Journal of Neuroscience Research 32:329-339 (1992)

Two-Site Enzyme Immunoassay for PNGF Applied to Human Patient Sera L. Lorigados, S. Soderstrom, and T. Ebendal Department of Developmental Biology, Uppsala University, Biomedical Center, Uppsala, Sweden

Nerve growth factor (NGF) supports sympathetic and sensory neurons in the peripheral nervous system and serves functions in the development and maintenance of cholinergic neurons in the basal forebrain. NGF distribution can be studied with the use of a sensitive two-site enzyme immunoassay (EIA). The monoclonal antibody 27/21to mouse NGF was recently shown to effectively block the activity of both recombinant human NGF and native mouse NGF, and a two-site EIA using monoclonal antibody 27/21was optimized. We have now applied this assay to examine NGF levels in normal human serum and serum from Parkinson, Alzheimer, and Huntington patients. To further test the specificity of conjugate binding, dilutions of the human sera were preincubated with an excess of monoclonal NGF antibody 27/21in solution. With this strategy it was possible to completely block the signal obtained using the two-site EIA. Furthermore, we show that recombinant BDNF and NT-3 do not cross-react with monoclonal antibody 27/21under our conditions. We found low levels of specific NGF immunoreactivity in normal human sera (0.4 f 0.1 ng/ml). Significantly lower levels of NGF were found in sera from patients with Parkinson’s and Huntington’s disease whereas sera from Alzheimer patients showed only slight reductions in the NGF level. Two patients who had received intracerebral NGF infusions (one with Parkinson’s and other with Alzheimer’s disease) showed significantly elevated serum levels of NGF during the period of infusion. Due to an inhibitory activity in human serum, it was impossible to demonstrate the low levels of NGF activity in the human serum samples using explanted embryonic sympathetic ganglia, even after concentration by pressure dialysis. Thus, the serum levels are below the limit to evoke a response in NGF-sensitive neurons and thus to expect any physiological effect. Nevertheless, the levels measured may be used as indicators in clinical conditions such as Parkinson’s and Huntington’s disease. o 1992 Wiley-Liss, Inc. 0 1992 Wiley-Liss, Inc.

Key words: monoclonal anti-NGF antibody 27/21,pgalactosidase conjugate, bioassay, sympathetic ganglion, two-site enzyme immunoassay, human serum, Alzheimer’s disease, Parkinson’s disease, Huntington’s disease, intracerebral NGF infusions INTRODUCTION PNGF is a basic 118 amino acid protein acting as a trophic factor for many sensory and sympathetic neurons in the peripheral nervous system (Levi-Montalcini, 1987; Levi-Montalcini and Angeletti, 1968; Thoenen and Barde, 1980). The levels of NGF and its mRNA correlate with the density of sympathetic innervation (Korsching and Thoenen, 1983), which is consistent with the notion that NGF has a trophic role for sympathetic neurons. Nerve growth factor has also been found in the brain (Korsching et al., 1985; Whittemore et al., 1986; Goedert et al., 1986) where it acts in a trophic capacity for the development and maintenance of cholinergic neurons of the basal forebrain (Korsching et al., 1985; Liirkfors et al., 1987, Ernfors et al., 1990; review by Ebendal, 1989a). The highest levels of NGF and its mRNA in the brain have been found in the hippocampus and cerebral cortex, to which the major cholinergic pathways in the brain project (for reviews, see Thoenen et al., 1987; Ebendal, 1989a). These neurons can be prevented from dying after axonal transection in vivo by the addition of exogenous NGF (Hefti and Weiner, 1986; Williams et al., 1986). A valuable method in NGF research is the sensitive two-site enzyme immunoassay (EIA; Fig. 1) introduced to allow reliable determinations of the low

Received March 3, 1992; accepted March 20, 1992. Address reprint requests to T. Ebendal, Department of Developmental Biology, Uppsala University, Biomedical Center, Box 587, S-751 23 Uppsala, Sweden.

L. Lorigados is now at the Iberolatinoamerican Center for Transplant and Regeneration of the Nervous System, Ave. 25 # 15805 entre 158 y 160, Cubanacan, Playa. C., Havana, Cuba.

330

Lorigados et al.

A

B

4 - methylumbelliferylO - D-galactoside (MUG)

EM1 4 5 0 n m

4-methylurnbelliferone

O-galactosidase conjugated anti-ONGF Mab 27/21

A

anti - O N G F Mab 27/21

Fig. 1. Schematic illustration of the principle of the two-site EIA (A) and the preincubation blocking step used as a specificitycontrol (B). It should be noted for panel B that a likely event is also the complexing of NGF between free anti-NGF antibodies without prior binding to the immobilized coat of Mab 27/21.

levels of NGF present in peripheral and CNS tissues (Korsching and Thoenen, 1983, 1987; Ltirkfors and Ebendal, 1987; Weskamp and Otten, 1987). It is now possible to detect NGF in tissue samples using a sensitive two-site enzyme immunoassay with monoclonal antibody 27/2 1, which has recently been shown to be able to effectively block the activity of both recombinant human NGF and native mouse NGF (Soderstrom et al., 1990). There are conflicting reports in the literature concerning the presence of nerve growth factor in serum (Johnson et al., 1971; Fabricant and Todaro, 1981; Skaper and Varon, 1982; Heinrich and Meier, 1988; Stephani et al., 1987; Stephani and Maurer, 1990) especially in relation to various human clinical conditions such as neurodegenerative diseases (Parkinson’s disease, Alzheimer’s disease, Huntington’s disease). Although NGF is the best characterized neurotrophic factor so far, little is known about its role in the pathogenesis of any disease. Recently it has been reported that long-lasting positive effects were observed in a patient with severe Parkinson’s disease who received adrenal medullary autografts initially supported by NGF infusion (Olson et al., 1991). Clearly, the establishment of a sensitive assay system for the demonstration of NGF in serum would have a significant impact on our ability to evaluate the possible involvement or therapeutic use of NGF in neurological diseases. The present paper describes the im-

munological detection of low levels of NGF in human serum samples and introduces a blocking test validating the specificity of the immunoreactivity for NGF in human serum.

MATERIALS AND METHODS Blood Samples Venous-blood samples from normal and diseased persons were taken in non-heparinized tubes, centrifuged after clotting and stored frozen. Serum samples were collected from 12 patients: 4 with Parkinson’s disease (PD), mean age 57 years; 4 with Alzheimer’s disease (AD), mean age 65 years; 4 with Huntington’s disease (HD), mean age 48 years. Normal human sera (NHS) were taken from five age-matched controls. Clinical diagnosis of PD was rated according to the presence of tremor, bradykinesia, rigidity and loss of postural reflexes. The AD was diagnosed according to the NINCDS/ADRDA-criteria, and for HD, clinical diagnosis according to familiar genetic analysis, choreics or choreoatetosic movements, and mental alterations type dementia tests were employed. In this study were also included sera from two patients, one with PD and another with AD that received intracerebral infusion of NGF as described (Olson et al., 1991, 1992). These serum samples were taken before and during treatment to

NGF Enzyme Immunoassay Applied to Human Serum

assess the possible appearance of NGF in the serum of the patients.

Other Sera Tested Pooled normal rat serum was obtained from Sprague Dawley rats (Alab, Stockholm). Fetal calf serum (FCS) was from a commercial source (Biochrom, Berlin). Bioassay Techniques Sympathetic ganglia from 9-day-old chick embryos (white leghorn) were used to estimate the nerve growthpromoting effect of the human serum. The method was essentially according to previous descriptions (Ebendal, 1989b; Soderstrom et al., 1990) in which sympathetic ganglia were embedded in a collagen gel matrix. One sample of normal serum was concentrated by pressure dialysis (Millipore PTGC filter, cut-off 10,000 MW) remove low molecular weight substances, and the samples (dialysed and non-dialysed sera) plus several dilutions of NGF (2, 1, 0.5, 0.25 ng/ml) were added over the collagen gels at an equal volume and the culture incubated at 37°C and 5% CO, for 2 days. Fibre outgrowth response was scored blindly in 0.1 increments, where 0 biological units (BU) represent total absence of neurites and 1 BU represents a dense, circular fibre halo. Each determination was repeated at least three times and the mean k SEM calculated. Purification of Mouse Salivary Gland PNGF PNGF was purified from male mouse submandibular gland as described previously (Mobley et al., 1976; Ebendal et al., 1984). Elution from the second column of CM-Sepharose was by a linear gradient of sodium chloride (Chapman et al., 1979; Ebendal et al., 1984) yielding a peak of highly purified PNGF. The concentration of the NGF was determined from its specific absorbance at 280 nm (1.6 at 1 mg/ml in a 1 cm cuvette). To test for cross-reactivity in the EIA with BDNF and NT-3 we used the recombinant rat proteins found in conditioned medium after transfection of COS cells, as described by Hallbook et al. (1988), Ernfors et al. (1990) and Ibaiiez et al. (1991). Enzyme Immunoassay for NGF A two-site enzyme-immunoassay (EIA) for NGF described previously (Soderstrom et al., 1990) was used for the present study. Immunoplates (black 96-well multidishes, Dynatech Microfluor) were coated with monoclonal anti-mouse-PNGF antibody 2712 1 (Korsching and Thoenen, 1983; obtained from Boehringer Mannheim Biochemica). The stock solution of purified antibody 27/ 21 (60 pg/ml) was diluted in 0.05 M carbonate buffer (pH 9.6) to 0.5 p,g/ml. Control wells were coated with

331

the same amount of normal immunoglobulin G (mouse IgG; Calbiochem). Each well received 50 pl and the dishes were incubated at 4°C overnight. The immunoplates were then blocked for 1 hr at room temperature using 1% bovine serum albumin (BSA) in the same buffer. After washing with Tris-buffered saline (TBS; comprising of 0.02 M Tris-HC1, pH 7.4, 0.5 M NaCl) containing 0.5% Tween 20, the samples were applied in a volume of 50 pl, the plates sealed, and incubated overnight at 4°C. To serve as standards, samples of purified mouse NGF at 0.1, 0.001-100 ng/ml in the same buffer were also included. The plates were then washed extensively in TBS with Tween 20 at room temperature before antibody 27/21-P-galactosidase-conjugate(Boehringer; 4 units of enzyme activity per ml) was added at a dilution of 1:200 in TBS, 0.1% BSA, 0.05% Tween 20, 2 mM MgCl,, for an overnight incubation at 4°C. An additional, novel specificity control consisted of adding the conjugate with an excess of unconjugated 27/2 1 monoclonal antibody (500nglml final concentration) in parallel wells. For this blocking test, 20 pl of 27/21 Mab (1.2 pg) from the stock flask (60 pg/ml) was diluted with 200 p1 of TBS sample buffer. To each of these control wells (50 pl) 5 pl of this dilution was added to give a final concentration of 0.5 pg/ml. After incubation, the immunoplates were washed for several hours in TBS, 0.5% Tween 20 with 2 mM MgCl,. Finally the dishes were washed with 0.1 M phosphate buffer (pH 7.2), 2 mM MgSO,, 0.4 mM MnSO,, 4 mM EDTA, and 0.1% azide. To start the enzyme reaction, 100 p1 of this buffer was added to each microwell with 100 p1 of 0.2 mM methylumbellifery1-Pgalactoside (MUG). The accumulation of methylumbelliferone (excitation wavelength 365 nm, emission wavelength 450 nm) was followed in a microplate fluorometer (Dynatech Microfluor). Matched background values (coat of normal mouse IgG) were subtracted from each reading and the result expressed as enzyme activity (fmol MUG per min) by comparison against an external standard of methylumbelliferyl. The readings from wells which contained sera pre-incubated with antibody 2712 1, were compared to the background values of wells coated with normal mouse IgG and the difference was subtracted from the assay values. The NGF concentrations in sera have therefore been corrected to account for background and non-specific binding. The EIA values for serum samples were determined from the regression line for the NGF standard incubated under similar conditions.

RESULTS NGF Immunoassay The capacity of the two-site enzyme immunoassay (Fig. 1A) to detect the standard of mouse PNGF is

332

Lorigados et al. 8oo

Detection of NGF in Human Serum The normal human serum was found to give a fairly strong, concentration-dependent signal using the two-site enzyme immunoassay (Fig. 3A, filled symbols). The problem was to distinguish non-specific background binding of the anti-NGF-conjugate from real signal from human NGF carried in the serum. We solved this by introducing a pre-incubation (Fig. 1B) in which dilutions of human serum were incubated with an excess of the monoclonal NGF antibody 27/21 in solution (0.5 p,g/ ml). Using this strategy it was proven possible to partially quench the signal obtained from human sera (Fig. 3A, open symbols), lending support to the notion that immunoreactive PNGF is present at low levels in the human serum samples.

1

L__L_IL I -

0.0

0.2

0.4

0.6

0.8

,

1

1.0

1.2

Comparison Between Human, Fetal Calf, and Rat Sera The human serum samples were serially diluted Fig. 2 . Dose-response curve for the standard of purified mouse and for comparison, fetal calf serum (FCS) was tested in NGF. A linear relationship is evident when the enzyme activity a similar manner. In contrast to the results from human is measured under the conditions used. The detection limit for sera, no NGF immunoreactivity was detectable in FCS NGF was routinely around 2 pg/ml. Open symbols show the curve obtained after the preincubation with an excess of rnono- using the two-site EIA (Fig. 3B). Adult rat serum clonal NGF antibody 27/21 (500 ng/ml). The result represents showed very low (40 2 15 pg/ml) but detectable immumean f SEM of three independent determinations. The SEM noreactivity to NGF in these samples (data not shown). is not shown when less than the width of the symbol. Recovery of NGF Added to Normal Human Serum A known amount of NGF (1 ng/ml) was added to different concentrations of normal human serum in order shown in Figure 2 (filled symbols). The method is based to determine recovery in the assay (Fig. 4). At the highon sandwiching the NGF dimer between an immobilized est concentration of human serum used (50%) an 80% coat of antibody 27/2 1 and a P-galactosidase-conjugated recovery of the added NGF was found. This study demantibody and is sensitive to a NGF concentration of ap- onstrated a fairly constant recovery of NGF added to proximately 1-2 pg/ml. The ability of the assay to detect each of the different dilutions of NHS sample tested. It human NGF at the same sensitivity as mouse NGF under was also shown that blocking the immunoreactivity by the present conditions has been described (Soderstrom et adding an excess of Mab 27/21 was effective also in the al., 1990). It was previously found that 5 ng/ml of NGF samples which had NGF added to the serum samples is completely blocked by 100 ngiml of antibody 27/21 in (Fig. 4). Thus we conclude that although there may be a bioassay (Soderstrom et al., 1990). The novel blocking binding proteins or carrier molecules for NGF in human step during which the sample is incubated with an excess serum, they do not interfere to any considerable extent in (500 ng/ml) of unlabelled monoclonal anti-NGF anti- the EIA under the optimized conditions used here. body 27/21 (Fig. 1B) was effective in reducing the signal in the EIA (Fig. 2, open symbols). Using this strategy, Test for Neurotrophic NGF Activity in implying that NGF complexed between excess of free Human Serum It was considered important to determine whether 27/21 antibodies was removed or that NGF bound to the the NGF immunoreactivity found in serum represents coat of monoclonal antibody 27/21 was masked, made it possible to completely block the signal obtained in the biologically active NGF. Ebendal et al. (1989) demonEIA under standard conditions. We also tested the pos- strated that the biological assays using explanted embrysible cross-reactivity of monoclonal antibody 27/2 1 with onic ganglia (under ideal conditions having a detection the two NGF-related neurotrophins BDNF and NT-3 limit for NGF of 0.1-0.2 ng/ml) are negatively influ(Ernfors et al., 1990; Ibhiiez et al., 1991): Under the enced by human serum present at levels above 10-20% present conditions the 27/2 1 EIA did not recognize these in the tissue culture medium. This toxic effect prevents assays by simple inclusion of human serum at levels proteins (data not shown). Conc. of N G F ( n g / m l )

333

NGF Enzyme Immunoassay Applied to Human Serum

v Normal h u m a n sera ( N H S ) (n=5) v NHS + 27/21 Mab

h

c

.I

E

1

h

.-C E

>

300

V

Fetal calf serum ( F C S ) + 27/21 Mab

V FCS

15

0

E

E

3v

L v

h ‘j: .*

B

20

A

400

i

200

i

0

m

E3 c

3

100

c

5

W

w

0

C 1

1

I

I

I

1

I

I

I

I

I

I

I

I

0

10

20

30

40

50

60

0

10

20

30

40

50

60

S e r u m c o n c . (%)

Serum conc. (%)

Fig. 3. The enzyme immunoassay (EIA) result for normal human serum (NHS) and fetal calf serum (FCS). A: Measurement of the immunoassay enzyme activity of NHS using normal conditions and with the preincubation step reduces the enzyme activity indicating the presence of NGF in serum. B: As in A but with FCS. No NGF could be detected in this serum. In both graphs, each data point represents the mean of 3 independent measurements. Note the difference in scales for enzyme activity in A and B .

where the amounts of NGF expected from the EIA re- Alzheimer’s disease, Parkinson’s disease, and Huntingsults should be detectable (Fig. 5). To circumvent this ton’s disease were tested (Fig. 6 ) . There were, in all problem we concentrated normal human serum (NHS) by cases, a trend for lower values of NGF detected in the pressure dialysis for subsequent bioassay aiming to re- patient sera. In the Alzheimer sera, the reductions were move inhibitory, possibly low-molecular weight sub- not statistically significant (Fig. 6A). However, the Parstances. The dose-response relationship between NHS kinson sera showed strikingly lower levels of NGF im(dialyzed and undialyzed with several dilutions of NGF) munoreactivity in the EIA (Fig. 6B) and there were also and the outgrowth in sympathetic ganglia was studied statistically significant reductions in NGF levels in sera and compared with the response obtained with mouse from Huntington patients when compared to normal ageNGF diluted in BME (Fig. 5 ) . NHS reduced NGF activ- matched sera (Fig. 6C). ity, the non-dialyzed serum reducing outgrowth more To test the Parkinson sera for possible increases in than the dialyzed sample. Since the lowest activity of endogenous NGF binding activity possibly leading to NGF detected in 25% human serum occurred at 0.5 ng/ apparent reductions in NGF levels, we added a known ml, the amount of endogenous NGF in concentrated se- amount of NGF also to these sera and determined recovrum needs to be at least 2 ng/ml to be detected by our eries. It was found that the recovery was the same as for bioassay. The EIA indicated lower values, in the range normal human sera (over 80%, data not shown). We thus of 0.3-0.5 ng/ml, and it is thus not surprising that they conclude that it is more likely that the lower NGF values are undetected in the bioassay. Skaper and Varon (1982) in the EIA of Parkinson sera are the result of reduced have previously used sensory neurons in bioassays in levels of NGF in the serum rather than from increases in attempts to reveal NGF activity in human serum (tested substances possibly masking the immunoreactive NGF. at 10% concentration in the culture medium) with negaIn the present study, we also included further analtive results. ysis of the unique sera from the first two patients that have received intracranial infusion of NGF (Olson et al., Application of the NGF EIA to Sera From Patients 1991, 1992). One case is a Parkinson patient that reWith Neurological Diseases ceived mouse PNGF into the putamen at 120 pg per day The next issue was to determine the quantities of for 23 days following autografts of adrenal medullary human NGF present in sera from patients afflicted by tissue to the putamen (Olson et al., 1991). The levels of various neurological diseases. Sera from patients with NGF in serum samples collected before and at the end of

334

Lorigados et al. 1000 V NHS

.?

E

V NHS

800

+ +

-

NGF ( I n g / m l ) N G F + 27/21 Mab

2

?.

0 NHS

kl

600

.+

n

A

0.6

0 0

z

2

400

0.4

5

;

!

(25%) + NGF

1

4

w

0 Non-dialysed NHS

0.8

t

.r

X

BME

5

.d

.4J

0 N G F in

0 Dialysed NHS (25%)+ NGF

3

> s

1.o

a

k

200

0.2

vl

0.0

-

0 1

,

,

L

0

10

20

30

I

,

J

40

50

60

1

,

0.0

0.5

1.0

1.5

I

1

2.0

2.5

Conc. of N G F (ng/ml)

Serum c o n c . ( % )

Fig. 4. Demonstration of the ability of the assay to measure a known quantity of exogenously added NGF (1 ng/ml) to NHS. A sample of NHS was divided, and to one half, NGF was added. The NGF-added sample was again divided and one half was preincubated with 27/21 (500 ngfml). The samples were diluted as indicated and EIA measurements were performed.

Fig. 5. Neurite outgrowth from 9-day-old chick embryo sympathetic ganglia elicited by different dilutions of control human sera (NHS) with NGF added. The dose-response relationship between dialysed and non-dialysed NHS and the outgrowth of sympathetic ganglia was studied and compared with the activity of NGF diluted in culture medium (BME). The NHS reduced NGF activity in both cases, but non-dialysed serum reduced activity to a greater degree than dialyzed sample. Growth was scored after 48 hr. Each point represents the mean of 3 determinations, SEM indicated by the bars.

the infusion period were determined. Over the entire DISCUSSION range of concentrations tested it was possible to see an The ability of the monoclonal antibody 27/21 to increase in the NGF level from the low levels noted in the Parkinson patients (Fig. 7A; cf. Fig. 6B). Most of this detect human NGF in an EIA (Soderstrom et al., 1990) increase could be blocked using an excess of Mab 27/21 of serum samples was studied here. It has been indicated in parallel sets of wells (Fig. 7A). Comparison with the also in other studies that the monoclonal antibody 27/21 external NGF standard in the EIA indicated a substantial to mouse PNGF (Korsching and Thoenen, 1983) recogand highly significant outflow of NGF into the blood of nizes human NGF in a two-site EIA (Heinrich and this patient as a result of the intraputaminal NGF infusion Meyer, 1988; Bruce and Heinrich, 1989). The similarity (Fig. 7B). Moreover, this finding validates the method of in amino acid sequence among NGF from different spedetermination of NGF in human serum based on this cies is high (66-98%) with the variable regions located controlled situation with known addition of excess in a few hydrophilic domains (Meier et al., 1986; Ebenamounts of NGF to the human body. The second case dal et al., 1989), likely to correspond to antigenic concerns an Alzheimer patient that received 75 pg of epitopes since they form three P hairpin loops and a NGF per day intraventricularly for 90 days in a first reverse turn exposed on the surface of the NGF protein clinical test (Olson et al., 1992). An increase in NGF (McDonald et al., 1991). The two-site EIA using rnonolevel in the serum, attributable to the infused mouse NGF clonal antibody 27/21 earlier optimized by us (Sodwas also observed in this case, although not statistically erstrom et al., 1990) was used in the present study. Unsignificant (Fig. 7C,D). For comparison, it can be men- der the conditions used, this EIA detected the human tioned that whereas the Alzheimer patient had NGF lev- recombinant NGF with the same sensitivity (1-2 pg/ml) els in serum raised to around 0.15 ng/ml (Fig. 7D), the as shown for the mouse NGF. It was thus considered levels of NGF in the cerebrospinal fluid serving as re- possible to apply this method in order to determine levels cipient for the NGF infusion were previously determined of NGF in samples of serum from patients or agein our laboratory to be around 150-200 ng/ml by the matched controls. We now demonstrate a NGF-like immunoreactivity two-site EIA and by bioassay (Olson et al., 1992).

335

NGF Enzyme Immunoassay Applied to Human Serum

-.‘ E

400

B

A

500

I 1

V

V

Parkinson’s disease V P D + 2 7 / 2 1 Mab

Alzheimer’s disease ( A D ) ( n = 4 ) AD + 27/21 M a b

:2

100

{PD) ( n = 4 )

I

C

w

I I I ‘ 1

o

in

20

30

40

60

50

Serum conc. (2)

800

,

0

10

20

30

40

50

60

Serum conc. (%)

C Normal h u m a n s P r a ( n = 5 ) Alzheirnrr s e r a ( n = 4 ) Huntington sera ( n = 4 ) [IIID Parkinson s e r a (n=4)

a

Serum samples

Fig. 6. Comparison of NGF levels among sera from patients suffering from different neurological diseases and age-matched healthy controls. A: The result from tests of four Alzheimer sera. A slight decrease in measured NGF levels is indicated compared to normal human serum (cf. Fig. 3A). In particular it is worth noting that the fraction of EIA signal possible to block with monoclonal antibody 27/21 is reduced. B: The re-

sult for Parkinson sera. A conspicuous reduction in apparent NGF immunoreactivity is present. C: Summary of the measured NGF levels in sera including also cases of Huntington’s disease. Both the Huntington and Parkinson sera show reduced NGF levels compared with normal human sera. Statistical analysis was by Student’s t test (**P