Medical Sciences: Corrections Proc. Natl. Acad. Sci. USA 87 (1990) 10133 Correction. In the article "Epididymis is a principal site of retrovirus expression in the mouse" by Ann Anderson Kiessling, Richard Crowell, and Cecil Fox, which appeared in number 13, July 1989, of Proc. Natl. Acad. Sci. USA (86, 5109-5113), the authors have requested that the following correction be made. The legend in Fig. 2 has an error in lane identification and should read as follows: Lanes: 1, testis.

Correction. In the article "Antibiotic proteins of human polymorphonuclear leukocytes" by Joelle E. Gabay, Randy W. Scott, David Campanelli, Joe Griffith, Craig Wilde, Marian N. Marra, Mary Seeger, and Carl F. Nathan, which appeared in number 14, July 1989, of Proc. Natl. Acad. Sci. USA (86, 5610-5614), the authors request that the following correction be noted. On page 5614, in Table 2, middle section (activity against S. faecalis), the LD50 for peak VII reported as 53 Ag/ml should be 5.3 tug/ml. Downloaded by guest on September 28, 2021 Proc. Nati. Acad. Sci. USA Vol. 86, pp. 5610-5614, July 1989 Medical Sciences Antibiotic proteins of human polymorphonuclear leukocytes (bacteria/lysosomes/) JOELLE E. GABAY*t, RANDY W. SCOTTt, DAVID CAMPANELLI*, JOE GRIFFITHt, CRAIG WILDEt, MARIAN N. MARRAt, MARY SEEGER*, AND CARL F. NATHAN* *Beatnce and Samuel A. Seaver Laboratory, Division of Hematology-Oncology, Department of Medicine, Cornell University Medical College, New York, NY 10021; and tInvitron Corporation, Redwood City, CA 94063 Communicated by Maclyn McCarty, May 1, 1989 (receivedfor review March 6, 1989)

ABSTRACT Nine polypeptide peaks with antibiotic activ- may have been blocked. Finally, we compared quantitatively ity were resolved from human polymorphonuclear leukocyte the individual activities of each rpHPLC peak against mi- azurophil membranes. All but 1 of the 12 constituent crobes of three types (Gram-positive and -negative bacteria polypeptides were identified by N-terminal sequence analysis. and fungi) under a range of assay conditions previously Near quantitative recovery ofprotein and activity permitted an reported to optimize the activity of one or another of the assessment of the contribution of each species to the overall previously characterized species. respiratory-burst-independent antimicrobial capacity of the cell. Three uncharacterized polypeptides were discovered, including two broad-spectrum antibiotics. One of these, a MATERIALS AND METHODS that we have designated human antimicro- PMN azurophil granule membranes were isolated and ex- bial peptide 4, was more potent than previously described tracted as described (11). Concentrated extracts were ana- but represented less than 1% ofthe total protein. The lyzed by rpHPLC using a Vydac C4 column (Rainin Instru- other, named azurocidin, was abundant and comparable to ments) and an elution gradient of 0-100% (vol/vol) acetoni- bactericidal permeability-increasing factor in its contribution trile in 0.1% trifluoroacetic acid. The resulting fractions were to the killing of Escherichia coli. subjected to SDS/PAGE on a 15% polyacrylamide gel (12). Automated Edman degradation was carried out using an Polymorphonuclear leukocytes (PMNs) defend against bac- Applied Biosystems model 477A pulse liquid-phase se- terial and fungal infections by two main mechanisms: pro- quencer. Phenylthiohydantoin amino acid analysis was per- duction of reactive oxygen intermediates through the respi- formed on-line using an Applied Biosystems liquid chromato- ratory burst and delivery of polypeptide antibiotics from graph, model 120A. The N-terminal sequences for peaks III, lysosomal granules into phagocytic vacuoles (1-3). Several VI, VII, and IX were also verified by direct sequencing of antimicrobial proteins have been isolated from human PMNs. Coomassie-stained bands electroblotted onto polyvinylidene Bactericidal permeability-increasing protein (BPI) (4) [prob- difluoride membranes as described (13), except that 0.05% ably identical with cationic antimicrobial protein (5) and SDS was added to the transfer buffer. Antimicrobial activity bactericidal protein (6)] specifically kills Gram-negative bac- against Escherichia coli K12 (strain MC4100), Streptococcus teria. G (7) and three mutually homologous pep- faecalis (ATCC 29212), and Candida albicans (clinical iso- tides termed defensins (8, 9) are active against Gram-positive late from Presbyterian Hospital, New York) was tested as bacteria, Gram-negative bacteria, and fungi. is described (14). Antimicrobial activity was expressed in total active against some pathogenic Gram-positive organisms killing units [(killing units/ml) x volume of the extract]. (10). Killing units are defined as the reciprocal of the dilution of Analysis of the role of polypeptide antibiotics in the granule extract necessary to kill 105 bacteria per ml in 30 min function of PMNs has been held back by three problems. (i) at 37°C or 104 fungi per ml in 60 min at 37°C. Procedures resulting in isolation ofone polypeptide antibiotic often failed to reveal other antibiotics. (ii) The widely diver- RESULTS gent conditions used for the bioassay of antibiotics were optimal for detection of some species but probably obscured Isolation and Identification of Azurophil Membrane Pro- the activity of others. (iii) There have been few if any studies teins. In an earlier study, nitrogen-bomb cavitates of diiso- comparing the known antibiotic species in the same assay. propyl fluorophosphate-treated PMNs were fractionated on Consequently, it has not been possible to assess the relative Percoll density gradients, and antibacterial activity was as- contribution of any one of these proteins to the overall sessed in the fractions enriched in nuclei, cytosol, specific respiratory-burst-independent antimicrobial potential of granules, azurophil granules, or plasma membranes (11). PMNs. Nor can it be excluded that previously undiscovered Almost all the activity of the cavitate was recovered in the antimicrobial polypeptides ofhuman PMNs may contribute a fractions enriched in azurophil granules and could be recov- significant proportion of the cells' activity. ered quantitatively from azurophil membrane material (11). We dealt with these problems by isolating from human In the present study, we subjected concentrated extracts of PMNs the smallest subcellular compartment that contained the azurophil granule membrane to further purification and most of the respiratory-burst-independent antimicrobial ac- analysis. tivity (11). From this compartment-the membrane-associ- ated fraction of the azurophil granule-we used reverse- Abbreviations: BPI, bactericidal permeability-increasing protein; phase HPLC (rpHPLC) to isolate all 12 of the major poly- BSA, bovine serum albumin; ECP, cationic protein; peptides and sequenced the N terminus of all but 1, which HNP, human neutrophil antimicrobial peptide; MBP, major basic protein; PMN, polymorphonuclear leukocyte; rpHPLC, reverse- phase HPLC. The publication costs of this article were defrayed in part by page charge tTo whom reprint requests should be addressed at: Box 57, Cornell payment. This article must therefore be hereby marked "advertisement" University Medical College, 1300 York Avenue, New York, NY in accordance with 18 U.S.C. §1734 solely to indicate this fact. 10021.

5610 Medical Sciences: Gabay et al. Proc. Natl. Acad. Sci. USA 86 (1989) 5611 closely spaced and had a single N-terminal sequence (Table 1.50 1); we assume they represent isoforms. The N-terminal sequences of the polypeptides in six ofthe l rpHPLC peaks matched previously recognized sequences / (Table 1). The major peak was peak I (45% of recovered / / protein); this consisted of a mixture of defensins termed / / human neutrophil antimicrobial peptide (HNP) 1, HNP-2, / / and HNP-3 (15), in a molar ratio of approximately 40:40:20. The peak III polypeptide shared 19 of 20 N-terminal residues '-e 0.75 I with eosinophil cationic protein (ECP) (16). The peak IV .1 polypeptide (18% ofrecovered protein) was identical over its 11 ,l first 16 residues with the N terminus ofcathepsin G (17). The peak V polypeptide shared its first 17 of 20 amino acids with .1 major basic protein (MBP) of (18). In some preparations, a small amount of lysozyme, a soluble constit- uent of the azurophil granule, was also found in peak V (up JU L to 20% of the protein found in this peak). The peak VIII 9 polypeptide shared 18 of 20 residues with elastase (19). The 20 40 60 80 peak IX polypeptide was identified as BPI on the basis of Retention time, min identity of 20 of 20 amino acids (20). Three polypeptides had N-terminal sequences that have FIG. 1. rpHPLC analysis of azurophil granule membranes. Pro- apparently not been reported. The peak II polypeptide ap- tein (1-2 mg) from azurophil fraction isolated from 300 ml of blood peared to be a member of the defensin family. It shared 11 of was concentrated 20-fold in a Centricon-10 ultrafiltration unit, made 32 or 33 residues with HNP-1, -2, and -3, including all six 0.1% in trifluoroacetic acid, and subjected to rpHPLC. cysteines, and has been designated HNP-4 (unpublished data). The 29-kDa polypeptide in peak VI was of particular Nine peaks were separated by rpHPLC (Fig. 1), with a total interest because its abundance (17% of recovered protein) protein recovery of 77% (Table 1). Fig. 2A shows the matched that of cathepsin G and was second only to the polypeptide profile of each peak after SDS/PAGE under defensins counted as a group. The polypeptides of both peak reducing conditions. The polypeptide in peak II, once re- VI and peak VII displayed N-terminal to duced, could not be visualized by silver staining, but staining cathepsin G, elastase, and other members of the serine of gels containing an unreduced sample revealed a single protease family (17, 19). Detailed studies of the 29-kDa species migrating at 4 kDa (Fig. 2B). Two species, migrating polypeptide found in peak VI, which has been named azuro- at 29 and 80 kDa, made up peak VI. These have been cidin, will be reported separately. separated by other procedures; only the 29-kDa polypeptide Activity Profiles of Each Species. In prior work, optimal has antimicrobial activity (unpublished results). The 80-kDa antimicrobial activity of defensins depended on low ionic species, for which no N-terminal sequence could be deter- strength and the presence of nutrients in the assay buffer (9). mined, will not be discussed further. The polypeptides pre- Another critical variable for defensins, as well as for cathep- sent in peaks III, IV, VII, VIII, or IX migrated in more than sin G and BPI, was maintenance ofthe assay buffer at neutral a single band. The bands within each of these peaks were to alkaline pH (4, 7, 9). Therefore, before testing each species Table 1. Protein recovery and N-terminal sequence analysis of polypeptides resolved by rpHPLC: Comparison to known PMN proteins Molecular Sequence Peak or comparable mass, Recovery, species kDa ug of protein 1 5 10 15 20 Ia 4 200 A C Y C R I P A C I A lb C Y C R I P A C I A Ic D C Y C R I P A C I A HNP-1(15) A C Y C R I P A C I A HNP-2 (15) - C Y C R I P A C I A HNP-3 (15) - D C Y C R I P A C I A II* 4 2 V C S C R L V F C R R T E L R V G N C L III 18 10 X P P Q F T R A Q W F A I Q H I S L N P ECP (16) - R P P Q F T R A Q W F A I Q H I S L N P IV 29 76 I I G G R E S R P H S R P Y M A Cathepsin G (17) I I G G R E S R P H S R P Y M A V 14 12 T X R Y L L V R S L Q T F S Q A X F T X MBP (18) - - T C R Y L L V R S L Q T F S Q A W F T C VI (azurocidin)* 29 80 I V G G R K A R P R Q F P F L A S I Q N VII* 29 36 I V G G H E A Q P H S R P Y M A S L E M VIII 29 18 I V G G R R A R P H A X P X M V S L Q L Elastase (19) - I V G G R R A R P H A W P F M V S L Q L IX 54 12 V N P G V V V R I S Q K G L D Y A S Q Q BPI (20) - V N P G V V V R I S Q K G L D Y A S Q Q Protein recoveries are shown for a typical HPLC run (580 ,ug of concentrated azurophil granule membrane extract was fractionated and 77% was recovered). rpHPLC-purified polypeptides were concentrated to 50 Il in a SpeedVac (Savant) and identified by N-terminal sequence analysis. The single-letter amino acid code is used. Numbers in parentheses are refs. *Sequences that apparently have not been reported. 5612 Medical Sciences: Gabay et al. Proc. Natl. Acad. Sci. USA 86 (1989)

A VI VIl Vil IX individually, we tested the effect of these variables on the S 11 III IV V bioactivity ofthe unfractionated azurophil granule membrane extract. In addition, we included one more variable that was 68 important for the recovery of antimicrobial activity in the 45- rpHPLC fractions: the presence of bovine serum albumin 36_ (BSA) in the tubes in which the fractions were collected. 24- Experiments (data not shown) indicated a marked increase in recovery of bioactivity if BSA was added to the tubes prior 20- to, rather than after, further handling of the samples; the optimal concentration of BSA was 0.02% (final concentra- 14- tion). As indicated in Fig. 3, the microbicidal activity of the granule extract against E. coli was higher at pH 5.5 than at neutral or alkaline pH. Within the range tested, ionic strength was immaterial, but the addition of nutrient broth was inhibitory. In contrast, with S. faecalis as a test organism, low ionic strength was critical and pH was no longer a key variable. Again, nutrient broth was inhibitory. B Thus, with the unfractionated extract, each variable tested M IR IIR IN IIN m affected the bioassays differently depending on the test organism. This made it necessary to test each fraction against each organism under each condition. The detailed results are depicted in Fig. 4. For purposes of a summary comparison (Table 2), we selected the set of assay conditions yielding the maximum activity of the unfractionated extract against each test organism. Because ECP and MBP probably represent proteins from contaminating eosinophilic PMNs, rather than from neutrophilic PMNs, the last column ofTable 2 compares the activities of the other peaks to the total activity obtained by summing all peaks except those representing ECP and MBP. For each of the organisms tested, the sum of activities of the isolated polypeptides was 80-100% of the activity of the unfractionated extract. This allowed us to assign relative potencies and activities to each antibiotic polypeptide, as discussed below. FIG. 2. SDS/PAGE analysis of rpHPLC peaks. Peaks collected Activity against E. coli. Two species accounted for almost from rpHPLC were dried by vacuum centrifugation, washed by all the activity against E. coli: BPI and azurocidin, the 29-kDa resuspension in 0.1% acetic acid, and returned to dryness. (A) Starting polypeptide in peak VI. Although the specific activity of BPI material (lane S) and peaks (lanes I-IX) were treated with 2% (wt/vol) was 10 times higher than that of azurocidin, the greater SDS and 5% (vol/vol) 2-mercaptoethanol at 100TC prior to SDS/ us to estimate that PAGE. Molecular mass marker migration is indicated on the left (in abundance of the latter (see Table 1) led kDa). (B) Reduced (lanes R) or non-reduced (lanes N) peak I and peak it contributes 75% as much to anti-E. coli activity as does II. Molecular mass markers (lane M) were the same as in A. Molecular BPI. Cathepsin G, MBP, and elastase were also active but mass markers (lane m) were myoglobin digest fragments of 17,200, contributed only a minor portion of the total activity. The 14,600, 8240, 6380, and 2560 Da. Both gels were silver stained. abundant defensins in peak I had a specific activity 15,000 Activity against: E co/i S. foeco/is C. olbicons a b c d f g a b c d e f 9 h i 200,000

150,0001 y

0 0- 2 100,000 o' c FIG. 3. Effect of assay conditions on =t 50,000 susceptibility of test organisms to azuro- phil granule extract. The activity of the extract against E. coli, S.faecalis, and C. albicans was tested as indicated. +, on 0'rV2 Component present; -, component ab- - sent. Assay buffers were sodium citrate and sodium phosphate, at the noted con- BSA 0.02% + + + + + -+ centration and pH. A supplement to Citrate 50 mM - - some of these assay buffers was trypti- mM 50 50 50 10 10 10 50 50 50 10 10 10 10 io 10 case soy broth (TSB). Results are ex- Phosphate pressed as means ± SEM for three to pH 5.5 5.5 7.0 5.5 7.4 7.4 5.5 5.5 7.0 5.5 5.5 7.4 7.4 5.5* 5.5* eight experiments, each performed with TSB 1% - - material from a separate fractionation. Medical Sciences: Gabay et al. Proc. Natl. Acad. Sci. USA 86 (1989) 5613 Peak: Activity against: E: co/i S foeco/is C albicons a b c d f g a b c d e f g h i 25ou r~

25- vvn 50 -I .03 .04 .01 1.6 ~i .1 .2 .06 - L 100 50 50 100 II 25 6 50 .03 oo0l .1 D1 J2 .075.03 .6 0 _- 50 25 100 III 25 .04 .3 ] 2{Z. 50 -.54 0 FIG. 4. Effect of assay conditions on 50 100l00 susceptibility of test organisms to puri- fied polypeptides. The activity ofthe nine IV 25 - I.p 8 I.2 50 - 6.A .1 94_ .4 j rpHPLC peaks against E. coli, S. faeca- lis, and C. albicans was tested under the ri. same conditions as in Fig. 3. The activity V 2 of each peak against a particular micro- s0 ffi84jU frjj3505 r n r) 50 -.E .8 organism in the various assay conditions 100- is represented as a percentage (shown on { ~424 the ordinate) of the activity of the total - 12 16 16 24 VI 25t2 25 .06.06.06 3 3 3 .4 granule extract measured in 50 mM cit- rate (pH 5.5) for E. coli and in 10 mM 50 - 2.41 phosphate (pH 5.5) for S. faecalis and C. 25 25 albicans. These assay conditions (bars b, VIl .3 2 2.8 3.2 .01 3 .03 .01 .03 .06 100 d, and i) resulted in the highest activity of 0 0 the unfractionated granule extract 50 50o against these microorganisms, as shown Vlill 25 -.4 4.8 8 3.2 2.4 2.4 25 .03 .04 .02 1:LJ in Fig. 3. Bars without an indicated per- 50 centage correspond to activity below the 50 32 28 50_ limit of detection. Results are means + SEM for three experiments, each per- IX 25 - 16 25- 2.4 formed with material from a separate A ~~nL fractionation. times lower than that of BPI under the conditions used for was more potent than the mixture of HNP-1, -2, and -3 comparison, and thus made a negligible contribution to the contained in peak I but represented less than 1% of the total total activity. Defensin HNP-4 found in peak II was 60-fold protein. The possibility has not been excluded that HNP-4 more potent than the mixture of HNP-1, -2, and -3. arose from eosinophils but this seems unlikely as no other Activity against S. faecalis. Low ionic strength was re- defensins have yet been identified in eosinophils. In contrast, quired for activity against this organism. Cathepsin G and azurocidin was both an abundant and potent species and may MBP together accounted for 88% of the total activity against contribute 75% as much to the killing of E. coli as BPI, which S. faecalis. Excluding the contribution of MBP, the largest was previously thought to account for almost all of this proportion of activity after that of cathepsin G was provided activity (21, 22). The localization of azurocidin to the azuro- by azurocidin. The contribution of defensins was almost phil granules of neutrophils has been confirmed by immun- negligible under these conditions. Under test conditions of ofluorescence and immuno electron microscopy with a higher ionic strength, the only species to show some activity monospecific antibody (D.C., P. Detmers, C.F.N., and against S. faecalis was cathepsin G. J.E.G., unpublished results). Activity against C. albicans. Antifungal activity was also A strength of this analysis was the high recovery in protein dependent on low ionic strength (results at high ionic strength (Table 1) and antimicrobial activity (Table 2) of the purified are not shown). Under these conditions, the major species to species. Under most assay conditions, the sum of activities contribute to killing of C. albicans were cathepsin G and of the resolved components closely matched (80-100%) that MBP, with far lower but still appreciable contributions by the of the unfractionated extract from which they were derived. peak I defensins and azurocidin. The specific activity ofMBP This implies that neither synergistic interactions nor mutual exceeded that of the peak I defensins by 70-fold. HNP-4, the inhibition among the component polypeptides need be in- defensin of peak II, was 7-fold more potent than the mixture voked. Nonetheless, either phenomenon could occur, in of those of peak I. particular, under suboptimal conditions for antimicrobial activity. Other antimicrobial polypeptides ofthe human PMN azurophil granule may remain to be discovered. We screened DISCUSSION against only one microbe from each of three classes. Addi- Twelve polypeptides were isolated from the azurophil mem- tional antimicrobial proteins may exhibit activities against brane fraction of human PMNs. Eleven of them were iden- other organisms or, like peak II, exist at levels that make tified by N-terminal sequence analysis and their antimicrobial detection difficult. activities were compared. Four of these polypeptides proved A limitation of this analysis is that it was restricted to to be potent, broad-spectrum antibiotics. Of these activities, proteins associated with the membrane of the azurophil only one-that of cathepsin G-was previously appreciated. granule. Other subcellular compartments were excluded, Another-that of MBP-probably arose from contamination based on their lack of bioactivity against E. coli (11). Thus, of our neutrophil preparation (97%) with eosinophils (3%). significant activities elsewhere in the cell against Gram- Two broad-spectrum antimicrobial polypeptides were dis- positive bacteria or fungi may have been overlooked or covered: a 4-kDa defensin termed HNP-4 and a 29-kDa underrated. For example, lysozyme, a soluble constituent of polypeptide named azurocidin. HNP-4 (contained in peak II) the azurophil granule, has activity against S. faecalis and C. 5614 Medical Sciences: Gabay et al. Proc. Natl. Acad. Sci. USA 86 (1989) Table 2. Summary of antimicrobial activities of each peak under sosomal fluid in PMNs. Divalent cations markedly affect the the conditions optimal for activity of the starting material against activity of PMN antimicrobial proteins; the levels of these the same organism cations in PMN phagolysosomes are also unknown. Activity, Activity, Activity, % of Purification of the antimicrobial proteins of PMNs, gener- LD50, % of % of total less peaks ation of monospecific antibodies against them, and isolation Peak ,ug/ml ext. total III and V of their cDNAs may lead to a better understanding of the regulation of granule proteins during and after myelopoiesis, Against E. coli insight into the molecular basis of clinical states marked by I 225 0.03 0.035 0.04 defective antimicrobial function, and design of antibiotics. II 3.7 0.03 0.03 0.034 III ND ND ND Note. After this manuscript was submitted, we became aware of a IV 0.38 8 10 11.3 prior report of isolation of HNP-4; its antimicrobial activity was not V 0.06 8 10 tested (25). VI 0.16 24 30.4 34 VII 0.72 2 2.5 2.8 We thank J. Snable for technical assistance and P. Detmers, R. VIII 0.15 4.8 6.1 6.8 Nachman, H. Shuman, and S. Wright for critically reading the IX 0.015 32 40.6 45.2 manuscript. This work was supported by Grant BC-586 from the American Cancer Society and Grants Al 23807 and CA 43610 from Against S. faecalis the National Institutes of Health. J.E.G. is a Cornell Scholar in I 42 0.1 0.14 0.26 Biomedical Sciences. II 0.6 0.075 0.08 0.16 III 9 0.01 0.01 1. Klebanoff, S. J. (1988) in Inflammation, eds. Gallin, J. I., IV 0.05 43.8 48 92.8 Goldstein, I. M. & Snyderman, R. (Raven, New York), pp. V 0.012 43.8 48 391 444. VI 1 3 3.4 6.6 2. Elsbach, P. & Weiss, J. (1988) in Inflammation, eds. Gallin, VII 53 0.006 0.007 0.01 J. I., Goldstein, I. M. & Snyderman, R. (Raven, New York), VIII 13 0.03 0.034 0.07 pp. 445-470. IX ND ND ND ND 3. Lehrer, R. I., Ganz, T. & Selsted, M. E. (1988) in Hematology/ albicans Oncology Clinics ofNorth America, ed. Curnutte, J. T. (Saun- Against C. ders, Philadelphia), pp. 159-163. I 2.8 6.4 5.3 6.9 4. Weiss, J., Elsbach, P., Olsson, I. & Odeberg, H. (1978) J. Biol. II 0.4 0.6 0.53 0.7 Chem. 253, 2664-2672. III 0.1 4.8 3.9 5. Shafer, W. M., Martin, L. E. & Spitznagel, J. K. (1984) Infect. IV 0.11 80 65.8 86.3 Immun. 45, 29-35. V 0.04 24 19.7 6. Hovde, C. J. & Gray, B. H. (1986) Infect. Immun. 54, 142-148. VI 1.6 4.8 3.9 5.2 7. Odeberg, H. & Olsson, I. (1975) J. Clin. Invest. 56, 1118-1124. VII 5.6 0.5 0.4 0.5 8. Drazin, R. E. & Lehrer, R. I. (1977) Infect. Immun. 17, 382- VIII 5.6 0.4 0.33 0.43 388. 9. Ganz, T., Selsted, M. E., Szklarek, D., Harwig, S. S. L., IX ND ND ND ND Daher, K., Bainton, D. F. & Lehrer, R. I. (1985) J. Clin. Activity is expressed in three ways: (i) as a percent of that of the Invest. 76, 1427-1435. unfractionated azurophil granule extract (% ofext.); (ii) as a percent 10. Spitznagel, J. K. (1984) in Contemporary Topics in Immuno- of the total activity recovered in all fractions (% of total); and (iii) as biology, ed. Snyderman, R. (Plenum, New York), pp. 283-343. a percent ofthe total activity recovered in all fractions except the two 11. Gabay, J. E., Heiple, J. M., Cohn, Z. A. & Nathan, C. F. activities suspected to arise from eosinophils, peaks III and V (% of (1986) J. Exp. Med. 164, 1407-1421. total less peaks III and V). ND, not detectable. 12. Laemmli, U. K. (1970) Nature (London) 227, 680-685. 13. Matsudaira, P. (1987) J. Biol. Chem. 262, 10035-10038. 14. McGrogan, M., Simonsen, C., Scott, R., Griffith, J., Ellis, N., albicans (23, 24). However, pure human lysozyme (Calbio- Kennedy, J., Campanelli, D., Nathan, C. & Gabay, J. (1988) J. chem) was 100-500 times less active than cathepsin G or Exp. Med. 168, 2295-2307. MBP against S.faecalis and 5-10 times less active than these 15. Selsted, M. E., Harwig, S. S. L., Ganz, T., Schilling, J. W. & two species against C. albicans on a weight basis (14) and did Lehrer, R. I. (1985) J. Clin. Invest. 76, 1436-1439. not act synergistically with pure cathepsin G or MBP (un- 16. Gleich, G. J., Loegering, D. A., Bell, M. P., Checkel, J. L., Ackerman, S. J. & McKean, D. J. (1986) Proc. Natl. Acad. Sci. published data). Quantitatively, the major antibiotic proteins USA 83, 3146-3150. in the soluble fraction of the azurophil granule are defensins. 17. Salvesen, G., Farley, D., Shuman, J., Przybyla, A., Reilly, C. Half of the total defensin content remained associated with & Travis, J. (1987) Biochemistry 26, 2289-2293. the granule membrane (data not shown). Thus, our estimates 18. Wasmoen, T. L., Bell, M. P., Loegering, D. A., Gleich, G. J., of the contribution of defensins to overall activity of the cell Prendergast, F. G. & McKean, D. J. (1988) J. Biol. Chem. 263, were probably not grossly biased by our focus on membrane- 12559-12563. 19. Sinha, S., Watorek, W., Karr, S., Giles, J., Bode, W. & Travis, associated proteins. Indeed, a survey of the entire azurophil J. (1987) Proc. Natl. Acad. Sci. USA 84, 2228-2232. granule for antimicrobial activity resulted in findings similar 20. Ooi, C. E., Weiss, J., Elsbach, P., Frangione, B. & Mannion, to those described here (unpublished data). Finally, results B. (1987) J. Biol. Chem. 262, 14891-14894. were similar when diisopropyl fluorophosphate was omitted 21. Weiss, J., Victor, M., Stendahl, 0. & Elsbach, P. (1982) J. Clin. (data not shown). Invest. 69, 959-970. This study points up the major impact of assay conditions 22. Weiss, J., Kao, L., Victor, M. & Elsbach, P. (1985) J. Clin. on the specific activity of individual antimicrobial polypep- Invest. 76, 206-212. 23. Laible, N. J. & Germaine, G. R. (1985) Infect. Immun. 48, tides. A major question is what set of assay conditions most 720-728. faithfully reproduces the milieu of the phagolysosome. Ex- 24. Lehrer, R. I., Ladra, K. M. & Hake, R. B. (1975) Infect. isting data do not permit an answer; instead, they suggest that Immun. 11, 1226-1234. key factors such as pH are likely to vary with time after 25. Singh, P., Bateman, A., Zhu, Q., Shimasaka, S., Esch, F. & phagocytosis and may also depend on the organism. There is Solomon, S. (1988) Biochem. Biophys. Res. Commun. 155, no published information on the ionic strength of phagoly- 524-529.