Proc. Natl. Acad. Sci. USAVol. 89, pp. 4358-4362, May 1992Medical Sciences

Immunostimulatory human urinary protein(urine/scid mice/macrophages/Listeria monocytogenes/Lewis carinoma)

ELISABETH FONTAN*, HtLtNE SAKLANI-JUSFORGUES, AND ROBERT M. FAUVEUnitt d'Immunophysiologie ceilulaire, Institut Pasteur, 25 rue du Dr. Roux, 75724 Paris Cedex 15, France

Communicated by Merrill W. Chase, January 28, 1992 (received for review September 5, 1991)

ABSTRACT We have previously extracted a protein frominflammatory mouse granuloma that fully protects normalmice against lethal Listeria monocytogenes infection. We nowreport that polyclonal antibodies directed against this proteinreact with a human urinary fraction that also provides fullprotection of normal mice against L. monocytogenes. Murinemonoclonal antibodies completely inhibit the protective activityof the human urinary fraction. We have purified to apparenthomogeneity a human glycoprotein of 43 kDa (HGP.43), pI =3.2 ± 0.2. HGP.43 injected i.v. at 250 jug/kg fully protectedmice against a lethal inoculum of L. monocytogenes. Suchresistance followed injection ofHGP.43, 4 days before and, stillsigniflcantiy, 8 hr after Listeria infection. In scid mice lackingT and B cells, similar resistance to L. monocytogenes wasobserved. Inflammatory murine macrophages became cyto-static against Lewis carcinoma cells after incubation with 1.5nM HGP.43.

An extract of a mouse granuloma protein induced by talc orBio-Gel P-100 was found to fully protect normal mice againsta lethal inoculum of Listeria monocytogenes and to activateperitoneal cells against Lewis carcinoma cells (1-3). A poly-clonal antibody prepared against mouse granuloma proteinwas shown to react with a human protein isolated from theurine of healthy donors. We therefore set out to purify thisprotein. We have obtained an immunostimulatory protein,separated and purified as a 43-kDa protein, free ofinactive alacid glycoprotein (a1-AGP), which fully protects mice againsta lethal inoculum of L. monocytogenes. This protein alsoactivates mouse macrophages against Lewis carcinoma cells.

MATERIALS AND METHODSUrinary Fractions. Twenty-four-hour urine specimens (2

liters) collected from healthy male and female human donorswere stored at -200C. After thawing, the urine was centri-fuged at 6900 X g at 40C and the supernatant was concen-trated by using an Amicon YM10 ultrafilter with a Mr 10,000cut-off, to a final volume of 10 ml. Subsequent procedures arediscussed in Results.

Mice. BALB/c, OF1 (Iffa Credo, St. Germain sur l'Ar-bresle, France), C57BL/6, and CB-17 Icr scid mice (InstitutPasteur) were kept under specific pathogen-free conditions.

Fate of L. monocytogenes in Spleen and Liver. L. monocy-togenes (serotype I, no. 54,149, Institut Pasteur Collection)were kept virulent by passage in mice and grown in tryptose/phosphate broth. Groups of four OF1 mice were injected i.v.with 1 x 101 Listeria 24 hr after the i.v. injection of theisolated fractions. Mouse survival was recorded 4 days afterinfection, or bacterial counts from spleen and liver weredetermined according to Pierce et al. (4).

Cells. Inflammatory macrophages were obtained by themethod of Fauve et al. (5). Briefly, 4 days after dorsal s.c.

injection ofC57BL/6 mice with Bio-Gel P-100 (Bio-Rad), thegel was withdrawn and the inflammatory cells were recov-ered by filtration through a 400-mesh screen. The cells werecentrifuged (250 x g for 10 min), washed twice with ice-coldDulbecco's modified Eagle's medium (DMEM), incubated (1x 106 cells per ml, 0.5 ml per culture tube) for 24 hr, andwashed to remove polymorphonuclear cells. Lewis tumorcells (3) were isolated from murine lung metastases andcultured in plastic Petri dishes in DMEM containing 10%(vol/vol) fetal calf serum at 3TC in an atmosphere of 7.5%C02/92.5% air.

Antibodies. Polyclonal rabbit anti-mouse granuloma pro-tein antibodies were prepared as described by Fontan et al.(1). Polyclonal antibodies against the human urinary proteinwere raised in rabbits by s.c. injection of 10 ,ug in Freund'scomplete adjuvant and twice-monthly booster injections.Immunoglobulins were purified by precipitation in(NH4)2SO4 at 33% saturation and DEAE-gel chromatogra-phy. Antibodies against human al-AGP were purified bymeans ofimmunoaffinity chromatography using pure a1-AGP(Sigma) coupled to CNBr-activated Sepharose. Monoclonalantibodies against the human urinary protein were preparedas follows. Female BALB/c mice were immunized by s.c.injection of 50 jug of a semipurified active human urinaryfraction. Spleen cells were fused with the SP2/0 BALB/cmyeloma cell line. Wells with hybrid growth were screenedfor reactivity with the urinary fraction by using an ELISA.Positive wells were subcloned twice by limiting dilution. Oneof the subclones (K52.5G2) was chosen for the neutralizationtest. This clone produces an IgG1 that was purified by(NH4)2SO4 precipitation and DEAE-gel chromatography.The supernatants and antibody-free controls were tested inmice challenged with a lethal inoculum ofL. monocytogenes.

All the materials, solutions, and tissue culture media werefree of endotoxin.

Statistics. Results are expressed as the mean ± SEM. Thesignificance of the differences between experimental groupswas analyzed by Student's t test. A P value of <0.05 wasconsidered significant.

RESULTSPurification and Characterization of the Urinary Protein.

After Amicon ultrafiltration concentration of the urines, thematerial was chromatographed on a DEAE-Sephacel anion-exchange column and eluted with 0.2 and 1 M NaCl. All theimmunostimulating activity was eluted with 0.2 M NaCI (Fig.1). The active fractions were pooled, dialyzed, and loadedonto a Cibacron blue column to remove contaminating humanalbumin. The unbound material was then subjected to Ultro-gel AcA 44 gel filtration. Two peaks containing the stimu-lating activity were obtained with apparent molecular massesof 92 and 43 kDa. As previous attempts to characterize this

Abbreviations: HGP, human glycoprotein; a1-AGP, a, acid glyco-protein; M-CSF, macrophage colony-stimulating factor.*To whom reprint requests should be addressed.

4358

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

7, 2

020

Proc. NatL. Acad. Sci. USA 89 (1992) 4359

A3

0CNC14

0

1 2 3 4 5

Fractions

B0.8

4'

0.6

0.4

0.2

0.0

C0.5

0

N 0.4

0.3

0.2

0.1

0.0

Ve 68 45 25kDa

60 so 100 120Fractions

Ve 68kDa 45kDa

X

0 10 20 30 40 50 60 70 80 90Fractions

1.0

az

:

kDa92 -

68 -

43 -

0.530 -

-a.-_ a...mK100

50 i

_ 0.0 070

100 I

50 "

- 0a v

25 kDa

I

00 _a

0 U

FIG. 1. Gel filtration. (A) DEAE-Sephacel chromatography.Concentrated normal human urine was loaded on a DEAE-Sephacelcolumn equilibrated in 0.07 M sodium phosphate (pH 6.8). Thecolumn was washed with this buffer and bound material was elutedwith the same buffer containing 0.2 M or 1 M NaCl (dashed line).Fractions (1 ml) were collected and the amount of protein in eachfraction was determined at 280 nm (solid line). OF1 mice wereinjected i.v. with each fraction (20 Ag of one fraction per mouse) 24hr before Listeria infection and the biological activity (hatched bar)was expressed as percentage of infected mice survival, 4 days later.(B) Gel filtration on an Ultrogel AcA 44 column. The active fractionseluted from the Cibacron blue column were applied to an UltrogelAcA 44 column equilibrated with 0.1 M sodium phosphate (pH 7.2),containing 0.1 M NaCl and calibrated with molecular mass standards.Fractions (1 ml) were collected. OF1 mice were injected i.v. witheach fraction (20 ,ug of one fraction per mouse) before infection andthe biological activity (hatched bars) was expressed as percentage ofinfected mice survival. V,, elution volume. (C) Gel filtration on anUltrogel AcA 54 column. After Pronase hydrolysis ofactive fractionseluted from the Ultrogel AcA 44 column and concentration by

FIG. 2. SDS/polyacrylamide gel of HGP.43. Lanes: 1, 20 ,ug ofconcentrated urine; 2, 10 pAg of active fractions from DEAE-Sephacel; 3, 10 &g of fractions from Ultrogel AcA 44; 4, activefractions after Pronase treatment; 5, 10 Ag of purified HGP.43.Proteins were loaded onto SDS/10%/ polyacrylamide gels and stainedwith Coomassie blue. Molecular mass standards used to calibrate thegel are shown on the left.

protein showed that the biological activity was not destroyedby trypsin or Pronase, to facilitate its isolation, we hydro-lyzed the active fractions with Pronase. Pronase was re-moved by ultrafiltration, and the concentrate was loaded ontoan Ultrogel AcA 54 column. Only fractions 59-69 fullyprotected mice against a lethal inoculum of L. monocyto-genes. When the purified material was analyzed by SDS/polyacrylamide gel electrophoresis and Coomassie bluestaining, one single band migrating with an apparent molec-ular mass of 43 kDa was detected under reducing andnonreducing conditions (Fig. 2).

Partial amino acid sequencing of this material revealedthat it also contained a1-AGP at 43 kDa, which had nostimulatory activity in our assay. Final purification wasperformed using immunoaffinity chromatography. The anti-a1-AGP immunosorbent, prepared according to Pharmaciaprocedures, was mixed with the urinary protein and incu-bated for 1 hr under agitation. The antibody beads weretransferred to a small column and washed with phosphate-buffered saline. The unbound fraction was pooled, dialyzed,and lyophilized. The absence ofa1-AGP was confirmed by anELISA.The immunostimulatory activity was removed by passage

through a Con A affinity column, indicating that the purified43-kDa molecule was a glycoprotein [named 43-kDa humanglycoprotein (HGP.43)]. By using anthrone reagent, we es-timated the amount of carbohydrate to be 15% of the nativemolecular weight. Preparative flat-bed isoelectric focusingindicated a pI value of 3.2 ± 0.2. The immunostimulatoryactivity was completely inhibited by the anti-urinary proteinmonoclonal antibody (Fig. 3). It is noteworthy that themonoclonal antibody recognized not only the purifiedHGP.43 but also the 92-kDa protein (Fig. 4).HGP.43 Increases Spleen and Liver Macrophage Bacteri-

cidal Activity. Groups of four OF1 mice were injected i.v.with various amounts ofHGP.43 and then were injected 24 hrlater by 1 x 105 L. monocytogenes. As shown in Fig. 5, thenumber of L. monocytogenes found in the spleen 48 hr afterinfection was significantly decreased by treatment with 0.2Ag of HGP.43. This effect was clearly dose-dependent. Tenmicrograms of macrophage colony-stimulating factor (M-CSF) had no stimulatory effect. The number ofListeria foundin the spleen and the liver was reduced as early as 6 hr afterinfection in HGP.43-treated mice (Fig. 6). As shown in Table

ultrafiltration, the material was run through an Ultrogel AcA 54column equilibrated in 0.1 M sodium phosphate (pH 7.2), containing0.1 M NaCl and calibrated with molecular mass standards. Frac-tions (1 ml) were collected. OF1 mice were injected i.v. with eachfraction (10 1&g of one fraction per mouse) and the biological activity(hatched bars) was expressed as percentage of infected mice sur-vival.

Medical Sciences: Fontan et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

7, 2

020

4360 Medical Sciences: Fontan et al.

c

0

Ei 1000) 75

.6 50

E 20

5

5 6 7L monocytogenes, log10o no. per spleen

FIG. 3. Antibody neutralization of the HGP.43 activity. HGP.43was incubated with various concentrations of monoclonal antibodyfor 1 hr at 370C and overnight at 40C (open bars). The samples wereincubated 2 hr at 370C with immobilized protein A-Ultrogel toremove antigen-antibody complexes. The samples were centrifugedat 1000 x g for 10 min, and OF1 mice were injected with supernatant24 hr before L. monocytogenes infection. Control mice (C; hatchedbars) received monoclonal antibodies (100 ,ug/ml) incubated in thesame conditions. A bacterial count was performed in spleens ofinfected mice 48 hr later.

1, this decrease was not due to a bactericidal activity ofHGP.43 or of serum from treated mice. These results showthat injection of HGP.43 increases the bactericidal activity ofresident macrophages. The magnitude of the effect ofHGP.43 was dependent on the time of treatment relative toinfection (Fig. 7) but occurred from 4 days before infectionuntil 8 hr after. The stimulatory activity of HGP.43 wasabrogated by incubation at 700C for 1 hr, showing that theobserved effects were not due to contamination by lipopoly-saccharide, which is highly heat-resistant (2).Immunostimulatory Activity of HGP.43 in scid Mice. To

determine whether the effect of HGP.43 was mediated bymacrophages activated by sensitized T lymphocytes (6), wetested whether HGP.43 enhanced anti-bacterial resistance inT- and B-cell-deficient mice. CB-17 scid mice were chal-lenged with 1 x 105 L. monocytogenes 24 hr after adminis-tration of 10 ,ug or 5 ,ug of HGP.43. Fig. 8 shows that suchpretreatment decreased bacterial growth in both spleen andliver, indicating that HGP.43 can increase the level of non-specific anti-L. monocytogenes activity not involving matureT cells.

Cytostatic Activity of Inflammatory Macrophages Treatedwith HGP.43 in Vitro. C57BL/6 inflammatory macrophages(Bio-Gel P-100) were incubated for 24 hr with various

1 2

M-CSF 5pg F-H

1 0Mg

HGP 0.1Mug43 0.214g

0.5#9g1MAg2Mg

5Mug10g

Controls

4 5 6 7 8

log10 Listeria monocytogenes

FIG. 5. Number of Listeria, 48 hr after infection, in spleens ofOF1 mice injected with either 0.1-10 ,ug of HGP.43 or 5-10 ,ug ofM-CSF. Control OF1 mice (hatched bar) received only saline.Twenty-four hours later, OF1 mice received 1 x 105 Listeria and 48hr later a bacterial count was made in spleens of these mice. Valuesare expressed as the mean ± SEM (P < 0.05).

amounts of HGP.43. Target cells 3LL were then added tomacrophages at a ratio of 1:10. As shown in Fig. 9, incubationof Bio-Gel P-100-induced macrophages with 0.06 ,ug ofHGP.43 decreased the amount of 1251I-labeled deoxyuridineincorporated into 3LL cells. This effect was clearly dose-dependent. The culture supernatant had no cytostatic effect.

DISCUSSIONWe have found that all the human urines tested, obtainedfrom 12 healthy donors, contain two peaks of activity cor-responding to proteins with apparent molecular masses of 43kDa and 92 kDa that are able to protect mice against a lethalinoculum of L. monocytogenes. These two proteins arerecognized by the monoclonal antibodies. After treatment byPronase, all the immunostimulatory activity was found in the43-kDa protein. Preliminary experiments have shown that the43-kDa protein found in normal urine results from the hy-drolysis of the 92-kDa molecule. Indeed, the 43-kDa proteinwas obtained by Pronase treatment of the purified 92-kDa

10 I Spleen

105[

0

0c0

0

c

E-J

HGP.92

Io3L0

HGP.43

1 2 3 4 5 6Hours after infection

Liver

104 V

1030 1 2 3 4 5 6

Hours after Infection

FIG. 4. Western blot analysis of HGP.92 and HGP.43. HGP.92obtained after Ultrogel AcA 44 column chromatography and purifiedHGP.43 were transferred to nitrocellulose after gel electrophoresisand identified by an anti-HGP.43 monoclonal antibody. Lanes: 1,HGP.92; 2, HGP.43.

FIG. 6. Fate of Listeria in spleens and livers of OF1 mice duringthe first 6 hr of infection. OF1 mice received either saline (0) or 101Lg of HGP.43 (o) 24 hr before infection. Bacteria were counted 1 hr,3 hr, and 6 hr after i.v. injection of 1 x 10 L. monocytogenes. Valuesare expressed as the mean ± SEM (P < 0.05).

Proc. Natl. Acad Sci. USA 89 (1992)

L

-1

I

T

Dow

nloa

ded

by g

uest

on

Aug

ust 2

7, 2

020

Proc. Natl. Acad. Sci. USA 89 (1992) 4361

Table 1. HGP.43 and serum from HGP.43-treated mice lackbactericidal activity against L. monocytogenes

Bactericidal activity afterincubation, no. cells

(x10-4) per ml

Addition 1 hr 8 hr

None 0.64 ± 0.14 102 ± 17HGP.43 0.68 ± 0.06 140 ± 4Serum control 0.73 ± 0.05 210 ± 8Serum HGP.43 0.77 ± 0.17 185 ± 3

Listeria (2 x 103) were incubated in tryptose phosphate brothcontaining either 10 pug of HGP.43 or saline. For incubation withserum, mice were injected with either saline or 10 1Lg of HGP.43 and24 hr later, control and HGP.43 sera were collected and diluted 1:2with tryptose phosphate broth before the addition of2 x 103 Listeria.After incubation at 370C for 1 hr or 8 hr, bacterial counts wereperformed.

protein. The 43-kDa species (HGP.43) is glycosylated andPronase-resistant, and its biological activity is completelyneutralized by a specific monoclonal antibody. Among hu-man urinary proteins of similar molecular weight, a1-AGP,also found in our preparations, had no stimulatory effect,since its removal with its specific antibodies did not diminishthe anti-L. monocytogenes activity. Furthermore, purifieda1-AGP had no stimulatory effect. A 43-kDa epidermalgrowth factor-related protein has been purified by Gregory(7) from the urine of breast cancer patients, but no biologicalactivity was detected in fractions with an apparent molecularmass of 92 kDa, contrary to the anti-L. monocytogenesactivity we describe. In addition, the partial amino acidsequence of HGP.43 differs from the reported sequences ofproteins with epidermal growth factor-like activity (8, 9).Although cytokines, interleukin la (10), interferons ,B (11)and y (12), tumor necrosis factor (13), granulocyte/macrophage colony-stimulating factor (14), have been re-ported to increase the resistance of mice toward somepathogens, it is unlikely that they account for the biologicalactivity ofHGP.43, because their molecular masses are muchlower. Since M-CSF, which was shown to have antitumoreffects (15), is a 90-kDa disulfide-bonded homodimer (16), wetested M-CSF in our Listeria infection model and found it hadno effect. It, therefore, appears that HGP.43 is distinct fromcytokines.

Since HGP.43 is not itself bactericidal but is active in L.monocytogenes-infected scid mice, its activity may be me-diated, at least in part, by spleen and liver macrophages. In

1 0

107

0

0 10

0

0

DoKHGP43 injected

before or after

L.monocytogenes

.+

FIG. 7. Number of Listeria in spleens and livers of OF1 miceinjected with saline (hatched bar) or 10 ,g of HGP.43 (open bars) atvarious times before or after infection. D, day; h, hr. Forty-eighthours after infection, a bacterial count was performed in spleens.Values are expressed as the mean + SEM (P < 0.05).

LIVER

HGP43 1 0

Controls

SPLEEN

HGP43 | ojug5,*9

Co ntro l s //////////

104 loll lol 107 lol loll

Listeria monocytogenes

FIG. 8. Increased resistance of scid mice to L. monocytogenes.CB-17 Icr scid mice were injected with saline (hatched bars) or 5-10Ag of HGP.43 and inoculated with 1 x 105 L. monocytogenes 24 hrlater. Bacteria were counted in spleens and livers 48 hr later. Valuesare expressed as the mean ± SEM (P < 0.05).

mice, macrophages are indeed known to be responsible forincreased resistance to Listeria (17). The protective effect ofHGP.43 was evident as early as 2 hr after a L. monocytogenesinoculation and was also observed when HGP.43 was in-jected 8 hr after inoculation. To our knowledge, this is thefirst report of an immunostimulant activity after L. monocy-togenes infection (18). HGP.43-induced activation of macro-phages was also shown in vitro by the induction of cytostaticactivity against Lewis carcinoma cells. This cytostatic activ-ity did not require the presence of T lymphocytes and doesnot appear to be mediated by tumor necrosis factor, which isnot active against Lewis carcinoma cells (3). HGP.43 mayinduce, in inflammatory macrophages, other molecules thatalone or in synergy could explain the increased resistance toL. monocytogenes and the cytostatic activity against Lewiscarcinoma cells. Indeed, preliminary findings suggest thatHGP.43 induces interleukins 1 and 6 and tumor necrosisfactor production by human monocytes. Thus the immuno-stimulatory properties of this protein suggest its potentialusefulness in human therapeutics.

Macrophages+ HGP43 0.011 5psg

0.03jkg

0.06J9g0.12j9g

0.25pg

.59g1+

Macrophagesalone

L I3LL

0 1 2 3 4 5 6 7 81251-laboled deoxyuridine Incorporation

in 3LL cells

FIG. 9. Cytostatic activity of P-100 macrophages against 3LLcells. C57BL/6 macrophages (2.5 x 105 cells) obtained after injectionofBio-Gel P-100 beads into mice were incubated with various amountsof HGP.43 for 24 hr. 3LL cells and macrophages were then added ata ratio of 1:10 and incubated for 24 hr. Cells were then pulsed-labeledwith 0.04 ,uCi of 125I-labeled deoxyuridine (specific activity, 5 Ci/mg;1 Ci = 37 GBq) for another 17 hr. In control tubes, 3LL cells alone(hatched bars) or with macrophages (open bars) were added. Valuesare expressed as cpm (X10-3) (mean ± SEM; P < 0.05).

Medical Sciences: Fontan et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

7, 2

020

4362 Medical Sciences: Fontan et al.

We thank Dr. J. Vandekerckhove (Belgium) for amino acid se-quencing and Dr. Marc B. Garnick (Genetics Institute) for providingthe M-CSF. We acknowledge M. Dehbi for breeding the mice usedduring these studies and C. Courcy for secretarial help in thepreparation ofthe manuscript. This research was supported by grantsfrom the Direction des Recherches, Etudes et Techniques (no.89-246), the Fondation pour la Recherche Mddicale, and the LigueNationale Francaise contre le Cancer.

1. Fontan, E., Fauve, R. M., Hevin, M. B. & Jusforgues, H.(1983) Proc. Natd. Acad. Sci. USA 80, 6395-6398.

2. Fontan, E. & Fauve, R. M. (1986) Ann. Inst. Pasteur!lmmu-nol. 137D, 93-107.

3. Fontan, E. & Fauve, R. M. (1988) let. J. Cancer 42, 267-272.4. Pierce, C. H., Dubos, R. J. & Schaefer, W. B. (1953) J. Exp.

Med. 97, 189-205.5. Fauve, R. M., Jusforgues, H. & Hevin, M. B. (1983) J. Immu-

nol. Methods 64, 345-351.6. North, R. J. (1973) J. Exp. Med. 138, 342-355.7. Gregory, H. (1975) Nature (London) 257, 325-328.8. Eckert, K., Granetzny, A., Fischer, J., Nexo, E. & Grosse, R.

(1990) Cancer Res. 50, 642-647.

Proc. NatM. Acad. Sci. USA 89 (1992)

9. Mount, C. D., Lukas, T. J. & Orth, D. N. (1987) Arch. Bio-chem. Biophys. 255, 1-7.

10. Czuprynski, C. J. & Brown, J. F. (1987) Infect. Immun. 55,2061-2065.

11. Kiderlen, A. F., Kaufmann, S. H. E. & Lohmann-Mathes,M.-L. (1984) Eur. J. Immunol. 14, 964-967.

12. Nakane, A., Minagawa, T. & Kato, K. (1988) Infect. Immun.56, 2563-2569.

13. Havell, E. A. (1989) J. Immunol. 143, 2894-2899.14. Mayer, P., Schutze, E., Lam, C., Kricek, F. & Liehl, E. (1991)

J. Infect. Dis. 163, 584-590.15. Lu, L., Shen, R.-N., Lin, Z.-H., Aukerman, S. L., Ralph, P.

& Broxmeyer, H. E. (1991) Int. J. Cancer 47, 143-147.16. Cosman, D. J. & Gillis, S. (1990) in Hematopoietic Growth

Factors in Clinical Applications (Dekker, New York), pp.95-96.

17. Mackaness, G. B. (1962) J. Exp. Med. 116, 381-406.18. Fauve, R. M. (1980) in Advances in Nephrology, eds. Ham-

burger, J., Crosnier, J., Grunfeld, J. P. & Maxwell, M. H.(Year Book Med., Chicago), Vol. 9, pp. 167-186.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

7, 2

020