Reduced Fertility of Drosophila melanogaster Hybrid male rescue (Hmr) Mutant Females Is Partially Complemented by Hmr Orthologs From Sibling Species

Abstract

The gene Hybrid male rescue (Hmr) causes lethality in interspecific hybrids between Drosophila melanogaster and its sibling species. Hmr has functionally diverged for this interspecific phenotype because lethality is caused specifically by D. melanogaster Hmr but not by D. simulans or D. mauritiana Hmr. Hmr was identified by the D. melanogaster partial loss-of-function allele Hmr¹, which suppresses hybrid lethality but has no apparent phenotype within pure-species D. melanogaster. Here we have investigated the possible function of Hmr in D. melanogaster females using stronger mutant alleles. Females homozygous for Hmr mutants have reduced viability posteclosion and significantly reduced fertility. We find that reduced fertility of Hmr mutants is caused by a reduction in the number of eggs laid as well as reduced zygotic viability. Cytological analysis reveals that ovarioles from Hmr mutant females express markers that distinguish various stages of wild-type oogenesis, but that developing egg chambers fail to migrate posteriorly. D. simulans and D. mauritiana Hmr⁺ partially complement the reduced fertility of a D. melanogaster Hmr mutation. This partial complementation contrasts with the complete functional divergence previously observed for the interspecific hybrid lethality phenotype. We also investigate here the molecular basis of hybrid rescue associated with a second D. melanogaster hybrid rescue allele, In(1)AB. We show that In(1)AB is mutant for Hmr function, likely due to a missense mutation in an evolutionarily conserved amino acid. Two independently discovered hybrid rescue mutations are therefore allelic.

DROSOPHILA melanogaster can hybridize to three sister species: D. simulans, D. mauritiana, and D. sechellia (Lachaise et al. 1986). F₁ hybrid males from the cross of D. melanogaster females to males from any one of these sister species are lethal, dying as late-stage larvae or early-stage undifferentiated pupae (Sturtevant 1920, 1929; Sanchez and Dübendorfer 1983; Hutter et al. 1990). This loss of fitness in F₁ hybrids is one characteristic of the reproductive isolation that defines our understanding of species (Dobzhansky 1937; Mayr 1942; Muller 1942).

The D. melanogaster Hmr¹ mutation was discovered because it can partially suppress the lethality of these F₁ hybrid males (Hutter and Ashburner 1987). Hmr has been cloned and shown to correspond to a single gene that is predicted to encode an ∼1400-amino-acid protein with multiple MADF domains (Myb/SANT-like domain in Adf1; Adf1 is Adh transcription factor 1) that may have DNA or chromatin binding functions (Barbash et al. 2003; Maheshwari et al. 2008). Hmr orthologs from the sibling species of D. melanogaster were found to have a high rate of average per-site nonsynonymous divergence (D_N > 8%) and further analyses showed that the pattern of divergence and polymorphism at Hmr is consistent with the action of positive selection (Barbash et al. 2003, 2004a).

This discovery of adaptive evolution raises the question of what phenotype is under selection that is driving the divergence of Hmr between species, and whether Hmr function is conserved among D. melanogaster and its sibling species. The D. melanogaster Hmr¹ mutation is viable and fertile in both sexes, and homozygous stocks can be easily maintained (Hutter and Ashburner 1987). Genetic and molecular analyses indicate, however, that Hmr¹ is a partial loss-of-function (hypomorphic) allele (Hutter et al. 1990; Barbash et al. 2000, 2003). This allele may therefore not have an apparent phenotype within D. melanogaster because it retains partial Hmr⁺ activity.

Here we use stronger Hmr alleles to investigate Hmr function within D. melanogaster. We find that Hmr is required for wild-type levels of viability and fertility in adult females. We then use this discovery to address whether D. simulans and D. mauritiana Hmr⁺ orthologs that have been transformed into D. melanogaster can complement these phenotypes.

In(1)AB is a second D. melanogaster rescue mutation that also suppresses F₁ male lethality (Hutter et al. 1990). Genetic analyses demonstrated that it has stronger rescue activity than Hmr¹, leading to the speculation that it might be a stronger loss-of-function Hmr allele (Hutter et al. 1990; Barbash et al. 2000). However, Hmr appears to be expressed at wild-type levels in an In(1)AB strain (Barbash et al. 2003). Two amino acid differences were found in the Hmr coding sequence compared to wild-type D. melanogaster strains but it is unknown if they are responsible for the rescue activity of In(1)AB (Barbash et al. 2003). A nonrescuing chromosome parental to In(1)AB is unfortunately not available because, as wistfully noted by Hutter et al. (1990, p. 916), the record of its origin is “lost in the mists of time.” Here we test whether the two amino acid variants encoded by the Hmr allele carried on In(1)AB eliminate Hmr function.

MATERIALS AND METHODS

Drosophila stocks and nomenclature:

Hmr¹ is a hypomorphic allele caused by a P-element insertion (Hutter and Ashburner 1987; Barbash et al. 2003). Df(1)Hmr⁻ is an imprecise excision of Hmr¹ that deletes the first 96 codons of the predicted coding sequence (Barbash and Lorigan 2007). Hmr^1r1 is a precise excision of Hmr¹ that was derived from the same screen as Df(1)Hmr⁻ and has lost hybrid rescue activity (Barbash and Lorigan 2007). We propose on the basis of evidence in this study that the In(1)AB hybrid rescue chromosome discovered by Hutter et al. (1990) contains a loss-of-function mutation in Hmr. We designate this allele as Hmr² and the chromosome as In(1)AB, Hmr². The stock y¹ w^67c23 P{w^+mC y^+mDint2=EPgy2}EY12237 was generated by the Berkeley Drosophila Genome Project Gene Disruption Project (Bellen et al. 2004) and obtained from the Bloomington Stock Center. Available sequence data from inverse PCR indicates that it is inserted 32 bp 5′ to the transcription start site of Hmr, at precisely the same position as the Hmr¹ insertion (Barbash et al. 2003). Data presented below demonstrate that this insertion is a loss-of-function allele of Hmr, and we refer to it as P{EPgy2}Hmr³. The p72 and p73 transgenes are described in Barbash et al. (2003). P{Dsim\Hmr^+t8.6} and P{Dmau\Hmr^+t9.4} transgenic lines are described in Barbash et al. (2004a).

Construction of Df(1)Hmr⁻ stocks containing transgenes:

Transgenes are referred to as P[w⁺] in the following sections. All transgenic lines used contain autosomal insertions. Df(1)Hmr⁻, y w v/FM6,w virgin females were crossed to FM6,w/Y; P[w⁺]/+ males. Df(1)Hmr⁻, y w v/FM6,w; P[w⁺]/+ virgin daughters were then crossed to FM6,w/Y males, and this backcrossing scheme continued for three more generations, to generate stocks referred to as backcross 4 (BC4).

These BC4 stocks were maintained by selecting in every generation for Df(1)Hmr⁻, y w v/FM6,w females carrying the transgene. Because the Hmr mutation remained heterozygous this scheme allowed the lines to be maintained without potential selection for increased viability and fertility of Df(1)Hmr⁻ homozygous females.

Female viability and fertility tests:

For crosses 1–8 in Table 2, BC4 females of genotype Df(1)Hmr⁻, y w v/FM6,w; P[w⁺]/+ were crossed to Df(1)Hmr⁻, y w v/Y males. Virgin daughters of genotypes Df(1)Hmr⁻, y w v/Df(1)Hmr⁻, y w v; P[w⁺]/+ and Df(1)Hmr⁻, y w v/Df(1)Hmr⁻, y w v; +/+ were collected and aged for 2–3 days at 27° and the rest of the experiment was carried out at 27°. Single females were crossed to two wild-type males (Oregon-R stock) on day 0 of the experiment. Vials were checked on days 5, 10, 15, and 20. If the female and at least one of the males were alive then the flies were transferred to a new vial, the old vial was kept, and all progeny counted. If the female was dead then she was recorded as such and the vial was discarded without counting any progeny. If both males were dead then the vial was discarded and considered as missing data, regardless of whether the female was alive. If the flies escaped during transfer to a new vial, then this new vial was considered as missing data. Because of these last two conditions, the number of females assayed (N) of each genotype could drop during the course of the experiment. On day 25 the vials were examined as described above except that no new transfer was made and the parents were discarded. To generate the genotypes tested in Table 3, In(1)AB, w/FM6 females were separately crossed to Df(1)Hmr⁻, y w v/Y and to Hmr^1r1/Y males. The experiment was performed identically to that described above except that males from a Canton-S stock were used.

TABLE 2.

Tests for complementation of Df(1)Hmr⁻ female viability and fertility defects

	Genotype (no. of females at start)	Viability relative to Na (mean no. of progeny per female ± SD)b
Cross			Day 5	Day 10	Day 15	Day 20	Day 25
1	Df(1)Hmr−/Df(1)Hmr−	0.74	0.32	0.08	0.03	0.01
	(N = 73)	(6.2 ± 11.9)	(1.22 ± 3.9)	(0)	(0)	(0)
	Df(1)Hmr−/Df(1)Hmr−; P{Hmr+tp72}11-1/+	0.70	0.42	0.31***	0.22***	0.14**
	(N = 73)	(54.5 ± 33.3)***	(43.9 ± 23.8)***	(52.9 ± 19.1)***	(49.6 ± 22.4)**	(55.8 ± 35.4)
2	Df(1)Hmr−/Df(1)Hmr−	0.77	0.40	0.22	0.05	0.02
	(N = 65)	(14.6 ± 15.3)	(5.4 ± 7.6)	(0.07 ± 0.3)	(0)	(0)
	Df(1)Hmr−/Df(1)Hmr−; P{Hmr+tp72}16-1/+	0.89	0.72***	0.67***	0.56***	0.38***
	(N = 61)	(54.0 ± 25.5)***	(48.6 ± 28.5)***	(56.7 ± 26.0)***	(50.4 ± 28.8)**	(67.0 ± 25.5)
3	Df(1)Hmr−/Df(1)Hmr−	0.18	0.05	0.03	0	0
	(N = 74)	(7.4 ± 10.0)	(0)	(0)
	Df(1)Hmr−/Df(1)Hmr−; P{Hmrtp73}2-2/+	0.24	0.09	0	0	0
	(N = 74)	(4.6 ± 5.0)	(0)
4	Df(1)Hmr−/Df(1)Hmr−	0.29	0.06	0.01	0	0
	(N = 108)	(1.4 ± 3.5)	(1.0 ± 2.5)	(0)
	Df(1)Hmr−/Df(1)Hmr−; P{Hmrtp73}5-1/+	0.32	0.11	0.04	0	0
	(N = 105)	(0.6 ± 1.7)	(0.09 ± 0.3)	(0)
5	Df(1)Hmr−/Df(1)Hmr−	0.61	0.40	0.24	0.11	0
	(N = 75)	(2.9 ± 5.3)	(0.1 ± 0.4)	(0)	(0)
	Df(1)Hmr−/Df(1)Hmr−; P{Dsim\Hmr+t8.6}2-4/+	0.64	0.45	0.36	0.18	0.10**
	(N = 75)	(17.6 ± 16.4)***	(5.9 ± 8.8)***	(1.6 ± 7.3)	(0)	(0)
6	Df(1)Hmr−/Df(1)Hmr−	0.68	0.46	0.28	0.22	0.14
	(N = 73)	(7.6 ± 14.8)	(0.1 ± 0.5)	(0)	(0)	(0)
	Df(1)Hmr−/Df(1)Hmr−; P{Dsim\Hmr+t8.6}4-1/+	0.70	0.58	0.40	0.32	0.22
	(N = 73)	(16.3 ± 15.5)***	(3.9 ± 9.9)*	(0)	(0)	(0)
7	Df(1)Hmr−/Df(1)Hmr−	0.71	0.46	0.33	0.22	0.14
	(N = 70)	(4.1 ± 7.2)	(0.2 ± 0.6)	(0)	(0)	(0)
	Df(1)Hmr−/Df(1)Hmr−;P{Dmau\Hmr+t9.4}3-4/+	0.80	0.63*	0.54*	0.46*	0.32*
	(N = 70)	(20.6 ± 16.2)***	(10.4 ± 16.0)***	(0.87 ± 3.2)	(0)	(0)
8	Df(1)Hmr−/Df(1)Hmr−	0.59	0.40	0.26	0.10	0
	(N = 70)	(4.3 ± 8.3)	(0.9 ± 3.2)	(0.3 ± 1.2)	(0)
	Df(1)Hmr−/Df(1)Hmr−; P{Dmau\Hmr+t9.4}1-5/+	0.51	0.32	0.26	0.10	0.04
	(N = 74)	(16.4 ± 19.9)***	(4.7 ± 11.1)*	(0.5 ± 1.8)	(0)	(0)

For each cross the two genotypes being compared were siblings from a single cross. Single females of the indicated genotypes were mated to two wild-type males. The females were examined for viability and the parents changed to fresh vials on days 5, 10, 15, and 20. Vials were also examined on day 25 and any remaining females discarded. All vials in which the female was alive when examined were kept and total progeny counted. All crosses were done at 27°. See materials and methods for further details.

^aN can drop during the course of the experiment due to death of the tester males or accidental escape of the flies. N went down slightly (<10%) in all experiments, except for the Df(1)Hmr⁻, y w v/Df(1)Hmr⁻, y w v; P{Hmr^+tp72}16-1/+ class in cross 2, where N dropped from 61 to 53 during the course of the experiment. The difference in the proportions of live and dead females between the w⁺ and w classes was tested by a one-tailed Fisher's exact test. *P < 0.05; **P < 0.01; ***P < 0.001, respectively.

^bFor average fertility, the difference between the w⁺ and w classes was tested by a one-tailed t-test. *P < 0.05; **P < 0.01; ***P < 0.001, respectively. SD, standard deviation.

TABLE 3.

Test for complementation of Df(1)Hmr⁻ female viability and fertility defects by In(1)AB

Genotype (no. of females at start)	Viability relative to Na (mean no. of progeny/female ± SD)b
		Day 5	Day 10	Day 15	Day 20	Day 25
In(1)AB,w/Hmr1r1	0.90	0.86	0.82	0.80	0.68
(N = 92)	(78.5 ± 14.8)	(51.6 ± 19.2)	(33.3 ± 21.1)	(10.6 ± 15.6)	(1.3 ± 4.1)
In(1)AB,w/Df(1)Hmr−	0.94	0.87	0.84	0.75	0.68
(N = 88)	(31.4 ± 18.4)***	(3.9 ± 5.6)***	(0.12 ± 0.56)***	(0)***	(0)*

Single females of the indicated genotypes were mated to two wild-type males and processed as in Table 2. Crosses were done at 27°.

^aN dropped during the course of the experiment [to 66 at day 25 in the crosses with In(1)AB, w/Hmr^1r1 and to 71 at day 25 in the crosses with In(1)AB, w/Df(1)Hmr⁻], primarily due to lethality of the males. The difference in the proportions of live and dead females between the w⁺ and w classes was tested by a one-tailed Fisher's exact test. All tests were nonsignificant (P > 0.05).

^bFor average fertility, the difference between the w⁺ and w classes was tested by a one-tailed t-test. *P < 0.05; **P < 0.01; ***P < 0.001, respectively. SD, standard deviation.

For the experiment shown in Table 4, y w P{EPgy2}Hmr³/y w P{EPgy2}Hmr³ females were crossed to w¹¹¹⁸ males. Heterozygous females were then crossed to y w P{EPgy2}Hmr³/Y males. Heterozygous and homozygous females were distinguished on the basis of the darker eye color of homozygotes and crosses set up with a single female and two w¹¹¹⁸/Y males. Vials were checked and changed every 5 days until day 25. If either the female or both males died the vial was discarded and considered as missing data. Because female mortality was not specifically recorded we could assay potential female fertility effects but not female viability effects in this experiment.

TABLE 4.

Test for fertility defects in P{EPgy2}Hmr³

	Mean no. of progeny/female ± SDa (no. of vials tested)b
Genotype (no. of females at start)	Day 5	Day 10	Day 15	Day 20	Day 25
P{EPgy2}Hmr3/+	52.2 ± 28.1	36.9 ± 21.1	21.7 ± 17.1	13.5 ± 15.2	3.1 ± 5.0
(N = 146)	(97)	(84)	(65)	(49)	(15)
P{EPgy2}Hmr3/P{EPgy2}Hmr3	16.4 ± 18.7***	12.5 ± 17.5***	4.6 ± 6.7***	1.6 ± 3.7***	0*
(N = 86)	(49)	(35)	(30)	(21)	(8)

Single females of the indicated genotypes were mated to two w¹¹¹⁸/Y males. Vials were changed and examined on days 5, 10, 15, 20 and 25 as described in materials and methods. Crosses were done at 27°. SD, standard deviation.

^aFor average fertility, the difference between the w⁺ and w classes was tested by a one-tailed t-test. *P < 0.05; **P < 0.01; ***P < 0.001, respectively.

^bN dropped during the course of the experiment due to death of the female and/or male parents.

Analyses of fertility defects and morphology of eggs and ovaries:

All experiments in this section were performed at 28°. To estimate fertility of Hmr⁻ and Hmr⁺ control females as shown in Table 5 and to examine egg morphology, between 19 and 40 2- to 3-day-old virgin females of each genotype were mated to between 30 and 45 w¹¹¹⁸/Y males. The next morning cultures were placed into egg-collection containers with 35 × 10 mm plates containing grape juice/agar and yeast paste. Plates were changed in the evening and eggs collected overnight for ∼15–17 hr. The number of eggs from this overnight collection was counted, and a subset followed for hatching into larvae and subsequent development to adulthood. Flies were then placed back into vials for 5 additional days and then returned to egg-collection containers for a similar regiment, except that egg hatching was not monitored for most crosses due to the small number of eggs laid by Hmr mutant females. In separate experiments to count ovarioles, 23–24 pairs of ovaries from 5-day-old virgin females were dissected in 1× PBS and the ovarioles counted under a stereo microscope. The experiment was repeated twice.

TABLE 5.

Egg laying defect in Hmr mutant females

Experiment	Genotype of femalesa	Age of females (days old)b	No. of femalesb	Duration of collection (hr)	No. of eggs laid	No. of eggs/ female/hrc	No. of eggs assayed for hatching	No. of eggs hatched	No. eclosedd
1. Df(1)Hmr−	Hmr+	3–4	27	15.75	246	0.58	70	14	12
Repetition 1	Hmr−	3–4	28	15.75	79e	0.18***	73	7	3*
	Hmr+	9–10	23	15.75	105f	0.29	ND
	Hmr−	9–10	19	15.75	18	0.06***	ND
2. Df(1)Hmr−	Hmr+	3–4	40	16	187	0.29	50	10	10
Repetition 2	Hmr−	3–4	35	16	47	0.08***	45	1	1*
	Hmr+	9–10	30	16.5	108	0.22	ND
	Hmr−	9–10	13	16.5	6	0.03***	ND
3. Hmr3	Hmr+	3–4	24	16	134	0.35	134	74	20
Repetition 1	Hmr−	3–4	23	16	41	0.11**	41	10	5
	Hmr+	9–10	20	15	45	0.15	45	11	1
	Hmr−	9–10	11	15	14	0.08	14	1	0
4. Hmr3	Hmr+	3–4	21	15	373	1.18	100	49	25
Repetition 2	Hmr−	3–4	19	15	142	0.50*	100	5	2***
	Hmr+	9–10	19	15	75	0.26	ND
	Hmr−	9–10	14	15	28	0.13	ND

See materials and methods for details of egg collections. Each experiment was performed twice. *P < 0.05; ***P < 0.0001.

^aFor Df(1)Hmr⁻ experiments Hmr⁺ equals Df(1)Hmr⁻/Df(1)Hmr⁻; P{Hmr^+tp72}11-1/+ and Hmr⁻ equals Df(1)Hmr⁻/Df(1)Hmr⁻, from the BC4 stock described in materials and methods. For Hmr³ experiments Hmr⁺ equals w P{EPgy2}Hmr³/w¹¹¹⁸ + and Hmr⁻ equals P{EPgy2}Hmr³/P{EPgy2}Hmr³.

^bAge and number of females at start of egg collection. Experiment 1 was started with 28 females of each genotype and experiment 2 was started with 40 females of each genotype. Experiment 3 was started with 23 (Hmr³/Hmr³) or 24 (Hmr³/+) females. Experiment 4 was started with 19 (Hmr³/Hmr³) or 21 (Hmr³/+) females.

^cWithin each of the two time points of each experiment the number of eggs laid per female was compared between collections from Hmr⁺ and Hmr⁻ females. This ratio was further normalized to the length of collections to facilitate comparisons among collections and experiments. *P < 0.01; **P < 0.001; ***P < 0.0001 using Fisher's exact test (one-tailed).

^dWithin each of the two time points of each experiment the number of eclosed progeny per number of eggs assayed for hatching was compared between collection from Hmr⁺ and Hmr⁻ females using Fisher's exact test (one-tailed).

^ePlus one first-instar larva.

^fPlus 14 first-instar larvae.

Cytological analyses of ovarioles:

Ovaries were dissected in 1× PBS and tips of the ovarioles tweezed apart. Samples were fixed for 15 min in a mixture of 0.3 ml 1× PBTF (1× PBS, 4% paraformaldehyde, 0.125% TWEEN 20) and 0.9 ml heptane. Samples were rinsed three times in 1× PBT (1× PBS, 0.1% TWEEN 20), washed four times for 10 min in 1× PBT, and incubated for 1 hr in 1× PBTA (1× PBT, 1.5% BSA). The solution was then replaced with 1× PBTA containing the appropriate dilutions for each antibody. Primary antibody incubations were done overnight at 4°. Rinses and washes in PBT were repeated, and samples were blocked in 1× PBTA for 1 hr. The solution was replaced with 1× PBTA with the appropriate dilutions of secondary antibodies and incubated at room temperature for 4–6 hr. Rinses and washes in PBT were repeated and the samples were washed overnight in PBT at 4°. Ovaries were incubated with TO-PRO3 iodide (Molecular Probes) diluted to 1:10,000 in PBT for 5 min and rinsed with PBT for 1 min. Samples were mounted in Vectashield mounting media (Vector Laboratories).

The following antibodies and dilutions were used: rat anti-Vasa [Developmental Studies Hybridoma Bank, University of Iowa, (DSHB); 1:25], mouse anti-Hts 1B1 (DSHB; 1:4), mouse anti-Orb 6H4 (DSHB; 1:25), and mouse anti-Hts-RC (DSHB; 1:1). Alexa Fluor secondary antibodies (Molecular Probes) were used at a dilution of 1:500. Hu-li tai shao (Hts) encodes multiple isoforms (Petrella et al. 2007). The anti-Hts 1B1 antibody recognizes Hts protein isoforms that are localized to the fusome and follicle cell membranes, while anti-Hts-RC recognizes Hts isoforms present in postmitotic differentiating cysts including both fusomes and ring canals.

RNA extraction and RT–PCR:

Total RNA was extracted and purified using Trizol Reagent (Invitrogen). First strand cDNA was prepared from 0.5 μg of total RNA using the SuperScript III first-strand synthesis system (Invitrogen) with the oligo(dT)₂₀ primer in a 20-μl reaction according to the manufacturer's instructions.

For transcript analysis from Hmr mutants RNA was extracted from 20 pairs of ovaries of 6-day-old virgin females and cDNA prepared from 1 μg of total RNA. Three separate PCR reactions were performed using 1 μl of 1:10 diluted cDNA template with primer pairs that span an intron to be sure that amplification comes from cDNA, not contaminating genomic DNA. Primers were: for exons 1–2, 5′-AGCCTCTCTGCCCTATAGGTTTGC-3′ and 5′-AAGAGTGAAGAGTGAAGCAGTGGT-3′; for exons 2–3, 5′-CTGGTATTTTAGCGCTATCTCGTC-3′ and 5′-TATCGCCAAGAAGTTTCACATGCCAG-3′; for exons 4–5, 5′-TAAGAAATTGCGTAGCCAACCAAACAG-3′ and 5′-ATTTAGAGAACAGCTGCGTCATCTTGC-3′. Gapdh1 was amplified as a control with the primers 5′-CAGTGCTGGAGCCGAGTATGTGGTGGAGTC-3′ and 5′-GTGGCGGCCGGGATGATGTTCTGG-3′.

Construction of a transgene containing the Hmr substitutions found in In(1)AB:

In(1)AB contains two missense mutations resulting in the substitutions E371K and G527A in Hmr relative to wild-type D. melanogaster Hmr (Figure 1) (Barbash et al. 2003). These two amino acid changes were introduced into the P{Hmr^+tp74} transgene construct of Barbash et al. (2003), which contains Hmr and the flanking gene Rab9D; the formal name of this construct is P{w^+mC Hmr^+t8.0 Rab9D^+t8.0=Hmr^+tp74}. A 981-bp fragment containing the two missense mutations was amplified by PCR from DNA prepared from In(1)AB flies using the oligonucleotides 5′-CAGCGCATGCGCGGCACCGTAT-3′ and 5′-ATACCCTGTCTTGAGTTTGA-3′. This fragment was digested with SpeI and BsiWI and used as the insert for ligation. P{Hmr^+tp74} was digested with SpeI and BsiWI, dephosphorylated with shrimp alkaline phosphatase, and used as the vector for ligation without gel purification. Recombinant plasmids were distinguished from religated P{Hmr^+tp74} plasmids by the loss of an AvaI restriction site at the site of the second amino acid substitution (Figure 1A). The correct structure of the recombinant plasmid was confirmed by DNA sequencing across the region containing both amino acid substitutions. The formal designation of this construct is P{w^+mC Hmr² Rab9D^+t8.0=Hmr^2t8.0}. Two independent autosomal transformants were obtained by P-element transformation.

Figure 1.—

Map of Hmr region and transgenic constructs and expression analysis of P{Hmr^2t8.0} transgenes in hybrids. (A) CG2124 and Rab9D are on the opposite strand relative to Hmr. For the five constructs shown the relevant Hmr genotype is written above the name of the transgenic construct. D. melanogaster Hmr contains an AvaI restriction site. The D. melanogaster Hmr² construct contains the two indicated amino acid substitutions; the second substitution mutates the AvaI site. This restriction site is absent in D. simulans and D. mauritiana Hmr. The D. melanogaster Hmr² construct is described in this study. The other D. melanogaster constructs are described in Barbash et al. (2003); the D. simulans and D. mauritiana constructs are described in Barbash et al. (2004a). (B) RT–PCR analysis of hybrids containing P{Hmr^2t8.0} transgenes. M, DNA marker showing bands of indicated sizes; lanes 1–3, AvaI-digested PCR products amplified from genomic DNA; lanes 4–7, AvaI-digested PCR products amplified from cDNA samples. Samples were run on a 2% agarose gel. Template DNAs and cDNAs were: (1) y; cn bw sp D. melanogaster DNA (wild-type control); (2) D. simulans w⁵⁰¹ DNA (wild-type control); (3) D. melanogaster In(1)AB, w DNA; (4) FM6,w/X_sim, w⁵⁰¹; P{Hmr^2t8.0}1-7/+ hybrid female cDNA; (5) FM6,w/X_sim, w⁵⁰¹; +/+ hybrid female cDNA; (6) FM6,w/X_sim, w⁵⁰¹; P{Hmr^2t8.0}2-1/+ hybrid female cDNA; and (7) FM6,w/X_sim, w⁵⁰¹; +/+ hybrid female cDNA. X_sim indicates the D. simulans X chromosome. Flies from lanes 4 and 5 are siblings from a single cross, as are those from lanes 6 and 7. The 323-bp region of Hmr amplified from D. melanogaster contains an AvaI site that when cleaved produces fragments of 240 and 79 bp, excluding single-strand overhangs (lane 1). This AvaI site is absent from the Hmr allele carried by In(1)AB (lane 3) due to one of the missense changes described in Barbash et al. (2003) (see Figure 1A); it is also absent from D. simulans Hmr (lane 2). cDNAs in lanes 5 and 7 show an approximately equal expression of the D. melanogaster allele from the FM6,w chromosome and the D. simulans allele. cDNAs in lanes 4 and 6 show an increase in the amount of the uncleaved 324-bp band relative to the 240-bp band, indicating expression derived from the P{Hmr^2t8.0} transgene.

To test for expression cDNA was prepared from 0.5 μg of total RNA isolated from ∼20 adult hybrid females as described above. PCR was performed from 1 μl of cDNA using the primers 5′- GCCATTGGCTATTGTCTATCC-3′ and 5′-GACACGCCCGTTCCCATAGT-3′, which produces a fragment of 324 bp from D. melanogaster and 318 bp from D. simulans.

RESULTS

Characterization of new Hmr alleles:

We compared new and previously available Hmr alleles for the magnitude of hybrid male rescue (supplemental Table S1). Hmr¹ rescued well with D. mauritiana at 18° and 25°, but only with D. simulans at 18° and poorly or not at all with D. sechellia at all temperatures. Both the species specificity and temperature sensitivity largely agree with the original report of Hutter and Ashburner (1987). The other three alleles tested generally rescued better, which is consistent with reports that Hmr¹ is hypomorphic (Hutter et al. 1990; Barbash et al. 2000, 2003). Df(1)Hmr⁻ and P{EPgy2}Hmr³ had similar rescue activity in most assays and rescued better than In(1)AB, Hmr² in D. simulans and D. sechellia at >18°. Surprisingly both Df(1)Hmr⁻ and P{EPgy2}Hmr³ rescued poorly with D. mauritiana at 25° and 28°. Hybrid females of genotype +/+ were poorly viable at 25° and 28° in crosses with D. simulans and at 28° in crosses with D. sechellia. This female lethality was suppressed by Df(1)Hmr⁻, as observed previously for larger multilocus Hmr deletions (Barbash and Ashburner 2003). P{EPgy2}Hmr³/+ females could not be distinguished by visible markers from +/+ female hybrids with either species, but the high viability of total females strongly suggests that P{EPgy2}Hmr³ also suppresses high-temperature hybrid female lethality.

We analyzed expression from three of these alleles by RT–PCR, using primer pairs that detect three different regions of Hmr (Figure 2). PCR assays were not quantitative but rather designed to identify a possible expression-null allele. The expression profile of Hmr¹ was indistinguishable from the revertant allele Hmr^1r1 or two wild-type controls, a finding consistent with previous Northern blot analysis (Barbash et al. 2003). cDNA prepared from Df(1)Hmr⁻ showed no amplification across the exon 1–2 region, which is expected because the deletion removes exon 1 and part of exon 2 (Barbash and Lorigan 2007). Amplification was achieved, however, using the two downstream primer pairs. Df(1)Hmr⁻ therefore appears to express a truncated mRNA (or possibly a chimeric mRNA); whether a corresponding N-terminal deleted protein is also produced remains to be investigated. Only trace amounts of amplified product were observed in P{EPgy2}Hmr³ samples, and only using the 3′-most primer pair. This allele is therefore the best candidate for being null by molecular criteria, with the above genetic data suggesting that Df(1)Hmr⁻ is also null.

Figure 2.—

RT–PCR analysis of Hmr mutants. RT–PCR from the indicated genotypes was performed with three sets of primer pairs that span the indicated adjacent exons in Hmr, and with Gapdh1 as a control. Expected sizes of products are indicated to the left of each gel. RT − and + refers to absence and presence of reverse transcriptase enzyme during cDNA preparation. Canton-S and Oregon-R are wild-type D. melanogaster strains. The first lane of each gel contains a DNA molecular weight ladder in 100-bp increments.

The missense mutations in In(1)AB, Hmr² abolish Hmr⁺ function:

Two amino acid differences distinguish the Hmr allele on the In(1)AB chromosome from Hmr⁺ alleles (Barbash et al. 2003). On the basis of results presented in this section we refer henceforth to this allele as In(1)AB, Hmr². It is unknown whether these polymorphisms affect Hmr⁺ function or cause the hybrid rescue phenotype of In(1)AB, Hmr². The hybrid rescue activity of In(1)AB, Hmr² is suppressed by Hmr⁺ transgenes such as P{Hmr^+tp74} (Barbash et al. 2003). One interpretation of these data is that one or both of the missense mutations make the Hmr² allele carried by In(1)AB, Hmr² nonfunctional, and the transgenic Hmr⁺ copy can complement the nonfunctional Hmr² allele. An alternative interpretation is that In(1)AB, Hmr² rescues hybrid males due to a mutation in a gene other than Hmr, and that Hmr⁺ transgenes are capable of complementing this as yet unknown hybrid rescue gene because it interacts with Hmr. In this scenario the two Hmr amino acid polymorphisms carried by In(1)AB, Hmr² would not be expected to affect Hmr⁺ function, and an Hmr transgene containing these two polymorphisms would retain the ability to suppress rescue by In(1)AB, Hmr² or other Hmr mutations.

To test directly whether or not these two amino acid changes are responsible for hybrid rescue we engineered them into the Hmr coding region of the Hmr⁺ construct P{Hmr^+tp74} and generated the transgenic construct P{Hmr^2t8.0} (see materials and methods). We found that two independent transformants of P{Hmr^2t8.0} have lost Hmr⁺ activity because they did not suppress the ability of In(1)AB, Hmr² to rescue hybrid males: In(1)AB, Hmr²/Y; P{Hmr^2t8.0}/+ hybrid males had equivalent viability to their In(1)AB, Hmr²/Y; +/+ siblings (Table 1). Hmr⁺ transgenes also reduce the viability of hybrid females (Barbash et al. 2003). In contrast, FM6,w/+ hybrid females tested here had equivalent viability with and without the P{Hmr^2t8.0} transgene (Table 1). RT–PCR analysis demonstrated that these transgenes are expressed in species hybrids (Figure 1B).

TABLE 1.

Test for suppression of hybrid rescue by the P{Hmr^2t8.0} transgene

	Hybrid female progeny						Hybrid male progeny
Line tested	In(1)AB/+; +/+	In(1)AB/+; P{w+}/+	Relative viability	FM6,w/+; +/+	FM6,w/+; P{w+}/+	Relative viability	In(1)AB/Y; +/+	In(1)AB/Y; P{w+}/+	Relative viability
P{Hmr2t8.0}1-7	284	244	0.86	270	276	1.02	95	120	1.26
P{Hmr2t8.0}2-1	181	172	0.95	195	208	1.07	68	75	1.10

In(1)AB, w/FM6,w; P{Hmr^2t8.0}/+ females were crossed to D. simulans w⁵⁰¹ males. Three classes of sibling progeny with and without the P{Hmr^2t8.0} transgene are produced and were compared for viability. All comparisons are not significantly different from a 1:1 ratio (χ² test, P > 0.05). Two independent transformant lines were tested.

We further found that the P{Hmr^2t8.0} construct does not suppress rescue by a second allele, Df(1)Hmr⁻. Df(1)Hmr⁻/FM6,w; P{Hmr^2t8.0}/+ females were crossed to D. simulans v males. Crosses produced 127 Df(1)Hmr⁻/Y; P{Hmr^2t8.0}1-7/+ and 160 Df(1)Hmr⁻/Y;/+ hybrid sons; a second set of crosses with an independent transformant produced 209 Df(1)Hmr⁻/Y; P{Hmr^2t8.0}2-1/+ and 203 Df(1)Hmr⁻/Y;/+ hybrid sons. The differences between the sibling classes were not significantly different from a 1:1 ratio in either set of crosses (χ² test, P > 0.05). We conclude that one or both of the amino acid changes identified in the Hmr allele carried by In(1)AB, Hmr² causes a loss-of-function Hmr hybrid rescue phenotype.

Hmr is required for female viability and fertility:

We noticed that homozygous stocks of Df(1)Hmr⁻, In(1)AB, Hmr² and P{EPgy2}Hmr³ can often be maintained at ∼23° or lower but become unhealthy at temperatures of 25° or higher. Initial observations suggested that these effects were due to poor viability and fertility of females. Because of the incomplete penetrance of these phenotypes we further examined all three of these independently generated alleles. We first examined Df(1)Hmr⁻ and designed complementation tests to address two questions: (1) Is this reduced fitness caused specifically by the deletion or by other unknown variation in the Df(1)Hmr⁻ stock?, and (2) If the reduced fitness is caused by the deletion, is it due to reduced function of Hmr or of the adjacent gene CG2124 (Figure 1)? In these tests Df(1)Hmr⁻ females were compared to siblings heterozygous for a potentially complementing transgene so that the comparisons were done between similar genetic backgrounds under identical environmental conditions (see materials and methods). Crosses in Table 2 were done at different times, so viabilities between crosses cannot be directly compared although we comment on general trends observed in the data.

To address the first question we tested whether the P{Hmr^+tp72} transgene containing Hmr⁺ and CG2124⁺ complements the Df(1)Hmr⁻ phenotype. This construct was chosen because it most closely matches in structure the P{Hmr^tp73} control construct and the D. mauritiana and D. simulans Hmr⁺ constructs assayed below. We showed previously that P{Hmr^+tp72} has Hmr⁺ activity because it complements (that is to say, suppresses) the hybrid rescue phenotypes of Hmr¹ and In(1)AB (Barbash et al. 2003). Df(1)Hmr⁻/Df(1)Hmr⁻ and Df(1)Hmr⁻/Df(1)Hmr⁻; P{Hmr^+tp72}/+ females were compared using a crossing scheme designed to minimize extraneous variation between these genotypes (see materials and methods). Two different insertions of the P{Hmr^+tp72} transgene were tested for complementation (Table 2, crosses 1 and 2). In both sets of crosses female viability was comparable at day 5 between the two genotypes. Df(1)Hmr⁻/Df(1)Hmr⁻ viability was decreased at day 10, and from day 15 onward, was significantly lower than Df(1)Hmr⁻/Df(1)Hmr⁻; P{Hmr^+tp72}/+ females in both sets of crosses. The difference in fertility between the genotypes was even more striking. Df(1)Hmr⁻/Df(1)Hmr⁻; P{Hmr^+tp72}/+ females averaged between ∼44 and 67 progeny each throughout the course of the experiment, while Df(1)Hmr⁻/Df(1)Hmr⁻ females had significantly fewer progeny at all time points and were essentially sterile by day 15. These data demonstrate that Df(1)Hmr⁻ specifically reduces female viability and fertility.

To address whether the Df(1)Hmr⁻ phenotype is due to reduced function of Hmr or CG2124, we next tested whether the P{Hmr^tp73} transgene can also complement Df(1)Hmr⁻. This construct is identical to P{Hmr^+tp72} except that it contains an engineered frameshift mutation in Hmr; it therefore is CG2124⁺ and Hmr⁻ (Barbash et al. 2003). Again, two sets of experiments were performed using two different insertions of P{Hmr^tp73} (Table 2, crosses 3 and 4). In both experiments the Df(1)Hmr⁻ control females were substantially less viable than the Df(1)Hmr⁻ control females used above (crosses 1 and 2). For example, at day 5 female viability was 18% and 29% in crosses 3 and 4 compared to 74% and 77% in crosses 1 and 2. More similar among these crosses was the low fertility of Df(1)Hmr⁻ females, particularly at day 10 and after. These data suggest that viability may be a more variable phenotype than fertility and emphasize the necessity of comparing sibling progeny from single crosses. In both crosses 3 and 4, the Df(1)Hmr⁻ females with and without the P{Hmr^tp73} transgene had similarly low viability and fertility, with no significant differences at any time points. We conclude that CG2124⁺ cannot complement Df(1)Hmr⁻ and that the low viability and fertility of Df(1)Hmr⁻ is caused by the deletion of Hmr.

We next tested whether In(1)AB, Hmr² is also mutant for Hmr intraspecific function by asking whether it can complement the reduced viability and fertility described above for Df(1)Hmr⁻. In(1)AB, Hmr²/Df(1)Hmr⁻ females were compared to In(1)AB, Hmr²/Hmr^1r1 females (Table 3). Hmr^1r1 is a wild-type derivative allele of Hmr¹ that was derived from the same screen that generated Df(1)Hmr⁻ (Barbash and Lorigan 2007). In this experiment In(1)AB, Hmr²/Hmr^1r1 is the control class and In(1)AB, Hmr²/Df(1)Hmr⁻ is the experimental class being tested for a potential reduction in viability and fertility. We found that both classes had equivalently high viability throughout the time course examined, with 68% of females still alive at day 25. This result suggests that In(1)AB, Hmr² complements the viability defect of Df(1)Hmr⁻. However the fact that both classes remained so highly viable throughout the experiment suggests an alternative possibility that viability may be higher in this genetic background compared to the other backgrounds tested above, so that loss of Hmr function might not be detectable.

The results for female fertility, however, were unambiguous. In(1)AB, Hmr²/Df(1)Hmr⁻ females had significantly reduced fertility compared to controls at all time points measured. Furthermore the pattern of fertility, with essentially complete sterility observed from day 15 onward, closely resembles that seen above for Df(1)Hmr⁻ homozygous females. These data demonstrate that In(1)AB, Hmr² is mutant for Hmr intraspecific function.

We then compared P{EPgy2}Hmr³/P{EPgy2}Hmr³ vs. P{EPgy2}Hmr³/+ control females, using a slightly different design that could detect female fertility differences but not viability effects. We found that P{EPgy2}Hmr³/P{EPgy2}Hmr³ females produced significantly fewer progeny at all time points (Table 4). In conclusion, we find that all three strong loss-of-function Hmr alleles tested have significantly reduced female fertility.

The nature of the Hmr-dependent fertility defect:

Number of eggs laid and viability of progeny from Df(1)Hmr⁻ and P{EPgy2}Hmr³ mutants were examined to determine why Hmr mutants have reduced fertility. Df(1)Hmr⁻ females aged 3–4 days or 9–10 days laid significantly fewer eggs than Hmr⁺ controls (Table 5). Eggs laid from 3- to 4-day-old Df(1)Hmr⁻ females also hatched into larvae and produced eclosed adults at lower frequencies than those from control siblings. A similar experiment was performed using P{EPgy2}Hmr³. As with Df(1)Hmr⁻, P{EPgy2}Hmr³ mutant females laid fewer eggs than P{EPgy2}Hmr³/+ controls at both time points although the difference was statistically significant only with 3- to 4-day-old females. Hatchability and eclosion of adults also occurred at lower frequencies in experiments with P{EPgy2}Hmr³ compared to the heterozygous controls.

Eggs laid from Df(1)Hmr⁻ females were slightly misshapen and had shorter chorionic appendages compared to those produced from control siblings (Figure 3). Although the appendages were often fused at their tips, spacing at the base appeared normal compared to that seen in eggs laid by females mutant for weakly ventralizing mutations (Schupbach 1987). Similar morphological defects were also observed in eggs laid from P{EPgy2}Hmr³ females (data not shown).

Figure 3.—

Egg structure defects in Df(1)Hmr⁻ females. Five eggs each from 9- to 10-day-old Df(1)Hmr⁻/Df(1)Hmr⁻; P{Hmr^+tp72}11-1/+ females (A) and Df(1)Hmr⁻/Df(1)Hmr⁻ females (B) collected from experiment 2 of Table 5. Eggs from Df(1)Hmr⁻/Df(1)Hmr⁻ females are slightly shorter than from Df(1)Hmr⁻/Df(1)Hmr⁻; P{Hmr^+tp72}11-1/+ females and have substantially shorter chorionic appendages that are fused at their distal ends. Note that one of the eggs from Df(1)Hmr⁻/Df(1)Hmr⁻; P{Hmr^+tp72}11-1/+ females in A also has a chorionic appendage defect.

Ovaries from 5-day-old virgin P{EPgy2}Hmr³ females grown at 28° appeared somewhat smaller than P{EPgy2}Hmr³/+ control siblings. We counted ovarioles from both ovaries from 47 flies of each genotype. P{EPgy2}Hmr³/P{EPgy2}Hmr³ females had 21.2 ± 5.1 ovarioles and P{EPgy2}Hmr³/+ had 29.2 ± 2.9 ovarioles. These differences are statistically significant (P < 0.0001, one-tailed t-test).

We further analyzed ovarioles from P{EPgy2}Hmr³ females raised at 29°. In ovaries from young flies (5 days old) we observed no major morphological defects. However in ovaries from flies that were aged 15–17 days, early ovariole development was defective in approximately half of those examined, while the other half appeared normal. The wild-type-appearing class of P{EPgy2}Hmr³ ovaries contained properly formed ovarioles where oocyte development appeared similar to P{EPgy2}Hmr³/+ controls (Figure 4, A, C, and E). The most anterior portion of the ovariole, the germarium, was clearly visible as were the developing egg chambers which increased in size as they matured. The mutant class of ovaries contained ovarioles in which the anterior end was stub-like rather than tapered in shape (Figure 4, B, D, and F). The stub-like ovarioles were more even in size along their entire length rather than tapered as in wild type, and egg chambers could not be distinguished using Nomarski optics.

Figure 4.—

Two classes of ovariole morphology in 15- to 17-day-old P{EPgy2}Hmr³/P{EPgy2}Hmr³ females. (A, C, and E) Ovarioles that resemble wild-type or P{EPgy2}Hmr³/+ controls. (B, D, and F) Ovarioles with severe morphological defects. Anterior is marked with an asterisk in all images. (A and B) Ovarioles stained with anti-Vasa (green), anti-Hts 1B1 (red), and TO-PRO3 (DNA, blue). (A) Ovariole showing the characteristic wild-type pattern where at the anterior of the ovariole developing cysts are present in which anti-Vasa stains the germ cells and anti-Hts 1B1 stains the fusome, which gets larger as the cyst grows [for example, see Zaccai and Lipshitz (1996)]. Anti-Hts 1B1 also stains the cytoskeleton. Egg chambers budding from the germarium show a normal pattern of polyploid nuclei. (B) Ovariole showing egg chambers squished into anterior end. Several cysts are not correctly enveloped by a layer of follicle cells. (C and D) Ovarioles stained with anti-Orb (green) and TO-PRO3 (blue). (C) Ovariole showing younger cysts with Orb evenly dispersed and older egg chambers where Orb has accumulated in the oocyte, as described by Lantz et al. (1994). (D) Despite obvious morphological defects in the ovariole, younger cysts in the anterior of the ovariole with nonpolyploid nuclei show Orb correctly present throughout and older egg chambers where polyploid nuclei are present show an accumulation of Orb in one or two cells. (E and F) Ovarioles stained with anti-Vasa (green), anti-Hts-RC (red), and TO-PRO3 (blue). (E) Ovarioles showing the accumulation of Hts-RC in postmitotically active cysts and in the ring canals of egg chambers as observed in Robinson et al. (1994). (F) While egg chambers are squished into the anterior of the ovariole and in some cases disrupted, Hts-RC can still be seen at ring canals of intact egg chambers as well as near the nuclei of dispersed egg chambers.

When stained with antibodies that mark various stages of oogenesis, it became clear that the stub-like ovarioles do contain developing egg chambers. However, the egg chambers were squished into the anterior end rather than extending posteriorly in a linear progression as occurs in the wild type. Some of the squished egg chambers were intact as can be seen in Figure 4F. However, many of the egg chambers seemed to be disrupted as can be seen by the scattered nurse cell nuclei (Figure 4, B and D). To determine if the stub-like egg chambers contain mature cysts, we used anti-Orb antibodies to determine if oocytes are specified and anti-Hts-RC to determine if ring canals are properly formed and maintained in egg chambers. In what appears to be younger cysts, Orb was evenly distributed throughout the entire cyst in the mutant ovarioles. However, some cysts showed accumulation of Orb in one or two cells, suggesting that these cysts have specified an oocyte (Figure 4D). Additionally, the stub-like egg chambers also showed Hts-RC correctly accumulating at ring canals in both intact and dispersed egg chambers. In the defective class of ovarioles we also noticed that the somatic follicle cells appeared to be smaller and reduced in number. These data suggest that egg chambers in Hmr mutant females are able to mature and express germ-line markers that characterize several different stages of oogenesis. The defects in ovariole structure may therefore result from a somatic ovariole function of Hmr. Such a somatic defect might be causing a failure of the developing egg chambers to migrate posteriorly.

Combined with the data in Table 5, these data suggest that the reduced fertility of Hmr mutant females stems from multiple causes, including reduced ovariole number, defects in ovariole structure, reduced egg laying, and reduced zygotic viability, with defective egg morphology possibly contributing to the latter two phenotypes.

Sibling-species orthologs of Hmr⁺ partially complement Df(1)Hmr⁻:

Having established a phenotype for Hmr within D. melanogaster, we next tested whether D. simulans and D. mauritiana Hmr⁺ transgenes transformed into D. melanogaster can complement the deleterious phenotypes of Df(1)Hmr⁻ (Table 2, crosses 5–8). We used transgenic constructs of D. simulans and D. mauritiana Hmr⁺ that closely match in structure the P{Hmr^+tp72} constructs tested above (Figure 1A) and that were previously shown to express Hmr when transformed into D. melanogaster (Barbash et al. 2004a). Compared to the above crosses with P{Hmr^+tp72} (Table 2, crosses 1 and 2), Df(1)Hmr⁻ control females had similar viability except for being somewhat higher at days 20 and 25. However, Df(1)Hmr⁻ females carrying either the D. simulans or D. mauritiana transgenes generally did not have higher viability. One exception was in cross 7 (Table 2) where the females carrying the D. mauritiana Hmr⁺ transgene were moderately more viable (∼1.4- to 2.3-fold) at days 10 though 25. Similar complementation was not seen with a second transformant of this construct (Table 2, cross 8). In total these data suggest that sibling-species Hmr⁺ orthologs have little ability to complement the reduced viability caused by Df(1)Hmr⁻.

Df(1)Hmr⁻ females carrying these transgenes did have significantly higher fertility than Df(1)Hmr⁻ control females at days 5 and 10 in all four experiments. However the mean number of progeny produced was rather low compared to experiments with the D. melanogaster P{Hmr^+tp72} transgene, averaging only 3.9–10.4 progeny/female at day 10. Additionally, no complementation was observed at days 15–25, with transgene-carrying Df(1)Hmr⁻ females being essentially sterile. We conclude that D. simulans and D. mauritiana Hmr⁺ transgenes partially complement the reduced fertility phenotype of Df(1)Hmr⁻.

DISCUSSION

The Hmr¹ mutation was discovered on the basis of its ability to rescue otherwise lethal F₁ hybrid males (Hutter and Ashburner 1987). This allele has no detectable phenotype within D. melanogaster, but genetic and molecular analyses confirmed the prediction of Hutter et al. (1990) that Hmr¹ is hypomorphic (Barbash et al. 2000, 2003). Three other Hmr alleles rescue hybrid males more effectively (supplemental Table S1). Based on RT–PCR analysis P{EPgy2}Hmr³ is likely a null allele. Df(1)Hmr⁻ is caused by a small deletion of exon 1 and part of exon 2 (Barbash and Lorigan 2007) and has a hybrid rescue profile very similar to P{EPgy2}Hmr³ (supplemental Table S1). Surprisingly Df(1)Hmr⁻ mutants produce a transcript so it remains possible that this allele might encode a truncated product (Figure 2).

Using these two alleles as well as In(1)AB, Hmr² we have shown here that Hmr is required for viability and fertility of D. melanogaster females. Both of these phenotypes are incompletely penetrant and become stronger as females age. Because of this incomplete penetrance, the observation of a statistically significant loss of fertility in three independent mutant backgrounds provides strong evidence that the effects are specifically due to Hmr loss of function.

In general viability appeared to be more variable than fertility. For example, among Df(1)Hmr⁻ homozygous females in Table 2, viability at day 10 ranged from 5% in cross 3 to 46% in crosses 6 and 7. In contrast, in all experiments Df(1)Hmr⁻ homozygous females had low fertility at day 10 (0–5.4 ± 7.6) and were essentially sterile at days 15 and beyond. Late-stage female sterility is therefore the most reliable diagnostic phenotype of Hmr loss of function in D. melanogaster. It is important to note that there may also be additional phenotypes of Hmr that remain to be discovered, for example in males.

Hmr mutant females have reduced fertility:

This fertility phenotype likely results from multiple causes, including reduced ovariole number, defective ovariole development, defective oocytes and eggs, and zygotic lethality of embryos derived from mutant Hmr mothers. Cytological examination of Hmr mutant ovarioles revealed two classes of Hmr mutant ovarioles, a wild-type class and a stub-like class. The wild-type class was indistinguishable from ovarioles from Hmr heterozygous siblings in stainings with several antibody markers. The mutant class showed defects in ovariole shape and structure. Rather than containing a germarium with budding egg chambers that gradually increase in size as they move posteriorly, the mutant ovarioles appeared to have multiple egg chambers squished into the anterior end of the ovariole. Antibody staining to proteins expressed in mature cysts showed that mature egg chambers were present in the anterior end and in most cases have burst open, leading to a dispersion of the polyploid nurse cells. One possible explanation is that the structure of the ovariole is weakened such that the developing egg chambers fail to move posteriorly. The small size and low number of follicle cells that we observed may contribute to this failure in egg chamber migration. The Hmr mutant phenotype does not closely phenocopy female-sterile mutations in other germline-expressed genes of which we are aware. However some similarities in egg chamber defects are observed in mutants that affect formation of follicle cells or signaling between follicle and germ-line cells (Tworoger et al. 1999; Smith and Cronmiller 2001; Song and Xie 2003; Willard et al. 2004). For this reason we suggest that the observed defects reflect a requirement for Hmr function in somatic ovarian cells.

Hybrid daughters from D. melanogaster females crossed to D. simulans males are sterile (Sturtevant 1920). Hybrid female sterility is partially rescued by Hmr⁻ deletions, In(1)AB, and Hmr¹ (Barbash and Ashburner 2003). Hybrid female sterility is also partially rescued in the reciprocal cross by In(1)AB (Davis et al. 1996). On the basis of our results here we suggest that this fertility rescue by In(1)AB results from the Hmr² allele that it carries. Ovaries of nonrescued Hmr⁺ hybrid females derived from D. melanogaster mothers contain no eggs or developing ovarioles (Barbash and Ashburner 2003). In the reciprocal cross, hybrid females carrying In(1)AB, Hmr² show a range of phenotypes, from highly disorganized ovarioles to wild-type-appearing ovarioles with normal patterns of the Orb protein (Hollocher et al. 2000). None of these hybrid phenotypes bear close similarity to the intraspecific Hmr⁻ phenotype reported here. These phenotypes may therefore be mechanistically distinct, which is perhaps unsurprising since the intraspecific fertility loss is caused by the absence of Hmr⁺, while hybrid sterility is caused by the presence of Hmr⁺.

In(1)AB, Hmr²—an Hmr allele from the mists of time:

The hybrid rescue effect of In(1)AB was discovered serendipitously (Hutter et al. 1990). The distal breakpoint of the inversion at cytological position 9E7-8 is fairly close to the estimated cytological position of the Hmr¹ rescue allele at 9D1-9E4 (Hutter et al. 1990). These findings raised the possibility that, allowing for some imprecision in the genetic mapping and cytological analyses, the breakpoint of In(1)AB was disrupting the same gene responsible for rescue in Hmr¹ (Hutter and Karch 1994). Subsequent analysis, however, refuted this possibility by showing that the gene broken by the In(1)AB distal breakpoint, sesB, has no role in hybrid lethality (Zhang et al. 1999).

The Hmr allele carried by In(1)AB contains two missense mutations relative to the wild-type D. melanogaster sequence (Barbash et al. 2003). By engineering these two mutations into an Hmr⁺ transgenic construct we have shown here that one or both of these mutations are likely responsible for the hybrid rescue phenotype associated with In(1)AB, and we have named the Hmr allele carried by In(1)AB as Hmr². Our experiments also demonstrate that In(1)AB, Hmr² has a reduced fertility similar to Df(1)Hmr⁻.

D. melanogaster Hmr encodes a predicted protein of 1413 amino acids and contains four predicted MADF domains, which are known or proposed to be involved in DNA or chromatin binding (England et al. 1990, 1992; Bhaskar and Courey 2002; Maines et al. 2007; Maheshwari et al. 2008). Most MADF domains have a predicted positive charge, which is a property consistent with them being able to bind negatively charged DNA (Maheshwari et al. 2008).

The first substitution in In(1)AB, Hmr², E371K, occurs in the third MADF domain encoded by Hmr. This glutamic acid is conserved in all 14 known Hmr orthologs (Maheshwari et al. 2008). The third MADF domain encoded by Hmr is unusual in having a predicted negative charge, suggesting that it may bind to positively charged histones rather than directly to DNA. The identification of this unusual third MADF domain led us to propose that the Hmr protein has potential chromatin binding activity. The discovery here that the E371K mutation may be associated with a loss-of-function Hmr phenotype suggests that this unusual MADF domain is required for Hmr function. In contrast, the second mutation, G527A, does not occur in an identifiable domain region of Hmr and is not a conserved position among Hmr orthologs. We suggest, therefore, that the E371K substitution is more likely to be responsible for causing the In(1)AB, Hmr² mutant phenotype.

How many genes cause F₁ hybrid lethality?

Mutations that rescue the lethality of F₁ hybrid males have been found fortuitously or in screens of natural lines, rather than by using mutagenesis. Three alleles have been reported in D. melanogaster (Hutter and Ashburner 1987; Hutter et al. 1990; Barbash et al. 2004b). From results obtained here one can conclude that two of them—Hmr¹ and In(1)AB, Hmr²—rescue hybrids due to mutations in the same gene. The gene or genes mutated to cause hybrid rescue in the third rescuing strain, Df(1)EP307-1-2, remain unknown (Barbash et al. 2003). Two rescuing alleles are also known in D. simulans, and tentative evidence suggests that they may both be mutated for the same gene (Watanabe 1979; Brideau et al. 2006). On the basis of these limited data one can predict that there are unlikely to be many additional genes that when mutated can suppress F₁ hybrid male lethality. If correct then one would further conclude that F₁ hybrid male lethality is caused by a small number of major-effect genes.

These data do not exclude the possibility that additional genes may have minor effects on F₁ hybrid lethality. The possible existence of multiple modifier loci is suggested by the highly variable penetrance of rescue and lethality in F₁ hybrids (Barbash et al. 2000; Presgraves 2003).

Lack of evidence for Rab GTPases causing hybrid lethality:

Hutter (2007) has recently suggested that Rab GTPases may contribute to hybrid lethality. Hutter claimed that the presence of four amino acid polymorphisms in the Rab9D allele found in the Hmr¹ rescue strain supports a role for Rab9D in causing hybrid lethality. However, hybrid rescue by Hmr¹ has been experimentally shown to result from a P-element insertion upstream of Hmr because a revertant strain lacking this P-element, Hmr^1r1, does not rescue hybrids (Barbash et al. 2003). This nonrescuing Hmr^1r1 strain almost certainly retains the same Rab9D allele as Hmr¹ because the genes are closely linked (<7 kb apart). Furthermore, hybrid rescue by Hmr¹ is suppressed by the p83 transgene that carries Hmr⁺ but not Rab9D⁺ (Barbash et al. 2003). Hybrid rescue by Hmr¹ is therefore clearly caused by the mutation of Hmr, not by Rab9D polymorphism. Data presented here describing molecular lesions in Hmr carried on the independent genetic backgrounds of In(1)AB and P{EPgy2}Hmr³ further support the conclusion that Hmr, not Rab9D, is responsible for hybrid lethality.

The only experimental evidence offered in support of the Rab hypothesis was derived from a stock containing a PiggyBac insertion in Rab9D that was suggested to produce hybrid males and females of increased viability (Hutter 2007). The purported female effects, however, are not statistically significant (P > 0.10 by χ² analysis in all three comparisons described). Male hybrids produced with this stock remained lethal but with an increased proportion (25% vs. ∼12–14% in controls) dying as pseudopupae rather than larvae. Whether this slight change in the lethal phase of the male hybrids is actually due to the PiggyBac insertion or instead reflects other differences in the genetic background was not tested. We conclude that the purported role of Rab GTPases in hybrid lethality is unsupported by any experimental evidence.

The relationship between the intraspecific and interspecific functions of hybrid incompatibility genes:

We have found here that D. simulans and D. mauritiana Hmr⁺ orthologs partially complement the reduced fertility phenotype of D. melanogaster Hmr⁻. Although the methods used preclude the ability to compare directly the activity of these foreign-species orthologs relative to D. melanogaster Hmr⁺, their complete failure to complement Df(1)Hmr⁻ sterility at day 15 or later suggests that they are not functionally equivalent to D. melanogaster Hmr⁺.

There are two possible interpretations of this partial complementation. The first is that Hmr has evolved functions in the sibling species and in D. melanogaster that are at least partially different. In this scenario an Hmr mutation made in the sibling species would have a different phenotype than the D. melanogaster mutation. Alternatively, Hmr may have the same function in all the species tested, but has diverged in sequence through coevolution with other genes. This sequence divergence makes Hmr less functional in its interactions with genes from a foreign species. Under this interpretation an Hmr mutation would have equivalent phenotypes in D. melanogaster and its sibling species. In the absence of Hmr⁻ mutations in the sibling species, investigating the molecular basis of Hmr function and identifying interacting genes may provide an alternative method to distinguish between these possibilities.

In contrast, these same D. simulans and D. mauritiana Hmr⁺ transgenes were previously found to have no phenotype in hybrids, judging by their inability to complement (suppress) the hybrid male rescue of In(1)AB or Hmr¹ mutations (Barbash et al. 2004a). The D. simulans Hmr⁺ transgenes were additionally shown to have no effect on hybrid-female fertility rescue. These data demonstrate that Hmr has functionally diverged between D. melanogaster and its sibling species for the interspecific phenotype of hybrid lethality.

How should one view these data in light of evidence that Hmr has diverged considerably both in its coding sequence and in flanking noncoding regions between D. melanogaster and its sibling species under the force of natural selection (Barbash et al. 2004a)? In the standard model of hybrid incompatibility evolution the incompatibility phenotypes are viewed as arising as secondary by-products of intraspecific evolution. Positive selection of Hmr would then presumably result from selective forces on its intraspecific function.

We previously identified a large number of amino acid substitutions between D. melanogaster and its sibling species (49–137, depending on the model considered) that are candidates for causing hybrid incompatibility (Barbash et al. 2004a). It will be interesting to identify experimentally the substitutions causing functional divergence of the Hmr interspecific function and then to test whether the same mutations also affect the intraspecific function. In addition, the flanking noncoding regions of Hmr were also found to have a high rate of substitution compared to synonymous sites and introns (Barbash et al. 2004a). These areas of divergence are candidates for causing interspecific differences in transcription rate or stability of the respective Hmr mRNAs.

We currently lack detailed knowledge about the mechanism of hybrid incompatibilities that will allow one to interpret the phenotypic effects of sequence divergence on both the intraspecific and interspecific functions of hybrid incompatibility genes. The receptor tyrosine kinase Xmrk-2 is probably the example where the relationship between the interspecific hybrid incompatibility phenotype and the molecular evolution of the hybrid incompatibility gene is most clear. Xmrk-2 causes tumors and lethality in Xiphophorus maculatus (platyfish)/X. helleri (swordtail) hybrids. Xmrk-2 is a gene duplicate in X. maculatus that appears to be absent in X. helleri (Weis and Schartl 1998). It thus presumably causes hybrid lethality because its expression is lethal in the X. helleri genetic background to which it is foreign. Hmr-dependent hybrid lethality in larvae correlates with a failure in cell proliferation but identification of the specific cause of this proliferation defect remains elusive (Orr et al. 1997; Bolkan et al. 2007). The D. simulans Lethal hybrid rescue gene (Lhr; also known as HP3) interacts with Hmr to cause hybrid lethality, but little is known about its intraspecific function, other than that its encoded protein associates and colocalizes with the heterochromatin-localized HP1 protein (Brideau et al. 2006; Greil et al. 2007). Hemizygosity for D. simulans Nup96 also causes lethality in D. melanogaster/D. simulans hybrids (Presgraves et al. 2003). Nup96 encodes an essential component of the nuclear pore; whether hybrid lethality results from a failure of nuclear pore function remains unknown. Odysseus (Ods) causes male sterility in D. simulans/D. mauritiana hybrids (Ting et al. 1998). The mechanism leading to sterility is unknown but it may be due to misexpression that occurs in the hybrid background compared to the pure species. A knockout of Ods in D. melanogaster causes a subtle reduction in male fertility; whether this phenotype involves a milder manifestation of the complete sterility phenotype seen in hybrids is not known (Sun et al. 2004).

While these cases argue for the need of a more mechanistic understanding of hybrid incompatibility phenotypes, observations such as the misexpression of Ods in hybrids reinforce the notion of the distinct nature of hybrid vs. pure-species genetic backgrounds. The data reported here are also consistent with this view. Substitutions in Hmr between D. melanogaster and its sibling species cause quantitative differences in fertility in a pure D. melanogaster background but have a discrete lethal vs. viable effect in species hybrids.