Reprinted from the PRocrEDINGS oF THE NATIONAL ACADEMY OF SCIENCES
Vol. 61, No. 4, pp. 1478-1493. December, 1968.

SUPPRESSION OF HYDROGEN EXCHANGE IN STAPHYLOCOCCAL
NUCLEASE BY LIGANDS

By Anan N. Scuecurer,* Lapisnav Mordvex, t
AND CurisTIAN B. ANFINSEN

LABORATORY OF CHEMICAL BIOLOGY, NATIONAL INSTITUTE OF ARTHRITIS AND
METABOLIC DISEASES, NATIONAL INSTITUTES OF HEALTH, BETHESDA, MARYLAND

Communicated October 10, 1968

In 1954, Hvidt and Linderstrgm-Lang! introduced the study of hydrogen
exchange in proteins as a probe of secondary and tertiary structure. By com-
paring the kinetics of exchange of deuterium atoms between protein and water
with that of model compounds, they and other workers were able to make in-
ferences about the conformation of polypeptide chains in solution.2. Subsequent
modifications of this method, using tracer amounts of tritium,* 4 and gel filtra-
tion‘ rather than lyophilization to separate the labeled protein from the aqueous
medium, have had practical and theoretical advantages. Differences in the
kinetics of hydrogen exchange between enzymes in the presence and in the ab-
sence of their prosthetic groups or substrates have been shown in several cases.5—3
A further modification of these techniques, described in this paper, has allowed us
to make a study of the interactions of proteins with various ligands and of poly-
peptide chains with each other.

The studies reported below were done with an extracellular nuclease pro-
duced by Staphylococcus aureus. This nuclease cleaves phosphodiester bonds in
both DNA and RNA." Its amino acid sequence has been determined," and
its physical and enzymatic properties have been studied.“ Calcium ions have
been found to be essential in the binding to this nuclease of substrates or inhibi-
tory substrate analogues, such as deoxythymidine-3’,5’-diphosphate (pdTp).
Studies involving this inhibitor have suggested that the nucleotides fit into a
groove in the molecule that contains several tyrosyl groups, causing the exclusion
of water but only a small conformational change. Accompanying these effects
is a stabilization of the protein to the action of several proteolytic enzymes. At
alkaline pH, in the presence of calcium chloride and pdTp, trypsin digestion
produces three polypeptide fragments. The longer two of these, designated
Nase-T-p; and Nase-T-p; (44 and 100 amino acid residues, respectively), associate
to give a protein, nuclease-T, that has about 10 per cent of the enzymatic
activity of the original nuclease.'? We have used the measurement of tritium-
hydrogen exchange to study the effects of calcium chloride and pdTp binding
on nuclease and on the interactions of the Nase-T-p. and Nase-T-p; fragments.

We have observed significant suppression of tritium exchange after the addi-
tion of ligands to tritium-labeled nuclease. This tritium trapping may reflect
changes in protein motility’? and the direct shielding of parts of the protein by
the ligands, as well as changes in mean conformation. As discussed below, this
trapping effect may be analogous to the effect caused by corking a bottle.

Methods.—Tritiated water (100 mc/gm, lot no. 380-15) was obtained from the New
England Nuclear Corp. Sephadex G-25 (Pharmacia) was “‘fine’”’ grade. Deoxythymi-
dine-3',5’-diphosphate was purchased from Calbiochem.

1478
Vou. 61, 1968 BIOCHEMISTRY: SCHECHTER ET AL. 1479

Proteins: Staphylococcal nuclease (Nase) was isolated from the growth medium of
Staphylococcus aureus (Foggi strain).

Nuclease-T was prepared by limited digestion of Nase with trypsin (at pH 8.1) in the
presence of calcium chloride and pdTp, and was purified by chromatography on phos-
phorylated cellulose. Nase-T-p, and Nase-T-p; were resolved by gel filtration in 50'%
acetic acid on a Bio-Gel P-20 (3 < 200 cm) column. The Nase-T-p, and Nase-T-ps, as
characterized by amino acid analysis and reconstitution of nuclease activity upon recom-
bination, corresponded to the previously described preparations. ?

A spectrophotometric Nase assay, using denatured DNA as substrate, was employed."

Labeling: Stock solutions of tritium-labeled Nase were prepared by dissolving 6.0-7.0
mg of lyophilized Nase per ml of 0.05 M Tris-HCl buffer, pH 8.1. Five ul of tritiated
water (THO) was added to each milliliter of the Nase solution. Stock solutions of the
Nase-T-p: and Nase-T-p,, at 4 mg/ml and 9 mg/ml, respectively, were prepared with
the same buffer and concentration of tritium. The protein solutions were kept for several
days at 2°C to allow the establishment of tritium~hydrogen exchange equilibrium.

Gel filtration: In order to separate the labeled protein from the tritiated water, gel
filtration on Sephadex G-25 was employed as suggested originally by Englander.*| We
have used columns of 2 X 15 cm, or longer, specially designed for minimal dead space,
with flow rates of about 1 ml/min.

The labeled protein solution was equilibrated at room temperature before each experi-'
ment. Aliquots of 0.5 ml, to which 0.05 ml of 0.1 4 CaCl, and 0.05 ml of a pdTp solution
(3.5 mg/ml) was added, were used. If no ligand solution was added, the sample volume
was made to 0.6 ml with the Tris-HCI buffer. With these concentrations, the molar ratio
of pdTp to Nase was about 1.5 to 1. In studies of the effects of ligands, the equilibrating
buffer contained the indicated concentrations of ligands.

Fractions of 2.0 ml each were collected. As determined by measurements of radio-
activity, optical density at 280 mp, and enzyme activity, complete resolution of the la-
beled protein from the free tritiated water was achieved with good reproducibility. It
was found -that the column could be re-used after being washed with several hundred
milliliters of water. Recovery of protein from the Sephadex columns usually exceeded
90%. We assumed similar recoveries in experiments with added ligands where the pro-
tein absorbancy was masked.

Counting: A 1.0-ml aliquot from each fraction was added to 15 ml of Bray’s liquid
scintillator..7 Counting, for periods sufficient to give less than 5% variance, was done in
a, refrigerated Packard Tri-Carb liquid scintillation spectrometer, model 3214. Counts
per minute were converted to disintegrations per minute (dpm) after internal standardiza-
tion with toluenc-H’.

To determine the quantity of tritium per molecule of Nase, the disintegrations per
minute in the excluded peak were summed; in the studies with pdTp the disintegrations
per minute in the four tubes (five with the 40-cm column) containing the peak were
summed. These values are listed in Table 1. The extent of labeling under any given
condition was calculated from these sums, and the specific activity in the labeling mixture,
using the formula of Englander.‘

Results —Trapping of H*® by addition of pdTp and calcium chloride: When
labeled nuclease is separated from the tritiated water on a Sephadex G-25
column, one observes a peak of radioactivity corresponding to the distribution of
the protein, then radioactivity due to back-exchanging tritium atoms, and then
a large peak of free THO. Those hydrogens back-exchanging “instantaneously”
are obscured by this large peak of free THO.

Figure 1A shows the elution pattern of the gel filtration of 3.0 mg of nuclease on
a 2 X 40-em column of Sephadex G-25. The peak of radioactivity associated
with the protein corresponded to about five tritium atoms per molecule of pro-
tein, as shown in Table 1. The pattern shows that the major part of the tritium
1480 BIOCHEMISTRY: SCHECHTER ET AL. Proc. N. A. 8.

06

a
T

Fig. 1.—~-(A) Gel filtration of
nuclease in 0.05 M Tris-HCl, pH
is ° 8.1, on a Sephadex G-25 column

[ 8 (2 X 40 em).

(B) Gel filtration of nuclease
after addition of calcium chloride
and pdTp, using the same column
equilibrated with the Tris-HCl
buffer containing 0.01 1 CaCl, and
6 X 1074 M pdTp.

Optical density at 280 my
* (- -O- -O--) and disintegrations per
5b | minute (—@—@—) are plotted.

OPM xX 1073

 

9 1 T T
10 20 30 40
TUBE NUMBER

has been shed from the protein during its passage through the column. Figure
1B gives the elution pattern of an identical aliquot of nuclease to which calcium
chloride and pdTp were added ten minutes before application to the Sephadex
column. The Tris-HCl buffer on the column contained these two ligands at the
indicated concentrations. In contrast to the results summarized in Figure 1A,
much more of the radioactivity was now associated with the protein peak and the
amount back-exchanged during filtration was lower. The radioactivity in the
protein-containing fractions corresponded to about 35 H’ atoms per molecule of
protein.

The relative difference between the free and liganded proteins depends on the
duration of the gel filtration. Figure 2A shows the radioactivity pattern of the
gel filtration of an aliquot of nuclease on a 2 X 15-cm Sephadex G-25 column.
The summed disintegrations per minute are listed in Table 1. In four separate
experiments this value averaged 17,300 (+3000). Figure 2B shows the elution
pattern from the same column after addition of ligands to the aliquot of nuclease
and to the column buffer. The trapping of tritium by the ligands occurred but
the relative effect was smaller than in the first experiment because under these
conditions there is less time for exchange-out. The trapping effect of the ligands
must be very rapid compared to the separation occurring in the gel because a
very similar pattern was obtained when nuclease, without ligands, was applied to
a column which had been previously equilibrated with buffer containing both
ligands (Fig. 2). On the other hand, omission of the ligands from the column
buffer resulted in a substantial decrease in the radioactivity associated with
nuclease to which ligands had previously been added. This was presumably
due to dissociation of the protein-ligand complex occurring within the column.
Vou. 61, 1968 BIOCHEMISTRY: SCHECHTER ET AL. 1481

Tape }. Radioactivity associated with proicin after get filtration.

Ligands
Expt. Protein solution added x dpm* T/molecule*

lt Nuclease —_ 3,900 5
Nuclease Catt, pdTp 28,400 35

2 Nuclease _— 20, 300 27
Nuclease Catt, pdTp 41,900 57
Nuclease Catt, pdTpt 36,600 50

3 Nuclease —_ 14,600 18
Nuclease Catt, pdTp 37,800 46
Nuclease, Cat+, pdTp§ Cat*, pdTp 16,800 20

4 Nuclease, EDTA — 9,200 13
Nuclease, EDTA Cat* 19,000 27
Nuclease, EDTA pdTp 17,600 30

5 Nase-T-p. a 0 0
Nase-T-ps — 0 0
Nase-T-p2 Cat*+, pdTp 300 <0.1
Nase-T-ps Catt, pdTp 200 <O.1
Nase-T-p2, ~ ps — 3,800 3
Nase-T-p2, —Dps Catt, pdTp 21,400 18

* The sum of disintegrations per minute associated with the protein peak and the gram atoms of
tritium per mole of protein is calculated as described in Methods.

{ A 40-em column rather than 15 cm as in other experiments.

t Ligands were added to column buffer only.

§ Ligands were first added to nuclease before the addition of THO, again 10 min before application
to the column, and to the column buffer.

Inhibition of ligand effect: If calcium chloride and pdTp were added to the
nuclease solution before labeling with THO, trapping of tritium by the further
addition of these ligands to the radioactive protein did not occur, as illustrated in
Figure 3 and Table 1. In terms of our analogy, it is difficult to pour water into a
corked bottle. These results also exclude the possibility that the ligands them-
selves contribute significantly to the increase of radioactivity associated with the
protein complex.

Effect of caletum chloride or pdTp alone: Preliminary gel filtration studies
showed that calcium chloride alone increased the binding of tritium to nuclease
very slightly. However, pdTp alone gave most of the effect of both ligands
in combination. With nuclease previously treated with ethylenediaminetetra-
acetate (EDTA) (final concentration of 0.01 M), different results were obtained.
As summarized in Table 1, each of the ligands alone did cause some trapping of
tritium, but not the full effect that was seen with the ligands in combination. A
full understanding of the effects of ligands, individually, will require further study.

Tritium binding by nuclease-T fragments: We have applied the method de-
scribed above to Nase-T-p2 and Nase-T-p; and to nuclease-T in the presence
and absence of calcium ions and pdTp.

Gel filtration of the separated, tritium-labeled Nase-T-p. and Nase-T-ps
showed no radioactivity over background (Fig. 4B and C). When identical
samples of labeled Nase-T-p: and Nase-T-p3 were mixed before application
to the column, there was a significant increase in bound radioactivity, as
shown in Figure 4A. Of interest was the high level of radioactivity following the
polypeptide peak, denoting tritium atoms with back-exchange rates of minutes.
1482 BIOCHEMISTRY: SCHECHTER ET AL. Proc. N. A. 8.

As with intact nuclease, the addition of
ligands resulted in a dramatic increase of
bound tritium in the nucleasc-T (Tig.
4D). It is apparent that the exchange
rates of a large number of tritium atoms
have been appreciably slowed. The
radioactivity associated with the single
Nase-T-p2 and Nase-T-p; after the addi-
tion of these two ligands, listed in Table
1, was insignificantly above background.

Discussion.—The study of hydrogen
exchange in macromolecules has poten-
tial for describing dynamic aspects of
the behavior of molecules in solution,
but present theory and available exper-
imental data do not yet allow a com-
plete interpretation of the kinetics of hy-
drogen exchange with respect to the de-
tails of macromolecular conformation.?:18
However, effects on the kineties of hy-

. 1 _ drogen exchange have been reported for
ons i Teticl pH a1 on a. ‘Sephadex a number of protein-ligand systems,*~°
G-25 column (2 X 15 em). (B) Gel filtra- and have been interpreted as evidence
tion of nuclease after addition of calcium for induced conformational changes in
ae Teer thetigen rien O. the protein components. DeLuca and
both using the same column equilibrated Marsh’ found that the addition of sub-
with the Tris-HCl ater connate nt uM strates to luciferase, after separation
Fat ce disintogrations : er mninnte. nes from free THO, decreased the exchange-

out rate of many hydrogens.

In our studies of such interactions, proteins were tritiated in the ‘‘relaxed”’
state and then separated from free THO by gel filtration at relatively low flow
rate. Radioactivity associated with the protein peak thus represents tritium
atoms with ‘‘intermediate” or ‘‘slow” exchange rates (see refs. 2 and 4 for the
classifications); the tritium atoms in “instantaneous” and “fast’’ classes ex-
change-out during the 15 minutes or more required for gel filtration. The
interactions studied here result in a marked increase in the class of measurably
exchanging hydrogens. Thus, under conditions of defined temperature, ionic
strength, pH, and time, the specific radioactivity of one protein peak is a useful
parameter of molecular associations.

The very few tritium atoms in unliganded nuclease (Table 1) suggest that most
of the labile hydrogens are accessible to solvent, presumably the result of
relatively loose folding or of great motility!’ in the protein. The addition of
calcium chloride and pdTp markedly increased the number of tritium atoms in
the “intermediate” and ‘‘slow”’ exchanging classes. This could be caused by a
shielding of part of the protein molecule from the solvent by the ligands, by a
decrease in motility, or by a combination of these factors. Although other

wn

30

DPM x 1073

20°

 

 

 

TUBE NUMBER
VoL. 61, 1968 BIOCHEMISTRY: SCHECHTER ET AL. 1483

fA 06
10 - pos
A ¢ 7
i ‘
, . . i 4 03
Fic. 3—(A) Gel filtration of ivy \ 7 8
nuclease in 0.05 M Tris-HCl, pH } Ss
8.1, on a Sephadex G-25 column o | Se=Ossiestesteet plese O 0

 

(2 X 15 cm). (B) Gel filtration of
nuclease after addition of calcium
chloride and pdTp (—e—e—), 307
and nuclease to which these ligands
had been added before labeling
and at the time of gel filtration
(--O--O-); both run on the same
column, equilibrated with the Tris-
HCl buffer containing 0.01 M
CaCh and 6 * 1074 M pdTp. In
(A), optical density at 280 my
(--O- -O--) and disintegrations per
minute (—@®—@—) are indicated;
in (8), both lines refer to disin-
tegrations per minute.

ppm x 1073

207

   

 

 

|_ era -8-0-nea a! 1 yp
© 5 10 15
TUBE NUMBER

studies have suggested that the major effect of these ligands is expressed in a
crevice region, they might not have detected a ligand-induced change in motility.

The addition of ligands before the addition of THO prevented an increase in
tritium binding upon further addition of ligands to the labeled complex. This
can be explained by assuming that during the time of this experiment the presence
of the ligands excluded from labeling those hydrogens-responsible for the differ-
ences seen during the trapping phenomenon.

The effects on tritium exchange of calcium chloride and pdTp used separately
are most easily explained on the assumption, in accord with the experience of
others,’ that some calcium is ordinarily bound to the protein preparations.
After EDTA treatment of nuclease, the accessibility of the protein’s hydrogens
to the solvent is increased. Addition of calcium chloride or pdTp alone in-
creased the bound radioactivity but to a much lower extent than with the ligands
in combination.

The absence of tritium atoms associated with the gel-filtered Nase-T-p2 and
Nase-T-ps3 indicates the ready accessibility of all of the exchangeable hydrogens
to solvent. Complete exchange-out would be expected in a random-coil poly-
peptide.? The fact that tritium remains with the mixed fragments after they are
gel-filtered is physical evidence for their interaction in solution and the establish-
ment of a solvent-excluding conformation. This finding is compatible with the
restoration of partial enzyme activity and changes in optical rotatory dispersion!”
and circular dichroism” measurements upon mixing the fragments. The absence
of tritium trapping after the addition of the ligands to the individual fragments
suggests weak or no binding but does not exclude the possibility of binding of
these ligands unaccompanied by changes in structure or motility. Conversely,
1484 BIOCHEMISTRY: SCHECHTER ET AL. Proc. N. ALS.

 

 

 

l2- D
itr
iOF \
gt
gk
> 2 ay
a 2s 2
2 2 zef
bod =~ s 5f
z e
g 4}
z 3 0.6
z >
27 0.4
a
iF ~ “ 02 >
? Pee No 0 =
0 5 10 15 0 5 10 (5
TUBE NUMBER TUBE NUMBER

Fic. 4.—(A) Gel filtration of the equimolar mixture of Nase-T-p, and Nase-T-p, on a Se-
phadex G-25 column (2 X 15 cm), in 0.05 M Tris-HCl, pH 8.1.

(B) and (C) represent the gel filtration of 2.3 mg of Nase-T-p3 and 1 mg of Nase-T-p,, respec-
tively, on the same column and in the same buffer.

(D) shows the results of the gel filtration of the equimolar mixture of Nase-T-p: and Nase-T-p;
to which calcium chloride and pdTp had been added before application to the same column,
previously equilibrated with the Tris-HCl buffer containing 0.01 M CaCl and 6 x 10-4 M
pdTp.

Optical density at 280 my (- -O- -O- -), disintegrations per minute (—e—@—), aud change
in absorbancy at 260 mu in nuclease assay’® (--O-- +) are indicated.

when ligands are added to a mixture of Nase-T-p: and Nase-T-ps, there is a six-
fold increase in the trapping of tritium in comparison to that seen with the mixed
fragments without ligands. The amount of bound tritium in the nuclease-T
preparations is less than that in intact nuclease, suggesting that the covalent
change has made the structures more accessible to solvent.

Since gel filtration methods exclude the determination of the “instantaneously”
exchanging hydrogens, the advantage of labeling in a relaxed state and then
measuring the tritium trapped (at zero time of exchange-out) in the ligand-
stabilized form should be noted. Thus, molecules may be labeled to different de-
grees at zero time of exchange-out but the kinetics of their measurably exchanging
hydrogens may be very similar. Such effects could explain our findings, using
the interrupted-flow gel filtration technique, that the sizes of the several classes
of exchangeable hydrogens in the conformationally dissimilar?! myoglobin and
apomyoglobin are similar, yet the addition of hematin to the labeled apoprotein
causes tritium trapping.??. The ligand effects shown in Figure 3 also illustrate
these points.

Under the standard conditions of our experiments, the average trapping caused
by calcium chloride and pdTp raises the number of tritium atoms bound per
molecule of nuclease from about 23 in the free form to about 52 in the presence of
the ligands. It is unlikely that this difference results solely from stabilization of
helices. The helix content as determined by optical rotation and circular di-
chroism, which docs not change with the addition of ligands, would account for
Vou. 61, 1968 BIOCHEMISTRY: SCHECHTER ET AL. 1485

only about 27 slowly exchanging hydrogens.’ The direct shielding of the mole-
cule from solvent, or a change in motility caused by ligands, must thus be con-
sidered.

Summary.—The addition of the ligands, calcium chloride and deoxythymidine-
3’,5’-diphosphate, to a staphylococcal nuclease previously labeled with tritiated
water resulted in a marked decrease in the exchange-out rate of some tritium
atoms. ‘The mixing of two associating polypeptide fragments derived from this
protein resulted in binding of tritium that was not observed in the separated
fragments. The trapping of tritium demonstrated by this method may be a
sensitive and simple probe for various interactions of macromolecules.

We thank Dr. Hiroshi Taniuchi for his advice and help in the preparation of nuclease-T.

* Special postdoctoral fellow, National Institute of Arthritis and Metabolic Diseases, U.S.
Public Health Service.

} Visiting scientist on leave from the Institute of Organic Chemistry and Biochemistry,
Czechoslovak Academy of Sciences, Prague, Czechoslovakia.
a= | Hvidt, Aa., and K. Linderstrgm-Lang, Biochim. Biophys. Acta, 14, 574 (1954).

2 Hvidt, Aa, and S. O. Nielsen, in Advances in Protein Chemistry (New York: Academic
Press, 1966), vol. 21, p. 287.

3 Leach, S. J., and P. H. Springell, Australian J. Chem., 15, 350 (1962).

4 Englander, 8. W., Biochemistry, 2, 798 (1963).

’ Hvidt, Aa., J. H. R. Kagi, and M. Ottesen, Biochim. Biophys. Acta, 75, 290 (1963).

6 DiSabato, G., and M. Ottesen, Biochemistry, 4, 422 (1965).

7 DeLuca, M., and M. Marsh, Arch. Biochem. Biophys., 121, 233 (1967).

8 Praissman, M., and J. A. Rupley, Biochemistry, 7, 2431, 2446 (1968).

9 Ulmer, D. D., and H. R. Kagi, Biochemistry, 7, 2710, 2718 (1968).

10 Cuatrecasas, P., H. Taniuchi, C. B. Anfinsen, and D. Ontjes, in Brookhaven Symposia in
Biology, in press.

4 Taniuchi, H., C. B. Anfinsen, and A. Sodja, J. Biol. Chem., 242, 4752 (1967).

' Taniuchi, H., C. B. Anfiusen, and A. Sodja, these Procnipinas, 58, 1235 (1967).
~ 13 We use the term mofility to indicate the possibility of rapid, small reversible chauges in
the conformation of all or parts of protein molecules in solution, as discussed originally by K, U.
Linderstrgm-Lang and J. A. Schellman, in The Enzymes, ed. P. D. Boyer, H. Lardy, and K.
Myrback (New York: Academic Press, 1959), 2nd ed., vol. 1, pp. 443-510. In this sense,
“a protein cannot be said to have ‘a’ secondary structure but exists mainly as a group of
structures not too different from one another in free energy, but frequently differing consider-
ably in energy and entropy.”

14 Mordvek, L., C. B. Anfinsen, J. L. Cone, and H. Taniuchi, J. Biol. Chem., in press.

% Fuchs, 8., P. Cuatrecasas, and C. B. Anfinsen, J. Biol. Chem., 242, 4768 (1967).

16 Cuatrecasas, P., 8. Fuchs, and C. B. Anfinsen, J. Biol. Chem., 242, 1541 (1967).

7 Bray, G. A., Anal. Biochem., 1, 279 (1960).

18 [englander, S. W., in Poly-a-Amino Acids, ed. G. D. Fasman (New York: Dekker, 1967),
pp. 339-367.

19 Cuatrecasas, P., personal communication.

2 Taniuchi, H., and C. B. Anfinsen, J. Biol. Chem., 243, 4778 (1968).

21 Schechter, A. N., and C. J. Epstein, J. Mol. Biol., 35, 567 (1968).

22 Mordvek, L., A. N. Schechter, and C. B. Anfinsen, in preparation.