The heterochromatic Y chromosomes of several Drosophila species harbor a small number of male fertility genes (fertility factors) with several unusual features. Expression of their megabase-sized loci is restricted to primary spermatocytes and correlates with the unfolding of species-specific lampbrush loop-like structures resulting from huge transcripts mainly derived from clusters of loop-specific Y chromosomal satellites. Otherwise, there is evidence from genetic mapping and biochemical experiments that at least two of these loops, Threads in Drosophila hydei and kl-5 in D. melanogaster, colocalize with the genes for the axonemal dynein β heavy chain proteins DhDhc7(Y) and Dhc-Yh3, respectively. Here, we make use of particular Threads mutants with megabase-sized deletions for direct mapping of DhDhc7(Y)-specific exons among the large clusters of satellite DNA within the 5.1-Mb Threads transcription unit. PCR experiments with exon-specific primer pairs, in combination with hybridization experiments with exon- and satellite-specific probes on filters with large PFGE-generated DNA fragments, offer a simple solution for the long-lasting paradox between megabase-sized loops and protein-encoding transcription units; the lampbrush loops Threads and the DhDhc7(Y) gene are one and the same transcription unit, and the giant size of the DhDhc7(Y) gene as well as its appearance as a giant lampbrush loop are merely the result of transcription of huge clusters of satellite DNA within some of its 20 introns.
MEGABASE-sized clusters of tandemly repeated DNA sequences (satellites) are normally synonymous for regions of transcriptionally inactive heterochromatin. They are mostly concentrated within the centromeric and telomeric regions of chromosomes and are especially enriched in major parts of sex chromosomes. Satellite sequences from the Y chromosomes of Drosophila, however, seem to represent some of the rare exceptions to this general rule. In spite of their location on completely heterochromatic chromosomes (Figure 1A), most satellites on the Y of Drosophila are heavily transcribed in primary spermatocytes of male flies. Their transcription is remarkably restricted to areas of large lampbrush loop-like structures (Bonaccorsiet al. 1981; Lifschytzet al. 1983; Trapitzet al. 1988). The unfolding of each loop is correlated with the genetic activity of one particular member of the socalled fertility genes, e.g., Ns, Cl, Tr, Ps, Co, and Th for Drosophila hydei in Figure 1B (Hess and Meyer 1968; Bonaccorsi and Lohe 1991). If any one out of these loops is missing, D. hydei males are sterile (Hacksteinet al. 1982).
Y chromosomal satellites were originally cloned during the course of screening for particular fertility gene probes (Lifschytz 1979; Vogt and Hennig 1983; Awgulewitsch and Bünemann 1986), and the correlation of satellite-specific transcripts with particular loops has been interpreted as evidence for a “new type of eukaryotic genes” on heterochromatic chromosomes (Henniget al. 1989). This model of protein storage by specific ribonucleoprotein particle (RNP) formation on the huge transcripts of repetitive DNA was favored mainly because the presence of giant Y chromosomal loops, visible with the electron microscope, corresponded well to the megabase size of the fertility genes, as derived from mutation rates and analysis of Giemsa banding pattern in Y-translocation chromosomes (Kennison 1981, 1983; Gatti and Pimpinelli 1983; Bonaccorsiet al. 1988). On the other hand, the model was completely unable to explain on a molecular level the correlation between loss of particular loops in sterile males, loss of outer dynein arms within sperm axonemes, and absence of a specific dynein β heavy chain protein, as shown by EM sections through sperm tails and SDS gels of protein extracts from testes of mutant males, respectively (Hardyet al. 1981; Goldsteinet al. 1982; Kureket al. 1998).
As an approach to understand these conflicting results, we concentrated our research on the so-called Threads loops of D. hydei (Th in Figure 1). The pulsed-field gel electrophoresis (PFGE)-mediated identification of well-separated, extended clusters of the three Threads-specific satellites YLII, YLI, and rally, in combination with two-color in situ transcript hybridization on several Threads deletion mutants, enabled us to construct a physical map for a putative Threads-specific transcription unit (Trapitzet al. 1992; Kureket al. 1996). The transcription unit comprises ≥5.1 Mb of satellite DNA. It is located within the subterminal region on the long arm of the Y chromosome and is transcribed toward the chromosomal end (indicated as an arrow in Figure 1A). We have also successfully applied the PCR technique for detecting, cloning, and sequencing the complete cDNA of the Y-chromosomal dynein β heavy chain protein DhDhc7(Y) (Kureket al. 1998). The cDNA contains an open reading frame encoding a putative protein of 4564 amino acids (Figure 2A).
To understand the Threads-specific colocalization of the YLII, YLI, and rally satellites, as well as that of the DhDhc7(Y) gene, as deduced from cytogenetic experiments (Kureket al. 1998), we used two primer pairs representing the amino- and carboxy-terminal ends of the putative DhDhc7(Y) cDNA sequences to perform PCR experiments on DNA samples of sterile males lacking different parts of the Y chromosome (Kureket al. 1998). In contrast to DNA of wild-type samples, samples with DNA of Threads− mutants, however, failed to yield amplification products for the carboxy-terminal DNA fragment. As a result, deletion of the complete chromosomal tip comprising the NOL and the complete set of YLI- and rally-specific clusters (box of dotted lines in Figure 1A) must also include sequences from the 3′-end of the DhDhc7(Y) cDNA comprising at least the codons for the last 33 amino acids from the carboxy terminus of the putative dynein β heavy chain protein. These results provided initial evidence for DhDhc7(Y)-specific exons within Threads-specific satellite clusters. They also suggested that immotile sperm tails of sterile Threads− mutants might be devoid of outer dynein arms because of the carboxy-terminal deletion of the dynein β heavy chain protein DhDhc7(Y).
Although these results can explain conclusively the presence of an active protein-encoding gene within the Threads-specific transcription unit, they did not solve the long-lasting question concerning the link between the transcription of DhDhc7(Y) and transcription of a putative megabase-sized fertility gene associated with unfolding of the Threads loops. To address this problem, we used fertile Threads deletion mutants to study the extension of DhDhc7(Y)-specific introns in more detail. Our results of combined PFGE and PCR analyses clearly demonstrate the presence of several unusually large introns within the DhDhc7(Y) gene. One of them, intron 20, is especially huge and is composed mainly of megabase-sized clusters of the YLII, YLI, and rally satellites. In spite of its gigantic size of ≥3.6 Mb, intron 20 has to be fully transcribed to enable the transcription of the last small exon encoding 33 amino acids from the carboxy terminus of the DhDhc7(Y) protein to ensure motility of the sperm tails and fertility of the fly male. In summary, the paradox of a protein-encoding, megabase-sized Y chromosomal fertility gene in Drosophila can be solved by creating another paradox, namely that the largest transcription unit known so far exists on a heterochromatic chromosome that is generally believed to be transcriptionally inactive.
MATERIALS AND METHODS
D. hydei cell line and fly stocks: The D. hydei cell line KUN-DH-33 (Sondermeijeret al. 1980) was grown at 23° in Shield and Sang medium (Shield and Sang 1977). D. hydei stocks from our institute's collection were kept at 23° on a medium containing cornmeal, malt, sugar-beet syrup, soy flour, agar, and yeast. Individuals with special Y chromosomal Threads mutations needed for the preparation of DNA samples for the described PCR analysis were obtained as follows: X/YTh− males missing major parts of Threads were obtained from crosses of wild-type females with males of stock 99: attached X/YTh− × X·YPsTh/YTh−. X·YPsTh(mut) males missing a 3.6-Mb large internal part of Threads and X·YTh/O males with the Threads-specific Y-fragment YTh were obtained from crosses of attached Xwm1/O females (Beck 1976) with males from stock 697/139: attached X/Y × X·YPsTh(mut)/Y and stock D40F: attached X/Y × X·YTh/Y, respectively. X/O males lacking the entire Y chromosome were obtained from crosses of attached Xwm1/O females with wild-type males.
Fertility test of Thmut males: The Thmut deletion has been generated by X-ray irradiation of the X·YPsTh translocation chromosome according to Hess (1965). The resulting X·YPsTh(mut) was kept as follows: attached X/Y × X·YPsTh(mut)/Y. For the sterility test, the missing Y chromosomal loci were contributed by females from stock: attached X/YTh− × X·YPsTh/YTh−. The resulting X·YPsTh(mut)/YTh−(Thmut) males were mated with X/X wild-type females. The vials of 50 matings were controlled for viable progeny and compared with those of wild-type crosses. Progeny were observed in 43 and 39 vials of Thmut and wild-type males, respectively. Eclosion time and number of offspring were not examined in more detail.
Cloning of additional genomic sequences from DhDhc7(Y) by PCR: On the basis of the cDNA sequence of DhDhc7(Y) (Kureket al. 1998), we designed a set of nine sense and antisense primers to detect and locate any additional unknown introns within the gap regions. The positions of the primers relative to the amino acid (aa) sequence are given in parentheses: (1) Dh7cDNA+1/5′ (aa 1) 5′-TGGCTGATGATGAAGAC A3′ and Dh7AA91-3′ (aa 91) 5′-GTAGATGCGTATAAAGTAAGCGAC-3′, (2) Dh7AA49-5′ (aa 49) 5′-GGAGTTTTTATTCAATCCC-3′ and Dh7RACE-6 (aa 187) 5′-CTATACGAGGTGCTTCCTTCATGACTTCATCAATGC-3′, (3) Dh7sPCR2A-5′ (aa 168) 5′-CGAACAATATTTCCCTTGCC-3′ and Dh7AA281-3′ (aa 282) 5′-CGATTCGTCGATAGGAATGCTCG-3′, (4) Dh7AA258-5′ (aa 249) 5′-ATTCGAATCAAGACAATTGG-3′ and Dh7RACE-5A (aa 451) 5′-TCCAATTTCATAAAATCACGGCCAG-3′, (5) Dh7AA738-5′ (aa 739) 5′-CGGAAAGAGCCCTTATTGC-3′ and Dh7AA811-3′ (aa 812) 5′-CCTTTCGCTGAAAGAGCGGAACC-3′, (6) Dh7FuPro2-5′ (aa 797) 5′-GGATCCAAGGAAAGTATTAGATCC-3′ and Dh7RACE-3 (aa 1011) 5′-G CAAAGAAGACAAAGCCTGCTTTGGAATTGGGGTC-3′, (7) Dh7sPCR5A-5′ (aa 1000) 5′-CTGGACCCCAATTCCAA AGC-3′ and Dh7AA1060-3′ (aa 1061) 5′-ATATCTGAACGTTGCTTGG-3′, (8) Dh7AA1054-5′ (aa 1055) 5′-CCAAGCAACGTTCAGATATAATG-3′ and Dh7SU1123-3′ (aa 1176) 5′-CTAATTTGACTGAGAACCGATC-3′, (9) Dh7AA1170-5′ (aa 1171) 5′-GATCGGTTCTCAGTCAAATTAG-3′ and Dh7RACE-1A (aa 1264) 5′-CTCAGTATCCCTAAACTCGTATCC-3′. Approximately 1 μg genomic D. hydei DNA was analyzed with the different primer combinations in a standard PCR reaction. The samples were treated as follows: An initial incubation step at 93° for 3 min was followed by 30 cycles of 2 min at 50°, 3 min at 72°, and 1 min at 93°. In a final step, the samples were incubated at 50° for 2 min and 72° for 10 min. Amplified DNA fragments were cloned with the pGEM-T Easy cloning kit (Promega, Madison, WI). The derived clones were picked at random and sequenced. Finally, we obtained clones for four different DhDhc7(Y)-specific genomic DNA sequences, which allowed us to identify two new small introns at the amino acid positions 103 and 303. However, in the case of primer pairs 1, 3, 5, 7, and 9, we did not obtain any PCR product at all.
Cloning of the missing exon-intron boundaries by supported PCR: To reveal the complete exon-intron structure in the 5′ region of the gene, we performed seven different supported PCR (Rudenkoet al. 1993) experiments. The positions of the primers relative to the amino acid sequence are given in parentheses: (1) Dh7AA91-3′ (aa 91) 5′-GTAGATGCGTATAAAGTAAGCGAC-3′, (2) Dh7AA156-5′ (aa 158) 5′-TAGCTCAGATGAAAGGATTAG-3′, (3) Dh7AA281-3′ (aa 282) 5′-CGATTCGTCGATAGGAATGCTCG-3′, (4) Dh7AA811-3′ (aa 812) 5′-CCTTTCGCTGAAAGAGCGGAACC-3′, (5) Dh7AA987-5′ (aa 987) 5′-CATGGAGCTTCAGGAACCAATCG-3′, (6) Dh7A A1119-3′ (aa 1120) 5′-GCCAGAACTTCCTTCAAGC-3′, (7) Dh7AA1170-5′ (aa 1171) 5′-GATCGGTTCTCAGTCAAAT TAG-3′. The different primers were applied under reaction conditions essentially as described previously (Kureket al. 1998). The amplified DNA fragments were cloned using the pGEM-T Easy cloning kit (Promega). The derived clones were picked at random and sequenced. The lengths of the clones ranged from 100 to 400 bp. In all cases, the obtained sequence information was sufficient to locate any additional introns and to identify their exon-intron boundaries. Finally, 2 new introns at amino acid positions 190 and 1046 were found, raising the total number of introns in DhDhc7(Y) up to 20. Although we cloned the entire cDNA sequence and the corresponding genomic DNA sequence, as well as the exon/intron boundaries of all 20 introns of DhDhc7(Y), we failed to bridge 6 introns. Five of them (2, 4, and 10–12) are clustered in the 5′ region of the gene, whereas only a single one (20) is located in the 3′-most end.
Detection of DhDhc7(Y)- and DhDhc3-specific sequences in samples of genomic DNA: To localize the 3′ end of DhDhc7(Y) within the physical map of the lampbrush loop Threads on the Y chromosome of D. hydei, we used the following primer pairs of DhDhc7(Y)-specific sequences for PCR amplification on genomic DNA from D. hydei. The positions of the primers relative to the amino acid sequence are given in parentheses: Dh7E12-5′ (aa 4041) 5′-GCACGCTGGCTGTCTAGCC-3′ and Dh7E12-3′ (aa 4285) 5′-AACCGCTAGACATACGAGG-3′, Dh7-3I-5′C (129 nucleotides upstream from the 3′ end of intron 20) 5′-CTTGAGTGCTCTCATTTCTTTC-3′and Dh7-3-3′B (116 nucleotides downstream from the stop codon) 5′-GAGTACTTAGTTATGTATAATGG-3′. Since major parts of the cDNA from the autosomal dynein heavy chain gene DhDhc3 used in control reactions are still unknown, the amino acid positions of the DhDhc3-specific primers are taken from the corresponding DhDhc7(Y) sequences by aligning the sequences of their putative protein products: Dh3E04A-5′ (aa 522) 5′-CGATTGCCACAATCTAGAGAG-3′ and Dh3E07A-3′ (aa 679) 5′-GTTGAGGATCAACGAGTGTAC-3′.
Preparation and restriction digest of agarose inserts with high-molecular-weight DNA from KUN-DH-33 cells and nuclei of adult D. hydei flies: Sedimented cells from the D. hydei cell line KUN-DH-33 (Sondermeijeret al. 1980), washed with phosphate-buffered saline (PBS), or nuclei from adult flies prepared essentially as described (Trapitz and Bünemann 1992) were resuspended in 125 mm EDTA, pH 7.5, incubated for 2 min at 40°, mixed with 1 vol of 40° prewarmed, low-melting-point agarose (1% agarose in 125 mm EDTA, pH 7.5) and pipetted into insert molds. To inactivate residual enzymatic activities, the inserts were incubated overnight at 50° in NDS buffer (0.5 m EDTA, 10 mm Tris-HCl, 1% Na-N-sarcosinate, pH 9.5) with proteinase K (2 mg/ml). After washing with TE buffer (10 mm Tris-HCl, 0.1 mm EDTA, pH 8.0), the proteinase K was inactivated by glycine treatment (2 mg glycine/ml TE buffer) for 3 hr at room temperature. The inserts were washed again several times with TE buffer and stored in TE buffer at 4° for further usage. Prior to the restriction digestion, the inserts, or parts of them, were equilibrated in an excess volume of the appropriate restriction buffer and then incubated overnight with 100–200 units of enzyme in a total volume of 300 μl. The next day, the digestion reaction was completed by addition of 20–50 units of enzyme and incubation for an additional 2 hr. The digestion was stopped by addition of 30 μl 0.5 m EDTA, pH 8.0. After 30 min at room temperature, the inserts were cooled on ice and loaded on the gel.
PFGE and Southern blot analysis: Standard PFGE running conditions were 1% agarose, 0.25× TBE, 13°, 210 V, 22 hr, pulses log ramped from 100 to 20 s, and angles log ramped from 110 to 100°. All experiments were performed at the Rotaphore type IV R22 (Biometra). After staining with ethidium bromide and photography, gels were soaked successively in 0.25 m HCl, 1.5 m NaCl for 2× 10 min, 0.5 m NaOH for 2× 30 min, and 0.5 m Tris-HCl, 1.5 m NaCl, pH 7.0, for 2× 30 min. Finally, the gels were blotted on Hybond membranes according to standard protocols (Sambrooket al. 1989).
In situ hybridization of specific probes for the repetitive sequences (GT)n, YLII and YLI on D. hydei testes: Preparation of testes from adult D. hydei males was made essentially as described previously for the nuclei of primary spermatocytes (Glätzer 1984). To keep the testes intact, the siliconized coverslip was applied very carefully and the slide was transferred into liquid nitrogen without any squashing. Transcript in situ hybridizations on Y chromosomal lampbrush loops Threads were performed according to Trapitz et al. (1988) with fluorescein- or digoxigenin-labeled antisense RNA prepared by in vitro transcription from the YLI- and YLII-specific clones HR 17/9.2 and HR 11/2.2, respectively (the antisense sequences are shown in Wlascheket al. 1988), and with a synthetic, fluorescein-labeled (CA)15 oligonucleotide according to Huijser et al. (1987). Restriction of hybridization experiments to antisense probes is justified by our results on Northern blots of testis RNA (Trapitzet al. 1988). In all cases where oligonucleotides were used as probes for in situ hybridizations, the slides were incubated overnight at room temperature instead of at 37°. Detection of digoxigenin-labeled probes was performed by incubation with rhodamine-conjugated sheep antidigoxigenin Fab fragments (Boehringer Mannheim, Indianapolis).
Laser scanning microscopy: Preparations from the in situ hybridizations were examined and photographed on the laser scanning microscope TCS/NT from Leica with an Acousto optical tunable filter.
The 20 introns of DhDhc7(Y) come in three different size classes: As a first step toward understanding the distribution of DhDhc7(Y)-specific sequences among the satellite clusters within the Threads-specific transcription unit, we tried to complete identification and positioning of the gene-specific introns by analysis of additional genomic clones. In spite of numerous screens on several libraries of D. hydei DNA, which had identified positions and sequences of 16 putative introns, we had failed to close two remaining gaps within the 5′ region of DhDhc7(Y) gene (Kureket al. 1998). Finally, PCR and supported PCR with sets of overlapping primer pairs derived from corresponding cDNA sequences enabled detection and localization of four residual introns within those regions of the DhDhc7(Y) gene not covered by our genomic clones (Kureket al. 1998). After the precise localization of all 20 introns (Figure 2A), we tried to amplify the missing intron-specific sequences by using pairs of exon-derived flanking primers (not shown). Although we also tested a variety of PCR parameters, we failed to bridge the introns 2, 4, 10–12, and 20 (gray bars interrupted by question marks in Figure 2A). This result was rather unexpected because out of the remaining 14 introns (black bars in Figure 2A), 12 were extremely small, ranging from 47 to 72 bp, or only medium sized, with 3.100 and 1.169 bp for introns 13 and 17, respectively (GenBank accession no. AF-031494). Beside their remarkable size distribution, all other parameters of DhDhc7(Y)-specific introns, e.g., splice site-specific sequences, are essentially identical with those listed for sets of introns from various euchromatic genes of Drosophila (Mountet al. 1992).
Thmut males with an internal 3.6-Mb Threads deletion are fertile: During the course of our PFGE experiments to physically map the three Threads-specifically transcribed clusters of YLII, YLI, and rally satellites, one fertile Threads mutant, Thmut, has been analyzed because of its peculiar loop morphology (Figure 1D). The most obvious difference was related to the position of the Pseudonucleolus (Ps). In contrast to the wild type (Figure 1B), the extended Threads-like structures Th, located between Ps and the NO, the nucleolar organizer at the base of the pear-shaped spermatocyte nucleus, seemed to be substantially reduced in size. PFGE analysis, in combination with two-color transcript FISH, revealed that this shortening of the Threads in Thmut was caused by a huge internal deletion comprising 3.6 Mb of YLII and YLI satellites, indicated as Δ in Figure 2B (Kureket al. 1996). In spite of this deletion, fertility of Thmut males was not impaired (results not shown). Accordingly, PCR analysis revealed the presence of the complete coding sequence of DhDhc7(Y) in Thmut males. Especially, amplification of exon-19- and exon-21-specific fragments of 736 bp in lane 5 and of 345 bp in lane 6 of Figure 3, respectively, demonstrated the presence of protein-coding sequences from both ends of intron 20. Consequently, the codons for the last 33 amino acids from the carboxy-terminal end are present in Thmut males, enabling the formation of a full-sized DhDhc7(Y) mRNA and, hence, normal fertility. The 3.6-Mb deletion within intron 20 obviously has no inhibitory influence on the splicing machinery. In contrast, exon-21-specific sequences are absent (lane 8 in Figure 3) in sterile Th− males (Figure 2B). Therefore, sterility of Th− mutants could result from the absence of functional, full-length DhDhc7(Y) mRNAs.
Intron 20 of DhDhc7(Y) is gigantic because of megabase-sized clusters of satellite DNA: Although a huge intron within the coding region of the Y chromosomal DhDhc7(Y) gene would provide an explanation for the megabase-size of the putative Threads-associated fertility gene, the results with Thmut could probably also result from the additional chromosomal rearrangements often observed in combination with large chromosomal deletions. Because we could not exclude such complex rearrangement during generation and maintenance of the Thmut mutation, we chose the PFGE approach to prove the presence of giant introns within the DhDhc7(Y) gene directly. To get complete digests of wild-type Y chromosomal DNA, we performed the PFGE experiments essentially on DNA of D. hydei KUN-DH-33 cells, which are especially well suited because of supernumerous copies of otherwise normal Y chromosomes (Sondermeijeret al. 1980; Trapitzet al. 1992). Digests of KUN-DH-33 DNA were made with a number of restriction enzymes rarely cutting within all DhDhc7(Y)-specific exons and within the 14 introns cloned so far (see Figure 4B). The resulting Southern blot (Figure 5A) was subsequently hybridized with specific probes for exon 19 (probe 7 in Figure 4C) and exon 21 (probe 8 in Figure 4C) flanking the 5′ and 3′ splice sides of the putative huge intron 20, respectively. Both probes of unique DNA hybridized mainly with large DNA fragments (Figure 5, B and C). Especially in the SacI lanes, hybridization signals in the fragment size range of several hundred kilobases were obtained (lanes 2 in Figure 5, B and C). Since SacI cuts several times within exons 15 and 16 and once in the 3′ untranslated region (Sa in Figure 4B), the observed fragments must result from very rare SacI restriction sites within repetitive DNA sequences in intron 20. This interpretation is fully supported by the results of subsequent hybridizations with probes of YLII, YLI, and rally satellites. Among a couple of large fragments for each particular satellite (Figure 5, D–F), in the SacI lanes, exactly one YLII signal colocalizes with that of exon 19 (arrowheads in Figure 5, B and E), and only one YLI-specific DNA fragment shows crosshybridization with the exon-21-specific probe (arrowheads in Figure 5, C and F). In contrast, no cross-hybridization is observed for the samples of unique DNA and the rally probe (Figure 5D). In summary, our hypothetical model for the existence and orientation of large satellite clusters within intron 20 found full support from the PFGE experiments.
Although similar experiments were performed with several additional samples of unique DNA from the 5′ half of the DhDhc7(Y) cDNA (probes 1–6 in Figure 4C), the lack of suitable intron-specific satellite probes prevented any clear demonstration of crosshybridization of exon- and satellite-specific probes with one and the same large DNA fragment. There is evidence, on the other hand, that especially the so-called Cones (Co in Figure 1) may result from strong transcription of (GT)n repeat clusters (Huijseret al. 1987). Although subclones from introns 11 and 12 indeed also contain short stretches of (GT)n repeats (results not shown), a direct correlation between (GT)n-specific large DNA fragments, e.g., the 320-kb ClaI fragment and the DhDhc7(Y)-specific probe 3 in Figure 4C, is not possible unequivocally. In addition, longer stretches of (GT)n sequences are distributed throughout the D. hydei genome and not restricted to the Y chromosome (Huijseret al. 1987). Therefore, hybridization of (GT)n probes on PFGE Southern blots of D. hydei DNA produces a lot of more diffuse high-molecular-weight bands on an intensive background smear (not shown). Nevertheless, restriction with ClaI and subsequent hybridization with probes 1–3 (Figure 4C) indicate that the 5′ end of intron 10 and the 5′ end of intron 13 are separated by ≥400 kb from each other (not shown). As a result, other introns of unusual size and repetitive DNA content must exist within DhDhc7(Y), beside the especially large intron 20.
Direct visualization of sequential transcription of (GT)n, YLII, and YLI satellites in D. hydei testes: Primary spermatocytes of Drosophila are generated from stem cells in the apical proliferative center of the adult testis by a series of cell divisions. As a primary gonial cell arises from the stem cell by division, a cyst progenitor also divides. By this process, each primary gonial cell and its progeny are enclosed by a pair of cyst cells throughout spermatogenesis (for a recent review see Fuller 1993). In D. hydei, three mitotic divisions of the primary spermatogonial cell give rise to a cyst with eight premeiotic spermatocytes. After premeiotic replication, the primary spermatocytes switch from a program of cell division to one of growth and gene expression. During the next 5 days, spermatocytes of D. hydei grow ~25 times in volume and a number of genes are transcribed for the first time. But only the expression of Y-linked fertility factors is visibly manifested in the unfolding of characteristic lampbrush-like loops within the enlarging nuclei (Figure 1B). During this time, growing spermatocytes form a gradient from younger and smaller nuclei toward older and larger ones within the apical tip of the testis tube (arrow in Figure 6G).
Our model assumes that transcription on the Threads-specific transcription unit is directed from the Ps toward the NO (as shown by Figure 1A) and that the successive unfolding of the Threads loops is necessary to enable the generation of full-length 5.1-Mb transcripts. Under normal transcription rates, however, generation of 5.1-Mb transcripts requires 2–3 days (LeMaire and Thummel 1990; Tennysonet al. 1995). As a result, growth of spermatocyte nuclei and unfolding of Y chromosomal loops are strictly coupled. Under these conditions the chronological order for the first appearance of particular satellites within the loops transcripts of growing spermatocytes should follow exactly their sequential order on the Y chromosome as predicted by the physical map in Figures 2B and 6H. To verify this hypothesis, we performed two series of simultaneous in situ transcript hybridizations with pairs of differentially colored satellite probes on whole mounts of testis tubes. As predicted by our model, in small nuclei of early spermatocytes, transcription of (GT)n repeats, most probably restricted to the large introns (2, 4, and 10–12) in the 5′-region (Figure 4C), is followed by transcription of YLII-specific repeats in intron 20, as confirmed by the time of appearance and increasing size of corresponding hybridization areas (Figure 6, A–C). During this period, YLI satellites are not transcribed (not shown). Instead, transcription of the YLI satellite in intron 20 is restricted to larger nuclei of more mature spermatocytes and follows transcription of YLII repeats (Figure 6, D–F). Whereas YLII gives rise to an increasing matrix area (Figure 6, D–F), YLI transcription yields the typical Threads-like structure (Figure 6F). In summary, the chronological order of satellite transcription in growing spermatocytes exactly mirrors their distribution in different introns within the DhDhc7(Y) transcription unit.
Although the overall organization of the DhDhc7(Y) transcription unit is well understood, precise localization of defined promoter and termination regions is presently impossible. Transcripts of Threads-specific sequences most probably start somewhere in the (GT)n-rich region of the Cones (Figure 1A). Whether Threads-specific, (GT)n-rich transcripts result from the Cones (Huijseret al. 1987) or from (GT)n repeats within DhDhc7(Y)-specific introns 2, 4, and 10–12, however, is not clear. Similarly, the available Threads deletions are not sufficient to localize exon 21 more precisely with respect to the rally and the smaller YLI cluster. All clusters are clearly transcribed on fully unfolded Threads (Kureket al. 1996). Otherwise, mutants with the completely deleted smaller YLI distal cluster contain exon 21 and are fertile. Whether transcription of the smaller YLI cluster is probably due to readthrough transcription of satellites (Varleyet al. 1980) or due to location of the complete satellite cluster within intron 20, as proposed by Figure 6H, is not known.
Our mega-intron model of the D. hydei fertility factor Threads interprets major parts of this lampbrush loop as nothing else but the transcription product of a gigantic intron that separates the exon encoding the last 33 amino acids at the carboxy terminus from the main body of the dynein β heavy chain protein DhDhc7(Y) comprising a total of 4564 amino acids (Kureket al. 1998). The model is supported by experimental results on sterile Th− and fertile Thmut males and the finding that Th− males are infertile due to immotility of their sperm. Loss of dynein outer arms and, hence, sperm motility by a small deletion at the carboxy terminus of the DhDhc7(Y) protein is unlikely, however, because C-terminal deletions in axonemal outer-arm dynein heavy chain proteins of Chlamydomonas enable the assembly of correctly placed dynein outer arms (Sakakibaraet al. 1993). Therefore, the missing of any outer dynein arms in Th− axonemes seems to be indicative for the complete absence of DhDhc7(Y) proteins. Most likely, DhDhc7(Y) proteins are not synthesized because the terminal deletion in the DhDhc7(Y) gene prevents the correct splicing of the primary Th− transcript and, hence, its export from the nucleus.
This model is also supported by data on lampbrush loops that indicate that spectacular lampbrush loops are often associated with transcribed repetitive sequences (for a comprehensive review, see Callan 1986). In situ transcript hybridization experiments with Notophthalmus viridescence DNA samples of low complexity, e.g., satellite DNA and cloned histone genes especially revealed the remarkable observation that satellite DNA and histone genes are cotranscribed along the same loops. Cotranscription occurs because transcripts start from promoters within the tandemly arranged 9-kb histone gene transcription units and proceed unterminated through the interspersed spacers of satellite clusters ≤100 kb in length (Diaz and Gall 1985). Interestingly, these loops with actively transcribing histone gene clusters seem to be associated with the so-called sphere loci, a small number of eye-catching marker structures on particular chromosomes of the newt. Similarly, the nucleoli are the result of genetic activity of hundreds of tandemly arranged rRNA genes clustered within the nucleolar organizer region of chromosomes. In both examples, microscopic visibility of active genes results from accumulation of RNP complexes assembled on nascent transcripts of several hundreds of simultaneously transcribed repetitive genes.
With these examples for RNP accumulation on transcripts of clustered repetitive sequences in mind, the Threads are looking exactly as spectacular as they should look, provided they were actively transcribed megabasesized clusters of satellite DNA (Figure 1, B and C). According to the differences of their underlying repeats, YLII and YLI, the areas of corresponding RNP complexes appear as a diffuse matrix or as compact threads, respectively (Figure 6, A–F).
Although cotranscription of repetitive DNA apparently seems to be rather common during the course of gene expression in premeiotic cells, the main peculiarity of Y chromosomal fertility genes, such as the Threads in D. hydei or kl-5 in D. melanogaster, derives from their gigantic size (Hacksteinet al. 1982; Gatti and Pimpinelli 1983) and disproportion between the transcription of “sense” and “nonsense” DNA along these two (and probably other) Y chromosomal lampbrush genes (Bonaccorsi and Lohe 1991; Kureket al. 1996). The same criteria actually also hold true for a small number of genes on euchromatic chromosomes. The best known example is the human dystrophin gene, spanning ≤2.3 Mb of genomic DNA and encoding 79 rather small exons for an mRNA of 14 kb (Tennysonet al. 1995). In this case, 99.4% of the gene is composed of introns. Unfortunately, nothing is known about the size distribution and DNA complexity of these introns. Otherwise, it is very well conceivable that the splicing machinery can handle megabases of tandemly arranged simple repeats of several hundred base pairs even less erroneously than the equivalent amount of transcribed unique DNA. On the basis of the assumption that transcription of human dystrophin mRNA lasts 16 hr, one can calculate that the DhDhc7(Y) gene can be transcribed in 35 hr. Whether the huge size of the Threads fertility gene is also causative for the delayed onset of fertility of D. hydei males, 9 days after eclosion (Pitnicket al. 1995), is unknown at this time. This delay obviously has no adverse effect on fertility, although there is a correlation between sperm length and onset of fertility (Pitnicket al. 1995). According to this correlation, the late onset of fertility in D. hydei is mainly a result of the ≤23-mm!long spermatozoa of this fly species.
In spite of the fact that some uncommon features of the DhDhc7(Y) gene can be also observed for a small number of other more conventional genes, the accumulation of megabase-sized male fertility genes on the heterochromatic Y chromosomes of some Drosophila species is striking and probably due to two unique characteristics of this chromosome. (i) Recombination in the male germ line of Drosophila is absent or extremely rare. Also, sequences shared by the X and Y are restricted to clusters of rDNA and to some additional families of tandem repeats in D. melanogaster (Gatti and Pimpinelli 1992). Therefore, elimination of repetitive DNA by recombination processes, as continuosly generated by transposable elements or as the result of occasional saltatory amplifications, can be expected to be especially ineffective in Drosophila Y chromosomes. As a result, insertion of additional DNA is strongly favored for the Y chromosome, as shown for the Neo-Y in D. miranda (Steinemann and Steinemann 1998). (ii) In addition, function of Y chromosomal fertility genes is restricted to terminal differentiation processes during the course of the development of fertile spermatozoa. Therefore, any invasion of additional DNA into introns of genes exclusively required for male fertility would not be harmful for the persistence of a fly population as long as transcription and precise splicing are taking place within the correct time scale.
Since the pioneering work of Heitz (1928), constitutive or α-heterochromatin is defined as transcriptionally inactive chromatin most commonly restricted to centromere and telomere regions of chromosomes and detected by positive Giemsa staining. Such regions of α-heterochromatin are composed principally of megabase-sized clusters of a limited number of more or less pure satellites (Miklos 1985). For these reasons, it is remarkable that special factors or conditions seem to exist within premeiotic spermatocytes (and oocytes) that enable genes to overcome the heterochromatic gene-silencing character of megabase satellite clusters that are especially well known from the phenomenon of position effect variegation (Karpenet al. 1994). At this time, these factors, which provide transcriptional competence to heterochromatic regions of chromosomes in premeiotic cells, are completely unknown. There might, however, be no need for special proteins. In contrast, the specific areas normally occupied by the heterochromatic parts of chromosomes within somatic cells (Karpenet al. 1994; Henikoff 1997) are dissolved on their way toward meiosis and, especially, toward chromatin reorganization by the replacement of histones with other more basic proteins during course of terminal spermatozoa differentiation. Whether the absence of histone H1 within nuclei of primary spermatocytes of D. hydei (Kremeret al. 1986) represents a first step within this chromatin reorganization is not known.
We thank O. Hess for providing us with the fly stocks, M. Gatti, J. H. P. Hackstein and R. Hochstenbach for stimulating discussions, U. A. O. Heinlein for critical reading of the manuscript, and C. Gieseler for skillful technical assistence. This work was supported by grants from the Deutsche Forschungsgemeinschaft (Bu 342/4-6, 6-1).
Communicating editor: S. Henikoff
- Received July 6, 1999.
- Accepted October 28, 1999.
- Copyright © 2000 by the Genetics Society of America