Nature © Macmillan Publishers Ltd 1997

letters to nature

NATURE | VOL 387 | 5 JUNE 1997 607

homeodomain. Tests on embryonic samples (see Fig. 6b) used PCR with twoMrx gene-specific primers and a neo-specific primer.

Received 25 February; accepted 18 April 1997.

1. Ton, C. C. T. et al. Positional cloning and characterization of a paired box and homeobox-containinggene from the aniridia region. Cell 67, 1059–1074 (1991).

2. Glaser, T. et al. PAX6 gene dosage effect in a family with congenital cataracts, aniridia, anophthalmiaand central nervous system defects. Nature Genet. 7, 463–471 (1994).

3. Hill, R. E. et al. Mouse Small eye results from mutations in a paired-like homeobox-containing gene.Nature 354, 522–525 (1991).

4. Quiring, R., Walldorf, U., Kloter, U. & Gehring, W. J. Homology of the eyeless gene of Drosophila to theSmall eye in mice and Aniridia in humans. Science 265, 785–789 (1994).

5. Halder, C., Callaerts, P. & Gehring, W. J. Induction of ectopic eyes by targeted expression of the eyelessgene in Drosophila. Science 267, 1788–1792 (1995).

6. Jamrich, M. & Sato, S. Differential gene expression in the anterior neural plate during gastrulation ofXenopus laevis. Devellopment 105, 779–786 (1989).

7. Sive, H. L., Harrori, K. & Weintraub, H. Progressive determination during formation of theanteroposterior axis in Xenopus laevis. Cell 58, 171–180 (1989).

8. Mathers, P. H., Miller, A., Doniach, T., Dirksen, M.-L. & Jamrich, M. Initiation of anterior head-specific gene expession in uncommitted ectoderm of Xenopus laevis by ammonium chloride. Dev.Biol. 171, 641–654 (1995).

9. Bopp, D., Burri, M., Baumgartner, S., Frigerio, G. & Noll, M. Conservation of a large protein

domain in the segmentation gene paired and in functionally related genes in Drosophila. Cell 47,1033–1040 (1986).

10. Noll, M. Evolution and role of Pax genes. Curr. Opin. Genet. Dev. 3, 595–605 (1993).11. Hemmati-Brivanlou, A., de la Torre, J. R., Holt, C. & Harland, R. M. Cephalic expression and

molecular characterization of Xenopus En-2. Development 111, 715–724 (1991).12. Holt, C. E., Bertsch, T. W., Ellis, H. M. & Harris, W. A. Cellular determination in the Xenopus retina is

independent of lineage and birth data. Neuron 1, 15–26 (1988).13. Stiemke, M. M. & Hollyfield, J. G. Cell birthdays in Xenopus laevis retina. Differentiation 58, 189–193

(1995).14. Wetts, R., Serbedzija, G. N. & Fraser, S. E. Cell lineage analysis reveals multipotent precursors in the

ciliary margin of the frog retina. Dev. Biol. 136, 154–163 (1989).15. Wetts, R. & Fraser, S. E. Multipotent precursors can give rise to all major cell types in the frog retina.

Science 239, 1142–1145 (1988).16. Younossi-Hartenstein, A., Tepass, U. & Hartenstein, V. Embryonic origin of the imaginal discs of the

head of Drosophila melanogaster. Wilhelm Roux Arch. Dev. Biol. 203, 60–73 (1993).17. Campos-Ortega, J. A. & Hartenstein, V. The Embryonic Development of Drosophila melanogaster

(Springer, Berlin, 1985).18. Sakaguchi, D. A. Neurosci. Abstr. 16, 479.5 (1990).19. Huang, S. & Moody, S. A. The retinal fate of Xenopus cleavage stage progenitors is dependent upon

blastomere position and competence: Studies of normal and regulated clones. J. Neurosci. 13, 3193–3210 (1993).

20. Hogan, B. L. M. et al. Small eye (Sey): a homozygous lethal mutation on chromosome 2 which affectsthe differentiation of both lens and nasal placodes in the mouse. J. Embryol. Exp. Morphol. 97, 95–110(1986).

21. Grindley, J. C., Davidson, D. R. & Hill, R. E. The role of Pax-6 in eye and nasal development.Development 121, 1433–1442 (1995).

22. Richter, K., Grunz, H. & Dawid, I. B. Gene expression in the embryonic nervous system of Xenopuslaevis. Proc. Natl Acad. Sci. USA 85, 8086–8090 .(1988).

23. Nieuwkoop, P. D. & Faber, J. Normal table of Xenopus laevis (Daudin), 2nd edn (North-Holland,Amsterdam, 1967).

24. Harland, R. M. In situ hybridization: An improved whole-mount method for Xenopus embryos.Methods Cell Biol. 36, 685–695 (1991).

25. Conlon, R. A. & Rossant, J. Exogenous retinoic acid rapidly induces anterior ectopic expression ofmurine Hox2 genes in vivo. Development 116, 357–368 (1992).

Acknowledgements. We thank M.-L. Dirksen, K. T. Ault, N. Papalopulu, M. Whiteley, J. Kassis, F. D.Porter, D. Feltner, D. Sakaguchi, S. Witta, M. Moos, I. Dawid, S. Moody, T. Sargent, G. Spirou, A. Berrebiand O. Sundin for materials and advice.

Correspondence and requests for materials should be addressed to M.J. (e-mail: jamrich@a1.cber.fda.gov).Genbank accession numbers for the homeobox genes are as follows: Xenopus Rx1A, AF001048; XenopusRx2A, AF001049; mouse Rx, AF001906; zebrafish Rx1, AF001907; zebrafish Rx2, AF001908; zebrafish Rx3,AF001909; Drosophila Rx, AF001910; and human Rx, AF001911.

Figure 6 a, Schematic representation of Mrx targeted deletion strategy. E, EcoRI;

B, BglII; N, NotI; HB, homeobox. b, Homozygous pups (right) are born without

visible eye structures, whereas heterozygous pups (left) show no altered

phenotype. PCR analysis of these pups show that the null ( 2 =2 ) genotype

correlateswith the eyeless phenotype. c, Knockout embryos (right) show no eyes

at E13.5. d, Knockout embryos (right) lack visible signs of optic cup formation

(arrow) in E10.5 embryos.e, E9.0 null embryos (right) lack the optic depressions in

the neural tube. f, Section through a heterozygous pup showing normal pattern-

ing of the forebrain. g, Section of a newborn anophthalmic homozygous pup

displaying severe ablations of the forebrain and midbrain.

Theputativechaperonecalmegin is requiredforsperm fertilityMasahito Ikawa, Ikuo Wada*, Katsuya Kominami,Daisuke Watanabe, Kiyotaka Toshimori†,Yoshitake Nishimune & Masaru Okabe

Research Institute for Microbial Diseases, Osaka University, 3-1 Suita, Osaka 565,Japan* Department of Biochemistry, Sapporo Medical University School of Medicine,Sapporo 060, Japan† Department of Anatomy, Medical College of Miyazaki, Miyazaki, 889-16, Japan. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

The proper folding of newly synthesized membrane proteins inthe endoplasmic reticulum (ER) is required for the formation offunctional mature proteins. Calnexin is a ubiquitous ER chaper-one that plays a major role in quality control by retainingincompletely folded or misfolded proteins1–5. In contrast toother known chaperones such as heat-shock proteins, BiP andcalreticulin, calnexin is an integral membrane protein1,6. Calmegin isa testis-specific ER protein that is homologous to calnexin7–9. Herewe show that calmegin binds to nascent polypeptides duringspermatogenesis, and have analysed its physiological functionby targeted disruption of its gene. Homozygous-null male miceare nearly sterile even though spermatogenesis is morphologicallynormal and mating is normal. In vitro, sperm from homozygous-null males do not adhere to the egg extracellular matrix (zonapellucida), and this defect may explain the observed infertility.These results suggest that calmegin functions as a chaperone forone or more sperm surface proteins that mediate the interactionsbetween sperm and egg. The defective zona pellucida-adhesion

Nature © Macmillan Publishers Ltd 1997

letters to nature

608 NATURE | VOL 387 | 5 JUNE 1997

Figure 1 Calmegin and calnexin interact with nascent glycoproteins during

spermatogenesis. a, b, Immunofluorescence staining of testicular sections with

anti-calmegin or anticalnexin antibodies. Calmegin was not detected in the

outermost layer (spermatogonia and Sertoli cells) or the innermost layer

(elongated spermatids) of the semineferous tubule, either while the ubiquitous

expression of calnexin was evident (a) or during spermatogenesis (b). c, Both

calmegin and calnexin transiently interact with nascent proteins in the testicular

cells. Testicular cells were labelled for 30min with 35S-methionine and chased for

the indicated times (0–4 h). Cell extracts were immunoprecipitated with anti-

calmegin or anti-calnexin antibodies. The immune complexes were resolved by

SDS–PAGE. Arrowheads indicate nascent proteins (of Mr 195K, 58K and 41K)

interacted with calmegin. When the cells were lysed in the presence of 1% SDS

(denaturing condition) and were immunoprecipitated, a single polypeptide of

calnexin and calmegin bands were observed, indicating that the antibodies were

monospecific. d, CAS, a glucosidase inhibitor, abolishes the interaction of the

calmegin and calnexin with nascent proteins. CM, calmegin; CN, calnexin.

Figure 2 Targeted disruption of calmegin gene. a, Targeting vector was made by

inserting a neomycin resistance (neo) gene into the second exon resulting in

disruption of the calmegin ATG start codon. For negative selection, a herpes

simplex virus thymidine kinase gene (tk) was introduced into the targeting

construct. b, Hybridization of the 59- and 39-external probes with SacI-digested

genomic DNA yielded 8.4-kb (wild-type) and 9.6-kb (mutant) bands. c, Northern

blot analysis of total testis RNA (20 mg) from wild-type, heterozygote and

homozygote. d, Proteins (5 mg) from testicular cells were analysed by immuno-

blotting for calmegin and calnexin expression in mutant mice.

Nature © Macmillan Publishers Ltd 1997

letters to nature

NATURE | VOL 387 | 5 JUNE 1997 609

phenotype of sperm from calmegin-deficient mice is reminiscentof certain cases of unexplained infertility in human males.

Calmegin was first identified as a male meiotic germ cell-specificantigen8, and was later found to be a Ca2þ-binding protein localizedon the ER membrane7,9. Sequence analysis demonstrated thatcalmegin is highly homologous to calnexin, with 54% identity inamino-acid sequence, including two sets of characteristic sequencerepeats necessary for Ca2þ binding7,9. We determined whethercalmegin functions as a calnexin-like molecular chaperone duringspermatogenesis. In the testis, calmegin was expressed exclusively ingerm cells of pachytene spermatocyte to spermatid stage8 (Fig. 1a),whereas calnexin was expressed throughout the stages (Fig. 1b). Inthe ER of somatic cells, calnexin associates transiently with nascentmembrane and soluble glycoproteins of the secretory pathway andfacilitates folding2–5. When calmegin was immunoprecipitated fromtesticular cells pulse-labelled with 35S-methionine for 30 min, avariety of newly synthesized proteins were found in the immuno-precipitates (Fig. 1c). Upon chase, these ligands, including p195,p58 and p41, were dissociated from calmegin at various rates.Similarly, the transient interaction of calnexin with nascent proteinswas also observed (Fig. 1c). Castanospermine (CAS) treatment ofthe cells inhibited the association in either case (Fig. 1d), suggestingthat the glucosidase activity is needed for the binding process, asshown for calnexin3,4. It should be noted that both chaperonesseemed to interact with distinctive ligands, although they may havesome ligands in common. We therefore suggest that both calmeginand calnexin interact transiently with nascent glycoproteins duringspermatogenesis.

To address the physiological role of calmegin in vivo, we generatedcalmegin-deficient mice by homologous recombination. A

calmegin-targeting construct was designed (Fig. 2a) to interruptthe second exon with a neomycin-resistance gene (neo). Thisinsertion disrupted the ATG translation start site. Both the targetingevent in D3 embryonic stem cells and the germline transmission oftargeted genes were confirmed by Southern blot analysis (Fig. 2b).Intercrosses between heterozygous F1 mice yielded offspring (24litters) that segregated in a mendelian distribution: 57 wild type, 104heterozygous, and 53 homozygous mutant weaning pups. Homo-zygous ( 2 =2 ) mutant mice did not show any overt developmentalabnormalities.

Testes derived from adult F2 male mice were subjected to analysisof the effect of the neo-cassette insertion on calmegin synthesis. Thelevel of full-length (2.4 kb) messenger RNA is reduced to approxi-mately half that of normal in the heterozygous mice (Fig. 2c), but nomRNA is detected in the homozygous mutant mice. When the blotswere rehybridyzed to a probe specific for the calnexin or b-actingene, constant expressions of both were detected in all F2 samples.We performed immunoblot experiments to determine whethercalmegin protein was present in the mutant mice. The anticalmeginmonoclonal antibody (TRA 369) recognized a protein of relativemolecular mass 93,000 (Mr 93K) in þ=þ and þ=2 males (Fig. 2d).Neither normal-sized calmegin protein nor any smaller fragmentwas detected in testes from 2 =2 mice. Expression of calnexin inthe testis was not significantly affected by calmegin disruption (Fig.2c,d).

The effect of calmegin disruption on spermatogenesis was ana-lysed histochemically. Adult testis from a 2 =2 mouse was fixed inBouin’s solution and thin sections were made; testicular sectionswere stained with haematoxylin and eosin or by periodic acid–Schiff (PAS) staining and observed under a microscope. Sperma-togenesis was normal in the calmegin 2 =2 tubules, as indicated bythe production of mature spermatozoa and the appropriate pro-portion of cells in each substage. There was no difference in the sizeof the mutant testis or in the number of germ cells per section of atubule counted separately according to the meiotic stage (50 6 8þ=þ and 54 6 9 2 =2 primary spermatocytes; 144 6 16 þ=þ and144 6 22 2 =2 round spermatids; 133 6 19 þ=þ and 137 6 282 =2 elongated spermatids in stages I–IV tubule10. Three micefrom each strain were used for the counting.)

Homozygous mutant males were nearly sterile, but copulationand vaginal plug formation were normal. Even though 11 2 =2

Figure 3 a, Sperm from calmegin-deficient and wild-type mice were stained with

anti-acrosomal monoclonal and polyclonal antibodies (OBF13, MC101, MN17,

MN9, sp56 and PH-20; see Methods). Staining patterns were the same as wild-

type sperm (data not shown). b, Flow-cytometric analysis of preincubated sperm

from calmegin þ=þ and þ=2 mice stained with anti-acrosomal antibodies

(OBF13, MC101, MN17, MN9). The less intense peaks correspond to the ‘negative’

population that appeared during incubation: ‘bright’ peaks represent the amount

of each antigen contained in the sperm. c, The sperm surface was biotinylated

and solubilized with RIPA or SDS and analysed by SDS–PAGE using avidin-

peroxidase conjugate and chemoluminescence. We detected no difference in

staining patterns between sperm from knockout and wild-type mice. Biotinylation

was in EDTA-PBS (method I) or sodium bicarbonate buffer, pH 8.6 (method II).

Table 1 In vitro fertilization with sperm from calmegin-deficient mice

GenotypeEggs with zona-

penetrated sperm (%)*Eggs with

pronuclei (%)†.............................................................................................................................................................................þ=þ, þ=2 87/99

(88:8 6 7:6)65/75

(86:7 6 8:6)2 =2 2/100

(1:3 6 2:2)0/72(0)

.............................................................................................................................................................................Three independent experiments were performed.* Fertilization events were observed 1h after insemination.† Fertilization events were observed 8h after insemination.

Nature © Macmillan Publishers Ltd 1997

letters to nature

610 NATURE | VOL 387 | 5 JUNE 1997

males were mated with 3 females for more than 3 months (morethan 120 plugs were observed), only one pregnancy (2 pups) wasdiscovered. There were no significant differences in the fecunditybetween þ=þ and þ=2 males; the average litter size in the wild-type female/heterozygote male cross and the wild-type female/wild-type male cross were (8:2 6 1:5, n ¼ 18) and (8:4 6 1:1, n ¼ 20),respectively. In contrast, all F2 females (þ=þ, þ=2 and 2 =2 ) werefertile and showed normal fecundity (data not shown), consistentwith our findings that calmegin is not expressed in the ovary7,8.

We collected sperm from the reproductive tracts of females thathad been mated with 2 =2 males. The recovered sperm weresimilar in number and viability to sperm from wild-type males. Themotility of sperm from both þ=þ and 2 =2 mice was maintainedat over 80% during the 120 min of incubation in TYH medium.Moreover, motile sperm from 2 =2 mice were present in theoviduct. Epididymal sperm from 2 =2 mice showed normalviability, morphology and motility under microscopic observation.When sperm from calmegin 2 =2 and þ=þ mice were stainedusing various polyclonal and monoclonal anti-sperm antibodies, nodifferences in staining pattern or strength measured by flow-cytometry were found between them (Fig. 3a,b). The biotinylationand subsequent SDS–PAGE analysis of solubilized sperm mem-brane using avidin–peroxidase also showed no significant dif-ference in sperm membrane from þ=þ and 2 =2 mice (Fig. 3c).Despite their normal appearance in the in vivo fertilization medium,sperm from 2 =2 males were generally unable to penetrate the eggextracellular matrix (Table 1). Removal of cumulus cells showedthat these sperm failed to adhere to the egg despite frequentcollisions with the zona pellucida (Fig. 4). These results suggestthat sperm from 2 =2 males may encounter eggs in vivo, but fail tofertilize because they cannot bind to zona pellucida.

Various proteins have been proposed as candidate zona pellucidaadhesion molecules on sperm, including sp56 (refs 11, 12), PH-20(ref. 13), zonadhesin (refs 14, 15), Gal-Tase (refs 16, 17), p95 (refs18, 19) and sp17 (refs 20, 21). Activity of one or more of thesemolecules is presumably important in adhesion of sperm to thezona pellucida. Because calmegin transiently interacts with nascentproteins (Fig. 1c) and its gene disruption seems to impair specifi-cally the sperm binding to the egg (Table 1, Fig. 4), we suggest thatcalmegin may be expressed during spermatogenesis to ensure theproper maturation of certain sperm surface protein(s) required forthe egg binding. The identification of protein(s) causing infertilityof calmegin knockout mice will provide important insight into the

mechanism of the interaction between sperm and egg. Indeed, it hasbeen reported that about one-third of men with unexplainedinfertility (semen parameters were within the normal ranges)suffered from no or low binding of sperm to the zona pellucida22,23,although the causes of this condition have remained unknown. Thecalmegin mutant mice may therefore also provide a model forstudying unexplained male infertility. M. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Methods

Immunofluorescence staining and flow-cytometry. Frozen sections of adultmouse testis were immunostained with antibodies (TRA369 for calmegin8 andalpha-CN IP for calnexin3) as described previously. Sperm were collected fromepididymis of calmegin-disrupted and wild-type mice. After fixation with 4%paraformaldehyde, sperm were air-dried on a slide glass and immunostained.The monoclonal antibodies used were OBF13 (ref. 24), MC101 (ref. 25), MN7(ref. 26), MN9 (ref. 27) and 7C5 (ref. 12) (anti-sp56; QED Bioscience).P. Primakoff provided the polyclonal rabbit anti-PH-20 antibody. For thesecond antibodies, FITC-labelled anti-mouse and anti-rabbit antibodies(DAKO-immunoglobulins, Denmark) were used. For flow-cytometric analysis,sperm were preincubated in TYH medium28 for 3 h and fixed with 4%paraformaldehyde before staining.Biosynthetic labelling, immunoprecipitation and biotinylation. Cellsisolated from testis were preincubated for 30 min at 33 8C with or without1 mM CAS in Dulbecco’s minimum essential medium deficient in methionine.The cells (,106 cells) were subsequently labelled for 30 min with 100 mCi of[35S]methionine with or without 1 mM CAS and chased. The cells werecollected by a centrifugation at 500 g for 5 min and lysed in 150 ml of 1%Triton X-100/100 mM KCl/10 mM Tris-Cl, pH 7.5, leupeptin (10 mg ml 2 1,pepstatin (10 mg ml 2 1). The lysates were precleared by incubating with 10 ml of20% Zysorb (Zymed) for 20 min on ice followed by centrifugation for 5 min at10,000 g. Immunoprecipitates were recovered by 1,000 g centrifugation for3 min after incubating the lysates with anti-calmegin or anti-calnexin anti-bodies for 1 h on ice followed by incubation with 10 ml Zysorb for 20 min. Aftera wash in 0.6 M KCl, 0.05% Triton X-100, 10 mM Tris-HCl, pH 7.5, theimmunoprecipitates were resolved on SDS–PAGE and labelled bands werevisualized by BAS2000 phosphorimager (Fujix) equipped with a pictrography.For biotinylation, live sperm were isolated from epididymis. The biotinylationof sperm surface proteins were performed using ECL protein biotinylationmodule (Amersham).Production of calmegin-deficient mice. A clone containing the 59 end of thecalmegin gene was isolated from a 129 genomic library. A 7.5-kb BamHI–XbaIfragment was subcloned into pBluescript SKIIþ. The SalI and XhoI sites wereintroduced into the second exon using a Transformer site-directed mutagenesis

Figure 4 Sperm from þ=þ mice successfully

adhered to the eggs (a), but those from 2 =2

mice failed to attach despite frequent collisions

with the zona pellucida (b). Original magnifica-

tion, 3 400.

Nature © Macmillan Publishers Ltd 1997

letters to nature

NATURE | VOL 387 | 5 JUNE 1997 611

kit (Clontech) with mutagenic primer; 59-AGAGTCGACATGCGTCTC-GAGGGTGTTGGA-39. A XhoI–SalI fragment of the neomycin-resistance(neo) gene from pMC1neo PolyA (Stratagene) was inserted into theSalI–XhoI sites in the same transcriptional orientation as the calmegin. AKpnI–NotI fragment was introduced into the pPNT29 vector including thethymidine kinase (tk) gene to produce the targeting construct. The plasmid waslinearized by NotI digestion before electroporation into D3 ES cells. Of 96G418-gancyclovir-resistant clones, 20 were found to have undergone homo-logous recombination by Southern blot analysis. Four targeted cell lines wereinjected into C57BL/6J blastocysts, resulting in the birth of male chimaericmice. One chimaera produced heterozygous F1 offspring when bred to eitherC57BL/6J or 129 females.In vivo fertilization. Females were superovulated by injection of 5 IU ofpregnant mare serum gonadotropin followed 48 h later by 5 IU of humanchorionic gonadotropin (HCG). Ovulated egg masses were collected from theoviducts 16 h after HCG injection. Eggs were placed in 200 ml TYH medium28

drop covered with paraffin oil. Male mice three months old were subjected to invitro fertilization test. Sperm collected as a clot from mechanically dissectedcaudae epididymidis were introduced into a 200-ml drop of TYH medium.After 1 h of incubation, about 5 ml of sperm suspension was added to the dropcontaining eggs at final concentration of 1–2 3 105 sperm ml 2 1. Viability,morphology and motility were observed at various times of incubation. After1 and 8 h of incubation, eggs were fixed in 4% paraformaldehyde andfertilization events were observed under a microscope. To assess the zonapellucida binding ability, cumulus cells were dispersed by 3–5 min of incuba-tion with hyaluronidase (1 mg ml 2 1, Sigma type VI-s) and soybean trypsininhibitor (10 mg ml 2 1, Sigma T9003) before insemination. After 20 min ofincubation, eggs were fixed and observed.

Received 1 November 1996; accepted 3 April 1997

1. Bergeron, J. J., Brenner, M. B., Thomas, D. Y. & Williams, D. B. Calnexin: a membrane-boundchaperone of the endoplasmic reticulum. Trends Biochem. Sci. 19, 124–128 (1994).

2. Vassilakos, A., Cohen Doyle, M. F., Peterson, P. A., Jackson, M. R. & Williams, D. B. The molecularchaperone calnexin facilities folding and assembly of class I histocompatibility molecules. EMBO J. 15,1495–1506 (1996).

3. Wada, I., Ou, W. J., Liu, M. C. & Scheele, G. Chaperone function of calnexin for the foldingintermediate of gp80, the major secretory protein in MDCK cells. Regulation by redox state and ATP.J. Biol. Chem. 269, 7464–7472 (1994).

4. Herbert, D. N., Foellmer, B. & Helenius, A. Glucose trimming and reglucosylation determineglycoprotein association with calnexin in the endoplasmic reticulum. Cell 81, 425–433 (1995).

5. Ou, W. J., Cameron, P. H., Thomas, D. Y. & Bergeron, J. J. Association of folding intermediates ofglycoproteins with calnexin during protein maturation. Nature 364, 771–776 (1993).

6. Wada, I. et al. SSR alpha and associated calnexin are major calcium binding proteins of theendoplasmic reticulum membrane. J. Biol. Chem. 266, 19599–19610 (1991).

7. Watanabe, D. et al. Molecular cloning of a novel Ca(2þ)-binding protein (calmegin) specificallyexpressed during male meiotic germ cell development. J. Biol. Chem. 269, 7744–7749 (1994).

8. Watanabe, D., Sawada, K., Koshimizu, U., Kagawa, T. & Nishimune, Y. Characterization of malemeiotic germ cell-specific antigen (Meg 1) by monoclonal antibody TRA 369 in mice. Mol. Reprod.Dev. 33, 307–312 (1992).

9. Ohsako, S., Hayashi, Y. & Bunick, D. Molecular cloning and sequencing of calnexin-t. An abundantmale germ cell-specific calcium-binding protein of the endoplasmic reticulum. J. Biol. Chem. 269,14140–14148 (1994).

10. Russell, L. D., Ettlin, R. A., Hikim, A. P. S. & Clegg, E. D. Histopathological Evaluation of the Testis(Cache River, Clearwater, 1990).

11. Cheng, A. et al. Sperm–egg recognition in the mouse: characterization of sp56, a sperm protein havingspecific affinity for ZP3. J. Cell. Biol. 125, 867–878 (1994).

12. Bookbinder, L. H., Cheng, A. & Bleil, J. D. Tissue- and species-specific expression of sp56, a mousesperm fertilization protein. Science 269, 86–89 (1995).

13. Hunnicutt, G. R. et al. Structural relationship of sperm soluble hyaluronidase to the sperm membraneprotein PH-20. Biol. Reprod. 54, 1343–1349 (1996).

14. Hardy, D. M. & Garbers, D. L. Species-specific binding of sperm proteins to the extracellular matrix(zona pellucida) of the egg. J. Biol. Chem. 269, 19000–19004 (1994).

15. Hardy, D. M. & Garbers, D. L. A sperm membrane protein that binds in a species-specific manner tothe egg extracellular matrix is homologous to von Willebrand factor. J. Biol. Chem. 270, 26025–26028(1995).

16. Lopez, L. C. et al. Receptor function of mouse sperm surface galactosyltransferase during fertilization.J. Cell. Biol. 101, 1501–1510 (1985).

17. Gong, X., Subois, D. H., Miller, D. J. & Shur, B. D. Activation of a G protein complex by aggregation ofbeta-1,4-galactosyltransferase on the surface of sperm. Science 269, 1718–1721 (1995).

18. Saling, P. M. How the egg regulates sperm function during gamete interaction: facts and fantasies. Biol.Reprod. 44, 246–251 (1991).

19. Burks, D. J., Carballada, R., Moore, H. D. & Saling, P. M. Interaction of a tyrosine kinase from humansperm with the zona pellucida at fertilization Science 269, 83–86 (1995).

20. Richardson, R. T., Yamasaki, N. & O’Rand, M. G. Sequence of a rabbit sperm zona pellucida bindingprotein and localization during the acrosome reaction. Dev. Biol. 165, 688–701 (1994).

21. Yamasaki, N., Richardson, R. T. & O’Rand, M. G. Expression of the rabbit sperm protein Sp17 in COScells and interaction of recombinant Sp17 with the rabbit zona pellucida. Mol. Reprod. Dev. 40, 48–55(1995).

22. Liu, D. Y. & Baker, H. W. A new test for the assessment of sperm-zona pellucida penetration: relationshipwith results of other sperm tests and fertilization in vitro. Hum. Reprod. 9, 489–496 (1994).

23. Mackenna, A., Barratt, C. L., Kessopoulou, E. & Cooke, I. The contribution of a hidden male factor tounexplained infertility. Fertil. Steril. 59, 405–411 (1993).

24. Okabe, M. et al. Effect of a monoclonal anti-mouse sperm antibody (OFB13) of interaction of mousesperm to zona-free mouse and hamster eggs. J. Reprod. Immunol. 13, 211–219 (1988).

25. Toshimori, K., Tanii, I. & Araki, S. Intra-acrosomal 155,000 dalton protein increases the antigenicityduring mouse sperm maturation in the epididymis: a study using a monoclonal antibody MC101.Mol. Reprod. Dev. 42, 72–79 (1995).

26. Tanii, I., Araki, S. & Toshimori, K. Intra-acrosomal organization of a 90-kilodalton antigen duringspermiogenesis in the rat. Cell Tissue Res. 277, 61–67 (1994).

27. Toshimori, K., Tanii, I., Araki, S. & Oura, C. Characterization of the antigen recognized by amonoclonal antibody MN9: unique transport pathway to the equatorial segment of sperm headduring spermiogenesis. Cell Tissue Res. 270, 459–468 (1992).

28. Toyoda, Y., Yokoyama, M. & Hoshi, T. Studies on the fertilization of mouse eggs in vitro. I. In vitrofertilization of eggs by fresh epididymal sperm. Jpn. J. Anim. Reprod. 16, 147–151 (1971).

29. Tybulewicz, V. L., Crawford, C. E., Jackson, P. K., Bronson, R. T. & Mulligan, R. C. Neonatal lethalityand lymphopenia in mice with a homozygous disruption of the c-abl proto-oncogene. Cell 65, 1153–1163 (1991).

Acknowledgements. We thank P. Primakoff for the anti-PH-20 antibody and B. Storey, G. Gerton andD. Hardy for comments.

Correspondence and requests for materials should be addressed to M.O. (e-mail okabe@biken.osaka-u.ac.jp).

Neurotactin, amembrane-anchoredchemokineupregulated inbraininflammationYang Pan, Clare Lloyd, Hong Zhou, Sylvia Dolich,Jim Deeds, Jose-Angel Gonzalo, Jim Vath,Mike Gosselin, Jingya Ma, Barry Dussault,Elizabeth Woolf, Geoff Alperin, Janice Culpepper,Jose Carlos Gutierrez-Ramos & David Gearing

Millennium Pharmaceuticals Inc., 640 Memorial Drive, Cambridge,Massachusetts 02139, USA. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chemokines are small secreted proteins that stimulate the direc-tional migration of leukocytes and mediate inflammation1–4.During screening of a murine choroid plexus complementaryDNA library, we identified a new chemokine, designated neuro-tactin. Unlike other chemokines, neurotactin has a uniquecysteine pattern, Cys-X-X-X-Cys, and is predicted to be a type 1membrane protein. Full-length recombinant neurotactin is loca-lized on the surface of transfected 293 cells. Recombinant neuro-tactin containing the chemokine domain is chemotactic forneutrophils both in vitro and in vivo. Neurotactin messengerRNA is predominantly expressed in normal murine brain and itsprotein expression in activated brain microglia is upregulated inmice with experimental autoimmune encephalomyelitis, as well asin mice treated with lipopolysaccharide. Distinct from all otherchemokine genes, the neurotactin gene is localized to humanchromosome 16q. Consequently we propose that neurotactinrepresents a new d-chemokine family and that it may play a rolein brain inflammation processes.

Three classes of chemokines have been categorized to date: a(CXC), b (CC) and g (C). The definition for the subclass ofchemokines is based on the distance between the first two of fourcharacteristic disulphide-forming cysteine residues. The a-chemo-kines, including interleukin-8 (IL-8), neutrophil-activating protein2 (NAP-2), GRO-a-g, epithelial-cell-derived neutrophil activatingprotein (ENA-78) and granulocyte chemotactic protein 2 (GCP-2),are mainly chemotactic for neutrophils. By contrast, the b- or CCchemokines are chemotactic for monocytes and, in some instances,for eosinophils or basophils, but not for neutrophils. b-Chemokinesinclude monocyte chemotactic protein (MCP) 1-5, the macrophageinflammatory proteins MIP-1a and MIP-1b, eotaxin and RANTES.Lymphotactin is the only g-chemokine that has only one pair ofcysteines and is thought to be chemotactic for lymphocytes5–7. Herewe report the discovery of a fourth class of chemokine withsignificant structural and functional differences to other chemokines.