Proc. Natl. Acad. Sci. USA Vol. 88, pp. 10327-10331, November 1991 Dominant male sterility in mice caused by insertion of a transgene (spermatogenesis/impenetrance) JEANNE MAGRAM*t AND J. MICHAEL BISHOP*4 *The G. W. Hooper Foundation and *Department of Microbiology and Immunology, University of California, San Francisco, CA 94143-0552 Contributed by J. Michael Bishop, August 20, 1991

ABSTRACT While examining a series oftransgenic mouse genetic locus affected by Lvs will illuminate a previously lines carrying the HCK protooncogene, we encountered one line undescribed and important function in spermatogenesis. in which males henmzygous for the transgene were sterile. The sterile males mated normally but failed to impregnate females. MATERIALS AND METHODS Light and electron microscopy revealed that spermatogenesis proceeds normally until nuclear condensation, which occurs Production of Transgenic Mice. A DNA fragment contain- but gives rise to a variety of abnormally shaped nuclei. ing an immunoglobulin enhancer, simian virus 40 (SV40) Expression of the transgene was not detectable. Thus, the , a hematopoietic cell kinase (HCK) cDNA (9, 10), insertion itself probably caused the abnormal by and a SV40 poly(A) addition site and splicing signals (referred disrupting a (or ) important in spermatogenesis. The to as plsh and shown in Fig. 1A) was generated by digestion mutation is genetically dominant, causing an abnormal phe- ofits parent by EcoRI and isolation on an agarose gel notype even though the sterile mice carry an ostensibly normal followed by electroelution (11) and CsCl gradient purification counterpart of the disrupted locus. The mutant phenotype is (12). The fragment was microinjected into the pronuclei of completely penetrant only in some genetic backgrounds, sug- fertilized (C57BL6/J x SJL/J)F2 embryos, which were used gesting a modifying influence from a second locus. Junctions to generate transgenic mice by standard protocols as de- between the inserted transgene and adjoining cellular DNA scribed (12). were cloned, allowing us to confirm the heterozygous nature of Identification of Transgenic Mice. High molecular weight the genetic disruption and to detect an associated deletion. We DNA from a small piece of mouse tail was isolated as have designated the mutation Lvs (lacking vigorous sperm) and described (12). DNA was digested with HindIII and then presume that it may define a previously undescribed locus analyzed by Southern blot analysis (13) using the SV40 early important in spermatogenesis. region as a probe. Hybridization was in 0.5 M Na2HPO4, pH 7.2/7% SDS/1 mM EDTA at 650C overnight, followed by two Transgenic mice offer a valuable means with which to eval- washes of20 min each at 650C in 0.5 M Na2HPO4, pH 7.2/1% uate the control and phenotypic effects ofgene expression (1, SDS/1 mM EDTA. 2). In addition, insertion of a transgene into the Genomic . Genomic DNA from a transgenic mouse occasionally disrupts a vital gene, providing a fortuitous was digested to completion with EcoRI and electrophoresed mutation that can reveal the physiological function of the on a 0.9o agarose gel. DNA ranging in size from 4.5 to 7 affected gene (3-5). Typically, an insertion mutation causes kilobases (kb) was electroeluted (11) and used to construct a loss offunction, is genetically recessive, and must be bred to partial genomic library in the phage vector AgtlO. The library homozygosity before its effects become evident. The virtue was screened using the SV40 early region as a probe. in such a mutation is that the presence of the transgene Subcloning and other standard molecular biological tech- provides a molecular "tag" that can facilitate cloning of the niques were performed as described (11). mutant gene and its normal counterpart. Microscopic Examination. For analysis by light micros- We have produced a series of 10 transgenic mouse lines copy, testes were dissected and then fixed in 4% formalin, with the protooncogene HCK. As the mice were bred, it dehydrated in a gradient ofethanol, cleared with toluene, and became evident that many males of one line were sterile and embedded in paraffin, followed by preparation of thin sec- that the sterility arose from a defect in spermatogenesis. tions and staining with hematoxylin and eosin. For analysis Since the sterility segregated with the hemizygous transgene by electron microscopy, mice were perfused through the and occurred in the absence of detectable expression of the heart with phosphate-buffered saline containing heparin (10 transgene, we concluded that the abnormal phenotype was units/ml) followed by perfusion with 1.5% glutaraldehyde in due to mutagenesis by insertion of the transgene. We desig- 0.1 M sodium cacodylate buffer. Testes were removed and nated the mutation Lvs for "lacking vigorous sperm." placed in additional fixative. The tissue was then further fixed Spermatogenesis has been well described morphologically, in 1% osmium tetroxide in acetate vernanal buffer for 1 hr at but little is known of the underlying molecular events (6-8). 4°C followed by 1 hr in 1% tannic acid at room temperature Therefore, we decided to pursue the nature of Lvs and the and then 1 hr at 37°C in Kellenburger buffer. The tissue was gene (or genes) that it presumably defines. Here we describe then dehydrated in a gradient of ethanol and embedded in the phenotype and genetic behavior of Lvs and the isolation Epox; ultrathin sections were then prepared. of DNA from the Lvs locus adjacent to the transgene. The Screening for Fertility. Males were tested for fertility by mutation appears to affect spermatogenesis at the time of mating with FVB/N females (mating was identified by the nuclear condensation and displays two unexpected proper- presence of a vaginal plug in the female after copulation) and ties: genetic dominance and a modifying effect of genetic then dissecting the females 7-10 days later to determine backgrounds. We anticipate that characterization of the Abbreviation: SV40, simian virus 40. The publication costs of this article were defrayed in part by page charge tTo whom reprint requests should be addressed at: G. W. Hooper payment. This article must therefore be hereby marked "advertisement" Foundation, Box 0552, University ofCalifornia, San Francisco, CA in accordance with 18 U.S.C. §1734 solely to indicate this fact. 94143-0552. 10327 Downloaded by guest on September 23, 2021 10328 Genetics: Magram and Bishop Proc. Natl. Acad. Sci. USA 88 (1991) whether the female was pregnant and, if so, the number of Table 1. Effect of genetic background on penetrance of sterility pups present. BALB/ RESULTS C57BL/6J cByJ SJL/J FVB/N Identification of Sterile Males Among Transgenic Mice. Ten F1 60% (6/10) 57% (4/7) 75% (3/4)* 100o (3/3) founder mice were generated by of fertilized F2 40% (2/5) 0%1 (0/6) 100lo (4/4) 10%to (4/4) eggs with the plsh construct (Fig. 1A). The founder female of F3 0% (0/3) 0%o (0/5) 100%o (4/4) 100%o (4/4) transgenic line pIsh 3 was mated to a (C57BL/6J x SJL/J)F1 Percentage of sterile transgenic male progeny derived from back- male (Fig. 1B). A transgenic female from the first cross was crossing two different transgenic females with males ofthe indicated subsequently mated to a different (C57BL/6J x BALB/ inbred strain. Data were pooled for preparation of the table. Num- bers in parentheses indicate sterile males of total number tested. F cByJ)Fl male due to the inability of the regular supplier to indicates the backcross generation number. provide the starting (C57BL/6J x SJL/J)F1 mice. Females *The one male that was not sterile was of very limited fertility; only from the resulting generation were also mated to (C57BL/6J one female of seven was pregnant and with only five pups. x BALB/cByJ)Fl males. Male progeny were tested for fertility. Because normal plugs were obtained with the ex- BALB/cByJ genetic backgrounds, transgenic males on pected frequency, it appeared that males exhibited appropri- SJL/J or FVB/N genetic background were sterile (Table 1). ate sexual behavior and copulated normally. Transgenic The differences in genetic background were apparent after a males fell into three different phenotypic classes. First, the single backcross and were completely established after three majority (2/3) were completely sterile: no pregnancy oc- backcrosses. These data suggest that the impenetrance ofthe curred in a minimum of five matings. The second class of Lvs mutation was due to the influence ofgenetic background. males had very limited fertility: no more than one female Testicular Hstology of Sterile Lvs Males. The testes of became pregnant in a minimum ofsix matings and that female nontransgenic and mutant adult males were fixed and pre- had no more than five pups (generally, one to three). The pared for light microscopy. Although overall organization third class of males had apparently normal fertility. In and numbers of germ cells within the tubules appeared contrast, all nontransgenic males displayed normal fertility, normal in sterile mice, mature spermatids appeared to be demonstrating cosegregation of the sterile phenotype with more darkly stained and misshapen (Fig. 2). Spermatids did the presence of the transgene. not appear to have the normal elongated shape but rather Effect of Genetic Background on Penetrance of the Pheno- were short and wide. The morphologies of both Sertoli and type. One possible explanation for the appearance of various Leydig cells were indistinguishable between sterile trans- phenotypic classes was differences in genetic background. genic testes and nontransgenic testes. To test this hypothesis, transgenic females were backcrossed Electron Microscopy of Spermatids from Sterile Lvs Males. to males of four different inbred strains and resultant female Further characterization by electron microscopy revealed progeny were used to continue backcrossing onto the same the absence of normally shaped mature spermatids in the inbred strain. Male progeny were tested for fertility. In testes of sterile males. Although there appeared to be normal contrast to transgenic males on both the C57BL/6J and numbers of spermatids and development appeared normal A

Eco RI Eco RI

FIG. 1. The transgene and mu- IgE SVp hck cDNA SVpA/sp tant pedigree. (A) pIsh consists of an immunoglobulin enhancer B (IgE), SV40 promoter (SVp), cDNA for human HCK (hck FO cDNA), and a SV40 poly(A) addi- B6SJL tion site and splicing signals (SVpA/sp; see ref. 14). An acti- vating mutation was introduced into the HCK cDNA, which re- Fbb bb ii sulted in a substitution of phenyl- alanine for tyrosine at position 501 and enabled the encoded to transform NIH 3T3 cells (15). B6BaIb c (B) Pedigree for line 3 of pIsh transgenic mice. The founder fe- male was mated to a (C57BL/6J x SJL/J)F1 male, whereas F1 and F2 F2 6666 female progeny were mated to Babc0/6 09 (C57BL/6J x BALB/cByJ)F1 males due to the unavailability of the starting strain. Shaded boxes B6Balb c B6alb.: c represent transgenic animals, which fall into three classes of fertility as defined in the text: completely sterile (solid boxes), 3;4 very limited fertility (stippled box- es), and normal fertility (hatched boxes). All nontransgenic animals ---- ec display normal fertility. nt, Not tested for the presence of the F3 transgene; *, only one pup present 2./ 2 0 ..,7 3:. 4 0/5 0/5 3/3 4:1`6 0/6 3 L. 3:'5 0 '6 r) 10 in the observed pregnancy. Downloaded by guest on September 23, 2021 Genetics: Magram and Bishop Proc. Natl. Acad. Sci. USA 88 (1991) 10329

throughout most of spermatogenesis, no mature spermatids generated by insertion ofthe transgene and serve as i marker had normal morphology. Instead, spermatids from sterile for the disrupted locus. A transgenic male and female dis- transgenic males had abnormally shaped nuclei, although the played the same pattern ofEcoRI-generated bands indicating acrosome and tail remain normal (Fig. 3). the presence of both the normal and disrupted alleles (Fig. Cloning of Junction Fragments Containing DNA Flanking 4C). Therefore, insertion is not into the X chromosome since the Transgene Insertion Site. In an effort to identify the that would predict the absence of the normal allele in sterile genetic defect in Lvs mice, junction fragments between males. In addition, ifthere was not a deletion at the target site mouse genomic DNA and transgene DNA were cloned. upon integration, then cellular DNA adjacent to the two ends These fragments were determined to be 6.0- and 6.5-kb of the transgene would be linked in the normal locus- (Fig. EcoRI fragments by Southern blot analysis of transgenic 4A). Since both junction probes do not recognize the sime mouse DNA using the SV40 portion of the transgene as a EcoRI fragment associated with the normal allele, there must probe (data not shown). EcoRI-digested transgenic DNA was be a deletion or rearrangement of DNA associated with the size fractionated and used to construct a library in phage insertion event. Initial efforts to detect an RNA transcript in AgtlO. The library was screened using the same SV40 probe testes by using thejunction probes described here have been and three clones were identified as positive. Two inserts were unsuccessful (data not shown). subcloned into the plasmid Bluescript KS+ (the third positive clone did not digest with EcoPJ and was therefore discarded) DISCUSSION and shown to have the correct insert sizes. The regions Lvs as an Insertional Mutation. The mutation designated indicated in Fig. 4A as Xba and Hinfl indicate portions ofthe Lvs was encountered during the initial breeding of a single genomic DNA associated with these junction fragments that line ofmice bearing a HCKprotooncogene as a transgene. On were used as probes on Southern blots to recognize EcoRI further analysis, it became apparent that the male sterility polymorphisms (Fig. 4 B and C). These polymorphisms were characteristic ofLvs segregated with the transgene,-although A B Vv

4'

444% *. r.I @ * * 5, As a. '' .; a! *fir* rS w * Eo''i?..f. '44,,

* *-,uU * .> e w its * * * X *xs *: As *

.@ :: s It: #' s * .2 4 v

> 'fig # > *~~~~~~~~~~~iev ; ..,v % *A7i-S!-;* !? * '8 "I'' w q '' o:- e § ° PSS * + <

,o* *_asbe ,,.r ,,4,s V I 4,- .. . ;. Vm #. 0, OF t^ aSlRfj 4 V .

, "Pk- -.,- 4-.~~ ~ ~ ~~~~~~4P q. fo n. s-v. .I FIG. 2. Light microscopy of testes from normal and sterile males. Low-power micrograph of testes from a normal adult male (A) and a sterile transgenic male (B) showing representative semi, niferous tubules. Overall testicu- lar organization and approximate numbers of germ cells appear identical. Higher-power micro- graph of a seminiferous tubule from a normal adult male (C) and a sterile transgenic male (D). Al- though germ cell numbers in the sterile mouse are normal, mature spermatids appear misshapen and w71 more heavily stained. Sections w~~~~~~~.Il have been stained with hematox- :'ek ylin and eosin. (A and B, x200; C and D, x400.) Downloaded by guest on September 23, 2021 10330 Genetics: Magram and Bishop Proc. Natl. Acad. Sci. USA 88 (1991) A

C D

FIG. 3. Electron microscopy of spermatids from normal and sterile males. (A) Representative spermatids from a normal adult male. (B-D) Examples of spermatids from a sterile transgenic male. A variety of spermatid shapes were observed, but there was no consistent representative shape. Instead, all spermatids appeared to have misshapen nuclei, although the acrosomes and tails appeared normal. (x 18,000.) penetrance of the mutation varied as a function of genetic that is effectively null. The correct answer will probably be background (see below). Thus, we attribute the occurrence of known only when the mutation responsible for Lvs has been Lvs to the transgene itself rather than to a coincidental characterized molecularly. mutation at another locus. In principle, the mutant phenotype Variable Penetrance of Lvs. In the initial breedings that could be due to either ectopic expression of HCK or disrup- revealed Lvs, the phenotype proved to be only partially tion of an intrinsic gene by insertion of the transgene. Since penetrant. Some males in the lineage were sterile, some we could detect no expression of the transgene, we con- displayed only very limited fertility, and some were fully cluded that Lvs represents an insertional mutation. fertile. One possible explanation for the variation was that it Genetic Dominance ofLvs. Insertional mutations caused by was due to the influence ofgenetic background. Backcrossing transgenes typically result in a loss of function and, thus, onto different inbred strains confirmed this explanation and become apparent only when bred to homozygosity (3, 5). Lvs became fully penetrant on particular genetic back- Why then is Lvs genetically dominant? A variety of expla- grounds within one to three generations (see Table 1). Thus, nations are possible. First, the transgene might be located on expression of the Lvs phenotype is apparently subject to the X chromosome and create a complete deficiency for the epistatic modification and the rapidity with which full pene- disrupted gene in males. This is demonstrably not the case for trance can be established suggests that a single gene may be Lvs (see below). Second, Lvs may indeed represent a het- responsible. Since we do not yet understand the nature ofthe erozygous loss offunction, leading to a haploinsufficiency of mutation responsible for Lvs (see above), it is impossible to a gene product. Haploinsufficiency has not been a commonly anticipate the mechanism of modification. Methylation of encountered genetic lesion in mammals, but it is well de- DNA and genomic imprinting are obvious possibilities, scribed in invertebrates and microbes. Third, insertion ofthe should the impact of the Lvs mutation rely upon expression transgene might truncate or otherwise alter the product of a of the mutant locus (see above). gene, resulting in a protein with an anomalous function. This Cloning the Disrupted Locus ofLvs. The transgene provided could be a gain of function mutation or an alteration of the a molecular probe for the cloning of adjoining DNA. Ensuing gene product such that it results in transdominant inactiva- analysis of the disrupted locus revealed a rearrangement or tion ofthe normal counterpart, a "dominant negative" effect deletion, the nature of which remains undetermined. Rear- (16). Fourth, the presence of the transgene might activate a rangements and deletions at the site oftransgene insertion are gene that is not normally expressed in testicular tissue, a common occurrence (5). Since a deletion could remove all although the transcriptional controls used in the transgene or part of the affected gene, the cloned DNA adjacent to the that engendered Lvs are not expected to be active in testes HCK transgene may not represent the disrupted gene itself. (immunoglobulin enhancer and SV40 promoter; see refs. 17 Efforts to detect a mRNA with the junction probes have so and 18), and the transgene itself is not detectably expressed far failed. in testes or other tissues. Fifth, insertion of the transgene Analysis of Southern blots revealed that both the normal might inactivate one allele at a locus that does not express the and disrupted versions of the Lvs locus are present in sterile other allele due to genomic imprinting, resulting in a mutation males. Thus, the locus is not situated on the X chromosome. Downloaded by guest on September 23, 2021 Genetics: Magram and Bishop Proc. Natl. Acad. Sci. USA 88 (1991) 10331 A sponsible for the phenotype is not intrinsic to spermatids but resides instead in surrounding somatic tissue, such as the RI-6.5 kb -R Fl Fl l 6.0 kb FR Leydig or Sertoli cells. Hinf probe-(, Xbrobe Little is known about the molecular basis of genetically Hinfl probe Xba probe determined male sterility (7, 19). There is one previous example of male sterility produced by insertion of the trans- gene (20). In contrast to Lvs, however, the lesion was genetically recessive and affected an earlier stage in germ cell C development. Further study of Lvs may illuminate a previ- B tg _ + ously undescribed and important function in spermatogene- sis. tg _ We thank Nancy Quintrell for the human HCK cDNA, Rudi 6 5kb- Grosschedl for the immunoglobulin enhancer, Cheryl Pedula and Paul Goldsmith for histology, Ivy Jacques and Dan Friend for electron microscopy, and Kiran Chada and Richard Lang for review 80kb- a-.* of the manuscript. We would also like to thank Bob Paulson for help with naming the mutation. The work reported here was supported by 3.5kb- 4| A. 60kb- , a grant from the National Institutes ofHealth (CA 44338), funds from the G. W. Hooper Foundation, and fellowships to J.M. from the Anna Fuller Fund and the American Cancer Society. 1. Palmiter, R. D. & Brinster, R. L. (1986) Annu. Rev. Genet. 20, 465-499. 2. Wagner, E. F. & Stewart, C. L. (1986) in Experimental Ap- proaches to Mammalian Embryonic Development, eds. Rossant, J. & Pederson, R. (Cambridge Univ. Press, Cam- bridge, U.K.), pp. 151-193. 3. Gridley, T., Soriano, P. & Jaenisch, R. (1987) Trends Genet. 3, FIG. 4. Identification of junction fragments containing DNA 162-166. flanking the transgenic insertion site. (A) Map oftransgenic insertion 4. Reith, A. D. & Bernstein, A. (1991) Genes Dev. 5, 1115-1123. site containing -12 copies ofthe transgene (open boxes). Solid boxes 5. Costantini, F., Radice, G., Lee, J. L., Chada, K. K., Perry, W. represent adjacent mouse cellular DNA. RI, EcoRI restriction site. & Son, H. J. (1989) Prog. Nucleic Acid Res. Mol. Biol. 36, (B and C) Restriction fragments designated as Hinfi probe and Xba 159-169. probe represent the regions of mouse DNA used as probes on 6. Hecht, N. B. (1986) in Experimental Approaches to Mamma- Southern blots to detect EcoRI polymorphisms in transgenic mouse lian Embryonic Development, eds. Rossant, J. & Pederson, DNA. Also note the same pattern ofrestriction fragments in male and R. A. (Cambridge Univ. Press, Cambridge, U.K.), pp. 151-193. female transgenic DNA, indicating that the transgene (tg) is not on 7. Handel, M.-A. (1987) Results Probl. Cell Differ. 15, 1-62. the X chromosome. 8. Willison, K. & Ashworth, A. (1987) Trends Genet. 3, 351-355. 9. Quintrell, N., Lebo, R., Varmus, H., Bishop, J. M., Pettenati, In work to be reported elsewhere, we have mapped the locus M. J., Le Beau, M. M., Diaz, M. 0. & Rowley, J. D. (1987) to the proximal region of mouse chromosome 9 (J.M., N. Mol. Cell. Biol. 7, 2267-2275. 10. Ziegler, S. F., Marth, J. D., Lewis, D. B. & Perlmutter, R. M. Jenkins, N. Copeland, and J.M.B., unpublished data). (1987) Mol. Cell. Biol. 7, 2276-2285. Lvs and Spermatogenesis. Spermatogenesis occurs in an 11. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989) Molecular ordered and synchronous manner within the seminiferous Cloning:A Laboratory Manual (Cold Spring Harbor Lab., Cold tubules of the testes (6-8). The sequence of events appears Spring Harbor, NY). normal in Lvs until a relatively late stage marked by nuclear 12. Hogan, B., Costantini, F. & Lacy, E. (1986) Manipulating the condensation, which takes place but results in a variety of Mouse Embryo: A Laboratory Manual (Cold Spring Harbor atypical products. Wehypothesize that the lesion responsible Lab., Cold Spring Harbor, NY). arrests the maturation of at the of 13. Southern, E. M. (1975) J. Mol. Biol. 98, 503-517. for Lvs sperm point nuclear 14. Subramani, S., Mulligan, R. C. & Berg, P. (1981) Mol. Cell. condensation, causing the sperm to be unable to develop Biol. 1, 854-864. further. Thus, the sterility that defines Lvs apparently results 15. Ziegler, S. F., Levin, S. D. & Perlmutter, R. M. (1989) Mol. from a failure of sterile males to produce mature sperm. The Cell. Biol. 9, 2724-2727. ability to fertilize in vitro cannot be directly tested because 16. Herskowitz, I. (1987) Nature (London) 329, 219-222. spermatozoa cannot be recovered from the epididymis ofLvs 17. Gillies, S. D., Morrison, S. L., Oi, V. T. & Tonegawa, S. males. (1983) Cell 33, 717-728. The defect associated with Lvs appears to affect only the 18. Magram, J., Niederreither, K. & Costantini, F. (1989) Mol. Cell. Biol. 9, 4581-4584. germ cells. No abnormalities of other tissues or cell types 19. Chubb, C. (1989) J. Androl. 10, 77-88. were found in sterile males; the mice appeared healthy at all 20. MacGregor, G. R., Russell, L. D., Van Beek, M. E. A. B., stages of life and survived to the expected age. The Lvs Hanten, G. R., Kovac, M. J., Kozak, C. A., Meistrich, M. L. phenotype is apparent in developing spermatids. It remains & Overbeek, P. A. (1990) Proc. Natd. Acad. Sci. USA 87, possible, however, that the physiological abnormality re- 5016-5020. Downloaded by guest on September 23, 2021