Proc. Nati. Acad. Sci. USA Vol. 87, pp.

9188-9192, December 1990 Medical Sciences

Recombinant human DNase I reduces the viscosity of cystic fibrosis sputum

(deoxyribonuclease/DNA/infection)

STEVEN SHAK*, DANIEL J. CAPON, RENATE HELLMISS, SCOT A. MARSTERS, AND CARRIE L. BAKER
Departments of Immunobiology and Molecular Biology, Genentech, 460 Point San Bruno Boulevard, South San Francisco, CA 94080

Communicated by David Botstein, August 27, 1990

Respiratory distress and progressive lung deABSTRACT struction in cystic fibrosis can be attributed to bacterial persistence and the accumulation of viscous purulent secretions in the airways. More than 30 yr ago it was suggested that the large amounts of DNA in purulent secretions contribute to its viscosity and that bovine pancreatic DNase I could reduce the viscosity. To evaluate the potential clinical utility of recombinant human DNase I (rhDNase) in the treatment of cystic fibrosis, we have cloned, sequenced, and expressed rhDNase. Catalytic amounts of rhDNase greatly reduce the viscosity of purulent cystic fibrosis sputum, transforming it within minutes from a nonflowing viscous gel to a flowing liquid. The reduction in viscosity is associated with a decrease in size of DNA in the sputum. Inhalation of a rhDNase aerosol may be a simple direct approach that will help individuals with cystic fibrosis and other patients with pneumonia or bronchitis to clear their airways of purulent secretions. The clinical consequences of bacterial infection in the lung are particularly common and serious in individuals with cystic fibrosis. Respiratory distress and progressive lung destruction can be attributed to the accumulation and persistence of viscous purulent secretions in the airways of the lung (1, 2). It is not now known why the airways of these patients remain persistently infected with bacteria such as Pseudomonas aeruginosa despite antibiotic therapy. The cloning of the gene responsible for cystic fibrosis (3-5) promises to lead to insights and, hopefully, therapies. It has been suggested that factors such as impaired clearance of secretions due to their increased viscosity (possibly due to an ion-transport defect), or reduced effectiveness of aminoglycoside antibiotics from reversible binding to polyvalent anions in the secretions, may predispose to bacterial persistence (6-8). Interestingly, a plausible molecular mechanism for these secondary phenomena was proposed >30 yr ago and largely forgotten. Experiments in the 1950s and 1960s revealed that DNA is present in very large amounts (3-14 mg/ml) in purulent, but not in uninfected, lung secretions (9-11). DNA, an extremely viscous polyanion, may contribute both to the increased viscosity of lung secretions and to the reduced effectiveness of aminoglycoside antibiotics (12-14). In fact, lung secretions incubated in vitro with partially purified bovine pancreatic DNase I (contaminated with proteases) show a large reduction in viscosity (15, 16). Based on these observations, bovine pancreatic DNase I (Dornavac or Dornase) was approved in the United States for human use in 1958. Numerous uncontrolled clinical studies in patients with pneumonia and one study in cystic fibrosis suggested that bovine pancreatic DNase I was reasonably safe and effective in reducing the viscosity oflung secretions (17-20). However,
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. 1734 solely to indicate this fact.
9188

severe adverse respiratory reactions did occasionally occur, perhaps as a consequence of allergic reactions to a foreign protein or "irritation" due to contaminating proteases (up to 2% trypsin and chymotrypsin was reported present in the final product) (21, 22). The role of this heavy protease contamination of bovine DNase I in reduction of the viscosity of purulent secretions remained uncertain. Bovine DNase I has been extensively characterized in classic biochemical studies. The amino acid sequence of the bovine protein was determined in its entirety by Liao et al. in 1973 (23), and its crystal structure has been solved (24, 25). Less is known about the human protein. Human DNase I has previously been partially purified from human pancreas, duodenal juice, serum, and urine, and the N-terminal amino acid (leucine) was determined (26-29). To evaluate the potential clinical utility of recombinant human DNase I (rhDNase) in the treatment of cystic fibrosis, we have cloned, sequenced, and expressed rhDNase.t Catalytic amounts of rhDNase greatly reduce the viscosity of purulent cystic fibrosis sputum, transforming it within minutes from a nonflowing viscous gel to a flowing liquid.

pared from freshly obtained and liquid N2-frozen human pancreas (30). By using oligo (dT) primers and EcoRI-Sal I-Xho I-Sst II adapters, a cDNA library of 0.9 x 106 independent isolates of >600 base pairs (bp) was obtained. Two long probes (probe 1: 5'-GTG-CTG-GAC-ACC-TAC-

MATERIALS AND METHODS cDNA Cloning. A human pancreatic cDNA library was constructed in AgtlO by using polyadenylylated mRNA pre-

CAG-TAT-GAT-GAT-GGC-TGT-GAG-TCC-TGT-GGC-

AAT-GAC-3' corresponding to amino acids 91-107 of the bovine protein; probe 2: 5'-TAT-GAC-GTC-TAC-CTG-

GAC-GTG-CAG-CAG-AAG-TGG-CAT-CTG-AAT-GATGTG-ATG-CTG-ATG-GGC-GAC-TTC-AAC-GC-3' corre-

sponding to amino acids 148-170 of the bovine protein) were synthesized. The probes were end-labeled with T4 polynucleotide kinase and adenosine [32P]triphosphate and were used separately to screen the human pancreatic cDNA library under low-stringency hybridization conditions: 20%o (vol/ vol) formamide/5x SSC (lx SSC is 0.15 M NaCl/0.015 M sodium citrate)/50 mM sodium phosphate, pH 6.8/0.1% sodium pyrophosphate/5x Denhardt's solution (lx Denhardt's solution is 0.02% polyvinylpyrrolidone/0.02% Ficoll/ 0.02% bovine serum albumin)/sonicated salmon sperm DNA at 50 ,ug/ml/0.1% SDS/10% (wt/vol) dextran sulfate at 420C. Low-stringency washes were done in lx SSC/0.1% SDS at 420C. Clones that hybridized with both probes were subcloned into M13 vectors and sequenced by the dideoxynucleotide chain-termination method.
Abbreviation: rhDNase, recombinant human DNase I. *To whom reprint requests should be addressed. tThe sequence reported in this paper has been deposited in the GenBank data base (accession no. M55983).

Medical Sciences: Shak

et

al.

Proc. Natl. Acad. Sci. USA 87 (1990)

9189

Expression and Characterization of rhDNase. Human embryonic kidney 293 cells were transfected with an expression plasmid pRK.DNase containing a human cytomegalovirus promoter (31) and the signal and coding sequence of clone hDNase 18-1. Cells were maintained in serum-containing medium for 36 hr and placed in serum-free medium; then supernatants were taken at various times. Fluid-phase deoxyribonuclease activity in supernatants was detected by the generation of acid-soluble cpm after incubation of supernatants with radiolabeled DNA (32). Radiolabeled doublestranded DNA was prepared by incubating an hDNase 18-1 single-stranded M13 template with DNA polymerase I (Klenow fragment) and radiolabeled nucleotides. Zymograms of in situ deoxyribonuclease activity in SDS/polyacrylamide gels were performed by preparing SDS/polyacrylamide gels (10%o) containing salmon sperm DNA at 10 ,ug/ml and 2 mM EDTA (33). After electrophoresis, gels were washed in calcium-free buffer to remove SDS and then incubated at 370C with buffer containing calcium, magnesium, and ethidium bromide. When the gel is illuminated by UV light, deoxyribonuclease activity is indicated by the development of a dark band against a fluorescent background of DNA-bound ethidium bromide. Viscosity of Cystic Fibrosis Sputum. Purulent cystic fibrosis sputum obtained from patients with acute or persistent airway infection was divided into 200- to 400-pg aliquots with a razor blade (1 ml = 1.33 g) and then placed in 1.5-ml polypropylene test tubes. After preincubation at 37C for 5 min, the test material was added, and after mixing for 1 sec, the mixture was incubated at 37C for up to 30 min. Viscosity was measured qualitatively by the pourability assay (34). Pourability was assessed by inverting the tubes and observing the movement of sputum after a tap of the side of the tube. Viscosity was measured quantitatively at 25C using a Brookfield cone-plate viscometer (35, 36). A consistent timed protocol of increased and then decreased rpm of the cone was routinely used. As has been reported (37), freezing and storing sputum at -70C did not influence measurement of viscosity. Agarose Gel Electrophoresis of Sputum DNA. Sputum DNA was solubilized by incubation with 0.5% SDS and proteinase K at 100 ,ug/ml in the presence of 5 mM EDTA, extracted with phenol/chloroform, and precipitated with ethanol. After resolubilization in 10 mM Tris/1 mM EDTA, pH 8, sputum DNA was subjected to electrophoresis in 0.5% agarose gels.

RESULTS AND DISCUSSION

Cloning, Expression, and Characterization of rhDNase. To clone human DNase I, we screened a human pancreatic cDNA library using two long oligonucleotide probes based on the amino acid sequence of bovine pancreatic DNase I. A full-length clone (hDNase 18-1) of 1039 bp was isolated, which, when sequenced (Fig. 1A), was found to contain a long open reading frame encoding a polypeptide of 260 amino acids with extensive homology to bovine DNase I (Fig. 1B). Two possible in-frame initiation codons (ATG) are present upstream of the nucleotides coding for the N-terminal leucine of human DNase I. The nucleotides between the first ATG and the N-terminal leucine encode a sequence with a hydrophobic core of 12 amino acids, which is a plausible secretion signal sequence. A stop sequence immediately follows the nucleotides encoding the C-terminal lysine. Thus, the cDNA sequence of human DNase I is consistent with what one might expect of a simple secreted enzyme. Human DNase I and bovine DNase I (Fig. 1B) both contain four cysteines (amino acids 101, 104, 173, and 209) and two potential N-linked glycosylation sites (amino acids 18 and 106). The active site histidine (amino acid 134) is also conserved (23). Despite extensive structural homology, there is significant divergence at the amino acid level with only 77%

overall amino acid identity. Shown in Fig. 1 C and D are schematic representations of rhDNase based on the published three-dimensional x-ray crystallographic structure and hydrogen-bonding scheme of bovine DNase I (24, 25). Two six-stranded ,-pleated sheets are packed against each other to form the core of the protein, which is surrounded by eight a-helices and six turns. The amino acid residues that differ between the human and the bovine proteins are shown in blue. Except for one region in the 8-sheet structure, almost all of the differences between the human and the bovine proteins are located in hydrophilic regions on the surface of the molecule. Thus, there is sufficient divergence between bovine and human pancreatic DNase I in potentially immunogenic regions to suggest that some of the adverse reactions to the bovine protein could be due to an immune response to a foreign protein. Indeed, in one study, half of the patients given multiple doses of bovine pancreatic DNase developed significant serum antibody titers (38). To characterize the properties of rhDNase, 293 cells were transfected with plasmid pRK.DNase containing the human DNase I signal sequence and coding sequence. Cell supernatants from transfected and untransfected cells were analyzed for deoxyribonuclease activity by two methods, fluidphase hydrolysis of radiolabeled DNA and zymography after PAGE. Supernatants from transfected cells, but not from untransfected cells, hydrolyzed radioactive DNA to acid soluble nucleotides (Fig. 2A). Increased amounts of deoxyribonuclease activity were observed as a function of the time after transfection. Moreover, an additional 35-kDa protein was observed when supernatants from transfected cells, but not from untransfected cells, were subjected to PAGE (Fig. 2B). Zymography confirmed that deoxyribonuclease activity was associated with this additionally secreted protein (Fig. 2C). The predicted molecular mass of human DNase I from the deduced amino acid sequence is 29,300 kDa. The higher molecular weight and broad band on PAGE of rhDNase expressed in mammalian cells is from glycosylation; N-glycanase treatment reduces the material to a single band of -30 kDa (data not shown). We purified rhDNase and have further characterized its enzymatic activity. Like bovine DNase I, rhDNase depends on divalent cations for optimal enzymatic activity. Hydrolysis of radioactive DNA by rhDNase in the presence of 4 mm Ca2' and Mg2+ is reduced >10-fold by the absence of either divalent cation. Moreover, like bovine DNase I, rhDNase is inactivated by heat (100C for 10 min), is specifically inhibited by actin (using a 1:1 molar ratio of rhDNase and rabbit muscle actin), and shows optimal enzymatic activity at neutral pH (5.5-7.5) (39). Both the reported concentrations of Ca2+ and Mg2+ in sputum and the pH of sputum are sufficient to support the activity of rhDNase (40, 41). Effect of rhDNase on Cystic Fibrosis Sputum. The effects of rhDNase on cystic fibrosis sputum were examined qualitatively using a "pourability" assay. The results of a representative experiment in which saline, purified rhDNase, protease-free bovine DNase I, and heat-inactivated rhDNase were incubated with cystic fibrosis sputum at 37C are shown in Table 1. Purified rhDNase and protease-free bovine DNase (Worthington) but not saline or heat-inactivated rhDNase dramatically increase the pourability of cystic fibrosis sputum in a time-dependent fashion. Similarly, supernatants from mammalian cells transfected with an expression plasmid containing the human DNase I signal sequence and coding sequence, but not supernatants from mammalian cells transfected with a control plasmid, reduced the viscosity of cystic fibrosis sputum. The qualitative results of the pourability assay were confirmed by quantitative measurement of viscosity with a Brookfield cone-plate viscometer. Cystic fibrosis sputum was obtained from four different individuals with either acute

9190
A

Medical Sciences: Shak et al.

Proc. Natl. Acad. Sci. USA 87 (1990)

1 TCCTGCACAGGCAGTGCCTTGAAGTGCTTCTTCAGAGACCTTTCTTCATAGACTACTTTTTTTTCTTTAAGCAGCAAAAGGAGAAAATGTCATCAAAGGATATTCCAGATTCTTGACAG

-20

-10
eu~euG1
a~eu~eu

t~rg~lyh~t~ys
10
20

nGIXA~a~a1SerLeuLysIleAlaAla
40

121 CATTCTCGTCATCTCTGAGGACATCACCATCATCTCAGGATGAGGGGCATGAAGCTGCTCGGGGCGCTGCTGGCACTGGCGGCCCTACTGCAGGGGGCCGTGTCCCTGAAGATCGCAGCC 30

PheAsnIleGlnThrPheGlyGluThrLysMetSerAsnAlaThrLeuValSerTyrIleValGlnIleLeuSerArgTyrAspIleAlaLeuValGlnGluValArgAspSerHisLeu
241 TTCAACATCCAGACATTTGGGGAGACCAAGATGTCCAATGCCACCCTCGTCAGCTACATTGTGCAGATCCTGAGCCGCTATGACATCGCCCTGGTCCAGGAGGTCAGAGACAGCCACCTG

50

60

70

80

ThrAlaValGlyLysLeuLeuAspAsnLeuAsnGlnAspAlaProAspThrTyrHisTyrValValSerGluProLeuGlyArgAsnSerTyrLyaGluArgTyrLeuPheValTyrArg
361 ACTGCCGTGGGGAAGCTGCTGGACAACCTCAATCAGGATGCACCAGACACCTATCACTACGTGGTCAGTGAGCCACTGGGACGGAACAGCTATAAGGAGCGCTACCTGTTCGTGTACAGG

90

100

110

120

ProAspGlnValSerAlaValAspSerTyrTyrTyrAspAspGlyqy GluProCyaGlyAsnAspThrPheAsnArgGluProAlaIleValArgPhePheSerArgPheThrGluVa1
481 CCTGACCAGGTGTCTGCGGTGGACAGCTACTACTACGATGATGGCTGCGAGCCCTGCGGGAACGACACCTTCAACCGAGAGCCAGCCATTGTCAGGTTCTTCTCCCGGTTCACAGAGGTC

130

140

150

160

601 AGGGAGTTTGCCATTGTTCCCCTGCATGCGGCCCCGGGGACGCAGTAGCC"GAkTCGACGCTCTCTATGACGTCTACCTGGATGTCCAAGAGAAATGGGGCTTGGAGGACGTCATGTTG 170 180 190 200

ArgGluPheAlaIleValProLeu6i0AlaAlaProGlyA1pAlaValAlaGluIleAspAlaLeuTyrAspValTyrLeuAspValGlnGluLysTrpGlyLeuGluAspValMetLeu

721 ATGGGCGACTTCAATGCGGGCTGCAGCTATGTGAGACCCTCCCAGTGGTCATCCATCCGCCTGTGGACAAGCCCCACCTTCCAGTGGCTGATCCCCGACAGCGCTGACACCACAGCTACA

MetGlyAspPheAsnAlaGlyqysSerTyrValArgProSerGlnTrpSerSerIleArgLeuTrpThrSerProThrPheGlnTrpLeuIleProAspSerAlaAspThrThrAlaThr
210 220 230
240

ProThrHis8y4AlaTyrAspArgIleValValAlaGlyMetLeuLeuArgGlyAlaValValProA1pSerAlaLeuProPheAsnPheGlnAlaAlaTyrGlyLeuSerAspGlnLeu
841 CCCACGCACTGTGCCTATGACAGGATCGTGGTTGCAGGGATGCTGCTCCGAGGCGCCGTTGTTCCCGACTCGGCTCTTCCCTTTAACTTCCAGGCTGCCTATGGCCTGAGTGACCAACTG

250

260

AlaGlnAlaIleSerAspHisTyrProValGluValMetLeuLysOP* 961 GCCCAAGCCATCAGTGACCACTATCCAGTGGAGGTtATGCTGAAGTGAGCAGCCCCTCCCCACACCAGTTGAACTGCAG

B
human

50 10 40 20 30 LKIAAFNIQTFGETKMSNATLVSYIVQILSRYDIALVQEVRDSHLTAVGK
******** ************ **** *
**** * ******

***

bovine
human

LKIAAFNIRTFGETKMSNATLASYIVRIVRRYDIVLIEQVRDSHLVAVGK
60
70

80
**

90
**

100
* * ****

LLDNLNQDAPDTYHYVVSEPLGRNSYKERYLFVYRPDQVSAVDSYYYDDG
*** **** * *********************

bovine
human

LLDYLNQDDPNTYHYVVSEPLGRNSYKERYLFLFRPNKVSVLDTYQYDDG
150 120 140 130 110 CEPCGNDTFNREPAIVRFFSRFTEVREFAIVPLHAAPGDAVAEIDALYDV ** .**** .* .**** .* .* * .* * .***** .**_** .****** ..**** 160

bovine
human

CESCGNDSFSREPAVVKFSSHSTKVKEFAIVALHSAPSDAVAEINSLYDV
170 180
190

200

YLDVQEKWGLEDVMLMGDFNAGCSYVRPSQWSSIRLWTSPTFQWLIPDSA *****.** *.**********.**** YLDVQQKWHLNDVMLMGDFNADCSYVTSSQWSSIRLRTSSTFQWLIPDSA

210 220 230

.********.**.********** _

bovine
human

240

250

bovine
human

DTTATPTHCAYDRIVVAGMLLRGAVVPDSALPFNFQAAYGLSDQLAQAIS *****.*.********** **..*** ** ** ******** -* *** DTTATSTNCAYDRIVVAGSLLQSSVVPGSAAPFDFQAAYGLSNEMALAIS

260 DHYPVEVMLK
DHYPVEVTLT

___

bovine

__ l

FIG. 1. Human DNase I nucleotide sequence and deduced amino acid sequence. (A) The nucleotide and deduced amino acid sequences of human DNase I (clone hDNase 18-1) are shown. Nucleotides are numbered at left. Amino acids are numbered above the line starting at Leu + 1 of the mature enzyme sequence and preceded by a 22-amino-acid putative signal sequence (underlined). The four cysteine residues are printed in boldface. Two potential N-linked glycosylation sites are indicated by lines above the amino acid sequence. (B) The deduced amino acid sequence of human DNase I is compared with the amino acid sequence of bovine DNase 1 (23). Identical amino acid residues are indicated by an asterisk (*). Conserved differences are indicated by a period. (C and D) A schematic representation of human DNase I was constructed from the published three-dimensional x-ray crystallographic structure and hydrogen-bonding scheme of bovine DNase 1 (24, 25). An a carbon trace is shown in C, where the locations of the amino acid residues that differ between the human and the bovine proteins are shown in blue. The corresponding structural features are highlighted in D, where the a-sheets are shown in red, the a-helices in blue, and the turns and loops in green.

Medical Sciences: Shak et al.

A
120004

E 0. CL 0
0

080004

3
400010-

~0

0
0

Time, hr
B
a) c

4* -T

.]w. I "i
5W4:i;
vF

:mc

Proc. Natl. Acad. Sci. USA 87 (1990)

9191

B
23.1 6.6 -4.4
--

s.

a)
(A
0

CL

8
0

to

12

18

24

l n,
I

-2.3 -2.2

I
2 4 8 20

0r1h
co
c_

2s4

8g20

0 1

rhDNase (jg/mi)

rhDNase (~Lg/ml)

co

o zoU
c

w o B

cn

,-:

'

-97-

68-43-

FIG. 3. rhDNase reduces the viscosity of cystic fibrosis sputum (quantitative viscometry) and is associated with a reduction in sputum DNA size. Cystic fibrosis sputum was incubated with various concentrations of rhDNase for 15 min at 37C. (A) Viscosity measured at 25C using a Brookfield cone-plate viscometer. Shown are data at a low shear rate (2.5 rpm) obtained as the rpm was decreased from 100 to 0.5 rpm (mean SEM, n = three to seven experiments). (B) Agarose gel electrophoresis of sputum DNA. The size in kilobases of DNA standards is indicated.

-29::. _-

t8-

FIG. 2. Expression of enzymatically active rhDNase. (A) Fluidphase deoxyribonuclease activity in supernatants from transfected (e) and untransfected (o) cells. (B) SDS/polyacrylamide gels (10%6) of bovine DNase I and untransfected and transfected cell supernatants. Gels were stained with silver. The molecular mass (MW) standards are indicated (in kDa). (C) Zymogram of in situ deoxyribonuclease activity in SDS/polyacrylamide gels of the same samples shown in B.
or persistent infection. The results of experiments in which cystic fibrosis sputum was incubated with various concentrations of rhDNase for 15 min at 37C are summarized in Fig. 3. The reduction of viscosity by rhDNase is concentrationdependent (Fig. 3A) and is associated with the reduction in size of sputum DNA as measured by agarose gel electrophoresis (Fig. 3B). The qualitative and quantitative effects of rhDNase on the viscosity of cystic fibrosis sputum were similar in samples obtained on several occasions from the same individual and in samples obtained from different individuals. No effect of rhDNase was observed on uninfected sputum.

Table 1. rhDNase reduces the viscosity of cystic fibrosis sputum (qualitative pourability assay) 0 min 5 min 15 min 30 min Sample 0 tr tr 0 Saline tr 4+ 0 3+ rhDNase (50 /ug/ml) 1+ 4+ 0 2+ Bovine DNase 1 (50 Ag/ml) rhDNase (100'C for 10 min; 50 Ag/ml) 0 0 0 0 Pourability was assessed qualitatively after various periods of incubation at 370C by inverting the tubes and observing the movement of sputum after a tap of the side of the tube, where 0 = no movement, tr = movement of <10%o of the sputum down the tube, 1+ = movement of 10-20% of the sputum, 2+ = movement of 20-50% of the sputum, 3 + = movement of >50%o of the sputum, and 4+ = movement of all the sputum freely down the tube.

Additional experiments were performed to examine the effects of other agents on the viscosity of cystic fibrosis sputum. Whereas small amounts of rhDNase (8 ,ug/ml) greatly reduced the viscosity of cystic fibrosis sputum after a 15-min incubation at 37C from 732 centipoise to 188 centipoise (at a 2.5 rpm shear rate), large amounts (up to 250 jtg/ml) of other enzymes such as lysosomal DNase II, RNase A, and trypsin and chymotrypsin had minimal effects, yielding viscosities of 596, 568, 616, and 636 centipoise, respectively. Addition of trypsin and chymotrypsin to purified rhDNase did not increase its ability to reduce sputum viscosity (data not shown). These results indicate that contaminating proteases are not required for rhDNase to reduce the viscosity of cystic fibrosis sputum. Finally, Mucomyst (Nacetylcysteine), a currently approved mucolytic agent, at concentrations likely to be achieved in sputum after aerosolization (2400 ,ug/ml) also had minimal effects on viscosity, yielding a viscosity of 608 centipoise. Lung secretions are complex nonhomogenous materials that form a viscous hydrophilic gel. The viscosity of uninfected lung secretions or mucus ("white" sputum) has been attributed primarily to mucus glycoproteins. The viscosity of infected or purulent lung secretions ("yellow" or "green" sputum) has been attributed to both mucus glycoproteins and DNA. The dramatic ability of rhDNase to reduce the viscosity of cystic fibrosis sputum indicates that DNA is the factor largely responsible for the increased viscosity of purulent sputum. Although the molecular mechanism responsible for the viscosity of mucus is not clear, it has been suggested that aggregation of mucus glycoproteins yields an "entangled network" of fibers that is responsible for its viscoelastic properties (42, 43). Our results suggest that large amounts of DNA released by neutrophils that accumulate and lyse in purulent secretions contribute substantially to the entangled network. Although the reported clinical effects of bovine pancreatic DNase I in uncontrolled human studies are encouraging (17-20), it has not yet been conclusively shown that reduction in the viscosity of lung secretions leads to improved lung function. However, studies in experimental animals demonstrate that the increased viscosity of purulent sputum is associated with decreased ciliary transport in isolated frog

9192

Medical Sciences: Shak et al.

Proc. Natl. Acad. Sci. USA 87 (1990)

15. Armstrong, J. B. & White, J. C. (1950) Lancet l, 739-740. 16. Chernick, W. S., Barbero, G. J. & Eichel, H. J. (1961) Pediatrics, 27, 589-5%. 17. Elmes, P. C. & White, J. C. (1953) Thorax 8, 295-300. 18. Salomon, A., Herschfus, J. A. & Segal, M. S. (1954) Ann. Allergy 12, 71-79. 19. Spier, R., Witebsky, E. & Paine, J. R. (1961) J. Am. Med. Assoc. 178, 878-886. 20. Lieberman, J. (1968) J. Am. Med. Assoc. 205, 312-313. 21. Raskin, P. (1968) Am. Rev. Respir. Dis. 98, 597-598. 22. Lieberman, J. (1962) Am. J. Dis. Child. 104, 342-348. 23. Liao, T.-H., Salnikow, J., Moore, S. & Stein, W. H. (1973) J. Biol. Chem. 248, 1489-1495. 24. Suck, D., Oefner, C. & Kabsch, C. (1984) EMBO J. 3, 24232430. 25. Oefner, C. & Suck, D. (1986) J. Mol. Biol. 192, 605-632. 26. Funakoshi, A., Tsubota, Y., Wakasugi, H., Ibayashi, H. & Takagi, Y. (1977) J. Biochem. 82, 1771-1777. 27. Murai, K., Yamanaka, M., Akagi, K., Anai, M., Mukai, T. & Omae, T. (1978) Biochim. Biophys. Acta 517, 186-194. 28. Love, J. D. & Hewitt, R. R. (1979) J. Biol. Chem. 254, 1258812594. 29. Ito, K., Minamiura, N. & Yamamoto, T. J. (1984) Biochemistry 95, 1399-1406. 30. Lauffer, L., Garcia, P. D., Harkins, R. N., Coussens, L., Ullrich, A. & Walter, P. (1985) Nature (London) 318, 334-338. 31. Gorman, C. M., Gies, D. R. & McCray, G. (1990) DNA Protein Eng. Tech. 2, 3-6. 32. Laskowski, M. (1946) Arch. Biochem. 11, 41-48. 33. Rosenthal, A. L. & Lacks, S. A. (1977) Anal. Biochem. 80, 76-90. 34. Keal, E. E. (1971) Postgrad. Med. J. 47, 171-177. 35. Lieberman, J. (1968) Am. Rev. Respir. Dis. 97, 654-661. 36. Lieberman, J. (1968) Am. Rev. Respir. Dis. 97, 662-672. 37. Charman, J. & Reid, L. (1973) Biorheology 10, 295-301. 38. Johnson, A. J., Goger, P. R. & Tillett, W. S. (1954) J. Clin. Invest. 33, 1670-1686. 39. Moore, S. (1981) in The Enzymes, ed. Boyer, P. D. (Academic, New York), Vol. 14, pp. 281-296. 40. Potter, J. L., Matthews, L. W., Spector, S. & Lemm, J. (1967) Am. Rev. Respir. Dis. 96, 83-87. 41. Levy, J., Smith, A. L., Kenny, M. A., Ramsey, B. & Schoenknecht, F. D. (1983) J. Infect. Dis. 148, 1069-1076. 42. Boat, T. F., Cheng, P. W., Iver, R. N., Carlson, D. M. & Polony, 1. (1976) Arch. Biochem. Biophys. 177, 95-104. 43. Verdugo, P. (1982) Cell Motil., Suppl. 1, 1-5. 44. Litt, M. (1970) Arch. Intern. Med. 126, 417-423. 45. Dulfano, M. J. & Adler, K. B. (1975) Am. Rev. Respir. Dis. 112, 341-347. 46. Dulfano, M. J., Adler, K. & Philippoff, W. (1971) Am. Rev. Respir. Dis. 104, 88-98. 47. Smith, A. L., Redding, G., Doershuk, C., Goldmann, D., Gore, E., Hilman, B., Marks, M., Moss, R., Ramsey, B., Rubio, T., Schwartz, R. H., Thomassen, M. J., Williams-Warren, J., Weber, A., Wilmott, R. W., Wilson, H. D. & Yagev, R. (1988) J. Pediatr. 112, 547-554.

pharyngeal palate (44, 45). Moreover, increase in viscosity of purulent sputum is correlated with difficulty in expectoration in patients with chronic bronchitis (46). Finally, a recent multicenter study of antibiotic therapy of exacerbations in cystic fibrosis reveals that a reduction in sputum DNA content was correlated with clinical improvement (47). Our studies suggest that inhalation of a rhDNase aerosol may reduce the viscosity of purulent secretions and thereby help individuals with cystic fibrosis and other patients with pneumonia or bronchitis to clear their airways and breathe more easily.
We thank L. Presta and C. Eigenbrot for preparing the schematic representations of human DNase I; A. Liu for technical assistance; our colleagues in cell genetics, process sciences, and assay development and services for providing purified rhDNase used in some of these studies; and B. Davis for obtaining cystic fibrosis sputum. We also thank D. V. Goeddel, E. G. Peralta, W. E. Holmes, and D. H. Smith for helpful discussions and D. Botstein for reviewing the manuscript.
1. Wood, R. E., Boat, T. F. & Doershuk, C. F. (1976) Am. Rev. Respir. Dis. 113, 833-878. 2. Marks, M. I. J. (1981) Pediatrics 98, 173-179. 3. Rommens, J. M., lannuzzi, M. C., Kerem, B., Drumm, M. L., Melmer, G., Dean, M., Rozmahael, R., Cole, J., Kennedy, D., Hidaka, N., Zsiga, M., Buchwald, M., Riordan, J. R., Tsui, L. & Collins, F. S. (1989) Science 245, 1059-1065. 4. Riordan, J. R., Rommens, J. M., Kerem, B., Alon, N., Rozmahel, R., Grzelczak, Z., Zielenski, J., Lok, S., Plavsic, N., Chou, J., Drumm, M. L., lannuzzi, M. C., Collins, F. S. & Tsui, L. (1989) Science 245, 1066-1073. 5. Karem, B., Rommens, J. M., Buchanan, J. A., Markiewicz, D., Cox, T. K., Chakravarti, A., Buchwald, M. & Tsui, L. (1989) Science 245, 1073-1080. 6. Boat, T. F. (1988) in Textbook of Respiratory Medicine, eds. Murray, J. F. & Nadel, J. A. (Saunders, Philadelphia), pp. 1126-1152. 7. Potter, J. L., Matthews, L. W., Lemm, J. & Spector, S. (1963) Ann. N. Y. Acad. Sci. 106, 692-697. 8. Davis, S. D. & Bruns, W. T. (1978) Am. Rev. Respir. Dis. 117, 176-178. 9. Chernick, W. S. & Barbero, G. J. (1959) Pediatrics 24, 739745. 10. Potter, J., Matthews, L. W., Lemm, J. & Spector, J. S. (1960) Am. J. Dis. Child. 100, 493-495. 11. Matthews, L. W., Spector, S., Lemm, J. & Potter, J. L. (1963) Am. Rev. Respir. Dis. 88, 199-204. 12. Potter, J. L., Matthews, L. W., Spector, S. & Lemm, J. (1965) Pediatrics 36, 714-720. 13. Vandaux, P. & Waldvogel, F. A. (1980) J. Infect. Dis. 142, 586-593. 14. Mendelman, P. M., Smith, A. L. & Levy, J. (1985) Am. Rev. Respir. Dis. 132, 761-765.