Mutant Alleles of the Drosophila trithorax Gene Produce Common and Unusual Homeotic and Other Developmental Phenotypes

Corresponding author: Thomas R. Breen

Communicating editor: R. S. HAWLEY

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

trithorax (trx) encodes chromosome-binding proteins required throughout embryogenesis and imaginal development for tissue- and cell-specific levels of transcription of many genes including homeotic genes of the ANT-C and BX-C. trx encodes two protein isoforms that contain conserved motifs including a C-terminal SET domain, central PHD fingers, an N-terminal DNA-binding homology, and two short motifs also found in the TRX human homologue, ALL1. As a first step to characterizing specific developmental functions of TRX, I examined phenotypes of 420 combinations of 21 trx alleles. Among these are 8 hypomorphic alleles that are sufficient for embryogenesis but provide different levels of trx function at homeotic genes in imaginal cells. One allele alters the N terminus of TRX, which severely impairs larval and imaginal growth. Hypomorphic alleles that alter different regions of TRX equivalently reduce function at affected genes, suggesting TRX interacts with common factors at different target genes. All hypomorphic alleles examined complement one another, suggesting cooperative TRX function at target genes. Comparative effects of hypomorphic genotypes support previous findings that TRX has tissue-specific interactions with other factors at each target gene. Some hypomorphic genotypes also produce phenotypes that suggest TRX may be a component of signal transduction pathways that provide tissue- and cell-specific levels of target gene transcription.

THE Drosophila trithorax gene (trx) encodes a large protein (TRX) that is required throughout development to maintain tissue- and cell-specific levels of homeotic and other gene transcription (CAPDEVILA and GARCIA-BELLIDO 1981 ; INGHAM 1981 ; DUNCAN and LEWIS 1982 ; CABRERA et al. 1985 ; INGHAM 1985A , INGHAM 1985B ; CAPDEVILA et al. 1986 ; MAZO et al. 1990 ; BREEN and HARTE 1991 , BREEN and HARTE 1993 ; SEDKOV et al. 1994 ). In trx mutants, transcription of homeotic genes of the Antennapedia complex (ANT-C) and bithorax complex (BX-C) is reduced or absent in a specific subset of cells within a gene's normal expression domain. Besides homeotic genes, the transcription of engrailed (BREEN et al. 1995 ), fork head (KUZIN et al. 1994 ), and polyhomeotic (FAUVARQUE et al. 1995 ) is also TRX dependent. TRX associates with at least 76 sites on salivary gland polytene chromosomes, suggesting many additional target genes (KUZIN et al. 1994 ; CHINWALLA et al. 1995 ; PARO and HARTE 1996 ). trx transcripts are found in all cells during embryogenesis and are similarly widely distributed in imaginal discs (MOZER and DAWID 1989 ; KUZIN et al. 1994 ; SEDKOV et al. 1994 ; STASSEN et al. 1995 ). Characterized TRX target genes, such as homeotic genes, encode transcriptional regulatory factors that specify cell fates. It is not known if this is a common feature of TRX target genes.

There are at least five differentially spliced trx mRNAs that encode two protein isoforms, TRXI of 3358 amino acids and TRXII of 3726 amino acids. The protein isoforms differ by 368 N-terminal amino acids that are encoded in an alternatively used exon (MAZO et al. 1990 ; BREEN and HARTE 1991 ; SEDKOV et al. 1994 ; STASSEN et al. 1995 ). The 10- and 12-kb mRNAs that encode only TRXI are maternally supplied to oocytes and are present at decreasing levels through embryogenesis (MOZER and DAWID 1989 ; BREEN and HARTE 1991 ; SEDKOV et al. 1994 ). A 14-kb mRNA that encodes TRXII, and could potentially translate TRXI, too, is expressed from early embryogenesis through pupation (MOZER and DAWID 1989 ; BREEN and HARTE 1991 ; SEDKOV et al. 1994 ). Only the larger mRNA encoding TRXII is expressed during imaginal cell proliferation. Western blot analysis showed that TRXI is the most prevalent isoform during early embryogenesis, while TRXII is the predominant isoform during the final third of embryogenesis (KUZIN et al. 1994 ). It has not been reported whether TRXII is the only isoform present during larval growth and imaginal proliferation.

Both TRX isoforms have a C-terminal SET domain found in other proteins known or suspected to modulate chromatin structure. These proteins include SU (VAR)3-9, which modulates heterochromatin-mediated repression (TSCHIERSCH et al. 1994 ), and E(Z), a Polycomb group (PcG) protein required for transcriptional repression of homeotic genes (JONES and GELBART 1990 , JONES and GELBART 1993 ). E(Z) is required for binding of TRX and other proteins to specific chromosomal sites where they may interact with other chromatin factors to alter target gene transcription (RASTELLI et al. 1993 ; KUZIN et al. 1994 ; PLATERO et al. 1996 ). The SET domain of HRX (aka MLL, ALL1, HTRX), the human homolog of TRX, interacts with human myotubularin, a dual-specificity phosphatase, and Sbf1 (CUI et al. 1998 ). Apparently, Sbf1 protects the SET domain from dephosphorylation by myotubularin. This protection delays cell maturation and differentiation, which are promoted after SET domain dephosphorylation effected by myotubularin (CUI et al. 1998 ; DE VIVO et al. 1998 ). The TRX SET domain associates with SNR1, a homolog of the yeast SWI/SNF protein SNF5 that participates in chromatin remodeling to facilitate transcription (ROZENBLATT-ROSEN et al. 1998 ). By this interaction, the SET domain of TRX may abet the recruitment of Drosophila SWI/SNF complexes to some target genes to help modulate transcription by altering chromatin.

The TRX isoforms also contain four centrally located PHD-finger motifs (MAZO et al. 1990 ; AASLAND et al. 1995 ; STASSEN et al. 1995 ) found in other proteins that appear to interact with chromatin (LONIE et al. 1994 ; ADAMSON and SHEARN 1996 ; TRIPOULAS et al. 1996 ). AASLAND et al. 1995 speculate that PHD fingers may mediate protein-protein interactions between regulatory factors or they may recognize specifically modified histones. N-terminal to the PHD finger motifs, TRX isoforms have a C4 zinc-finger motif similar to the DNA-binding domain (DBD) of nuclear receptors (STASSEN et al. 1995 ), though it is not known whether TRX can bind DNA. TRX shares three additional conserved motifs with ALL1 and the closely related ALR (DJABALI et al. 1992 ; GU et al. 1992 ; TKACHUK et al. 1992 ; DOMER et al. 1993 ; ROWLEY 1993 ; PRASAD et al. 1997 ). One motif, of unknown significance, precedes the C-terminal SET domain. Two other short motifs, located on either side of the DBD homology (STASSEN et al. 1995 ), promote nuclear localization and may be necessary for chromosomal binding (YANO et al. 1997 ).

TRX and other members of the trithorax group (trxG) of homeotic gene positive regulators behave genetically as antagonists of PcG transcriptional silencing (CAPDEVILA and GARCIA-BELLIDO 1981 ; INGHAM 1983 ; CAPDEVILA et al. 1986 ; SATO and DENELL 1987 ; KENNISON and TAMKUN 1988 ; SHEARN 1989 ). TRX and another trxG protein, ASH1, colocalize with some PcG proteins at many sites along salivary gland polytene chromosomes (CHINWALLA et al. 1995 ; TRIPOULAS et al. 1996 ). Genetic and cytological data suggest that TRX and PC assemble relatively near each other at Polycomb response elements (PREs) near the genes Ultrabithorax (Ubx) and Sex combs reduced (Scr; CHAN et al. 1994 ; CHANG et al. 1995 ; CHINWALLA et al. 1995 ; GINDHART and KAUFMAN 1995 ). More recently, it was shown that TRX colocalizes with PC at the major PREs in the BX-C (ORLANDO et al. 1998 ). These observations suggest that the genetic behavior of some trxG and PcG genes may reflect direct functional interactions between their proteins. However, the significance of TRX and PC colocalization remains unknown. The genetic evidence cited above suggests some trxG and PcG proteins colocalize at PREs in other tissues to exert their regulatory effects. The mechanism by which TRX stimulates transcription and preempts PcG silencing is unknown. However, PcG silencing appears to occur by default if a target gene is not transcriptionally active at the time PcG silencing is implemented during germband elongation (PIRROTTA et al. 1995 ; POUX et al. 1996 ; PIRROTTA 1997 , PIRROTTA 1998 ).

The genetic, cytological, biochemical, and structural analyses of TRX suggest it may interact with a variety of proteins at its target genes to stimulate transcription through chromatin modulation. TRX is widely distributed during embryogenesis and imaginal development, as are its PcG antagonists. Either may exert its effect on a target gene depending on other factors that determine the target gene's transcriptional status during germband elongation. Within a homeotic gene's expression domain, TRX is used to different extents to stimulate that gene's transcription. Depending on the target gene, TRX may be generally required to boost its transcription in specific cells or tissues. For some target genes, TRX is essential for transcription in specific cells or tissues. The differential use of TRX at its target genes strongly suggests that tissue- or cell-specific combinations of regulatory factors interact with unique combinations of TRX functional groups to elicit appropriate use of TRX in a cell. As a first step to identifying distinct functions of TRX, I performed a detailed analysis of phenotypes associated with inter se combinations of 21 trx alleles.

Eclosing and pharate adults were examined for phenotypes associated with reduced expression of the trx homeotic target genes Scr of the ANT-C and Ubx, abdominal-A (abd-A), and Abdominal-B (Abd-B) of the BX-C. I hoped to identify allele-specific phenotypes that may reveal different functional domains of TRX. I characterized eight hypomorphic alleles, seven of which primarily affect imaginal development. Results demonstrate that TRX is used in tissue-specific contexts at the target genes examined. Some trx genotypes appear to have almost no Scr function in T1 leg discs, no Ubx function in T3 leg discs, and greatly reduced function of the other genes examined in their respective imaginal tissues. Some trx genotypes exhibited additional phenotypes, some of which are also seen in trx^- somatic clones (INGHAM 1981 , INGHAM 1985A ). These latter phenotypes are similar to ones produced by mutations in elements of signal transduction pathways. I suggest that the differential effects of trx mutations on different tissues and cells may be due in part to the differential regulation of TRX by cell-signaling mechanisms. I present a model of TRX regulation consistent with its signal transduction and homeotic mutant phenotypes.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Fly crosses:
Crosses were set in vials containing USB/Amersham Fly Diet. About five male and five virgin female flies were placed in each vial. Each cross consisted of at least four vials. Fly crosses were maintained at 22° except for those involving trx¹, which were maintained at 25°, the temperature at which trx¹ has its highest penetrance and expressivity (INGHAM and WHITTLE 1980 ). Adults were discarded from vials after 10 to 12 days. Fly stocks used in the crosses were trx^M17 red e/TM6B, trx^Z32 red e/TM6B, st trx¹/st trx¹, trx^E3/TM6B, trx^Z16 red e/TM6B, trx^Z11 red e/TM6B, trx^M18 red e/TM6B, cu trx^JY16 red e/TM6B, mwh trx^6.1 red e/TM6B, trx^D sr e/TM6B, trx^B11 red e/TM6B, trx^A7 red e/TM6B, trx^M14 red e/TM6B, trx^Z15 red e/TM6B, T(Y;3)JY25, cu trx^JY25 red e/red cv-c sbd², mwh trx^7.1 red e/TM6B, trx^Z44 red e/TM6B, st trx³ red/TM6B, cu trx^JY21 red e/TM6B, Df(3R)redP52/TM1, and Df(3R)redP6/TM1. TM1 is In(3LR)TM1, Me ri sbd^l. TM6B is In(3LR)TM6B, Hu e Tb ca. See references in Table 1 for origins of trx mutant chromosomes.

Lethal phase:
Vials were inspected for dead embryos, small or sickly larvae, and dead pupae as a component of lethal phase determination. A lethal phase in this report indicates the latest phase in which animals of an indicated genotype are seen. For each trx mutant genotype, the lethal phase is embryonic if no second instar larvae were detected, larval if no pupae were detected, and pupal if no adults were detected. Except in crosses involving trx^JY25, Df(3R)redP52, and Df(3R)redP6, transheterozygotes beyond the first larval instar are identifiable as Tb⁺ animals and as Hu⁺ adults because they are the only offspring that do not carry TM6B. Also, except in crosses involving trx¹, trx^E3, and trx^D, trx transheterozygotes and hemizygotes are red homozygotes or hemizygotes, respectively, and identifiable by their red Malpighian tubules after the first larval instar and red mutant eye color as adults. Crosses that involve the intersection of the two exceptional categories can produce heteroallelic larvae and pupae that cannot be distinguished by being either Tb⁺ or red mutants. Their lethal phases were determined by other criteria. Crosses between st trx¹ flies and flies carrying trx^JY25, Df(3R)redP52, and Df(3R)redP6 produced mutant adults distinguished by their phenotypes. Crosses between trx^E3/TM6B flies and flies carrying trx^JY25, Df(3R)redP52, and Df(3R)redP6 produced disproportionate numbers of dead pupae as did other crosses using trx^E3/TM6B flies in which the dead pupae were scored as mutant transheterozygotes because they were Tb⁺. Crosses between trx^D/TM6B flies and flies carrying trx^JY25, Df(3R)redP52, and Df(3R)redP6 produced disproportionate numbers of dead embryos as did other crosses using trx^D/TM6B flies in which transheterozygotes were determined to die as embryos because they did not produce Tb⁺ second instar larvae.

Alleles examined:
I examined phenotypes produced by inter se combinations of 21 of 67 available trx alleles (Table 1). The 21 alleles used in this study were chosen because (1) they have previously described phenotypic effects, (2) they correlate with described molecular lesions, (3) they were used in previous developmental studies, and (4) they have comparatively high penetrance of haploinsufficient phenotypes, or a combination of these characteristics.

The eight hypomorphic alleles described in this study are trx^M17, trx^Z32, trx¹, trx^E3, trx^Z16, trx^Z11, trx^M18, and trx^JY16. MORTIN et al. 1992 reported that trx^M17 and trx^Z32 are hemizygous viable at 22°. Homozygous trx¹ adults from homozygous trx¹ mothers show an array of transformation phenotypes associated with reduced homeotic gene expression in imaginal tissues. The penetrance of transformation phenotypes increases with increasing temperature to 25°. trx¹ is associated with an ~9-kb insert in the region encoding the first intron of trx (Figure 1). trx¹ probably has no qualitative effect on trx proteins. trx^E3 is associated with an in-frame deletion that removes 271 amino acids from TRXI and TRXII (Figure 1). The deleted residues are located on the C-terminal side of the central cysteine-rich domain that contains PHD fingers. trx^E3 is homozygous embryonic lethal, but is a pupal lethal allele in combination with null alleles. trx^Z16 and trx^Z11 correlate with point mutations in the central cysteine-rich and terminal SET domains, respectively. They are pupal lethal alleles in combination with null alleles. trx^M18 has low penetrance of haploinsufficient phenotypes and is a strong hypomorphic, pupal lethal allele in combination with null alleles. trx^JY16 is associated with a chromosomal break in the region encoding exon 3 (Figure 1). The resulting fusion gene can express unaltered TRXI and perhaps a form of TRXII with novel N-terminal residues, assuming a fusion initiation codon is used. It is a larval lethal allele in combination with null alleles.

View larger version (16K): In this window In a new window Download PPT slide   Figure 1. Characterized mutations within the trx transcription unit. (A) The map depicts 32 kb containing the trx transcribed region. It is derived from previously published reports (MOZER and DAWID 1989 ; MAZO et al. 1990 ; BREEN and HARTE 1991 ; SEDKOV et al. 1994 ; STASSEN et al. 1995 ). The thin line with vertical tick marks represents an EcoRI map of the region. Above it are nine rectangles that indicate regions encoding exons. Unfilled areas depict 5'- and 3'-untranslated sequences; filled areas show translated sequences. Connectors between the exons represent splicing alternatives. The M's on the 5' sides of the third and fourth exons label positions of likely initiation codons. The TRXII isoform of 3726 amino acids is translated from mRNAs that contain the third exon. The TRXI isoform of 3358 amino acids initiates within the fourth exon. The tick marks labeled "... AAA" show the positions of alternatively used polyadenylation signals. The positions of the point mutations associated with the trxZ11 and trxZ16 alleles are indicated above the exon rectangles, as is the approximate location of a rearrangement breakpoint associated with the trxJY16 allele. The gradient shaded boxes show the approximate sizes and locations of deletions associated with the trxE3 and trxB11 alleles. The proximal breakpoint of Df(3R)redP6 is located within the region represented by the open box labeled redP6. The rightward arrow indicates that Df(3R)redP6 deletes distally beyond the extent of the map. The inverted triangle depicts the approximate position of the 9-kb insert associated with the trx1 allele. The open box labeled trx1 indicates the region of uncertainty within which the insert is located; the base of the triangle represents the size of the insert. (B) TRXII and TRXI mutant isoforms are depicted. The C termini of these protein representations are to the left, consistent with the orientation of the transcription unit in A. The trxZ11 allele is associated with a missense mutation causing a G- to S-substitution at amino acid 3601 in the conserved SET domain (lightly shaded region). The trxZ16 allele is associated with a missense mutation causing an R- to W-substitution at amino acid 1753 in the conserved Cys-rich, PHD finger domain (medium gray region). The black region labeled DBD shows the region with similarity to DNA-binding domains of steroid receptors. The trxE3 allele is associated with an in-frame deletion that leads to the removal of 271 amino acids from both isoforms. The region removed is indicated by the gradient shaded box. The trxB11 allele is associated with an 833-bp deletion that encodes truncated isoforms with 83 novel, C-terminal residues (shaded boxes at left of B11 isoforms). The trxJY16 allele is associated with a rearrangement (possibly a small inversion) breakpoint that occurs in the region encoding the third exon. The resulting fusion gene must be transcribed as determined in this analysis. Fusion mRNAs encode normal TRXI. A fusion form of TRXII is also possible that would have the N-terminal ca. 172–276 amino acids replaced by residues encoded by the fusion partner.

The nine null alleles examined are trx^6.1, trx^D, trx^B11, trx^A7, trx^M14, trx^Z15, trx^JY25, trx^7.1, and trx^Z44. They are homozygous and hemizygous embryonic lethals. TRIPOULAS et al. 1994 reported that trx^6.1 and trx^7.1 enhance homeotic transformation phenotypes in double heterozygous combination with ash1. trx^D is the Rg-bx of LEWIS 1968 used in many developmental studies. trx^B11 is associated with an 833-bp deletion. It could encode truncated proteins consisting of 8.6% of the N terminus of TRXI and 17.7% of the N terminus of TRXII (Figure 1). It has been used as a null allele in several studies. trx^A7 and trx^M14 have low penetrances of haploinsufficient phenotypes for null alleles, whereas trx^Z15 and trx^Z44 have relatively high penetrances of haploinsufficient phenotypes. trx^JY25 is associated with a Y;3 translocation, but cytological examination showed the breakpoint is distant from trx (not shown).

trx³ and trx^JY21 have relatively high penetrance of haploinsufficient phenotypes (Table 1). INGHAM 1985A reported that trx³ may have some antimorphic characteristics. This study shows trx^JY21 also may be slightly antimorphic.

Df(3R)redP52 and Df(3R)redP6 are cytologically visible deletions of the region encoding trx. Df(3R)redP52 completely removes trx and at least 10 other surrounding complementation groups. The centromere proximal break of Df(3R)redP6 maps between the second and third trx exons (Figure 1). Df(3R)redP6 removes at least 5 more distal complementation groups. It was examined because a remaining trx fusion gene could express TRX. However, Df(3R)redP6 is a trx amorph.

Quantified trx mutant phenotypes:
Reduced Scr expression in first thoracic segment (T1) leg discs can lead to transformation of ventral T1 segmental structures into homologous ventral second thoracic (T2) segmental structures (LEWIS et al. 1980 ; STRUHL 1982 ). trx mutant males were scored for 10 Scr-related transformations, and females for 8. A large anterior preapical bristle, a large posterior apical bristle, or both can develop distally on a T1 tibia. One animal can have 1 to 4 such transformations. Anterior bristle transformations occur more often than posterior bristle transformations. Genotypes that produce successively more T1 transformations increasingly produce posterior bristle transformations. Males can have reduced numbers of sex comb teeth on the basitarsus of one or both T1 legs. Proximal transformations include development of T1 sternopleural and mesosternal bristles (Figure 2B, Figure E, and Figure I). One or two of both of these structures can appear in a trx mutant.

View larger version (98K): In this window In a new window Download PPT slide   Figure 2. trithorax mutant phenotypes. All flies shown have combinations of trx alleles that are lethal by the end of pupal development. Because these flies did not eclose, their wings and halteres partially transformed to wings are not expanded. Flies in A–E are trxJY21/trxZ16, the fly in F is trxZ16/trxE3, flies in G and I are trx3/trxZ11, and the fly in H is trx6.1/trxM17. (A) The arrow indicates ventral T3 with mesosternal bristles normally found only in T2. Mesosternal bristles are also visible in ventral T1. Above the arrow is a haltere partially transformed to a wing. Anterior sternites fail to fuse at the ventral midline. (B) The arrow indicates a transformation of dorsal T3 to mesonotum typical of T2. The A3 tergite is not fused at the dorsal midline. (C) The lower arrows show sternopleural bristles in T1 and T3. These are normally found only in T2. The upper arrow indicates wing tissue that developed in association with a T1 spiracle. The A6 spiracle has unusual material protruding from it. (D) The labels 6 and 7 indicate sixth and seventh tergites of this male. Males normally have fully pigmented fifth and sixth tergites and no seventh tergite. This male has his external genitalia transformed to a leg complete with terminal claws not visible in this focal plane. (E) The arrow shows a T3 leg with a large anterior preapical bristle normally found only on T2 legs. (F) The arrow indicates the presence of a right ocellus and ocellar bristle and the absence of left and center ocelli and a left ocellar bristle. The dorsal anal plate of this female is incomplete. The posterior border of the A1 tergite has Uab-like large bristles and dark pigmentation. (G) Arrows show A2 tergites with patches of small bristles normally found only in A1 tergites. Similar patches are also found in the A3 tergites of these flies. The left fly has Uab-like dark pigmentation and large bristles at the posterior of its A1 tergite. The right fly has a small head and anterior thorax compared to the one on the left. (H) Top arrow indicates a mirror image duplication of the right eye. The lower arrow shows several abnormal bristles on the lateral labellum. (I) Ventral view of same flies as G. The right fly has complete complements of T1 sternopleural bristles. Almost all of its ventral T1 is transformed to ventral T2. The maxillary palps of the right fly did not develop. Its antennae have dpp-like abnormal outgrowths and reduced aristae.

Reduced Ubx expression in third thoracic segment (T3) haltere and leg discs can lead to transformation of T3 segmental structures into homologous T2 segmental structures (LEWIS 1963 , LEWIS 1978 ). trx mutant males and females were scored for four dorsal and eight ventral Ubx-related transformations. Dorsally, haltere disc transformations include development of wing tissue in place of normal haltere (Figure 2C and Figure G) and mesonotal tissue in place of metanotum (Figure 2B). One or both halteres can be affected in a trx mutant, and the metanotal transformation can be unilateral or bilateral. Ventrally, transformed T3 legs can have a large anterior preapical bristle, a large posterior apical bristle, or both, on a distal tibia (Figure 2E). One animal can have one to four transformed T3 leg bristles. Proximal T3 ventral transformations include development of sternopleural and mesosternal bristles (Figure 2A, Figure C, and Figure E). One or two of both of these can appear in a trx mutant. Dorsally and ventrally, transformations that affect only anterior compartment structures are more frequent than those that also include posterior compartment structures, and genotypes that produce successively more T3 transformations increasingly develop larger transformed regions that extend into posterior compartments.

Reduced abd-A expression in dorsal histoblasts can lead to abdominal tergite transformations (SANCHEZ-HERRERO et al. 1985 ). trx mutant males and females were scored for five abd-A-related transformations. Tergites of abdominal segments two through seven (A2–A7) can develop patches with small bristles and lightly pigmented cuticle normally found in A1 tergites (Figure 2G). A trx mutant can have one to six transformed tergites. However, transformations of A6 to A1 tergite were so rare that they were not scored. Transformed patches that encompass only anterior tergite are more frequent than patches that extend into posterior tergite. Genotypes that produce successively more tergite transformations have successively larger transformed patches that extend into the posterior.

Abd-B is required for normal development of adult posterior segments including A5–A7 (KARCH et al. 1985 ; SANCHEZ-HERRERO et al. 1985 ; TIONG et al. 1985 ). trx mutant males were scored for two Abd-B-related transformations, and females for one. Reduced Abd-B expression in dorsal histoblasts can cause males to develop A7 tergites (Figure 2D) that are normally suppressed by Abd-B and females to develop enlarged A7 tergites that are similar to more anterior tergites. Males normally have dark pigmentation in A5 and A6 tergites. Reduced Abd-B expression can cause loss of this pigmentation (Figure 2D), indicating transformation of underlying cuticle-secreting cells into more anterior identities. Again, transformations that encompass only anterior tergite tissue are more common than those that also include posterior tissue, and genotypes that produce successively more tergite transformations have successively larger transformed patches that include posterior tergite structures.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Penetrance and expressivity of trx homeotic phenotypes:
Reduced trx function leads to reduced expression of homeotic genes during embryogenesis (MAZO et al. 1990 ; BREEN and HARTE 1991 , BREEN and HARTE 1993 ; SEDKOV et al. 1994 ) and imaginal development (CABRERA et al. 1985 ; INGHAM 1985B ). For each trx mutant genotype that produced pharate and eclosing adults, I measured the penetrance and expressivity of that part of the trx mutant phenotype that correlates with reduced expression of homeotic genes in imaginal tissue.

Each fly of each genotype was scored for transformations associated with reduced expression of Scr, Ubx, abd-A, and Abd-B. trx mutant expressivity was measured as the total number of transformed structures, and among those described in MATERIALS AND METHODS and in Figure 3, that a fly developed. As a partial example, the number of large T1 leg preapical and apical bristles, diminished sex combs, T1 sternopleural bristle groups, and T1 mesosternal bristles a trx mutant male can develop ranges from 0 to 10 and is a measure of trx function at Scr in T1 leg discs. Males were scored for 29 possible transformed structures, and females for 26. The penetrance of each trx mutant transformation is the frequency with which it appears in flies of a particular genotype. Therefore, each genotype produces a combined penetrance and expressivity (P&E) that is expressed as the average number of transformed structures it produces per fly. P&E measurements are used to make quantitative comparisons among the genotypes (Table 2). Male and female transformation averages were combined to obtain a single average for each genotype taking into account the male:female ratio. For each genotype, male:female distributions were within expected values (not shown). P&E values for informative hypomorphic heteroallelic combinations are graphically depicted in Figure 3 Figure 4 Figure 5 Figure 6 to illustrate their different qualitative effects.

View larger version (93K): In this window In a new window Download PPT slide   Figure 3. Mutant penetrance and expressivity of hemizygous hypomorphs. Vertical bars show the percentage of flies with the indicated number of transformed structures associated with reduced expression of Scr, Ubx, abd-A, or Abd-B. The number of transformed structures for a category measures its expressivity. The percentage of flies with transformed structures in a category measures its penetrance. Transformation penetrance and expressivity infer a level of trx function affecting homeotic gene expression specifying cell fate identity. Scr group transformations. Decreased Scr function causes development of T1 structures with T2 identity. T1 to T2 leg transformations (L1/L2) include development of an anterior preapical bristle, a posterior apical bristle, or both (a.b.) on the distal tibia of one or two T1 legs and reduced numbers of sex comb teeth in males only (Scr ) on one or two legs. Other ventral transformations of T1 to T2 include development of one or two sets of T1 sternopleural bristles (T1sp) and one or two T1 mesosternal bristles (T1ms). Ubx group transformations. Decreased Ubx function leads to development of T3 structures with T2 identity. Dorsal T3 to T2 transformations include development of one or two halteres to wing (h/w) and one or two hemimetanota to hemimesonota (d3/d2). Ventral T3 to T2 transformations include development of one or two sets of T3 sternopleural bristles (T3sp), an anterior preapical bristle, a posterior apical bristle, or both on the distal tibia of one or two T3 legs (L3/L2), and one or two T3 mesosternal bristles (T3ms). abd-A group transformations. Decreased abd-A function results in development of A2–A7 structures with A1 identity. Dorsal transformations are evidenced by development of small hairs typical of an A1 tergite in more posterior tergites (An/A1, n = 2–7). So few A6 tergites were transformed, the category was excluded. Vertical bars show the percentage of each transformed tergite. Transformation of more posterior tergites correlates with increased expressivity. Abd-B group transformations. Decreased Abd-B function causes development of posterior abdominal segments with more anterior abdominal identities. Transformation phenotypes include anterior abdominal pigmentation in A5 and A6 tergites in males (A5/A4 ) and enlargement of A7 tergites (A7/A6). These phenotypes were scored for penetrance only. The genotype and number of flies examined are indicated at the top left of each chart. The chromosome indicated to the left of the slash was maternally inherited. The trxM17 allele supplies significant function at the target genes examined except Ubx in ventral structures. Other transformations can be attributed to haploinsufficiency. The trxZ32 allele is a weak hypomorph at the four loci examined. The trx1 allele is a strong hypomorph at the loci examined. It has a noticeably greater effect on abd-A expression than the other hypomorphs. This effect is also seen when trx1 is paternally inherited. The trxE3 allele is a moderate hypomorph at the loci examined. Few trxE3 hemizygotes develop to the pharate stage. Of those that do, most have unchitinized cuticles and fail to evert head structures. The trxZ16 and trxZ11 alleles are strong hypomorphs at the loci examined. trxZ16 has some residual function at Scr and Ubx compared to trxZ11.

View larger version (102K): In this window In a new window Download PPT slide   Figure 4. trxJY21 may be a weak antimorph. Bar graphs are as in Figure 3. Compare hypomorph/trxJY21 profiles to those of the same hypomorphic hemizygotes in Figure 3. Penetrance and expressivity are higher in heteroallelic combinations with trxJY21. trxE3/trxJY21 is pupal lethal and not available for comparison. Note that trxJY21 impairs the already decreased function of trx1 at abd-A. trxM18 is a stronger hypomorph than trxZ11. It may be a near amorph in imaginal cells at the loci examined.

View larger version (78K): In this window In a new window Download PPT slide   Figure 5. trx1/hypomorph heterozygotes. Bar graphs are as in Figure 3. Compare trx1/hypomorph profiles to those of hemizygous hypomorphs in Figure 3. Penetrance and expressivity are lower in heteroallelic combinations with trx1. Thus, trx1 retains significant function. The effect shown here is almost the same as when trx1 is paternally inherited. The maternal effect of trx1 is more pronounced when penetrance and expressivity of trx1/amorph heterozygotes from trx1 homozygous mothers to amorph/trx1 heterozygotes from amorph/trx+ mothers are compared (Table 2). Amorph/trx + mothers supply more trx function to eggs than trx1/trx1 mothers. trxJY16 hemizygotes die as small, third instar larvae. trxJY16 is a strong imaginal hypomorph as seen when it is heterozygous with other strong hypomorphs (Figure 6 and Table 2). trx3 is an amorph when paternally inherited, but slightly antimorphic when maternally inherited (Table 2). Assuming trx1 reduces the amount of normal TRXI and TRXII made, Abd-B and Scr functions in developing T1 legs are most easily affected by reduced levels of these proteins.

View larger version (90K): In this window In a new window Download PPT slide   Figure 6. Penetrance and expressivity of hypomorphs heterozygous with a strong hypomorph, trxZ11. Bar graphs are as in Figure 3. Compare with profiles in Figure 3 and Figure 4. Hypomorphs other than trx1 and trxE3 proportionally have a greater effect on Ubx expression in developing T3 legs than at the other loci examined in other tissues. Compare the profile for trx1/trxZ11 to those of trx1/trxJY21 heterozygotes and trx1 hemizygotes. trxZ11 supplies substantial trx function to the loci examined. Weaker hypomorphs, trxM17, trxZ32, trx1, and trxE3 complement trxZ11 to an extent expected on the basis of their penetrance and expressivity when hemizygous and in combination with trxJY21. trxZ16 complements trxZ11 more than expected on the basis of the same comparisons, suggesting it has true functional complementation of trxZ11 at Scr, Ubx, and abd-A. This effect is only seen when trxZ16 is maternally inherited. All hypomorph/trxZ11 genotypes have nearly wild-type function at abd-A.

Below, I describe phenotypes for heteroallelic combinations of eight hypomorphic mutations, nine amorphic mutations, two possible antimorphic mutations, and two deletions. Flies of most combinations that include at least one hypomorphic allele can survive from pupal stages through adulthood. Amorphic combinations are embryonic lethal. The hypomorphic trx¹ allele appears to encode normal TRX that is produced in reduced amounts. Phenotypes of flies with heteroallelic genotypes that contain trx¹ reveal that TRX is used differently among the homeotic genes examined, and it is used differently at Ubx in haltere discs compared to T3 leg discs. The relative sensitivities to reduced trx function are summarized: Abd-B > Scr > Ubx in haltere discs > abd-A > Ubx in T3 leg discs. The seven other hypomorphic alleles complement trx¹ and, to some extent, each other. This indicates that the hypomorphic alleles encode impaired proteins, and TRX cooperates at target genes. Hypomorphic mutant proteins are sufficient for embryogenesis. Compared to trx¹, the other hypomorphic mutations have reduced function at Ubx in T3 leg discs but greater function at abd-A. One hypomorphic mutation, trx^JY16, complements loss-of-trx function in haltere discs. trx^E3 has a marked effect on Scr and produces head and growth defects. These observations show that TRX has unique interactions at different target genes. trx^M17, trx^Z11, and trx^E3 produce unusual phenotypes reminiscent of those produced by hypomorphic signal transduction mutations.

Effects of trx¹:
The ~9-kb insert associated with trx¹ is located outside coding sequences. It is likely that the mutant gene produces normal TRXI and TRXII but at reduced levels due to either impaired RNA processing, such as splicing of the first intron, or impaired translation of an mRNA with an unspliced first intron. Consequently, the effect of trx¹ is probably quantitative, and it supplies insufficient TRX to accumulate properly at affected target genes. This is consistent with the findings of CHINWALLA et al. 1995 , who showed that fewer TRX-binding sites on polytene chromosomes are occupied by TRX in trx¹/Df(3R)redP52 mutants compared with trx¹/trx¹ mutants.

trx¹ has measurably different effects on Abd-B, Scr, abd-A, Ubx in haltere discs and Ubx in T3 leg discs compared with effects produced by genotypes with high P&E values (Figure 3 and Figure 4). Abd-B is most sensitive to reduced levels of TRX in trx¹ hemizygotes followed by Scr, Ubx in haltere discs and abd-A, and Ubx in T3 leg discs. This suggests concentration-dependent differences in the ability to assemble or maintain sufficient TRX at different PREs and in different tissues.

The seven other trx hypomorphic alleles complement trx¹. Transheterozygotes of these alleles with trx¹ have significantly lower P&E values than trx¹ homozygotes and hemizygotes (Table 2). P&E values of trx¹/hypomorph transheterozygotes form an increasing series that suggests progressively decreasing function among the hypomorphic alleles: trx^M17 > trx^Z32 > trx^Z16 > trx^Z11 > trx^M18 = trx^JY16 > trx^E3. As hypomorphic complementation decreases in this series, Abd-B is most sensitive to declining trx function, followed by Scr and, to a lesser extent, Ubx and abd-A (Figure 5). This agrees with the relative sensitivities of the target genes seen in trx¹ hemizygotes. It is also consistent with previous reports of trx^- haploinsufficient phenotypic frequencies (CAPDEVILA and GARCIA-BELLIDO 1981 ; CAPDEVILA et al. 1986 ; CASTELLI-GAIR and GARCIA-BELLIDO 1990 ; T. R. BREEN, unpublished results) that show that transformations associated with decreased Abd-B and Scr functions are most frequent, followed by transformations associated with decreased Ubx and abd-A functions. Slight differences in target gene sensitivities seen among trx¹/hypomorph transheterozygotes can be attributed to allele-specific effects seen in hypomorph hemizygotes (Figure 3).

The hypomorphic alleles must encode proteins that cooperate with normal TRX supplied by trx¹ in imaginal cells. Hemizygous P&E values for trx^Z16, trx^Z11, trx^M18, and the larval lethality of trx^JY16 (Table 2) show that they are strong hypomorphs, yet these alleles provide substantial imaginal trx function in combination with trx¹ (Table 2) except at Abd-B (Figure 5). Thus, the defective proteins encoded in the hypomorphic alleles must be present in sufficient quantity, together with low levels of normal TRX, to form and maintain functional TRX structures at many PREs other than those associated with Abd-B.

GARCIA-BELLIDO and CAPDEVILA 1978 , INGHAM and WHITTLE 1980 , and CAPDEVILA and GARCIA-BELLIDO 1981 recognized that reduced trx function during imaginal cell determination and proliferation produces patches of tissue with cell identity transformations that occur most frequently in anterior compartments of segments. In this study, all heteroallelic combinations with trx¹ similarly produce patchy transformations that occur most frequently in anterior compartments (not shown; see MATERIALS AND METHODS). All other hypomorphic genotypes also produce transformed patches, but a greater percentage of them extend into posterior compartments than are seen in trx¹ genotypes. Furthermore, genotypes that produce higher P&E values concomitantly produce larger transformed patches, sometimes encompassing most of a segment (Figure 2).

INGHAM and WHITTLE 1980 and INGHAM 1985A suggested that transformed patches are clones derived from progenitor cells that have lost the ability to transmit functional TRX structures to their descendants. The preponderance of anterior compartment transformations produced by trx¹ genotypes suggests that production of heritable or functional TRX assemblies in anterior cells is more susceptible to reduced TRX concentration than in posterior cells. Strong hypomorphic hemizygotes can form some functional and heritable TRX structures (Figure 3), but they are more often lost during early imaginal cell proliferation and increasingly in posterior cells. This is seen in the increased penetrance of flies with three and four transformed leg bristles that include posterior apical bristles (Figure 3). These flies also show increased penetrance of posterior haltere to wing transformations and posterior tergite transformations (Figure 2).

Effects of other hypomorphic alleles:
trx^M17, trx^Z32, trx^Z16, trx^Z11, and trx^M18 in heteroallelic combination with amorphic or antimorphic alleles have proportionally less function at Ubx in T3 leg discs than at Ubx in haltere discs and at Scr, abd-A, and Abd-B in their tissues compared with equivalent trx¹ and trx^E3 heteroallelic combinations (Figure 3 and Figure 4). They have a relatively small effect on abd-A expression compared with that caused by trx¹ genotypes. These effects are also evident in flies with heteroallelic combinations of hypomorphic alleles as exemplified in Figure 6. trx^JY16 shows a slightly more exaggerated example of this profile in heteroallelic combinations with trx^M17, trx^Z32, trx^Z16, trx^Z11, or trx^M18, as demonstrated in Figure 6.

trx^M17, trx^Z32, trx^E3 trx^Z16, trx^Z11, trx^M18, and trx^JY16 complement each other to varying extents (Table 2 and compare Figure 3 and Figure 4 with Figure 6), but almost all combinations supply sufficient function at abd-A for normal tergite development. The complementation supplied by these defective proteins shows that they can cooperate at PREs. Differing levels of complementation among the hypomorphic genotypes reflect that the mutant proteins have different abilities to assemble at PREs, associate with each other, or interact with other factors to influence target gene transcription.

Molecular lesions associated with trx^Z16, trx^Z11, and trx^JY16 affect different regions of TRX (Figure 1), yet they have similar proportional effects on Scr, Ubx, abd-A, and Abd-B in imaginal cells. It is likely that mutations associated with these and some of the other hypomorphic alleles disrupt separate functions that have the same consequence on development. This is supported by the ability of trx^M17, trx^Z32, trx^E3, and trx^Z16 to complement trx^Z11 (Figure 6), suggesting that one mutant protein supplies the missing function of the other and vice versa. Other cases of heteroallelic complementation can be gleaned from Table 2. trx^M18/trx^Z11 and trx^JY16/trx^Z11 are examples of genotypes for which it is difficult to determine if there is complementation. For point mutations, lack of complementation may indicate both alleles affect the same functional domain. Other combinations are more difficult to interpret.

trx^JY16 homozygotes and hemizygotes die as small, lethargic third instar larvae. They do not have obvious segmental transformations, but their tracheae are convoluted and may be disjointed (not shown) as can occur when decapentaplegic protein (DPP) or epidermal growth factor (EGF) tracheal signaling is disrupted (VINCENT et al. 1997 ; WAPPNER et al. 1997 ). Homozygotes or hemizygotes of the other hypomorphic alleles develop through the third larval instar with no obvious defects. Thus, all eight hypomorphic alleles supply sufficient trx function for normal embryogenesis with one dose of maternally supplied, wild-type trx, and only trx^JY16 is insufficient for normal larval development. It is unknown how imaginal cell proliferation is affected in trx^JY16 homozygotes and hemizygotes. As stated above, trx^JY16 in combination with trx^M17, trx^Z32, trx^Z16, trx^Z11, or trx^M18 disproportionally complements Ubx in haltere discs compared with equivalent heteroallelic combinations of the other hypomorphs, as typified in Figure 6. This reveals a different use of TRX to regulate Ubx in haltere vs. T3 leg discs.

trx^E3 hemizygotes occasionally develop to pharate adults. More often they die as pupae with some head and thorax chitinization but little in the abdomen, and their heads often fail to evert. In hemizygotes, trx^E3 has a proportionally greater effect on Scr than Ubx and abd-A compared with other hypomorphic alleles (Figure 3). Its effect on Scr is more evident in transformed distal T1 leg structures (preapical and apical bristles and sex combs) than in proximal structures (sternopleural and mesosternal bristles). trx^E3 also proportionally provides greater complementation at Ubx than at Scr in heteroallelic combination with other hypomorphic alleles, as seen in Figure 6. These observations are consistent with those of SEDKOV et al. 1994 who noted that trx^E3 reduces ANT-C, but not BX-C, expression during embryogenesis. Pharate adult trx^E3 hemizygotes do have reduced BX-C expression, but much of it may be attributed to haploinsufficiency. This seems likely because trx^E3 substantially complements reduced Ubx, abd-A, and Abd-B function caused by other hypomorphic alleles (Figure 5 and Figure 6). The ability of two doses of trx^E3 to complement BX-C function more fully cannot be observed because trx^E3 homozygotes die as embryos. This lethality is probably not caused solely by trx^E3 because trx^E3 hemizygotes develop to pharate adults. As noted below, trx^E3 genotypes frequently have anterior dorsal eye-disc defects including missing ocelli, ocellar bristles, and postvertical bristles. This phenotype is not associated with ANT-C function (MERRILL et al. 1987 , MERRILL et al. 1989 ; PULTZ et al. 1988 ), but it may be related to deficient epidermal growth factor receptor (EGFR; CLIFFORD and SCHUPBACH 1989 ; FINKELSTEIN et al. 1990 ; GABAY et al. 1996 ) or hedgehog (hh) protein (HH) signaling (ROYET and FINKELSTEIN 1996 ) in eye imaginal discs.

Effects of amorphic and antimorphic alleles:
trx^6.1, trx^D, trx^B11, trx^A7, trx^M14, trx^Z15, trx^JY25, trx^7.1, and trx^Z44 homozygotes and hemizygotes die as embryos (Table 2). Inter se combinations of these alleles are also embryonic lethal. trx^6.1, trx^B11, trx^A7, and trx^M14 variably complement trx¹ (Table 2), but otherwise behave as amorphic alleles. Thus, these alleles may encode defective proteins that supply some imaginal function with wild-type TRX, though the complementation of trx^B11, trx^A7, and trx^M14 is marginal enough that it may be attributed to background effects (CAPDEVILA and GARCIA-BELLIDO 1981 ; T. R. BREEN, unpublished results).

trx^6.1 provides greater complementation of trx¹ than do hypomorphic alleles other than weak trx^M17 (Table 2). Thus, trx^6.1 mutant proteins probably form functional TRX structures with normal TRX supplied zygotically and maternally by trx¹ or zygotically by trx¹ and maternally by one dose of trx⁺ from trx^6.1/TM6B mothers. Proteins encoded in trx^6.1 do not complement functions affected by the hypomorphic alleles other than trx¹. trx^6.1 may encode a truncated protein that easily interacts with intact TRX but cannot compensate for a defective function.

trx^D and trx^A7 weakly complement trx^M17 (Table 2). They proportionally rescue trx^M17 function (not shown). Thus, defective proteins encoded in trx^D and trx^A7 may weakly interact with deficient trx^M17 proteins, but they are unable to rescue mutations that remove successively greater trx function.

trx^JY21/hypomorph genotypes typically have higher P&E values (Table 2) than equivalent amorphic genotypes, suggesting trx^JY21 is slightly antimorphic. Furthermore, trx^JY21 genotypes often produce phenotypes not used for P&E analysis (Figure 2; see below) that are not seen in hypomorphic hemizygotes. trx^JY21 genotypes also produce larger transformed patches, sometimes encompassing entire structures (Figure 2, A–D), than equivalent amorphic genotypes. Indeed, trx^M18/trx^JY21 pharate adults have nearly completely transformed ventral T1 and T3 structures, suggesting little or no trx function at Scr in T1 leg discs or at Ubx in T3 leg discs. In combination with trx¹, trx^JY21 disproportionally reduces Scr expression compared with what is seen in trx¹ hemizygotes (compare Figure 3 and Figure 4). trx^JY21 similarly reduces the function of strong hypomorphic alleles at Scr and Ubx, but not at abd-A.

In a few combinations with hypomorphic alleles, trx³ produces higher P&E values than equivalent amorphic genotypes; otherwise it behaves as an amorphic allele. Like trx^JY21, trx³ genotypes often produce phenotypes not seen in hypomorphic hemizygotes, including male genitalia to leg, humerus to wing, and others mentioned below. By this criterion, trx³ may be slightly antimorphic.

Additional trx mutant phenotypes:
Some trx mutant genotypes produce unusual phenotypes similar to those reported by INGHAM and WHITTLE 1980 or that develop in amorphic trx mutant clones (INGHAM 1981 , INGHAM 1985A ).

Flies transheterozygous for the weakly antimorphic trx^JY21 or trx³ and any of the strong hypomorphs trx^Z11, trx^Z16, or trx^M18 occasionally develop wing tissue adjacent to their prothoracic spiracles (Figure 2C). This is consistent with decreased Scr expression in humeral disc anterior compartments (INGHAM and WHITTLE 1980 ; INGHAM 1985A ; ROGERS et al. 1997 ). Males of the same genotypes occasionally develop T2 legs from their genital discs (Figure 2D) that may be caused by a relative increase in Antp expression compared with Abd-B r expression in male genital discs (CASARES et al. 1997 ).

Flies transheterozygous for trx^E3 and one of the strong hypomorphs frequently have missing dorsal head structures including different combinations of ocelli, ocellar bristles, and postvertical bristles (Figure 2F). This phenotype is occasionally seen in amorph/trx^Z11 pharate adults, too. The dorsal head phenotype is similar to that seen in Egfr (Drosophila EGF receptor) and ocelliless (oc) hypomorphs (CLIFFORD and SCHUPBACH 1989 ; FINKELSTEIN et al. 1990 ). Many trx^E3 hemizygotes develop as incomplete pharate adults whose heads fail to evert (and appear headless). They also have incomplete chitinization. Homeotic phenotypes could not be scored for these animals and they did not contribute to the P&E values in Table 2. Three of five trx³/trx^Z11 pharate adults developed small heads and anterior thoracic segments (Figure 2G and Figure I). They also had abnormal antennae (Figure 2I) similar to decapentaplegic (dpp) disc III mutants (SPENCER et al. 1982 ) and reduced or absent maxillary palps that may be associated with reduced Dfd, proboscipedia (pb), or labial (lb) expression in antennal discs (KAUFMAN 1978 ; MERRILL et al. 1987 , MERRILL et al. 1989 ; PULTZ et al. 1988 ). A total of 6 out of 17 trx^M18/trx^JY21 pharate adults also had small heads.

Flies transheterozygous for trx^M17 and amorphic alleles (Figure 2H) occasionally develop large bristles on their labial palps that may be associated with reduced pb and Scr expression in labial discs (KAUFMAN 1978 ; PULTZ et al. 1988 ; PERCIVAL-SMITH et al. 1997 ). They also develop mirror image eye duplications that have been seen in trx¹ homozygotes from trx¹ homozygous mothers (T. R. BREEN, unpublished results). Rarely, trx^M17/amorph flies develop posterior wing abnormalities and anterior wing duplications (not shown). Similar posterior wing disruptions are seen in en mutant clones (MORATA and LAWRENCE 1975 ; LAWRENCE and STRUHL 1982 ) and in mutant clones that remove function downstream of dpp protein (DPP) signaling (SINGER et al. 1997 ). Similar anterior wing duplications are seen also in DPP receptor mutant clones (PENTON and HOFFMANN 1996 ; SINGER et al. 1997 ).

Some phenotypes described above probably arise due to overall lack of trx function at target genes such as is produced by combinations of trx^JY21 with strong hypomorphs. Others may be due to altered trx function at specific target genes or in specific cell types such as those associated with trx^E3, trx^Z11, trx^M17, and perhaps trx^M18.

Many trx mutant genotypes produce additional phenotypes. Frequently, A1 tergites develop with dark pigmentation and large bristles at their posterior borders (Figure 2F and Figure G), similar to Ultra-abdominal (Uab) phenotypes (LEWIS 1978 ; KARCH et al. 1985 ). Other frequent phenotypes include incomplete dorsal fusion of tergites (Figure 2B and Figure F), abnormal abdominal spiracles (Figure 2C), and abnormal sternites (Figure 2A). These phenotypes are associated with decreased abd-A and Abd-B function in abdominal histoblasts (KARCH et al. 1985 ).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Comparative effects of trx¹ on homeotic phenotypes:
The P&E profiles of trx¹ genotypes suggest different quantitative requirements for TRX at the four homeotic loci examined. Three factors may contribute to the different sensitivities of target genes to decreased levels of TRX: (1) Each target gene accumulates a unique amount of TRX at its PRE(s). Genes that normally accumulate less TRX may be more sensitive to decreased levels of TRX. (2) Target genes with normally equal TRX accumulation may have different threshold TRX levels below which they no longer function for proper structural determination. (3) Different tissues undergo different numbers of cell divisions, which may lead to differential loss of limiting quantities of TRXs. These three factors are further described below.

1: Polytene chromosomal analyses show that different target genes do accumulate different amounts of TRX at their PREs (KUZIN et al. 1994 ; CHINWALLA et al. 1995 ). The level of accumulation of TRX proportionally decreases at all target genes in trx¹ mutants, which shows that target genes that normally accumulate more TRX recruit limiting amounts of TRX more efficiently. Even if there is not a linear relationship between the amount of TRX accumulated at a target gene and the amount needed for function, target genes that normally accumulate less TRX should be more susceptible to reduced nuclear TRX concentration than those that normally accumulate more. Due to low resolution, polytene chromosomal analysis did not determine if Scr in the ANT-C and Ubx, abd-A, and Abd-B in the BX-C accumulate different amounts of TRXs (CHINWALLA et al. 1995 ).

2: Some trx target genes in specific tissues may require relatively small amounts of TRX to be transcribed sufficiently for normal development, though they may normally be supplied with abundant TRX. Only very low TRX levels would elicit a mutant phenotype in such tissues. Reduced trx function would produce low P&E of phenotypes associated with such a relatively TRX-insensitive target gene.

3: trx¹ mutants have reduced maternal and/or zygotic production of TRX. Thus, target genes in imaginal precursor cells of trx¹ mutants initially accumulate less TRX than those in wild-type cells. During subsequent cell divisions, target genes are increasingly susceptible to insufficient TRX accumulation caused by continually impaired trx transcription or translation. Different imaginal tissues begin with different numbers of precursor cells and proliferate to different extents (COHEN 1993 ). Target genes expressed in imaginal tissues with more precursor cells and greater proliferation would be more likely to have TRX accumulation fall below threshold levels than those in tissues with fewer precursors and less proliferation. This scenario requires that all imaginal precursor cells initially have similarly reduced levels of TRX and all proliferating imaginal cells have the same reduced transcription or translation of trx.

Comparative effects of other hypomorphic alleles on homeotic phenotypes:
Hypomorphic trx alleles encode proteins that can assemble and provide some function at target genes. They function in combination with one wild-type maternal dose of trx for seemingly normal embryogenesis, which is consistent with their complementation of trx¹ during imaginal development. However, a maternal dose of trx⁺ is not sufficient to complement hypomorphic mutant function in imaginal precursors whose progeny cells produce only mutant protein.

P&E qualitative profiles produced by different hypomorphic genotypes, excluding trx¹ and trx^E3, are proportionally similar. This suggests that the mutant proteins equivalently impair TRX function at the homeotic genes examined. However, instances of complementation among hypomorphic alleles suggest different mutations alter different functional domains. These observations infer, not surprisingly, that different hypomorphic mutant proteins inefficiently interact with different factors present at many, if not all, target genes.

trx^Z16 and trx^Z11 are associated with point mutations in the PHD finger and SET domains, respectively (STASSEN et al. 1995 ). As mentioned previously, the SET domain of HRX at least mediates protein-protein interactions used in signal transduction and maturation (CUI et al. 1998 ; DE VIVO et al. 1998 ). PHD fingers may also mediate protein-protein interactions (AASLAND et al. 1995 ). trx^M17, trx^Z32, and trx^M18 are point or pseudopoint mutations (BREEN and HARTE 1991 ) that impair different functional interactions as determined by complementation. Disruption of any TRX protein-protein interaction would reduce the ability of a mutant protein to function at any target gene. Results of this study show that Ubx transcription in T3 leg discs is most easily affected by any of several small changes in different regions of TRX. The observed different sensitivities of trx target genes to a variety of mutant TRX infers that each target