................................................................. Innate antimicrobial peptide protects the skin from invasive bacterial infection Victor Nizet*, Takaaki Ohtake²³, Xavier Lauth²³, Janet Trowbridge²³, Jennifer Rudisill²³, Robert A. Dorschner²³, Vasumati Pestonjamasp²³, Joseph Piraino§, Kenneth Huttner§ & Richard L. Gallo*²³ * Department of Pediatrics; and ² Division of Dermatology, University of California, San Diego, California 92161, USA ³ Veterans Affairs San Diego Healthcare System, San Diego, California 92161, USA § Division of Neonatology, Massachusetts General Hospital, Boston, Massachusetts 02114, USA .............................................................................................................................................. In mammals, several gene families encode peptides with antibac- terial activity, such as the b-defensins and cathelicidins 1±3 . These peptides are expressed on epithelial surfaces and in neutrophils, and have been proposed to provide a ®rst line of defence against infection by acting as `natural antibiotics' 4,5 . The protective effect of antimicrobial peptides is brought into question by observa- tions that several of these peptides are easily inactivated 6±8 and have diverse cellular effects that are distinct from antimicrobial activity demonstrated in vitro 9±13 . To investigate the function of a speci®c antimicrobial peptide in a mouse model of cutaneous infection, we applied a combined mammalian and bacterial genetic approach to the cathelicidin antimicrobial gene family 14 . The mature human (LL-37) 15 and mouse (CRAMP) 16 peptides are encoded by similar genes (CAMP and Cnlp, respectively), and have similar a-helical structures, spectra of antimicrobial activity and tissue distribution. Here we show that cathelicidins are an important native component of innate host defence in mice and provide protection against necrotic skin infection caused by Group A Streptococcus (GAS). To assess directly the function of cathelicidins in vivo, we generated mice that are null for Cnlp by targeted recombination. A targeting vector was constructed in which exons 3 and 4, encoding the entire mature domain of CRAMP , were replaced with PGK-neo ¯anked 59 by a genomic Xba/R1 fragment and 39 by a 5.5-kilobase (kb) R1 fragment of Cnip (Fig. 1a). This construct was introduced into 129/SVJ embryonic stem cells by electroporation and subjected to G418 selection. Recombinant clones were screened by polymer- ase chain reaction (PCR) using a forward primer 59 to the deletion construct and reverse primer within the neomycin resistance gene. Embryonic stem cell clones were injected into C57BL/6 blastocysts and transferred to foster mothers. Chimaeric offspring were crossed with C57BL/6 females, whose heterozygous progeny were identi®ed by PCR (Fig. 1b), and these were immediately backcrossed into 129/SVJ. Northern blot analysis con®rmed the absence of CRAMP in Cnlp-null bone marrow (Fig. 1c). Cnlp-null mice had normal fetal development, were fertile, survived into adulthood, and demonstrated no obvious phenotype when housed under aseptic barrier-controlled conditions. The cathelicidins CRAMP and LL-37 (refs 15, 16) are greatly increased in the skin after wounding, owing to their release from neutrophil granules and increased synthesis by keratinocytes 17 . We chose to use GAS for our mouse infection model because the injury accompanying this pathogen leads to a large increase in the local accumulation of CRAMP 17 . Moreover, GAS are highly sensitive to cathelicidin antimicrobial action, even under culture conditions that inactivate peptide killing of other bacteria frequently described as sensitive (for example, Escherichia coli) 17 . Subcutaneous injection of GAS in mice induces a necrotic lesion that is histopathologically similar to that seen with invasive human infections. After identical injections of cathelicidin-sensitive GAS in wild-type, heterozygous and homozygous-null mice, CRAMP-de®cient mice were observed to develop much larger areas of infection (Fig. 2a, b). Lesion areas increased more rapidly, reached larger maximal size, and persisted longer in CRAMP-de®cient mice than in normal littermates while heterozygotes tended to have lesions of intermediate size (Fig. 2c). Cultures of equal amounts of tissue from lesions biopsied at day 7 after injection demonstrated persistent infection with b-haemolytic GAS in CRAMP-de®cient mice but not in normal mice (Fig. 2d). No difference in GAS lesion size was seen when wild-type parental strains C57BL/6 and 129/SVJ were compared. A complementary approach to demonstrating the importance of cathelicidins in host defence is to examine the effects in vivo of altering bacterial sensitivity to CRAMP. If the antimicrobial action of cathelicidin is essential to control a GAS skin infection, as suggested by the experiments with Cnlp-null mice, then CRAMP- resistant GAS should be more pathogenic than CRAMP-sensitive GAS in normal mice. A transposon mutant library was generated from wild-type GAS strain NZ131 by random integration of Tn917 into the bacterial chromosome. In contrast to the parent strain, bacterial growth was observed in a pooled library of transposon mutants exposed to increasing concentrations of the cathelicidin antimicrobial peptide. Southern blot analysis of several isolated colonies demonstrated clonality. The sequence ¯anking the single chromosomal integration of Tn917 in the cathelicidin-resistant mutant (NZ131-CR)was identi®ed and compared to the recently completed GAS genome database 18 . The Tn917 insertion mapped to an open reading frame (GenBank AAK34584) encoding a predicted product of relative molecular mass 28,200 (M r 28.2K) with the signature helix-turn-helix motif of the GntR family of bacterial transcription regulation proteins 19 (Fig. 3a). To demonstrate that Tn917 disruption at this locus was reproducibly associated with an inducible cathelicidin-resistance phenotype, targeted plasmid inte- grational mutagenesis of the gntR-related open reading frame was letters to nature 454 NATURE | VOL 414 | 22 NOVEMBER 2001 | www.nature.com Exon 1 Exon 2 Exon 3 Exon 4 X R R R X R PGK-neo (5.5 kb) X PGK-neo X +/+ +/– –/– +/+ –/– 28S Cnlp a b c Figure 1 Disruption of the Cnlp gene encoding CRAMP in mice. a, Gene targeting strategy: structure and partial restriction map of Cnlp. Top line, ®lled boxes represent exons 1±4. X, Xho I; R, Eco RI. Second line, Targeting construct replaces exon 3±4 R fragment with PGK-neo and retains 59 X±R fragment and 5.5-kb 39 R fragment. Dotted lines, gene fragment replaced by PGK-neo; crosses, ¯anking regions in construct to target homologous recombination; bottom line, inactivated allele; arrowheads are PCR primers. b, PCR genotype results of tail DNA ampli®ed with primers shown in a. c, Northern blot results of total RNA extracted from bone marrow. Cnlp probe is directed to exon 4 alone. 28S RNA is shown as a loading control. © 2001 Macmillan Magazines Ltd