COMPLEX
FORMATION
BETWEEN
BASIC
ANTIBIOTICS
AND
DEOXYRIBONUCLEIC
ACID
IN
HUMAN
PULMONARY
SECRETIONS
Joseph 1. Potter, M.D., Ph.D., LeRoy W. Matthews, M.D.,
Samuel Spector, M.D., and Joy Lemm, BA.
Department of Pediatrics, Babies and Children.s Hospital, Western Reserve
Unwer.sity School of Medicine, Cleveland, Ohio
(Submitted March 19; accepted for publication May 12, 1965.)
Presented in part at the Thirty-second Annual Meeting of the Society for Pediatric Research, Atlantic
City, New Jersey, May 8-10, 1962.
This investigation was supported by grants from the U.S. Public Health Service (AM-08305 and
AM-0385), and a grant from the National Cystic Fibrosis Research Foundation.
ADDRESS: (J.L.P.) Bal)ies and Childrens hospital, 2103 Adelbert Road, Cleveland, Ohio 44106.
714
PEDIATRICS, Vol. 36, No. 5, November 1965
T
HE Precil)itatioil of Ilucleic aci(ls bystreptomycin, a basic antibiotic, was
reported by Cohen in 1947.’ More recently,
the same author publisileci a method for the
isolation of deoxyribonucleic acid
(
DNA)and of nibonticleic acid
(
RNA)
frombac-terial extracts with the use of the same
anti-biotic.2 In our laboratory, human pulmonary
secretions were initially collected either by
direct aspiration from the tracheostomies
of laryngectomized patients, or by the
transfer of expectorated material from pa-tients with cystic fibrosis and
bronchiec-tasis illto prepared vials containing a
mix-ture of neomycin and polymixin. 4 Tllese
antibiotics were introduced to minimize bacterial alteration of tile pulmonary glyco-proteins. It was Iloted that after
lyophiliza-tion of the secretions and reconstitution as
a 1% suspension, the supernatant solutions
obtained by centnifugation demonstrated
antibacterial activity in samples from the
broncinectasis and laryngectomy groups.
However, comparable supernatants from
the secretions of patients with cystic fibrosis
showed no antibacterial activity. It has been subsequently demonstrated that the absence of antibacterial activity in
the cystic fibrosis secretions is directly
re-lated to the larger amounts of DNA present
as compared with tile other two groups of
secretions.14 Tile present work is concerned with a sttidy of complex formation between
basic antibiotics and DNA in human
pul-monary secretions and in simple systems
containing only DNA and antibiotic.
Bind-ing afl(l PreciPitatioll of RNA iy tllese
antibiotics is also described. The influence
of pH, ionic strength, and nuclease on tile
complex have been studied. The minimtim
chain lengtil of polynucleotide required for precipitation has been determined. The
possible significance of the reaction in
whole pulmonary secretions and in relation
to tile treatment of purtilent foci of
infec-tioll is considered.
M ETHODS
DNA was Prel)arel from various sotirces
by the method of Kay, Simmons, and
Dotince.5 RNA was prepared from yeast
by the metilod of Kay and Dounce.C
Trans-fer RNA0 and deoxynibonticlease I
(
DNase I) f
were commercial products. A series of polynticleotide chains composed exclusively of pynimidine nucleotides was preparedfrom calf thymus DNA by the method of
Burton and 2 A sample of DNA
was treated with Burton’s reagent for 16
ilours at room temperature. Formic acid
and diphenylamine were removed by ether
extraction, and the pH of the aqueous layer
adjusted to 8.5 with ammonium
hydrox-ide. The mixture was chromatograpiled
* Nutritional Biochemical Company, Cleveland, Ohio.
4.
l2
>. 1.0 H
En
z 0.8
0
-J
0.6
0
H 0.
0 0.4
02
\
\
\\
\\
\
220 240 260 280 300
ARTICLES
on a Dowex-1X2 coltimn, 15 by 0.9 cm,
in the formate form. Elution was effected
with increasing molarity of ammonium
formate at pH 4.5. Chain lengths were
determined with snake venom
phosphodi-esterase as previously described.5’
Human pulmonary secretions were
col-lected in vials containing 40 mg neomycin
sulfate and 400 .g polymixin stilfate, or in the absence of the antibiotic mixture. There were approximately 60 gm of whole
pulmonary secretions in each vial. The
secretions were lyophilized, and the dried material suspended in water at a
concen-tration of 10 mg/mi. After stirring for
one hour at 5#{176},the suspension was
centni-fuged at 13,000 RPM for one hour, and
the supernatant removed. The sediment
was washed with a little water and
re-centrifuged as above. The combined
supernatants were designated water-solu-ble fraction
(
WS ), and the sedimentdesignated water-insoluble fraction (WI).
The antibiotic activity of the
(
WS) fractions was assayed by wettingWilat-man No. 3 MM filter paper discs in the
solution. Tile discs were lightly blotted to remove excess fluid, and then applied to agar plates which had been previotisly streaked with strains of staphylococci sen-sitive to the particular antibiotic. The plates
were incubated at 37#{176}overnight, and a
clear area of at least 2.0 mm around the
disc was interpreted as evidence of anti-bacterial activity.
RESULTS
The
(
WS)
fraction of ten specimens of secretions collected in vials witil tileanti-biotic mixture from the laryngectomized
group and of six bronchiectatic specimens exhibited antibacterial activity. Only two of
thirteen samples of the
(
WS)
fractions ofthe cystic fibrosis secretions similarly col-lected exhibited antibacterial activity. The
(WS) fractions of a series of samples from the bronchiectasis and laryngectomized groups collected in the absence of the anti-biotic mixttire did not exhibit antibacterial activity. This indicated that tile secretions
WAVELENGTH mji
FIG. 1. The ultraviolet spectra of the (WS) fraction
and (WI) fraction from the pulmonary secretions
of patients with cystic fibrosis. Open circles =
(‘tVS) fraction; closed circles = ( WI) fraction after solubilization by treatment with DNase I.
had 110 inherent antibacterial activity, at
least under these experimental conditions.
The failure to demonstrate the activity of
added neomycin-polymixin in the case of
the cystic fibrosis secretions indicated
eitiler inactivation or removal from the
soltible fraction.
On the basis of Cohen’s early study
con-cerning tile precipitation of DNA by strep-tomycin, it seemed likely tilat tile larger
amounts of DNA present in the cystic
fibrosis secretions1 might account for the
observed phenomenon . The ultraviolet
spectrum of the (WS
)
fraction of samplesfrom the grotip of patients with cystic
fibrosis is shown in Figure 1, and exhibits
a peak at 275 ., characteristic of protein.
The
(
WI)
fractions from these secretions were stispended ill 10 ml of distilled waterand 2.0 .g crystalline DNase I added to
each sample. After incubation for two hours
at 370, the suspensions were centrifuged at
I-z
I-z
LiJ
a-(I)
z
z
a !00
80
60
40
20
0 20 40 60 80
Antibiotic
Neomycihl*
Polymixin
Kanamycin
Colistint Streptomycin
Jg NEOMYCIN BASE
FIG. 2. The precipitation of calf thvmus DNA by
neomycin.
spectrum of the resultant superriatant
solu-tions is also shown in Figure 1, and exhibits
a peak at 260 p., characteristic of nucleic acid. Assay of these supernatant solutions
showed distinct antibacterial activity in all
of the specimens examined. Next the (WI)
fractions of six samples of cystic fibrosis
secretions collected in the
neomycin-poly-mixin mixture were suspended in M sodium
chloride and stirred for one hour at 25#{176}.
The suspensions were centrifuged as above,
and the ultraviolet spectrum of the
resul-tant supernatant solutions was found to
TABLE I
ANTIBIoTIcs PRECIPITATED WITh DNA
Molar Ratio Antibiotic/DNA
phosphorus
0.17-0.19
0.14
0.57-0.60
0.14
0.6 -0.625
* Supplied by the Upjohn Company, Kalamazoo,
Michigan.
fSupplied by the Warner-Lambert Research
Insti-tute, Morris Plains, N. J., research affiliate of Warner Chilcott Laboratories.
have a typical nucleic acid spectrum with
a peak at 260 ti.. Assay of these supernatants again showed distinct antibacterial activity
in all of the specimens examined.
Complex formation was next investigated
in systems containing antibiotic and highly
purified DNA, RNA, or polynucleotides of
varying chain lengths. To determine the
amount of antibiotic bound experiments
were carried out as follows. Increasing
amounts of antibiotic were added to a series
of tubes containing a constant amount of
DNA. The tubes were centrifuged at 13,000
RPM for one hour, and the optical density
of the supernatant solution measured at
260 &. In this manner the minimum amount
of antibiotic required to precipitate all the
DNA was determined, and from this value
the ratio of antibiotic to DNA-phosphorus calculated. Figure 2 shows the results of a
typical experiment carried out with calf
thymus DNA and neomycin. The curve is
linear and intersects the abscissa at a value
of 80 neomycin base. Table I presents
the results of experiments carried out with
other antibiotics. The data is presented as mole of antibiotic bound per mole of DNA
phosphorus. Table II lists several
antibi-otics which did not precipitate with DNA
in the test system.
Preparations of DNA isolated from thy-mus, liver, and spleen of calf demonstrated
identical precipitation curves with
neomy-cm, and were thus indistinguishable in the test system. The characteristics of the re-action were not altered by prior treatment of the DNA with heat.1#{176}A sample of DNA isolated from Ehrlich Ascites cells yielded
the same molar ratio of antibiotic to
DNA-phosphorus as found for DNA prepared
TABLE II
ANTIBIOTIcs NOT PRECIPITATED WITh DNA
Chloramphenicol*
Tetracycline
Methicillin
Penicillin G
Bacitracin
* Supplied by Parke, Davis, and Company, Detroit,
.2
-1.0
-08
-0.6
-0.4
-0.2
-0 ,0
c.,J
TUBE NO.
Fic. 3. Fractionation of polynucleotide fragments produced from calf thymus DNA by the formic
acid-diphenylamine reagent. Peak 1 = pCp; peak
2 = pTp; peak 12 = octonucleotides to
dodeca-nucleotides.
from the various tissues of the calf. It was of interest to examine the behavior of RNA
in the test system with neomycin, even
though RNA has not been detected by us
in human pulmonary 4 Both
yeast RNA and transfer RNA yielded molar
ratios of antibiotic to nucleic acid phos-phorus identical to that obtained with DNA
in the same system.
Since DNA and RNA exhibited identical precipitation curves with neomycin, it
be-came of interest to determine whether
chain length governed the precipitation
phenomenon. Figure 3 shows the elution
pattern of a sample of calf thymus DNA
after treatment with Burton’s reagent.7 The chain length of the pyrimidine
frag-ments varied from the monomers
pCp(5’-0-phosphoryldeoxycytidine 3’-phosphate)
and pTp(5’-O-phosphorylthymidine
3’-phosphate), eluted with 0.25 M ammonium
formate, to fragments with an average chain
length of 10, eluted with 2 M ammonium
formate. It was found that pTp and pCp
did not form precipitates with neomycin, nor did the polynucleotide chains in peaks
3 to 11.
Figure 4 shows the reaction of the
poly-nucleotides from peak 12 (eluted with 2 M
ammonium formate and having an average
chain length of 10) with neomycin. About two-thirds of the polynucleotide chains in this fraction are precipitated by neomycin,
0 80 60 240 340 420 500
pg NEOMYCIN SO4
FIG. 4. The precipitation of polynucleotides
corn-posed exclusively of prirnidine residues (average length = 10) by neomycin sulfate.
whereas the remainder are non-precipitable
even in the presence of an excess of the
antibiotic.
The digestion of calf thymus DNA by
streptodornase produces a spectrum of pro-ducts ranging in size from the monomer to
polynucleotides of a chain length of about 10.8 Figure 5 shows that about 50% of the
decanucleotide fraction eluted with 2 NI
ammonium formate buffer from a
Dowex-1-formate column was precipitated by
neo-mycin, while the remainder of the
poly-nucleotide chains in the fraction were
non-precipitable even in the presence of a
large excess of antibiotic.
The influence of ionic strength on
com-plex formation and precipitation of DNA
by neomycin was examined. The ionic
strength was adjusted by the addition of
sodium chloride to the reaction mixture. Figure 6 shows that at an ionic strength of 0.225 or less the added electrolyte had no
718
0
(0 c’J
0.4
-0.2
-50
00
80
60
40
20
I-z
I-z a::
w
a-c/I
z
z a
JLQ NEOMYCIN BASE
Fic. 6. The tirect of ionic strength on the precipi-tation of DNA 1) neomvcin sulfate. crosses =
water; open circles = ionic strength 0.225; half-shaded circles = ionic strength 0.250; closed
circles = ionic strength 0.3.
0 20 40 60 80
I.2#{149}
0.8
0.6
0 25 50 75 100 125
NEOMYCIN SO4
Fic. 5. The precipitation of polvnucleotides
(aver-age chain length = 10) obtained from a strepto-dornase digest of DNA by neornycin sulfate.
tile right, and at an ionic strength of 0.3 or
higher the marked effect of the added
dee-trolyte in the direction of dissociation of
the complex completely prevented precipi-tation of the DNA by neomycin.
It was of interest to sttidy the effect of
pH on the precipitation of DNA by
neo-mycin. The reaction was carried out in
dilute buffers
(
ionic strength = 0.1 or less)of the desired pH. At pH values from 4 to
7, there was no detectable influence of
hydrogen ion concentration on the reaction.
At pH8, and higher values, precipitation of DNA by added neomycin was not observed.
COMMENT
In 1947, Cohen used streptomycin to
study some aspects of bacteriophage
sur-face structure. In this study, he observed
that the addition of streptomycin to neutral
solutions of high polymer DNA resulted in
the formation of opalescence or of floccu-lent precipitates.1 Cohen studied some of
the parameters of the reaction and showed
that whereas slight depolymerization of the DNA witil nuclease did not alter its pre-cipitability, alkali-degraded DNA or RNA
showed distinctly less precipitability with
the antibiotic. He also discovered that the complex could be dissociated in M sodium
chloride. In 1960, Cohen and Lichtenstein used streptomycin to isolate DNA and RNA from bacterial extracts. The DNA precipi-tates were found to contain 0.5 to 0.7 mole of streptomycin per mole of phosphorus.
The failure to demonstrate
neomycin-polymixin activity in the soluble fraction
of the 1% reconstituted secretions of patients with cystic fibrosis, in contrast to the easily demonstrable activity in less purulent secre-tions of patients with bronchiectasis and
laryngectomy, suggested that the higher
levels of DNA present in cystic fibrosis
secretions might account for the observed
phenomenon. That this was the case was
conclusively demonstrated by the liberation of the antibiotic from the insoluble fraction
with DNase I or by soltibilization of both
I)NA and antibiotic in NI NaCl.
The reaction of nticleic acids with
ileo-mycin has 1)een extensively studied. All
of tile samples of l)NA which were
cx-amined had extremely high molectilar
weights and were indistinguishable in the
reaction witil the antibiotic. Heat
treat-ment,1#{176} which presumably results in the conversion of DNA to the single-stranded form with a resultant llalving of tile
molec-ular weight, had no effect on the
cilaracter-istics of the precipitation reaction. Similarly,
two samples of RNA, of much lower molec-ular weight, yielded precipitates with the
same molar ratio of antibiotic to nucleic
acid-phosphorus as obtained with high
polymer DNA.
The preparation and isolation of pyrimi-dine polynucieotide fragments, and of poly-nucleotide fragments from a streptodornase
digest of DNA permitted a determination of tile minimum cilain length required for
precipitation to occur. The average chain
length of the precipitable fragments was
ten, and probably the range was from
octonucleotides to dodecanucleotides. Since
one-half to two-thirds of the polynucleotide
chains formed precipitates with neomycin,
it is concluded tilat tile critical chain length
is about 10. The fact that fragments
con-taming only pyrimidine nucleotide
pre-cipitate with neomycin strongly suggests that base composition and sequence are not important in the reaction.
The crucial clinical question is whether
the type of complexing described is of
significance at the site of infection.
Assum-ing there are about 100 gm wet weight of
secretions in the lung, the cystic fibrosis
secretions would contain 400 mg of DNA
capable of binding 100 mg of neomycin
suffate. The magnitude of the binding in
bronchiectasis secretions would be of the order of 40 mg of the same antibiotic. The
pH of both types of secretions varies from 6.8 to 7.2, and is entirely compatible with
complex formation. The experiments
car-ried out with reconstituted secretions
rep-resented five- to tenfold (hlutiOlls of \VilOie
secretions, and thus ionic strengtil was
clearly hot d factor. \Vllcn whole, fresilly
exl)ectOrated secretions froill 1)cttients vith
cystic fibrosis receiving an aerosol of
neo-mycin sulfate four times a day 1 were
ex-amined, both free and complexed antii)iotic
were detected. This finding suggests that
at least high in the tracheobronchial tree,
there was some degree of dissociation, or
that tile complexing ability of the DNA
had been exceeded. The chemical milieu at
the actual site of infection, determined by the degree of cellular necrosis, leucocytic
response, and bacterial action is difficult to
determine. However, it is likely that the
concentration of DNA is highest at that
site.
In consideration of the basic difficulty of
introducing antibiotics into the lung,
whether by systemic administration or
by aerosol, it would appear that the level
of DNA in the secretions might serve as an
effective barrier to therapy with basic anti-biotics. In order to overcome this barrier,
drainage to remove purtilent material, or
tile use of nucleases or salt solution to break
tile complex might be effective. It is evident
that increasing tile level of the antibiotic
to exceed tile complexing ability of the
DNA, or the use of antimicrobial agents
not complexed are also stlltal)le approaches
to the problem.
SUMMARY
1. Complex formation and precipitation
of DNA by neomycin, with consequent
in-activation of the antibiotic, has been shown
to occtir in the pulmonary secretions of
patients \Vitil cystic fibrosis.
2. Highly polymerized l)NA from a
variety of sources, as well as two samples
of RNA, were precipitated by neomycin in
vitro. Polymixin, kanamycin, colymycin,
and streptomycin similarly co-precipitate with DNA in vitro.
3. The minimum chain length of
poiy-nucleotide required for precipitation in the
720
4. The complex is readily attacked by
DNase I resulting in the splitting of the DNA and the liberation of the antibiotic.
Both components of the complex are solubi-lized in NI sodium chloride.
5. The resistance of purulent foci of
in-fection to therapy with basic antimicrobial agents may, in part, be due to the complex
formation with the high levels of DNA
found at the site of infection.
REFERENCES
1. Cohen, S. S.: Streptomycin and desoxyribonu-clease in the study of variations in the prop-cIties of a bacterial virus.
J.
Biol. Chern.,168:511, 1947.
2. Cohen, S. S., and Lichtenstein,
J.:
The isolation of deoxyribonucleic acid from bacterialex-tracts by precipitation with streptornycin.
J.
Biol. Chern., 235:PC55, 1960.3. Potter, J. L., Matthews, L. W., Lernrn, J., and Spector, S.: Human pulmonary secretions in health and disease. Ann. N.Y. Acad. Sci.,
106:692, 1963.
4. Matthews, L. W., Spector, S., Lemm,
J.,
and Potter, J. L.: Studies on pulmonary secre-tions. I. The overall chemical composition ofpulmonary secretions from patients with
cystic fibrosis, bronchiectasis, and laryngec-tomy. Amer. Rev. Resp. Dis., 88: 199, 1963. 5. Kay, E. R. M., Simmons, N. S., and Dounce
A. L.: An improved preparation of sodium desoxyribonucleate. J. Amer. Chem. Soc.,
74:1724, 1952.
6. Kay, E. R. M., and Dounce A. L. : The
prepa-ration of sodium ribonucleate with the use
of sodium dodecyl sulfate.
J.
Amer. Chem.Soc., 75:4041, 1953.
7. Burton, K., and Petersen, C. P. : The quantita-tive distribution of pyrimidine nucleotides in caff thymus deoxyribonucleic acid.
Bio-chem. Biophys. Acta (Amsterdam) 26: 667, 1957.
8. Felix, F., Potter, J. L., and Laskowski, M.:
Ac-tion of venom phosphodiesterase on deoxy-ribooligonucleotides carrying a
monoesteri-fled phosphate on carbon 3.
J.
Biol. Chem.,235:1150, 1960.
9. Potter,
J.
L., and Laskowski, M.: Concerning the specfficit of streptococcal deoxyribonu-clease.J.
Biol. Chem., 234: 1263, 1959.10. Bollum, F.
J.
: Thermal conversion ofnon-priming deoxyribonucleic acid to primer. J.
Biol. Chem., 234:2733, 1959.
11. Matthews, L. W., Doershuk, C. F., Wise, M.,