Myristyl Acylation of the Tumor Necrosis Factor ce
Precursor on Specific Lysine Residues
By Frazier T. Stevenson,* Stuart L. Bursten,~ Richard M. Locksley, g
and David H. Lovett*
From "The Medical Service, San Francisco Veterans Administration Medical Center, San
Francisco, California 94121; the *Department of Medicine, University of Washington~Seattle
Veterans Administration Medical Center, Seattle, Washington 98108; and the SDepartment of
Medicine, University of California, San Francisco, San Francisco, California 94143
NH2-terminal glycine myristyl acylation is a cotranslational modification that affects both protein
localization and function. However, several proteins that lack NH2-terminal glycine residues,
including the interleukin 1 (Ibl) precursors, also contain covalently linked myristate. To date,
the site(s) of acylation of these proteins has not been determined. During an evaluation of Ibl
acylation, it was observed that [3H]myristate-labded human monocyte lysates contained a
prominent 26-kD myristylated protein, which was identified as the tumor necrosis factor ce (TNF)
precursor protein on the basis of specific immune precipitation. Radioimmunoprecipitates from
the supernates of labeled monocytes indicated that the processed or mature 17-kD form of TNF
does not contain myristate, suggesting that the site of acylation occurs within the 76-amino
acid propiece of the precursor molecule. As the TNF precursor does not contain an NH2-
terminal glycine, we hypothesized that myristyl acylation occurs on the N-e-NH2 groups of
lysine, of which two are present in the propiece (K19K20). Synthetic peptides were designed to
include all seven lysine residues present within the entire 26-kD TNF precursor, and used in
an in vitro myristyl acylation assay containing peptide, myristyl-CoA, and monocyte lysate as
a source of enzyme. Analysis of reaction products by reverse phase high performance liquid
chromatography and gas phase sequencing demonstrated the exclusive myristyl acylation of K19
and K20, consistent with the presence in monocytes of a specific lysyl N-e-NH2-myristyl
transferase activity. The acylated lysine residues are located immediately downstream from a
hydrophobic, probable membrane-spanning segment of the propiece. Specific myristyl acylation
of the TNF propiece may facilitate membrane insertion or anchoring of this critical inflammatory
mong the many modifications of newly synthesized pro-
teins, cotranslational acylation with myristic acid has
received considerable attention as an important determinant
of protein function and intracellular localization (for review,
see reference 1). For most myristylated proteins studied thus
far, acyhtion occurs via the formation of an amide bond linking
the fatty acid to an NHz-terminal glycine residue after the
removal of the initiator methionine. This process has been
weU characterized and the enzyme responsible, myristyl CoA:
protein N-myristyl transferase (NMT), 1 has been cloned (2).
However, in a few cases myristylated proteins have been
identified that lack the correctly positioned NHz-terminal
glycine strictly required for acylation by NMT. These pro-
teins include the insulin receptor, the/~ Ig heavy chain, and
1 Abbreviation used in this paper: NMT, N-myristyl transferase.
the Ibl oc and B precursors (3-5). All of these proteins are
myristylated by an undescribed enzymatic mechanism that
does not involve acylation on NH2-terminal glycines. One
potential alternative mechanism for myristyl acylation would
be the myristylation of internal lysine residues, using the free
e-amino groups to form the characteristic amide bonds. Ac-
ylation of internal lysine residues with long chain fatty acids
has been shown to enhance binding of pancreatic phospholi-
pase A2 to its substrate (6, 7), but as yet, the myristylation
of internal lysine residues as a discrete, cotranslational pro-
tein modification has not been demonstrated.
TNF is a cytokine active in mediating cachexia, tumor
regression, septic shock, autoimmunity, and complications
in infections such as HIV-1, cerebral malaria, and bacterial
meningitis (reviewed in reference 8). TNF is translated as a
26-kD precursor molecule that is subsequently processed by
unclear mechanisms to an extracellularly active, 17-kD ma-
1053 The Journal of Experimental Medicine i Volume 176 October 1992 1053-1062
ture protein (9). In addition, the 26-kD TNF precursor pro-
rein can act as a plasma membrane-associated protein that
mediates inflammation by direct cell-to-cell contact (10, 11).
We report here that 26-kD TNF is myristylated and that
the acylation occurs via amide bond formation with two
specific e-amino groups of internal lysine residues present
within the 76-amino acid propiece of the molecule. Myristyl
acylation within this region may facilitate the membrane lo-
calization or insertion of the precursor molecule.
Materials and Methods
Reagents. Lipid A, purified from the Salmonella minnesota lk595
strain, was obtained from Ribi Immunochem (Hamilton, MT).
The lipid A was prepared as a stock solution of 1 #g/ml in RPMI
1640 supplemented with 0.1% defatted BSA. Immediately before
use, the stock solution of lipid A was briefly sonicated on ice. Pan-
sorbin (fixed protein A-bearing Staphylococcus aurens) for radioim-
munoprecipitation was obtained from Calbiochem-Behring Corp.
(La Jolla, CA). Media and heat-inactivated FCS were from Gibco
Laboratories (Grand Island, NY). [3sS]Methionine (1,300 Ci/
mmol), [3SS]cysteine (800 Ci/mmol), and [3H]myristate (22
Ci/mmol) were obtained from New England Nuclear (Boston,
MA). The murine monoclonal anti-human TNF antibody TNF-E
was obtained from Genentech Inc. (S. San Francisco, CA). This
IgG1 antibody had a neutralization titer of >5 x 10 s U/ml and
an endotoxin concentration of 4 EU/ml by Limulus antilipopolysac-
charide assay. Human recombinant TNF was obtained from Gen-
zyme (Cambridge, MA).
Preparation of Cells and Cytosol. Heparinized blood from normal
donors was diluted 1:1 with PBS, pH 7.2, before separation over
Ficoll-Hypaque cushions (Sigma Chemical Co., St. Louis, MO)
by centrifugation for 15 min at 2,000 g. The mononuclear cell in-
terface was collected, washed three times in calcium-free PBS, and
distributed to 100-ram plastic dishes at a concentration of 5 x
106 cells/ml in RPMI 1640 containing 5% FCS. After incubation
at 37~ in 5% CO2 for 2 h, nonadherent cells were removed by
vigorous washing.' The adherent monolayers consisted of >95%
monocytes as assessed by nonspecific esterase staining and phago-
cytosis of zymosan. For preparation of crude cytosolic fractions,
monocytes were suspended in TE buffer (10 mM TRIS-C1, pH
7.6, 1 mM EDTA) containing protease inhibitors (5 mM EDTA,
0.2 mM PMSF, 2/~M pepstatin), and subjected to three cycles of
freeze-thawing at -80~ Unlysed cells and nuclei were removed
by centrifugation at 2,000 g and lysates were stored at -80~
Experimental Protocol. For labeling with [3SS]methionine or
[3sS]cysteine, washed layers of monocytes were incubated for 60
min in methionine- or cysteine-deficient medium. The medium was
removed and replaced with fresh medium containing 100/~Ci/ml
of labded methionine or cysteine as indicated. Experimental groups
were stimulated for 4 h with 100 ng/ml lipid A. After incubation,
the culture supernatants were removed, centrifuged at 400 g for
10 min, supplemented with protease inhibitors as above, and frozen
at -80~ Monocyte layers were washed twice with cold PBS,
scraped into microfuge tubes, and centrifuged at 10,000 g for 5
min. The cell pellets were washed an additional two times with
cold PBS before preparation for electrophoretic analysis or immune
precipitation. For labeling with [3H]myristate, the fatty acid was
dried under argon and brought into solution at 25/zCi/ml by soni-
cation in medium supplemented with 0.1% defatted BSA. After
incubation, the cell layers and supernatants were processed as above.
Immune Precipitation. Cell pellets from [3H]myristate-labeled
human monocytes were lysed in a equal volume of 2 x lysis buffer
consisting of (final concentrations) 2% NP-40, 0.15 M NaC1, 10
mM NaH2PO4, pH 7.2, 0.68 M sucrose, and protease inhibitors
as above. The lysates were incubated on ice for 30 min, sonicated
with three 10-s bursts on ice at a power setting of five (Bronson
Ultrasonics, Danbury, CT), and centrifuged at 10,000g for 15 min.
For immune precipitation of radiolabeled cellular TNF, the cell 1)-
sates were precleared by incubation at 4~ overnight with 10%
(vol/vol) Pansorbin. The cleared lysates were then incubated for
18 h at 4~ with 2.5/~g/ml of murine monoclonal anti-TNF IgG.
TNF antigen-antibody binding was competed by addition of purified
recombinant human TNF in concentrations from 0.1 to 1/~g/ml.
Nonimmune routine IgG was used as a negative control. After in-
cubation, 10% Pansorbin was added and the bound immune com-
plexes were recovered by centrifugation. The pellets were washed
five times in large volumes of wash buffer containing 0.5 % NP-40,
0.45 M NaC1, and 50 mM Tris/HC1, pH 8.3. The pellets were
resuspended in IEF sample buffer (9.3 M urea, 5 mM K2CO3, 2%
NP-40, 2% 3-10 ampholines, and 30 mM DTT). After a 2-h incu-
bation at 4~ the samples were centrifuged and analyzed by two-
dimensional electrophoresis. The proteins were separated in the first
dimension on 0.4 mm ultrathin polyacrylamide gels according to
Goldsmith et al. (12), using 1.4% 4-6 ampholines and 0.6% 3-10
ampholines (LKB Instruments, Inc., Bromma, Sweden). Focused
proteins were separated in the second dimension on 12.5% discon-
tinuous gels according to O'Farrell (13). After electrophoresis, the
gels were fixed in 10% TCA/30% ethanol for 45 min, washed
twice in 5% TCA/30% ethanol, and soaked in Amplify (Amer-
sham Corp., Arlington Heights, IL). Dried gels were analyzed by
autoradiography at -800C using preflashed Kodak X-O-Mat film
with intensifying screens. The isoelectric points given were deter-
mined by comparison with stained protein standards (Pharmacia
LKB, Uppsala, Sweden). Isoelectric points given are not corrected
for the inhibitory effects of the 9.2 M urea on hydrogen ion ac-
tivity and should not be considered identical to isoelectric points
obtained under physiologic (i.e., aqueous) conditions (14, 15). Im-
mune precipitation of biosynthetically labeled TNF from the mono-
cyte culture supernatants was performed in an identical fashion and
analyzed by one-dimensional electrophoresis on 12.5% SDS-PAGE
gels, followed by fixation and autoradiography as above.
Analysis of Acylated Proteins. Recombinant human TNF (10/zg)
was added as a carrier protein to [3H]myristate-labeled TNF im-
mune precipitants and separated by SDS-PAGE on 12.5% gels. The
TNF protein was identified by Coomassie blue staining, cut out,
electroeluted from the gel slice, and concentrated by lyophiliza-
tion. The samples were subjected to acid methanolysis by heating
to 110~ for 60 h in 83% methanol/2 M HC1, containing 200
/~g each of myristic and palmitic acid. The reaction products were
extracted three times with petroleum ether, and 400 #g each of
methyl myristate and methyl palmitate was added. The samples
were evaporated under argon, resuspended in methanol, and
identified by analytic HPLC on a 4.6 x 25-mm ODS-5 column
(Bio-Rad Laboratories, Richmond, CA). The column was devel-
oped with 80% (vol/vol) acetonitrile (ACN)/0.1% TFA/0.06%
triethylamine (TEA) at a flow rate of I ml/min. Serial fractions
were collected, and the radioactivity was quantitated by liquid scin-
tillation counting. The dution profile of the radioactivity was com-
pared with the absorbance (210 nm) elution profiles of standard
palmitic and myristic acids and the respective methyl esters.
In Vitro Aoflation of Synthetic Peptides. A series of synthetic pep-
tides containing potentially reactive lysine residues (Table 1) were
prepared and purified by reverse-phase HPLC. Chemically my-
1054 Myristyl Acylation of Tumor Necrosis Factor
ristylated standards were prepared by reaction of the synthetic pep-
tides with the symmetric anhydride of myristic acid, according to
Towler and Glaser (16). The chemically acylated standard peptides
were treated with 1 M hydroxylamine to cleave any ester-linked
fatty acid, extracted with petroleum ether, and analyzed by reverse-
phase HPLC and gas-phase sequencing (see below).
The enzymatic myristyl acylation of synthetic peptides was based
on the method of Towler and Glaser (16). In brief, myristyl CoA
was prepared by reacting 5 nmol myristic acid with 10 nmol LiCoA,
in an acylation buffer containing 10 mM Tris/HCl, pH 7.4, 0.1
mM EDTA, 1 mM DTT, 5 mM MgCI2 and 250 nM ATP. There-
after, 15 mU Pseudomonas CoA synthetase (Sigma Chemical Co.)
was added and the mixture incubated for 30 rain at 30~ in a final
reaction volume of 50/~1. To this was subsequently added 10 nmol
of synthetic peptide and 50/~g of monocyte cellular lysate in a buffer
of 10 mM Tris/HC1, pH 7.4, 0.1 mM EDTA, and 1 mM DTT.
A battery of protease inhibitors (8 #M leupeptin, 1 mM PMS-F,
and 10/~g/ml pepstatin) was added and the reaction volume was
brought to 110/~1. The enzymatic acylation of the synthetic pep-
tides was continued for 10 min at 30~ followed by the addition
of 110/~1 methanol and 10/~1 saturated TCA. This mixture was
incubated on ice for 10 min to precipitate cellular proteins, cen-
trifuged for 10 rain at 10,000 g, and the supernates (containing
the synthetic peptides) were extracted with petroleum ether three
times to remove unreacted myristic acid before analysis by reverse-
phase HPLC. Standard and chemically acylated synthetic peptides
were used to calibrate a 4.6 x 250-mm C4 RP304 (Bio-Rad Labora-
tories) reverse-phase HPI.C column using a linear gradient of ACN
(1%/rain) in 0.1% TFA. The column eluates were monitored at
To confirm the sites of peptide myristyl acylation after the enzy-
matic reaction, gas phase sequencing was performed. As myristyhted
lysine residues are hydrophobic and elute from the sequencer at
a higher solvent concentration than nonacylated residues, it is pos-
sible to localize within the peptide sequence the actual site of ac-
ylation (as determined by the reduced yield of nonderivatized amino
acid). In addition to this indirect determination, a fraction of the
products from the gas-phase sequencer was diverted during each
cycle and directly analyzed for N-e-NH2-myristyl lysine content,
using a quantitative HPLC assay. For this assay, standard N-e-
NH2-myristyl lysine was prepared by reaction of N-c~-BOC-lysine
(Sigma Chemical Co.) with the symmetric anhydride of myristic
acid as above, followed by removal of the N-cc-BOC protecting
group by TFA hydrolysis. Standard N-e-NHz-myristyl lysine was
used to calibrate a 150 x 2.l-ram ODS-222 column (Brownlee
Labs, Santa Clara, CA), using a linear gradient of ACN (1%/rain)
in a buffer consisting of water/0.1% TFA/0.06% TEA. The column
Table 1. Synthetic Peptides for TNF In Vitro Myristylation Assay
E14 E ALKK
tLs3 T P S DK P VAH92
Y136 S QVL F K GQG145
Glss AE AK PWYEP194
L L S A I K S P177
The amino acid sequences of the five synthetic peptides designed to span
all lysine residues (boldface) in the TNF precursor are shown.
elutate was monitored at 214 nm; N-e-NH2-myristyl lysine elutes
at 51% ACN.
Prior studies from our laboratories have documented the
cotranslational myristyl acylation of the intracellular mono-
cyte II.-1 o~ and 3 precursor proteins (5). Two-dimensional
autoradiograms revealed that multiple intracellular monocyte
proteins are myristyl acyhted, in addition to the II:l precursors.
As shown in Fig. 1 A, incubation of freshly isolated human
monocytes with [3H]myristate for 4 h leads to the labding
of a prominent 25-26-kD protein with an apparent pI of
"~6.3. Stimulation of the monocytes for 4 h with 100 ng/ml
lipid A before electrophoretic analysis resulted in a three- to
fourfold increase in [3H]myristate labeling of the 25-26-kD
protein, as determined by quantitative densitometry (Fig. 1
B). The myristyl labeling of this particular protein is a cotrans-
lational, or rapid posttranslational event, as no myristyl ra-
diolabding occurred when the incubations were performed
in the presence of 10 #g/ml cycloheximide (Fig. 1 C). Con-
cordant labeling of the 25-26-kD protein with [3H]myri-
state and [3SS]methionine was readily apparent using incu-
bation times as short as 1 h (not shown).
Studies by Panuska and colleagues (15), using a similar two-
dimensional dectrophoresis protocol, have identified a 25-26-
kD, pI 6.3 protein present within [3SS]methionine-labded
human monocytes as the TNF precursor protein, based on
specific immune precipitations. This observation suggested
that the prominent 25-26-kD myristate-labeled protein ob-
served in our experiments represented the precursor of TNF.
This was confirmed, as shown in Fig. 2 A. Using a murine
monoclonal anti-human TNF antibody, it was possible to
specifically recover the [3H]myristate-labeled 25-26-kD, pI
6.3 protein from monocyte lysates. This recovery was specific
and was completely inhibited by competition experiments
using an excess of nonlabeled recombinant TNF protein (not
shown). We next determined whether the secreted, or
processed, 17-kD form of TNF is myristyl acylated. Freshly
isolated monocytes were stimulated for 4 h with 100 ng/ml
lipid A, during which time radiolabeling was accomplished
with either [3SS]cysteine or [3H]myristate. The culture su-
pernates were subjected to immune precipitation using the
monoclonal anti-TNF antibody and the products analyzed
by one-dimensional SDS-PAGE and autoradiography. A
[3SS]cysteine-labeled, 17-kD protein was readily immune
predpitated from the supernates of stimulated monocytes using
the monoclonal TNF antibody (Fig. 2 B, lane I). The mo-
lecular mass of this protein is consistent with its identification
as the mature, or processed, extracellular form of TNF. In
contrast, it was not possible to recover from such supernates
a [3H]myristateqabeled 17-kD protein (Fig. 2 B, lane 2). Ex-
posure of such material for periods of 3 mo and greater did
not reveal any significant radioactivity in these fractions. These
experiments suggest that myristyl acylation is restricted to
a component of the 25-26-kD TNF precursor that is removed
during the process of secretion.
To confirm that the radiohbel present on the TNF precursor
1055 Stevenson et al.
Figure 1. Two-dimensional autoradiograms of human
monocyte l)sates labeled with [3H]myristate. (A) Lysates
from freshly isolated monocytes after a 4-h incubation with
[3H]myristate. The prominent 26-kD protein (denoted
with an asterisk) falls within the pI 6.5-6.0 range. (B)
Lysates from monocytes incubated for 4 h with 100 ng/ml
lipid A in the presence of [3H]myristate. (C) L)sates from
monocytes incubated for 4 h with 100 ng/ml lipid A,
[3H]myristate, and 10/~g/ml cycloheximide.
protein represented intact, covalently linked myristate, acid
methanolysis of the isolated [3H]myristate-labeled protein
was performed. After recovery of labeled TNF from SDS-
PAGE gels, covalently bound fatty acid was hydrolyzed in
acid/methanol to form methyl esters suitable for HPLC
identification and quantitation. The resultant HPLC separa-
tion of the recovered TNF-bound fatty acid is shown in Fig.
3. Approximately 85% of the radioactivity was recovered with
the myristyl and methyl myristate fractions. A small amount
of radioactivity was identified as either palmitate or methyl
palmitate. The myristyl-labeled TNF protein was resistant
to hydrolysis with 1 M hydroxylamine (not shown), com-
patible with linkage in the amide as opposed to thioester form.
These data suggested that myristic acid is covalently linked
to the NHrterminal propiece of the 26-kD TNF precursor
by an amide bond. Nearly all myristylated proteins studied
to date have been found to be acylated on NH~.-terminal gly-
cine residues. Examination of the amino acid sequence of TNF
did not reveal a glycine in position no. 2 that could function
as a substrate for NH2-terminal myristylation. We therefore
considered the possibility that myristylation occurs at an avail-
able internal amino group, i.e., the e-NHz side group of ly-
sine. The 26-kD TNF precursor contains a total of seven ly-
sine residues. Only two of these (K19K20) are contained
within the 76-amino acid NH2-terminal propiece, which we
considered the likely site of acylation. To determine which
of these seven lysines were myristylated, we used an in vitro
assay developed by Towler and Glaser (16) for the character-
ization of the glycine specific, N-myristyl transferase. In this
assay, synthetic peptides containing potentially reactive amino
acids were combined with myristyl-CoA and a source of
acyltransferase (usually a cell lysate). Cellular proteins were
Figure 2. (A) Two-dimensional autoradiograrn of [3H]myristate-hbeled
monocyte lysates after immune precipitation with monoelonal anti-TNF
IgG. Specifically recovered is the 26-kD, pI 6.5-6.0 TNF precursor pro-
tein. (B) One-dimcusional autoradiogram of [3sS]c)steine-labeled mono-
cyte supernates after immune precipitation with monoclonal anti-TNF IgG
(lane 1) with recovery of the 17-kD extracelhlar TNF protein. A radiola-
beled 17-kD TNF protein was not recovered from the supernates of mono-
cytes incubated with [3H]myristate (lane 2).
A I / A
l 'l ?9
......,...,1'\ ll., Ill ,,/\
5 I 0 15 20 25 30 35 40
Figure 3. Acid methanol)sis of [3H]myristate-labeled TNE Radiola-
beled 26-kD TNF precursor protein was recovered as detailed in Materials
and Methods, and subjected to acid methanolysis. The resulting methyl
esters and fatty acids were separated by analytic reverse-phase HPLC, and
the radioactivity in each fraction was quantified. The bulk (85%) of the
radioactivity was recovered as myristate and methyl myristate.
1056 Myristyl Acylation of Tumor Necrosis Factor
Figure 4. In vitro acylation ofTNF peptide
no. 1. (A) Elution pattern of TNF peptide no.
1 (filled arrow) by reverse-phase HPI.C. (B) Elu-
tion pattern of chemically bis-acylated TNF
peptide no. 1 (open arrow), which is more hy-
drophobic than the unmodified peptide (filled
arrow). (C) Elution pattern of enzymatic, ally
acylated TNF peptide no. 1, revealing the for-
mation of mono- and bis-acylated TNF pep-
tides (open arrows).
precipitated with TCA/MeOH, free myristic acid was ex-
tracted with petroleum ether, and the resultant supernatants
were analyzed by reverse-phase HPLC. Myristylated peptides
are more hydrophobic and elute later than their unmodified
forms. Enzymatically myristylated peptides were identified
by comparison with chemically myristylated standards, which
had been made by reacting each peptide with the symmetric
anhydride of myristic acid. For the TNF analysis, five syn-
thetic peptides of 10-15-amino acid length, which spanned
all the lysine-containing sequences in the 26-kD TNF
precursor, were synthesized (see Table 1) and evaluated in the
above assay. As a source of a potential N-e-NH2-myristyl
transferase, cell lysates from LPS-stimulated human mono-
cytes were used. The reaction mixtures were then analyzed
1057 Stevenson eta|.
Figure 5. Results of the enzymatic in vitro acylation of TNF peptides 2-5 (A-D, respectively). In no case was an acylated end-product detected.
for the presence of enzymatically myristylated peptides by
reverse-phase HPLC. In Fig. 4 A, the elution profile of TNF
peptide no. 1 is shown. Chemical acylation of this peptide,
which contains two contiguous lysine residues, yielded a later
eluting peak (Fig. 4 B), which was shown by gas-phase se-
quencing and myristyl-lysine determination to consist entirely
of bis-acylated peptide. Fig. 4 C shows the HPLC analysis
ofTNF peptide no. 1 after reaction in the enzymatic myristy-
lation assay. The elution profile shows the unreacted peptide
peak, as well as a doublet that coelutes with the myristyl
peptide standard. This doublet represents the mono- and bis-
acylated forms of the peptide (see below). In contrast to the
results obtained with TNF peptide no. 1, it was not possible
to demonstrate any enzymatic acylation of TNF peptides nos.
2-5 (Fig. 5). To unambiguously confirm the myristyl acyla-
tion of the TNF peptide no. 1, gas phase sequencing and
direct quantitation of N-e-myristyl lysine was performed on
the later eluting, bis-acylated product. As shown in Fig. 6,
the yields of nonderivatized lysine in cycles 6 and 7 were
significantly decreased, consistent with the conversion of the
lysines to the acylated forms. Direct quantitation of N-e-
NH2-myristyl lysine confirmed the presence of the deriva-
tized forms in sequencing cycles 6 and 7 (Fig. 7). Analysis
of the mono-acylated product of the enzymatic reaction in-
dicated the preferential acylation of the second lysine residue
(K20; not shown). A time kinetic analysis of the enzymatic
acylation of TNF peptide no. 1 is shown in Fig. 8, which
demonstrates the rapid synthesis of the bis-acylated end-
product. The reaction rate is significantly blunted by 5 rain,
at which time nearly 3 nmol of end-product have accumulated.
In this paper the specific enzymatic myristylation of two
lysine residues contained within the 76-amino acid propiece
of the 26-kD TNF precursor protein has been demonstrated.
In addition, the studies utilizing synthetic peptides provide
evidence for the existence of a lysyl peptide N-e-NH2-
1 O0 t I I
0 5 10 15
Figure 6. Recovery of nonderivatized lysine by gas-phase sequencing
of enzymatically acylated (bis) TNF peptide no. 1. The yield of nonderiva-
tized (i.e., nonacylated) lysines is decreased due to the conversion to acylated,
hydrophobic forms. Although the sequencing was performed on HPI.C-
purified bis-acylated TNF peptide no. 1, there is a partial conversion of
the acylated lysine residues during the sequencing reaction (hydrolysis) to
the nonacylated forms.
1058 Myristyl Acylation of Tumor Necrosis Factor
Figure 7. Identification by reverse-phase HPLC of N-e-NH2-myristyl lysine in bis-acylated TNF peptide no. 1 gas-phase sequencing cycles 6 and
7 (A-D, cycles 5-8, respectively).
myristyl transferase activity. Myristylation of internal lysine
residues joins a short list of co- or posttranslational protein
acylations that includes N-glycyl myristylation, ester-linked
palmitylation, and modification with complex glycosylated
phospholipid (17). Further knowledge concerning the struc-
ture, substrate spedficities, and relationship of the lysyl N-e-
pc'pride no. 1. The in vitro acylation assay was performed as detailed in
Materials and Methods, with the c~ception that the reaction was ended
at the time points indicated. Results are given as the means of triplicate
determinations and are expressed as nanomoles of bis-acylated end-product.
I ~ I L I i I
5 10 15 20
Time kinetics of the enzymatic formation of bis-acylated TNF
NH2-myristyl transferase to the N-glycyl myristyl transferase
awaits its purification and characterization. The detailed sub-
strate characterization of the N-glycyl myristyl transferase has
indicated a complete lack of activity against lysine (18), and
it must therefore be assumed that the lysyl-specific activity
observed here represents a distinct, and previously uurccog-
nized, enzymatic entity. The careful quantitative study of
Towler and Glaser (19), concerning the acylation of cdlular
proteins, indicated that ",70-80% of total amide-linked
myristate was in the form of myristyl glyciue. Interestingly,
a significant amount of radioactivity was also present in an
undefined fraction with the HPLC dution properties charac-
teristic of myristyl lysine, suggesting that acylation of this
residue may not be a rare event.
The enzymatic (octanoyl) acylation of two internal lysine
residues as a consequence of the activation ofAgkistrodon phos-
pholipase A2 has been described (6). The functional conse-
quence of this event was the conversion of the inactive phos-
pholipase A2 monomer to a catalytically effective enzyme
dimer that exhibited enhanced interaction with phospholipid
monolayers. Similarly, chemical acyhtion with a series of fatty
acids of lysyl 6-NH2 groups in pancreatic phospholipase A2
converted the soluble enzyme into a membrane-penetrating
form (20). These studies also demonstrated that attachment
of acyl groups to hydrophobic regions, as opposed to
hydrophilic regions, significantly enhanced the degree of mem-
brane penercation, and that this penetration was optimized
by utilizing fatty acids with smaller molecular areas. Further
insights into the potential role of the myristyl acylation of
lysine residues may be deduced from work done on N-glycyl-
1059 Stevenson et al.
Table 2. Interspecies Homology~ of the TNF Myristflation Site
M L N
M F N
Diagram demonstrating the interspecies amino acid homology for the region surrounding the myristylated lysine residues (boldface). Vertical lines
denote strictly conserved residues, dots denote functional conservation.
myristylated proteins. One extensively studied function has
been the plasma membrane targeting of myristylated pro-
teins. The viral proteins p60 ..... , Pr65gag, and a variant of
p21 Ras are all myristylated, membrane-associated proteins
(21-24). Point mutations that abolish myristylation convert
them into soluble, cytosolic proteins, with consequent im-
pairment of transforming ability or viral assembly. A similar
role for myristylation has been demonstrated for several mam-
malian proteins. The protein kinase C substrate proteins and
the c~ subunit of the GTP inhibitory binding protein all re-
quire binding of myristic acid in order to associate with the
cell membrane (25-27).
TNF-o6 active extracellularly as a 17-kD protein, is trans-
lated as a 26-kD precursor molecule. Initially, it was thought
that the NH2-terminal propiece represented an unusually
long signal peptide, and that the molecule was processed to
its mature form through the classical secretory pathway (28).
However, Muller et al. (29) found that microsomes failed to
process the TNF precursor when analyzed in an in vitro trans-
lation system. In 1988, Kriegler et al. (9) demonstrated that
the 26-kD TNF precursor protein exists as an integral, trans-
membrane protein. The orientation of the TNF precursor
(N~yto/C~xo) was determined by differential proteolytic diges-
tions and is consistent with classification as a type II integral
membrane protein (30). Type II proteins contain signal/anchor
domains, as opposed to cleavable signal peptides, and this pre-
sumably explains the failure of microsomal preparations to
process the 26-kD precursor to the 17-kD form (9, 29).
The existence and biologic activity of plasma mem-
brane-associated TNF has been confirmed by subsequent
studies. Chensue et al. (31) showed the presence of membrane-
associated TNF on mouse peritoneal macrophages by immuno-
histochemical and electron microscopic techniques. Perez et
al. (11) found that transfected cells expressing a noncleavable
mutant of 26-kD TNF on the cell surface were active in cell-
to-cell killing, which did not require processing to the 17-kD
form. In contrast to the findings of Kriegler et al. (9), Bak-
ouche et al. (32) failed to detect an integral membrane 26-
kD TNF protein in activated human monocytes, but instead
recovered a membrane-associated, salt-elutable 17-kD protein,
suggesting that the processed TNF molecule was linked to
a discrete, membrane-associated TNF-binding protein. Leuttig
et al. (33) found evidence for both mechanisms in murine
macrophages. Plasma membranes contained both an acid-
elutable 17-kD form of TNF with the characteristics of a
receptor-bound protein, and a 26-kD integral transmembrane
form. The transmembrane form possessed about 60% of total
TNF bioactivity. In summary, there appear to be two forms
of biologically active membrane-associated TNF: an integral
membrane protein of 26 kD and a processed, 17-kD form
presumably bound to a receptor or binding protein. Processing
of the 26-kD transmembrane form to the 17-kD form appar-
ently involves the action of proteolytic enzymes located on
the cellular surface, (9), leaving behind the propiece within
the membrane (34). Structural analysis of the site of myristyl
acylation of the TNF propiece places the target lysine residues
almost immediately adjacent to a hydrophobic stretch of
sufficient length (24 residues) to act as a membrane-spanning
or anchoring sequence. Examination of the interspecies ho-
mology of the TNF myristylation site shows significant amino
acid conservation (Table 2). In particular, the preferentially
myristylated lysine20 is conserved across all species, consis-
tent with its having a conserved functional role. This site
also conforms to the "positive-inside rule" of yon Heijne (30),
as calculation of the charge distribution across the putative
membrane-spanning region is consistent with the experimental
delineation by Kriegler (9) of a type II N~o/C,xo orienta-
tion. Given these considerations, we hypothesize that the func-
1060 Myristyl Acylation of Tumor Necrosis Factor
tional significance of the myristylation of these lysines is to
facih'tate the membrane insertion or anchoring of this sequence.
This event could occur primarily as a consequence of a phys-
icochemical interaction of the acyl group with membrane phos-
pholipids or via binding to receptors specific for myristylated
TNF, analogous to that identified for myristyl-p60 src by
Resh and Ling (35).
This work was supported by Research Grants from the Department of Veterans Affairs and by Public
Health Service grant DK AI-26918.
Address correspondence to Dr. David Lovett, 111J Medical Service, San Francisco VAMC, 4150 Clement
Street, San Francisco, CA 94121.
Received for publication 12 May 1992.
1. Gordon, J.I. 1990. Protein n-myristylation: simple questions,
unexpected answers. Clin. Invest. 38:517.
2. Duronio, R.J., D.A. Towler, R.O. Heuckeroth, and J.I.
Gordon. 1989. Disruption of the yeast N-myristyl transferase
gene causes recessive lethality Science (Wash. DC). 243:796.
3. Hedo, J., E. Collier, and A. Watkinson. 1987, Myristyl and
palmytyl acylation of the insulin receptor. J. Biol. Chem. 262:
4. Pillai, S., and D. Baltimore. 1987. Myristylation and the post-
translational acquisition of hydrophobicity by the membrane
immunoglobulin heavy-chain polypeptide in B lymphocytes.
Proc Natl. Acad. Sci. USA. 84:7654.
5. Bursten, S.L., R.M. Locksley, J.L. Ryan, and D.H. Lovett.
1988. Acyhtion of monocyte and glomerular mesangial cell
proteins. Myristyl acylation of the Interleukin 1 precursors.
f Clin. Invest. 82:1479.
6. Cho, W., A. TomasseUi, R. Heinrikson, and F.J. Kezdy. 1988.
The chemical basis for interracial activation of monomeric phos-
pholipases A2. Autocatalytic derivatization of the enzyme by
acyl transfer from substrate. J. Biol. Chem. 263:11237.
7. Tomasselli, A.G., J. Hui, J. Fisher, H. Zuercher-Neely, I.M.
Reardon, E. Oriaku, F.J. Kezdy, and R.L. Heinrikson. 1989.
Dimerization and activation of porcine pancreatic phospholi-
pase A2 via substrate level acylation of lysine 56.J. Biol. Chem.
8. Fiers, W. 1991. Tumor necrosis factor: characterization at the
molecular, cellular, and in vivo level. FEBS (Fed. Eur. Biochem.
So~) Lett. 285:199.
9. Kriegler, M., C. Perez, K. DeFay, I. Albert, and S.D. Lu. 1988.
A novel form of TNF/cachectin is a cell surface cytotoxic trans-
membrane protein: ramifications for the complex physiology
of TNF. Cell. 53:45.
10. Decker, T., M. Lohmann-Matthes, and G.E. Gifford. 1987.
Cell-associated tumor necrosis factor as a killing mechanism
of activated cytotoxic macrophages. J. Immunol. 138:957.
11. perez, C., I. Albert, K. DeFay, N. Zachariades, L. Gooding,
and M. Kriegler. 1990. A nonsecretable cell surface mutant
of tumor necrosis factor (TNF) kills by cell-to-cell contact.
12. Goldsmith, M., E. Rattner, M. Koehler, S. Balikov, and C.
Bock. 1979. Two-dimensional electrophoresis of small-mol-
ecular-weight proteins. Anal. Biochem. 99j:248.
13. O'Farrell, P.H. 1975. High resolution two-dimensional dec-
trophoresis of proteins. J. Biol. Chem. 250:4007.
14. Ui, N. 1971. Isoelectric points and conformation of proteins.
I. Effect of urea on the behavior of some proteins in isoelectric
focusing. Biochim. Biophys. Acta. 229:567.
15. Panuska, J.R., K. Fukui, and C.W. Parker. 1988. Secreted pro-
teins of human monocytes. Analysis by two-dimensional dec.
trophoresis and effect of lipopolysaccharide. Biochem.J, 249:501.
16. Towler, D., and L. Glaser. 1986. Protein fatty acid acylation:
enzymatic synthesis of an n-myristylglycyl peptide. Proa Natl.
Acad. Sci. USA. 83:2812.
17. Olsen, E.N. 1988. Modification of proteins with covalent lipids.
Prog. Lipid Res. 27:177.
18. Towler, D.A., S.R. Eubanks, D.S. Towery, S.P. Adams, and
L. Glaser, 1987. Amino-terminal processing of protein by
N-myristylation. J. Biol. Chem. 262:1030.
19. Towler, D., and L. Glaser. 1986. Acylation of cellular proteins
with endogenously synthesized fatty adds. Biochemistry. 25:878.
20. Van der Weile, F.C., W. Atsma, K. Dijkman, A.M.M. Schrenrs,
A.J. Slotboom, and G.H. De Haas. 1988. Site specific e-NH2
monoacyhtion of pancreatic phospholipase A~. 1. Preparation
and properties. Biochemistry, 27:1683.
21. Kamps, M.P., J.E. Buss, and B.M. Sefton. 1985. Mutation of
n-terminal glycine of p60 s~ prevents both myristylation and
morphologic transformation. Proc, Natl. Acad. Sci. USA.
22. Rein, A., M.K. McClure, N.R. Rice, R.B. Luftig, and A.M.
Schultz. 1986. Myristylation site in Pr65 ~g is essential for
virus particle formation by Moloney murine leukemia virus.
Proc. Natl. Acad. Sci. USA. 83:7246.
23. Jacobs, E., P. Gheyson, P. Thines, M. Francott, and M. deWilde,
1989. The HIV-1 Gag precursor Pr55gag synthesized in yeast
is myristylated and targeted to the plasma membrane. Gene
24. Buss, J.E., P.A. Solski, J.P. Schaeffer, M.J. MacDonald, and
cO. Def. 1989. Activation of the cellular proto-oncogene
product p21Ras by addition of a myristylation signal. Science
(Wash. DC). 243:1600.
25. Graft, J.M., J.I. Gordon, and P.J. Bhckshear. 1989. Myristylated
and nonmyristylated forms of a protein are phosphorylated by
protein kinase C. Science (Wash. DC). 246:503.
26. Aderam, A.A., K.A. Albert, M.M. Keum, J.K.T. Wang, P.
1061 Stevenson et al.
Greengard, and Z.A. Cohn. 1988 Stimulus-dependent myristy- Download full-text
lation of a major substrate for protein kinase C. Nature (Lond.).
27. Jones, T.L.Z., W.F. Simonds, J.J. Merendino, M.R. Brann,
and A.M. Spiegel. 1990. Myristylation of an inhibitory GTP-
binding protein c~ subunit is essential for its membrane attach-
ment. Proc. Natl. Acad. Sci. USA. 87:568.
28. Wang, A.M., A.A. Creasey, M.B. Ladner, L.S. Lin, J. Strickler,
J.M. VanArsdell, R. Yamamoto, and D.F. Mark. 1985. Mo-
lecular cloning of the complementary DNA for human tumor
necrosis factor. Science (Wash. DC). 228:149.
29. Muller, R., A. Marmenout, and W. Fiefs. 1986. Synthesis and
maturation of recombinant human tumor necrosis factor in
eukaryotic cells. FEBS (Fed. Eur. Biochem. SoL) Lett. 197:99.
30. yon Heijne, G., and Y. Gavel. 1988. Topogenic signals in inte-
gral membrane proteins. Eur. J. Biochem. 174:671.
31. Chensue, S.W., D.G. Remick, C. Shmyr-Porsch, T.F. Beals,
and S.L. Kunkel. 1988. Immunohistochemical demonstration
of cytoplasmic and membrane-associated tumor necrosis factor
in routine macrophages. Amer. J. Pathol. 133:564.
32. Bakouche, O., Y. Ichinose, R. Heicappell, I.J. Fidler, and L.
Lachman. 1988. Plasma membrane-associated tumor necrosis
factor: a non-integral membrane protein possibly bound to its
own receptor. J. Immunol. 140:1142.
33. Leuttig, B., T. Decker, and M. Lohmann-Matthes. 1989. Evi-
dence for the existence of two forms of membrane tumor
necrosis factor: an integral protein and a molecule attached to
its receptor. J. Immunol. 143:4034.
34. Jue, D.M., B. Sherry, C. Luedke, K.R. Manogue, and A.
Cerami. 1990. Processing of newly synthesized cachectin/tumor
necrosis factor in endotoxin-stimulated macrophages. Biochem-
35. Resh, M.D, and H. Ling. 1990. Identification of a 32K plasma
membrane protein that binds to the myristylated amino-
terminal sequence of p60 v-~'c. Nature (Lond.). 346:84.
1062 Myristyl Acylation of Tumor Necrosis Factor