A combination and slight modification of different methods, originally used for spreading polytene chromosomes (Burkholder, 1976; Kalisch and Hägele, 1981), surface studies of chromosome clusters with scanning EM (Welter, Black and Hodge, 1984), and the use of a stable acid-resistant supporting film on grids (Felluga and Martinucci, 1976) or mica mounting technique (Engelhardt and Plagens, 1980a), resulted in new methods which appeared suitable for analysing the higher-order structure of whole-mounted chromosomes and nuclei in transmission EM studies (cf. Ris, 1981).
The original purpose was to develop more useful methods for ultrastructural cytogenetical analyses (karyotyping) than those earlier available (reviewed in Goyanes, 1985); they were often not reproducible or otherwise unsatisfactory for detailed EM inspection. When such methods were worked out, it was recognized that the chromosomes obtained could be used for high-resolution structural analyses; this was not the original intention, indeed the acid isolation methods were thought to be too destructive. Several more gentle "physiological" methods (e.g. Wray and Stubblefield, 1970; Wray, 1973; Spandidos and Siminovitch, 1977; Adolph, 1980; Paulson, 1982) to isolate metaphase chromosomes were also tried. They did not produce any satisfactory results. They were time-consuming, cumbersome and caused unavoidable clumping in pelleting and, more importantly, they caused cytoplasmic contamination that made any closer examinations unsatisfactory and unreliable. The acid procedures, however, did not cause any clumping, retained excellent details and were completely clean of cytoplasmic structures and debris. This has also been noted in the scanning EM study of chromosome clusters (Welter, Black and Hodge, 1984). In addition, acid-isolated chromosomes seem very stable and can be stored in solution (in a refrigerator) until used.
It has been generally considered earlier that chromosomes, nuclei and whole cells are all too bulky to be inspected in an ordinary (60-100 kV) transmission EM and can only be examined in high voltage (MeV) EM to resolve relevant details (cf. Ris, 1981). This is not true when some preparative and other technical steps are slightly changed and carefully controlled. The essential points are, besides the combination of techniques mentioned: the preparations are stained in very diluted uranyl acetate, critical point-dried, photographed on soft film material (high-resolution pan film) and preferably examined as stereo-pair transparencies (negatives or positives). It is not necessary, however, to follow every technical step for Pt rotary-shadowed preparations and replicas. They can, for instance, be photographed on the ordinary film material used in EM photography; in these preparations only the surface of the material is being examined, just as in scanning EM studies.
In the acid-isolated chromosome preparations, the gross and fine structural morphology of the chromosomes, beyond the nucleosome level, is presumably perfectly preserved, because chromatid coiling, scaffold structures and their subunits can be identified in detail. This preservation is not due to histones, because they should have been extracted in the acid washings. It could also be proposed that structures that kept chromosomes together also preserved the shape of the interphase nuclei. Neither were polytene chromosomes released, which is easily understood as they are highly amplified interphase chromosomes. Something that is shared by metaphase chromosomes and interphase nuclei withstands the acidic treatments.
In the scanning-EM investigation reported previously (Welter, Black and Hodge, 1984) a lattice of interwoven fibres or lamina-like structure was recognized on the surface of interphase nuclei and clusters of ana-telophase chromosomes. In the metaphase configuration no interwoven fibres could be detected on the surface of, or between, adjacent chromosomes, except at the "centromeric ring." It was concluded that the lattice structure is presumably the "dense lamina," as it has also been shown that the lamina proteins are disassembled into the cytoplasm during mitosis and not identified in metaphase chromosomes (Gerace and Blobel, 1982). Yet because of the technique (scanning EM) Welter, Black and Hodge (1984) could not look inside the metaphase chromosomes and see the internal network of pore or ring-like structures as described in the present study. More crucially, the internal network of the whole-mounted chromosomes and nuclei was resistant to DNase and 2 M salt solution (before and after DNase digestion) and devoid of any detectable DNA with a sensitive DNA-specific fluorochrome (DAPI). The best candidate for this internal network, so far, seems to be the matrix-scaffold, which can therefore be described as a pore or ring-perforated network, which appears identical to the similarly treated NE; this confirms the results obtained with manually isolated unfixed polytene chromosomes and NEs (Results I).
It could be argued that extraction and digestion were affected by the prefixation steps. However, the prefixation step did not seem to be necessarily involved, because similar pilot preparations of metaphase chromosomes could be made without this step and generally polytene chromosomes were not prefixed. The prefixation step was found most useful, mainly for practical reasons, in collecting material from different sources, which is the standard method in cytogenetical techniques. The obvious reason for the success of releasing chromosomes from fixed material lies in the fact that "fixing" with methanol/acetic acid is not a real fixation: no cross-linking occurs, some dissolution and extraction by acetic acid takes place, and the "fixation" is essentially a precipitation.
It could also be argued that the network of pore and ring-like structures as described here in stereo pairs was an artifact produced by the acidic preparative conditions. This claim would be valid if such structures had not been described earlier, or could not be detected in a re-examination of other whole-mount preparations of chromosomes from other procedures. Pores and annuli connected with chromatin fibres have been noticed previously and are found even in the earliest literature of spreading techniques (DuPraw, 1965; Comings and Okada, 1970; Lampert, 1971). Yet generally they have not been described as embedded inside condensed, morphologically intact, whole-mounted chromosomes. This may surely result from the difficulty of detecting them if the preparation steps were not carefully executed (as discussed above); stereoscopic EM techniques are an exellent aid in final inspection and interpretation.
In a re-examination, for instance, of some published stereo pair pictures of whole-mounted metaphase chromosomes, isolated by the hexylene glycol method (Stubblefield and Wray, 1971), pore or ring-like structures, in the same size range, can clearly be detected (e.g. in their Figs. 4-7, 11, 17; that is when one knows what to look for), and in stereo pairs of whole-mounted chromosomes produced by other techniques (H. Ris 1977, pers. comm. concerning stereo pair pictures and diapositives taken with a high voltage EM and presented in a poster session at the Helsinki Chromosome Conference; see also Ris , 1981).
These few examples suffice to warrant the conclusion that the chromosomes isolated with the acidic procedures do not contain more artifacts than chromosomes isolated with other procedures. The structures described are something real, they just have to be recognized. What they represent, is still not definitely established but it is not far-fetched to claim that they represent cyclomeres, the nucleoplasmic parts of the NPCs, as originally suggested from thin section studies (Engelhardt and Pusa, 1972a, b, 1974a, b, 1975, 1977, 1978; Pusa and Engelhardt, 1977; Engelhardt, 1976, 1979).
It is about a hundred years since the chromomeres and the helical coiling of chromosomes were first described (reviewed by Haapala, 1985; Nokkala, 1985; Sorsa, 1986, 1988). Yet no definitive explanation of their structures has been achieved. The reason is that there is as yet no unifying chromosome model from which would automatically follow all the chromosome features from the molecular DNA-nucleosome-string level to the gross-structural level of the chromatid.
New efforts to explain these aspects of chromosome coiling have followed from the finding of coiling in some electron microscopic whole-mount preparations (Golomb and Bahr, 1974; Harrison, Britch, Allen and Harris, 1981; Haapala and Nokkala, 1982; Mullinger and Johnson, 1983; Rattner and Lin, 1985; Taniguchi and Takayama, 1986, 1987). A number of suggestions have been raised to solve the problem (Haapala and Nokkala, 1982; Haapala, 1983, 1984a, b, 1985; Nokkala, 1985; Rattner and Lin, 1985; Sorsa 1986; Taniguchi and Takayama, 1986, 1987). The main issue is that none of the many structural models presented earlier (e.g. DuPraw, 1965, 1970; Comings, 1977; Bak, Zeuthen and Crick, 1977; Bak and Zeuthen, 1978; Sedat and Manuelidis, 1978; Marsden and Laemmli, 1979; Adolph and Kreisman, 1983; Pienta and Coffey, 1984) give a satisfactory explanation of the appearance of chromatid coiling.
It is unnecessary here to go into the details and difficulties of earlier chromosome models as these have been thoroughly discussed recently (Haapala and Nokkala, 1982; Haapala, 1983, 1984 a, b, c, 1985; Nokkala, 1985; Sorsa 1986). Only some relevant suggestions and models will be dealt with here. Now, many models (e.g. Bak, Zeuthen and Crick, 1977; Bak and Zeuthen, 1978; Sedat and Manuelidis, 1978) seem invalid simply because they introduce a "total coiling hierarchy of DNA packing", which according to Haapala (1983) does not agree with known ultrastructural details. Such models also disagree with measurements and calculations (Haapala and Nokkala, 1982), which show that only a fraction of "true DNA" fits into a chromatid in such "space-filling models." Some recent models (Taniguchi and Takayama, 1986, 1987) introduced mainly to explain chromatid coiling and banding seem to suffer from the same difficulties.
On the basis on some controversial staining results of whole-mounted chromosomes, Haapala and Nokkala (1982) have suggested an axial core (filament) in chromatid macrocoils that would be asymmetrically localized. The contraction and coiling and unilateral localization of this axial filament in the chromatid would cause bending of the chromatid, i.e. (macro)coiling with chromomeric loops protruding radially outwards. The contraction of the axial filament is thought to be a condensation of an inter-loop (inter-chromomere) nucleosome string into the next-level solenoid and to be stabilized by scaffolding proteins. The asymmetrically situated axial core in mitotic chromosomes would correspond to the lateral elements in the synaptonemal complex in meiotic chromosomes (Haapala and Nokkala, 1982; Nokkala, 1985). According to the authors their model excludes neither the radial loop model (Marsden and Laemmli, 1979; Adolph and Kreisman, 1983; Pienta and Coffey, 1984) nor an earlier model in which chromomere loops were proposed to be clustered (Comings, 1977) to form the basis of chromosome banding. Their model essentially states that the chromatid "macro" coiling, which earlier models have totally missed, is caused by the unilateral positioning of the chromatid axial core.
Based on the suggestion of Haapala and Nokkala (1982) on the asymmetric localization of the chromatid core, Sorsa (1986) has presented a model concerning this asymmetry and the macrocoiling. According to this model, no asymmetry would exist if all the chromomere loops were of equal size, a point which seems to have been ignored by Haapala and Nokkala (1982). When this kind of chromonema (axial filament) is coiling (microcoiling; for the terms, see Nokkala, 1985), no asymmetry and therefore no bending into macrocoils would occur. Sorsa (1986) finds a periodicity in the distribution of prominent bands in LM and EM maps of salivary gland chromosomes in Drosophila melanogaster (and other Diptera and even some Collembola and Protozoa ; Sorsa pers. comm.; Sorsa, 1988), which according to his concept would orient the largest chromomere loops mainly on one lateral half and the smallest chromomere loops on the opposite side of the axial core, at the chromonema stage, making it to bend and form the chromosomal "macro" coil in metaphase chromosomes.
Some specific details of the previous viewpoints (Haapala and Nokkala, 1982; Sorsa, 1986) will be considered in relation to the loop-and-rosette model. As regards the unilateral axial core, in the first place, the existence of the axial core has been clearly demonstrated, but its asymmetric localization has not. It is based on rather less convincing staining properties of whole-mounted chromosomes (Haapala and Nokkala, 1982). But accepting the suggestion for the sake of the argument, it does give one reason why the chromatid bends into macrocoils as the axial core contracts (microcoiling), orienting radially the chromomere loops outwards. However, Haapala and Nokkala (1982) have not given a structural reason for the unilateral localization of the axial core in the chromatid or for the asymmetric orientation of the radially oriented chromomeric loops at the chromonema stage. To these Sorsa (1986) finds an explanation in the periodicity of bands as mentioned above. But it is somewhat hard to believe that the distribution of small and prominent bands (chromomeres) would follow a very strictly ordered lengthwise distribution; one would expect them to be somewhat irregular.
Some insufficiencies can thus be found in both these models (Haapala and Nokkala, 1982; Sorsa, 1986). They do stress relevant directions in the question of chromatid coiling. Discrepancies should be replaced by more natural causes, such as will be suggested in the following.
The loop-and-rosette model does not attempt to explain in a straightforward manner the chromatid coiling, but in the first instance simply the higher-order DNA folding, beyond the nucleosome fibre. Yet, from this single folding principle, a natural explanation easily follows for a variety of situations: for size differences in chromomeres; for the axial fibre (core); for the asymmetric distribution of chromomeres in coiling, or the unilateral (inner side) localization of the axial core in the chromatid (macro)coil. Moreover there is no need for new assumptions to arrive at these consequences.
The origin of the different-sized chromomeres, amplified by local duplication from the single loops during evolution, and their bases or anchoring sites, as derived from NE connections, has been discussed (see Discussion I. 8. The loop-and-rosette model). It also follows that the axial fibre is the interchromomeric strand connecting the loop bases or anchoring sites of the different-sized chromomeres in the model (Figs. 36, 37).
In the first place, differences from the radial loop model (Marsden and Laemmli, 1979) must be pointed out: the radial loop model does not give any explanation to the (evolutionary) origins to the loop bases or anchorage sites and to the size differences between the loops. The chromomere loop model (Comings, 1977), on the other hand, has presented the clustering of loops by anchoring sites in the matrix-scaffold as an explanation to the existence and size variation of bands (chromomeres and their clusters). These models come closest to our loop-and-rosette model, from which they may have originated (Engelhardt and Pusa, 1972a, 1974b, 1977, 1978; Engelhardt, 1979; Comings, 1977 pers. comm.).6
6 Personal discussion at the Helsinki Chromosome Conference, l977, with Comings, who had received a drawing of the model, description and review (Engelhardt and Pusa, l974a, b). A great number of authorities have received this material during the years. The drawing was sent to Laemmli and coworkers in l977.
In both of these models (Comings, 1977; Marsden and Laemmli, 1979) the loops and their clusters can freely rotate around the axis with no preferential localization on any side, and thus with no unilateral positioning. However, the loop-and-rosette model clearly defines that the core of the rosette type chromomere is the cyclomere and the folding principle places the entry and exit point of interchromomeric strand asymmetrically on the cyclomere periphery (see the details of the folding principle in Figs. 35 A, 36 B). This is analogous to the bases of the loops in the radial loop model. Essentially, the cyclomere imposes a higher-order configuration on the chromatin fibre and gives the resulting rosette further potentials of interaction. During condensation this configuration will fix the freedom of rotation around the axial fibre. The cores of the rosette, i.e. the cyclomeres, will interact with each other or assemble, by analogy with NPC formation, at the NE. This characteristic feature is not an extra assumption, added only to explain the chromatid coiling, but an inherent consequence from the origin (NPC) of the core, due to specific interacting and anchoring proteins. It is not difficult to imagine that the axial fibre, in condensation, will have a unilateral (inner side) position in the chromatid (macro)coil.
Accordingly, macrocoiling is not a consequence of a real microcoiling, but a succession and assembly of ring-like cores, with their condensed or protruding chromatin loops (see below). Apparently, this kind of configuration can easily and erroneously be taken for real coiling, at least at LM level and even at EM level, if careful stereoscopic inspections are not done. Still, even this could be claimed to be a question of interpretation if a network of pore or ring-like structures did not remain as the scaffold, as has been shown, after extraction and digestion.
Theoretically, real coiling along the chromatid could be the consequence if shorter or longer stretches of the chromatid were occupied by chromomere-loops and their multiples that have not evolved around cyclomeres. In these places a real (micro)coiling could be the consequence. The alternative would be that single monomeric loops would cluster and form a ring-like chromomere (cyclomere) and groups of chromomeres.
There is the other possibility that the cores of ring-like chromomeres (cyclomeres) fuse into a continuous true microcoiling, and consequently no macrocoiling would follow. Depending on the cell division stage, mitotic or first or second meiotic metaphase, macro and microcoiling can be observed interchanging in the same species (for review see Nokkala, 1985).
The sizes of the chromomeres can vary in the same chromosome, between chromosomes and between chromosomes of different organisms, depending on the amount of local chromomeric amplification (successive duplications) of DNA loops in evolution. It follows that in big chromosomes (e.g. in some plants) with lots of DNA and with typical chromomeres, more typical chromatid coiling can be found than in certain small, holocentric chromosomes (for references, see Nokkala, 1985), where an evolution to a cyclomere-type configuration of chromomeres may not have evolved.
Light microscope observations of minor coiling inside major coils imply a coiling hierarchy. Yet hierarchial coiling models need not be considered until the coiling-within-coiling is demonstrated also at the EM level. Hierarchical coiling is by no means incompatible with the present model. The conclusion is warranted that chromatid (micro)coiling is not a coiling at all but a stacking of the (ring-like) cores, i.e. cyclomeres, of different-sized chromomeres and their assemblies, to which the folded chromatin loops, the rosette loops, are anchored. The chromatid will totally or only locally be twisted into a (macro)coil by the condensation of the peripherally located intrerchromomere strand. According to this view the "axial core" is not the interchromomere strand but rather the result of an assembly of the anchoring particles on this strand. This axial core and the individual component cores (cyclomeres, only about 0.1 mm in diameter) are beyond light microscopic resolution when the preparations are properly preserved and fixed.
As a consequence of the present view, different chromosome banding techniques (reviewed, e.g. in Goyanes, 1985) can be understood and explained as differential decondensation effects, affecting either chromatin fibres or the stacks of (chromomere) cores. Then either the protruding chromatin fibres or the alternating stacks of chromomere cores of the "macro"coils will represent bands or interbands, depending on the technique. Also, the variable behavior of chromosomes in banding techniques gives more strength to the loop-and-rosette model than to chromosome models that recognize only a coiling of a uniform axial core.
It would seem that a number of behaviour patterns of the chromosomes can be derived from the basic characteristics of the present chromosome model. This is not mere speculation, because many of the basic assumptions themselves have been shown to be valid, by whole mounting and other methods.
Computer reconstructions of chromosomes and nuclei from whole mounts and serial sections are likely to contribute to the solving of many of these questions (Engelhardt and Mäntylä, in progress). That recent computer reconstructions (Harauz, Borland, Bahr, Zeitler, and van Heel, 1987; Belmont, Sedat, and Agard, 1987) have yielded fairly little is probably caused by transitory problems of specimen preparation and the difficulties in their interpretation.
The contraction of the loops from the 10-nm thick DNA-nucleosome string into the 30-nm thick chromatin fibres, in the loop-and-rosette chromomeres, needs further elucidation with reference to metaphase chromosomes (see the chapter on The loop-and-rosette model), although this level of processes is beyond the scope of the present investigation. It is generally attributed solely to histones. The problem has not been solved and numerous models have been presented, which can roughly be classified as: (1) the original solenoid model, (2) some kind of a helical coil of the nucleosome fibre, and (3) a beaded (superbead, supranucleosome, nucleomer) arrangement of the nucleosome subunits (for references see Walker and Sikorska, 1987a, b).
The confusion is obvious from descriptions of the various models: "a zig-zag pattern of nucleosomes; an interdigitised foldback of the nucleosome zig-zag; a regular distorted helix of planar orientated zig-zag monofilaments; a solenoid coil; a dis-orderly worm-like coil; a double helical crossed linker coil; and the superbead structure" (for references see Drinkwater, Wilson, Skinner, and Burgoyne, 1987). The only consensus seems to be in the first order of packing: "beads on a string" or DNA wound into a polynucleosomal fibre. But beyond this order, the confused situation regarding the condensation to the 30-nm chromatin fibre is in principle the same as that regarding the gross structural metaphase chromosome.
The first question is the length and possible dynamical changes of the DNA-nucleosome loops interacting with the scaffold elements. According to the data from polytene chromosomes (Results I), which could represent the situation in some interphase cells, the loops would be short, not more than a few (1.2-2.2) kb, and harbour not more than a few (6-11) nucleosomes, enough only to form one superbead or one supranucleosome (as discussed under The loop-and-rosette model; this despite the fact that the "superbeads" have largely been disregarded, see Drinkwater, Wilson, Skinner, and Burgoyne, 1987). Little new information on the loop length could be obtained from the acid-isolated metaphase chromosomes, but they seemed to have similar-sized loops (Results II: Fig. 5 f). In any case, it can be estimated from spreading data on rosettes that they fall in the same size range in metaphase chromosomes (Okada and Comings, 1979, 1980).
If topoisomerase II is the prominent master in the higher-order folding machinery in the packing of DNA, short loops should be the result. The active interphase chromatin (euchromatin) may undergo dynamic changes, but even here the topoisomerases (I and II) have recently been suggested as the driving force behind transcription (Liu and Wang, 1987).
In sections of meiotic prophase cells in Tradescantia, it can be shown that besides cyclomeres, in all their detachment steps from the NE (Pusa and Engelhardt, 1977; Engelhardt and Plagens, 1984), the prominent structures seen are in fact superbeads of chromatin (Engelhardt, unpublished). These results were based on improved fixation with ruthenium red techniques and also stereoscopic methods in the inspection of sections.
The reason why the real configuration in the 30-nm chromatin fibre has been difficult to study is to be found in its dynamic changes during the cell cycle and in the fact that "chromatin" is biochemically a poorly defined entity varying according to isolation sources and conditions. Chromatin fibres isolated from their natural environment, the interphase matrix or the metaphase chromosome scaffold, will lose their original configuration and must be considered artifacts as soon as they are isolated, i.e. torn apart. Thus the contradictory results and the different opinions have arisen.
If the ultimate DNA-nucleosome fibre that folds up were only 1.2-2.2 kb long, it could not harbour more than 6-11 nucleosomes. A recent report on calf liver gives precisely seven nucleosomes per satellite DNA repeat of 1.4 kb (Pagès and Roizès, 1988). It follows that a superbead arrangement of chromatin fibres around the scaffold units would be the only possibility in metaphase chromosome packing also, just because of the length of the loop. This choice is supported by the irregular-shaped or "knobby" 30-nm chromatin fibres in electron microscopic images, by the presence of peaks in sucrose density gradient profiles of nuclease digests of chromatin (references in Walker and Sikorska, 1987a, b), and by the description of "microconvules" (superbeads?) in a high-resolution scanning EM study of metaphase chromosomes (Daskal, Mace, Wray, and Busch, 1976).
According to the present results, it was difficult to distinguish scaffold elements and chromatin fibre loops in Pt-shadowed preparations in undigested critical point-dried whole-mount preparations of acid-isolated metaphase chromosomes. Another curiosity was that after extensive acid, salt dehistonization and nuclease digestions the gross structure of spread polytene chromosomes still looked remarkably intact, though they were depleted of all DNA (shown by the absence of DAPI fluorescence). The sponge-like network of the scaffold fibres (cables) and of the chromatin fibres, separately inspected, are practically indistinguishable. Both fibres are in the range of 30-nm thickness. As shown in the results (Figs. 27, 30), the scaffold "cables" are composed of globular scaffold particles (25-30 nm average diameter, total range 20-45 nm) but they are depleted of histones and DNA. Superbeads of chromatin fibres are of the same size, but on the contrary they are considered to be composed solely of histones and DNA (6-11 nucleosomes, or for 1.4 kb DNA precisely 7 nucleosomes, Pagès and Roizès, 1988).
The number of DNA loops in a ring-like chromomere could be roughly estimated as 8-20 in well spread rosettes (Results I, Fig. 34); the theoretical value would be 8 loops for a rosette chromomere, according to the model (Figs. 36, 37). The rosettes found could be two superimposed cyclomeres with a total of 16 loops. When a single rosette-like chromomere would be fully condensed there would be a ring of 8 globular scaffold elements concentrically surrounded or superimposed by 8 superbeads (shown in the model of the polytene chromosome, Figs. 36 C, D). This could also be the final condensation form of the rosette-chromomeres in the metaphase chromosomes.
The scaffold elements have a dual appearance, either globular or ring-like (globosomes or cyclosomes), and very flat when air dried (results I). The flattening and dual appearance could be due to the fact that when intact they are empty spheres with a hole. Depending on which way the hole is oriented, either up or down, they would appear either as ring-like or globular.
There is a final possibility of combining chromatin and scaffold elements and their duality. As the globular scaffold element is approximately of the size of the superbead, it is tempting to consider that a scaffold element, when empty, could take a (DNA-nucleosome) superbead into its interior. This virus-like packing would also agree with the size variation (20-45 nm) of the globular elements (cf. chromosome pellicle, Shi et al., 1987; Fields and Shaper, 1988). Lack of specific details, such as charge distributions, does not permit a decision for or against this possibility.
A more plausible expectation would be that, in some chromomeres at least, the chromatin loops are very long, as in fact suggested in the radial loop model: they would somehow condense into 30-nm fibres and would by their sheer mass effectively hide the ring-like core elements. Another difficulty in distinguishing these core elements would arise from the common thickness of loops and cores. In Pt-shadowed critical point-dried metaphase chromosomes it was indeed difficult to differentiate between core elements from protruding chromatin fibres. It is to be noted that acid-isolated, dehistonized chromosomes must represent an "unpacking" state and that the questions of loop size and packing cannot be fully settled until truly intact chromosomes can be studied.
The matrix-scaffold was indicated in the early work of Mayer and Gulick (1942) and in the classical demonstration of residual protein (Mirsky, 1947). The relationship of this material to the NE was also shown (Zbarsky, 1972, and earlier references there; Harlow, Tolstoshev and Wells, 1972).
In structural and comparative terms the matrix-scaffold was anticipated also in our earlier thin section studies (Engelhardt and Pusa, 1972a, b), as a system of nonhistone-based structures. These consist of the NPCs in the NE and their detachable nucleoplasmic part called the cyclomere, the cyclomere-forming or independent subunits which were termed cyclosomes (Engelhardt and Pusa, 1978), and "striated microfilaments" derived from the "internal dense lamella." They all are elements of the synaptonemal complex (Engelhardt and Pusa, 1972a, b). Some of these relationships in the formation of the synaptonemal complex were independently confirmed (Fiil and Moens, 1973; Moens, 1973; Fiil, 1976a, b).
Earlier works with spreading techniques had shown that annular structures are connected with chromatin fibres (DuPraw, 1965; Comings and Okada, 1970; Lampert, 1971). NPC-like structures of chromatin, if real, are difficult to show with spreading methods which cannot distinguish between natural detachment (here from the NE) and artificial disrupture by spreading forces. Yet in thin sections with ruthenium red we could establish that such ring-like structures are truly embedded in chromatin (Engelhardt and Pusa, 1972a, b). It was suggested that the ring-like structure (called a cyclomere) is the nucleoplasmic half of the NPC, which would be bipartite (the cytoplasmic part functioning as the receptor half, Engelhardt and Pusa, 1972a, b; for review see Maul, 1977). This is presented in a detailed model of the NPC in Fig. 35 (reprinted from Engelhardt and Pusa, 1974a, Engelhardt, 1976). That the NPCs can actually be split into two halves has since been spectacularly demonstrated (Schatten and Thoman, 1978).
The presence of rosettes and ring-like structures in chromatin has been observed also in spread alkali-urea treated and thin-sectioned squash preparations of polytene chromosomes (Sorsa, 1973, 1974, and review in 1988). In telophase, in thin sections, such rings participate in the reconstruction of the NE (Engelhardt and Pusa, 1975; Maul, 1977). Cyclomeres occur in various kinds of chromatin (Engelhardt and Pusa, 1972a, 1974a, b, 1977, 1978; Pusa and Engelhardt, 1977; Engelhardt, 1976, 1979). Rosette chromomeres have been described later in sections of hypotonically treated interphase, metaphase and polytene chromosomes and in spreadings of nuclei with condensed chromatin (Zatsepina, Polyakov and Chentsov, 1983; Prusov, Polyakov, Zatsepina, Chentsov and Fais, 1983; Zatsepina, Polyakov, Lozovskaya and Chentsov, 1983 ).
Astonishing new results (Sheehan, Mills, Sleeman, Laskey, and Blow, 1988) have now been obtained using cell-free systems which show that in the reconstruction of demembranated sperm chromatin with two fractions from Xenopus eggs (both a soluble and a washed vesicular fraction is needed), a NE is formed with nuclear pore complexes. Mixing various proportions of the fractions allowed the observation, in sections, of immature nuclear pores termed "prepores," in incomplete single membranes surrounding, or "free" within, the chromatin. These pore-like structures are present throughout the chromatin mass. An interesting observation is also that these prepores reach the right size fully only when united to the outer membrane and forming normal pore complexes (cf. Results I, Figs. 7, 29, 30, 33, where there is a diversity in size of the ring-like oligomers). Only after a complete reassembly of the nuclei (including finished nuclear pore complexes) an initiation of DNA replication, without lag, can be observed in vitro.
These structures were originally described by Engelhardt and Pusa (1972a, b) in sections of different chromatin in many different species, and were named "cyclomeres", and suggested to be involved in the regulation of replication and transcription. The authors only deal with spreading data that cannot distinguish between remnants of pore complexes artificially torn from the NE by the spreading forces as discussed and those naturally embedded in the chromatin. Spreading data do confirm that fibres which are indistinguishable from chromatin are clearly and firmly connected with the remnants of the pore complex annuli. In these spreading preparations no indications can be found that chromatin fibres could be connected to the lamina, i.e. the interpore region, that is generally thought to mediate the anchorage of chromatin fibres to the NE (Aaronson and Blobel, 1975; Gerace, 1986).
Having these demonstrations of chromatin "annuli" it may be surprising that with few exceptions (e.g. in Physarum, Wille and Steffens, 1979; in rat liver, Kuzmina, Buldyaeva, Troitskaya and Zbarsky, 1981; in Drosophila, Fisher, Berrios and Blobel, 1982) annular structures have not been described in the matrix except those clearly seen in the periphery as the NPC-lamina (Berezney and Coffey, 1977). However, rosette-like structures could be detected in thin sections of high-salt-resistant (2 M NaCl) calf liver nuclear residue after limited DNase digestion (Wanka, Pieck, Bekers and Mullenders, 1982; Bekers, 1982), and in spread fragments of the same material DNA threads emerge from residual nuclear pores and, interestingly, also from "fibrilloglobular" material (Bekers, 1982, his Fig. 9, p. 58).
Yet study of some published photographs will show that generally the appearance within the matrix looks in no way different from the NPC-lamina (Comings and Okada, 1976, their Fig. 2; Berezney and Coffey, 1977, their Fig. 11). In many places truly annular structures (cyclomeres) can be seen in these photographs, and for instance, also in sections of metaphase chromosomes (Marsden and Laemmli, 1979, their Figs. 1, 2, 3; Adolph, 1981, his Figs. 1 a-f, 2 a-d, 5 a, b, f-h.) In some pictures they were described as amorphic material (Comings and Okada, op. cit., Figs. 4 e, 5 e). The impression of their absence may be conditioned by the conventional location of pore complexes at the NE. According to earlier protein analyses, it may be noted, nuclear protein matrix (Berezney and Coffey, 1974, 1976) and the isolated NPC-lamina (Aaronson and Blobel, 1975) had nearly identical band composition on SDS-PAGE.
Hyaluronidase treatment and the mica technique may help dispel the confusion about the nuclear materials, as they pinpoint the common, basic units of the chromosome scaffold and nuclear envelope scaffold and unravel their relation to DNA.
The "DNA membrane complex" or "the origin/terminus region of replication" in a colicinogenic plasmid (Sparks and Helinski, 1979, see their Fig. 2 d) is similar in size and appearance to the ring assembly of scaffold elements. The replication complex may be fundamentally the same in plasmids and in eukaryotes. At least two of the residual chromosomal proteins comigrate with the DNA-bound protein of E. coli (Werner and Petzelt, 1981) and it is therefore probable that the structural proteins involved in folding DNA into loops are phylogenetically older and even more universal than the proteins needed for the further compaction of these loops, i.e. histones. The characterization of such scaffold elements may also help to locate the as yet mysterious site of DNA (RNA)-membrane interaction (Moyer, 1979).
The implications of the similarity between NE elements and matrix-scaffold elements have finally to be considered. The ring-like assemblies in the matrix-scaffold and their globular elements are indistinguishable from NPCs and their subunits. This suggests a common origin. It was presumed in the original working hypothesis that in eukaryotes the initiation sites of DNA replication, i.e. anchoring elements, would reside at the NE, by analogy with prokaryotes (Engelhardt and Pusa, 1972a, b). It was also claimed that carbohydrate moieties contributed to the anchoring elements, which motivated the use of specific stains such as ruthenium red in their visualization at the NE. Moreover, in the pairing of the chromosomes at meiosis, these elements would detach and associate with the formation of the synaptonemal complex. To our surprise, NPCs and their subunits and also the "chromatin connecting fibres," were found to be distinctly stained by the ruthenium red methods at the NE in interphase. Similar ring-like structures were found embedded inside the condensed premeiotic X chromosome of Acheta, the corresponding evagination of the NE being depleted of NPCs. In the lateral elements of the synaptonemal complex in Acheta and Lilium ring-like and arch-like elements, and granular or ring-like subunits could be found in accordance with the NPC and its components. In addition, the lateral elements contained a lining of "striated microfilaments." All these elements were independently described as involved in the formation of multiple synaptonemal complexes (Fiil and Moens, 1973). The "striated microfilaments" presumably correspond to the lamina part of the NPC-lamina fraction (Aaronson and Blobel, 1975). Correspondingly, the NEs of meiotic cells have been shown to be devoid of a nuclear lamina detectable by an immunofluorescence and EM study (Stick and Schwartz, 1982, 1983).
Presumably the original prokaryotic DNA-membrane interaction sites have in eukaryotes evolved to higher-order clusters like the bipartite NPCs, with a detachable nucleoplasmic moiety, to be able to control and fullfil demands of the eukaryotic nucleus, where the surface of the NE is quite small in relation to the enormous amount of DNA. It is interesting to note that the attachment sites in prokaryotes have been shown only recently to be temporary and very delicate, being controlled by "hemimethylation" of DNA (Ogden, Pratt, and Schaechter, 1988). In view of this, it is not surprising to find that the interphase matrix and the scaffold of metaphase chromosomes are in all details indistinguishable from the NPCs, especially the nucleoplasmic part. An essentially similar though more thorough discussion of the origin of the matrix-scaffold has lately appeared (Cavalier-Smith, 1982). Nuclear matrix is gaining increased support for being the "actual milieu" where DNA replication takes place (for references see Tsutsui, Tsutsui and Muller, 1988). The cytoplasmic part would then be the receptor site in the control not only of the synchronization of replication, as has been recently demonstrated (Sheehan, Mills, Sleeman, Laskey, and Blow, 1988), but more generally, the NPCs are "press stud elements of chromosomes in pairing and control" (Engelhardt and Pusa, 1972a). As regards the control aspects a similar view, presented as "gene gating," has been been developed as recently as 1985 by Blobel.
Eukaryotic chromosome structure [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11]