Advantages of branched peptides in serodiagnosis

Journal of Immunological Methods, 147 (1992) 65-72 @ 1992 Elsevier Science Publishers B.V. All rights reserved 0022-1759/92/$05.00

65

JIM 06189

Advantages of branched peptides in serodiagnosis Detection of HIV-specific antibodies and the use of glycine spacers to increase sensitivity H.S. Marsden, A.M. Owsianka, S. Graham, G.W. McLean, CA. Robertson and J.H. Subak-Sharpe MRC VU'Ology Unit, Institute of VU'Ology, Church Street, Glasgow GIl 5JR, UK

(Received 19 June 1991, revised received 16 September 1991, accepted 7 October 1991)

The reactivities of antibodies with branched and monomeric peptides were compared in ELISA assays. We found that lower amounts of antibodies could be detected with branched peptides than with monomeric peptides. This was observed with a monoclonal antibody and with antibodies in the sera of various HIV-positive individuals. To investigate the physical aspects of branched peptides important for the observed increase in sensitivity, glycine spacers of different lengths were introduced between the branched lysine core and the epitope reacting with the monoclonal antibody, The effect of the number of glycine residues, both on the sensitivity of antibody detection and on the amount of branched peptide needed to produce a given signal, was studied and the optimum was found at 4-5 residues. We discuss the basis for these findings and conclude that the routine use of branched peptides for serodiagnosis will give both greater sensitivity and appreciable cost savings. Key words: Branched peptide; Serodiagnosis; AIDS; EUSA

Introdudion Synthetic peptides corresponding to short stretches of protein sequences can be recognised by some antibodies directed against the whole protein. Peptides are attractive reagents for detection of serum antibodies directed against proteins because they are relatively easy to synthesise and are generally cheaper to produce than the whole protein. Peptide-based ELISA kits have been found to be sensitive and specific for the

Correspondence to: H.S. Marsden, MRC Virology Unit, Institute of Virology, Church Street, Glasgow G11 5IR, UK. (Tel.: 041-339.8855, ext. 4639; Fax: 041-337.2236).

diagnosis of HIV specific antibodies in AIDS patients (Niirviinen et aI., 1988) and are now commercially available. Recently, branched peptides (Possnett et aI., 1988; Tam, 1988) have been used to detect antibodies (Tam and Zavala, 1989; Del Giudice et al., 1990). These branched peptides consist of a small, two-fold bifurcating, polylysine core onto which the peptide of interest is synthesised. Tam and Zavala (1989) compared the amount of branched and monomeric peptides needed to detect monoclonal antibodies by RIA and found that the amount of antibody detectably bound to branched pep tides was considerably greater than that bound to monomeric peptides within the range of peptide concentrations used. In this report we have

66

used a variety of antibodies and peptides to provide evidence for the general applicability of their findings. From a clinical viewpoint, an important aspect of any serodiagnostic technique is the sensitivity with which antibodies can be detected. We show, for the first time, that lower concentrations of antibodies can be detected with branched peptides than with monomeric peptides. The increased sensitivity of branched peptides was demonstrated for a monoclonal antibody and, most importantly, for antibodies present in the sera of HIV-positive individuals. To optimise the manner of epitope presention we investigated the effect of varying the distance between the reactive epitope on the branched peptide and the polylysine core. We conclude from our findings that branched peptides have considerable potential and compelling advantages for peptide-based serodiagnosis. Materials and metbods

Oligopeptides Monomeric and branched peptides (Possnett et al., 1988; Tam, 1988) were synthesised by continuous-flow Fmoc chemistry using a Novabiochern peptide synthesiser (Mclean et al., 1991). Following synthesis, side chain protecting groups were removed by treatment with 95% trifluoroacetic acid in water (plus 5% w/v phenol for the arginine-containing peptides, or 3% w/v phenol plus 2% ethanedithiol for the peptide containing cysteine and arginine). Relative molecular masses of the monomeric peptides were determined by mass spectrometry (M-Scan, Cambridge) which gave values identical to those expected. The amino acid compositions of branched peptides, determined by amino acid analysis (Cambridge Research Biochemicals), were in agreement with expectations. Immunological reagents The rabbit antisera used in tbis study have been described elsewhere (Mclean et al., 1991). Monoclonal antibody ZIFll, specific for the herpes simplex virus type 1 (HSV-1) 65 kDa DNA binding protein (Schenk et al., 1988) was kindly

provided by Dr. P. Schenk and Professor H. Ludwig. This antibody recognises the sequence GDPEDLD corresponding to residues 360-366 of the protein (Murphy et al., 1989). Monoclonal antibody HCMV-3 is specific for a 72 kDa early protein induced by human cytomegalovirus (HCMV) and was obtained from Bioscott, Edinburgh, Scotland. Human HIV-positive sera H0457 (Department of Genetics, University of Edinburgh), ESI (Regional Virus Laboratory, Ruchill, Glasgow), EBH2 and EBH3 (Regional Virus Laboratory, East Birmingham Hospital) were obtained from HIV-1 infected individuals. Control sera were obtained from healthy HIV antibody-negative individuals. All sera were beat inactivated at 60°C for 30 min prior to use. Enzyme-linked immunosorbent assay (ELISA) Peptides were dissolved at 10 mg/ml in water (bubbled with ammonia vapour for peptides 161, 172, 177 and 204 and at 20 mg/ml in 30% acetic acid for peptide 259). The peptides were then diluted in PBS to solutions ranging in concentration between 25 pg/ml and 1 mg/ml. 100 ILl of antigen solution were added to each well of a microtiter plate Ommulon 1 or 2, as specified in the text, Dynatech) and allowed to adsorb at 37"C overnight. Excess antigen solution was removed and the plates were blocked with a solution containing 1% BSA for 1 h at 37°C. The plates were then washed three times in PBS containing 0.01 % Tween 20 (PBS-Tween) and incubated with 100 ILl of the appropriate antiserum dilution at 37"C for 1 h. For the detection of bound antibody, plates were washed five times in PBS-Tween and incubated for 1 h at 37"C with 100 ILl/well of horseradish peroxidase-protein A conjugate (Sigma) (diluted lOOO-fold in PBS) or horseradish peroxidase-sheep anti-human conjugate (SAPU) (diluted lOOO-fold in PBS). Plates were then washed seven times in PBS-Tween and reacted with a 50 mg/ml solution of the enzyme substrate 2' 2-azino-bis(3-ethylbenzthazoline-6-su Iphonic acid) (ABTS) in citrate phosphate buffer containing 0.01 % hydrogen peroxide. After 15-30 min of colour development the plates were read on a Titertek Multiscan plate reader at 405 nm. All results given are the mean of duplicate determinations.

67 TABLE I PEPTIDES USED IN TIllS STUDY Peptide number

Peptide sequence •

Gene predicted to contain the sequence

Position of peptide in protein

Number of amino acids in predicted protein

161 171 172 173 175 177

(Y)LYATFDEFPPP b DLGQESAFEKEY b (Y)GGFVQFVTSTRNA b (Y)GAAALRAHVSGRRA b (Y)LTPANLIRGDNA b (Y)THHLVKRRGLGA b YRPLGPGTPPMRALPA b EFGLRNCQFLAVGPD b LQARlLAVERYLKDQQL b VSHGDPEDLDGAARAGE f (GDPEDLDG,,) g.h

HSV-l UL31 c HCMVUL45 e HSV-1 UL45 c HSV-l UL47 c HSV-l UL46 c HSV-1 UL41 c HSV-1 UL45 c HCMVUL45 e HIV-1 gp41 d HSV-l UL42 c HSV-l UL42 c

290-301 514-525 156-168 617-684 17- 28 13- 23 10- 26 760-774 67- 83 357-373 360-366

301 770 172 693 718 480 172 770 347 488 488

202

204 259 216 Gil

Tyrosine residues in parentheses (Y) are not part of the natural protein sequence and were added to enable them to be coupled to a carrier protein for immunisation purposes (McLean et aI., 1991). b Both monomeric and branched peptides were synthesised. Branched peptides consisted of eight copies each synthesised onto a branching lysine core, general formula: (monomeric peptide)gK 7 A. Branched peptides did not contain non-natural tyrosine residues. C McGeoch et aI., 1988. d Desai et aI., 1986. e Chee et aI., 1990. f Only the monomeric form was synthesised. I Only the branched form was synthesised. h The carboxy Gil residues (where n = 0, 1, 2, 4, 5, 7, 10, 15 or 20) are not in the natural sequence but were added to provide spacers of different lengths. a

1.0

Results

Previously, Tarn and Zavala (1989) had reported that two monomeric peptides with 12 and 17 residues showed no reactivity at concentrations up to 30 p.g/ml with specific polyclonal or monoclonal antibodies while the corresponding branched peptides were immunoreactive at 0.011 p.g/ml. The availability of a large number of rabbit sera raised against both branched peptides and protein-conjugated monomeric peptides (McLean et aI., 1991) permitted us to test the general applicability of their finding. The reactivities in ELISA assays of sera raised against the monomeric peptide DLGQESAFEKEY (peptide 171 in Table I) and the branched form (DLGQESAFEKEY)8K7A with both monomeric and branched peptides are shown in Fig. 1. The

e-c

0.8

10

0

~

•u

.

0.6

C

0.4

.D ~ 0 til .D

0.2

0(

Amount of peptIde per well (1'9)

Fig. 1. ELISA reactivities of rabbit antiserum raised against the monomeric peptide DLGQESAFEKEY coupled to KLH (0, e), rabbit antiserum raised against the branched peptide (DLGQESAFEKEY)gK 7A not coupled to carrier protein (A, .) and pre-immune serum (0, . ) all used at 1/8 dilution, with monomeric (open symbols) and corresponding branched (closed symbols) peptides.

68

amount of both monomeric and branched peptide was varied from 2 X 10- 6 ILg to 10 2 ILg to cover and extend the range used by Tam and Zavala (1989). The results show that the amount of branched peptide needed to produce a given absorbance in the assay was about 10 3-10 4-fold lower than that of the corresponding monomeric peptide. The reactivities of a further seven peptides (161, 172, 173, 175, 177, 202 and 204, Table 1) with antisera raised against them were tested and all exhibited similar behaviour in that the antibodies were detected with 10 3 -10s-fold lower amounts of the branched peptides than of the corresponding monomeric peptides (data not shown). Invariably this effect was independent of 20

A

whether the antisera had been raised against monomeric peptides (coupled to carrier protein) or branched peptides (not coupled to carrier). These observations suggested to us that branched peptides could afford considerable savings in cost and materials in peptide-based serodiagnosis. To test this possibility and assess the ability of branched peptides to detect antibodies in the serum of HIV-infected individuals, branched and monomeric peptides of sequence LQARILAVERYLKDQQL corresponding to amino acids 67-83 of gp41 of HIV-l (Table I) were synthesised and tested for reactivity with four different sera from HIV-positive individuals. Comparative reactivity with branched and monomeric peptides was measured at half the maximum optical den2.0

•

1.6

16

12

-~ O~ -

08

~

J

20

C

2.0

16

1.6

12

1.2

08

0.8

04

0.4

D

Amount of peptide per well ()Jg) Fi•. 2. EUSA reactivity of HIV-positive sera at 1/100 dilution with the monomeric peptide LQARILAVERYLKDQQL 01' branched peptide (LQARILAVERYLKDQQL)sK 7 A (peptide 259 in Table I). A: reactivity of control HIV-neptive serum with monomeric (0), and branched (D) peptides, or serum H0457 with monomeric (e) and branched (.) peptides. B: reactivity of control HIV-ne.ative serum with monomeric (0), and branched (D) peptides, or serum EBH2 with monomeric (e) and branched (.) peptides. C: reactivity of serum EBH3 with monomeric (e) and branched (.) peptides. D: reactivity of serum ESt with monomeric (e) and branched (.) peptides.

69

sity obtained above background. The results demonstrate that with all four sera the branched peptide was reactive at lower concentrations than the corresponding monomeric peptide (Fig. 2). To quantitate these data the concentrations of peptide needed to give the mean of the minimum and maximun absorbance was determined for both branched and monomeric forms and indicated that between 20-1oo-fold less branched peptide was needed to produce a given signal. The relative sensitivity of monomeric and branched peptides for serodiagnosis was next investigated. Serum samples ES1, EBH2 and EBH3 from three HIV-1 infected individuals were serially diluted and reacted with peptides 259 (Table I) in an ELISA assay. The results (Fig. 3) demonstrate that the branched peptide was more efficient at detecting antibodies in these sera. For serum ES1, the dilution at which the monomeric peptide ceases to detect antibodies is 2s whereas for the branched peptide it is at least 212: a difference in sensitivity of at least 128-fold (Fig. 3A). Similarly, branched peptides detect antibodies in sera EBH2 and EBH3 with a 256 and 16-fold greater sensitivity respectively than do the equivalent monomeric peptides (Figs. 3B and 3C). The effect of varying the distance between the epitope and the core on the reactivity of the branched peptide with the detecting antibody was then investigated. To do this we employed monoclonal antibody ZlFll (Schenk et aI., 1988), previously shown to recognise the sequential epitope GDPEDLD (Murphy et aI., 1989). A series of peptides were synthesised in which different numbers of spacer glycine residues were added to the lysine core followed by the sequence GDPEDLD on each of the branches. These peptides can be represented by the general formula {GDPEDLDGn)8K7A. Nine peptides were made with n, the number of glycine residues, having the values 0, 1, 2, 4, 5, 7, 10, 15 and 20. In the first experiment, shown in Fig. 4, different amounts of peptides, ranging from 100 ILg down to 100 fg were tested in ELISA assays for reactivity with ZlFll (diluted 1/2(00). The controls used included: monoclonal antibody HCMV-3, peptide 171 (Table I) in both monomeric (A) and branched (C) forms and monomeric peptide 216 (VSHGDPEPLD.

2.0

A

1.6

'.2 0.8

~

0." 0.0

E c:

-8

2

0

4

6

9$*~ 8

10

12

14

.0

.2

.4

~

14

•

2.0 1.6

10

~

1.2

i

0.8

U)

0.4

-g ~

0.0

~,

0

2.0

2

4

6

8

C

1.6

1.2 0.8 0.4 0.0

0

2

4

6

8

10

12

L092 dilution of human serum Fig. 3. Effect of diluting control HlY·negative (open symbols) or HIY.positive (closed symbols) serum on reactivity with monomeric LQARlLAVERYLKDQQL (0, e) peptide or with branched (LQARILAVERYLKDQQL)8K7A (0, .) peptide. A: serum ES1; B: serum EBH2; C: serum EBH3.

GAARAGE) which corresponds to amino acids 357-373 of the protein encoded by HSV-1 gene VIA2 (McGeoch et aI., 1988) and contains the ZIFll-reactive epitope. The figure shows that all branched peptides containing the sequence GDPEDLD reacted more strongly with ZlFll than did the monomeric peptide 216. However,

70

there was a marked difference between the branched peptide with no glycine spacer (Go), which was only about tenfold more reactive than the monomeric peptide, and the branched peptide with one glycine spacer (G.): at 100 ILg peptide per well G. gave twice the amount of absorbance produced by Go while 10 4-fold more Go than G. was needed to produce an absorbance of 0.5. Additional glycine residues progressively increased the absorbance for amounts of peptide above 10- 5 ILg/well although peptides G 5 and G 7 were most sensitive at low concentrations. The controls behaved as expected: firstly, the non-related antibody HCMV-3 did not react with any peptide at any concentration (only the data for GIO is shown); secondly, ZIFll did not react with the unrelated 171A (monomeric) and 171 C (branched) peptides; and thirdly, in the absence of peptide, ZIFll was not reactive. In the second experiment (Fig. 5) the effect of glycine spacer length on the sensitivity with which branched peptides could detect monoclonal antibody ZIFll was examined. Three different _

GO

- 0 - 01 ••••• -. 01/

.

_at

ic

-"0--' 04

_05

o ~

•

"C

~

I

-.-. .-. 010 -'-0--' 015

~

_

A

C

02D

- - 0 - 211 - _ _•

171A

-_--. I71C _

0

•

!!

_ 0

-

-

_1111>

0

-

Amount of peptide per~)

Fig. 4. Effect of spacer glycines on recognition of monoclonal antibody ZIFll by peptides containing the epitope GPDEDLD. Peptides G o-G2I) represent branched peptides containing 0-20 spacer glycines as discussed in the text. Peptide 216, VSHGDPEDLDGAARAGE, is monomeric and represents pan of the natural sequence of HSV-l protein and contains tbe ZIFll-reactive epitope. Control peptides 171A (monomeric) and 171C (branched) (Table I) do not react with ZIFl1. HCMV 3 is an unrelated antibody.

0.•••0.

..

~

"'u..'.

E C

---e-

0

~

• •. Q

C

GIl

- 0 - G1

OR

--0- QI

--A-

•.,

--a-

GIS

~

GIlD

I

--A-

..

A 0

A

C

- 0 - III

0

.. .. 0

$

0 0

';::-

0 0

0

0 0

0

~ ~

0 0

0

0 0 0

~ ~

-.. .. 0 0 0

0 0 0

~

Dilution of MAb Z1F11

Fig. S. Effect of spacer glycines on the sensitivity of detection of monoclonal antibody ZIFII by branched peptides containing the epitope GPDEDLD used at I !-II/well. Peptide 216. VSHGDPEDLDGAARAGE, is monomeric and represents pan of the natural sequence of HSV-I protein UlA2 and contains the ZIFll-reactive epitope.

amounts of peptide were tested: 100 ILl, 1 ILl and 0.01 ILg. As all lave essentially the same results only the data for 1 ILg/well is presented. The reactivity of the peptides was tested against doubling dilutions of ZIFll ranging from 1/250 to 1/64.000 and all forms of the branched peptide were significantly more reactive than the control monomeric peptide 216. The effect was particularly marked with four or more glycine spacer residues. When diluted 64,OOO-fold ZIFll yielded an absorbance of 0.5 with G4 peptide whereas to produce the same absorbance with peptide Go the antibody could only be diluted 250-fold. Additional glycine residues did not significantly increase the reactivity of the branched peptide. We conclude from this experiment that a spacer of four or more glycine residues increases the sensitivity for detection of antibody ZlFll by at least 256-fold. Discussion An important new finding reported here is that branched peptides can detect lower amounts

71

of antibodies than can monomeric peptides. The increased sensitivity was demonstrated for both sera from HI V-infected individuals (Fig. 3) and a monoclonal antibody (Fig. 5). Our findings have obvious relevance to clinical serodiagnosis. Our use of the word sensitivity should not be confused with that of Tam and Zavala (1989) who used it to describe their finding that antibodies could be detected by lower amounts of branched peptides than monomeric peptides. In their experiments antibodies specific for two distinct epitopes were employed to demonstrate that branched peptides readily detected the appropriate antibody whereas monomeric pep tides, even at a 256-fold higher concentration, did not. In addition to our new finding that branched peptides detect lower amounts of antibodies we have also provided evidence for the general applicability of the findings of Tam and Zavala (1989). In our studies, the reactivities of three HIV-positive human sera (Fig. 2), rabbit sera directed against eight different herpesvirus peptide sequences (Fig. 1 and Table I), and one HSV-l monoclonal antibody (Fig. 4), with peptides covering a 108-fold range of concentrations, were investigated. Without exception a given amount of the antibodies could be detected with less branched peptides. The physical characteristics of branched peptides important for the detection of antibodies were investigated by utilising a monoclonal antibody (ZlFll) reactive with a known epitope and by introducing glycine spacers of different lengths between the epitope and the polylysine core. The effect of these glycine spacers on the amount of peptide needed to produce a given signal and on the sensitivity with which the antibodies could be detected was examined. The optimum length of spacer was found to be four or five residues. Amino acids other than glycine (for example alanine, non-natural amino acids or part of the natural sequence of the protein) could form the spacer perhaps with advantage. This possibility is under investigation. What makes branched peptides so superior? Several factors could contribute. First, branched peptides may bind more readily to the microtitre wells than do monomeric peptides (Tam and Zavala, 1989). Second, multivalent binding' be-

tween antibody and branched peptide would result in a considerable increase in stability, compared to simple monovalent binding of monomeric peptide (Tam and Zavala 1989). Thus there may be up to a l03-fold increase in the binding energy of IgG when both valencies (binding sites) are used (Roitt et aI., 1989). Third, to be immunoreactive monomeric peptides have to both bind to the plastic well and remain available for binding to the antibody, whereas one or more arms of a branched peptide could bind to the plastic and leave the others free to bind to the antibody. It is possible that in binding to the plastic, physical constraints are imposed on a monomeric peptide to prevent it from interacting freely with the antibody. Tam and Zavala (1989) showed that monomeric peptides were less immunoreactive than were branched peptides and suggested that this was because the branched pep tides bound more readily to the plastic. However, the third possibility above could equally account for their findings. We think that to unambiguously decide amongst these possibilities the amounts of peptide bound will have to be directly measured. Our data are concordant with, but do not prove, that the increased sensitivity of branched peptides derives in part from their multivalent nature. The effect (Fig. 5) of the glycine spacer could be to increase the separation of the reactive epitopes on two arms sufficiently to contact the two antigen binding sites of an antibody molecule simultaneously. In our experience the synthesis of branched peptides is no more difficult or expensive than that of monomeric peptides. The large savings potentially available when using branched peptides for peptide-based serodiagnostic kits and their greater sensitivity lead us to suggest they will become the reagents of choice.

Acknowledgements

We thank Drs. P. Simmonds and A. LeighBrown for supplying serum H0457, Dr. E. Follet for serum ES 1 and Dr. U. Desselberger for sera EBH2 and EBH3.

72

Project grant support for A.O. from Glaxo Group Research Limited is gratefully acknowledged. GMcL and CR were supported by MRC studentships for training in research methods. SG was supported by a project grant from the MRC AIDS directed research programme. References Chee, M.S., Bankier, A.T., Beck, S., Bohni, R., Brown, C.M., Cerny, R., Horsnell, T., Hutchison III, c.A., Kousarides, T., Martigenetti, J.A., Preddie, E., Satchwell, S.C., Tom· linson, P., Weston, K.M. and Barrell, B.G. (1990) Analysis of the protein coding content of the sequence of human cytomegalovirus strain AD169. Curro Top. Microbio!. Immuno!. 154,125-169 (HCMV sequence in EMBL Release 23, May 1990 accession no. XI7403). Del Giudice, G., Tougne, c., Louis, J.A, Lambert, P.-H., Bianchi, E., Bonelli, F., Chiappinelli, 1.. and Pessi, A. (1990) A multiple antigen peptide from the repetitive sequence of the Plasmodium malariae circumsporozoite protein induces a specific antibody response in mice of various H-2 haplotypes. Eur. J. Immunol. 20, 1619-1622. Desai, S.M., Kalyanaraman, V.S., Casey, J.M., Srinivasan A, Anderson, P.R. and Devare, S.G. (1986) Molecular cloning and primary nucleotide sequence analysis of a distinct human immunodeficiency virus isolate reveal significant divergence in its genome sequences. Proc. Nat. Acad. Sci. U.S.A. 83, 8380-8384.

McGeoch, 0.1., Dalrymple, M.A., Davison, AJ., Dolan, A, Frame, M.C., McNab, D., Perry, L.J., Scott, J.E. and Taylor, P. (1988) The complete DNA sequence of the 10lIl unique region in the genome of herpes simplex virus type 1. J. Gen. Viro!. 69, 1531-1574. McLean, G.W., Owsianka, A.M., Subak-Sharpe, J.K. and Marsden, H.S. (1991) Generation of anti-peptide and anti-protein sera: effect of peptide presentation on Unmunogenicity. J. Immuno!. Methods 137, 149-157. Murphy, M., Schenk, P., Lankinen, H.M., Cross, AM., Taylor, P., Owsianka, A., Hope, R.G., Ludwig, H. and Marsden, H.S. (I 989) Mapping of epitopes on the 6SK DNAbinding protein of Herpes simplex virus type 1. J. Gen. Viro!. 70, 2357-2364. Nitvinen, A., Korkolalnen, M., Suni, J., Korpela, J., Kontio, S., Partanen, P., Vaheri, A. and Huhtala, M.-L. (1988) Synthetic env gp41 peptide as a sensitive and specifIC diagnostic reagent in different stages of human immunodeficiency virus type 1 infection. J. Med. Virol. 26, 111-118. Posnett, D.N., McGrath, H. and Tam, J.P. (1988) A novel method for producing anti-peptide antibodies. J. BioI. Chern. 263, 1719-1725. Roitt, I., Brosnoff, J. and Male, D. (1989) Immunolo8)', 2nd edn. Gower Medical Publishing, London, pp. 7.2-7.3. Tam, J.P. (1988) Synthetic peptide vaccine design: Synthesis and properties of a high-density multiple antigenic peptide system. Proc. Nat. Acad. Sci. U.S.A. 85, 5409-5413. Tam, J.P. and Zavala, F. (1989) Multiple antigen peptide. A novel approach to increase detection sensitivity of synthetic peptides in solid-phase immunoassays. J. Immunol. Methods 124,53-61.