JOURNAL OF PATHOLOGY, VOL.

180: 188-193 (1996)

AMPLIFICATION OF SPECIFIC mRNA FROM A

SINGLE HUMAN RENAL GLOMERULUS, WITH AN
APPROACH TO THE SEPARATION OF EPITHELIAL
CELL mRNA
GARETH R. BICKNELL, JACQUI A. SHAW, J. HOWARD PRINGLE AND PETER N. FURNESS

Univcrsify of Leicesfeu Department qf' Pathology, Robert Kirkpatrick Clinical Sciences Building, Lvicester Royal Infivnu~~.y,
P.O. Box 65, Leicester LE2 7LX, U.K.

SUMMARY
A method has been developed by which single human glomeruli may be plucked from fresh renal biopsies under direct vision, followed
by separation of mRNA using oligo-dT-linked paramagnetic beads. The mRNA was amplified by reverse transcription and polymerase
chain reaction (RT-PCR). Primers for a variety of human and rat proteins have been developed. The quantity of the amplified cDNA
was measured by an enzyme-linked immuno-sorbent assay (ELISA), where biotinylated forward strands of DNA were captured, probed
with a fluorescein-conjugated DNA oligomer, and then assayed with an enzyme-linked anti-fluorescein antibody. The cDNA-linked
beads are reported to be stable and can be reused with different primer sets, thus forming a 'bank' of samples from cases with defined
glomerular disorders, which can be used to address new questions as they arise. Using rat glomeruli, a method has been devised which
permits at least partial separation of epithelial cell mRNA from mesangial and endothelial cell mRNA.
KEY WORDS-reVerSe transcription-polymerase chain reaction; RT-PCR; glomerulus; renal; mRNA; matrix; collagen; metalloprotein-
ase; TIMP

INTRODUCTION sample for RT-PCR, as trauma such as mechanical

sieving or plucking of isolated glomeruli causes them to
Measurement of thc synthetic activities of cells in break off at a consistent point at the vascular p01e.l.~
biopsy material has many applications in research and Application of this method to glomeruli plucked from
diagnostic work. Measurement of the amount of a nephrectomy specimens has recently been d e ~ c r i b e d . ~ - ~
substance present, however, gives no indication of its Quantitation of this approach using animal models has
rate of synthesis. The best available methods of evaluat- been sufficient to detect changes of mRNA levels as low
ing synthetic rates in human tissue biopsies measure the as 1.7-f0ld.~
amount of specific mKNA sequences; the amount of In this paper we report that, as expected, human renal
inRNA coding for a specific protein is assumed to reflect biopsies can also be used as a source of single glomeruli
the rate at which the cell is making that protein. as starting material for semi-quantitative RT-PCR. We
Although some exceptions to this assumption have been describe the application of a convenient method of
reported, this approach has been widely accepted. mRNA extraction using oligo-dT-linked paramagnetic
Unfortunately, the methods used to detect tiny beads, and sets of primers, optimized for this method for
amounts of mRNA have serious limitations. In situ the detection of mRNA for human and rat actin, human
hybridization gives good localization, but it cannot collagen IV a5 (a type IV collagen chain which is
detect molecules which are present in very low copy relatively specific to the glomerular basement mem-
number and it does not give consistent quantitation. brane9), human matrix metalloproteinase 2, and human
Reverse transcription of the mRNA to cDNA followed tissue inhibitor or metalloproteinase (TIMP) 1 and 2.
by use of the polymerase chain reaction (RT-PCR) is These genes were studied to facilitate our investigation
more sensitive and can give better quantitation, but does of the extracellular matrix in the glomerulus.
not provide localization. The development of in sitti Whole glomeruli are not an ideal source of mRNA for
PCR methods may in due course provide excellent this purpose, as they normally contain three cell types:
localization, but quantitation is unlikely to be precise. epithelial, endothelial, and mesangial cells. We have
In view of the ability of PCR to detect specific taken advantage of the natural compartmentalization of
sequences in tiny samples, one strategy has been micro- the glomerulus provided by the basement membrane
dissection of tiny areas of tissue, often from tissue (GBM) to devise a strategy to obtain two separate
sections to provide a relatively well-defined starting mRNA preparations, from cells outside and inside the
sample.',' Renal glomeruli provide an appropriate GBM respectively. We therefore also detected rat
megalin (the core protein of gp330, the main Heymann
Addressee for correspondence: Dr P. N . Furiiess, University of
Leicester Department of Pathology, Robert Kirkpatrick Clinical
nephritis antigen) and rat Thy 1.1, as markers of
Sciences Building, Leicestcr Royal Infirmary, P.O. Box 65, Leicester glomerular epithelial and mesangial cells respectively, to
LE2 7LX. U.K. test our efforts to separate the mRNA from these cells.
CCC 0022-34 17/96/100188.-06 Reterved 13 December 1995
1996 by John Wiley & Sons, Ltd.
:(zx;j Accepted 29 Murch 1996
 RT-PCR AND SINGLE HUMAN GLOMERULl 189

MATERIALS AND METHODS 10 min at room temperature and washed as described

above.
Extraciion of mRNA from single human ~lo~neruli
Single glomeruli were plucked from fresh human renal
biopsies by hand, using forceps which had been ground Reverse transcription polymerase chain reaction
to a fine point, under direct vision using a dissection mRNA-linked Dynabeads were washed three
microscope. Glomeruli were obtained in less than 1 min times with reverse transcriptase buffer (Promega,
after the biopsy had been taken, after the adequacy of Southampton, UK) and then resuspended in the
the biopsy for diagnostic purposes had been confirmed. same buffer containing 1 mM DEPC-treated dNTPs
Glomeruli were immediately dropped into a lysisl (Pharmacia Biotech, St. Albans, UK), 25 U of RNAsin
binding buffer [Tris-HC1, pH 8, 100 mM; LiCl, 500 mM; (Promega), and 5 U of AMV reverse transcriptase
EDTA, pH 8, 10 mM; lithium dodecyl sulphate (LIDS) 1 (Promega). Priming was by the oligo-dT Dynabeads and
per cent; dithiothreitol (DTT), 5 mM; Dynal Ltd., incubation was for 1 h at 42°C. As a control for DNA
Bromborough, U.K.] and incubated with 50pg/ml pro- contamination, a similar reaction was set up without the
teinase K (Boehringer Mannheim, Lewes, UK) for 1 h at reverse transcriptase.
37°C. The lysate was centrifuged for 30 s at 10 000 g and cDNA-linked beads were washed once in Tris-EDTA
the supernatant mixed with oligo-dT-linked Dynabeads and once in PCR buffer [Tris, pH 8.8, 45mM;
(Dynal). The mRNA was allowed to anneal to the (NH4),S04, I 1 mM; MgCI,, 4.5 mM; dNTPs, 200pM;
Dynabeads for 10 min at room temperature. mRNA- ultrapure BSA (Advanced Protein Products Ltd.,
linked Dynabeads were washed twice in a buffer con- Brierley Hill, UK), 110 pg/ml; P-mercaptoethanol,
taining LiDS (Tris-HC1, pH 8, 1 0 m ~ LiC1, ; 0.15 M; 6.7 mM; EDTA, pH 8, 4 . 4 ~ before ~ 1 resuspension in
EDTA, 1 mM; LIDS, 0.1 per cent; Dynal) and three PCR buffer containing 10 pmol of forward primer and
times in the same buffer but without LIDS. 10 pmol of reverse primer. PCR was ‘hot started’ under
the following regime: one cycle of denaturation at 98°C
for 5 min, holding at 59°C pending addition of 1 U of
Diferential extraction of mRNA from rat glomeruli Taq (Gibco BRL, Paisley, UK), and then primer exten-
sion at 72°C for 30 s; four cycles of denaturation at 95°C
Pieces of kidney cortex from freshly killed rats were for 1 min, annealing at 59°C for 30 s, and primer
pressed sequentially through 50-, loo-, and 200-gauge extension at 72°C for 30 s; 35 cycles of denaturation at
stainless steel meshes, with liberal flushing with ice-cold 95°C for 30 s, annealing at 59°C for 30 s, and primer
PBS. The glomeruli were trapped on the 200-gauge mesh extension at 59°C for 30 s.
with less than 5 per cent contamination with tubular The primer sequences are shown in Table I. All
material. Glomeruli were flushed into a centrifuge tube primers, except those for actin, were designed using
and allowed to settle. They were then resuspended gently GCG Prime (Genetics Computer Group, Madison, WI,
to a small volume, such that the density of glomeruli was U.S.A.) on sequences obtained from the EMBL
10-20 in 1Opl. A sample of lop1 (approximately 15 database. They were designed preferentially from the 3’
glomeruli) was gently lysed for 2 min in modified end of the molecules in view of the use of oligo-dT
lydbinding buffer (Tris-HCl, pH 8, 100 mM; NaC1, Dynabeads as the mRNA capture method, and were
500 mM; EDTA, pH 8, 10 mM; Triton X-100, 0.1 per tested for homology with up to two mismatches against
cent; DTT, 5 mM). The buffer was modified so that this all human sequences in the EMBL database, Primers for
step would not disrupt the glomerular basement mem- p-actin, Thy 1.1, and megalin were synthesized by Oswel
brane. The lysate was separated from the glomeruli by DNA Sciences, Edinburgh, UK. All other primers and
gravity filtration through a 10pm polypropylene mesh probes were synthesized by Gibco BRL. Forward
(Whatman, Maidstone, U.K.). The glomeruli were then primers were biotinylated at source.
resuspended in modified lysidbinding buffer for 10 min,
and this second lysate was separated as above. Alterna-
ELISA quantitation of PCR product
tively, this second lysis was performed in modified
lysdbinding buffer for 1 h at 37°C in the presence of CovaLink plates (Life Technologies, Paisley, UK)
proteinase K (50puglml). The second lysate was separ- were biotinylated by overnight incubation with
ated from the glomeruli by filtration through the mesh. n-hydroxy-succinyl biotin (20 puglml in PBS) at room
The second, longer/more vigorous lysis was to release temperature. Plates were then washed three times with
mRNA from cells inside the glomerular basement mem- ‘buffer I’ (NaC1 2 M;MgSO, 40 mM; Tween 20,0.05 per
brane, either through increased leakage of mRNA cent vlv in PBS) and treated with avidin (50puglml in
through the hole at the base of the glomerulus or buffer 1) for 30 min at room temperature, with agitation.
through disruption of the basement membrane. Once Following three washes in ‘buffer 2’ (Tween 20, 0.02 per
separated from the glomeruli, any lysate that had not cent in PBS), plates were treated with 0.5 per cent
already been treated with proteinase K was incubated normal PBS/BSA for 15 min at room temperature, with
with 50pg/ml proteinase K for 1 h at 37°C to separate agitation. Amplified biotinylated DNA was then dis-
proteinaceous material from the mRNA. The lysates solved ]/I00 in PBS/BSA and allowed to bind to the
were centrifuged for 30 s at 10 000 g and the super- avidin-coated CovaLink plates for 1 h at room tempera-
natants mixed with oligo-dT-linked Dynabeads. The ture, with agitation. Unbiotinylated (reverse strand)
mRNA was allowed to anneal to the Dynabeads for DNA was denatured from the biotinylated (forward)
190 G . R. BICKNELL ET AL.

Table I-Primer and probe sequences

PCR primers
P-Actin" 5' GGAGACAAGC TTGCTCATCA CCATTGGCAA TGAGCG
/$Actin 3' GCGAATTCGA GCTCTAGAAG CATTTGCGGT GGACG
Thy- 1 5' CCGAGAGAAGAAGAAGCACG
Thy-1 3' TGGAGGAAGGAGAGGGAAAG
Megalin" 5' CCCAAATGCAAGTGTTCCAG
Megalin 3' CACCCCATTTCCATTTTCAG
C O L ( I V ) U ~5'' ~ TACAAGTGCA GGGGCAGAAG
COL(1V)US 3' TTGACATCGG CTAATTCGTG
MMp-215,'6 5' ATTGATGCGG TATACGAGGC
MMP-2 3' GGCACCCTTG AAGAAGTAGC
TIMP-Ii75' TGGGGACACC AGAAGTCAAC
TIMP-I 3' CAGGGGATGG ATAAACAGGG
TIMP-2I8 5' AACGACATTT ATGGCAACCC
TIMP-2 3' ACCTGTGGTT CAGGCTCTTC

ELISA probes
,&Actin GGAGTACTTG CGCTCAGGAG G
MMP-2 CTCCAGAATT TGTCTCCAGC
COL*(IV)aS GCTGTAGGAG TTGGCATAGT
TIMP-I GTAGTGATGT GCAAGAGTCC
TIMP-2 TCTATATCCT TCTCAGGCCC

I = Biotin

+HRP
# = Avidin
i
= Anti-fluorescein HRP-Antibody
= Biotinylated cDNA
f = Fluorescein Probe

= CovaLink Plate
Fig. 1---Schematic representation of the detection of amplified cDNA using ELISA. Biotin is covalently
linked to the CovaLink plate. Avidin is then bound to the biotin and is used to capture biotinylated cDNA.
A fluoresein-labelled probe specific to the cDNA in question is detected with an anti-fluorescein antibody, to
which horseradish peroxidase (HRP) has been conjugated. The HRP catalyses a colorimetric reaction

strand by addition of an equal volume of 0.25 M NaOH substrate for 10 min at room temperature. The colour
for 10 min. Probes had previously been labelled reaction was stopped by addition of an equal volume
with a 1:2 ratio of fluorescein-1 I-dUTP (Amersham of 1 M H,SO,. The optical density of the resulting
International, Amersham, UK): dATP by terminal solutions was read at 450 nm with a differential reading
deoxynucleotidyl transferase (Gibco) in a buffer contain- at 630 nm.
ing MnCI, (1 mM), sodium cacodylate ( 1 2 0 m ~ ) ,and
dithiothreitol (100 p ~ ) Plates
. were then washed three
times in buffer 2 and incubated with fluorescein- RESULTS
conjugated DNA probes (0.2 pmol/,ul in RapidHyb
buffer; Amersham) for li-2 h at 42°C (Fig. 1). Probe The five human sequences tested all gave bands on
sequences are shown in Table I. gel electrophoresis after 40 cycles of PCR, as shown in
Following three washes in buffer 2, plates were treated Fig. 2. These results were reproduced on five occasions.
with anti-fluorescein antibody conjugated with horse- The absence of a significant actin band in samples pro-
radish peroxidase (1:500 in PBS/BSA) for 30 min at cessed without reverse transcription (not shown) con-
room temperature, with agitation. After a final three firms that the bands are the result of amplified cDNA,
washes in buffer 2, plates were incubated with TMB rather than amplified genomic DNA contaminants.
 RT-PCR AND SINGLE HUMAN GLOMERULI 191

1.803
1.600 P

-eActin
+ w 2
-A- COL Na5

2 I +nwi
d
0.600
0.400
i -m- nw-2

Total 1st 2nd

Fig. 2-Amplification of five genes from cDNA extracted from a single

human glomerulus. DNA from the PCRs was mixed with a loading
buffer (1 nigh1 bromophenol blue, 1 mg/ml xylene cyanol FF, 10 per
cent glycerol, 1 x Tris-acetate EDTA) and loaded onto a gel of 2 per
cent ‘NuSieve GTG’ (Flowgen) / 1 per cent ‘Agarose GTG’ (ICN
Biomedicals) in Tris-acetate EDTA. DNA was visualized under ultra-
violet illumination with ethidium bromide. The sizes of fragments are
consistent with those predicted from the gene sequences. A DNA Fig. &Differential extraction of mRNA from glomeruli (short/long
standard of (pX174 Hue 111 digest (Promega) was also loaded in a lysis). mRNA extracted from whole glomeruli (Total) was compared
similar buffer, but without the xylene cyanol FF. The standard bands with that in the first extract (1st) and the second extract (2nd). The first
represent molecular weights equivalent to 1353, 1078, 872, 603, 310, extract contains a clear band for megalin (Mega: a podocyte marker),
271/281,234, 194, 118, and 72 base pairs. Abbreviations as used in text but Thy 1.1 (mesangial cell marker) is barely detectable. The second
extract contains a clear band for Thy 1.1, but not megalin. The sizes
of the bands are as predicted from the gene sequences. Act: Actin.
Using the ELISA assay, actin was clearly detectable Std: Molecular weight markers, as in Fig. 2
after only 30 cycles of PCR and a plateau effect was
observed at 3 5 4 0 cycles. COL(IV)a5, TIMP-1, and
TIMP-2 were detectable by 35 cycles, with further results consistently; contamination of the first extract
amplification at 40 cycles, while MMP-2 needed 40 with Thy 1-1 occurs occasionally and to a variable
cycles to give a signal significantly higher than the extent. This is presumably due to leakage of mesangial
background (Fig. 3). For relative quantitation of mRNA through the hole generated by separation of the
mRNA levels, our initial data suggest that actin will be glomerulus from the vascular pole. There is also a
an appropriate ‘housekeeping’ gene for comparison with problem with these markers, as Thy 1.1 is present in
high copy number mRNAs. Preliminary experiments much higher copy number than megalin.
suggest that porphobilinogen deaminase (PBGD) may
be more appropriate for low copy number rnRNAs
(data not shown). DISCUSSION
Using rat glomeruli, the use of an initial brief non-
ionic lysis buffer enabled differential separation of We believe that this is the first report of RT-PCR
glomerular epithelial cell mRNA (Fig. 4). The initial analysis of single human renal glomeruli from renal
lysate contained detectable megalin (a glomerular epi- biopsies, rather than pooled samples of glomeruli from
thelial cell marker) but little or no Thy 1.1 (a mesangial nephrectomy specimens5 or animal models.* The ability
cell marker), whereas the second extraction released to detect and measure the levels of specific mRNA
Thy 1.1 but contained no detectable megalin. Unfortu- species from glomeruli plucked from human renal
nately, we have not yet been able to produce such clean biopsies represents an exciting new approach to the
192 G. R. BlCKNELL ET AL.

investigation of renal disease. There are obvious appli- We can at present see no way in which the mRNA
cations, not only to improve our understanding of from the endothelial and mesangial cells can be
renal diseases, but also to investigate renal disease in separated, but consideration of the pathological pro-
individual patients. cesses seen in most glomerular disorders suggests
Comparison of mRNA levels for one gene product that this may be somewhat less important than the
with mRNA levels for a ‘housekeeping’ gene can provide separation which we have already partly achieved.
an approach which may be applicable to small biopsies Limitations of this technique must be recognized. It is
of other tissues. However, the renal glomerulus provides likely to be of less value in conditions which do not effect
an additional opportunity, because it represents a all glomeruli to a similar extent, as the morphology of
defined anatomical unit. If quantitation is made abso- the individual glomeruli used cannot be studied. Even
lute by applying competitive RT-PCR,8 then not only with the development of ELISA-based assay methods of
can one measure rates of protein synthesis by the whole product quantitation and competitive PCR techniques,
glomerulus, but also the ‘housekeeping gene’ becomes a quantitation is less precise than most biochemical assays
rapid measure of glomerular cellularity. The level of and is subject to occasional method failures. Neverthe-
genes constitutively expressed by one cell type (e.g., less, we feel that the method has such potential that we
macrophages) may become a measure of the number of are now collecting cDNA samples from most renal
that type of cell in one glomerulus. We hope that the biopsies. Dynabead-linked cDNA is relatively stable,
level of expression of collagen mRNA may correspond so we are able to develop a ‘bank’ of reusable cDNA,
to the rate of glomerular sclerosis, thus giving a new with each sample reflecting the synthetic activity of a
measure of rate of disease progression in conditions single glomerulus from a single patient with a defined
which are currently notoriously unpredictable. Expres- condition at a known time. This will be an invaluable
sion of tenascin may also be of prognostic value, as resource with which to address new questions about
it has been proposed as a marker of active matrix specific conditions as they arise in the future.
remodelling. It is conceivable that expression of some
mRNA species may allow one to distinguish between
different glomerular diseases. The potential exists for an ACKNOWLEDGEMENT
expansion of the way in which we investigate renal We are grateful for financial support for this work
biopsies, comparable to that which occurred with the from the U.K. Medical Research Council, Grant No.
introduction of immunofluorescence. 9429414.
The use of the ELISA for quantitation of the PCR
product has several advantages. It appears to be more
sensitive. It certainly permits the use of much smaller REFERENCES
samples, which is essential if repeated sampling of the
I . Pan LX. Diss TC, Peng HZ, Isaacson PG. Clonality ailalysis of defined
PCR is to be carried out; this may be necessary if B-cell populations in archival tissue sections uaing microdissection and the
multiple primer sets have been used in one reaction, or if polymerase chain reaction. Hisrqmfhology 1994; 24: 323--327.
quantitation rather than detection is required and the 2 . Pelayo JC, Mobilia MA, Tjio S , ~ ‘ rill. A method for isolation of rat
renal microvessels and mRNA localization. Am .I Phy\iol 1994; 267:
appropriate number of cycles is not known. Finally it F497-503.
is at least as easy and rapid as electrophoresis and 3. Raij L, Keane WF‘, Oswald H, Michael AF. Mesangial function in ureteral
obstruction in the rat; blockage of the eKcrcnt limb. J C h i Invesl 1979; 6 4
densitometry, and can handle more samples. 1204-1 2 12.
The use of oligo-dT-linked paramagnetic beads to 4. Furnesa I”. The formation and fxte of gloiner-ular immune complex
extract the mRNA from glomeruli has several advan- deposits. J Puthol 1991; 164 195-202.
5 . Peten EP, Striker LJ, Carome MA, er al. The contribution of increased
tages. The extraction is simple, clean, and very quick; collagen synthesis to human glomerulosclerosis: a quantitative analysis of
from proteinase K-treated glomerulus to purified alpha 21V collagen mRNA expression by competitive polymerase chain
mRNA takes as little as 20 min. The potential for reaction. J Exp Med 1992; 176 1571-1576.
6. Yang CW, Vlassara H, Peten EP, er a/. Advanced glycation end products
multiple amplifications is greatly improved by the small up-regulate gene expression found in diabetic glomerular disease. Proc N o r /
volumes required. Currently, our PCR amplification Acild Sci U S A 1994; 91: 9436-9440.
7. Peten EP, Striker I J . Progrehsion of glomerular &eases. J Intern n z d 1994;
uses only 2 pl of a 25 pl reverse transcription reaction, so 236 241-249.
12 genes can be amplified separately from each glomeru- 8. Yang CW, Hattori M, Vlassara H, er ul. Overexpression of transforming
lus before reuse of the beads becomes necessary, even if growth factor-beta I rnRNA is associated with up-regulation of gloinerular
tenascin and laminin gene expression in iioiiobese diabetic mice. J A m SIC
multiplex PCR analyses are not employed. Nephrt.111995; 5: 1610-1617.
Separation of mRNA from the different cell types 9. Tryggvason K , Zhou J, Hostikka SL, Shows TB. Molecular genetics of
would be a great advantage, but further development of Alport syndrome. K i d n q lrrf 1993; 43: 3844.
10. Wallace WA, Howie SE, Lamb D, Saller DM. Tenascin immunoreactivity
this methodology is need. We have found that mRNA in cryptogenic fibrosing alveolitis. J Pnfhol 1995, 175: 415420.
can be extracted from a single cell type, the podocyte, I I . Ponte P, Ng SY, Engel J, Gunning P, Kedes L. Evolutionary conservation
but with variable and unpredictable contamination by in the untranslated regions of actin mRNAs: DNA sequence of a human
beta-actin cDNA. Nucleic Acids Res 1984; 1 2 1687-1696.
mesangial cell mRNA. It is conceivable that if specific 12. Moriuchi T, Chang HC, Denome R, Silver J Thy-I cDNA sequence
epithelial and mesangial markers were to be measured in suggests a novel regulatory mechanism. Nufurr 1983; 301: 80-82.
13. Saito A, Pietromoaaco S, Loo AKC, Farquhar MG. Complete cloning and
every case, the amount of contamination in each sample sequencing of rat gp330lmegalin, a distinctive member of the LDL receptor
could be assessed, allowing a mathematical correction, gene family. P r m .Val/ Acud Sci USA 1994; 91: 9725-9729.
in the manner employed by a gamma counter with 14. Zhou J, Leinonen A, Tryggvason K. Structure of the human type IV
collagen COL4A5 gene, J B i d C‘hem 1994; 269 6608-6614.
simultaneous measurement of two isotopes with 15. Collier IE, Wilhelni SM, Eisen AZ, rt d.H-ru.s oncogene-transformed
overlapping emission spectra. huinan bronchial epithelial cells (TBE-I ) sccrete a single mctalloprotease
 RT-PCR AND SINGLE HUMAN GLOMERULI 193

capable of degrading basement membrane collagen. J Biol Chem 1988; 2 6 3 17. Docherty AJP, Lyons A, Smith BJ, rt a/. Sequence of huinan tissue inhibitor
6579-6581. of metalloproteinases and its identity to erythroid-potentiating activity.
16. Huhtala P, Tuuttila A, Chow LT, r f ul. Complete structure of the human Natur-r 1985; 318: 66-69.
gene for 92-kDa type IV collagenase: divergent regulation of expression for 18. Stetler-Stevenson WG, Brown PD, Onisto M, Levy AT, Liotta LA. Tiaauc
the 92- and 72-kilodalton enzyme genes in HT-1080 cell. J B i d Chem 1991; inhibitor of metalloproteinase-2 (TIMP-2) mRNA expression in tuniour cell
266 16845-16849. lines and human tumour tissues. J B i d Chem 1990; 265: 13933~13938.