Protein Science (1995). 4:1507-1515. Cambridge University Press. Printed in the USA.
Copyright 0 1995 The Protein Society
C-terminal specific protein degradation:
Activity and substrate specificity
of the Tsp protease
KENNETH C. KEILER,' KAREN R. SILBER,' KEVIN M. DOWNARD,'
IOANNIS A. PAPAYANNOPOULOS,* KLAUS BIEMANN,* AND ROBERT T. SAUER'
' Department of Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139
'Department of Chemistry, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139
(RECEIVED February 28, 1995; ACCEPTED May 4, 1995)
The activity of Tsp, a periplasmic endoprotease of Escherichia coli, has been characterized by assaying the cleavage
of protein and peptide substrates, determining the cleavage sites in several substrates, and investigating the kinetics
of the cleavage reaction. Tsp efficiently cleaves substrates that have apolar residues and a free a-carboxylate at the
C-terminus. Tsp cleaves its substrates at a discrete number of sites but with rather broad primary sequence speci-
ficity. In addition to preferences for residues at the C-terminus and cleavage sites, Tsp displays a preference for
substrates that are not stably folded: unstable variants of Arc repressor are better substrates than a hyperstable
mutant, and a peptide with little stable structure is cleaved more efficiently than a protein substrate. These data
are consistent with a model in which Tsp cleavage of a protein substrate involves binding to the C-terminal tail
of the substrate, transient denaturation of the substrate, and then recognition and hydrolysis of specific peptide
Keywords: intracellular proteolysis; mass spectrometry; periplasmic protease; substrate specificity
Proteases are ubiquitous in biological systems. The structures,
properties, and mechanisms of a large number and variety of
extracellular proteases are known (Beynon & Bond, 1989), but
our knowledge of intracellular proteases is far less complete. For
example, intracellular proteolysis is extremely selective
1992; Gottesman & Maurizi, 1992) and yet the simple question
of what determines the substrate specificity of an intracellular
protease has in no case been clearly answered. In yeast and
higher eukaryotes, some proteins are marked for degradation
by modification with ubiquitin (Hershko & Ciechanover, 1992),
but in bacteria there is no ubiquitin system. Factors that have
been shown to be important for the proteolytic fate of specific
bacterial proteins include thermal stability (Parsell & Sauer,
1989), the identity of N-terminal sequences (Tobias et al., 1991),
and the identity of C-terminal sequences (Bowie & Sauer, 1989;
Parsell et al., 1990). A better understanding of proteolytic se-
lectivity requires purification of the relevant proteases and bio-
chemical studies of their interactions with substrates.
Tsp is a periplasmic protease of Escherichia coli that was pu-
rified based upon its ability to selectively degrade a variant of
the N-terminal domain of X repressor with the apolar C-terminal
Reprint requests to: Robert T. Sauer, Department of Biology, Mas-
sachusetts Institute of Technology, 68-571, 77 Massachusetts Avenue,
Cambridge, Massachusetts 02139; e-mail: firstname.lastname@example.org.
sequence WVAAA, but not to degrade a variant with the polar
C-terminal sequence RSEYE (Silber et al., 1992). The gene en-
coding Tsp has been called both fsp and prc (Hara et al., 1991 ;
Silber et al., 1992), and Tsp may be identical with protease Re
(Park et al., 1988) and with a protease purified based upon its
ability to cleave oxidized glutamine synthetase (Roseman & Le-
vine, 1987). The X repressor variant that is a substrate for Tsp
in vitro is still degraded in cells deleted for the tsp gene, indi-
cating that another protease must be responsible for degrada-
tion of this substrate in the deletion strain (Silber & Sauer, 1994).
However, cells deleted for the fsp gene are defective in the
C-terminal processing of two periplasmic proteins, penicillin-
binding protein 3 (Nagasawa et al., 1989) and tonB (R. Larsen
& K. Postle, pers. comm.), suggesting that these proteins may
be natural substrates of the Tsp protease. Tsp has no sequence
homology to any characterized protease and is not inhibited by
a wide array of small-molecule protease inhibitors (Silber et al.,
1992). ctpA, a gene with homology to fsp, has recently been
identified in the cyanobacterium Synechocystis sp. PCC 6803
and is believed to encode a protease that is responsible for
C-terminal processing of the Dl protein (Anbudurai et a l . , 1994).
In this paper, we probe the characteristics of peptides and pro-
teins that make them substrates for Tsp in vitro, determine the
location of cleavage sites in several protein and peptide sub-
strates, and present a preliminary characterization of the kinetic
K. C. Keiler et al.
the resulting gene product, called TspH,, using nickel chelate
chromatography. The CD spectra of Tsp and TspH, are
identical and consistent with approximately 35-40% a-helix (see
Fig. I). The molecular weight of TspH, was determined to be
78.5 k 5.3 kDa by sedimentation equilibrium centrifugation.
The polypeptide molecular weight calculated from the sequence
is 75.0 kDa, indicating that TspH, is monomeric, as has previ-
ously been shown for Tsp (Silber et al., 1992).
As shown in Figure 2, Tsp and TspH, have very similar ac-
tivities in cleaving several protein substrates. In addition, time
courses of the cleavage of Arc repressor and the #I05 variant of
the N-terminal domain of X repressor by TspH, and Tsp are in-
distinguishable (not shown). These results indicate that the his-
tidine tag does not affect the specificity or kinetic properties of
Tsp. Moreover, because Tsp and TspH, are purified
pletely different methods and yet display the same enzymatic ac-
tivity, it is extremely unlikely that the observed proteolytic
activity is due to a contaminant. The experiments
the following sections were performed using the TspH, enzyme.
Fig. 1. CDspectraoftheTsp (filled diamonds) and TspHh(opn squares)
properties of Tsp. These studies confirm the importance of
C-terminal sequence in determining substrate preference but
show that additional factors are also
tors appears to be thermodynamic stability because experiments
with a set of related substrates with differing stabilities indicate
that the denatured protein is the only proteolytically suscepti-
ble form. Another factor may be the presence of appropriate
cleavage sites because we find that Tsp makes endoproteolytic
cuts in its substrates at a limited number of discrete bonds.
involved. One of these fac-
Comparison of Tsp and TspH,
To simplify the purification of Tsp, we constructed a tsp gene
encoding six additional C-terminal histidine residues and purified
Importance of C-terminal sequences in different
protein and peptide substrates
Four sets of proteins or peptides were tested as potential sub-
strates for Tsp (Table I). Within each set, the amount of cleav-
age is sensitive to the identity of residues or functional groups
at the C-terminus. For example, the two X repressor variants and
four Arc variants shown in Table IA differ only
C-terminal sequences and yet show dramatically different sen-
sitivities to Tsp cleavage. In single point cleavage assays, the
variants with relatively apolar C-terminal tails (X #105, Arc, and
Arc-st7) are good substrates for Tsp, whereas those with charged
or highly polar tails (X N-terminal domain, Arc-st5, and Arc-
st6) are poor substrates. In addition, even though each of the
Arc variants has the apolar sequence GRIGA
tail, only those with this sequence at the C-terminus (Arc and
Arc-st7) are good substrates.
in its C-terminal
- + "
h 1 1
+ " + -
- + -
+ - - +
Tsp% - - + "
Fig. 2. Degradation of X repressor and Arc sub-
strates by Tsp and TspH,. Reaction mixtures
were incubated at 37 "C for 4 h, and cleavage was
assayed by gel electrophoresis. At shorter incu-
bation times the same cleavage products were ob-
served, but at reduced levels. Arrow indicates the
position of the Tsp and TspH, protein bands.
C-terminal specific protein degradation
Table 1. Cleavage of protein and peptide substrates by Tsp
A. Protein substrates
X N domain
X #I05 N domain
Arc 1-48: GRIGA
Arc 1-48: GRIGAKNQHE
Arc 1-48: GRIGAHHHHHH
Arc 1-48: GRIGAHHHHHHGRIGA
~ . _ _ _ _ _ _ _ _ _
B. Peptide substrates
a Protein substrates were incubated with Tsp for 8 h and cleavage
was quantified by laser densitometry of SDS gels.
Sequences of peptides tested as substrates of Tsp. B represents a
non-natural amino acid amino: butyric acid. Peptides
inus blocked by acetylation (Ac) or the C-terminus blocked by amida-
tion (CONH2) are indicated. Peptides were incubated with Tsp for 5 h,
chromatographed on an HPLC reverse-phase column, and the amount
of cleavage was quantified by integrating the peak corresponding to in-
AC WARAAARAAARBAAB CONH,
AC WARAAARAAARBGGB CONH2
with the N-terrn-
The presence of a free a-carboxyl group is also important in
determining whether closely related peptides with apolar
C-terminal sequences are cleaved efficiently by Tsp. As shown
in Table 1 B, peptides BAS7 and BAS9 differ only in whether the
C-terminus is a carboxyl or carboxyamide group; the same is
true for peptides BAS8 and BAS10. In both cases, the peptide
with the a-carboxyl group is cleaved readily by Tsp, whereas the
variant with the a-carboxyamide group is cleaved poorly or not
cleaved at all (Table 1B; Fig. 3).
Site of proteolytic cleavage
To determine the sites at which Tsp cleaves its substrates, the
digestion products of Arc, Arc-st7, the #lo5 variant of the
N-terminal domain of X-repressor, and peptide BASlO were an-
alyzed by MALDI-TOF mass spectrometry. The major products
from the digestion of each substrate are listed in Table 2 and are
shown schematically in Figure 4. Several conclusions emerge
from these studies. First, Tsp is clearly an endoprotease. Al-
though Tsp recognizes determinants at the C-terminus of sub-
strates, it cleaves these substrates at a limited number of discrete
sites within each polypeptide chain (Table 2; Fig. 4). These sites
occur at different distances in the sequence from the C-terminus
and map to diverse locations in the native secondary and tertiary
structures. The P1 position of the scissile peptide bond is Ala,
Ser, or Val in 14 of the 20 cleavage sites; Ile, Leu, Arg, and Lys
occupy the P1 position in the remaining 6 sites. Hence, Tsp ap-
pears to prefer PI residues that are small and uncharged or that
have aliphatic side chains extending through the C, , C,, or C,
BAS1 0 BAS10 i .
Fig. 3. Cleavage of peptides BASlO (free a-carboxyl group) and BAS8
(a-carboxyamide group) by TspH,. Reverse-phase HPLC traces (A226
versus time) are shown after 0 and 2 h of incubation with TspH, (pH 8,
positions. The same general preferences are evident at the P1'
position of the scissile peptide bond except there is no discern-
ible preference for a small residue and the repertoire of allowed
residues includes Met, Tyr, and Trp.
Kinetic properties and the importance
of substrate stability
To determine approximate kinetic parameters for Tsp cleavage
of a protein and a peptide substrate, initial rates of cleavage were
determined for different substrate concentrations of Arc and
peptide BAS9 by using CD to monitor the loss of cr-helicity that
occurs upon degradation. Figure 5A and B shows the rate ver-
sus substrate concentration plots for Tsp cleavage of Arc and
BAS9, respectively. Fitting of these data yields the apparent ki-
netic parameters shown in Table 3. The apparent K, for Tsp
cleavage of peptide BAS9 is comparable to the K,,, observed for
Tsp cleavage of Arc, but kc,, and k,,,/K, are almost 30-fold
higher for cleavage of the peptide. Thus, Tsp-catalyzed degra-
dation of peptide BAS9 is significantly more efficient than
Tsp-catalyzed degradation of Arc. By contrast, Tsp and chymo-
trypsin have similar values of k,,/K, using Arc as a substrate
One factor that might influence the efficiency with which Tsp
cleaves different proteins or peptides is the equilibrium between
the native and denatured forms of the substrates. This could oc-
cur, for example, if the denatured substrate were actually the
form in which the peptide bond was cleaved (see Discussion).
To test this idea, the susceptibility to cleavage of Arc repressor
K. C. Keiler et al.
Y) .- .
O r ,
.- e -
. d E $
4 - :;
w , "
- i w
C-terminal specific protein degradation
Table 2. Products of Tsp cleavage of substrates determined
by MALDI-TOF mass spectrometry
X #lo5 11,177.15
+ I .95
for the [M+H]+ ion.
Arrows indicate cleavage between the residues shown.
The difference between measured and calculated m/z values for the
assigned sequence is ~0.18% (except for sequence marked by *, which
was confirmed independently; see footnote e).
In addition to 1-102, the starting material contained small amounts
of 1-100 and 1-99. Thus, fragments that end at position 99 or 100 are
assumed to result from these truncated substrates and not from cleav-
age by Tsp.
e Sequence confirmed by an independent MS/MS experiment (Bie-
N-terminal sequence confirmed by Edman degradation.
a Measured m/z values represent average (i.e., polyisotopic) values
variants with decreased stability (Arc-NK29, lysine for aspara-
gine at position 29) or increased stability (Arc-PL8, leucine for
proline at position 8) were compared with that of wild-type Arc.
Arc-NK29 and Arc-PL8 each differ from wild-type Arc at a sin-
gle amino acid residue and have the same apolar C-terminal se-
quence as Arc. Moreover, neither of the sites of mutation is
adjacent to a scissile bond. The melting temperatures of these
three proteins under comparable conditions are approximately
0 . 1 0
50 100 150
Fig. 5. Rate versus substrate concentration plots for Tsp cleavage of
peptide BAS9 and Arc. Cleavage was assayed by changes in CD ellip-
ticity. Error bars indicate the uncertainty in determining the rate in each
assay. Points below 25 p M have errors of similar magnitude that have
not been shown.
37 "C (Arc-NK29), 50 "C (Arc), and 70 "C (Arc-PL8) (M. Milla,
pers. comm.). As shown in Figure 6, the hyperstable PL8 Arc
variant is cleaved significantly more slowly than Arc, whereas
the unstable NK29 variant is cleaved significantly more rapidly
than Arc. These data suggest that substrates are cleaved only in
a denatured form.
In this work, we have investigated the substrate specificity of
Tsp and probed some of its basic enzymatic properties. Previ-
ous studies had shown that Tsp cleaves a variant of X repres-
sor's N-terminal domain with an apolar, but not
C-terminal tail sequence (Silber et al., 1992). Here we have
shown that Tsp also selectively cleaves variants of Arc repressor
depending upon their C-terminal sequences. Arc variants with
relatively apolar C-terminal residues are good substrates for Tsp,
whereas those with polar C-terminal residues are poor substrates.
The studies reported here also show that a free a-carboxyl group
at the C-terminus of an appropriate substrate is important in de-
termining whether it will be cleaved by Tsp. Peptides BAS9
and BASlO, which have apolar C-terminal sequences and free
K. C. Keiler et al.
Table 3. Apparent kinetic parameters for the cleavage of
Arc and the peptide BAS9 by Tsp and chymotrypsin (Chy)=
Protease Substrate Apparent kinetic parameters
K,,, = 50 (k36) pM
k,,, = 0.19 (k0.05) s-'
k,,,/K,,, = 3.8 (k2.9) X lo3 M" s-'
k,,/K, = 4.4 (k0.S) X IO3 M" s-'
K," = 35 (k 13) pM
k,, = 3.7 (k0.5) s-'
= 1 . 1 (t0.4) X lo5 M" s-I
-~ " .
a K,,, and k,, values were determined by fitting the data in Figure 4
to the hyperbolic function u = k,,,[E,] ([SI /(K,,? + [SI)).
a-carboxyl groups, are cleaved readily by Tsp, whereas other-
wise identical peptides with amidated C-termini are cleaved very
poorly. The finding that C-terminal determinants are important
in determining cleavage susceptibility in several unrelated sets
of proteins and peptides strongly suggests that this is a general
mechanism by which Tsp discriminates among potential substrates.
How does the chemical nature
C-terminal region influence whether it will be a good substrate
for Tsp? Because Tsp is an endoprotease with some cleavage
sites far from the C-terminus, it is unlikely that the C-terminal
region of a substrate directly affects the chemistry of
There are several plausible models by which the C-terminal se-
quence of a substrate could affect Tsp
inal sequences that are not strongly polar
as a site to which Tsp can initially bind to the substrate, with
the a-carboxyl group forming part of this binding site. Such
binding might be important simply for tethering Tsp to the sub-
strate but might also be required to activate the protease in some
fashion. Activation does not occur in trans because neither the
#IO5 variant of the N-terminal domain of X repressor nor a pep-
tide corresponding to the last nine residues of #I05 stimulates
of a protein's or peptide's
cleavage. First, C-term-
or charged could act
5 0.60 !
wild type Arc
Fig. 6. Tsp cleavage of wild-type Arc and stability variants assayed by
loss of CD ellipticity at 222 nm. Melting temperatures of PL8, Arc, and
NK29 are approximately 70, 50, and 37 "C, respectively.
cleavage of the wild-type N-terminal domain of X repressor (data
not shown). We note that the tail sequences of Arc and the
N-terminal domain of X repressor are partially disordered and
accessible in the native structures of these proteins (Weiss et al.,
1987; Breg et al., 1990), which would allow them to act as bind-
ing sites for Tsp even in the native proteins. Second, the tail
quences could act indirectly by promoting unfolding or other
conformational changes in the substrate that result in increased
susceptibility of the substrate to Tsp cleavage. This model, how-
ever, is not consistent with experiments showing that Tsp-
sensitive variants with hydrophobic tails have
conformations comparable to Tsp-insensitive proteins with hy-
drophilic tails in both the
X repressor and Arc proteins (Parsell
& Sauer, 1989; Milla et al., 1993). Moreover, although Tsp can
discriminate between variants of X repressor or Arc differing
only at their C-termini, nonspecific proteases cleave variants
with hydrophobic or hydrophilic tails equally well (Silber, 1992).
The peptide pairs examined here (BAS7/BAS9 and BAS81
BASIO) have little stable structure as assayed by CD, but their
spectra are identical, indicating that amidation of the terminaI
carboxylate in peptides BAS7 and BAS8 does not lead to sig-
nificant conformational changes. Because there is no evidence
that apolar C-terminal sequences alter substrate conformation
or stability, we favor the idea that these sequences act as bind-
ing sites for Tsp.
Mapping of the sites of Tsp cleavage in the primary structures
of several protein and peptide substrates reveals a limited num-
ber of discrete sites of proteolysis that occur throughout the sub-
strate sequences. Tsp cleaves peptide bonds with Ala, Ser, or Val
(and to a lesser extent He, Leu, Lys, or Arg) at the P1 position
and with these same side chains plus Met, Tyr, or Trp at the PI'
position. These specificity preferences appear to be sufficiently
lax that most proteins and peptides would be expected to con-
tain some cleavage sites. We note, however, that many of the
substrates contain peptide bonds that fit these broad rules but
are not sites at which cleavage is observed. For example, pep-
tide BAS10 is cleaved at the Ala2-Arg3 bond but not at the
Ala7-Args or Ala12-Arg13 bonds. These latter bonds may not
be cleaved because of additional side-chain preferences at ad-
joining residues (P2, P2', etc.) or because other factors are op-
erative. For example, if Tsp contains a site for binding the
C-terminus of the substrate that is distinct from the active site,
then the distances between these sites may introduce a require-
ment for some minimal distance between the C-terminus of the
substrate and a cleavage site.
We find that an Arc repressor variant that is less stable than
wild type is cleaved more rapidly and a variant that is more sta-
ble than wild type is cleaved more slowly. These results are con-
sistent with a model in which denaturation of substrate
required prior to cleavage. Indeed, the positions of the cleav-
age sites in Arc and the N-terminal domain of
to many different locations in the native secondary and tertiary
structure, including the hydrophobic core, surface loops, un-
structured regions, and the middle of a-helices. All of these sites
cannot be recognized in the context of the native structures of
Arc or the N-terminal domain of X repressor because many of
the scissile peptide bonds would be inaccessible to the protease
in the native proteins. If global rather than local denaturation
of protein substrates is a prerequisite for cleavage, then the
global unfolding rate constant would
equal to k,, for Tsp cleavage (>0.2 s-' for Arc at 37 "C). In
h repressor map
need to be greater than or
C-terminal specific protein degradation
fact, the unfolding rate constant for Arc is -3 s-' at 37 "C (C.
Waldburger, pers. comm.) and thus global unfolding of Arc
could precede cleavage.
Kinetic analysis of substrate cleavage by Tsp is complicated
because there are often multiple cleavage sites, and the proteo-
lytic products may themselves be substrates for further cleavage
or even inhibitors. For these reasons, only apparent values
of kinetic parameters can be measured. Cleavage of Arc by
Tsp at 37 "C and pH 8 is relatively slow (apparent kca,/Km =
3.8 X lo3 M" s") but is similar in rate to cleavage of the
same substrate by chymotrypsin (apparent k,,,/K,,, = 4.4 X
IO3 M-' s" ). The cleavage of peptide BAS9 (kca,/Km = 1.1 X
IO5 M-' s-' ) by Tsp is almost 30-fold more efficient. We pre-
sume that BAS9 is a better substrate for Tsp than Arc
it is largely unfolded at 37 "C (B. Schulman, pers. comm.), al-
though other factors may also contribute. For comparison,
cleavage of pentapeptide substrates by elastase have k,,,/K,,,
values of approximately 2 X lo4 M" s-' (Thomson & Blout,
1973; Bauer et al., 1976). Thus, Tsp does not appear to be any
less efficient in cleaving susceptible protein and peptide sub-
strates than well-characterized enzymes such as chymotrypsin
and elastase, but it is far more specific in its choice of substrates.
A protease such as Tsp almost certainly comes in contact with
a wide variety of different proteins in the periplasm of the bac-
terial cell and must be able to discriminate reliably between ap-
propriate and inappropriate substrates. Our studies suggest that
the chemical nature and structural accessibility of the C-terminal
sequence of potential substrates play a central role in determin-
ing whether such proteins are cleaved by Tsp. Proteins with po-
lar C-terminal residues or with C-terminal sequences that are
buried within the native structure should not be cleaved by Tsp.
Similarly, even if a protein has an accessible and apolar C-term-
inus that allowed transient Tsp binding, it could probably still
escape cleavage if its native structure were sufficiently stable.
By using a combination of specificity determinants for substrate
recognition, Tsp is apparently able to avoid degrading the wrong
proteins in the cell.
Materials and methods
Proteins and peptides
Wild-type Tsp was purified from E. coli strain X90 containing
the Tsp overproducing plasmid pKS6-lw (Silber et al., 1992).
Cells were grown in 6-12 L of LB broth at 37 "C with gentle
shaking to an absorbance of 0.8 at 600 nm and induced by ad-
dition of 1 mM IPTG. The culture was grown for an additional
3 h, harvested by centrifugation, and resuspended in 10 volumes
of a sphereoplasting buffer containing 100 mM Tris-HCI, pH 8,
5 mM EDTA, 500 mM sucrose. After resuspension, the cells
were collected again by centrifugation, resuspended in 10 vol-
umes of cold water to burst the outer membrane and release the
periplasmic proteins, and the cytoplasmic and membrane por-
tions were removed by centrifugation. The soluble periplasmic
fraction was chromatographed on a 20-mL DEAE Sephacel col-
umn equilibrated in 10 mM Tris-HCI, pH 8, 100 mM KCI. The
flow-through fraction was collected, concentrated to 0.5 mL by
ultrafiltration, and chromatographed on a 24-mL Superose 6
column in 10 mM Tris-HCI, pH 8, 100 mM KCI, using a Phar-
macia FPLC apparatus. Fractions containing Tsp at greater than
95% purity, as assayed by SDS-PAGE, were pooled. Tsp was
stored in column buffer at 0 "C with no loss of activity over a
period of several months.
A gene encoding a variant of Tsp (TSpH6) containing six ad-
ditional histidine residues at the C-terminus was constructed by
subcloning the Eco RI-Hind 111 fragment from pKS6-lw (which
contains all of the tsp gene with the exception of 66 bases at the
3' end of the gene) into Eco RI and Xba I cut pBluescript I1
KS(+) together with a Hind 111-Xba I oligonucleotide cassette
designed to restore the wild-type bases at the 3' end of the gene
followed by the His6 tag and normal termination codon. The
structure of the resulting plasmid (pKKlO1) was confirmed by
restriction mapping and DNA sequencing. The TSpH6 protein
was purified from E. coli strain KS1000, which contains a de-
letion of the chromosomal tsp gene (Silber & Sauer, 1994), trans-
formed with pKKlOl. Cells were grown, induced, and harvested
as described above. The cell pellet was resuspended in five vol-
umes of 100 mM Tris-HCI, pH 8,200 mM KCI, 1 mM EDTA,
2 mM CaCI,, 10 mM MgC12, 1.4 mM 2-mercaptoethanol, lysed
by sonication, and the lysate was cleared by centrifugation. The
supernatant was applied to a 20-mL Ni-NTA column (Qiagen)
equilibrated in 100 mM NaH,PO,, 10 mM Tris-HCI, pH 8.
Bound proteins were eluted with a 500-mL gradient formed by
linear mixing of the column buffer at pH 8 and column buffer
titrated to pH 4.5 with HCI. Fractions containing the TspHd
protein, which eluted between pH 5.3 and pH 4.8, were pooled,
concentrated to 0.5 mL by ultrafiltration, and chromatographed
on a 24-mL Superdex 75 column equilibrated in 10 mM Tris-
HCI, pH 8,
100 mM KCI. Fractions containing TspH, at
greater than 95% purity were pooled.
The Arc-st7 protein differs from wild-type Arc in having 11
additional C-terminal residues (HHHHHHGRIGA). The arc-
st7 gene was constructed by synthesis of an appropriate Hind
111-Cla I cassette that was inserted between the Hind I11 and
Cla I sites of plasmid pSA600 (Milla et al., 1993). Arc repres-
sor, the Arc-st5, Arc-st6, and Arc-st7 variants, the N-terminal
domain of h repressor (residues 1-102), and the #lo5 variant of
the N-terminal domain were purified by ion-exchange chroma-
tography, affinity chromatography, and gel filtration as de-
scribed (Lim & Sauer, 1991; Milla et al., 1993). The NK29 and
PL8 mutants of Arc were gifts of Marcos Milla. Peptides BAS7-
BAS10 were the generous gift of Brenda Schulman. The ex-
pected molecular weights of all peptides were confirmed by mass
Characterization of Tsp
The CD spectra of Tsp and TspH6 (50 pg/mL) were recorded at
5 "C in CD buffer (50 mM phosphate, pH 7.0, 100 mM KCI)
using an Aviv model 60DS circular dichroism spectrometer. The
a-helical content was estimated using the program Varselec (Man-
avalan & Johnson, 1987). Sedimentation equilibrium centrifu-
gation was performed in 10 mM Tris-HCI, pH 8, 100 mM KC1
using a Beckman XL-A analytical ultracentrifuge. Molecular
weight values were determined for 0.5 pM, 1.7 pM, 3.4 pM, and
5 pM TspH6 at rotor speeds of 15,000 rpm and 13,000 rpm by
nonlinear least-squares fitting of the data to a single species mo-
lecular weight function (C(r) = C(a)exp[w2M(1 - up)(r' - a')/
2RT1) (Laue et al., 1992) using the program Nonlin (Johnson
& Frazier, 1985; Brenstein, 1989).
K.C. Keiler et al.
Assays for proteolysis
For assays of Tsp or TspH, cleavage of protein substrates by
gel electrophoresis, 0.5 pg of protease and 2 pg of substrate in
30 pL of reaction buffer (10 mM Tris-HC1, pH 8.0, 200 mM
KCI) were incubated at 37 "C for 4 h and the reaction products
were electrophoresed on 15% polyacrylamide-SDS-Tris-tricine
gels (Schagger & von Jagow, 1987). Cleavage of peptides by
TspH, was monitored by HPLC on a reverse-phase column
(gradient from 0 to 100% acetonitrile in 0.1% TFA), following
incubation of 500 pg peptide with 3.4 pg TspH, in 1 mL of re-
action buffer for 5 h at 37 "C. The extent of cleavage was quan-
tified by loss of area in the peak corresponding to the intact
peptide and by appearance of additional peaks in the chromato-
graph. Reverse-phase column chromatography was also used to
monitor the time course of Tsp cleavage of protein substrates
such as Arc and the #lo5 variant of the N-terminal domain of
X repressor. For these assays, 10 pg of substrate was mixed with
2 pg TspH, in 300 pL of reaction buffer at 37 "C, and 50-pL
aliquots were removed from the reaction at 20-min intervals,
quenched with 0.1 Vo TFA, and analyzed by reverse-phase chro-
matography as described above.
Kinetic constants for cleavage of Arc repressor and peptide
BAS9 were determined by mixing 50 nM TspH, with 2.5-220 pM
Arc repressor or 1.6-150 pM BAS9 in reaction buffer at 37 "C
and monitoring changes in CD ellipticity at 222 nm. The rate
of cleavage in mdeg/s was determined from a linear fit of the
ellipticity versus time data and converted to units of pM sub-
strate cleaved/s/pM enzyme assuming that cleaved substrate has
no ellipticity at 222 nm. K, and k , ,
were determined by fitting the rate versus substrate concentration
data to the hyperbolic function v = kc,, [E,] [[SI
using the program Nonlin. The relative rates of cleavage of Arc
repressor variants were determined by mixing 45 nM Tsp with
Arc, Arc-PL8, or Arc-NK29 at concentrations of 8 pM, and
monitoring cleavage using the CD assay described above.
values for Arc and BAS9
/(Km + [SI))
The peptides produced by Tsp cleavage of X #105, Arc, Arc-st7,
and the peptide BAS10 were analyzed by MALDI-TOF mass
spectrometry (Hillenkamp et al., 1991). Typically 50-150 nmol
of substrate were digested with 100 pmol of Tsp or TspH, for
1-6 h. In some cases, the digestion products were partially puri-
fied by reverse-phase chromatography or nickel-NTA affinity
chromatography prior to mass spectrometric analysis. A sample
of each digest was diluted with a 10 mM solution of 2,5-dihy-
droxybenzoic acid or 3,5-dimethoxy-4-hydroxycinnamic
a concentration of approximately 10 pM. One microliter of each
matrix/analyte solution was deposited and dried on the sample
probe of a VESTEC 2000 linear time-of-flight mass spectrometer
(PerSeptive Biosystems). Samples were irradiated at 337 nm with
a nitrogen laser (model VSL337ND, Laser Science Inc.). Mea-
sured m/z values for the [M+H]+ ions detected were matched
with segments of the protein using a computer program (COM-
POST) that has been previously described (Papayannopoulos
& Biemann, 1991). In cases where more than one peptide from
a given substrate could correspond to the measured mass, se-
quence assignments were made based on the consistency of the
peptide termini with those from other peptide digestion prod-
ucts. In this regard, peptides formed through a single cleavage
of the intact protein were favored over those requiring two cleav-
ages. In several cases, sequence assignments were confirmed in-
dependently from the collision-induced dissociation mass spectra
of the peptide products (see Biemann, 1990) or by N-terminal
sequencing using the Edman degradation.
We thank Bronwen Brown, Brenda Schulman, and Marcos Milla for
gifts of proteins and peptides. We also thank Ray Larsen and Kathleen
Postle for communicating unpublished data. This study was supported
by NIH grants AI-I5706 and RR-00317.
Anbudurai PR, Mor TS, Ohad I, Shestakov SV, Pakrasi HB. 1994. The CtpA
gene encodes the C-terminal processing protease for the Dl protein of
the photosystem I1 reaction center complex. Proc Nafl Acad Sci LISA
Bauer CA, Thompson RC, Blout ER. 1976. The active centers of Sfrepto-
myces griseus protease 3, alpha-chymotrypsin, and elastase: Enzyme-sub-
strate interactions close to the scissile bond. Biochemisfry 15:1296-1299.
Beynon RJ, Bond JS, eds. 1989. Proteolyric enzymes: A practical approach.
Oxford, UK: IRL Press.
Biemann K. 1990. Sequencing of peptides by tandem mass spectrometry and
high-energy collision-induced dissociation. Methods Enzymol 193:455-
Bowie JU, Sauer RT. 1989. Identification of C-terminal extensions that pro-
tect proteins from intracellular proteolysis. J Biol Chem 264:7596-7602.
Breg JN, van Opheusden JHJ, Burgering MJM, Boelens R, Kaptein R. 1990.
Structure of Arc repressor in solution: Evidence for a family of beta-sheet
DNA-binding proteins. Nature 346:586-589.
Brenstein RJ. 1989. Nonlin for Macinfosh. Carbondale, Illinois: Robelko
Goldberg AL. 1992. The mechanisms and functions of ATP-dependent pro-
teases in bacterial and animal cells. Eur J Biochem 203:9-23.
Gottesman S, Maurizi MR. 1992. Regulation by proteolysis: Energy-
dependent proteases and their targets. Microbiol Rev 56:592-621.
Hara H, Yamamoto Y, Higashitani A, Suzuki H, Nishimura Y. 1991. Clon-
ing, mapping, and characterization of the Escherichia coliprc gene, which
is involved in C-terminal processing of penicillin-binding protein 3. J Bac-
teriol 173 :4799-48 13.
Hershko A, Ciechanover A. 1992. The ubiquitin system for protein degra-
dation. Annu Rev Biochem 61:761-807.
Hillenkamp F, Karas M, Beavis RC, Chait BT. 1991. Matrix-assisted desorp-
tionhonization mass spectrometry of biopolymers. Anal Chem 63: I 193A-
Johnson ML, Frazier S. 1985. Nonlinear least-squares analysis. Mefhods En-
Laue TM, Shah BD, Ridgeway TM, Pelletier SL. 1992. Computer-aided in-
terpretation of analytical sedimentation data for proteins. In: Harding
S, Rowe A, Horton J, eds. Analytical ulfracentrifugafion in biochemis-
try andpolymer science. Cambridge, UK: Royal Society of Chemistry.
Lim WA, Sauer RT. 1991. The role of internal packing interactions in de-
termining the structure and stability of a protein. JMol Bio1219:359-376.
Manavalan P, Johnson WC Jr. 1987. CD secondary structure analysis using
variable selection. Anal Biochem 167:76-85.
Milla ME, Brown BM, Sauer RT. 1993. P22 Arc repressor: Enhanced ex-
pression of unstable mutants by addition of polar C-terminal
Protein Sei 2:2198-2205.
Nagasawa H, Sakagami Y, Suzuki A, Suzuki H, Hara H, Hirota Y. 1989.
Determination of the cleavage site involved in C-terminal processing of
penicillin-binding protein 3 of Escherichia coli. J Bacteriol 171:5890-
Pabo CO, Lewis M. 1982. The operator-binding domain of lambda repres-
sor: Structure and DNA recognition. Nature 298:443-447.
Papayannopoulos IA, Biemann K. 1991. A computer program (COMPOST)
for predicting mass spectrometric information from known amino acid
sequences. J Am Soc Mass Specfrom 2:174-177.
Park JH, Lee YS, Chung CH, Goldberg AL. 1988. Purification and char-
acterization of protease Re, a cytoplasmic endoprotease in Escherichia
coli. J Bacteriol 170:921-926.
C-terminal specific protein degradation
Parsell DA, Sauer RT. 1989. The structural stability of a protein is an im-
portant determinant of its proteolytic susceptibility in Escherichia coli.
J Biol Chem 264:1590-1595.
Parsell DA, Silber KR, Sauer RT. 1990. Carboxy-terminal determinants of
intracellular protein degradation. Genes & Dev 4:277-286.
Roseman JE, Levine RL. 1987. Purification of a protease from Escherichia
coli with specificity for oxidized glutamine synthetase. JBiol Chem 262:
Schagger H, von Jagow G. 1981. Tricine-sodium dodecyl sulfate-plyacryl-
amide gel electrophoresis for the separation of proteins in the range from
1 to 100 kDa. Anal Biochem 166:368-379.
Silber KR. 1992. The role of carboxyl termini in selective protein degrada-
tion [thesis]. Cambridge, Massachusetts: Massachusetts Institute of
Silber KR, Keiler KC, Sauer RT. 1992. Tsp: A tail-specific protease that se-
lectively degrades proteins with nonpolar C termini. Proc Nut1 Acud Sci
Silber KR, Sauer RT. 1994. Deletion of theprc (tsp) gene provides evidence
for additional tail-specific proteolytic activity in Escherichia coli K-12.
Mol Gen Genet 242:237-240.
Thomson RC, Blout ER. 1973. Dependence of the kinetic parameters for
the elastase-catalyzed amide hydrolysis on the length of peptide substrates.
Tobias JW, Shrader TE, Rocap G, Varshvsky A. 1991. The N-end rule in
bacteria. Science 254:1374-1377.
Weiss MA, Karplus M, Sauer RT. 1987. 1H NMR aromatic spectrum of the
operator binding domain of the lambda repressor: Resonance alignment
with application to structure and dynamics. Biochemistry 26:890-897.