Ž .Journal of Neuroimmunology 116 2001 233–237www.elsevier.comrlocaterjneuroin

Development of an ultrasensitive enzyme immunoassay for thež /determination of matrix metalloproteinase-9 MMP-9 levels in normal

human cerebrospinal fluid

Malgorzata Maliszewska a, Michael Mader a,), Ulrike Scholl a, Ivo Azeh a, Rudiger Hardeland b,¨ ¨ ¨Klaus Felgenhauer a, Wolfgang Beuche a, Frank Weber a,c

a Department of Neurology, Georg August UniÕersity, Robert-Koch-Str. 40, D-37075 Gottingen, Germany¨b Institute of Zoology and Anthropology, Georg August UniÕersity, Berliner-Str. 8, D-37073 Gottingen, Germany¨

c Section of Neurology, MPI of Psychiatry, Kraepelinstr. 2-10, D-80804 Munich, Germany

Accepted 22 March 2001

Abstract

Ž . Ž .Determination of matrix metalloproteinase-9 MMP-9 in human cerebrospinal fluid CSF to study blood–brain barrier impairmentand immune cell migration in inflammatory neurological diseases recently became a matter of major interest. Regularly, MMP-9 was

Ž .determined qualitatively or semi-quantitatively by zymography gelatin gel electrophoresis or quantitatively by enzyme immunoassayŽ . ŽEIA . As yet, it was not possible by either method to detect MMP-9 in CSF of controls patients without pathologically increased CSF

. Ž .parameters . We developed an ultrasensitive two-side enzyme-linked immunosorbent assay ELISA which allows for the first time tomeasure reliably MMP-9 concentrations in CSF of controls. This ELISA uses a monoclonal as capture and a polyclonal as detectorantibody. The detection limit of the assay is below 10 pgrml and the assay range is 15–2000 pgrml. Intra-assay precision is 2.5% forlow and 3.7% for high, inter-assay precision is 11% for low and 10.7% for high values, respectively. The determination of the MMP-9concentration in 50 control CSF gave the following results: range, 22–146 pgrml; median, 76 pgrml. The measurement of native andrecombinant MMP-9 was carried out with three commercially available ELISAs, most widely employed in MMP-9 research, andcompared to the newly developed one. All ELISAs recognize recombinant MMP-9 by factors of 5–20 less sensitively than nativeMMP-9. q 2001 Elsevier Science B.V. All rights reserved.

Keywords: Blood; Cerebrospinal fluid; Enzyme-linked immunosorbent assay; Matrix metalloproteinase-9; Sensitivity; Zymography

1. Introduction

Ž .The matrix metalloproteinases MMPs represent a fam-ily of 20 endopeptidases which contain a zinc-atom in theiractive center. They are synthesized by various cell types,particularly, by tumor cells and cells of the immune sys-tem. After secretion, they are activated extracellularly byserin proteinases or membrane-bound MMPs. The MMPsexhibit a broad and overlapping substrate specificity and intheir totality they are capable of degrading the extracellular

Žmatrix proteins completely Matrisian, 1992; Birkedal-.Hansen et al., 1993; Woessner, 1994 . Accordingly, they

) Corresponding author. Tel.: q49-551-39-9493; fax: q49-551-39-9493.

Ž .E-mail address: mmaeder@gwdg.de M. Mader .¨

are involved in various physiological and probably patho-Žlogical processes Masure and Opdenakker, 1989; Ries and

.Petrides, 1995; Yong et al., 1998; Kieseier et al., 1999a .Ž .The gelatinases MMP-2 and MMP-9 represent a sub-

group of MMPs which are structurally characterized by afibronectin II insert within the catalytic domain whichlacks all other MMPs. Furthermore, they are catalytically

Ž .specific for gelatin allowing zymographic analysis andŽcollagens type IV and V Birkedal-Hansen et al., 1993;

.Woessner, 1994 . The catalytic activity against the men-tioned non-fibrillar collagens which are major compoundsof the basement membrane caused researchers to studytheir involvement in migration of immune cells through

Žcapillary endothelium Huber and Weiss, 1989; Leppert et.al., 1995; Delclaux et al., 1996 .

In neurological sciences, their role in blood–brain bar-Žrier impairment was additionally addressed Rosenberg et

0165-5728r01r$ - see front matter q 2001 Elsevier Science B.V. All rights reserved.Ž .PII: S0165-5728 01 00304-6

( )M. Maliszewska et al.rJournal of Neuroimmunology 116 2001 233–237234

.al., 1995; Paul et al., 1998 . Many investigators haveunequivocally shown that a proform of MMP-2 is constitu-tively expressed in CSF of patients and control subjectsŽpatients without CNS inflammation or blood–brain barrier

. Ž .disturbance as well as in healthy animals rat and rabbit ,and does not change dramatically under pathological con-

Žditions Yushchenko et al., 2000; Paul et al., 1998; Azeh et.al., 1998 . In contrast, MMP-9 was never found in CSF of

controls or healthy animals, independently of the detectionŽ . Žmethod used EIA or zymography Gijbels et al., 1992;

.Kolb et al., 1998; Perides et al., 1998 . However, the onsetof an inflammatory CNS disease usually characterized by

Ž .pleocytosis, is paralleled by a rise of MMP-9 proformconcentration in human and animal CSF above the detec-tion threshold and sometimes followed by dramatical in-

Žcreases, particularly in severe bacterial infections Azeh et.al., 1998; Kieseier et al., 1999b; Valenzuela et al., 1999 .

Recently, it has been shown that in this process MMP-9concentration in CSF is related to the number of CSFinfiltrating leukocytes, whereas blood–brain barrier im-pairment does not seem to play an essential roleŽ .Yushchenko et al., 2000 . Here, we report on the develop-ment of a two-side ELISA, which allows for the first timeto detect MMP-9 in CSF of controls reliably and to exactlyquantify concentrations.

2. Materials and methods

2.1. CSF samples

Lumbar CSF samples were obtained from the Depart-ment of Neurology of the Georg August University ofGottingen. The lumbar puncture was part of the regular¨diagnostic procedure. CSF cell count was immediately

determined in the neurochemical laboratory. The controlŽ .group ns50 consisted of patients with headache or

backpain who underwent lumbar puncture for exclusion oftumor, subarachnoidal hemorrhage, or inflammatory dis-ease. These patients exhibited normal CSF parameters, i.e.

Žabsence of inflammatory signs pleocytosis, intrathecal.production of immunoglobulins and CSFrserum albumin

Ž . y3ratio Q values below 8=10 . Furthermore, CSFAlb

samples were collected from six patients with variousneurological diseases having cell counts above 5 cellsrml.The ages of the individuals were from 19 to 87 years,median was 52 years; 30 females and 26 males.

( )2.2. Enzyme-linked immunosorbent assay ELISA

For quantification of MMP-9 a monoclonal antibodyŽ .MMP-9 AB-1;Oncogene, Cambridge, MA, USA , whichrecognizes the pro- and the active-form of the enzyme, wasused as capture reagent in concentrations of 0.5 or 2.0

Žmgrml. Ninety-six-well plates Maxisorb; Nunc, Wies-.baden, Germany were coated with the monoclonal anti-

Ž .body dissolved in 100 ml phosphate-buffered saline PBSfor 16 h at 48C. The wells were blocked with PBS

Ž .containing 1% bovine serum albumin BSA for 1 h at258C. Subsequently, the wells were incubated overnight at

Ž .48C with antigen CSF or standard appropriately dilutedŽin PBSrBSArTween containing 1% BSA and 0.05%

. ŽTween 20 . After four washes with PBSrTween contain-.ing 0.2% Tween 20 , the plates were incubated with a

Ž .polyclonal MMP-9 antibody Biotrend, Koln, Germany¨diluted 1:5000 in PBSrBSArTween for 2 h at 258C. Afteranother four washes with PBSrTween, the wells wereincubated with an affinity purified peroxidase-conjugated

Žrabbit IgG antibody Jackson Immuno Research, West.Grove, PA, USA diluted 1:10,000 in PBSrBSArTween

Fig. 1. Standard curve obtained by twofold dilutions of native MMP-9.

( )M. Maliszewska et al.rJournal of Neuroimmunology 116 2001 233–237 235

Fig. 2. Above: MMP-2 and MMP-9 in seven CSF control samples afterzymography. Lane 1, serum standard showing MMP-9 and MMP-2; lanes2 and 7, empty lanes; lanes 3–6 and 8–10, various CSF control samples.Below: detection limit of recombinant MMP-9 by zymography. LanesŽ .ngrml : 1s37.5; 2s25.0; 3s20.0; 4s10.0; 5s5.0; 6s2.5; 7s1.25; 8s0.625; 9s0.313; 10s0.156.

for 1 h at 258C. After five washes, the plates were incu-Ž X wbated with 1 mgrml substraterH 0 2.2 -azino-bis - 3-2 2

xethylbenzthiazoline-6-sulfonate ; ABTS; Roche,.Mannheim, Germany and absorbance was read at 405r495

nm. A standard curve was obtained using native MMP-9Ž .Biotrend in twofold dilutions from 2000 to 3.9 pgrml.

2.3. Zymography

Zymography was carried out as described by Azeh et al.Ž .1998 . Gelatin concentration in the gel was 0.05% and thetime of gelatin digestion after SDS-PAGE was 24 h.

3. Results

In Fig. 1, the standard curve of the ELISA is presented:the optimal assay range is between 30 and 500 pgrml, but

Žreliable results can be obtained from 15 to 2000 pgrml cf..Table 3 . The detection limit of the assay is below 10

Žpgrml. The inter-assay precision is 11% for low 40. Ž .pgrml and 10.7% for high 850 pgrml and the intra-as-

Ž .say precision is 2.5% for low 40 pgrml and 3.6% forŽ .high 850 pgrml values, respectively.

Table 1 shows the results of a recovery experiment:200, 100 or 50 pgrml of native MMP-9 were dissolved ina CSF pool which comprised 10 CSF control samples. Themean recovery was 99%. Subsequently, the MMP-9 con-

Ž .centration was analyzed in 50 CSF samples controls ofpatients without pathologically elevated CSF parameters.The concentration range of normal CSF samples was22–146 pgrml and the calculated mean was 77 pgrml.

Ž .The cut-off mean plus three standard deviations forpathologically increased concentrations was 176 pgrml.All 50 samples were negative when analyzed by gelatingel electrophoresis. The detection limit of this assay was

Ž .found to be 300 pgrml Fig. 2 .For comparison, 10 of these samples were measured

with the three most widely used ELISA kits provided byvarious companies, but, as expected, the MMP-9 concen-

Žtration was below the detection limit of these assays Table.2 .

In a further experiment, all ELISAs were comparedŽ .using different concentrations of native MMP-9 Table 3 .

Our own assay reflected quite well the amounts of proteinadded and, at low concentrations of MMP-9, proved to bemore sensitive than the others. However, the commerciallyavailable ELISAs indicated much higher values as long asthe recombinant MMP-9, provided with the respective kit,was used for establishing the calibration curves. This,however, turned out to represent enormous over-estima-tions resulting from a substantial difference in detection

Table 1Recovery experiment

Ž .A total of 200, 100 or 50 pgrml of native MMP-9 were dissolved each in a CSF pool pool of 10 different CSF control samples and the recovery wasdetermined by the ELISA introduced in this publication.

Ž .pgrml Mean OD Minus Calculated Minus CSF = dilution RecoveryŽ . Ž .value background amount pool 33.9 factor %

Ž . Ž . Ž . Ž .3 replicates pgrml pgrml pgrml

200 0.9150 0.6040 227.2 193.3 193.3 97100 0.6860 0.3750 131.4 97.5 195.0 9850 0.5590 0.2480 84.8 50.9 203.7 102

Ž .CSF pool 0.4095 0.0985 33.9 – – 99 mean

Table 2MMP-9 determination in 10 CSF control samples by various ELISAs

Ž .The MMP-9 concentration values determined by the commercially available ELISAs Calbiochem, R&D Systems, Amersham were below the lowestvalue of the respective standard curve.

Ž .Ours pgrml 105.8 74.6 86.7 74.0 98.7 22.2 48.7 27.5 51.9 115.9Calbiochem R&D Systems 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Ž .Amersham pgrml

( )M. Maliszewska et al.rJournal of Neuroimmunology 116 2001 233–237236

Table 3Determination of native MMP-9 by three purchasable and our own ELISAThe numbers in bracket represent measured values=dilution factor. The values were calculated from the standard curve of each ELISA kit.

ELISA, Calbiochem R&D Systems Amersham Oursa a aŽ . Ž . Ž . Ž .Native 0.6–20 ngrml , 0.3–20 ngrml , 1–32 ng , 15–2000 pgrml ,

MMP-9 ngrml ngrml ngrml pgrmlpgrml

Ž . Ž . Ž . Ž .2000 13.65 13.7 16.40 16.4 34.54 34.5 2000.7 2001Ž . Ž . Ž . Ž .1000 6.85 13.7 9.25 18.5 17.74 35.5 1002.0 2004Ž . Ž . Ž . Ž .500 3.28 13.1 4.77 19.1 8.70 34.8 495.7 1983Ž . Ž . Ž . Ž .250 1.58 12.6 2.16 17.3 4.35 34.8 252.6 2021Ž . Ž . Ž . Ž .125 0.89 14.3 0.99 15.9 2.13 34.1 125.2 2003Ž . Ž . Ž . Ž .62.5 0.49 15.6 0.40 12.9 1.06 34.0 63.3 2027Ž . Ž . Ž . Ž .31.25 0.29 18.8 0.17 10.7 0.43 27.6 30.2 1933Ž . Ž . Ž .15.63 0.18 23.1 0.07 9.3 0.00 14.8 1894

Ž .7.81 0.00 0.00 0.00 7.1 1818

aAssay range as indicated in the kit protocol.

between native and recombinant MMP-9 proteins. Whenrecombinant MMP-9 was determined by our ELISA, about20-fold of the amount was required to reach similar OD

Ž .values as with native standard Table 4 .In CSF samples with elevated cell count, the accord-

ingly increased MMP-9 concentrations were determinedŽ .with the new ELISA Table 5 . The new assay also

recognizes MMP-9 in serum, where MMP-9 is constitu-tively present, and in conditioned cell culture media assensitively as in CSF. Plasma values were by 60% lower

Table 4Recovery of recombinant MMP-9 by our ELISARecombinant MMP-9 in twofold dilutions was measured by the newlydeveloped ELISA and MMP-9 concentration values were calculated fromthe OD values according to the standard curve established with nativeMMP-9 as shown in Fig. 1.

OD minus Recombinant Calculated Difference-background MMP-9 native fold

pgrml MMP-9pgrml

1.5038 24,000 1052.2 22.81.0793 12,000 562.3 21.40.6583 6000 275.9 21.70.3788 3000 141.6 21.10.2068 1500 73.1 20.50.1138 750 39.3 19.2

Table 5Determination of elevated MMP-9 levels in CSF samples with increasedcell count

Cell count 6 8 13 224 316 896MMP-9 0.564 0.673 0.249 4.702 60.488 32.500concentrationin CSFŽ .ngrml

Žcompared with serum of the same individual data not.shown .

4. Discussion

The highly sensitive two-side ELISA has been designedfor reliable determination of MMP-9 in the 15–2000 pgrmlrange. The components of this ELISA were all commer-

Ž .cially available cf. Section 2 . The ELISA uses a mono-clonal antibody as capture reagent, which recognizes thepro- and active-form of the enzyme. This antibody wasoriginally raised against MMP-9 purified from conditioned

Ž .media of phorbol 12-myristate 13-acetate PMA -stimu-lated HAT-1080 human fibrosarcoma cells. For detection apolyclonal antibody was used which was raised against

ŽMMP-9 purified from human leukocytes mainly granulo-.cytes . As standard, native MMP-9 was employed, which

was purified from human granulocytes. Thus, both anti-Ž .bodies of the assay were raised against generic native

human MMP-9, which was also used for generating thestandard curve. Compared with a previously described

Ž .ELISA Yushchenko et al., 2000 , which used recombinantMMP-9 as standard, incubation with CSF was carried outovernight at 48C instead of 2 h at 228C. To reach thehighest sensitivity, it was important to take care that thesecondary antibody was not out of date. These facts ex-plain well the high sensitivity and reliability of the assay,

Ž .particularly at very low 15–500 pgrml concentrations.Furthermore, compared with recombinant MMP-9 regu-larly used in commercially available ELISA kits, about20-fold of the amount of this protein is necessary to obtainsimilar OD-values as with native MMP-9. This means thatthe affinity of the antibodies is much higher for native thanfor recombinant MMP-9 and consequently, much morerecombinant than native MMP-9 is required to reach simi-

Ž .lar binding efficiency Tables 3 and 4 . Thus, the use of

( )M. Maliszewska et al.rJournal of Neuroimmunology 116 2001 233–237 237

native MMP-9 shifts the standard curve from the ngrmlrange to the pgrml range. This is also true for the com-

Ž .mercially available kits cf. Table 3 . A standard curvegenerated by use of recombinant MMP-9 does not seem tobe appropriate for measuring absolute MMP-9 concentra-tions in body fluids.

Ž .Human blood and CSF contain generic native MMP-9which is mainly derived from granulocytes and possibly

Ž .monocytes Yushchenko et al., 2000 . Therefore, it isrecognized very sensitively by the ELISA and the determi-nation of MMP-9 concentrations in CSF of controls be-came possible for the first time. Additionally, the newlydeveloped ELISA is more sensitive than zymography,which as yet has been believed to be the most sensitive

Žtechnique for detection of MMP-9 Leppert et al., 1998;.Yushchenko et al., 2000 . However, it had never been

possible to detect MMP-9 in normal CSF by zymographyas reported unequivocally by all researchers studying this

Žbody fluid Kieseier et al., 1999b; Kolb et al., 1998;.Gjibels et al., 1992; Yushchenko et al., 2000 . The detec-

tion limit of zymography, as evaluated by ourselves, was300 pgrml and thus, zymography was about 15–30-fold

Ž .less sensitive than the above ELISA Fig. 2 . Zymographyby itself is a semiquantitative technique like other gel andscanner based approaches since inter-assay variation isconsiderable. Furthermore, renaturation of the enzyme af-ter non-reducing SDS-PAGE is probably not complete andmay vary additionally between experiments.

The commercially available ELISA kits, most widelyused for quantification of MMP-9, employ recombinantMMP-9 as standard. However, this standard is by factorsof 10–20 less sensitively recognized than native MMP-9

Ž .by these ELISAs as well as by ours Tables 3 and 4 .Thus, it can be assumed that all MMP-9 concentrationsdetermined previously have to be corrected by these fac-tors as far as the absolute amounts of MMP-9 in bodyfluids are concerned.

The MMP-9 ELISA introduced by this work consists ofŽ .components antibodies, standard which are all commer-

cially available at reasonable costs. The monoclonal cap-ture antibody, the most expensive part, can be coated at

Ž .very low concentrations 0.5 mgrml , lowering costs enor-mously. Coating of higher concentrations, e.g. 2 mgrml,did not improve the ELISA substantially and the standard

Ž .curve was nearly identical data not shown . Thus, theMMP-9 ELISA described herein will be a useful andaffordable tool for researchers to quantitate MMP-9 invarious body fluids and cell culture media since the ELISArecognizes MMP-9 in serum and conditioned medium assensitively as in CSF.

Acknowledgements

This work was supported in part by DeutscheForschungsgemeinschaft WE 1439r4-1.

References

Azeh, I., Mader, M., Smirnov, A., Beuche, W., Nau, R., Weber, F., 1998.¨Experimental pneumococcal meningitis in rabbits: the increase ofmatrix metalloproteinase-9 in cerebrospinal fluid correlates withleukocyte invasion. Neurosci. Lett. 256, 127–130.

Birkedal-Hansen, H., Moore, W.G.I., Bodden, M.K., Winsor, L.J.,Birkedal-Hansen, B., DeCarlo, A. et al., 1993. Matrix metallopro-teinases: a review. Crit. Rev. Oral Biol. Med. 4, 197–250.

Delclaux, C., Delacourt, C., d’Ortho, M.-P., Boyer, V., Lafuma, C., Harf,A., 1996. Role of gelatinase B and elastase in human polymorphonu-clear neutrophil migration across basement membrane. Am. J. Respir.Cell Mol. Biol. 14, 288–295.

Gijbels, K., Masure, S., Carton, H., Opdenakker, G., 1992. Gelatinase inthe cerebrospinal fluid of patients with multiple sclerosis and otherinflammatory neurological disorders. J. Neuroimmunol. 41, 29–34.

Huber, A.R., Weiss, S.J., 1989. Disruption of the subendothelial base-ment membrane during neutrophil diapedesis in an in vitro constructof a blood vessel wall. J. Clin. Invest. 83, 1122–1136.

Kieseier, B.C., Seifert, Th., Giovannoni, G., Hartung, H.-P., 1999a.Matrix metalloproteinases in inflammatory demyelination. Neurology53, 20–25.

Kieseier, B.C., Paul, R., Koedel, U., Seifert, T., Clements, J.M., Gearing,A.J.H. et al., 1999b. Differential expression of matrix metallopro-teinases in bacterial meningitis. Brain 122, 1579–1587.

Kolb, S.A., Lahrtz, F., Paul, R., Leppert, D., Nadal, D., Pfister, H.-W. etal., 1998. Matrix metalloproteinases and tissue inhibitors of metallo-proteinases in viral meningitis: upregulation of MMP-9 and TIMP-1in cerebrospinal fluid. J. Neuroimmunol. 84, 143–150.

Leppert, D., Waubant, E., Galardy, R., Bunnet, N.W., Hauser, S.L., 1995.T cell gelatinases mediate basement membrane transmigration invitro. J. Immunol. 154, 4379–4389.

Leppert, D., Ford, J., Stabler, G., Grygar, C., Lienert, C., Huber, S. et al.,Ž .1998. Matrix metalloproteinase-9 gelatinase B is selectively elevated

in CSF during relapses and stable phases of multiple sclerosis. Brain121, 2327–2334.

Masure, S., Opdenakker, G., 1989. Cytokine-mediated proteolysis intissue remodelling. Experienta 45, 542–549.

Matrisian, L.M., 1992. The matrix degrading metalloproteinases. BioEs-says 14, 455–463.

Paul, R., Lorenzl, S., Koedel, U., Sporer, B., Vogel, U., Frosch, M. et al.,1998. Matrix metalloproteinases contribute to the blood–brain barrierdisruption during bacterial meningitis. Ann. Neurol. 44, 592–600.

Perides, G., Charness, M.E., Tanner, L.M., Peter, O., Satz, N., Steere,A.C. et al., 1998. Matrix metalloproteinases in the cerebrospinal fluidof patients with lyme neuroborreliosis. J. Infect. Dis. 177, 401–408.

Ries, C., Petrides, P.E., 1995. Cytokine regulation of matrix metallopro-teinase activity and its regulatory dysfunction in disease. Biol. Chem.Hoppe-Seyler 376, 345–355.

Rosenberg, G.A., Kornfeld, M., Estrada, E., Kelly, R.O., Liotta, L.A.,Stetler-Stevenson, W.G., 1995. TIMP-2 reduces proteolytic openingof blood–brain barrier by type IV collagenase. Brain Res. 576,203–207.

Valenzuela, M.A., Cartier, L., Collados, L., Kettlun, A.M., Araya, F.,Concha, C. et al., 1999. Gelatinase activity of matrix metallopro-teinases in the cerebrospinal fluid of various patient populations. Res.Commun. Mol. Pathol. Pharmacol. 104, 42–52.

Woessner, J.F., 1994. The family of matrix metalloproteinases. Ann.Natl. Acad. Sci. U. S. A. 732, 11–21.

Yong, V.W., Craig, A., Krekoski, P., Forsyth, P.A., Bell, R., Edwards,D.R., 1998. Matrix metalloproteinases and diseases of the CNS.Trends Neurosci. 21, 75–80.

Yushchenko, M., Weber, F., Mader, M., Scholl, U., Maliszewska, M.,¨ ¨Ž .Tumani, H. et al., 2000. Matrix metalloproteinase-9 MMP-9 in

Ž .human cerebrospinal fluid CSF : elevated levels are primarily relatedto CSF cell count. J. Neuroimmunol. 110, 244–251.