Article available online at http://www.idealibrary.

com on Microbial Pathogenesis 2000; 29: 231–244

doi:10.1006/mpat.2000.0383

MICROBIAL
PATHOGENESIS

Induction of MMP-9 mediated gelatinolytic

activity in human monocytic cells by cell wall
components of Mycobacterium tuberculosis
Carlos A. Rivera-Marrero, William Schuyler, Susanne Roser &
Jesse Roman∗
Department of Medicine, Pulmonary and Critical Care Division, Atlanta VA Medical Center and
Emory University School of Medicine, Atlanta, GA 30033, U.S.A.

(Received March 2, 2000; accepted in revised form May 28, 2000)

Mycobacterium tuberculosis (Mtb) infection induces the expression of host matrix metallo-
proteinases (MMPs) capable of tissue degradation. We show that infection of mice with Mtb results
in differential expression of MMPs in the lung. MMP-9 activity increased by week 1 post-infection,
while MMP-2 activity increased after week 2. RT-PCR analysis for gene expression of gelatinases
and their respective inhibitors showed: a small increase in MMP-9 by week 1, no change in TIMP-
1 and MMP-2, and a significant decrease in TIMP-2 by week 4. The increase in MMP-2 could be
due to a decrease in TIMP-2 expression. Addition of 4-aminophenylmercuric acid to lung extracts
increased MMP-9 activity, suggesting that its regulation could be due to endogenous activation
by proteases. In vitro, attenuated and virulent Mtb strains equally induced MMP-9 expression in
U937 monocytes. The inducer of MMP-9 in Mtb was present in culture filtrates, and was active
after paraformaldehyde fixation. LAM stimulated MMP-9 expression in THP-1 cells, but not U937
cells. However, LAM-free extracts also induced MMP-9 activity in THP-1 cells. Fractionation of
Mtb extracts by chromatography revealed fractions of 17 and 156 kDa with MMP-9 inducing
activity. In conclusion, LAM and other components of Mtb induce the expression of MMP-9.
Key words: gelatinases, matrix metalloproteinases, tuberculosis, granuloma, lung cavitation.

Introduction cases of tuberculosis (TB) and almost 3 million

Mycobacterium tuberculosis (Mtb) is the leading
cause of death due to an infectious disease
worldwide with an estimated 8 million new
∗ Author for correspondence: Atlanta VA Medical Center,
Pulmonary & Critical Care Section, Room 12C191, 1670
Clairmont Rd, Decatur, GA 30033, U.S.A. E-mail: roman-
rodriguez.jesse@atlanta.va.gov
0882–4010/00/100231+14 $35.00/0
deaths from this disease yearly [1]. One of the
most devastating consequences of infection with
Mtb is the formation of caseating granulomas
followed by tissue destruction with liquefaction
causing cavity formation [2, 3]. Despite the pre-
dominant role these events play in the pro-
gression of disease and dissemination of
mycobacteria, very little is known about how
Mtb infection elicits this destructive process.
232
Early studies have indicated that cellular hy-
persensitivity responses to Mtb are responsible
for the tissue inflammation and massive caseous
tissue necrosis and liquefaction observed during
Mtb infection [4, 5]. Proteolitic damage by
macrophage-secreted proteases has been im-
plicated since macrophages are the predominant
cell in Mtb granulomas and contain a large
number of proteases capable of degrading the
extracellular matrix (ECM) [6–9]. There is also
evidence that secreted proteases are upregulated
in the granulomatous reaction [10]. Based on this
information, we hypothesize that mycobacteria
elicit a host tissue response characterized by
the exaggerated expression of bioactive matrix-
degrading enzymes which are responsible for
much of the tissue destruction observed in in-
fected lungs. Our work is directed to the iden-
tification of matrix-degrading enzymes that
could play significant roles in Mtb-induced lung
tissue destruction.
Of all the candidates potentially involved in
this process, we have focused on the gelatinases
(also known as type IV collagenases) members
of the Matrix Metalloproteinase (MMP) family.
MMPs are a group of zinc- and calcium-de-
pendent endopeptidases capable of degrading
collagens, elastin, fibronectin, proteoglycans and
other connective tissue matrices [11, 12]. These
enzymes have not been studied extensively in
the setting of Mtb infection, but have been im-
plicated in many disorders characterized by lung
tissue destruction including emphysema and
sarcoidosis, another systemic illness, that like TB,
is characterized by pulmonary granulomatosis
[13–17]. Recently, Chang and colleagues dem-
onstrated that bacilli from an attenuated strain
of Mtb (H37Ra) are capable of stimulating the
production of gelatinases by transformed mono-
cytic THP-1 cells [18].
Our studies revealed that infection with vir-
ulent Mtb Erdman strain induces the differential
expression of gelatinolytic activity mediated by
the gelatinases MMP-9 and MMP-2 in the lungs
of C57BL/6 infected mice. To investigate the
cellular mechanisms involved in MMP induction
by Mtb, we examined their induction in vitro in
the human monocytic cell lines U937 and THP-
1. We show that U937 and THP-1 cells express
MMP-9 when infected with Mtb. The stim-
ulatory effect of Mtb is observed with both
attenuated and virulent strains, with live and
formalin-fixed dead bacilli, and is dependent
upon cell-to-cell contact between Mtb bacilli and
monocytic cells. Furthermore, we show that
C. A. Rivera-Marrero et al.
purified cell membrane liporabinomannans
manLAM and araLAM stimulate the expression
of MMP-9 in THP-1 cells but not in U937 cells.
Finally, we demonstrate that the inducer of MMP
expression in Mtb is found in long term culture
filtrates, detergent extracts of bacilli, LAM-free
cell lysates, and can be separated by column
chromatography. These studies suggest that
LAM glycolipids and other components of the
cell wall of Mtb induce MMPs which could in
turn play important roles in the pathogenesis of
mycobacterial infection.
Results
Induction of MMP activity in the lungs of
mice infected with Mtb
C57BL/6 mice were infected with an intra-
venous inoculum (1×106 cfu) of virulent Mtb
Erdman strain and killed at 1, 2 and 3 weeks
post-infection. The Mtb infected lungs were har-
vested, extracted in detergent buffer, and ex-
amined by gelatin zymography. Fig. 1(a) shows
that a band of 92 kDa, corresponding to MMP-
9, present in lungs of normal mice, increased in
intensity in lungs of Mtb infected mice as early as
1 week post-infection. Increased MMP-9 activity
was also seen after 2 and 3 weeks of infection,
with the highest levels detected at 3 weeks when
the experiment was ended. A band of 72 kDa
corresponding to MMP-2 was detected only after
2 and 3 weeks of infection. As shown in Fig. 1(b),
analysis of this gel by densitometry scanning
confirmed the observed increase in the ex-
pression and/or activity of MMP-9 and MMP-
2 in the lungs of mice during infection. The
increase in MMP-9 and MMP-2 activity cor-
related with the increase in bacillary load seen
in the lungs of infected mice at weeks 2 and 3
post-infection [Fig. 1(c)].
In a similar experiment, groups of seven
C57BL/6 mice were infected with Mtb Erdman,
the lungs harvested at 1, 2, 3 and 4 weeks post-
infection, and total RNA prepared for analysis
of gene expression by quantitative RT-PCR. The
levels of MMP-9 and MMP-2 mRNA were de-
termined by the use of gene specific primers
and probes as described. In addition, we ex-
amined the expression of two MMP inhibitors,
Tissue Inhibitor of MMP-1 (TIMP-1) and Tissue
Inhibitor of MMP-2 (TIMP-2). As shown in Fig.
2(a), during the course of infection, a small
Induction of lung gelatinases by Mycobacterium tuberculosis 233

(a) course of infection, a significant decrease in the

Weeks post-infection mRNA for the MMP-2 inhibitor TIMP-2 was

MMP
kDa
N 1 1 2 2 3 3 seen at week 4 post-infection. The latter suggests
that the increase in MMP-2 activity seen by
gelatin zymography in lung extracts after week
92
2 and 3 post-infection could be due to a relative
decrease in TIMP expression.
72 Another possible level of regulation is ac-
tivation of MMPs by cleavage of their pro-peptide
(b)
7000
domain after secretion. To determine if the in-
creased gelatinolytic activity observed in lungs
Densitometry values

6000 92 kDa 72 kDa

5000
of infected mice could be due to exogenous ac-
4000
tivation of resident proteases by pro-peptide
3000
cleavage, lung extracts were treated with 4-
2000
aminophenylmercuric acid (APMA, an agent
1000 capable of cleaving the pro-peptide and thereby
0 activing the MMP) and then examined by gelatin
N 1 2 3 zymography. Fig. 2(c) shows that addition of
Weeks post-infection
10 mM APMA to lung extracts of mice obtained
1 [Fig. 2(c), 1–1 to 1–7] and 2 [Fig. 2(c), 2–1 to
(c) 2–6] weeks post-infection, resulted in a significant
8 increase in MMP-9 activity. Normal mice [Fig.
2(c), N-1 and N-2] showed very low levels of
6
5

MMP-9 activity in their lungs and addition of

cfu/ml × 10

4 APMA also resulted in increased activation of

2 active MMP that can be activated by pro-peptide
0
1 2
Weeks post-infection
Figure 1. Gelatinolytic activity in lung extracts of
Mtb infected C57BL/6 mice. (a) C57BL/6 mice in-
fected with Mtb Erdman strain, killed at 1, 2 and 3
weeks after infection, and lung extracts analysed for
MMP activity by gelatin zymography. A band of
gelatinolytic activity of 92 kDa, corresponding to
MMP-9, is seen prior to infection (N), and increases
in intensity during the course of infection (weeks 1,
2 and 3). A band of activity of 72 kDa, corresponding
to MMP-2, is only detected after 2 and 3 weeks
infection. (b) Densitometry scan of bands obtained
in gelatin zymograms showing the differential ex-
pression of MMP-9 and MMP-2 in lungs during Mtb
infection. (c) cfu counts in lung extracts obtained
after weeks 1, 2 and 3 of infection.
increase in MMP-9 trascription was seen at week
1 post-infection but the levels of expression re-
turned to baseline at weeks 2, 3 and 4. The
expression of the MMP-9 inhibitor TIMP-1
showed no significant change during the course
of infection. Fig. 2(b) shows that, while the ex-
pression of MMP-2 did not change during the
MMP-9. This suggests that the lungs contain in-
cleavage during the course of Mtb infection.
3
Mtb induces the expression of MMP-9 in
vitro in U937 monocytic cells
To determine the mechanisms responsible for
induction of gelatinases after infection with Mtb,
we compared the effects of infection of cultured
U937 human monocytic cells with an attenuated
vs a virulent strain of Mtb. U937 cells were
infected with the attenuated strain H37Ra and
the virulent strain Erdman at a multiplicity of
infection (moi) of 1–10 for a period of 12 h.
Culture supernatants and detergent cell extracts
of infected U937 cells were analysed by gelatin
zymography. As shown in Fig. 3(a), both at-
tenuated and virulent Mtb strains induced the
expression of MMP-9 in U937 cells and the
activity of MMP-9 was higher in culture super-
natants than in detergent cell extracts. Consistent
with its extracellular secretion after synthesis,
densitometric analysis of the zymograms re-
vealed a two-fold increase in MMP-9 activity
with H37Ra and Erdman Mtb strains over the
unstimulated control and a 0.3-fold higher ac-
tivity in culture supernatants over detergent cell
extracts [Fig. 3(b)]. To determine if the observed
234 C. A. Rivera-Marrero et al.

200 gelatinolytic activity was derived from the U937

(a)
cells and not from Mtb, detergent extracts and
broth culture supernatants of Mtb bacilli were
150 tested by gelatin zymography. No gelatinolytic
Relative light units

activity was detected in these Mtb fractions (data

not shown) indicating that the gelatinolytic ac-
100 tivity was originated from the U937 cells upon
exposure to Mtb.

50
MMP-9 induction by Mtb occurs with
formalin fixed Mtb and is dependent on
cell to cell contact
0
N 1 2 3 4 To investigate the nature of the inducer(s) of
Weeks of infection gelatinolytic activity present in attenuated and
virulent Mtb strains, we tested Mtb bacilli and
200 extracts under various conditions and in two
(b) types of human monocytic cells, the U937 and
THP-1 cell lines. Our initial experiments dem-
onstrated that the MMP inducer in Mtb was
150
found in Mtb bacilli and in long term culture
Relative light units

filtrates of Mtb suggesting the secreted nature

of this molecule (not shown). However, in tran-
100 swell plate assays, where Mtb bacilli and mono-
cytes were separated by a 0.4
M semipermeable
membrane, no induction of MMP activity was
50 detected during 24 h of exposure (not shown).
These experiments suggested that the inducer
of MMP in Mtb bacilli is a cell associated
0 molecule that acts by cell-to-cell contact and that
N 1 2 3 4 it is shed into the culture medium during long
Weeks of infection term growth. To determine how labile is the
inducer present in the cell envelope of Mtb, Mtb
bacilli were killed by fixation in 4% para-
5000
(c) formaldehyde and tested for MMP induction
by U937 and THP-1 monocytes. As shown by
Densitometry units/µg protein

4000 gelatin zymography, fixed killed Mtb H37Ra

(attenuated) and Erdman (virulent) bacilli in-
duced MMP-9 expression in U937 and THP-1
3000

2000

Figure 2. Expression of MMPs and TIMPs in lung

1000 tissue. C57BL/6 mice (6–7 per group) were infected
with Mtb Erdman strain, killed at 1–4 weeks after
infection, and lung extracts analysed by RT-PCR. (a)
0 mRNA levels for MMP-9 (Χ) and TIMP-1 (Φ) in
lung tissue during weeks 1–4 post-infection. (b)
7
2
3
4
5
6

2
3
4
5
6
-2
1

1
-1

mRNA levels for MMP-2 (Ε) and TIMP-2 (Α) in

1-
1-
1-
1-
1-
1-

2-
2-
2-
2-
2-
1-

2-
N
N

Mouse lung extract lung tissue during weeks 1–4 post-infection. (c) Ac-
tivation of MMP activity in lung extracts of normal
(N) and Mtb infected mice at week 1 (1–1 to 1–7)
and week 2 (2–1 to 2–6) post-infection by addition
of 10 mM APMA. No APMA (Φ), APMA (Ε).
Induction of lung gelatinases by Mycobacterium tuberculosis 235
(a) (a) U937 THP-1
U937 Sup. U937 Cell Ext.

Unstimulated

Erdman dead

Erdman dead
Erdman live

Erdman live
H37Ra dead
H37Ra live
Unstimulated

Unstimulated

Blank
Erdman

Erdman
H37Ra

H37Ra
kDa
kDa

92
92
72 72

(b) U937 THP-1

(b)
3000

manLAM

manLAM
araLAM

araLAM

MMP-9
Blank
PMA
2500
kDa
Densitometry units

2000

1500 92

1000

500 Figure 4. MMP-9 is induced by formalin-fixed Mtb

0 of U937 and THP-1 monocyte culture supernatants
Unstimulated H37Ra
Treatment
Figure 3. Mtb bacilli stimulate MMP-9 in U937 cells.
(a) Gelatin zymography of cell supernatants and
extracts of U937 cells infected with attenuated H37Ra
and virulent Erdman Mtb strains, showing a band
of gelatinolytic activity of 92 kDa corresponding to
MMP-9. Both attenuated and virulent Mtb strains
induced the expression of MMP-9 in U937 cells and
the activity of MMP-9 was higher in culture super-
natants than in detergent cell extracts. (b) Densi-
tometry scan of gel zymography showing a two-fold
increase in MMP-9 activity with H37Ra and Erdman
Mtb strains over the unstimulated control and a
0.3-fold higher activity in culture supernatants over
detergent cell extracts. (Φ) U937 Sup.; (Ε) U937 Cell
ext.
monocytes at levels similar to live bacilli [Fig.
4(a)]. Unstimulated U937 (shown) and THP-1
(not shown) monocytes produced very low
levels of MMP-9. These experiments dem-
onstrate that the inducer of MMP activity on the
cell surface of Mtb bacilli is biologically active
even after crosslinking the cell envelope proteins
with paraformaldehyde.
bacilli and LAM glycolipids. (a) Gelatin zymography
Erdman 12 h after incubation with live and para-
formaldehyde-fixed Mtb bacilli, showing a band of
gelatinolytic activity of 92 kDa corresponding to
MMP-9. (b) Gelatin zymography of U937 and THP-
1 monocyte culture supernatants 12 h after culture
in the presence of purified LAM glycolipids (1 mg/
ml) manLAM and araLAM, showing a band of gel-
atinolytic activity of 92 kDa corresponding to MMP-
9. As a control, U937 cells stimulated with PMA show
induction of MMP-9. The 92 kDa band of gelatinolitic
activity is shown to comigrate with purified MMP-
9.
To further determine which of the components
of the cell wall in Mtb is capable of inducing
MMP activity in monocytic cells, we tested puri-
fied Mtb cell wall lipoarabinomannan glyco-
lipids araLAM and manLAM. As shown in Fig.
4(b), araLAM or manLAM did not induce the
expression of MMP-9 in U937 cells, but induced
high levels of MMP-9 in THP-1 monocytes. How-
ever, both U937 (shown) and THP-1 (not shown)
monocytes were responsive since they produced
high levels of MMP-9 when stimulated by PMA.
MMP-9 expression in U937 cells could only be
induced by exposure to long term culture fil-
trates of Mtb and live or fixed killed bacilli. These
results demonstrate that while LAM glycolipids
236 C. A. Rivera-Marrero et al.

can induce MMP-9 expression in THP-1 cells, (a)

other components or molecules in their native 0.25
conformation might be needed for stimulation 0.20
670 K

Protein (mg/ml)
of U937 cells. 158 K
0.15 44 K

0.10 17 K
Separation of MMP-9 inducing activity by
column chromatography 0.05
1.35 K
Further characterization of the components in 0
Mtb with gelatinase-inducing activity was car- 5 10 15 20 25 30
ried out by size exclusion column chromato- Volume (ml)
graphy. A total protein detergent extract of Mtb (b)
was fractionated on a Sephacryl S-200 column
500
and two main peaks of protein material were

Densitometry units
obtained [Fig. 5(a)]. Fractions eluted from the 400
column were incubated with THP-1 cells and
MMP activity determined by gelatin zym- 300
ography and densitometry analysis. As shown 200
in Fig. 5(b), fractions eluted at 6.3–12.9 ml, with
an estimated molecular weight larger than 100
156 kDa and obtained before elution of the first
0
protein peak, showed MMP-9 inducing activity

19 .2
18.8

20 .5
22 .8
11.2
12.5
14.9
15.2
16 5

23.1
24.4
26.8
27.0
.4
6
10.9
3

in THP-1 cells. However, the highest activity

.
7.
6.

was detected at an elution volume of 11.5 ml and

corresponded to an estimated m.w. of 156 kDa.
Fractions obtained at an elution volume of
20.8–24.8 ml and with an estimated m.w. of
17 kDa also induced MMP-9 activity in THP-1
cells.
Due to the pervasive nature of lipo-
arabinomannan, lipomannan (LM) and pho-
phatidylinositol mannosides (PIMs), it is
possible that the fractions obtained by size ex-
clusion chromatography contained these
lipoglycans. To separate these from the gly-
coprotein fraction, protein extracts were sep-
arated by ion exchange chromatography (DEAE
Sephacel). The unbound material was eluted in
PBS, while the bound material was eluted with
0.5 M NaCl phosphate buffer. Analysis of these
fractions by gelatin zymography showed that
the peak of bound material, devoid of neutral
lipoglycans and containing the glycoprotein
fraction, induced approximately three-fold more
MMP-9 activity in comparison to the unbound
material (data not shown). These results dem-
onstrate that besides LAM glycolipids, other
components of the mycobacterial cell wall can
induce the expression of MMPs in monocytic
cells.
To further examine components of Mtb that
are capable of inducing monocytic cells to pro-
duce MMP-9, several purified Mtb cell mem-
brane and cytosolic materials (obtained from Dr
MMP-9
Column fractions (ml)
Figure 5. Separation of Mtb proteins by column
chromatography and identification of fractions with
MMP-9 inducing activity. (a) Elution profile of Mtb
soluble protein extract after separation by Sephacryl
S-200 size exclusion chromatography. Two major
peaks of protein with molecular weight in the
10–200 kDa size range are shown. A calibration curve
with known molecular weight markers is indicated.
(b) Gelatin zymography (bottom) and densitometry
scan (top) of THP-1 cell extracts after incubation
with Mtb protein fractions obtained by size exclusion
chromatography. The highest MMP-9 inducing ac-
tivity was detected at elution volumes of 11.5 ml
and 20.8 to 24.8 ml, corresponding to proteins with
estimated m.w. of 158 and 17 kDa, respectively.
Belisle, CSU, CO, U.S.A.) were tested for their
induction of gelatinolytic activity in U937 and
THP-1 cells using gelatin zymography. Table 1
summarizes our results with these reagents as
well as those described above for LAM and
whole bacilli. As expected, cell wall fractions
of H37Rv were found to induce gelatinolytic
activity in both U937s and THP-1. LAM-free
Induction of lung gelatinases by Mycobacterium tuberculosis
Table 1. Induction of MMP-9 mediated gelatinolytic
activity by components of M. tuberculosis1
Cells
Component U937 THP-1
Cell wall + +++
Culture filtrate (-LAM2) + +
Cytosol (-LAM2) — +
AraLAM — +
ManLAM — +
TDM3 — —
Bacilli ++ +++
1
Data based on densitometric scanning of gel zymograms.
2
Lipoarabinomannan.
3
Trehalose-6,6′-dimycolate.
culture filtrates of M. tuberculosis also induced
gelatinolytic activity in both cell lines but to
a lower extent. LAM-free cytosolic fractions,
however, stimulated gelatinolytic activity in
THP-1 cells, but not in U937 cells. This is similar
to our findings with LAM. The polar cell wall
glycolipid trehalose-6,6′-dimycolate (TDM), also
known as mycobacterial cord factor, had no
activity against either cell line. Overall, these
data suggest that LAM as well as LAM-free
components can induce gelatinolytic activity.
Discussion and Conclusions
Tissue inflammation and destruction due to con-
nective tissue degradation is an important con-
sequence of mycobacterial infection and has
therefore become the target of intense in-
vestigation [7–9, 19–21]. Some clues about the
mechanisms involved are derived from studies
performed in animal models of Mtb infection.
For example, pulmonary cavities have been
demonstrated in rabbits 2–4 weeks after injection
of living as well as dead mycobacteria, or the
particulate and lipid-protein fractions of tubercle
bacilli [22]. These treatments were all effective
in producing pulmonary cavitation in animals
previously sensitized by subcutaneous injection
of heat-killed bacilli in adjuvant. However, in
non-sensitized and desensitized animals, cavity
formation was prevented [5, 23]. Similar results
were obtained in guinea-pigs. These early stud-
ies suggested that the cellular hypersensitivity
237
response elicited in the host to destroy the ba-
cillus was responsible for much of the de-
struction of host tissue by promoting caseous
necrosis and liquefaction [4, 23].
The aforementioned observations led several
researchers to investigate the tissue enzymes/
proteases involved in the host response to Mtb
infection. The studies performed in the laborat-
ories of Weiss [8, 9, 20, 21] and Dannenberg [6,
7] focused on the hydrolytic enzymes involved
in liquefaction. Weiss and others demonstrated
that tuberculous granulation tissue of rabbits
infected with virulent M. bovis contained high
concentrations of protease and nuclease ac-
tivities. Normal lung tissue and recently ca-
seating tissue showed moderate enzyme
activities, while older caseous tissue and li-
quefied tissue showed low or absent activities.
Dannenberg et al. demonstrated the presence of
cathepsin D, RNase, DNase, hyaluronidase and
other enzymes in dead and live macrophages
surrounding the caseous centers. The activity
of these enzymes diminished in fully liquefied
areas. Lipases and phospholipases have also
been identified in tuberculous tissues [19]. More
recently, others found evidence for activation of
the plasminogen/plasmin system in granu-
lomatous inflammation caused by Mtb [10].
Some authors have also postulated that enzyme
inhibitors derived from the bacilli and necrotic
host tissues exist in caseous material and that
these prevent tissue liquefaction for a number
of weeks [9, 21]. However, to our knowledge,
no such inhibitors have been identified.
Although toxic components released by dead
bacilli are considered important, lymphokines
produced from T cells and other activated host
cells (e.g. monocyte/macrophage), and com-
ponents such as hydrogen peroxide, toxic fatty
acids, cationic protein, lipases, phospholipases,
nucleases and antigen–antibody complexes,
have all been implicated in the tissue destruction
process [24]. More recently, attention has been
given to apoptosis, the process by which several
external signals (e.g. proinflammatory cyto-
kines) can activate a pathway of cell death in
the absence of necrosis [25, 26].
This report implicates yet another group of
molecules, the MMPs, in the destructive process
that ensues during mycobacterial infection.
MMPs are zinc- and calcium-dependent en-
dopeptidases capable of degrading connective
tissue matrices [11, 12]. These enzymes are con-
sidered important for the normal tissue re-
modelling processes that occur during
238
development and wound healing, and for much
of the tissue destruction that characterize dis-
eases of the lung such as tumor metastasis and
emphysema [13, 27, 28]. To date, at least 14
MMPs have been identified [12] and these have
been classified into four major sub-families
including the interstitial collagenases, stro-
melysins, gelatinases and membrane-type
MMPs (MT-MMPs). Our work examines the po-
tential role of the gelatinases MMP-2 and MMP-
9 in Mtb infection. These proteases are known
to cleave denatured collagens (i.e. gelatin), type
IV collagen typically found in basement mem-
branes, types V, VII, X and XI collagens, elastin
and other ECM components [29]. Both of these
MMPs are encoded by genes located on human
chromosome 16. Like other MMPs, these pro-
teases are secreted as inactive zymogens and
secreted into the extracellular space where they
become active. Although these two MMPs have
similar substrate specificities, they are dif-
ferentially regulated at the transcriptional and
post-translational level [30].
A role for the gelatinases in Mtb infection was
suggested by Chang et al. (1996) who showed
that THP-1 human monocytes exposed to LAM
from attenuated mycobacteria or heat killed at-
tenuated H37Ra bacilli produced MMP-9 ex-
pression [18]. Our studies corroborate these
findings and expand them to include in vivo
studies in mice and the isolation of molecular
determinants in Mtb which induce MMP ex-
pression in monocytic cells. First, we observed
that the lungs of mice infected with a lethal
intravenous dose of virulent Mtb Erdman con-
tained increased amounts of gelatinolytic ac-
tivity. These observations suggest that MMPs
may indeed be involved in the host response of
the lung to Mtb infection in vivo. It should be
noted that low levels of MMP-9, but not MMP-
2, were detected in lungs prior to infection.
This is consistent with studies showing that
embryonic murine lungs show a progressive
increase in the production of MMP-9 and less
of MMP-2 with the former predominating in the
newborn (Roman et al., unpublished ob-
servations). In Mtb infected lungs, the observed
increase in MMP-9 at week 1 post-infection,
before bacilli are found elevated in lung tissue,
suggests that its expression and/or activation
could be due to a generalized inflammatory
response to the intravenous inoculum. It is
known that intravenous inoculation of Mtb ba-
cilli in mice results in lodging of more than 90%
of the bacilli into the liver and spleen. Only after
C. A. Rivera-Marrero et al.
2–4 weeks of infection, bacilli are increased in
the lung while multiplication within the liver
and spleen is suppressed [31]. The observed
increase in MMP-2 at 2 and 3 weeks post-in-
fection, which coincided with the highest ba-
cillary load (cfu) in the lungs, may therefore
be a specific response to cellular invasion and
intracellular replication of Mtb in lung tissue.
To determine if the differential expression of
MMPs occurs by regulation of message pro-
duction, the mRNA levels for MMP-9, MMP-2
and their respective inhibitors TIMP-1 and
TIMP-2, were determined by quantitative RT-
PCR during Mtb infection. Only a small increase
in MMP-9 mRNA was detected after week 1,
while no changes in TIMP-1 were detected dur-
ing the course of infection. MMP-2 mRNA levels
were also unchanged, but a one-fold decrease
in TIMP-2 expression was detected at week 4
post-infection. The decrease in TIMP-2 ex-
pression could explain the observed increase
in MMP-2-related activity in lung extracts 2–3
weeks post-infection. It is possible that during
the course of Mtb infection, a delicate balance
in MMP and TIMP expression plays a significant
role in the pathogenesis of tuberculosis.
Another possible mechanism of regulation of
MMP activity is in situ activation by cleavage
of their pro-peptide domain after secretion by
extracellular enzymes. MMP pro-peptides con-
tain a free cysteine residue that is thought to
associate with the active-site zinc molecule re-
sulting in latency [11]. Activation of these pro-
teinases occurs when this association is
disrupted causing the so-called cysteine switch,
which results in cleavage of the bond and ac-
tivation of the enzyme. Cleavage and activation
can occur via the action of serine proteases as
well as by other MMPs such as the membrane-
bound metalloproteinases (MT-MMP) [32, 33].
This type of activation can be mimicked in vitro
by addition of APMA. We showed that addition
of APMA to lung extracts of Mtb infected mice
results in a significant increase in MMP-9 ac-
tivity. This demonstrates that in lung extracts,
the total levels of MMP-9 are actually higher
than detected by gelatin zymography and that
activation of inactive MMPs by proteolytic cleav-
age at the site of inflammation could be involved
in their regulation during the course of Mtb
infection.
Although MMPs are produced by many dif-
ferent cell types [16, 32], we postulated that
much of the MMP production observed in the
lungs of Mtb-infected animals was derived from
Induction of lung gelatinases by Mycobacterium tuberculosis
monocytic/macrophages resident in the lung
or recruited after infection. To investigate the
cellular mechanisms responsible for induction
of MMPs by Mtb, Mtb bacilli and cell free ex-
tracts were tested under various conditions in
two types of human monocytic cells, the U937
and THP-1 cell lines. The gelatinolytic activity
originated in the monocytic cells and was in-
duced by both attenuated and virulent strains.
Our initial experiments demonstrated that the
MMP inducer in Mtb was found in Mtb bacilli
and in long term culture filtrates of Mtb sug-
gesting the secreted nature of this molecule (not
shown). However, in transwell plate assays,
where Mtb bacilli and monocytes were separated
by a 0.4
M semipermeable membrane, no in-
duction of MMP activity was detected during
24 h of exposure (not shown). Also, the inducer
of MMP-9 activity in Mtb bacilli was active after
fixation of bacilli with paraformaldehyde. These
experiments suggest that the inducer of MMP
in Mtb bacilli is a cell associated molecule that
acts by cell to cell contact that is shed into
the culture medium during long term growth.
Furthermore, they show that the inducer of
MMP activity on the cell surface of Mtb bacilli
is biologically active even after crosslinking the
cell envelope proteins with paraformaldehyde.
To further determine which of the components
of the cell wall in Mtb is capable of inducing
MMP activity in monocytic cells, we tested the
purified Mtb cell wall glycolipid lipo-
arabinomannan (LAM) in uncapped (araLAM)
and mannose capped form (manLAM). LAM, a
phosphatidyl-inositol membrane bound glyc-
olipid [34] consisting of a linear backbone of 1-6
and 1-2 linked mannoses with highly branched
arabinose side chains [35, 36], is considered a
key cell surface molecule in host–pathogen inter-
actions and its biological and immunological
properties of LAM have been studied ex-
tensively [37–44]. Initial structural studies dem-
onstrated that the non-reducing end of the
arabinose chains in the virulent strain Erdman
were capped with 1-2 linked mannose residues
[36], while the capping was missing from LAM
isolated from a rapidly growing Mtb strain ini-
tially described as the avirulent H37Ra strain
[35]. In vitro, uncapped LAM (AraLAM)
triggered tumor necrosis factor release from in-
fected murine macrophages thereby enhancing
their bacteriostatic activity, whereas the man-
nose capped LAM (ManLAM) from the virulent
strain elicited low levels of TNF that did not
239
restrain bacterial growth [45]. However, the im-
plication that terminal 1,2 mannose residues
could function as virulence factors has not been
established since in further studies it was dem-
onstrated that LAM from avirulent M. bovis
BCG were mannose capped [46, 47]. In our
studies, araLAM and manLAM equally stim-
ulated MMP-9 expression in THP-1 cells, but
not in U937 monocytes. These results dem-
onstrate that the terminal 1-2 linked mannose
residues in LAM are not critical for MMP-9
induction in THP-1 cells. The finding that MMP-
9 expression in U937 cells is not induced by
LAM, but only by exposure to long term culture
filtrates of Mtb and live or fixed-killed bacilli,
suggest that other components or molecules in
their native conformation might be needed for
stimulation of these cells. It also suggests that
subtle differences in cellular differentiation be-
tween U937 and THP-1 monocytes, perhaps de-
termining the presence of specific receptors, are
important for cell responsiveness to Mtb and
MMP expression.
We attempted to isolate the inducer of MMP
in Mtb by gel filtration and ion exchange
chromatography. A detergent extract of Mtb was
prepared and the protein fraction precipitated
in cold acetone and separated by Sephacryl S-
200 chromatography. Fractions with an es-
timated m.w. of 158 and 17 kDa induced MMP-
9 expression in THP-1 cells. In addition, protein
fraction isolated by DEAE ion exchange
chromatography and presumably devoid of
LAM and other neutral lipoglycans was shown
to induce MMP-9 in THP-1 cells. These results
demonstrate that Mtb may contain more that
one type of molecule capable of inducing MMP
expression in monocytic cells. To further ex-
amine the possible components of Mtb that are
capable of inducing monocytic cells to produce
MMP-9, several purified Mtb cell membrane
and cytosolic materials were tested for their
induction of gelatinolytic activity in U937 and
THP-1 cells using gelatin zymogrphy. The ob-
servation that LAM-free culture filtrates and
cytosolic fractions were capable of inducing gel-
atinolytic activity in both U937s and THP-1 sug-
gests that other components must be involved.
Further studies are underway to define the mo-
lecular determinants in Mtb, the mechanisms of
MMP induction, and role of gelatinases in the
pathogenesis of tuberculosis.
Overall, our work indicates the activation of
MMPs as one possible mechanism of connective
tissue destruction in Mtb infection. Further work
240
defining the mechanisms of MMP activation in
Mtb infection may accelerate the development
of novel strategies aimed at preventing or, at
least ameliorating, lung destruction and the loss
of lung function in subjects with mycobacterial
infection.
Materials and Methods
Bacterial cultures and growth conditions
M. tuberculosis laboratory strains H37Ra (ATCC
25177) and Erdman (ATCC 35801) were obtained
from American Type Culture Collection (Rock-
ville, MD, U.S.A.). Mycobacteria were grown
in Middlebrook 7H9 broth supplemented with
ADC enrichment (Difco Laboratories, Detroit,
MI, U.S.A.), 2% glucose and containing 1% gly-
cerol. Growth was at 37°C with slow shaking to
mid-log growth phase (7–10 days) or to sta-
tionary phase (21 days). Bacteria were passaged
in broth cultures once and quantified with
McFarland equivalence turbidity standards (Re-
mel, Lenexa, KS, U.S.A.). Cultures of 106–108
bacilli/ml were used immediately or stored at
−80°C in 7H9 broth containing 20% glycerol.
Killing of Mtb was done by fixation in para-
formaldehyde. Broth cultures of 10 ml con-
taining 106–108 bacilli/ml were centrifuged for
10 min at 3500×g and the bacterial pellets re-
suspended in PBS pH 7.2. After re-cen-
trifugation, the pellets were resuspended in
10 ml of PBS containing 4% paraformaldehyde
and incubated at 37°C for 1 h. After fixation, the
bacteria were centrifuged at 3000×g and the
pellets were resuspended in PBS pH 7.2. After
washing three times in PBS, the fixed myco-
bacteria were resuspended in 10 ml of serum
free RPMI 1640 and tested in MMP induction in
in vitro assays. Viability of fixed mycobacteria
was tested by plating on 7H9 agar plates.
Mouse infections
Groups of 6–8 week old C57BL/6 female mice,
3–7 mice/group, were injected in the tail vein
with 0.2 ml of Mtb Erdman culture, ap-
proximately 106 bacilli. Mice were killed after 1,
2, 3 and 4 weeks post-infection and the lungs
were removed for quantitation of mycobacteria,
determination of gelatinolytic activity, and de-
termination of gene expression by quantitative
C. A. Rivera-Marrero et al.
RT-PCR. Quantitation of mycobacteria was car-
ried out by homogenization of the organs in PBS
and plating serial dilutions (1:10, 1:100 and
1:1000) of the tissue homogenates in 7H9 agar
plates. After 3 weeks of incubation at 37°C, the
number of cfu was recorded for each dilution.
For determination of gelatinolytic activity in
lungs, the organs were surgically removed,
rinsed in sterile PBS and approximately a 1 mm
portion of tissue was dissected and placed in
2.0 ml of CHAPS buffer (0.15 M NaCl, 10 mM
CHAPS, 20 mM HEPES, pH 7.5, 1 mM EDTA,
2 mM PMSF) at 4°C for 1 h. For lungs, only the
upper left lobe was taken. The tissue was broken
up with a tissue grinder, centrifuged at 12 000×g
and the supernatants were collected. Total pro-
tein was determined and aliquots were stored
frozen at −80°C until tested by gelatin zym-
ography (see below).
M. tuberculosis extracts
M. tuberculosis extracts from long term culture
filtrates and total soluble protein, were prepared
and analysed for MMP induction. Long term
culture filtrates were prepared as follows: Mtb
4–6 week old cultures were harvested by cen-
trifugation for 10 min at 3500×g and the super-
natants filtered through a 0.22
M sterile filter.
The culture filtrates were concentrated on a
Nanospin 4K centrifuge filter (Gelman Sciences,
Ann Arbor, MI, U.S.A.) and protein con-
centration determined by the Bradford method
(Bio-Rad, Richmond, CA, U.S.A.). Culture fil-
trates were stored at −80°C until tested for
MMP induction.
Total soluble protein was extracted from Mtb
strains as follows: 10 ml of a bacterial broth
culture (3–6×108/ml) was centrifuged (10 min
at 3500×g) and the pellets washed twice in PBS,
pH 7.2. The bacterial pellets were resuspended
in 1.0 ml of PBS and sonicated (5 pulses/20 s
each) in ice in a bath sonicator. The cell extracts
were mixed with 1.0 ml of 2% SDS in PBS and
incubated at room temperature for 1 h. After
centrifugation (10 min at 3500×g), the super-
natant was collected, filtered through a 0.2
m
filter, and proteins precipitated by addition of 7
volumes of acetone at −70°C. Total protein,
devoid of lipids and detergent, was recovered
by centrifugation at 20 000×g for 30 min. The
protein pellet was air dried, resuspended in
Induction of lung gelatinases by Mycobacterium tuberculosis
200
l PBS, and the protein concentration de-
termined. Protein samples were stored at −80°C
for analysis of MMP induction.
The following purified Mtb cell membrane and
cytosolic materials were obtained from Dr John
T. Belisle, CSU, Fort Collins, CO, U.S.A.) through
NIH contract ‘‘TB Research Materials and Vaccine
Testing’’. Briefly, AraLAM and ManLAM were ob-
tained by breaking the bacilli with 4% Triton X-
114. After heating, the detergent layer was har-
vested and macromolecules including LAM were
collected by acetone precipitation. The proteins
and carbohydrates were then separated by ex-
traction with phenol. The crude carbohydrate mix-
ture was fractionated by size exclusion
chromatography and the pure LAM was pooled.
Buffer contaminants were removed by extensive
dialysis against water, and the solution was sterile
filtered and stored dry at −70°C until use. Endo-
toxin contamination was 12.3 ng/mg as de-
termined by the Limulus Amebocyte Assay.
AraLAM was isolated from rapid growing Myco-
bacterium spp. and filter sterilized (endotoxin con-
tamination, 2.44 ng/mg). These glycolipids were
resuspended in PBS to a concentration of 1 mg/
ml. M. tuberculosis cell wall fractions were obtained
from H37Rv. The bacilli were disrupted with
French press or probe sonication, followed by isol-
ation of the cell walls by centrifugation at 27 000×g
for 1 h, resuspension in PBS, and storage at −70°C
(Lot #97.Rv.2.8.6.9.CW). LAM-free culture filtrate
proteins were isolated by resuspending H37Rv
culture filtrates with Triton X-114, and the aqueous
and detergent layers separated. The aqueous
phase was re-extracted several times with 4%
Triton X-1124, precipitated with 10 volumes of
cold acetone, and the acetone precipitate washed
with 100% acetone to remove residual detergent.
The dried pellet was resuspended and dia-
lysed against 0.01 ammonium bicarbonate,
lyophilized and stored at −70°C until use (Lot
#98.Rv.2.7.8.10.Aq1). LAM-free cytosolic fractions
from H37Rv were obtained by suspending the cy-
tosolic fraction of gamma-irradiated bacilli in 4%
Triton X-114 as described above. Trehalose-6,6′-
dimycolate, obtained by petroleum ether ex-
traction from M. smegmatis, was a gift of Dr Kunu
Takayama (Madison, WI, U.S.A.).
Macrophage cultures and MMP induction
assays
U937 and THP-1 human monocytic cell lines were
grown at 37°C/5% CO2 in RPMI 1640 sup-
241
plemented with 10% heat inactivated fetal bo-
vine serum and 1% antibiotic-antimycotic
(100 U/ml penicillin G sodium, 100 U/ml strep-
tomycin sulfate, 0.25 g/ml amphotericin B). The
cells (2×106/ml) were washed in serum free
media and seeded in 15 ml T flasks containing
serum free media. Cells were infected with Mtb
strains H37Ra and Erdman at a multiplicity of
infection (moi) of 1–10. For studies with para-
formaldehyde-killed Mtb, bacilli were fixed for
1 h at room temperature, washed three times in
PBS pH 7.2, resuspended in serum free media
and added to the cells at a similar moi. Cells
and Mtb bacilli were incubated for 6, 12 and
24 h at 37°C/5% CO2. After these time points, a
20
l aliquot of culture was collected to deter-
mine cell viability by Trypan blue exclusion. The
infected cell cultures were centrifuged at 500×g
to pellet the cells and the culture supernatant
was collected and filtered through a 0.22
M
filter. Cell pellets were resuspended in 0.5 ml of
CHAPS buffer at 4°C, sonicated, centrifuged at
12 000×g for 5 min, and the supernatants were
collected and filtered through a 0.22
M filter.
Total protein in the culture supernatants and
detergent cell extracts was determined and ali-
quots were stored frozen at −80°C until ana-
lysed for the presence of MMPs by gelatin
zymography.
Mtb culture filtrates, total soluble protein and
purified glycolipids (AraLAM, ManLAM) were
tested for in vitro induction of gelatinolytic ac-
tivity in U937 and THP-1 cells. Culture filtrates
were added to cells (2×106 cells/ml) at a protein
concentration of 30
g/ml per mL. Purified
glycolipids AraLAM and ManLAM were tested
at a concentration of 1.0
g/ml. Cells were in-
cubated with the different Mtb fractions for 24 h
and viability was determined. After incubation,
the cells and culture supernatant were collected
and tested for the presence of gelatinolytic ac-
tivity by gelatin zymography.
Transwell (Costar, Cambridge, MA, U.S.A.)
plate assays were used to evaluate the role of
cell–cell contact in MMP induction (0.4
M,
24 mm Transwell polycarbonate membrane cell
culture chambers; Cat. No. 3412). U937 and THP-
1 cells, 2×106 cells in 1.0 ml of serum free media,
were added to the bottom chambers. Mtb bacilli
were added to the top chamber at an moi of
1–10 in 1.0 ml of serum free media. Culture
chambers were incubated at 37°C/5% CO2 for
24 h and the culture media in the bottom cham-
bers collected for analysis of gelatinolytic ac-
tivity.
242
Gelatin zymography
To measure gelatinolytic activity, gelatin zym-
ography was performed using a 9% SDS-PAGE
gel saturated with 1mg/ml gelatin (Sigma, 300
bloom), as previously described [48]. Samples
with equal protein concentration (30
g) were
loaded onto the gel and electrophoresed at a
constant 150 V for 1.5 h. The gels were incubated
for 1 h at room temperature in 2.5% Triton X-
100, followed by an overnight incubation at
37°C in gelatinase substrate buffer (50 mM Tris,
10 mM CaCl2, 0.02% NaN2, pH 8.0). The gels
were stained with 0.5% Coomasie Blue followed
by subsequent destaining with 50% methanol.
The gels were dried onto cellophane and
scanned under a densitometer for semi-
quantitative determination of MMP activity.
Quantitative RT-PCR
Determination of mRNA levels was carried out
by a quantitative bioluminescence-based RT-
PCR, as previously described [49, 50]. RNA was
extracted from lung tissue and cultured cells
using the reagent RNAzolTM (Tel-test Inc.,
Friendswood, TX, U.S.A.) followed by puri-
fication with serial extractions in chloroform,
isopropanol and ethanol. Biotinylated PCR
primers and digoxigenin-labelled probes for
murine MMP-9, MMP-2, TIMP-1 and TIMP-2
were synthesized by Genosys Biotechnologies,
Inc. (The Woodlands, TX, U.S.A.). Primers and
probes for MMP-9 and TIMP-1 were synthesized
based on GeneBank published sequences.
Reverse transcription of mRNA was done
using the Promega Access RT-PCR kit (Promega
Corp., Madison, WI, U.S.A.). The reaction mix-
ture consisted of: RNA sample (1
g), 2.5
l
dNTP mix (2.5 mM), 5.0
l RT buffer (250 mM
Tris-HCl pH 8.3, 375 mM KCl, 15 mM MgCl2),
2.0
l DTT (0.1 M), 0.5
l RNase inhibitor (20 U),
2.0
l random hexamer oligonucleotides and
DEPC treated sterile water, in a total volume of
25
l. The reaction mixture was heated at 70°C
for 5 min, cooled in ice, and 1
l (50–100 U) of
Superscript II reverse transcriptase was added.
The first strand cDNA synthesis reaction was
carried out at 42°C for 1 h. After cDNA synthesis,
the reaction mixture was diluted by adding
175
l of sterile water and a 5
l aliquot with-
drawn and added to PCR reaction tubes con-
taining: 2.5
l of 10× PCR buffer (100 mM Tris-
HCl pH 8.3, 500 mM KCl, 0.01% w/v gelatin),
C. A. Rivera-Marrero et al.
1
l of each biotinylated primer set (2
M), 1
l
dNTP mix (2.5 mM), 1.52 mM MgCl2, and 0.25
l
(0.5 U) of Taq DNA polymerase (AmpliTaq,
Perkin Elmer) in a total volume of 25
l. The
thermal cycling parameters consisted of: 95°C
(3 min), 24 cycles of 94°C (30 s), 54°C (30 s), 72°C
(1 min) and a final extension at 72°C (7 min).
After amplification, the biotinylated PCR
products were denatured by addition of de-
naturing solution (6.5
l of 1 M NaOH, 0.2 M
EDTA), neutralized by addition of neutralizing
buffer (31.5
l of 0.15 M sodium phosphate
pH 6.0), and added to streptavidin-coated plates
(Boehringer Mannheim). To each well 100
l of
hybridization buffer (62.5 mM Na2HPO4, 94 mM
citric acid, 15 mM sodium azide, 0.0625% BSA,
94 mM NaCl, 0.0125% Tween 20, 10 mM MgCl2)
containing 2 ng of the digoxigenin-labelled
probes. Hybridization was carried out for 2 h at
42°C temperature. After hybridization, the plates
were washed four times in wash buffer (PBS,
0.15% Tween 20, 2 mM EDTA, pH 7.4) and 100
l
of assay buffer (25 m Tris Base, 10 mM EDTA,
0.15 M KCl, 15 mM sodium azide, 2 mg/ml BSA)
containing 1
l of Anti-digoxigenin aequorin
conjugate (AquaLite, SeaLite Sciences, Nor-
cross, GA, U.S.A.) was added to each well. The
plates were incubated for 30 min at room tem-
perature, washed in wash buffer four times and
100
l of trigger buffer added to each well. Lu-
minescence was measured on a Dynatech ML
3000 Microtitre Plate Luminometer after trig-
gering with 0.1 M calcium. Results were plotted
as Relative Light Units (RLU). Standardization
reactions were done for the -actin gene. Op-
timization studies were performed to ensure
that for the amounts of cDNA being amplified
the reaction had not reached the plateau of the
amplification curve at a constant number of
cycles for any primer pair. Negative controls
consisted of RNA without RT-PCR products for
MMPs and -actin cDNA.
Column chromatography
Gel filtration chromatography was performed
on a 1×30 cm column of Sephacryl S-200 equi-
librated in 0.01 M PBS, pH 7.0, and calibrated
with known molecular weight markers (Sigma,
St. Louis, MO, U.S.A.). The void volume, de-
termined with blue dextran 2000 (Sigma), was
5.0 ml. A soluble protein extract prepared from
Mtb Erdman (4.0 mg in 2.0 ml PBS, pH 7.2) was
applied to the column followed with 3.0 ml of
Induction of lung gelatinases by Mycobacterium tuberculosis
3% sucrose in PBS/0.02% sodium azide. Frac-
tionation was carried out with an Isco Wizard
peristaltic pump and degassed PBS at a flow
rate of 20 ml/min. The elution profile was mon-
itored by absorbance at 280 nm with an Isco V4
monitor and recorder. Fractions of 3.3 ml were
collected and the protein concentration meas-
ured by the Bradford method [51]. Fractions
corresponding to the various peaks obtained
were pooled and stored at −20°C for MMP
activity determination by gelatin zymography.
Mtb cell extracts (0.5 mg in 0.2 ml) PBS were
also separated in a DEAE Sephacel (Sigma) ion
exchange chromatography column containing
25 ml of bead volume and pre-equilibrated in
PBS pH 7.2. After elution of unbound material
with 30 ml PBS pH 7.2, the column was washed
with 50 ml of phosphate buffer and the bound
material eluted with 0.5 M NaCl in phosphate
buffer at a flow rate of 1 ml/min. The peak
fractions of unbound and bound material were
collected, concentrated by filtration and stored
at −20°C for gelatinolytic activity determination
by gelatin zymography.
Acknowledgements
We thank Dr Frederick D. Quinn from the Centers
for Disease Control and Prevention (Atlanta, GA,
U.S.A.) for critical review of the manuscript and for
helpful discussions throughout the course of this
study. We thank Mark Burroughs and Geoffrey Ste-
vens for technical assistance in mouse infections and
cfu determinations. This work was supported by
National Institutes of Health Grant 1 RO1 AI37937
(to JR) and a minority supplement award (to C.A.R-
M.).
 2000 U.S. Government
References
1 Enarson DA, Murray JF. Global epidemiology of tuber-
culosis. Tuberculosis. In: Rom WN, Gray SM, Eds.
Tuberculosis. Boston: Little Brown, 1996: 57–75.
2 Garay S. Pulmonary tuberculosis. In: Rom WN, Gray S,
Eds. Tuberculosis. Boston: Little & Brown, 1996: 373–412.
3 Jagirdar J, Zagzag D. Pathology and Insights into Patho-
genesis of Tuberculosis. In: Rom WN, Gray S, Eds.
Tuberculosis. Boston: Little & Brown, 1996: 467–91.
4 Dannenberg AM. Chemical and enzymatic host factors
in resistance to tuberculosis. In: Kubica GP, Wayne LG,
Eds.The Mycobacteria. New York: Marcel Dekker, Inc.,
1984: 721–60.
243
5 Yamamura Y. The pathogenesis of tuberculous cavities.
Adv Tuberc Res. 1958; 9: 13–37.
6 Rojas-Espinosa O, Dannerberg JAM, Sternberger LA,
Tsuda T. The role of cathepsin D in the pathogenesis
of tuberculosis: A histochemical study employing un-
labeled antibodies and the peroxidase-antiperoxidase
complex. Am J Pathol 1974; 74: 1–12.
7 Tsuda T, Dannerberg JAM, Ando M, Rojas-Espinosa O,
Shima K. Enzymes in tuberculous lesions hydrolyzing
protein, hyaluronic acid and chondroitin sulfate. A
study of isolated macrophages and developing and
healing rabbit BCG lesions with substrate film tech-
niques: The shift of enzyme pH optima towards neut-
rality in ‘‘intact’’ cells and tissues. J Reticuloendothel Soc
1974; 16: 220–31.
8 Weiss C, Tabachnik J. Mechanisms of softening of
tubercles. IV. Digestion of caseous tubercles by a pro-
teinase extracted from polymorphonuclear leukocytes.
Arch Pathol 1956; 61: 76–83.
9 Weiss C, Tabachnik J, Cohen HP. Mechanism of soft-
ening of tubercles. III. Hydrolysis of protein and nucleic
acid during anaerobic autolysis of normal and tuber-
culous lung tissue in vitro. Arch Pathol 1954; 57: 179–93.
10 Gordon S, Cohn ZA. Bacille-Calmette-Guerin infection
in the mouse: Regulation of macrophage plasminogen
activator by T lymphocytes and specific antigen. J Exp
Med 1978; 147: 1175–87.
11 Birkedal-Hansen H, Moore WG, Windsor LJ. Matrix
metalloproteinases. A review. Crit Rev Oral Biol Med
1993; 4: 197–250.
12 Murphy G, Docherty AJP. The matrix metallo-
proteinases and their inhibitors. Am J Respir Cell Mol
Biol 1992; 7: 120–5.
13 D’Armiento J, Dalal SS, Okada Y, Berg RA, Chada K.
Collagenase expression of the lungs of transgenic mice
causes pulmonary emphysema. Cell 1992; 71: 955–61.
14 Delacourt C, Bourgeois ML, D’Ortho MP. Imbalance
between 95 kDa type IV collagenase and tissue inhibitor
of metalloproteinases in sputum of patients with cystic
fibrosis. Am J Respir Crit Care Med 1995; 152: 765–74.
15 Montano M, Ramos C, Gonzalez G, Vadillo F, Pardo A,
Selman M. Lung collagenase inhibitors and spon-
taneous and latent collagenase activity in idiopathic
pulmonary fibrosis and hypersensitivity pneumonitis.
Chest 1988; 106: 1115–9.
16 O’Connor C, Odium C, Van Breda A, Power C, Fitz-
gerald MX. Collagenase and fibronectin in broncho-
alveolar lavage fluid in patients with sarcoidosis. Thorax
1988; 43: 393–400.
17 Sepper R, Konttinen YT, Sorsa T, Koski H. Gelatinolytic
and type IV collagenolytic activity in bronchiectasis.
Chest 1994; 106: 1129–33.
18 Chang JC, Wysocki A, Tchou-Wong K-M. Effect of
Mycobacterium tuberculosis and its components on
macrophages and the release of matrix metallo-
proteinases. Thorax 1996; 51: 306–11.
19 Oshima Y, Hagihara T, Mizuta Y, Hattori S, Mishima
J. Histochemical study on the relationship between
tuberculosis and lipase activity. I. Lipase and lipids of
tuberculous lesions. Kekkaku 1955; 30: 716–21.
20 Weiss C, Boyar-Manstein ML. On the mechanism of
liquefaction of tubercles. I Am Rev Tuberc 1951 63:
694–705.
21 Weiss C, Singer FM. Mechanisms of softening of
tubercles. II. Behavior of desoxyribonuclease in
244
tubercles developing in the lungs of rabbits. Arch Pathol
1953; 55: 516–30.
22 Maeda H, Yamamura Y, Ogawa Y, Maeda J, Yamamura
Y. Mycobacterial antigens relating to experimental pul-
monary cavity formation. Am Rev Respir Dis 1977; 115:
617–23.
23 Yamamura Y, Ogawa Y, Maeda H, Yamamura Y. Pre-
vention of tuberculous cavity formation by de-
sensitization with tuberculin-active peptide. Am Rev
Respir Dis 1974; 109: 594–601.
24 Munger JS, Chapman JHA. Tissue destruction by pro-
teases. In Rom WN, Gray SM, Eds. Tuberculosis. Boston:
Little, Brown & Company, 1996: 353–61.
25 Gan H, Newman GW, Remold HG. Plasminogen ac-
tivator inhibitor type 2 prevents programmed cell death
of human macrophages infected with Mycobacterium
avium, serovar 4. J Immunol 1995; 155: 1304–15.
26 Keane J, Balcewicz-Sablinska MK, Remold HG, Chupp
GL, Mee BB, Fenton MJ, Kornfeld H. Infection by
Mycobacterium tuberculosis promotes human alvelar
macrophage apoptosis. Infect Immun 1997; 65: 298–304.
27 Brown CC, Hembry RM, Reynolds JJ. Im-
munolocalization of metalloproteinases and their in-
hibitors in the rabbit growth plate. J Bone Joint Surg
1989; 71-A: 580–93.
28 Henney AM, Wakeley PR, et al. Localization of the
stromelyzin gene in atherosclerotic plaques by in situ
hybridization. Proc Natl Acad Sci 1991; 88: 8154–8.
29 Murphy G, Crabbe T. Gelatinases A and B. Methods in
Enzymology. New York: Academic Press, 1995: 470–83.
30 Corcoran ML, Hewitt RE, Kleiner DE, Stetler-Stevenson
WG. MMP-2: Expression, activation and inhibition. En-
zyme & Protein 1995; 49: 7–19.
31 McMurray DN, Collins FM, Dannenberg Jr, AM, Smith
DW. Pathogenesis of experimental tuberculosis in
animal models. In: Shinnick TM, Ed. Tuberculosis. New
York: Springer, 1996: 157–79.
32 Sato H, Takino T, Okada Y, Cao A, Shinagawa A,
Yamamoto E, Seiki M. A matrix metalloproteinase ex-
pressed on the surface of invasive tumour cells. Nature
1995 370: 61–5.
33 Will H, Hinzmann B. cDNA sequence and mRNA tissue
distribution of a novel human matrix metalloproteinase
with a potential transmembrane segment. Eur J Biochem
1995; 231: 602–8.
34 Hunter SW, Brennan PJ. Evidence for the presence of a
phosphatidylinositol anchor on the lipoarabinomannan
and lipomannan of Mycobacterium tuberculosis. J Biol
Chem 1990; 265: 9272–9.
35 Chatterjee D, Bozic C, McNeil M. Structural features of
the arabinan component of the lipoarabinomannan of
Mycobacterium tuberculosis. J Biol Chem 1991 266: 9652–
60.
36 Chatterjee D, Lowell K, Rivoire B, McNeil MR, Brennan
MJ. Lipoarabinomannan of Mycobacterium tuber-
culosis: Capping with mannosyl residue in some strains.
J Biol Chem 1992; 267: 6234–9.
37 Besra GS, Chatterjee D. Lipids and carbohydrates of
Mycobacterium tuberculosis. In: Bloom B, Ed. Tuber-
culosis, pathogenesis, protection and control. American So-
ciety for Microbiology: Washington, DC, 1994.
C. A. Rivera-Marrero et al.
38 McNeil MR, Brennan PJ. Structure, function and bio-
genesis of the cell envelope of mycobacteria in relation
to bacterial physiology, pathogenesis and drug re-
sistance; some thoughts and possibilities arising from
recent structural information. Res Microbiol 1991; 142:
451–63.
39 Ellner JJ, Daniel TM. Immunosuppression by myco-
bacterial arabinomannan. Clin Exp Immunol 1979; 35:
250–7.
40 Sibley LD, Hunter SW, Brennan PJ, Krahenbuhl JL.
Mycobacterial liparabinomannan inhibits gamma-in-
terferon-mediated activation of macrophages. Infect Im-
mun 1988; 56: 1232–6.
41 Moreno C, Mehlert A, Lamb J. The inhibitory effects of
mycobacterial lipoarabinomannan and polysaccharides
upon polyclonal human T-cell proliferation. Clin Exp
Immunol 1988; 74: 206–10.
42 Adams LB, Fukutomi Y, Krahenbuhl JL. Regulation of
murine macrophage effector functions by lipa-
rabinomannan from mycobacterial strains with dif-
ferent degrees of virulence. Infect Immun 1993; 61:
4173–81.
43 Barnes PF, Chatterjee D, Abrams JS. Cytokine pro-
duction induced by Mycobacterium tuberculosis lipo-
arabinomannan. Relationship to chemical structure. J
Immunol 1992; 149: 541–7.
44 Roach TIA, Barton CH, Chatterjee D, Blackwell JM.
Macrophage activation: lipoarabinomannan from avir-
ulent and virulent strains of Mycobacterium tuber-
culosis differentially induces the early genes c-fos, KC,
JE, and tumor necrosis factor-alpha. J Immunol 1993;
150: 1886–96.
45 Chatterjee D, Roberts AD, Lowell K, Brennan PJ, Orme
IM. Structral basis of capacity of lipoarabinomannan to
induce secretion of tumor necrosis factor. Infect Immun
1992 60: 1249–53.
46 Prinzis S, Chatterjee D, Brennan PJ. Structure and anti-
genicity of lipoarabinomannan from Mycobacterium
bovis BCG. J Gen Microbiol 1993; 139: 2649–58.
47 Venisse A, Berjeaud JM, Chaurand P, Gilleron M, Puzo
G. Structural features of lipoarabinomannan from
Mycobacterium bovis BCG. Determination of molecular
mass by laser desorption mass spectrometry. J Biol Chem
1993; 268: 12401–11.
48 Kleiner DE, Stetler-Steven. Quantitative zymography:
Detection of picogram quantities of gelatinases. Ana-
lytical Biochemistry 1994; 218: 325–9.
49 Svetic A, Finkleman FD, Jain YC, Diefenbach CW, Scott
E, McCarthy KF, Steinberg AD, Gause WC. Cytokine
gene expression after in vivo primary immunization
with goat antibody to mouse IgD antibody. J Clin
Immunol 1991; 147: 2391–7.
50 Wynn TA, Eltown I, Cheeves AW, Lewis FA, Gause
WC, Sher A. Analysis of cytokine mRNA expression
during primary granuloma formation induced by eggs
of Schistosoma mansoni. J Immunol 1993; 151: 1430–40.
51 Bradford MM. A rapid and sensitive method for the
quantification of microgram quantities of protein util-
izing the principle of protein-dye binding. Analytical
Biochemistry 1976; 72: 248–54.