Mammalian chromosome structure

Analysis of the organisation and localisation of the FSHD-associated tandem array in primates: implications for the origin and evolution of the 3.3 kb repeat family

The D4Z4 locus is a polymorphic tandem repeat sequence on human chromosome 4q35. This locus is implicated in the neuromuscular disorder facioscapulohumeral muscular dystrophy (FSHD). The majority of sporadic cases of FSHD are associated with de novo DNA deletions within D4Z4. However, it is still not known how this rearrangement causes FSHD. Although the repeat contains homeobox sequences, despite exhaustive searching, no transcript from this locus has been identified. Therefore, it has been proposed that the deletion may invoke a position effect on a nearby gene. In order to try to understand the role of the D4Z4 repeat in this disease, we decided to investigate its conservation in other species. In this study, the long-range organisation and localisation of loci homologous to D4Z4 were investigated in primates using Southern blot analysis, pulsed field gel electrophoresis and fluorescence in situ hybridisation. In humans, probes to D4Z4 identify, in addition to the 4q35 locus, a closely related tandem repeat at 10qter and many related repeat loci mapping to the acrocentric chromosomes; a similar pattern was seen in all the great apes. In Old World monkeys, however, only one locus was detected in addition to that on the homologue of human chromosome 4, suggesting that the D4Z4 locus may have originated directly from the progenitor locus. The finding that tandem arrays closely related to D4Z4 have been maintained at loci homologous to human chromosome 4q35-qter in apes and Old World monkeys suggests a functionally important role for these sequences.

Analysis of extrachromosomal structures containing human centromeric alphoid satellite DNA sequences in mouse cells

Yeast artificial chromosomes (YACs) spanning the centromeric region of the human Y chromosome were introduced into mouse LA-9 cells by spheroplast fusion in order to determine whether they would form mammalian artificial chromosomes. In about 50% of the cell lines generated, the YAC DNA was associated with circular extrachromosomal structures. These episomes were only present in a proportion of the cells, usually at high copy number, and were lost rapidly in the absence of selection. These observations suggest that, despite the presence of centromeric sequences, the structures were not segregating efficiently and thus were not forming artificial chromosomes. However, extrachromosomal structures containing alphoid DNA appeared cytogenetically smaller than those lacking it, as long as yeast DNA was also absent. This suggests that alphoid DNA can generate the condensed chromatin structure at the centromere.

Chromosome manipulation: a systematic approach toward understanding human chromosome structure and function

Images

Identification of overlapping DNA-binding and centromere-targeting domains in the human kinetochore protein CENP-C

The kinetochore in eukaryotes serves as the chromosomal site of attachment for microtubules of the mitotic spindle and directs the movements necessary for proper chromosome segregation. In mammalian cells, the kinetochore is a highly differentiated trilaminar structure situated at the surface of the centromeric heterochromatin. CENP-C is a basic, DNA-binding protein that localizes to the inner kinetochore plate, the region that abuts the heterochromatin. Microinjection experiments using antibodies specific for CENP-C have demonstrated that this protein is required for the assembly and/or stability of the kinetochore as well as for a timely transition through mitosis. From these observations, it has been suggested that CENP-C is a structural protein that is involved in the organization or the kinetochore. In this report, we wished to identify and map the functional domains of CENP-C. Analysis of CENP-C truncation mutants expressed in vivo demonstrated that CENP-C possesses an autonomous centromere-targeting domain situated at the central region of the CENP-C polypeptide. Similarly, in vitro assays revealed that a region of CENP-C with the ability to bind DNA is also located at the center of the CENP-C molecule, where it overlaps the centromere-targeting domain.

Gene targeting to the centromeric DNA of a human minichromosome

A human supernumerary minichromosome (MC), previously identified as a derivative of chromosome 9, has been introduced into Chinese hamster ovary (CHO) cells by means of cell fusion. A hybrid clone containing the MC as the only free human chromosome was isolated. A selectable marker gene (neo) inserted into a yeast artificial chromosome (YAC) has been successfully targeted to the MC centromeric DNA via co-transfection with chromosome-9-specific alpha satellite DNA. In situ hybridization and Southern blotting experiments demonstrated that the intact neo gene was integrated into the MC centromeric DNA. Studies on the clonal distribution and on the stability of the MC either in the presence or in the absence of the selective agent have been carried out. The MC is susceptible to further manipulations and may thus represent a model for the construction of a large-capacity vector for somatic gene therapy.

Distribution and linkage of repetitive clusters from the heterochromatic region of human chromosome 22

The pericentric regions of eukaryotic chromosomes consist of several types of repetitive DNA families. In human chromosome 22, the organization of such families was studied in more detail. In addition to the known families of alpha and beta repeats, an additional repeat with a 48-bp motif was previously assigned to 22pter-q11. Here, we report in more detail the distribution of these repeat families, applying pulsed-field gel electrophoresis, fluorescence in situ hybridization and physical linkage on cosmid recombinants. At least two clusters of 48-bp repeats are localized on chromosome 22, one on the distal p-arm and one in the region 22cen-q11. Cosmids from a chromosome 22 library, containing both 48-bp and beta-repeats, link both arrays on 22p and define their maximum distances to less than 44 kb. Loss of 48-bp repeat sequences in a Dl-George cell line carrying a deletion in 22q11 suggests the presence of a second cluster in 22q11, a distribution supported by (fluorescene in situ hybridization)-FISH signal analysis. As additional members of the 48-bp repeat family can be found on all acrocentric chromosomes. It remains to be determined whether the distribution seen on chromosome 22 is also common in other human acrocentric chromosomes.

Mitotic recombination among acrocentric chromosomes' short arms

Routine Q-banding chromosome analysis detected the jumping behaviour of bright fluorescent chromosome 22 satellites (22s+) in two unrelated males (case 1 ascertained for recurrent abortions and case 2 for infertility), and in the mother of one of them, all with a normal karyotype. The 22s+ was present in more than 90% of the cells. In a minority of the cells the polymorphism was present alternatively on another acrocentric, on one chromosome 22 and on another acrocentric, on both chromosomes 22 or it was absent. We take these findings as evidence of mitotic exchanges between the short arms of the acrocentric chromosomes. The presence of a stable 22s+ in the fibroblasts of case 1 and in the lymphocytes of his son indicates that acrocentric short arm exchanges depend both on the type of tissue and on the genetic content of all the other acrocentrics.

Microsatellites and PCR genomic analysis

Microsatellites form a significant proportion of the growing family of repetitive DNA sequences, widely dispersed in the human genome. Due to their ubiquity, PCR (polymerase chain reaction) typability, Mendelian co-dominant inheritance, and extreme polymorphism, microsatellites have assumed an increasingly important role as markers in the genome. Apart from their obvious applications in genome mapping and positional cloning, these markers have been applied in fields as disparate as tumour biology, personal identification, population genetic analysis, and the construction of human evolutionary trees. Microsatellites are associated with human disease, not only as markers of risk but also directly in disease aetiopathogenesis, providing new insights into non-Mendelian inheritance; the replication, repair, and mutation of eukaryotic DNA; the regulation of gene transcription; and protein-protein interactions. These insights have resulted in novel paradigms for oncogenesis and neurological disease.

Molecular mapping of the centromeres of tomato chromosomes 7 and 9

The centromeres of two tomato chromosomes have been precisely localized on the molecular linkage map through dosage analysis of trisomic stocks. To map the centromeres of chromosomes 7 and 9, complementary telo-, secondary, and tertiary trisomic stocks were used to assign DNA markers to their respective chromosome arms and thus to localize the centromere at the junction of the short and long arms. It was found that both centromeres are situated within a cluster of cosegregating markers. In an attempt to order the markers within the centric clusters, genetic maps of the centromeric regions of chromosomes 7 and 9 were constructed from F2 populations of 1620 Lycopersicon esculentum x L. pennellii (E x P) plants and 1640 L. esculentum x L. pimpinellifolium (E x PM) plants. Despite the large number of plants analyzed, very few recombination events were detected in the centric regions, indicating a significant suppression of recombination at this region of the chromosome. The fact that recombination suppression is equally strong in crosses between closely related (E x PM) and remotely related (E x P) parents suggests that centromeric suppression is not due to DNA sequence mismatches but to some other mechanism. The greatest number of centromeric markers was resolved in the L. esculentum x L. pennellii F2 population. The centromere of chromosome 7 is surrounded by eight cosegregating markers: three on the short arm, five on the long arm. Similarly, the centric region of chromosome 9 contains ten cosegregating markers including one short arm marker and nine long arm markers. The localization of centromeres to precise intervals on the molecular linkage map represents the first step towards the characterization and ultimate isolation of tomato centromeres.

SINE and LINE within human centromeres

A number of the Alu and L1 elements present within the centromeric regions of the human chromosomes have been analyzed by polymerase chain reaction amplification. The oligonucleotide primers were homologous to the 3' end consensus sequences of either Alu or L1 in conjunction with an oligonucleotide primer homologous to alphoid sequences specific to different chromosomes. This allowed one to detect an unusual number of Alu and L1 polymorphisms at different loci. It is proposed that this results from molecular rearrangements which occur within the alpha-satellite DNA in which they are embedded (Marçais et al. J. Mol. Evol. 33:42-48, 1991) and not because the centromeric regions are targets for new insertions of such elements. The same analyses were made on cosmids and YACs originating from the centromeric region of chromosome 21 as well as on a collection of somatic hybrids containing chromosome 21 centromere as unique common human genetic material. The results were consistent with the above hypothesis.

Centromeres of human chromosomes

The centromere, recognized cytologically as the primary constriction, is essential for chromosomal attachment to the spindle and for proper segregation of mitotic and meiotic chromosomes. Considerable progress has been made in identifying both DNA and protein components of the centromere and kinetochore complex in mammalian chromosomes, including definition of specific motor proteins with demonstrable functions in chromosome movement. Searches for possible environmental influences on chromosome disjunction might logically be based on known components of the segregation apparatus, both intrinsic and extrinsic to the chromosomes themselves. This article reviews available information on both DNA and protein components of the centromere of mammalian, particularly human, chromosomes and summarizes our current understanding of their role(s) in facilitating normal chromosome behavior in mitosis and meiosis.

Chromosomal differences in susceptibility to meiotic aneuploidy

A basic question concerning the origins of germ cell aneuploidy is whether the same mechanisms operate for all chromosomes, or whether there are chromosome-specific factors influencing the susceptibility to nondisjunction. Although selective loss of some trisomies in early gestation may contribute to the observed differences in trisomy frequency, data from spontaneous abortions, early embryos and gametes strongly suggest that there are real differences in the frequency with which different trisomies arise. In particular the preponderance of trisomy 16 and acrocentric trisomy appears to be present at conception. Maternal and paternal age relationships also differ among trisomies, as do the extent of maternal and paternal contributions, and the relative frequency of meiosis I and meiosis II errors. Recombination patterns associated with nondisjunction also show chromosomal differences. Chromosomal differences in length, centromere position, pericentromeric and other repetitive sequences, recombination patterns and chromatin characteristics might all be related to a differential susceptibility to aneuploidy, but no current explanation accounts for the excess of maternally derived trisomy 16. The existence of chromosome-specific factors makes extrapolation from observations on one chromosome to all aneuploidy unwise, both for investigations into the causes of aneuploidy, and for surveillance of aneuploidy frequency.

The centromere: hub of chromosomal activities

Centromeres are the structures that direct eukaryotic chromosome segregation in mitosis and meiosis. There are two major classes of centromeres. Point centromeres, found in the budding yeasts, are compact loci whose constituent proteins are now beginning to yield to biochemical analysis. Regional centromeres, best described in the fission yeast Schizosaccharomyces pombe, encompass many kilobases of DNA and are packaged into heterochromatin. Their associated proteins are as yet poorly understood. In addition to providing the site for microtubule attachment, centromeres also have an important role in checkpoint regulation during mitosis.

CENP-B binds a novel centromeric sequence in the Asian mouse Mus caroli

Minor satellite DNA, found at Mus musculus centromeres, is not present in the genome of the Asian mouse Mus caroli. This repetitive sequence family is speculated to have a role in centromere function by providing an array of binding sites for the centromere-associated protein CENP-B. The apparent absence of CENP-B binding sites in the M. caroli genome poses a major challenge to this hypothesis. Here we describe two abundant satellite DNA sequences present at M. caroli centromeres. These satellites are organized as tandem repeat arrays, over 1 Mb in size, of either 60- or 79-bp monomers. All autosomes carry both satellites and small amounts of a sequence related to the M. musculus major satellite. The Y chromosome contains small amounts of both major satellite and the 60-bp satellite, whereas the X chromosome carries only major satellite sequences. M. caroli chromosomes segregate in M. caroli x M. musculus interspecific hybrid cell lines, indicating that the two sets of chromosomes can interact with the same mitotic spindle. Using a polyclonal CENP-B antiserum, we demonstrate that M. caroli centromeres can bind murine CENP-B in such an interspecific cell line, despite the absence of canonical 17-bp CENP-B binding sites in the M. caroli genome. Sequence analysis of the 79-bp M. caroli satellite reveals a 17-bp motif that contains all nine bases previously shown to be necessary for in vitro binding of CENP-B. This M. caroli motif binds CENP-B from HeLa cell nuclear extract in vitro, as indicated by gel mobility shift analysis. We therefore suggest that this motif also causes CENP-B to associate with M. caroli centromeres in vivo. Despite the sequence differences, M. caroli presents a third, novel mammalian centromeric sequence producing an array of binding sites for CENP-B.

Elements of chromosome structure and function in fission yeast

The investigation of fission yeast chromosome structure and function has moved rapidly over the past 10 years. The isolation of replication origins, telomeres and centromeres has allowed the development of minichromosomes, a yeast artificial chromosome (YAC)-like cloning system and investigations into chromosome segregation and behaviour during mitosis and meiosis. Many mutants have been isolated which are defective in chromosome segregation. The development of the fluorescent in-situ hybridization (FISH) technique for use in S. pombe has allowed the localization of centromeres and telomeres throughout mitosis and meiosis. In combination with indirect immunofluorescence to detect spindle and chromosomal proteins, the FISH technique should further advance our understanding of fission yeast chromosome structure and function. The recent discovery of a heterochromatin-like structure mediating transcriptional repression at centromeres reinforces the notion that fission yeast centromeres are similar to those of larger eukaryotes. Further characterization of such phenomena will accelerate the genetic dissection of this important chromosomal element.

A tightly bound chromosome antigen is detected by monoclonal antibodies in a ring-like structure on human centromeres

Monoclonal antibodies (Mabs) were raised against isolated Chinese hamster protein-depleted chromosomes Chromosome scaffolds) in order to probe for components involved in the higher-order structure of mammalian chromosomes. One of the Mabs detected a ring-like structure in metaphase at the centromere, which is conserved between Chinese hamster and human cells. Additionally, the Mab stained the centrioles in interphase cells in these two species. The antigen was enriched in chromosomal protein preparations by comparison with nuclear protein samples, and has an apparent Mr = 170,000. The centromere antigen remained present in chromosome scaffold preparations, indicating that it was tightly associated with DNA. The antigen was distinct in its centromeric localisation from any of the centromere antigens reported to date. A possible role of the antigen in stabilising the centromere, by holding the sister chromatids together until their separation at the metaphase-anaphase transition is presented.

Mutations derepressing silent centromeric domains in fission yeast disrupt chromosome segregation

The ura4+ gene displays phenotypes consistent with variegated expression when inserted at 11 sites throughout fission yeast centromere 1. An abrupt transition occurs between the zone of centromeric repression and two adjacent expressed sites. Mutations in six genes alleviate repression of the silent-mating type loci and of ura4+ expressed from a site adjacent to the silent locus, mat3-M. Defects at all six loci affect repression of the ura4+ gene adjacent to telomeres and at the three centromeric sites tested. The clr4-S5 and rik1-304 mutations cause the most dramatic derepression at two out of three sites within cen1. All six mutations had only slight or intermediate effects on a third site in the center of cen1 or on telomeric repression. Strains with lesions at the clr4, rik1, and swi6 loci have highly elevated rates of chromosome loss. We propose that the products of these genes are integral in the assembly of a heterochromatin-like structure, with distinct domains, enclosing the entire centromeric region that reduces or excludes access to transcription factors. The formation of this heterochromatic structure may be an absolute requirement for the formation of a fully functional centromere.

Centromeric dodeca-satellite DNA sequences form fold-back structures

The evolutionarily conserved centromeric dodeca-satellite DNA has an asymmetric distribution of guanine and cytosine residues resulting in one strand being relatively G-rich. This dodeca-satellite G-strand contains a GGGA-tract that is similar to the homopurine tracts found in most telomeric DNA sequences. Here, we show that the dodeca-satellite G-strand forms intramolecular hairpin structures that are stabilized by the formation of non-Watson-Crick G.A pairs as well as regular Watson-Crick G.C pairs. Special stacking interactions are also likely to contribute significantly to the stability of this structure. This hairpin conformation melts at relatively high temperature, around 75 degrees C, and is detected under many different ionic and pH conditions. As judged by electron microscopy visualization, these structures can be formed in a B-DNA environment. Under the same experimental conditions, neither the C-strand nor the double-stranded dodeca-satellite DNA were found to form any unusual DNA structure. A protein activity has been detected that preferentially binds to the single-stranded dodeca-satellite C-strand. The biological relevance of these results is discussed in view of the similarities to telomeric DNA.

Analysis of centromeric activity in Robertsonian translocations: implications for a functional acrocentric hierarchy

Approximately 90% of human Robertsonian translocations occur between nonhomologous acrocentric chromosomes, producing dicentric elements which are stable in meiosis and mitosis, implying that one centromere is functionally inactivated or suppressed. To determine if this suppression is random, centromeric activity in 48 human dicentric Robertsonian translocations was assigned by assessment of the primary constrictions using dual color fluorescence in situ hybridization (FISH). Preferential activity/constriction of one centromere was observed in all except three different rearrangements. The activity is meiotically stable since intrafamilial consistency of a preferentially active centromere existed in members of six families. These results support evidence for nonrandom centromeric activity in humans and, more importantly, suggest a functional hierarchy in Robertsonian translocations with the chromosome 14 centromere most often active and the chromosome 15 centromere least often active.

The evolutionary dynamics of repetitive DNA in eukaryotes

Repetitive DNA sequences form a large portion of the genomes of eukaryotes. The 'selfish DNA' hypothesis proposes that they are maintained by their ability to replicate within the genome. The behaviour of repetitive sequences can result in mutations that cause genetic diseases, and confer significant fitness losses on the organism. Features of the organization of repetitive sequences in eukaryotic genomes, and their distribution in natural populations, reflect the evolutionary forces acting on selfish DNA.

Cloning and molecular characterization of a novel chromosome specific centromere sequence of Chinese hamster

We isolated and characterized the first chromosome-specific satellite DNA (HC2sat) of Chinese hamster. This novel satellite was localized to the pericentric region of hamster chromosome 2. The 2.8 kb long repeat unit of HC2sat was identified and two such units were sequenced. Extended short range periodicity could not be revealed in repeat units. These elements are amongst the largest satellite repeat units reported from mammals to date. HC2sat is a major constituent of the pericentric region of CHO chromosome 2 representing a 7-14 Mb long DNA segment. Studies of long range organization of HC2sat indicated that the formation of the satellite array might occur in different phases and involved different amplification mechanisms.

Interstitial deletions of repetitive DNA blocks in dicentric human Y chromosomes

Cytogenetic analysis of aberrant human Y chromosomes was done by fluorescence in situ hybridization (FISH) with Y specific repetitive DNA probes. It revealed an interstitial deletion of different DNA blocks in two dicentric chromosome structures. One deletion includes the total alphoid DNA structure of one centromeric region. The second deletion includes the total repetitive DYZ5 DNA structure in the pericentromeric region of one short Y arm. Both dicentric Y chromosomes were iso(Yp) chromosomes with break and fusion point located in Yq11, the euchromatic part of the long Y arm. Their phenotypic appearance was "abnormal", resembling small monocentric Yq-chromosomes in metaphase plates. Mosaic cell lines, usually included in karyotypes with dicentric Y chromosomes, were not observed. It is assumed that both deletion events suppress the kinetochore activity in one Y centromeric region and thus stabilize its dicentric structure. Local interstitial deletion events had not been described in dicentric human Y chromosomes, but are common in dicentric yeast chromosomes. This raises the question of whether deletion events in dicentric human chromosomes are rare or restricted to the Y chromosome or also represent a general possibility for stabilization of a dicentric chromosome structure in human.

YACs, BACs, PACs and MACs: artificial chromosomes as research tools

Yeast artificial chromosomes (YACs) have become essential research tools as they enable large fragments of DNA to be cloned. In order to overcome several disadvantages of YACs, including chimaerism and instability, several complementary bacterial artificial chromosome (BAC) vectors have been developed. More recently, attempts are being made to construct artificial chromosomes in mammalian cells (MACs).

Unusual chromosome structure of fission yeast DNA in mouse cells

Chromosomes from the fission yeast Schizosaccharomyces pombe have been introduced into mouse cells by protoplast fusion. In most cell lines the yeast DNA integrates into a single site within a mouse chromosome and results in striking chromosome morphology at metaphase. Both light and electron microscopy show that the yeast chromosome region is narrower than the flanking mouse DNA. Regions of the yeast insert stain less intensely with propidium iodide than surrounding DNA and bear a morphological resemblance to fragile sites. We investigate the composition of the yeast transgenomes and the modification and chromatin structure of this yeast DNA in mouse cells. We suggest that the underlying basis for the structure we see lies above the level of DNA modification and nucleosome assembly, and may reflect the attachment of the yeast DNA to the rodent cell nucleoskeleton. The yeast integrant replicates late in S phase at a time when G bands of the mouse chromosomes are being replicated, and participates in sister chromatid exchanges at a high frequency. We discuss the implications of these studies to the understanding of how chromatin folding relates to metaphase chromosome morphology and how large stretches of foreign DNA behave when introduced into mammalian cells.

Mouse centromere mapping using oligonucleotide probes that detect variants of the minor satellite

Cytologically, the centromere is found at the very end of most Mus musculus chromosomes, co-localizing with an array of minor satellite sequences. It is separated from the euchromatin of the long arm by a large domain of heterochromatin, composed in part of arrays of major satellite sequences. We used oligonucleotide probes that specifically detect regions of sequence variation found in certain cloned minor satellite sequences. They detect a limited subset of the minor satellite arrays in the mouse genome, based on both pulsed-field gel electrophoresis and in situ hybridization data, and provide direct molecular genetic markers for individual centromeres in some inbred mouse strains. Array size polymorphisms detected by these probes map to positions consistent with the centromeres of chromosomes 1 and 14 in the BXD recombinant inbred (RI) strains. The genetic distances between these minor satellite arrays and loci on the long arms of chromosomes 1 and 14 are consistent with repression of meiotic recombination in the heterochromatic domains separating them. The existence of chromosome-specific minor satellite sequences implies that the rate of sequence exchange between non-homologous chromosomes relative to the rate between homologous chromosomes is much lower than has previously been postulated. We suggest that the high degree of sequence homogeneity of mouse satellite sequences may instead reflect recent common ancestry.

Biparental inheritance of chromosome 21 polymorphic markers indicates that some Robertsonian translocations t(21;21) occur postzygotically

Robertsonian translocations between acrocentric chromosomes are the most common structural chromosomal rearrangements in humans and many other organisms, and several mechanisms for their formation have been proposed. We have analyzed highly informative DNA polymorphisms in a family with a non-mosaic de novo Robertsonian translocation 21q;21q, to determine the parental origin of the two 21q arms of the rearranged chromosome. The genotypes indicated a biparental origin, i.e. one 21q was paternal and the other maternal. These results imply that in some cases the formation of the rob(21q;21q) occurs in the zygote or in the first few postzygotic mitotic divisions.

5-azacytidine produces differential undercondensation of alpha, beta and classical human satellite DNAs

Fluorescence in situ hybridization employing human alphoid, beta and classical satellite DNA probes was performed on 5-azacytidine treated and untreated chromosomes obtained from human lymphocytes. The individual used in this study presented a polymorphism of constitutive heterochromatin of chromosomes 1 and 9 as revealed by in situ digestion with the restriction endonuclease Alul. Neither the alphoid nor the beta satellite DNA domains were susceptible to condensation-inhibition by 5-azacytidine. Only the classical satellite localized on chromosome 9 was affected. The constitutive heterochromatin size polymorphism was shown to depend mainly on variations of the classical satellite DNA domain. Therefore, condensation-inhibition, as a phenomenon which may modify the natural folding of the chromatin fibre, regionally affects human constitutive heterochromatin and seems to be dependent on the heterochromatic family.

The storage of energy as a cause of malignant transformation: a 7-phase model of carcinogenesis

Although many theories for the development of cancer exist, a new hypothesis for carcinogenesis is suggested and a new 7-phase model of malignant transformation described. Both the hypothesis and the model are based on the principle of a critical point in local energy (entropy?) storage, at a certain level of structural organisation within the cells. This principle has been previously formulated by the author from the rules of non-equilibrium thermodynamics. The model introduces a new terminology and explores the concepts of both high work value of energy and bifurcation. The ability (A) of cells is suggested to be the most important cellular feature in respect to cell survival. This ability implies that the cells follow the requirement Kir > 1, even in dangerous situations and under harmful environmental influence. But, when the cells have lost their ability (A) and all levels of the cell defence machine have been exhausted, then the local energy storage may provoke a cascade of harmful events within the cells. The S-stage is the most unstable state of the cell cycle. If such harmful events take place during DNA synthesis within the 'premalignant' cell, in the phase 'promotion' at the 4th bifurcation, they lead ultimately to carcinogenesis (i.e. malignant transformation). The 7-phase model of carcinogenesis is consistent and offers many advantages in comparison to the previous hypotheses. This model could help the design of experiments, development of new drugs and optimization of medical treatment. The new hypothesis could contribute to a better understanding of the processes of carcinogenesis and the uncontrolled division, growth and lability of tumour cells.