Supporting Online Material for

Transcription

1 Supporting Online Material for Epigenetic Licensing of Germline Gene Expression by Maternal RNA in C. elegans Cheryl L. Johnson and Andrew M. Spence* *To whom correspondence should be addressed. Published 2 September 2011, Science 333, 1311 (2011) DOI: /science This PDF file includes: Materials and Methods SOM Text Figs. S1 to S10 Tables S1 to S6 References

2 Supporting Online Material Materials and Methods: Nematode maintenance and alleles Nematodes were cultured as described (24). All experiments were conducted at 20 C. The N2 Bristol isolate served as the wild-type strain. Except as noted, the following mutations are described in Wormbase ( Linkage group II: dpy-1(e1), unc-45(r450), fem-2(e2105). IV: dpy-13(e184sd), mor-2(e1125), dpy- 20(e1282ts), egl-23(n601dm), fem-1(e2195), fem-1(e2196), fem-1(e2267), fem-1(e2268), him-6(e1423), unc-5(e53), unc-24(e138), unc-30(e191). V: rde-1(ne219). X: egl- 36(n728dm), lon-2(e678), unc-7(e5). The rearrangement nt1dm is nt1[let(n754) unc(dm)]iv;v. The fem-1 deficiencies iddf1 and iddf2 were previously described as fem-1 alleles e2044 and e2382 respectively (3, 25). A third deficiency iddf3 was isolated as a spontaneous unc-5 allele (formerly ev447) in strain RW7097 and kindly provided by J. Culotti. Each deficiency was outcrossed to wild type at least six times. The extent of each deficiency was mapped by Southern blotting and PCR, and the break points were sequenced. Their coordinates are available from WormBase ( Quantification of maternal-effect embryonic lethality iddf2 and iddf3 cause incompletely penetrant maternal-effect embryonic lethality. The penetrance of this phenotype was measured by mating fem-1 unc-24 females with wild type males. Parents were transferred to new plates every twelve hours, and the number of embryos laid was recorded. Dead embryos were counted after 24 hours, and surviving adults were counted after an additional 48 hours. Results are shown 1

3 in Table S4. No additional zygotic contribution to lethality was observed in self-progeny of fem-1(df) unc-24 hermaphrodites. iddf1 complements the other deficiencies for lethality, which therefore does not result from elimination of fem-1. Presumably it is a consequence of deletion of neighbouring genes including drp-1 (26). Quantification of germ-line feminization fem-1(y) unc-24; unc-7 females were crossed to wild-type males, where y represents a given fem-1 allele. Adult F 1 XX females (fem-1(y) unc-24/+; unc-7/+) and XO animals (fem-1(y) unc-24/+; unc-7/o) containing oocyte-like cells were scored as Fog. The X-linked unc-7 mutation prevents XO animals from fertilizing their XX siblings. Occasional XX animals were sterile, with germ-line proliferation defects (~1% of progeny of iddf2 and iddf3 females); they were not included in counts of Fog and non-fog F 1. Experiments in which fem-1 was marked with mor-2 rather than unc-24 produced similar results. In some experiments, a paternally inherited egl-23 or egl-36 allele conferred a dominant egg-laying defective phenotype to ensure that XX hermaphrodites producing a small number of embryos were not incorrectly scored as females. fem-1 alleles tested included e2268, e1965, e1991, e2195, e2196, and e2267: all gave the same result as presented in Table 1 for fem-1(e2268). Heteroallelic females were generated by crossing fem-1(y) unc-24/+ males with fem-1(z) unc-24(e138)/nt1dm hermaphrodites. XX cross-progeny of these females were transferred to new plates as L4 larvae to avoid fertilization by F 1 males. Isolating chromosome IV paternal disomics Paternal disomic animals were identified as Dpy non-unc animals among the F 1 progeny of crosses between fem-1(+) dpy-20 him-6 males and iddf2 him-6 unc-30 2

4 females. The him-6 mutation causes frequent non-disjunction leading to production of aneuploid gametes (27). Paternal disomic fem-1(+) dpy-20 him-6 XX animals were distinguished from dpy-20 him-6/iddf2 him-6 unc-30 XXX animals, which are also Dpy, by crossing to dpy-20 males and verifying that their progeny were all Dpy. Heritable effects on germ-line feminization The persistence and increased penetrance of the Fog phenotype over several generations was observed by performing backcrosses. dpy-13/+ males were crossed to fem-1(y) unc-24 females. Semi-Dpy F 1 males (dpy-13/fem-1(y) unc-24) from each series were mated with fem-1(y) unc-24; unc-7 females. Germ-line feminization of their F 2 semi-dpy progeny was scored. To continue the experiment over several generations, semi-dpy males were crossed to fem-1(y) unc-24 females. Recombinant progeny were identified by their Unc semi-dpy or non-unc non-dpy phenotypes and excluded from the analysis. Heritable germ-line feminization of non-fog dpy-13/fem-1(y) unc-24 series IV F 1 animals was also assessed. Semi-Dpy XX animals were cloned during the L4 larval stage. The Dpy, Unc and Fog phenotypes were scored for all the self-progeny of each F 1 hermaphrodite. Semi-Dpy Unc and non-dpy non-unc recombinants were not analyzed. RNA injection Linearized plasmid DNA served as a template for in vitro transcription. Plasmids are available on request. Roche T3 or T7 RNA polymerase was used according to the manufacturer s instructions. In most experiments, template DNA was removed by treatment with DNaseI (Ambion). Transcripts were purified following electrophoresis on a non-denaturing agarose gel using Bio 101 Systems RNaid Spin Kit. 3

5 Injection mixes contained 100 nm RNA, and in some experiments, 0.125% Lucifer Yellow dye to aid in monitoring the injection. Injection into the germ line of iddf2 unc- 24; unc-7 or iddf2 mor-2; unc-7 females was performed as described (28). Females were crossed to wild-type males and transferred to new plates daily. The requirement for rde-1 was tested by injecting females of genotype iddf2 mor-2; rde-1; unc-7 and crossing them to rde-1 males. The Fog phenotype was scored in cross-progeny born hours after injection, after they reached adulthood. Preliminary experiments indicated that the effects of RNA injection were strongest among animals born during that period. The offspring of at least 20 injected females were scored for each RNA tested. Plates containing fewer than 15 animals were not counted. Injected animals were assigned to bins according to whether they produced <25%, 26-50%, 51-75%, or >75% Fog progeny during the period scored. The bin distribution of animals injected with each type of RNA (plotted in Figs. 1, S3, S5, S9) was compared to that of uninjected controls, and to that of animals injected with the sense transcript of fem-1 cdna, using the Mann-Whitney U test in the SPSS statistical software package. Sample sizes and p values are listed in Table S5. Measuring fem-1 maternal rescue lon-2 males were crossed to fem-1(y) unc-24; lon-2 females (Series I, Fig. 2) and to fem-1(y) unc-24/nt1dm; lon-2 hermaphrodites (Series II, Fig.2). Non-Unc F 1 XX (fem- 1(y) unc-24/+; lon-2) were crossed to fem-1(e2268) unc-24/+ males. Unc, Lon F 2 XO animals (fem-1(y) unc-24/fem-1(e2268) unc-24; lon-2/o) were identified. The X-linked marker differentiates Lon XO animals from non-lon XX animals. 4

6 Somatic masculinization of each F 2 XO animal was assessed using four categories. Level 1: two-armed gonad, vulva, truncated hermaphrodite tail. Level 2: twoarmed gonad, partial vulva, rudimentary fan, crumpled spicules, < 4 rays. Level 3: abnormal gonad, partial vulva, small fan, short spicules, 4 to 8 rays. Level 4: one-armed gonad, no vulva, near-normal fan, normal spicules, > 8 rays. Measuring fem-2 maternal rescue unc-45 fem-2/dpy-1; lon-2 males were crossed to dpy-1; iddf2 unc-24; lon-2 females (Fig. S6, Series III). Non-Dpy F 1 XX (dpy-1/unc-45 fem-2; iddf2 unc-24/+; lon- 2) were crossed to unc-45 fem-2/+ males. Unc, Lon F 2 XO animals (unc-45 fem-2; +/+ or iddf2 unc-24; lon-2/o) were scored. To demonstrate the degree of feminization shown by fem-2 animals in the absence of a maternal contribution of fem-2(+), m-z- animals were produced by crossing unc-45 fem-2; lon-2 females to unc-45 fem-2/+ males (Fig. S6, Series IV). Unc, Lon XO progeny (unc-45 fem-2; lon-2/o) were scored. The full extent of maternal rescue provided by uncompromised fem-2(+) alleles was measure by crossing unc-45 fem-2/ dpy-1; lon-2 hermaphrodites to unc-45 fem-2/+ males (Fig. S6, series V) and scoring their Unc, Lon XO progeny (unc-45 fem-2; lon-2/o). Somatic masculinization was assessed using the same criteria described for fem-1 maternal rescue. RNA in situ hybridization fem-1 RNA levels were measured in the dissected germ lines of adult fem-1(y) unc-24 animals and the adult F 1 progeny of fem-1(y) unc-24 females crossed to wild-type males. Each hybridization tube also contained wild-type animals of the opposite sex as a control to ensure that the hybridization reactions were equally effective. Gonad dissection, fixation and RNA in situ hybridization were performed as described (29, 30). 5

7 EDTA (0.1mM) and 1mM aurintricarboxylic acid were also included during dissection. Sense and antisense fem-1 DNA probes were synthesized with DIG-dUTP using a PCR DIG Probe Synthesis Kit (Roche). An alkaline phosphatase-conjugated anti-dig antibody (Roche) was used at 1:2500. Control sense probes produced no signal. A 2x2 contingency chi-square test was used to compare the frequencies of fem-1 RNA staining in fem-1(e2268)/+ and iddf2/+ gonads (n=280). Expected values were calculated using the null hypothesis that the proportion of stained gonads for each genotype is not different. The chi-square value was 93; given one degree of freedom, we reject the null hypothesis at p< Restoring activity to compromised alleles Paternally contributed fem-1(+) alleles marked with dpy-13 were crossed to fem- 1(y) unc-24 females for two generations. F 2 semi-dpy males were then crossed to fem- 1(e2268) unc-24; unc-7 females, and the Fog phenotype of their semi-dpy progeny was assessed. RT-PCR analysis RNA was extracted using Trizol (Invitrogen) from samples of 50 young adults of genotypes fem-1(+), fem-1(e2268), iddf1 or iddf2. To ensure that all animals analyzed were female, those carrying fem-1(+) were also homozygous for fem-3. cdna was synthesized from 1 g of RNA using the Superscript First Strand cdna Synthesis Kit (Invitrogen). Serial dilutions of cdna samples served as templates for PCR amplification of fem-1, drp-1, and T12E12.2 products. Primer sequences are available on request. Amplification of him-3, mex-3, and pgk-1 cdnas provided standards. Following electrophoretic separation of the products and ethidium bromide staining, product levels 6

8 were measured by densitometry of images of reactions in which amplification was proportional to cdna input. The level of each product was normalized first to the geometric mean of the levels of the standards in the same sample and then to the normalized level of gene product detected in the fem-1(+) sample. Characterization of fem-1 transcript from iddf1 chromosome Total RNA (1 g) from iddf1 females, fem-1(+) hermaphrodites, or fem-3 females was amplified by 3 RACE-PCR using the Invitrogen 3 RACE kit according to the manufacturer s recommendations. Products specific to the iddf1 sample were cloned and sequenced (The Centre for Applied Genomics, Hospital for Sick Children, Toronto). The cloned cdna fragments included fem-1 exons 2-4 and extended through the iddf1 breakpoint in intron 4 to poly(a) addition sites located nucleotides downstream. In 6 of 11 clones the poly(a) tail immediately followed the complete complement of 21UR (Fig. S9B). The cloned fem-1 iddf1 3 RACE product and a wild type fem-1 cdna were used as templates in recombinant PCR to produce a cdna extending from the normal fem-1 mrna 5 end to the most frequent poly(a) site observed in iddf1 fem-1 cdnas. That cdna (Fig. S9B) served as a template for in vitro transcription to prepare RNA for injection into iddf2 females. Supplementary Results and Discussion: Factors contributing to the penetrance of the fem-1(df) maternal effect. iddf1 causes less frequent maternal-effect germ-line feminization than either iddf2 or iddf3. We tested two hypotheses, which are not mutually exclusive, that might 7

9 explain the difference: the first is that whereas iddf2 and iddf3 prevent fem-1 expression by respectively deleting the promoter region or the entire gene, iddf1 could allow production of a truncated fem-1 transcript that retains partial licensing activity. We found that iddf1 females indeed produce polyadenylated fem-1 transcripts that extend through exons 1-4 and beyond the deletion breakpoint, which joins the sequence of fem-1 intron 4 to sequence from the fem-1 unc-5 intergenic region. This region includes many sequences on both strands that encode 21U RNAs (13), named for their 21-nucleotide length and their tendency to have a 5 U residue. fem-1 transcripts from the iddf1 chromosome are polyadenylated nucleotides downstream of the deletion breakpoint, and they include, immediately upstream of their poly(a) tails, nucleotides complementary to the sequence of one 21U RNA, 21UR-9867 (Fig. S9). 21U RNAs associate with the worm Piwi protein PRG-1 and may be analogous to pirnas from other organisms (31-33). Some 21U RNAs contribute to silencing the transposable element Tc3, but the functions of the great majority remain unknown (32). Levels of the truncated fem-1 transcripts in iddf1 females were comparable to those of the full-length transcript in wild type hermaphrodites. Hence the iddf1 maternal effect is not likely to result from reduced accumulation of maternal transcript, but might instead reflect the smaller size or altered sequence of the maternal RNA. Injection of iddf1 fem-1 RNA into fem-1(e2268) females did not induce any germ-line phenotype in their progeny, arguing that the abnormal transcript did not interfere with zygotic fem-1 expression. The iddf1 fem-1 transcript only weakly rescued spermatogenesis in the heterozygous progeny of injected iddf2 females (compare Fig. 1A,B with Fig. S9, and see Table S5). These observations support the idea that fem-1 is susceptible to silencing in the progeny of 8

10 iddf1 females because the truncated, chimeric iddf1 fem-1 transcript is relatively ineffective at licensing its expression. iddf2 and iddf3 eliminate expression of fem-1, but they also differ from iddf1 in affecting the drp-1 operon, which lies upstream of fem-1 in the opposite orientation. iddf2 deletes the fem-1-drp-1 intergenic region, which contains the operon promoter, and iddf3 deletes the entire operon (Fig. S1). Therefore a second hypothesis to explain the severity of the maternal-effect of these deficiencies is that one or more genes in the operon contribute to the effect. To test this idea, we used RNAi to inhibit the activity of these genes in iddf1 females and measured the frequency of the Fog phenotype among their heterozygous cross-progeny. Three of the four genes did not significantly affect the phenotype, but inhibition of T12E12.2, the third gene in the operon, enhanced the severity of the maternal effect of iddf1 to levels comparable to those of iddf2 or iddf3 (Table S6). Loss of T12E12.2 does not entirely account for the fem-1(df) maternal effect, however. RNAi targeting T12E12.2 did not feminize wild-type animals or the heterozygous cross-progeny of fem-1(e2268) females, nor did it enhance the maternaleffect germ-line feminization caused by mutations in fem-3 (Table S6). Moreover, iddf1 does not affect expression of T12E12.2 (Fig. S10). Therefore we interpret the maternal effect of iddf1 to be a consequence of compromising RNA-mediated licensing. It is important to note that although the penetrance of the effect is low (~15-20%) in the F 1 cross-progeny of iddf1 females, it increases in their backcross progeny, exceeding 90% in the third generation, despite normal maternal activity of T12E12.2 (Fig. 3E). These results suggest that T12E12.2 is not generally required for fem-1 expression or for 9

11 spermatogenesis; instead it appears to offer some protection from silencing when maternal RNA-mediated licensing is compromised. T12E12.2 cannot be essential for fem-1 licensing, because the injection of fem-1 RNA into iddf2 females licenses fem-1 expression in the absence of T12E12.2. The product of T12E12.2 is a chromodomain protein, suggesting that it associates with chromatin, consistent with the idea that fem-1 silencing involves regulation of chromatin. T12E12.2 has previously been implicated in transgene-induced cosuppression (22), but its specific roles in that process and in preventing fem-1 silencing await further analysis. 10

12 References for Supporting Online Material 24. S. Brenner, Genetics 77, (1974). 25. A. M. Spence, A. Coulson, J. Hodgkin, Cell 60, (1990). 26. A. M. Labrousse, M. D. Zappaterra, D. A. Rube, A. M. van der Bliek, Molecular Cell 4, (1999). 27. J. Hodgkin, H. R. Horvitz, S. Brenner, Genetics 91, (1979). 28. C. Mello, A. Fire, in Methods in Cell Biology: Caenorhabditis elegans: Modern Biological Analysis of an Organism, H. F. Epstein, D. C. Shakes, Eds. (Academic Press, San Diego, 1995), vol. 48, pp M. H. Lee, T. Schedl, Genes Dev. 15, (2001). 30. R. Francis, M. K. Barton, J. Kimble, T. Schedl, Genetics 139, (1995). 31. P. J. Batista et al., Mol Cell 31, (2008). 32. P. P. Das et al., Mol Cell 31, (2008). 33. G. Wang, V. Reinke, Curr Biol 18, (2008). 11

13 Legends for Supplementary Figures Figure S1 fem-1 Deficiencies. The extent of each of the three deficiencies, iddf1, iddf2, and iddf3 is indicated by a heavy barred line beneath a map of the corresponding region of chromosome IV. Boxes represent genes. The region between the dashed lines is enlarged in the lower part of the diagram to indicate the extent of overlap between iddf1 and iddf2 with respect to fem-1. Boxes represent exons in the enlarged map. A bent arrow indicates the direction of transcription of fem-1. The position of the nonsense mutation e2268 is indicated (6). Figure S2 Maternal-effect feminization of paternally disomic fem-1(+) progeny of fem-1(df) females. In the schematic diagrams of chromosome IV, filled boxes represent wild-type alleles, and open boxes represent mutant alleles. Rare paternal disomics are identified by their Dpy, non-unc phenotype (4). Figure S3 Injection of transcripts from fem-1 flanking sequences and from unrelated genes. Frequency of Fog progeny produced by iddf2 females injected with RNA transcribed from the indicated templates. The progeny of at least 20 females injected with RNA from each template were scored. Asterisks indicate distributions significantly different from that of the uninjected control females shown in Fig. 1A: * 10-4 < p < 10-2, Mann-Whitney U test. Figure S4 Features of fem-1 RNA and injected fragments. In the schematic diagram of fem-1 mrna, open boxes represent untranslated regions; yellow indicates the region encoding ANK repeats, and blue indicates the remainder of 12

14 the coding region. Black vertical lines denote exon boundaries. The coding potential of each reading frame is shown below, with arrowheads indicating start codons and vertical lines stop codons. Sequences deleted by iddf1 and iddf2 are indicated by black bars. Red bars represent the fem-1 sequences included in the RNA fragments injected into iddf2 females (Fig. 1D-F). Figure S5 rde-1(+) activity is not required for the fem-1(df) maternal effect or for rescue of spermatogenesis in the progeny of iddf2 females by RNA injection. Fraction of rde-1; iddf2 females producing indicated fractions of Fog progeny without injection (A) or following injection with RNA transcribed from fem-1 cdna (B). n = number of injected females. Asterisks indicate distribution significantly different from that of the uninjected control females: *** p < 10-6, Mann-Whitney U test. Figure S6 fem-2(+) maternal rescuing activity is not reduced in descendants of fem-1(df) homozygotes. A. Similarly to series I F 1 XX animals (Fig. 2), the series III F 1 XX are descended from iddf2 homozygotes; however, the series III F 2 XO animals scored are fem- 2(-); fem-1(+) rather than fem-2(+); fem-1(-). B. In the absence of maternal fem-2(+), fem-2 m-z- XO animals show extensive feminization. C. Maternal fem-2(+) activity completely rescues male development in fem-2 m+z- XO animals. The results of B and C are consistent with previous reports (3). 13

15 Figure S7 Reduced fem-1(+) activity in non-fog descendants of fem-1(df) females. A. Crossing scheme to examine self-progeny of non-fog descendants of deficiency homozygotes. B. Heritably reduced fem-1(+) activity in F 1 suggested by reduced F 1 brood size, occurrence of the Fog phenotype in F 2 Dpy [fem-1(+)/fem-1(+)] and semi-dpy [fem-1(+)/fem-1(y)] animals, and failure of maternal rescue of F 2 Unc [fem- 1(y)/fem-1(y)] animals. Median brood sizes of these F 1 hermaphrodites are presented in Table S1. Figure S8 Activity of a compromised allele can be restored by crossing it to a fem-1(e2268) female. Mean frequency of the Fog phenotype in at least four crosses, ± standard deviation. Over 450 F 3 animals of each genotype were scored. Figure S9 Structure of iddf1 chromosome and its fem-1 transcript. A. The extent of iddf1 is shown by the heavy barred line beneath a map of the fem-1 region of wild type chromosome IV. Boxes denote exons, and flag symbols represent loci encoding 21U RNAs. Genes transcribed from top and bottom strands are respectively indicated with magenta and cyan. Below is a map of the iddf1 chromosome and a diagram of the truncated fem-1 transcript that it encodes The triangle represents sequence antisense to 21U RNA B. The sequence of fem-1 cdna from iddf1 females. Uppercase and lowercase letters respectively indicate fem-1 coding and noncoding sequence. The yellow 14

16 box indicates the iddf1 breakpoint, and the red underlined sequence indicates the antisense 21U RNA 9867 sequence. C. Frequency of Fog progeny produced by iddf2 females injected with RNA transcribed from the fem-1_iddf1 cdna shown in B. Asterisk indicates distribution significantly different from uninjected control females (Fig. 2A) (p = 8 x 10-3, Mann-Whitney U test). Figure S10. Levels of drp-1 and T12E12.2 transcripts in fem-1 females. Transcripts of drp-1 and T12E12.2 were detected by RT-PCR in total RNA from young adult females of the indicated genotypes. [fem-1(+) animals were homozygous for fem- 3(e1996), so that all comparisons involved females.] Bars indicate mean relative transcript level ± standard deviation. See Methods for details. 15

17 Figure S1. 16

18 17

19 Figure S3 18

20 Figure S4 19

21 Figure S5 20

22 Figure S6. 21

23 Figure S7. 22

24 Figure S8. 23

25 Figure S9 24

26 Relative transcript level fem-1(+) fem-1(e2268) iddf1 fem-1 Genotype drp-1 T12E12.2 Figure S10 25

27 Table S1. Brood sizes of +/fem-1(df) m-z+ hermaphrodites Maternal genotype* n Median brood size of m-z+ heterozygote Range iddf iddf iddf * See Materials and Methods for complete genotypes. Brood size in C. elegans hermaphrodites is limited by the number of sperm produced, and the hermaphrodite offspring of fem-1(df) females had smaller brood sizes than wild type, suggesting decreased spermatogenesis. Table S2: fem-1(df)/+ progeny of heterozygotes rarely exhibit germ-line feminization Maternal Genotype * Percent Fog fem-1(df)/+ F 1 b XX n c XO n c iddf1/ iddf2/+ 0.1 ± iddf3/+ 0.3 ± nd * See Materials and Methods for complete genotypes. Mean ± standard deviation. n = Number of crosses. Total number of animals scored for each genotype was

28 Table S3. Comparison of fem-1 epigenetic licensing to other genetic and epigenetic phenomena Phenomenon Description Similar to fem-1(df) effect? (reason) Maternal effect The phenotype of an individual depends on its mother s genotype rather than, or in yes (1, but see 2) addition to, its own. Paternal effect The phenotype of an individual depends on its father s genotype rather than, or in no (3) addition to, its own. Haploinsufficiency +/lf mutants exhibit a mutant phenotype. no (1) Transvection Phenotype is affected by pairing between homologous chromosomes. no (1,4) Paramutation The activity of a paramutable allele is heritably reduced as a result of interaction no (1,4,5) with a paramutator allele. Imprinting Whether an allele is expressed depends on whether it is inherited from mother or no (1) father. Allelic Exclusion Only one of two alleles of a locus is expressed. no (1) Meiotic Silencing of Unpaired DNA In meiotic prophase, chromosomal regions that lack a homologous pairing partner in meiosis accumulate chromatin modifications associated with heterochromatin. no (1,4) Segregation Distortion Genes in those regions are silenced. SD is revealed by inheritance of one allele at a locus by greater than 50% of the progeny of a heterozygote. 1. +/fem-1(df) heterozygotes exhibit Fog phenotype only if descended from fem-1(df) homozygous females. +/fem-1(df) descendants of crosses between wild type males and +/fem-1(df) hermaphrodites or fem-1(e2268)/fem-1(df) females do not exhibit the Fog phenotype. These crosses demonstrate the importance of maternal genotype and show that a single paternally contributed fem-1(+) allele is sufficient to support complete male development in animals not descended from fem-1(df) homozygous females. 2. Other (non-df) fem-1 alleles, including several that prevent expression of FEM-1 protein, do not show such a maternal effect. 3. +/fem-1(df) heterozygotes have similar phenotypes whether descended from +/+ male or +/fem-1(df) male, as long as the maternal genotype is the same in each cross. (We can t make males homozygous for fem-1(df), as fem-1(+) is required for male development.) 4. Paternal disomic fem-1(+) homozygotes descended from fem-1(df) female show the Fog phenotype. 5. iddf3 entirely lacks fem-1 and its regulatory region any deleterious interaction would have to involve sequences more than 20 kb downstream or more than 100 kb upstream of fem-1. no (6) 27

29 6. The fem-1(df) maternal effect is observed in the heterozygous progeny of a cross between inbred homozygotes, in which there is no opportunity for distortion. Both fem-1(+) and fem-1(df) alleles segregate from the affected heterozygotes at the expected frequency: fem-1(+) can be detected by its masculinizing activity in somatic tissues even when it has been silenced in the germ line. 28

30 Table S4: Maternal-effect lethality of fem-1 deficiencies Maternal fem-1 genotype* % Embryonic lethality (n) fem-1(+) 1 (410) iddf1 n.d. iddf2 36 (887) iddf3 62 (701) iddf1/iddf2 0.2 (554) iddf1/e (490) iddf2/e (428) * See Materials and Methods for complete genotypes. n.d. not determined 29

31 Table S5. Statistical significance of effect of RNA injection* Template None (uninjected iddf2) Number of Animals p vs. Uninjected iddf2 50 n.a. 3.2 x fem-1 cdna x n.a. fem-1_xatg x fem-1_ x fem-1_ x fem-1_ x fem-1_genomic x fem-1_anti x x 10-4 fem-1_5 flank x x 10-3 fem-1_3 flank x x 10-4 fem-1_intron x x 10-3 ds_fem x x fem-2 cdna x fog-3 cdna x 10-5 GFP cdna x 10-6 mom-5 cdna x 10-6 iddf1 fem x x 10-2 p vs. fem-1 cdna Injection * Results of these injections are shown in Fig. 1, S3, S9. probability that the observed distribution does not differ from uninjected controls. Mann- Whitney U test. (n.a. not applicable) 30

32 probability that the observed distribution does not differ from animals injected with fem-1 sense cdna transcript. Mann-Whitney U test. (n.a. not applicable) 31

33 Table S6. RNAi targeting T12E12.2 enhances the maternal effect of iddf1 Maternal genotype* RNAi target Percent Fog F 1 XX XO iddf1 none 21 ± 5 10 ± 2 drp-1 18 ± 7 T12E ± 13 T12E ± 8 75 ± 2 T12E ± 6 iddf2 none 93 ± 4 85 ± 9 T12E ± 2 83 ± 6 iddf3 none 92 ± 1 80 ± 3 T12E ± 4 86 ± 4 fem-1(e2268) none 0 ± 0 0 ± 0 T12E ± 0 0 ± 0 fem-3(e1996) none 14 ± 1 T12E ± 3 * See Materials and Methods for complete genotypes and procedure. None refers to worms grown on bacteria carrying empty feeding RNAi vector. Mean ± SD. p < 10-6 compared to no RNAi control (t-test, unequal variance). For all others, p > 0.1 compared to corresponding no RNAi control. 32