Archives of insect Biochemistry and Physiology 143-155 (1986) Into The Lair of the Gene Geoff Richards Laboratoire de G6n6tique Moliculaire des Eucaryotes du CNRS, et Unit6 184 de Biologie Molkculaire et de G6nie Ginetique de I’INSERM, Institut de Chimie Biologique, Faculty of Medecine, Strasbourg A new picture of gene and chromosome structure is emerging in Drosophila that differs considerably from that largely derived from polytene chromosome banding patterns and saturation mutagenesis. In particular, gene transfer has enabled us to more clearly limit the functional unit of a number of genes. Gene regulation may be studied at the molecular level with such techniques. The possible complexity of regulatory elements that may pose problems in their analysis is discussed. Key words: Drosophila, P elements, regulatory sequences, polytene chromosomes INTRODUCTION ”With these ... discoveries before us ... it was clear that we had within our grasp the material of which every one had been dreaming. We found ourselves out of the woods and upon a plainly marked highway with by-paths stretching in every direction. It was clear that the highway led to the lair of the gene.”T.S. Painter, Salivary chromosomes and the attack on the gene. J Hered 25:465 (1934). In his early review of Drosophila melanogaster polytene chromosome studies, Painter [l] summarized the evidence that the giant banded salivary gland chromosomes provided a physical counterpart to the genetic maps derived from meiotic recombination distances obtained in breeding experiments. Although the nature of the genetic material was unknown at that time, and was to remain so for ten more years, it was a clear vindication of the working model of ”beads on a string” in which genes were considered as discrete units joined together in a linear array by linker material to form chromosomes. Although Painter was careful to conclude that genes might correspond to a band or part of a band, the combined results of Alikhanian , who calculated that the Drosophila X chromosome contained 968 genes capable of mutating to a lethal state, and Bridges , who showed that the same chromosome contained 1,024 polytene bands, were largely responsible for the adoption of the one band-one gene hypothesis. For nearly 40 years the Address reprint requests to Dr. C. Richards, LCME du CNRS, 11 rue Hurnann, 67085 Strasbourg Cedex, France. 0 1986 Alan R. Liss, Inc. 144 Richards majority of experimental evidence was interpreted in the light of this hypothesis, which was extended by polytene chromosome puffing studies that showed at least in some cases, either in development or following experimental treatments, that individual bands decondensed to form puffs that were sites of intense RNA synthesis. This led to a simple model of one gene-one band-one puff in which an active gene gives rise to a puff. In cases in which large puffs involving several bands were formed, it was considered likely either that adjacent bands contained genes with similar developmental regulation or that a single band decondensed actively and that its decondensation spread passively into neighboring bands. Later, following estimations of the potential informational content of the average chromosome band (with 20-30 kb* of DNA), a modification of this concept was proposed in which the band or chromomere remained the functional unit but might contain related genes in an arrangement similar to the prokaryotic cistron . In this review I will consider the genetic and molecular studies, including gene transfer, of the past few years that demand a complete reappraisal of our views of Drosophila gene organization and discuss some features of regulation that may complicate our final analysis of the lair of the gene. PUFFS, BANDS, AND GENES; THE CYTOGENETIC APPROACH If the alternating dark and light bands of the polytene chromosomes were really genes and their spacers and these genes were the same in all cell types of an organism, then it followed that banding patterns should remain constant. The essential constancy of banding patterns in different Dipteran tissues was demonstrated by Beermann, Pavan and Breuer, and Mechelke (see Richards  for references), and the notion that their genetic content remained constant in development derived support from the totipotency experiments of Gurdon using nuclear transplants in Xenopus . While there was thus compelling circumstantial evidence supporting the hypothesis, a number of heretical views had already been expressed in recent years from a purely cytogenetic approach. The most critical reappraisal of the significance of the chromosomal banding patterns was that of Zhimulev and Belyaeva and their colleagues in Novosibirsk [7,8]. The original alternative to the constancy of the banding pattern was the ”Istanbul hypothesis” of Sengun 191, who proposed that banding patterns reflected the condensation of inac- *Abbreviations: Adh = alcohol dehydrogenase; B helix = B-form of DNA having a righthanded double helix with about 10 bp per complete turn; bp = base pairs of DNA; HSE = heat shock regulatory sequence elements; hsp-lacZ = heat shock puff 70-linked to the @galactosidase structural gene; kb = kilobase; kd = kilodalton; P element = a type of transposable element of Drosophila melanogaster characterized by frequent insertion and deletion in flies with cytotype M; ry = rosy eye color of Drosophila melanogaster; sgs3 = salivary gland secretion protein 3; TATA box = an adeninekhymine-rich sequence located about 25 base pairs before the first nucleotide of a transcriptional unit; YP1,2,3 = yolk proteins; Z DNA = form of DNA with the left-handed helix with 12 bp per turn; -100 bp = negative sign indicates the 5’ direction , upstream, from the first nucleotide transcribed; + I 6 bp = positive sign indicates the 3’ or downstream direction from the site of initiation of transcription. Into the Lair of the Gene 145 tive chromatin and that tissue-specific gene activity, or potential gene activity, would lead to tissue differences in banding pattern. The Novosibirsk group have reformulated this hypothesis, adding their considerable experimental results, by suggesting that interbands contain "housekeeping" genes that are active in most tissues. This gives a basic banding pattern common to all tissues that is modified by decondensations related to tissue-specific gene activity. These modifications may occur before the gene is active (puffed) and will reflect a tissue-specific chromatin structure determined during differentiation of the cell type. Regions containing genes that will not be expressed in this cell type may form bands containing one or many inactive genes. Technical limitations preclude a wholly satisfactory conclusion that all interbands are sites of active mRNA synthesis; however, there are small but consistent differences in banding patterns in Drosophilu tissues [10,11] and even more dramatic differences in banding patterns were found by Ribbert , studying secondary polytene chromosomes of ovarian nurse cells and normal polytene chromosomes of trichogen cells of Culliphora erythrocephalu. That bands can contain more than one gene is best illustrated by the genetic and electron microscope studies of chromosome band 10A1.2, which contains at least three distinct genes that can be physically separated by chromosomal rearrangements and yet function independently in their new positions . A different experimental approach, also important for the problem of gene content in bands, is that undertaken by Keppy and Welshons , who synthesized new compound (multifunctional) bands in Drosophilu by X-ray deletions of interband material. Combining their genetic analysis with early biochemical analyses of transcription in Drosophila, Zhimulev and Belyaeva [7l suggested in 1975 that 20,000-30,000 was a more realistic minimum number of genes in Drosophila than the 5,000 derived from band counts. They have since reviewed many of the saturation mutation studies whose results did not differ appreciably from the one band-one gene hypothesis . In agreement with Young and Judd  they suggest that problems derive both from inadequate cytology of the region studied and from the failure of the majority of screens to detect mutations other than lethals or clear visibles. That many characterized genes have viable null alleles suggests that this class escaped detection and that previous results should be reconsidered as showing an approximate correlation of one essential gene per band. This, however, differs significantly from the original hypothesis of a strict one-to-one correspondence of bands to genes. Using the electron microscope, the Novosibirsk group have made important advances in studies of puffing [16,17l. They have recently reported examples both of the decondensation of part of a band and the simultaneous decondensation of adjacent bands to form a large puff, in addition to many more traditional cases where either a single band undergoes uniform decondensation or adjacent bands appear to be drawn passively into a puff as the principal band decondenses. The former exceptions clearly support their notion that the chromosome band is not a genetic unit whose expression is simply dependent upon its decondensation in puff formation. One refinement of the basic model that we should bear in mind in later analyses is the 146 Richards two-step model of puffing, suggested by the large number of ecdysoneinducible puffs found in more than one tissue [18-201. In this model it is envisaged that the hormone acts first to cause the decondensation of a chromosome region into a puff and that a second level of regulation, perhaps also hormonally dependent, is necessary to select the particular sequence to be transcribed in a given tissue [20,21]. BANDS AND GENES: THE MOLECULAR APPROACH The one band-one gene hypothesis was born in a period when the genome was considered highly stable. Indeed the fact that polytene chromosome banding patterns could be used in phylogeny studies of Drosophilu groups was itself strong evidence for such a stability. The earliest molecular cloning studies in Drosophila were to upset this view as the middle repeat sequences, such as copiu, not only proved to be present at multiple chromosome sites, but also were highly polymorphic in their sites both within and between populations. Other genes coding for abundant products such as the histones or tRNAs were also immediate exceptions to the one band-one gene hypothesis, the former being present in tandem repeats covering several chromosome bands, whereas the latter were in multiple mixed clusters in extremely short lengths of DNA (see Richards  and Spradling and Rubin,  for reviews). As to unique genes, studies of Dmsophiln gene structure have so far revealed two general classes of transcription units: 1)small, compact genes with short introns (e.g., those for alcohol dehydrogenase , larval serum proteins , dopa decarboxylase 1251, yolk proteins , larval cuticle proteins [ 2 7 , and salivary gland glue proteins [28,29]) and 2) complex transcription units with short coding sequences and large introns that may extend over 50-100 kb of genomic sequence (e.g., ultrubithorax  and untennupediu ).The genes of the first class are typical protein-coding genes where the coding sequence and introns span no more than 2-3 kb of DNA. Much larger introns have been found in many vertebrates, and initially we might have expected that Dvosophilu genes with large introns would account for the 2030 kb of a chromosome band. This is not the case for these genes, although some of them are found in clusters that might suggest coordinate regulation. However one such cluster, that of the four heat shock genes at 67 B, was shown to be interspersed by three other transcription units that are not induced by heat shock. In this case, there are seven transcription units in about 14 kb of DNA, all of which are taken into the 67 B heat shock puff, which forms at most from two bands and their neighboring interbands . Similar high densities of transcripts have been detected following chromosomal walks in Drosophilu (e.g., Bossy et al. ). If there are also regions where lower densities of transcripts have been detected, we cannot exclude that this is a technical problem of detecting low-abundance transcripts, including transcripts restricted to a few highly specialized cells. The evidence so far is that the figure of 20,000-30,000 genes proposed from the combination of genetic and message complexity studies will be justified by transcript analyses. There are two further points to be made regarding such analyses: Into the Lair of the Gene 147 One is the detection of alternative splicing to produce different messengers from a single region (e.g., Adh, ), and the other is evidence from a number of studies in progress that there may be genes within genes in DrosophiZu, i.e., the exons of one gene may be contained within an intron of another. Such results are contrary to the "beads on a string" model. In some respects this model needed modification once it was proposed that bands contained coordinately regulated genes as that implied that each gene was not self-contained but could depend on more distal sequences regulating the coordinated cluster. As we shall see in the next section, the coordinated cluster model is not necessary to explain results from gene transfer experiments. We should not, however, eliminate the possibility that elements regulating a gene may be found beyond, or within, neighboring transcription units. GENES: THE GENE TRANSFER APPROACH To test gene function, it is important to be able to reintroduce cloned genes into cells. Two systems are currently available: permanent cell lines or transgenic animals. Cell line transformation has certain disadvantages for the study of developmentally specific gene regulation. One of the most difficult aspects is controlling the number of sequences entering a cell, which may be as many as several thousand. Another is that we do not always have cell lines equivalent to those cells in which the gene is normally expressed. Even when these are available we cannot be sure that the incoming DNA will be packaged in a conformation identical to the resident gene, or indeed, if there are specific regulatory molecules being produced in a cell type, that these will be sufficient to regulate the multiple copies of the introduced gene. All these points are avoided in whole-animal transformation with I' elements in Drosophilu [XI, where it is possible to rapidly isolate lines containing a single nonrearranged copy of the gene per haploid genome. The major problem that exists in following the activity of a transformed gene is that of the activity of the homologous gene in its normal chromosomal site in the recipient strain, the resident gene. If the introduced gene and the resident gene are identical, then we are unable to distinguish the RNA transcripts or the protein products of the former from those of the latter, unless the introduced gene displays an abnormal stage or tissue distribution. A number of different strategies to overcome this problem have been used in Drosophilu studies, all of which have used small genes of the first class discussed in the previous section. The strategies include: 1)the introduction of a wild-type gene into a mutant strain where the resident gene is inactive (e.g., rosy (ry) [34,35] or alcohol dehydrogenase ) and where, under permissive conditions, the homozygous recessive strain is viable; 2) the rescue of an otherwise sterile genotype by the insertion of a wild-type gene (e.g., K10 [37J);3) the insertion of a modified gene, in which the transcribed region has been altered by the introduction of homologous or heterologous DNA so as to produce either an altered RNA or indeed a fusion protein that has an enzymatic activity not normally found in DrosophiZu (e.g., the hsp-lucZ fusion of Lis et al. ); and 4)the insertion of a wild-type gene where there 148 Richards exists a series of alleles producing recognizable variants either at the RNA or protein level. This last approach is the most useful if one wishes to study the regulation of transcription, as the resident gene serves as an internal control in all assays. It is the approach we have chosen with sgs3, a choice that derives essentially from the studies of Korge , which showed the existence of extensive size polymorphisms in the salivary gland secretion proteins in a number of strains of D. melunoguster. The first successful gene transfers in the P system were those with the ry gene coding for xanthine dehydrogenase [34,35] using genomic segments of 7.2 or 8.1 kb. While all inserts showed correct developmental expression, there was a fivefold range in levels of expression. These differences are considered a consequence of position effects. In particular, genes inserted on the X chromosome showed at least partial dosage compensation in males, that is, they acquired a regulation of the level of expression at their site of insertion. A number of other genes have been transformed using similarsized segments, dopa decarboxylase (7.5 kb, ), Adh (11.8 kb, ), and sgs3, (6.7 kb, ). All showed correct developmental regulation but overall levels ranged from wild-type to 10% of wild-type as judged in the various RNA or protein assays. This raises the question of the nature of the position effects on expression. Apparently, these genomic segments have the elements necessary for stage- and tissue-specific responses in the sequences flanking the gene but cannot always ensure normal levels of expression. This may result either from a property of the chromosomal region in which they are inserted or from a lack of sufficient flanking sequences. Using the imagery of the "beads on a string" model, we must ask whether we have transferred the whole bead or not. If so, will it be expressed equally in all chromosomal sites? If not, can elements flanking the insertion sometimes substitute for its normal sequences? This first round of experiments showed that genes transferred with P elements, using pieces of DNA much smaller than those previously involved in classic chromosomal rearrangements, could be regulated in new chromosomal sites. In addition, the ry constructs are unlikely to contain all the DNA of band 87Dll or its adjacent interbands (see Bossy et al. ), while the sgs3 construct g l has only two of the three glue genes clustered at 68C (see Fig. 1).This suggests that for ry the band is not a functional unit and that for sgs3 the cluster is not contained within a functional domain. Other members of clusters have also been successfully expressed in isolation following transforH Xb H X b E Xb H I l l CI sgs8 g1' lkb , I I Xb S Xh S S E I I I I I I 111, Iu17; sgs7 sgs3 1 Fig. 1. The glue gene cluster at 68C in Drosophila mehogaster. The 6.7 kb EcoRl fragment of the Formosa strain, gl, used in transformation experiments  i s indicated by a bar. The transcribed regions of sgs3, sgs7, and sgs8 are shown by arrows 5'+3'; introns are denoted by an open space. The central region of sgs3, which contains an array of 15 bp repeats whose length differs between strains [29,33] i s shown as a hatched region. E = EcoRI, H = Hindlll, S = Sall, X h = Xhol, Xb = Xbal. Into the Lair of the Gene 149 mation, e.g., chorion genes  and heat shock genes [38,43]. In the case of sgs3 the 6.7-kb construct resulted in normal levels of expression in the two sites of insertion studied. A second series of experiments showed that while most of the 3’ flanking sequences could be removed, there were elements between 1.4 and 2.7 kb 5’ to the gene necessary for normal levels of expression . With 1kb of 5‘ sequence, low but developmentally specific levels of sgs3 RNA were detected, suggesting as with other systems that this specificity is determined close to the gene. In a related set of sgs3 constructs, using a different P-element vector, Meyerowitz and his co-workers find that as little as 127-bp of 5’ sequence will confer developmental specificity of transcription (see Cherbas et a1 ). Our own experiments have shown that upstream elements in the region -127 bp to -2.7 kb will act bidirectionally on this minimum construct to restore normal levels of transcription . Evidence for rather remote sequences determining tissue specificity has been obtained by Garabedian et al.  with the yolk protein genes ypl and yp2, which share two distinct cis-acting elements in the 1.2 kb region between their 5‘ ends, one responsible for ovarian transcription and the other for fat body transcription. One element is at least 880 bp 5’ to ypl. These and other experiments have shown that DNA sequences more than 1 kb away from the capsite may have important roles to play in gene regulation and this is rather more than was previously expected. In contrast to these results, the regulation of transcription and puffing of heat shock genes has been shown by gene transfer to be relatively simple, in that for hsp70 the necessary elements lie between -89 and +65 from the transcription start site . This is in accord with the functional and evolutionary argument that heat shock genes must maintain a simple and rapid response to environmental stress and have not evolved a differential regulation such as that required for genes showing stage- and tissue-specific activity. In the next section I shall speculate on the elements that contribute to the regulation of such genes in a eukaryotic chromosome. WHAT LIES IN THE LAIR OF THE GENE? Initially it was assumed that the regulatory elements of eukaryotic genes would be found in the first 1-200 bp 5’ to the capsite and that elements such as hormone-receptor complex binding sites would be revealed by sequence comparisons between genes subjected to the same developmental regulation. Indeed sequence comparisons of a number of cloned genes led to important discoveries regarding elements such as the TATA box. We can attempt to extend this kind of analysis further upstream, either for members of closely related gene families, such as cuticle genes  or chorion genes , where the members are clustered and appear to derive from recent duplication events, or dispersed functionally related families, such as the glue genes. While such sequence comparisons have revealed conserved sequences in the upstream regions, not all of the members of the family have necessarily retained these sequences, and their positions relative to the transcribed regions and each other can be highly variable [47,48]. To date this approach has not delineated functional elements and results suggesting that elements 150 Richards may extend over a number of kilobases have not simplified the analysis. The lengths of DNA sequence that may regulate transcription in even these relatively simple genes also imposes limitations on the use of experimental approaches. Thus the fine analysis of promoters by base-by-base mutation as applied in viral systems for regions of 100-200 bp is clearly not practical with P-element analysis. The following questions may be helpful in arriving at a new view of the gene: 1)How many types of regulatory elements are there? 2) Are these elements independent? 3) Can there be more than one element of a given type? 4) Is the order of the elements fixed? 5) How important are the interactions or the spacings between different elements? 6) How is the ultimate level of gene transcription determined? It is not clear whether the elements that we are searching for are unique, both in the sense of being limited to a single copy of a given type of element for each gene and in the sense that a gene may be regulated correctly in development only by its own flanking sequence. There is already preliminary evidence on both these points that suggest that the elements may not be unique in either sense. Murine mammary tumor virus DNA sequences have been shown to contain many regions that bind the receptor for glucocorticoids in vitro [49-511. These regions are found both upstream of the transcription start site and within the transcribed regions and are all of sufficiently high affinity to be mapped by nuclease footprinting. While it has been shown by Chandler et al.  that a single region will confer glucocorticoid regulation in a gene fusion, this does not exclude the possibility that the other sites may have a functional role in normal regulation. If there are many functional sites, it is clear that deletion analysis alone may be misleading, as the deletion of one of a number of sites may either result in minor alterations in transcript level or lead to the substitution of another sequence as the major regulatory element. A similar multiplicity of sites has been observed in the heat shock regulatory sequence elements (HSE) (see Bienz  for review), and there is evidence that this also leads to difficulties of interpretation under noncritical conditions. In transfected cells, where there are multiple copies of the introduced gene, a single HSE will give rise to a clear temperature-dependent induction of transcription, whereas in transformed flies, further upstream sequences, containing a second HSE, are necessary for normal levels of transcription . It would not be surprising if similar complications exist in ecdysteroid regulatory sequences. In particular the two-step model of puffing would predict two kinds of binding site, one responsible for the opening of the puff and the second, close to the gene, for the induction of transcription. With regard to the substitution of regulatory elements there are two classes to be considered, the first consisting of developmentally specific elements and the second of a more general nature. We suspect that in the case of one sgs3 construct  sgs3 is benefiting from specific regulatory elements normally associated with sgs7 and sgs8, and experiments are in progress to test this hypothesis. In the case of regulatory elements of a more general nature, chromosome puffing may still be a source of ideas. It seems reasonable that a major problem in eukaryotic gene expression is decondensing chromatin so that the transcription complex can recognize a 1- or 2-kb sequence nor- Into the Lair of the Gene 151 mally contained in a 30,000-kb DNA molecule. A number of possible elements important for this process can already be suggested: a) attachment sites to the nuclear matrix; b)topoisomerase sites for nicking or cutting the helix, thereby facilitating the rotations necessary for unwinding chromatin from its condensed form, and c) alternating purine-pyrimidine tracts, which appear to stimulate transcription; also perhaps by facilitating conformational changes. In Dvosophila, studies of the role of topoisomerases in transcription are preliminary but encouraging. Using an indirect immunoflourescence approach, Fleischmann et al.  have shown that there are high concentrations of topoisomerase I, which makes single-strand nicks in DNA,in the region of salivary gland puffs, and in a related study Steiner et al.  showed that in a larva heterozygous for an sgs4 wild-type and a mutant allele, only the former was brightly stained. Using a different approach, Udvardy et al.  have shown that, in vitro, topoisomerase 11, which makes transient doublestrand breaks, shows major binding and cleavage sites close to the 5’ and 3’ boundaries of hsp70 gene sequences. These authors note that this distribution is similar to the attachment sites to the nuclear scaffold described for this gene by Mirkovitch et al. . The loop model of chromosome organization of Laemmli et al.  predicts that there are specific attachment sites in DNA that result in the formation of loops or domains and enable active genes to be positioned close to the nuclear matrix. The possible role of alternating purine-pyrimidine tracts is more complicated. Some of these sequences appear to have the potential to form Z DNA and this suggests that even short sequences may disrupt the normal B helix. Certain of these sequences appear to stimulate transcription (e.g., Karin et al. ) and have been found in a number of viral enhancer sequences. Whether they do in fact form Z DNA in vivo is not known, nor is it known if their primary role is to disrupt chromatin structure perhaps thereby facilitating the entry of transcription complexes, or if they act as recognition sites for specific regulatory proteins. The fine tuning of the level of transcription of a gene may depend on the number and placing of such tracts. I have summarized these ideas schematically in Figure 2 to try to give an idea of the complexity that may be present. We do not know how the regulation of a gene evolves but we may speculate that logically it is a progressive refinement of an initial differential regulation. The specific regulation afforded by a hormone-receptor complex is at least initially close to the transcription unit that it regulates. This does not prevent later duplications - -- - n ..... n n 0 ..... TAA- lkb I Fig. 2. A general scheme indicating some of the regulatory sequences that may be associated with ecdysteroid-regulated genes. A model gene is denoted as in Figure 1, with i t s TATA box. Symbols are as follows: open circles, hormone-receptor binding sequences; open boxes, sites of topoisomerase I or II activity; solid boxes, enhancer-like sequences; hatched boxes, nuclear matrix attachment sites. Lines connecting elements imply interactions that may require strict spacing in the DNA sequence- see text. 152 Richards of hormone-binding sites or refinements in the regulation such as a hormonedependent enhancer or indeed a repressor sequence. It is not clear whether the more general elements will be intimately associated with the specific regulatory elements or whether they are distributed in a random fashion in chromosomes. In the latter case, we can expect that in different chromosomal insertions transferred genes will benefit from such adjacent sequences although their effects on transcription may be highly variable. There may be at least three factors involved in the ability of these sequences to modulate gene activity, their sequence, their length, and their distance from the gene. As the first is not necessarily the most important factor, they may prove difficult to detect simply by sequence analysis, as a short sequence at -100 bp may have an effect equivalent to a much longer sequence at -1.5 kb. We must also break away from a tendency to see gene regulation as a linear problem. That is, regulation may require interactions between sequences that are separated by considerable linear distances in the primary sequence but in the folded chromatin may be in close association. I have indicated such associations in Figure 2 by joining elements by line sections. Even with these considerations we have not resolved the basis of tissue specificity, as the same DNA sequence is inactive in some tissues and not in others. This must be a consequence of differences in proteins in different cell types, which, by binding in the regulatory region of the gene, dictate the chromatin structure of the gene. Indeed, perhaps all the DNA sequences indicated should be seen as binding sites for regulatory proteins either of a specific or a general nature. If regulation evolves in the manner I have suggested, then it is possible that the order in the DNA sequence of the various elements is not the same for all genes. There is already evidence for this from the study of Garabedian et al.  (see above), where ypl has the fat body-specific element proximal, whereas yp2 has the ovary-specific element proximal, both genes presumably sharing both of these elements. Such a possibility will lead to problems in gene constructs where upstream sequences are exchanged between genes, as there may not be an equivalent site that will allow an exchange of regulatory elements. We have recently discussed other aspects of such gene fusions in the context of our attempts to establish whether units of regulation may be exchanged between genes . While there is every hope that hormonereceptor binding sites will be rapidly determined for a number of ecdysteroidregulated genes, if there is a fraction of the complexity I have suggested for other elements, our future analyses will reveal a number of surprises. LITERATURE CITED 1. Painter TS: Salivary chromosomes and the attack on the gene. J Hered 25,465 (1934). 2. Alikhanian SI: A study of the lethal mutations in the left end of the sex chromosome in Drosophilu rnelunoguster. Zoo1 Zhur (Moscow) 16, 247 (1937). 3. Bridges CB: A revised map of the salivary gland X-chromosome of Drosophilu rnelunoguster. J Hered 29, 11(1938). 4. Judd BH, Young Mw: An examination of the one cistron: one chromomere concept. Cold Spring Harbor Symp Quant Biol38, 573 (1974). into the lair of the Gene 153 5. Richards G: Polytene chromosomes. In: Comprehensive Insect Physiology, Biochemistry and Pharmacology. Kerkut GA, Gilbert LI, eds. Pergamon, Oxford, Vol. 2, pp 255-300 (1985). 6. Gurdon JB: Gene Expression During Cell Differentiation. Oxford University Press, Oxford (1973). 7. Zhimulev IF, Belyaeva ES: Proposals to the problem of structural and functional organization of polytene chromosomes. Theor Appl Genet 45, 335 (1975). 8. Zhimulev, IF, Belyaeva ES, Semeshin VF: Informational content of polytene chromosome bands and puffs. CRC Crit Rev Biochem 11, 303 (1981). 9. Sengun A:-Variability of the banding patterns of giant chromosomes. J Hered 45, 119 (1954). 10. Zhimulev IF, Belyaeva ES: Variation of banding pattern in polytene chromosomes of Drosophila melunoguster larvae. Genetics (USSR) 13, 1398 (1977). 11. Richards G: The polytene chromosomes in the fat body nuclei of Drosophilu melunoguster. Chromosoma 79, 241 (1980). 12. Ribbert D: Chromomeres and puffing in experimentally induced polytene chromosomes of Calliphora erythrocephala. Chromosoma 74, 269 (1979). 13. Zhimulev IF, Pokholkova GV, Bgatov AV, Semeshin VF, Belyaeva ES: Fine cytological analysis of the band 10A1.2 and the adjoining regions in the Drosophilu melanoguster Xchromosome. 11. Genetical analysis. Chromosoma 82, 25 (1981). 14. Keppy DO, Welshons WJ: The synthesis of compound bands in Drosophila melunogasfer salivary gland chromosomes. Chromosoma 76, 191 (1980). 15. Young M W , Judd BH: Nonessential sequences, genes and the polytene chromosome bands of Drosophila melanoguster. Genetics 88, 723 (1978). 16. Semeshin VF, Baricheva EM, Belyaeva ES, Zhimulev IF: Electron microscopical analysis of Drosophilu polytene chromosomes. 11. Development of complex puffs. Chromosoma 91, 210 (1985). 17. Semeshin VF, Baricheva EM, Belyaeva ES, Zhimulev IF: Electron microscopical analysis of Drosophilu polytene chromosomes. 111. Mapping of puffs developing from one band. Chromosoma 91, 234 (1985). 18. Berendes HD: The hormone ecdysone as effector of specific changes in the pattern of gene activities of Drosophila hydei. Chromosoma 22, 274 (1967). 19. Holden JJ, Ashburner M: Patterns of puffing activity in the salivary gland chromosomes of Drosophila. IX. The salivary and prothoracic gland chromosomes of a dominant temperature sensitive lethal of Drosophila melunogusfer. Chromosoma 68, 205 (1978). 20. Richards G: Sequential gene activation by ecdysteroids in polytene chromosomes of Drosophilu melunogaster. VII. Tissue specific puffing. Wilhelm Roux's Arch Dev Biol 191, 103 (1982). 21. Ashburner M: Happy Birthday -Puffs. In: Chromosomes Today. de La Chapelle A, Sorsa M, eds. ElsevieriNorth-Holland, Amsterdam, Vol. 6, pp 213-222. (1977). 22. Spradling AC, Rubin GM: Drosophila genome organization: Conserved and dynamic aspects. Annu Rev Genet 15, 219 (1981). 23. Benyajati C, Spoerel N, Haymerle H, Ashburner M: The messenger RNA for alcohol dehydrogenase in Drosophila rnelanogaster differs in its 5' end in different developmental stages. Cell 33, 125 (1983). 24, McClelland A, Smith FS, Glover DM: Short intervening sequences close to the 5' ends of the three Drosophilu larval serum protein 1genes. J Mol Biol 153, 257 (1981). 25. Hirsh J, Davidson N: Isolation and characterization of the Dopa-decarboxylase gene of Drosophilu rnelunoguster. Mol Cell Biol 1, 475 (1981). 26, Barnett T, Pachl C, Gergen JP, Wensink PC: The isolation and characterization of Drosophila yolk protein genes. Cell 21, 729 (1980). 27, Snyder M, Hirsh J, Davidson N: The cuticle genes of Drosophilu melunogusfer: A developmentallv regulated gene cluster. Cell 25, 165 (1981). . , 28. Muskavitch MAT, Hogness DS: Molecular analysis of a gene in a developmentally regulated puff of Drosophila melunoguster. Proc Natl Acad Sci USA 77, 7362 (1980). 29. Garfinkel MD, Pruitt RE, Meyerowitz EM: DNA sequences, gene regulation and modular protein evolution in the Drosophila 68C glue gene cluster. J Mol Biol 168, 765 (1983). ~I 2 " v 154 Richards 30. Beachy PA, Helfand SL, Hogness DS: Segmental distribution of bithorux complex proteins during Drosophilu development. Nature 313, 545 (1985). 31. Levine M, Hafen E, Garber RL, Gehring WJ: Spatial distribution of Antennupediu transcripts during Drosophilu development. EMBO J 2, 2037 (1983). 32. Bossy B, Hall LMC, Spierer P: Genetic activity along 315kb of the Drosophilu chromosome. EMBO J 3, 2537 (1984). 33. Mettling CM, Bourouis M, Richards G: Allelic variation at the nucleotide level in Drosophilu glue genes. Mol Gen Genet, 201, 265 (1985). 34. Rubin GM, Spradling AC: Genetic transformation of Drosophilu with transposable element vectors. Science 218, 348 (1982). 35. Spradling AC, Rubin GM: The effect of chromosomal position on the expression of the Drosophilu xanthine dehydrogenase gene. Cell 34, 47 (1983). 36. Goldberg DA, Posakony JW, Maniatis T: Correct developmental expression of a cloned alcohol dehydrogenase gene transduced into the Drosophilu germ line. Cell 34, 59 (1983). 37. Haenlin M, Steller H, Pirrotta V, Mohier E: A 43 kilobase cosmid transposon rescues the fs(Z)KZ0 morphogenetic locus and three adjacent Drosophilu developmental mutants. Cell 40, 827 (1985). 38. Lis JT, Simon JA, Sutton CA: New heat shock puffs and ,8-galactosidase activity resulting from transformation of Drosophilu with an hsp70-lucZ hybrid gene. Cell 35, 403 (1983). 39. Korge G: Chromosome puff activity and protein synthesis in larval salivary glands of Drosophilu melunoguster. Proc Natl Acad Sci USA 72, 4550 (1975). 40. Scholnick SB, Morgan BA, Hirsh J: The cloned dopa decarboxylase gene is developmentally regulated when reintegrated into the Drosophilu genome. Cell 34, 37 (1983). 41. Richards G, Cassab A, Bourouis M, Jarry B, Dissous C: The normal developmental regulation of a cloned sgs3 ‘glue’ gene chromosomally integrated in Drosophilu melunoguster by P element transformation. EMBO J 2, 2137 (1983). 42. Kalfayan L, Wakimoto B, Spradling A: Analysis of transcriptional regulation of the s38 chorion gene of Drosophilu by P element-mediated transformation. J Embryo1 Exp Morphol 83 (Suppl), 137 (1984). 43. Hoffmann EP, Corces VG: Correct temperature induction and developmental regulation of a cloned heat shock gene transformed into the Drosophilu germ line. Mol Cell Biol 4, 2883 (1984). 44. Bourouis M, Richards G: Remote regulatory sequences of the Drosophilu glue gene sgs3 as revealed by P element transformation. Cell 40, 349 (1985). 45. Cherbas L, Benes H, Bourouis M, Burtis K, Chao A, Cherbas P, Crosby M, Garfinkel M, Guild G, Hogness D, Jami J, Jones CW, Koehler M, Lepesant J-A, Martin C, Maschat F, Mathers P, Meyerowitz E, Moss R, Pictet R, Rebers J, Richards G, Roux J, Schulz R, Segraves W, Thummel C, Vijayraghavan K: Structural and functional analysis of some ecdysone-responsive genes from Drosophilu. Insect Biochem 16, 241 (1986). 46. Garabedian MJ, Hung M-C, Wensink PC: Independent control elements that determine yolk protein gene expression in alternative Drosophilu tissues. Proc Natl Acad Sci USA 82, 1396 (1985). 47. Snyder M, Hunkapiller M, Yuen D, Silvert D, Fristrom J, Davidson N: Cuticle protein genes of Drosophilu: Structure, organization and evolution of four clustered genes. Cell 29, 1027 (1982). 48. Wong Y-C, Pastel1J, Spoerel N, Kafatos FC: Coding and potential regulatory sequences of a cluster of chorion genes in Drosophilu melunoguster. Chromosoma 92, 124 (1985). 49. Payvar F, DeFranco D, Firestone GL, Edgar B, Wrange 0, Okret S, Gustafsson J-A, Yamamoto KR: Sequence-specific binding of glucocorticoid receptor to MMTV DNA at sites within and upstream of the transcribed region. Cell 35, 381 (1983). 50. Scheidereit C, Geisse S, Westphal HM, Beato M: The glucocorticoid receptor binds to defined nucleotide sequences near the promoter of mouse mammary tumor virus. Nature 304, 749 (1983). 51. Govindan M Y , Spiess E, Majors J: Purified glucocorticoid receptor-hormone complex from rat liver cytosol binds specifically to cloned mouse mammary tumor virus long terminal repeats. Proc Natl Acad Sci USA 79, 5157 (1982). 52. Chandler VL, Maler BA, Yamamoto KR: DNA sequences bound specifically by glucocorticoid receptor in nitro render a heterologous promoter hormone responsive in vivo. Cell 33, 489 (1983). Into the Lair of the Gene 155 53. Bienz M: Transient and developmental activation of heat shock genes. TIBS 10, 157 (1985). 54. Simon JA, Sutton CA, Lobell RB, Glaser RL, Lis JT: Determinants of heat shock induced chromosome puffing. Cell 40, 805 (1985). 55. Fleischmann G, Pflugfelder G, Steiner EK, Javaherian K, Howard GC, Wang JC, Elgin SCR: Drosophilu DNA topoisomerase I is associated with transcriptionally active regions of the genome. Proc Natl Acad Sci USA 82, 6958 (1984). 56. Steiner EK, Eissenberg JC, Elgin SCR: A cytological approach to the ordering of events in gene activation using the Sgs-4 locus of Drosophilu rnelanogusfer. J Cell Biol99, 233 (1984). 57. Udvardy A, Schedl P, Sander M, Hsieh T-S: Novel partitioning of DNA cleavage sites for Drosophilu topoisomerase 11. Cell 40, 933 (1985). 58. Mirkovitch J, Mirault M-E, Laemmli UK: Organization of the higher-order chromatin loop: Specific DNA attachment sites on nuclear scaffold. Cell 39, 223 (1984). 59. Laemmli UK, Cheng SM, Adolph KW, Paulson JR, Brown JA, Baumbach WR: Metaphase chromosome structure: The role of non-histone proteins. Cold Spring Harbor Symp Quant Biol42, 109 (1978). 60. Karin M, Haslinger A, Holtgreve H, Richards RI, Krauter P, Westphal HM, Beato M: Characterization of DNA sequences through which cadmium and glucocorticoid hormones induce human metallothionein-IIA gene. Nature 308, 513 (1984). 61. Bourouis M, Richards G: Hybrid genes in the study of glue gene regulation in Drosophilu. Cold Spring Harbor Symp Quant Biol50 (in press).