Current Genetics

Curr Genet (1989) 15:327-334

© Springer-Verlag 1989

Premeiotic disruption of duplicated and triplicated copies of the Neurospora crassa am (glutamate dehydrogenase) gene J. R. S. Fincham, I. E Connerton*, E. Notarianni, and K. Harrington Department of Genetics, University of Cambridge, Downing Street, Cambridge CB2 3EH, UK

Summary. Premeiotic inactivation of duplicated sequences (the RIP phenomenon of Selker et al.) was studied by tetrad analysis using ectopic copies of a m + (coding for NADP-specific glutamate dehydrogenase) and a missense allele am 3, coding for a distinctive form of the enzyme, at the normal locus. In duplication crosses either both gene copies were inactivated or neither. Two inactivated a m 3 derivatives were shown to have undergone methylation and numerous base-pair changes, reflected in losses and gains of restriction sites, but without sequence rearrangement. Cutting at restriction sites within the disrupted sequences was incomplete but became almost complete following growth in the presence of 5-azacytidine. In a triplication cross in which one parent carried two unlinked ectopic gene copies together with a m 3 at the normal locus, premeiotic inactivation, when it occurred, tended to affect two of the three copies in any one ascus, but there were a few asci in which all three were inactivated. Key words: N e u r o s p o r a crassa - ectopic genes - methylation - RIP phenomenon

Introduction Efficient methods of transformation with cloned DNA have been devised for filamentous fungi, especially N e u r o s p o r a crassa (Case et al. 1979; Kinsey and Rambosek 1984; Vollmer and Yanofsky 1986) and Aspergillus nidulans (Tilburn et al. 1983). Unlike the situation in S a c c h a r o m y c e s cerevisiae, transformants

Department of Microbiology, University of Reading, London Road, Reading RG1 5AQ, UK * Present address:

Offprint requests to:

J. R. S. Fincham

in filamentous fungi may be due either to homologous or to ectopic integration of the transforming DNA, with the ectopic class usually in the majority. It is a common experience that transformants, even when they seem to be stable during haploid growth, often transmit their transformed character inefficiently through sexual crosses. In N. crassa, Case (1986) showed that ectopically integrated qa-2 + often lost its activity when put through a cross, even though the DNA sequences that should have contained the gene were still detectable by molecular probing of the progeny cultures. Selker et al. (1987a) have produced compelling evidence that duplicated Neurospora DNA sequences that are normally present in single copy can both be subjected to premeiotic disruption (called RIP, for rearrangements induced premeiotically) manifesting itself both in new patterns of restriction fragments and in heavy methylation of the duplicated sequences; the nature of the hypothetical rearrangements was not demonstrated. An obvious hypothesis is that the premeiotic disruption is triggered by premature pairing of repeated sequences in the still haploid nucleus. This hypothesis would predict that, with a triplication in one parent of a cross, only two of the three gene copies would usually be inactivated at the same time. The availability in our laboratory of transformant strains with ectopically integrated a m + (NADP-specific glutamate dehydrogenase coding) sequences, as well as normally located am mutant alleles with distinctive products, provided a good opportunity for further investigation of the RIP effect.

Materials and methods The standard wild-types were STA (74A) and STa, a closely-related strain of opposite mating type. The am 132 mutant, which is deleted with respect to the entire am + transcribed sequence, was present in the two strains 132-5-6A and 132-5-16a, Neurospora strains.

328 Table 1. Ascus analysis of crosses involving derivatives of transformants T75 and T85. Derivation of strains: T75-1, T75-2 and T85-1 were am +like ascospore derivatives from crosses of the primary transformants to am 132a. T75-1-1 and T75-1-2 were derived from a further cross ofT75-1 to am 132 a. 92-12 was derived from cross (4) and 93-56 from the analogous cross of a T85 derivative to am3al a. 277-78 was from cross 9 Cross

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

"

T75-2 X am+a T85-1 ;< am+a T75-2 X am132a T75-1-1 X am3a T75-1-2 X am3a T85-1 X am132a 92-12 (amT75 + am3a) X am132A 93-56 (amT85 + am3alA) )< am132ab 92-12 X 93-56 b 277-78 (am 3 amT75 + amT85+ al) X am132a b

Ratio growth/non-growth on minimal glycine or (cross 9) presence/absence of normal G D H 4:0

3:1

2:2

1:3

0 0 0 0 0 0 0 0 2 3

5 2 0 1a 0 0 0 0 1 3

0 1 8 11 1 10 22 16 6 9

0 0 0 0 0 0 0 0 0 0

0:4 0 0 0 0 0 0 8 1° 6 16

Classified by G D H assay

Parental ditype

Tetratype

Non-parental ditype

12 2:2 tetrads from cross (7) 11 2:2 tetrads from cross (8)

1 2

7 7

4 2

a Presumed to be the result of a dissection error b All tetrads segregated 2:2 for a/+ vs al c This unusual tetrad segregated 2:2 for am 3 vs am 132 and showed significant amT85 + activity in one product (determined by G D H assay)

both isolated after five rounds of crossing to STA. The mutant am 3 has a single base-pair substitution which has the effect of replacing glutamate with glycine (Brett et al. 1976) and renders the G D H product inactive under physiological conditions while leaving it capable of full activation in vitro at p H 8.5 in the presence of 0.15 M L-glutamate (Fincham 1962). The strain am3al a was from our own stock collection and is not closely inbred with the other strains used. Origins of derived strains are detailed in Table 1. DNA clones, p8-3 consists of the 2.64-kb BamHI fragment that

includes the whole of the am+-transcribed sequence (Kinnaird and Fincham 1983) cloned into the Co/El-derived plasmid pUC8. It was constructed and kindly donated by Dr. J. A. Kinnaird and is apparently identical to the plasmid previously used by Grant et al. (1984) to transform am I32. The lambda clone kcl0 (Kinnaird et al. 1982) contains a longer HindIII fragment of about 9.1 kb that includes the 2.64 kb BamHI fragment (Fig. 5). Transformation. Protoplasts ofaml32-5-6A were prepared and transformed with p8-3 DNA essentially as described by Grant et al. (1984) and Vollmer and Yanofsky (1986). Regeneration was obtained in minimal medium supplemented with 0.02 M glycine to inhibit untransformed growth (Grant et al. 1984). Strongly growing colonies present after 4 days of incubation were picked, grown up on glycineminimal and immediately crossed to am132-5-16a. Ascospores produced in these crosses were screened on unsupplemented minimal agar for those (usually a small minority if present at all) with the vigorous branching growth characteristic of am + cultures. Enzyme assays. Mycelia were grown from heavy conidial inocula in

unshaken 250-ml conical flasks containing 50 ml liquid minimal medium (Vogel's) supplemented with 0.5 mg/ml L-glutamate to permit equal growth of am + and am-. After 16h at 30°C and a further 24 h at 25 °C, mycelia were recovered by filtration, washed with water and extracted with about 8 times their blotted weight of 0.05 M p H 7.4 sodium phosphate by grinding with sand. Supernarants obtained after centrifuging at 10,000 rpm for 15 rain contained 3.0-4.5 mg protein/ml. 15 or 20 ~tl samples were assayed for G D H in two different systems at 35°C: (a) 0.15 M sodium L-glutamate, 6X 10-5 M NADP, 0.10 M Tris-HCl pH 8.5; (b) 10 mM disodium c~-ketoglutarate, 22 mM NH4C1, 2.5 X 10-s M NADPH, 0.10 M Tris HC1 pH 8.5. The mycelial extract was added as final addition to the second assay but pre-incubated for 2 min in the first assay with all the components except the NADPH, which was added to start the reaction. Wild-type G D H catalyses the oxidation of N A D P H (decrease in OD at 340 nm) in system (b) about 2.5 times as fast as it catalyses the reduction of NAD (increase in OD at 340 nm) in system (a). The am3-type GDH, on the other hand is inactive in system (b), while giving about the same activity as wild-type in system (a). DNA preparation and Southern analysis. DNA was prepared from

mycelia by the ethanolic perchlorate method of Stevens and Metzenberg (1982). Restriction digests were carried out according to the procedures recommended by the manufacturers of the enzymes. The digested DNA was fractionated by eiectrophoresis in 0.9 % agarose gels, blotted on to Hybond nylon membrane and hybridised to either p8-3 or )~cl0 DNA labelled by nick-translation, following the procedures described by Maniatis et al. (1982) with the modifications for Hybond membrane recommended by the manufacturers (Amersham 1985).

329 Table 2. GDH activities in non-disrupted progeny from duplication

crosses Strain no.

Specificactivities (btmol/min/mg protein)

Inferred genotype

Glutamate Glutamate oxidation synthesis Wild-type STA STa 92-12 93-56 277-78 207-1a 207-3 207-5 207-7 209-1b 209-4 209-5 209-7

115 97 185 265 170 20 210 n.d. 185 18 165 n.d. 242

270 315 83 41 183 62 37 n.d. n.d. 52 46 n.d. n.d.

am + am + am 3 amT75 + am ~ amT85 + am 3 amT75 + amT85 + am 132 amT85 + am 3 amT85 + am 132 am 3 am 132 amT75 + am 3 amT75 + a m 132

am 3

not detectable, ~<2 a Ascus 207 (spores 1, 3, 5, 7) from 93-56 × am 132 b Ascus 209 (spores 1, 4, 5, 7) from 92-12 X am13Z: cf Fig. 1 Neither ascus shows gene inactivation

Results of crosses involving T75 and T85 derivatives are shown in Table 1. Crosses to a m 132 gave regular 2:2 segregation of growth/no-growth on minimal-glycine, indicating that the transforming sequence was stably integrated and remained stable through meiosis. Crosses to wild-type gave a preponderance of 3: 1 segregations, consistent with integration at an ectopic site unlinked to the normal a m locus. Since the transforming sequence had no homology with the genome of the arn 132 deletion mutant, ectopic integration was expected to account for all stable transformants. The ectopically integrated gene copies were designated a m T 7 5 + and a m T 8 5 +. A number of asci were also analysed from crosses of T75 derivatives to a m 3, a missense rather than a deletion mutant. Again, each meiotic tetrad had two functional a m + members with the exception of one, most probably due to a dissection error, which appeared to have three. The a r n T 7 5 + sequence was evidently not destabilised by the presence of a second gene copy in the same nucleus following karyogamy.

n.d.,

Results Origin a n d initial breeding behaviour o f t r a n s f o r m a n t s

Twenty-six strongly growing transformed colonies, designated T71-T96, were picked from the initial selective plates of regenerating a m 1 3 2 - 5 - 6 A protoplasts treated with p8-3 DNA. Only ten of the resulting cultures continued to grow strongly on selective medium after a further conidial transfer. Of these, eight yielded at least a few apparently a m + ascospores on crossing to am132-5-6a but only two, T75 and T85, produced such ascospores at a frequency of more than a few per cent. A third transformant, T73, yielded only about 1% phenotypically a m + spores from the initial cross to a m 132 and continued to transmit the transformed phenotype at a low frequency (5 % or lower) through subsequent crosses to a m 132 or a m 3. Insofar as it was transmitted, the transforming sequence in T73 appeared to be linked to markers on linkage group 1. A total of 24 phenotypically a m + ascospores that were isolated from crosses of various T73 derivatives to a m 3 a l included 17 that showed no recombination between a m + and either a l or mating type and 3, 3 and 2 with a m + recombined with al, mating type and both markers, respectively.

Analysis of duplication crosses

F r o m crosses ofT75 and T85 derivatives to a m 3, strains were obtained that evidently carried both a m 3 and an ectopic copy of a m +. These had relatively low levels of G D H activity: typically about 20% and 25-30% of wild-type for T85 and T75 derivatives, respectively. Such levels of activity are characteristic of the ectopically integrated 2.64 kb B a m H I a m + fragment (Kinsey and R a m b o s e k 1984); T73 had similarly low G D H activity. The presence of a m 3 in these strains was indicated by high (sometimes even higher than wildtype) levels of glutamate-oxidizing G D H activity in an assay system in which the enzyme was preincubated with 0.15 M glutamate at p H 8.5, but without significant glutamate-synthesizing activity (Fincham 1962). When an a m 3 a m T 7 5 + strain (92-12) was crossed to a m 132 (cross 7, Table 1), 22 of the 30 dissected asci from which at least one representative of each spore-pair germinated showed 2: 2 segregation of ability/inability to grow on minimal-glycine. Twelve of these tetrads were assayed for G D H and all showed 2:2 segregation for presence/absence of the low-level wild-type activity characteristic of a m T 7 5 + and also, independently, for presence/absence of am3-type activity (Table 2). The ratio of 1 parental ditype tetrad, 4 non-parental ditypes and 7 tetratypes shows conclusively that a r n T 7 5 + cannot be on the same chromosome as the normal a m 3 locus. The preponderance of first-division segregation of a m T 7 5 + (17 out of 21 scorable asci - data not tabulated) indicated a location relatively close to a centromere.

330 Table3. Analysis o f s o m e cross 9 asci and all cross 10 asci by enzyme assay

Analysis o f disrupted derivatives by Southern blotting: restriction site changes and methylation

Cross

DNA samples from six of the eight aberrant tetrads from the a m 3 / a m T 7 5 + X am 132 cross, and from six of those showing regular meiotic segregation, were digested with H i n d I I I and analysed by Southern blotting, )~cl0 DNA being used as probe. Figure 1 shows some of the results. Wild-type gave the expected 9-kb hybridising fragment and am 132the much reduced fragment of about 2 kb (Kinnaird et al. 1982). T75 had already been shown (D. Bradley, unpublished undergraduate project) to have an am + hybridising sequence linked to some pUC8 sequence within a larger H i n d I I I fragment of about 13-14 kb, as well as the characteristic am 132 deletion fragment. The regularly segregating asci (e.g., ascus 209, Fig. 1) all showed the 9 kb fragment in segregants with am 3 enzyme activity and the am 132deletion fragment in the others. The presence of amT75 + activity correlated completely with the presence of the larger H i n d I I I fragment, which hybridised less strongly, presumably because of the reduced extent of homology with the probe (2.64 kb, as compared with 9 kb for normally located am + or am3). The six aberrant tetrads each showed regular segregation of a 9-kb hybridising fragment versus the am 132 deletion fragment, and also of a fragment similar in size to that associated with amT75 +. In three asci either the am3-1ike or the amT75-1ike fragment was accompanied by one or more additional hybridising fragments not present in either parental strain. Our initial analysis has been focussed upon two asci, 217 and 246, which conform to the usual RIP pattern of apparently identical modifications in both members of the meiotic tetrad inheriting a particular disrupted sequence. The two single-ascospore products, 217-4 and 246-7, each contained a disrupted am 3 sequence but no ectopic am sequence, the latter having segregated into other meiotic products. DNA from these strains was digested with H i n d I I I , E c o R I , BarnH1, BglII and combinations of these enzymes. In each case the total size of the H i n d I I I am+-hybridising H i n d I I I or B a m H I fragments exceeded the 9 kb of the wild-type H i n d I I I fragment. It was evident, however, that not all the hybridising bands could have been generated from non-overlapping genomic sequence, since all the hybridising sequence was contained within a single BglII fragment of about 12 kb, the same size as in the wild-type (results not shown here). Comparisons of wild-type, 217-4 and 246-7 am +hybridising fragments formed in single and double digests with H i n d I I I , B a m H I and E c o R I are shown in Fig. 2. The results strongly suggested that the multiple H i n d I I I bands yielded by 217-4 and 246-7 were due in

9

10

Segregation of a m 3

4:0 2:2 0:4 2:2 0:4

N u m b e r s of asci Segregation of am +: a m 4:0

3:1

2:2

1a. 0 0 3e 0

0 0 0 3f 0

1b 3c 0 2g 7h

0:4 0 0 2d 13i 3k

Interpretation: a No inactivation in either parent - parental ditype for ectopic copies b As a, except non-parental ditype for ectopic copies (ascus 277) ° Inactivation of a m 3 and ectopic a m + in one parent only d Inactivation of a m 3 and ectopic a m + in both parents e All three gene copies transmitted intact; nonparental ditype (NPD) for ectopic a m + copies f As e, except tetratype for ectopic a m + ' s g As e, except parental ditype (PD) for ectopic am+'s; or, less probably, inactivation of one a m + alone h Inactivation of a m 3 together with one a m + ; or, less probably, inactivation of a m 3 alone and PD for the ectopic a m + ' s i Inactivation of both ectopic a m + copies; a m 3 unaffected k All three gene copies inactivated

The other eight asci from 92-12 X aml32-5-6A gave a 0:4 ratio with regard to growth phenotype. Enzyme assays showed neither a m T 7 5 + nor am 3 activity in any of the products of these asci. A cross of an am 3 amT85+al strain (93-56) to am 132 (cross 8, Table 1) gave a rather different result. Of 17 dissected asci, 16 showed 2:2 segregation for growth/ no-growth on minimal-glycine medium. The remaining tetrad consisted of four non-growers. Of the 17, 12 (including the one showing the anomalous 0:4 ratio) were scored by GDH assay and all were found to segregate 2:2 with respect to am 3 activity. All 17 showed 2:2 segregation with respect to the albino marker. The anomalous 0:4 tetrad needs further analysis (see Table 1, footnote), but overall, the frequency of gene disruption (RIP) in this cross appeared to be lower than in the corresponding crosses involving a m T 7 5 + (difference significant at the 5% level). The ratio of 2 parental ditype : 2 non-parental ditype : 7 tetratype segregations with respect to amT85 + and am 3 are consistent with amT85 + being unlinked to the normal am locus. Analysis of the cross 92-12 X 93-56 (cross 9, Table 1) showed ectopic gene inactivation (accompanied, so far as it was tested, by am 3 inactivation) in at least one parent nucleus in at least 9 out of 15 asci, and in both parents in 6 out of 15 (Table 3).

331

Fig. la, b. Southern blot analysis of asci from the cross 92-12 (am3/amT75 ~) X am 132 (deletion mutant). DNA samples were digested with HindIII. The radioactive probe was cl0 DNA containing the 9.1-kb HindIII wild-type fragment including the complete am + sequence (see Fig. 4). The DNA tracks are labelled with ascus number (above) and ascospore number (1-8 from tip to base of ascus) below: a ascus 209, showing regular segregation of intact gene copies; b asci 217,246, both showing loss of activity of both am 3 and am T75 +. 3, T and 132 indicate bands of normal size for am 3, amT75 + and am 132, respectively; 3', T' indicate novel fragments attributable to am 3 and amT75 +, respectively

Fig. 2a, b. Southern analysis of wild-type (1), 246-7 (2), and 217-4 (3) DNA digested with HindIII (H), B a m H I (B), EcoRI (E) and pairwise combinations of these enzymes. The probe was the 2.64-kb wild-type B a m H I fragment. The size calibration was obtained from marker tracks (gel a) or the known sizes of the wild-type fragments (gel b)

332

Fig. 3a-e. Effects of methylation in the arn3-derived sequences in 217-4 and 246-7. a Odd-numbered and even-numbered tracks contain DNA digested with HpaII and MspI, respectively. Tracks 1 and 2, wild-type 74A; 3 and 4, 217-4; 5 and 6, 246-7; 7 and 8, 217-4 grown in presence of 5-azacytidine; 9 and 10, 246-7 grown in presence of 5-azacytidine. b DNA digested with HindIII (tracks 1, 2, and 3), BamHI (4, 5, and 6) or BamHI plus HindlII (tracks 7, 8, and 9). DNA was from wild-type (1, 4, and 7), 5-azacytidine-grown 217-4 (2, 5, and 8) and 5-azacytidinegrown 246-7 (3, 5, and 9). c 246-7 DNA digested with BamHI (tracks 1 and 4), BamHI plus HindIII (tracks 2 and 5) and BamHI plus EcoRI (tracks 3 and 6), from mycelium grown without (1, 2, 3) or with (4, 5, 6) azacytidine

each case to a single novel partially digestible HindIII site within the 2.64 BamHI fragment and that the multiple BamHI bands were due to incomplete digestion of the normally placed BamHI sites delimiting this fragment. A new EcoRI site, also only partially digestible, appeared to have been generated within the 2.64-kb BamHI fragment in 246-7. The EcoRI and BglII sites normally found within the 2.64-kb BamHI fragment had disappeared in both 217-4 and 246-7. Further analysis utilising the )~cl0 probe showed no changes in the positions or digestibility of restriction sites within the 9-kb HindIII fragment but outside the 2.64 kb BamHI fragment that had been duplicated in the parent strain (data not shown here). Paired digestions with HpaII and MspI, and with MboI and Sau3A, showed that, as expected (Selker et al. 1987), the disrupted sequences were extensively methylated. In order to see whether the partial digestion of the restriction sites within the disrupted sequences was due to methylation, and to test the interpretation of the multiple restriction fragments as partial products, 217-4 and 246-7 were grown in medium containing either 50 btM or 100 btM 5azacytidine. These concentrations appeared to be about equally inhibitory to growth, an additional 1 or

am +

<

Ba

E

Bg

Ba' I

I

I L

)

I

I I

Ba' !

I

1 kb

H' !

Ba

H'

Ba'

:

;

I

217-4

Ba' i

I

246*7

E' ;

I

Fig. 4. Restriction site maps deduced from the data shown in Figs. 2 and 3. The 2.64-kb BamHI fragment duplicated in the parent strain 92-12 is boxed Ba, BamHI; Bg, BglII; H, HindIII; E, EcoRI. Broken lines indicate sites only partially cleaved by the corresponding enzymes

11/2 days of incubation at 30°C being required to obtain yields of mycelium in the normal range. DNA isolated from such cultures was substantially but not completely demethylated (results for MspI/HpaII shown in Fig. 3) and gave simpler patterns of restriction fragments indicative of nearly complete cutting of the new internal HindIII and flanking BarnHI sites and substantially increased, though still far from complete, cutting of the novel 246-7 EcoRI site (Fig. 3).

333 Restriction maps summarising the conclusions drawn from this analysis are shown in Fig. 4.

A n a l y s i s o f a triplication cross

From the cross 92-12 ( a m 3 / a m T 7 5 + a ) 3< 93-56 (am3~ a m T 8 5 + A albino) one ascus was identified in which all products carried a m 3 (shown by GDH assay, Table 3) but only 2 were phenotypically a m +. This result was most easily explained as due to a non-parental ditype segregation of the two ectopic sequences, with the two am + products each carrying both a m T 7 5 + and a m T 8 5 +. This interpretation was confirmed by the relatively high level of GDH in these derivatives. One of them, 277-78, which carried the albino marker, was crossed to am 132, and 31 asci were dissected and the products analysed by growth tests and enzyme assay. The scoring is summarised in Table 1 (cross 10) and in more detail, with an interpretation of the data, in Table 3. It is clear that in 6 of the asci all three gene copies had been transmitted without loss of activity and, more surprisingly, that in three asci all three copies had been inactivated. The remaining 22 asci are interpretable as due to inactivation of two copies out of three, but, because of the difficulty of distinguishing with certainty between a m T 7 5 +, a m T 8 5 + and the combination of the two by GDH assay, some of these are susceptible to alternative interpretation. Two of the 22 [category (g) in Table 3] could well belong to the no-inactivation class with parental ditype segregation of a m T 7 5 + and a m T 8 5 +, but the rather low level of am + GDH in one of these asci suggested the possibility of inactivation of just one ectopic gene copy in this case (Table 3). The category (h) asci, from the levels of GDH activity in their a m + products, seemed likely to be segregating for only one active ectopic copy, the other having been inactivated in concert with a m 3.

Discussion

With the doubtful exception of one ascus from a m 3 a m T 8 5 + X am 132, the results from the duplication crosses were consistent with the general rule that premeiotic disruption affects both gene copies or neither. The triplication cross gave 3 out of 31 asci with all three copies inactivated, but overall there was a strong tendency for disruption to affect two out of three copies. The two-out-of-three prediction would imply (i) that all asci with am 3 inactivated would show 2: 2 ratios for presence/absence of a m + GDH, and (ii) that all asci with both ectopic a m + copies clearly surviving (3:1 and 4:0 ratios) would have surviving am 3 activity, as would (iii) all asci with both ectopic

copies inactivated. In fact, prediction (i) holds in 7 out of 10 asci, prediction (ii) holds in 6 out of 6 asci, and prediction (iii) holds in 13 out of 16 asci. The departures from prediction are entirely due to the 3 asci with all three copies inactivated. This encourages the idea that it is pairwise interaction, perhaps premature homologous synapsis, that triggers the methylation and basepair changes. The three asci in which all three gene copies were inactivated could be due to successive rounds of pairing or perhaps associations-of-three with exchanges of partner, analogous to trivalent association of chromosomes in triploid meiosis. It is not clear from the present data whether a m T 7 5 + and a m T 8 5 + differ consistently from each other in their respective frequencies of inactivation in interaction with a m 3. Crosses 5 and 6 (Table 1) show significantly different frequencies of disrupted asci (8 out of 30 and 0 or 1 out of 17 respectively), but a m T 8 5 + is inactivated in at least 6 out of 15 asci from cross 9, in which both parents have duplicated sequences. In the triplication cross (10) interaction between a m T 7 5 + and a m T 8 5 + seems to be more frequent than interaction of either with am 3. This could be due to the greater degree of DNA homology between the two ectopic sequences; both include some plasmid sequence as well as the 2.64kb am-specific sequence shared with a m 3 (results not shown). In interpretation of the enzyme-based scoring it must be borne in mind that retention 0f some degree of GDH activity does not necessarily imply no change in the a m DNA sequence. The two possible cases of apparent inactivation of one gene copy on its own could be due to a pairwise interaction in which both partners suffered some damage but only one was damaged so severely as to totally eliminate the activity of the gene product. Further analysis at the DNA level may clarify these apparent exceptions to the general rule. A main conclusion to be drawn from the DNA analysis is that, in the two disrupted products examined, there was no evident rearrangement of DNA sequence (as might be implied by the R in the RIP acronym) but only methylation and base-pair changes within the duplicated sequences. Selker and Stevens (1985, 1987) found numerous GC-to-AT transitions in naturally occurring duplicated sequences in Neurospora; this is the type of mutation expected from deamination of 5-methyl cytosine. The new restriction sites in 217-4 and 246-7 could have arisen in this way. The known sequence of a m + (Kinnaird and Fincham 1983) does contain, in appropriate positions, sequences convertible to the new observed restriction sites by single GC-to-AT transitions. The partial cleavage of the sites and their more complete cleavage following demethylation by growth

334 on 5-azacytidine are strongly suggestive of partial methylation of specific cytosines, with some more completely methylated than others; the 246-7 HindIII site was more completely cut than the 217-4 HindIII site, and the downstreamBamHI site in 246-7 was more completely cut than the upstream BamHI site. Available evidence indicates that D N A methylation in Neurospora is confined to cytosines, and so it is paradoxical to find an apparent effect of methylation on the digestibility of HindIII sites, which are usually considered to be protected only by adenine methylation. It may be that HindlII sites can be partly protected by heavy C-methylation in adjacent sequence. The maintenance of methylation of RIPed sequences is an intriguing problem. If these sequences are only partially methylated at any one C residue, as our results suggest, one cannot postulate a maintenance methylase acting in a precise manner (Holliday and Pugh 1975). Unpublished results in this laboratory show that methylation persists through crosses, even when the duplication that originally provoked it has been segregated away in a preceding generation. We have also found that the original high level of methylation is restored when substantially but not completely demethylated mycelium is transferred back to medium without azacytidine. One possibility is a methylase that binds to incompletely methylated sequence and then methylates, with something less than 100% probability, any cytosine within reach. On the other hand, Selker et al. (1987b) obtained evidence that, at least in one case, a RIPed sequence was remethylated in Neurospora after complete demethylation in E. coli; the signal for methylation may then be nothing more than a high proportion of A-T base-pairs.

References Amersham International plc (1985) Membrane transfer and detection methods. Amersham, UK Brett M, Chambers GK, Holder JA, Fincham JRS, Wootton JC (1976) J Mol Biol 155:1138-1146 Case ME (1986) Genetics 113:569-587 Case ME, Schweizer M, Kushner SR, Giles NH (1979) Proc Natl Acad Sci USA 76:5757-5763 Durrens P, Green PM, Arst HN Jr, ScazzocchioC (1986) Mol Gen Genet 203:44-549 Fincham JRS (1962) J Mol Biol 4:257-274 Grant DM, Lambowitz AM, Rambosek JA, Kinsey A (1984) Mol Cell Biol 4:2041-2051 Holliday A, Pugh JE (1975) Science 187:226-232 Kinnaird JA, Keighren MA, Kinsey M, Eaton M, Fincham JRS (1982) Gene 20:387-396 Kinnaird JA, Fincham JRS (1983) Gene 26:253-260 Kinsey JA, Rambosek JA (1984) Mol Cell Biol 4:117-122 Maniatis T, Fritsch EF, Sambrook J (1982) "A Manual of Cold Spring Harbor Laboratory, Cold Sring Harbor Laboratory, NY Selker EU, Cambareri EB, Jensen BC, Haack KA (1987a) Cell 51:741-752 Selker EU, Jensen BC, Richardson GA (1987b) Science238:48-53 Selker EU, Stevens JN (1985) Proc Natl Acad Sci USA 82:81148118 Selker EU, Stevens JN (1987) Mol Cell Biol 7:1032-1038 Stevens JN, Metzenberg RL (1982) Neurospora Newslett 29:27-28 Tilburn J, Scazzocchio C, Taylor GG, Zabicky-Zissman JG, Lockington RA, Davies RW (1983) Gene 26:205-221 Vollmer SJ, Yanofsky C (1986) Proc Natl Acad Sci USA 83:48694873

Communicated by B. S. Cox Received February 20, 1989