. Further, we suggest that RPA-p34 is hyperphosphorylated as a participant in the recombinational postreplication repair of these replication products. Successful resolution of these replication intermediates reduces the accumulation of chromosomal aberrations that would otherwise occur as a consequence of UV radiation.

Comparison of the Ability of Partially N-Acetylated Chitosans and Chitooligosaccharides to Elicit Resistance Reactions in Wheat Leaves1

Chitin, a linear polysaccharide composed of (1→4)-linked 2-acetamido-2-deoxy-β-d-glucopyranose (GlcNAc) residues, and chitosan, the fully or partially N-acetylated, water-soluble derivative of chitin composed of (1→4)-linked GlcNAc and 2-amino-2-deoxy-β-d-glucopyranose (GlcN), have been proposed as elicitors of defense reactions in higher plants. We tested and compared the ability of purified (1→4)-linked oligomers of GlcNAc (tetramer to decamer) and of GlcN (pentamer and heptamer) and partially N-acetylated chitosans with degrees of acetylation (DA) of 1%, 15%, 35%, 49%, and 60% and average degrees of polymerization between 540 and 1100 to elicit phenylalanine ammonia-lyase (PAL) and peroxidase (POD) activities, lignin deposition, and microscopically and macroscopically visible necroses when injected into the intercellular spaces of healthy, nonwounded wheat (Triticum aestivum L.) leaves. Purified oligomers of (1→4)-linked GlcN were not active as elicitors, whereas purified oligomers of (1→4)-linked GlcNAc with a degree of polymerization ≥ 7 strongly elicited POD activities but not PAL activities. Partially N-acetylated, polymeric chitosans elicited both PAL and POD activities, and maximum elicitation was observed with chitosans of intermediate DAs. All chitosans but not the chitin oligomers induced the deposition of lignin, the appearance of necrotic cells exhibiting yellow autofluorescence under ultraviolet light, and macroscopically visible necroses; those with intermediate DAs were most active. These results suggest that different mechanisms are involved in the elicitation of POD activities by GlcNAc oligomers, and of PAL and POD activities by partially N-acetylated chitosan polymers and that both enzymes have to be activated for lignin biosynthesis and ensuing necrosis to occur.

Clues to epidermal cancer proneness revealed by reconstruction of DNA repair-deficient xeroderma pigmentosum skin in vitro

Sun exposure has been clearly implicated in premature skin aging and neoplastic development. These features are exacerbated in patients with xeroderma pigmentosum (XP), a hereditary disease, the biochemical hallmark of which is a severe deficiency in the nucleotide excision repair of UV-induced DNA lesions. To develop an organotypic model of DNA repair deficiency, we have cultured several strains of primary XP keratinocytes and XP fibroblasts from skin biopsies of XP patients. XP skin comprising both a full-thickness epidermis and a dermal equivalent was succesfully reconstructed in vitro. Satisfactory features of stratification were obtained, but the expression of epidermal differentiation products, such as keratin K10 and loricrin, was delayed and reduced. In addition, the proliferation of XP keratinocytes was more rapid than that of normal keratinocytes. Moreover, increased deposition of cell attachment proteins, α-6 and β-1 integrins, was observed in the basement membrane zone, and β-1 integrin subunit, the expression of which is normally confined to basal keratinocytes, extended into several suprabasal cell layers. Most strikingly, the in vitro reconstructed XP skin displayed numerous proliferative epidermal invasions within dermal equivalents. Epidermal invasion and higher proliferation rate are reminiscent of early steps of neoplasia. Compared with normal skin, the DNA repair deficiency of in vitro reconstructed XP skin was documented by long-lasting persistence of UVB-induced DNA damage in all epidermal layers, including the basal layer from which carcinoma develops. The availability of in vitro reconstructed XP skin provides opportunities for research in the fields of photoaging, photocarcinogenesis, and tissue therapy.

Assembly of RecA-like recombinases: Distinct roles for mediator proteins in mitosis and meiosis

Members of the RecA family of recombinases from bacteriophage T4, Escherichia coli, yeast, and higher eukaryotes function in recombination as higher-order oligomers assembled on tracts of single-strand DNA (ssDNA). Biochemical studies have shown that assembly of recombinase involves accessory factors. These studies have identified a class of proteins, called recombination mediator proteins, that act by promoting assembly of recombinase on ssDNA tracts that are bound by ssDNA-binding protein (ssb). In the absence of mediators, ssb inhibits recombination reactions by competing with recombinase for DNA-binding sites. Here we briefly review mediated recombinase assembly and present results of new in vivo experiments. Immuno-double-staining experiments in Saccharomyces cerevisiae suggest that Rad51, the eukaryotic recombinase, can assemble at or near sites containing ssb (replication protein A, RPA) during the response to DNA damage, consistent with a need for mediator activity. Correspondingly, mediator gene mutants display defects in Rad51 assembly after DNA damage and during meiosis, although the requirements for assembly are distinct in the two cases. In meiosis, both Rad52 and Rad55/57 are required, whereas either Rad52 or Rad55/57 is sufficient to promote assembly of Rad51 in irradiated mitotic cells. Rad52 promotes normal amounts of Rad51 assembly in the absence of Rad55 at 30°C but not 20°C, accounting for the cold sensitivity of rad55 null mutants. Finally, we show that assembly of Rad51 is induced by radiation during S phase but not during G1, consistent with the role of Rad51 in repairing the spontaneous damage that occurs during DNA replication.

UV-induced ubiquitination of RNA polymerase II: a novel modification deficient in Cockayne syndrome cells.

Damage to actively transcribed DNA is preferentially repaired by the transcription-coupled repair (TCR) system. TCR requires RNA polymerase II (Pol II), but the mechanism by which repair enzymes preferentially recognize and repair DNA lesions on Pol II-transcribed genes is incompletely understood. Herein we demonstrate that a fraction of the large subunit of Pol II (Pol II LS) is ubiquitinated after exposing cells to UV-radiation or cisplatin but not several other DNA damaging agents. This novel covalent modification of Pol II LS occurs within 15 min of exposing cells to UV-radiation and persists for about 8-12 hr. Ubiquitinated Pol II LS is also phosphorylated on the C-terminal domain. UV-induced ubiquitination of Pol II LS is deficient in fibroblasts from individuals with two forms of Cockayne syndrome (CS-A and CS-B), a rare disorder in which TCR is disrupted. UV-induced ubiquitination of Pol II LS can be restored by introducing cDNA constructs encoding the CSA or CSB genes, respectively, into CS-A or CS-B fibroblasts. These results suggest that ubiquitination of Pol II LS plays a role in the recognition and/or repair of damage to actively transcribed genes. Alternatively, these findings may reflect a role played by the CSA and CSB gene products in transcription.

Ultraviolet-induced movement of the human DNA repair protein, Xeroderma pigmentosum type G, in the nucleus.

Xeroderma pigmentosum type G (XPG) is a human genetic disease exhibiting extreme sensitivity to sunlight. XPG patients are defective XPG endonuclease, which is an enzyme essential for DNA repair of the major kinds of solar ultraviolet (UV)-induced DNA damages. Here we describe a novel dynamics of this protein within the cell nucleus after UV irradiation of human cells. Using confocal microscopy, we have localized the immunofluorescent, antigenic signal of XPG protein to foci throughout the cell nucleus. Our biochemical studies also established that XPG protein forms a tight association with nuclear structure(s). In human skin fibroblast cells, the number of XPG foci decreased within 2 h after UV irradiation, whereas total nuclear XPG fluorescence intensity remained constant, suggesting redistribution of XPG from a limited number of nuclear foci to the nucleus overall. Within 8 h after UV, most XPG antigenic signal was found as foci. Using beta-galactosidase-XPG fusion constructs (beta-gal-XPG) transfected into HeLa cells, we have identified a single region of XPG that is evidently responsible both for foci formation and for the UV dynamic response. The fusion protein carrying the C terminus of XPG (amino acids 1146-1185) localized beta-gal specific antigenic signal to foci and to the nucleolus regions. After UV irradiation, antigenic beta-gal translocated reversibly from the subnuclear structures to the whole nucleus with kinetics very similar to the movements of XPG protein. These findings lead us to propose a model in which distribution of XPG protein may regulate the rate of DNA repair within transcriptionally active and inactive compartments of the cell nucleus.

Segregation of DNA polynucleotide strands into sister chromatids and the use of endoreduplicated cells to track sister chromatid exchanges induced by crosslinks, alkylations, or x-ray damage.

The method of Matsumoto and Ohta [Matsumoto, K. & Ohta, T. (1992) Chromosoma 102, 60-65; Matsumoto, K. & Ohta, T. (1995) Mutat. Res. 326, 93-98] to induce large numbers of endoreduplicated Chinese hamster ovary cells has now been coupled with the fluorescence-plus-Giemsa method of Perry and Wolff [Perry, P. & Wolff, S. (1974) Nature (London) 251, 156-158] to produce harlequin endoreduplicated chromosomes that after the third round of DNA replication are composed of a chromosome with a light chromatid and a dark chromatid in close apposition to its sister chromosome containing two light chromatids. Unless the pattern is disrupted by sister chromatid exchange (SCE), the dark chromatid is always in the center, so that the order of the chromatids is light-dark light-light. The advent of this method, which permits the observation of SCEs in endoreduplicated cells, makes it possible to determine with great ease in which cell cycle an SCE occurred. This now allows us to approach several vexing questions about the induction of SCEs (genetic damage and its repair) after exposure to various types of mutagenic carcinogens. The present experiments have allowed us to observe how many cell cycles various types of lesions that are induced in DNA by a crosslinking agent, an alkylating agent, or ionizing radiation, and that are responsible for the induction of SCEs, persist before being repaired and thus lose their ability to inflict genetic damage. Other experiments with various types of mutagenic carcinogens and various types of cell lines that have defects in different DNA repair processes, such as mismatch repair, excision repair, crosslink repair, and DNA-strand-break repair, can now be carried out to determine the role of these types of repair in removing specific types of lesions.

Peroxo-iron and oxenoid-iron species as alternative oxygenating agents in cytochrome P450-catalyzed reactions: switching by threonine-302 to alanine mutagenesis of cytochrome P450 2B4.

Among biological catalysts, cytochrome P450 is unmatched in its multiplicity of isoforms, inducers, substrates, and types of chemical reactions catalyzed. In the present study, evidence is given that this versatility extends to the nature of the active oxidant. Although mechanistic evidence from several laboratories points to a hypervalent iron-oxenoid species in P450-catalyzed oxygenation reactions, Akhtar and colleagues [Akhtar, M., Calder, M. R., Corina, D. L. & Wright, J. N. (1982) Biochem. J. 201, 569-580] proposed that in steroid deformylation effected by P450 aromatase an iron-peroxo species is involved. We have shown more recently that purified liver microsomal P450 cytochromes, including phenobarbital-induced P450 2B4, catalyze the analogous deformylation of a series of xenobiotic aldehydes with olefin formation. The investigation presented here on the effect of site-directed mutagenesis of threonine-302 to alanine on the activities of recombinant P450 2B4 with N-terminal amino acids 2-27 deleted [2B4 (delta2-27)] makes use of evidence from other laboratories that the corresponding mutation in bacterial P450s interferes with the activation of dioxygen to the oxenoid species by blocking proton delivery to the active site. The rates of NADPH oxidation, hydrogen peroxide production, and product formation from four substrates, including formaldehyde from benzphetamine N-demethylation, acetophenone from 1-phenylethanol oxidation, cyclohexanol from cyclohexane hydroxylation, and cyclohexene from cyclohexane carboxaldehyde deformylation, were determined with P450s 2B4, 2B4 (delta2-27), and 2B4 (delta2-27) T302A. Replacement of the threonine residue in the truncated cytochrome gave a 1.6- to 2.5-fold increase in peroxide formation in the presence of a substrate, but resulted in decreased product formation from benzphetamine (9-fold), cyclohexane (4-fold), and 1-phenylethanol (2-fold). In sharp contrast, the deformylation of cyclohexane carboxaldehyde by the T302A mutant was increased about 10-fold. On the basis of these findings and our previous evidence that aldehyde deformylation is supported by added H202, but not by artificial oxidants, we conclude that the iron-peroxy species is the direct oxygen donor. It remains to be established which of the many other oxidative reactions involving P450 utilize this species and the extent to which peroxo-iron and oxenoid-iron function as alternative oxygenating agents with the numerous isoforms of this versatile catalyst.

Ozone-induced responses in Arabidopsis thaliana: the role of salicylic acid in the accumulation of defense-related transcripts and induced resistance.

Exposure of Arabidopsis thaliana to ozone results in the expression of a number of defense-related genes that are also induced during a hypersensitive response. A potential common link between the activation of defense gene expression during a hypersensitive response and by ozone treatment is the production of active oxygen species and the accumulation of hydrogen peroxide. Here we report that salicylic acid accumulation, which can be induced by hydrogen peroxide and is required for the expression of both a hypersensitive response and systemic acquired resistance, is also required for the induction of some, but not all, ozone-induced mRNAs examined. In addition, we show that ozone exposure triggers induced resistance of A. thaliana to infection with virulent phytopathogenic Pseudomonas syringae strains. Infection of transgenic plants expressing salicylate hydroxylase, which prevents the accumulation of salicylic acid, or npr1 mutant plants, which are defective in the expression of systemic acquired resistance at a step downstream of salicylic acid, demonstrated that the signaling pathway activated during ozone-induced resistance overlaps with the systemic acquired resistance activation pathway and is salicylic acid dependent. Interestingly, plants expressing salicylate hydroxylase exhibited increased sensitivity to ozone exposure. These results demonstrate that ozone activates at least two distinct signaling pathways, including a salicylic acid dependent pathway previously shown to be associated with the activation of pathogen defense reactions, and that this latter pathway also induces a protective response to ozone.

A polygenic mouse model of psoriasiform skin disease in CD18-deficient mice.

Previously, a hypomorphic mutation in CD18 was generated by gene targeting, with homozygous mice displaying increased circulating neutrophil counts, defects in the response to chemically induced peritonitis, and delays in transplantation rejection. When this mutation was backcrossed onto the PL/J inbred strain, virtually all homozygous mice developed a chronic inflammatory skin disease with a mean age of onset of 11 weeks after birth. The disease was characterized by erythema, hair loss, and the development of scales and crusts. The histopathology revealed hyperplasia of the epidermis, subcorneal microabscesses, orthohyperkeratosis, parakeratosis, and lymphocyte exocytosis, which are features in common with human psoriasis and other hyperproliferative inflammatory skin disorders. Repetitive cultures failed to demonstrate bacterial or fungal organisms potentially involved in the pathogenesis of this disease, and the dermatitis resolved rapidly after subcutaneous administration of dexamethasone. Homozygous mutant mice on a (PL/J x C57BL/6J)F1 background did not develop the disease and backcross experiments suggest that a small number of genes (perhaps as few as one), in addition to CD18, determine susceptibility to the disorder. This phenotype provides a model for inflammatory skin disorders, may have general relevance to polygenic human inflammatory diseases, and should help to identify genes that interact with the beta2 integrins in inflammatory processes.

A biomarker that identifies senescent human cells in culture and in aging skin in vivo.

Normal somatic cells invariably enter a state of irreversibly arrested growth and altered function after a finite number of divisions. This process, termed replicative senescence, is thought to be a tumor-suppressive mechanism and an underlying cause of aging. There is ample evidence that escape from senescence, or immortality, is important for malignant transformation. By contrast, the role of replicative senescence in organismic aging is controversial. Studies on cells cultured from donors of different ages, genetic backgrounds, or species suggest that senescence occurs in vivo and that organismic lifespan and cell replicative lifespan are under common genetic control. However, senescent cells cannot be distinguished from quiescent or terminally differentiated cells in tissues. Thus, evidence that senescent cells exist and accumulate with age in vivo is lacking. We show that several human cells express a beta-galactosidase, histochemically detectable at pH 6, upon senescence in culture. This marker was expressed by senescent, but not presenescent, fibroblasts and keratinocytes but was absent from quiescent fibroblasts and terminally differentiated keratinocytes. It was also absent from immortal cells but was induced by genetic manipulations that reversed immortality. In skin samples from human donors of different age, there was an age-dependent increase in this marker in dermal fibroblasts and epidermal keratinocytes. This marker provides in situ evidence that senescent cells may exist and accumulate with age in vivo.

Human immunodeficiency virus reverse transcriptase substrate-induced conformational changes and the mechanism of inhibition by nonnucleoside inhibitors.

A combination of transient kinetic and equilibrium titration methods has been used to show that both primer/template and nucleotide binding to human immunodeficiency virus type 1 (HIV-1) reverse transcriptase are two-step processes. In both cases, after initial formation of relatively weakly bound states, isomerization reactions lead to tightly bound states. In the case of deoxynucleotide binding to the reverse transcriptase-primer/template complex, the second step in the interaction is rate-limiting in the overall reaction during processive polymerization. Discrimination against incorrect nucleotides occurs both in the initial weak binding and in the second step but is purely kinetic in the second step (as opposed to thermodynamic in the first step). Nonnucleoside inhibitors have a relatively small effect on nucleotide-binding steps (overall affinity is reduced by a factor of ca. 10), while the affinity of the primer/template duplex is increased by at least a factor of 10. The major effect of nonnucleoside inhibitors is on the chemical step (nucleotide transfer).

Expression of 5-lipoxygenase in differentiating human skin keratinocytes.

We studied the expression of arachidonate 5-lipoxygenase (5-LO) in a cell line of human keratinocytes (HaCaT) and in normal human skin keratinocytes in tissue culture. In undifferentiated keratinocytes 5-LO gene expression was low or undetectable as determined by 5-LO mRNA, protein, cell-free enzyme activity, and leukotriene production in intact cells. However, after shift to culture conditions that promote conversion of prokeratinocytes into a more differentiated phenotype, 5-LO gene expression was markedly induced in HaCaT cells and, to a lesser extent, in normal keratinocytes. These results show that 5-LO gene expression is an intrinsic property of human skin keratinocytes.

A splice variant of alpha 6 integrin is associated with malignant conversion in mouse skin tumorigenesis.

The epithelial-specific integrin alpha 6 beta 4 is suprabasally expressed in benign skin tumors (papillomas) and is diffusely expressed in carcinomas associated with an increase in the proliferating compartment. Analysis of RNA samples by reverse transcriptase-PCR and DNA sequencing revealed that chemically or oncogenically induced papillomas (n = 8) expressed a single transcript of the alpha 6 subunit, identified as the alpha 6 A splice variant. In contrast, carcinomas (n = 13) expressed both alpha 6A and an alternatively spliced form, alpha 6B. Primary keratinocytes and a number of keratinocyte cell lines that vary in biological potential from normal skin, to benign papillomas, to well-differentiated slowly growing carcinomas exclusively expressed alpha 6A. However, I7, an oncogene-induced cell line that produces highly invasive carcinomas, expressed both alpha 6A and alpha 6B transcript and protein. The expression of alpha 6B in I7 cells was associated with increased attachment to a laminin matrix compared to cell lines exclusively expressing alpha 6A. Furthermore, introduction of an alpha 6B expression vector into a papilloma cell line expressing alpha 6A increased laminin attachment. When a papilloma cell line was converted to an invasive carcinoma by introduction of the v-fos oncogene, the malignant cells expressed both alpha 6A and alpha 6B, while the parent cell line and cells transduced with v-jun or c-myc, which retained the papilloma phenotype, expressed only alpha 6A. Comparative analysis of alpha 6B expression in cell lines and their derived tumors indicate that alpha 6B transcripts are more abundant in tumors than cell lines, and alpha 6B is expressed to a greater extent in poorly differentiated tumors. These results establish a link between malignant conversion and invasion of squamous tumor cells and the regulation of transcript processing of the alpha 6 beta 4 integrin.