optimisation of a peptide-based indirect elisa for the detection of antibody in the serum of hiv-1...

ELSEVIER Journal of !mmunological Methods 200 (1997) 79-88

JOUIINAL OF IM ICAL METtiOUS

Optimisation of a peptide-based indirect ELISA for the detection of antibody in the serum of HIV-l seropositive patients

Louise Brown a, Mike Westby a, Bernard E. Souberbielle ‘, Paul W.S. Szawlowski b, Graham Kemp b, Phillip Hay ‘, Angus G. Dalgleish a. *

’ G-U Medicirte. St. George’s Hospiial, London SWI 7 ORE. UK

Received I6 February 1996; revised I 1 September 1996; accepted 15 September 1996

Abstract

A modified peptide-based indirect ELISA technique for the detection of HIV-I specific antibodies in the sera of HIV-l seropositive individuals is described. We found that the reduction of non-specific binding of HIV-l seropositive sera to the ELBA plate was essential for the reliable detection of serum antibodies in the peptide based indirect ELISA. Optimal results were obtained using Immulon microtitre plates, different concentrations of denatured, purified grade of casein in the blocking ( 1%) and washing (0.25%,) solutions and by diluting HIV-I seropositive sera 1 in 1600. These conditions reduced non-specific binding and improved assay sensitivity. We show that the inclusion of a control peptide is essential to reducing the incidence of false positive and false negative results. Taken together, the modifications described in this report improve reliability of the peptide-based indirect ELBA without compromising its sensitivity and have particular relevance for those wishing to apply the peptide-based indirect ELISA technique to serum samples which exhibit high levels of non-specific binding. To illustrate this, levels of antibody in the sera of HIV-I seropositive and seronegative donors that are specific for peptides derived from a conserved region of HIV-I gpl20 sharing homology with the FAS apoptosis antigen were analysed

using this technique.

Krword~: ELISA: Synthetic peptide: Human serum; Anti-HIV- I antibody detection: Non-specific binding; Virology

1. Introduction

Abbreviations: A,,,,. absorbance read at 450 nm; ELISA.

enzyme linked immunosorbent assay: HIV- I. human immunodeficiency virus type 1: gp130. I20 kDa envelope glycoprotein; OPD.

o-phenylenediamine dihydrochloride: SLE. systemic lupus erythe-

matosus.

Corresponding author. Tel.: +44-(0)181-725-5815: Fax:

+ -V-(O)1 8 l-725-2992.

Sera from patients infected with HIV contain a variety of autoantibodies. It is unclear whether the

antibodies can be explained purely by the activation of B cells with resultant hypergammaglobulinaemia. Alternative explanations include cross-reactivity between HIV and self protein sequences such as HLA, IgG, FAS, etc. (Dalgleish, 1995). Whilst seeking the presence of antibodies to specific sequences in order

0022.1759/97/$17.00 Copyright 8 1997 Elsevier Science B.V. All rights reserved

P/l S002’~1759(96)00192-5

80 L. Brown et al. / Journal of Immurmlogical Methods 200 f 1997) 79-88

to examine this possibility we became aware that published methods resulted in an assay with unac-

ceptably high background levels using HIV positive sera. Moreover, it does not appear to be the rule to

include a control peptide. The purpose of this study was to optimise a peptide based indirect ELISA for

detecting subtle differences in epitope recognition in

sera from patients with hypergammaglobulinaemic

sera which is often connected with chronic infectious and chronic autoimmune diseases.

2. Materials and methods

2.1. Peptide antigens

Descriptions of the peptides used in this study are

summarised in Table 1. ARP7022 and ARP740-27 were obtained from the MRC AIDS Reagent Project

(H. Holmes, NIBSC, UK); p12 was supplied by E.F. Hounsell (Department of Biochemistry, UCL, UK); HSP65 derived peptide was synthesized by B. Austen

(Department of Surgery, St. George’s Hospital Medi- cal School, UK) and CV-FAS was synthesised in our

laboratories. Purity of the peptides was assessed by HPLC analysis. The optimum concentration of

ARP7022 peptide antigen was tested, in the range

l-32 pg/ml, in the ELISA system.

2.2. Human serum samples

HIV seropositive blood samples were obtained from patients attending the Genito-Urinary Depart- ment of St. George’s Hospital. HIV-l seropositive

Table I Peptides used in this study

Peptide Amino acid sequence Description

sera used in the optimisation of the peptide-ELISA

were from donors with > 300 cu/mm3 CD4+ T

cells. HIV-l seropositive sera analysed for anti-

ARP740-27 antibodies were randomly chosen from both asymptomatic and symptomatic HIV- 1 infected

donors, from HIV-l positive donors who have re-

cently seroconverted and from HIV-l positive

African subjects living in Uganda. Sera from both

the recent seroconverter and African groups were obtained from the Cambridge HIV Repository, with the help of U. Desselberger, Clinical Microbiology

and Public Health Laboratory, Addenbrooke’s Hospi- tal, Cambridge. Dates of the last negative and first

positive HIV-l results. and the date of the serum

sample collection for each recent HIV-l serocon-

verter are summarised in Table 2. HIV seronegative

serum samples screened for both HIV-l and HIV-2 were obtained from the South Thames Blood Trans-

fusion Service, London, in collaboration with T. Kaye. Serum was prepared from whole blood by

centrifuging at 1000 X g for 20 min, in a serum separation vial, and stored as aliquots at -20°C.

2.3. Rabbit immunisation

Peptide CV-FAS was coupled to the carrier protein human serum albumin (HSA), following the

method described by Scheidtmann (1989). Rabbit

immunisation was carried out as previously described by Webster et al. (1993).

2.4. ELISA method

Peptide antigens were dissolved in 0.05 M sodium carbonate, pH 9.6 buffer and dispensed into dupli-

CV-FAS CVEINCTR 8 residue peptide derived from HIV-l gp120 (aa296-303). Used in ELISA

to construct a standard curve.

ARP740-27 VEINCTRPNNNTRKRIRIQ 19 residue peptide derived from HIV- I gp 120 (aa297-3 15).

HSP65 MAKTIAYDEEARRGL HLA-DR3 restricted T cell epitope derived from Myobacterium leproe heat

shock protein. Used in the ELISA to detect non-specific binding of rabbit serum.

ARP7022 DQQLLGIWGCSGKLICTTAVPWNC 24 residue peptide derived from a conserved immunodominant region of HIV- 1

gp41 (aa593-616). recognised by most European and African HIV+ sera

(Lange et al., 1993). Used in the ELBA as a positive control peptide

for HIV seropositive sera.

PI2 RAKTVERKVERRK Scrambled sequence of the C5 region of HIV- 1 gp 120. Used in the ELISA as

a control peptide to detect non-specific binding of HIV seropositive sera.

Table 2

L. Brown et al. /Journal of Immunological Methods 200 C 1997) 79-88 81

Dates of last negative and first positive HIV-I tests, and date of sample collection for each of the recent seroconverter donors tested in the

peptide-ELISA

Donor Last negative HIV result

2061 2/3/94

2063 18/5/94

2113 22/5/95

2127 15/7/95

2034 14/l/94

First positive HIV result Date of sample collection

2/8/94 8/8/94

10/8/94 I8/8/94

17/6/95 28/6/95

2/10/95 1 l/10/95

s/9/94 5/10/94

cate wells of a 96 well flat based microtitre plate at 50 ~1 per well. Peptides were bound to the mi-

crotitre plate wells overnight, at 4°C. The plates were blocked overnight with 200 ~1 casein solution (see

below). Serum was diluted with casein solution (see below) and added in a volume of 100 ~1 per well for 1 h at 37°C. The plates were then washed six times

with wash solution (see below) using the WWO04

Wellwash 4 microplate washer (Denley, UK). Immo-

bilised peptide-specific serum antibodies were de- tected indirectly using a biotinylated species specific

anti-IgG polyclonal antibody (anti-rabbit: RPN 1003 and anti-human: RPN 1004, both Amersham, UK) followed by streptavidin linked horseradish peroxi-

dase conjugate (RPN 123 1, Amersham). Both the anti-IgG antibodies and the enzyme conjugate were used at a dilution of 1 in 1000 in diluent (see below),

dispensed into each well in a volume of 100 ~1 and

incubated for 1 h at 37°C. Each of these incubations were followed by a washing step, as described be-

fore. The reaction was measured after addition of the

substrate, o-phenylenediamine dihydrochloride

(OPD) (Sigma), following the manufacturer’s in-

structions at 450 nm, using the WSO50 Wellscan automatic microplate reader (Denley).

The peptide-based indirect ELISA was modified

by optimising the following experimental parame- ters.

2.4. I. Microtitre plate types The following microtitre plate types were tested:

Immulon 2 and 4, both Dynatech Laboratories, VA, USA; MaxiSorp (PS flat well), Gibco BRL, UK; 655061 (MLN high bind, flat well) and 655001 (MLN medium bind, flat well), both Greiner

Labortechnik, UK and Falcon (flexible PVC flat well) manufactured by Becton Dickinson Labware, CA, USA.

2.4.2. Blocking solutions Blocking solutions were prepared by dissolving

10 g casein powder in 100 ml 75 mM NaOH, with

further dilutions being made using phosphate buffered

saline (PBS). The stock solution was stored at 4°C and used within a week. Denatured regular grade

casein, from Sigma, and more purified grade caseins, from BDH (Hammersten grade) and J.T. Baker

(Baker grade), were compared for high specific to

non-specific binding ratios, using HIV seropositive

sera. In addition, each make of casein was tested at

three different concentrations: 1%. 2.5% and 5%.

2.4.3. Diluent and wash solution

Diluent/wash solution was prepared as described for the blocking solution, except for the addition of Tween 20 (BDH) to a final concentration of 0.05%.

Different denatured casein solutions (Sigma, BDH

and J.T. Baker, as before) were compared for high specific to non-specific binding ratio using HIV

seropositive sera, at three different concentrations:

O.l%, 0.25% and 0.5%.

2.4.4. Data analysis Included in each ELISA were ‘blank’ (serum

omitted in serum incubation step) and ‘peptide blank’ (wells containing control peptide, in which serum

was incubated) wells. Absorbance values at 450 nm (A,,,) obtained f or wells containing test and control

peptides were both corrected for non-specific binding of anti-IgG/enzyme conjugate to peptide/blocking protein by subtracting the mean blank well A,,

value. Final absorbance values were corrected for

non-specific binding of sera to the control peptide p 12 by subtracting the mean A,,, peptide blank well value from the A,,, values for the wells containing the test peptide. Specific to non-specific binding ratios were calculated by dividing the mean AGSo

82 L. Brown et nl. / Journal qf Immtmologicul Methods 200 C 19971 79-88

value for test peptide containing wells with the mean

A 450 value for peptide blank wells.

2.4.5. Quantifwation of serum antibody leaels To permit comparisons of antibody levels be-

tween different microtitre plates, a method of quanti- fying immobilised specific antibody using sera from

a rabbit immunised with CV-FAS was devised. The concentration of protein in the CV-FAS specific sera

was first determined using the BCA protein assay microtitre plate method (Pierce, IL, USA). During the ELISA assay, a standard curve of absorbance

values corresponding to doubling dilutions (from l/400 to l/12 800) of the CV-FAS specific rabbit

sera binding to CV-FAS peptide was included in each ELISA plate. The absorbance values for each

sample were then expressed as corresponding arbi-

trary protein concentration values using this standard

curve.

3. Results

3.1. Optimisation of the ELISA for the detection of

antibody in the sera of HIV-I seropositiue donors

The optimal dilution factor for HIV- 1 seropositive sera, in terms of low non-specific and high specific

binding, was tested using sera from two HIV-l positive and two HIV negative donors (Fig. la>. The detection of anti-ARP7022 serum antibodies increased as the serum dilution increased between 1 in

Fig. I. Optimisation of the peptide-ELISA. a illustrates the opti-

mum serum dilution factor in the ELISA to detect anti-ARP7022

antibodies in the sera of two HIV-infected individuals, HIV#I

and HIV#2. ARP7022 was used at 20 pg/ml and bound to an

Immulon 2 microtitre plate. BDH casein was used in the blocking

and wash/diluent solutions at 2.5% and 0.25%, respectively.

Absorbance values obtained with increasing ARP7022 peptide

concentrations are shown in b. In this assay, sera were diluted

l/1600 and the microtitre plate used was Immulon 2. Blocking

and wash/diluent solutions were as before. Absorbance values

using different microtitre plate types in the ELISA to detect

anti-ARP7022 antibodies from one high and one low binding

HIV-seropositive sera are shown in c. ARP7022 was bound to

microtitre plate wells at 16 pg/ml. Blocking and wash/diluent

solutions as before. Standard deviations are represented by error

bars in each panel.

400 and I in 800. The highest AJ5,, values were

observed at around 1 in 1600 serum dilution using sera from HIV-l seropositive donors #l and #2. At serum dilutions greater than 1 in 1600, A,,, values

decreased for both sera tested. Binding of HIV-l seropositive sera to the negative control peptide, ~12, also decreased with increasing serum dilution. The greatest differential between specific and non-specific

A 450 values was obtained using serum diluted to 1 in 1600. which was used in the following experiments.

(a) 1.5

1 T n ADP7022 / HIV+ dciwr ll 0 ADP7022 / HIV+ doror X2 0 p12 I HIV* danor #I Opl2/HIV+donor#2 A ADP7022 I HIV- &nor Xl l ADP7022 / HIV- donw #2

,,u#) 11800 1/1600 113200 1/54001H250m/25500

serum dilution

(W 2.0

1 q ADP7022 / HIV+ sem

0 p12 I HIW aam

0 4 5 12 16 20 24 25 32

peptida concenfration (p@nl)

l/800 l/l600 113200

serum dilulion

ADP7022 / HIV

0 Muiso~ HlVX3

l Maarorp: HW4

0 lmmular 2: HlV#3

0 lmmuka 2: HlWf4

A lmmulon 4: HIM3

A lmmulon 4: HIV44

L. Brown et al. /Journal of Immunological Methods 200 (1997) 79-88 83

The optimum concentration of ARP7022 was

tested, in the range l-32 Kg/ml, using sera from an HIV-l seropositive donor and, as a control, sera from a seronegative donor. Absorbance values, corrected for non-specific binding, are shown in Fig. lb.

The highest A,,, values were obtained when a 16

pg/ml ARP7022 peptide solution was used (A,,,: 1.4835 + 0.03 IO). At peptide concentrations below 16 pug/ml, HIV- 1 seropositive sera also bound to

the wells containing the control peptide. Low level binding of HIV seronegative sera to ARP7022 coated

wells was also observed at peptide concentrations below 16 pug/ml, but not at levels seen with HIV- 1 positive sera. ARP7022 was used at a concentration

of 16 pg/ml in subsequent experiments. Six different microtitre plates were tested in the

ELISA to detect anti-CV-FAS antibodies in CV-FAS specific rabbit sera. The flexible microtitre plate was

found to give false positive results due to the increased incidence of cross-contamination between

adjoining wells (results not shown) and was there- fore rejected as a possible candidate microtitre plate for the ELISA using human sera. The microtitre

plates Immulon 2 and 4, and MaxiSorp performed well in the assays using CV-FAS specific rabbit sera

(results not shown) and were selected for testing in the ELISA using human sera. Both high and low ARP7022 binding HIV-I seropositive sera were as-

sayed for anti-ARP7022 antibodies using each of the

three microtitre plates. A,,, values, corrected for

non-specific binding, are shown in Fig. lc. For the high ARP7022 binding sera (from donor HIV#3) the highest AqSo values were obtained using Immulon 4, at all three serum dilutions tested. Similarly. using Immulon 4 resulted in the best absorbance values for the low binding sera (HIV#4), but only at the high-

est serum concentration. At 1 in 1600 and 1 in 3200, highest absorbance values for the low binding sera were obtained using the Immulon 2 microplate. Max-

iSorp performed poorly in the detection of peptide specific antibody in the low binding sera. Based on these results Immulon 2 and Immulon 4 microtitre plates appeared to be suitable for use in the peptide- ELISA and Immulon 4 was used in subsequent experiments.

The effect of using a higher grade of casein in the blocking, diluent and wash solutions on the reduction of non-specific binding of HIV-I seropositive sera

(4 A J.T.Baker casein

0 BDH casein

I2 Sigma casein

0

(b)

0 0.35 0.5

% case/n

Fig. 2. Comparison of the specific to non-specific binding ratios

obtained in the ELISA to detect anti-ADP7022 antibodies in the

sera of HIV seropositive donors. In a, specific to non-specific

binding ratios are compared for wells blocked with 0, 0.25, 0.5,

1.0, 2.5 and 5.0% casein solutions, using caseins from three

different manufacturers (Sigma - regular grade, BDH and J.T.

Baker - both purified grades). In each case. the same casein as

used for blocking was also used for the wash/diluent solution, at

a concentration of 0.25%. The amount and type of casein in the

wash/diluent solution are compared for the purified grades, at

four concentrations (0, 0.1, 0.25 and 0.5%) in b. Wells were

blocked with a 1% solution of the same casein as was used in the

wash/diluent solutions.

was tested. In Fig. 2a, the ratios of specific to non-specific binding of HIV-l seropositive sera are compared for 18, 2.5% and 5% blocking solutions of Sigma, BDH and J.T. Baker caseins. In the ELISAs using a higher grade of casein (from BDH and J.T.

Baker), the ratios of specific to non-specific binding of HIV-l seropositive sera were higher than those where a technical grade of casein (Sigma) was used, the latter giving very unsatisfactory results. The low sensitivity obtained when Sigma casein was used to

84 L. Brown et al. /Journal of Immunological Methods 200 f 1997) 79-88

block wells was a result of high non-specific binding

of HIV-l seropositive sera (not shown). In addition,

higher non-specific binding of HIV seronegative sera was also seen when Sigma casein was used, compared with BDH and J.T. Baker caseins. The opti-

mum concentration of casein in the blocking solution was 1% for each type of casein tested.

The optimum concentration of casein in the dilu-

ent and wash solution was compared between BDH and J.T. Baker caseins at O.l%, 0.25% and 0.5%.

Ratios of specific to non-specific binding of HIV-l

seropositive sera are shown in Fig. 2b. From these results, the optimum concentration of casein in the

diluent and wash solution is O.l%, for BDH casein, and OS%, for J.T. Baker casein. Combining these

results with those obtained in the blocking solution experiments, denatured purified grades of casein im-

prove the sensitivity of this ELISA system. J.T. Baker casein was used in subsequent experiments, at 1% in the blocking and 0.5% in the diluent and wash

solutions.

3.2. Reduction of false-positive / negative results in the peptide-ELBA for the detection of antibody in

the sera of HIV-l seropositive donors.

Serum antibodies specific for the peptides ARP7022, CV-FAS and ARP740-27 were assayed for 14 HIV-l seropositive and 14 HIV seronegative donors. Levels of antibodies before and after correc-

tion for binding to the control peptide p12 are com-

W (cl loo00

loo0

,w

IO

1 1 0 I

10000

,000

,oo

IO

1

/

0 -r

pared between the two donor groups in Fig. 3a, 3c, 3e and Fig. 3b, 3d, 3f, respectively: levels of anti-

bodies specific for peptide ARP7022 are compared in Fig. 3a and b; levels of antibodies specific for

CV-FAS in Fig. 3c and d, and levels of antibodies specific for ARP740-27 in Fig. 3e and f. Antibodies specific for ARP7022 were found in all HIV-l seropositive donors and none of the HIV seronegative donors tested. In the HIV-l seropositive group,

the mean level of antibodies binding to ARP7022 decreased after correcting for non-specific binding to

p12 (from 3 136 to 2958 arbitrary protein concentra-

tion). Before correction for non-specific binding, three HIV-I seropositive and two HIV seronegative

donors’ sera bound to CV-FAS coated wells. How- ever, after subtracting absorbance values due to

non-specific binding, none of the HIV-l positive or HIV negative donors were found to have anti-CV- FAS antibodies. Thirteen out of fourteen sera in each

donor group bound to ARP740-27 coated wells with no significant difference between the means of the two groups (Student t test, p < 0.1) when the analy-

sis was done without subtracting absorbance values due to binding to ~12. None of the HIV seronegative

donors had anti-ARP740-27 antibodies and only six

HIV-l seropositive sera were found to contain anti-

ARP740-27 specific antibodies after correcting for non-specific binding.

Serum antibodies specific for the peptides ARP7022, CV-FAS and ARP740-27 were also assayed for five HIV-l positive donors who had re-

(d) ‘WOO

Fig. 3. Comparison of levels of anti-ARP7022 antibodies before (a) and after (b) subtracting absorbance values for binding to ~12 from

absorbance values for binding to ARP7022. Similarly, levels of anti-CV-FAS and anti-ARP740-27 antibodies before and after correcting for

non-specific binding to p12 are shown in (c), Cd) and (e), (f ), respectively. 14 HIV-seropositive and 14 HIV-seronegative donors’ sera

were tested in each ELISA. Results are shown as arbitrary protein concentrations.

L. Brown et al. /Journal of Immunological Methods 200 (1997) 79-88 85

(a) (W (d) (e)

Fig 4. Comparison of levels of anti-ARP7022 antibodies before (n) and after (b) subtracting absorbance values for binding to p12 from

absorbance values for binding to ARP7022. Similarly, levels of anti-CV-FAS and anti-ARP740-27 antibodies before and after correcting for

non-specific binding to p12 are shown in cc). Cd) and (e), (f), respectively. Sera from five HIV-I infected recent seroconverters (RS: see

Table 2). five HIV- I positive Ugandan (AS) and ten HIV-negative donors were tested in each ELISA. Results are shown as arbitrary protein

concentration.

cently seroconverted, five HIV- 1 seropositive donors living in Uganda and ten HIV-negative donors. Lev-

els of antibodies before and after correcting for non-specific binding to control peptide p 12 are com-

pared between these three donor groups in Fig. 4. High levels of non-specific binding of the African

sera (AS) against each of the peptides tested were found. After correcting for binding to ~12, levels of

anti-ARP7022 and anti-ARP740-27 were reduced, and binding to CV-FAS was shown to be non-

specific. Each of the African sera tested reacted against peptide ARP7022 and three out of the five contained ARP740-27 specific antibodies. In con- trast, the sera from recently seroconverted donors (RS) had lower levels of non-specific binding although the binding of sera from two donors to

ARP740-27 was shown to be non-specific after cor-

recting for binding to ~12. Non-specific binding to peptide ARP740-27 was also seen with seven out of

ten HIV-negative donors before correcting for binding to ~12. Two out of the five sera from recent seroconverters did not contain ARP7022 specific antibodies.

4. Discussion

From our own experience, the main difficulty in modifying a peptide-based ELISA method for the

detection of antibody in human HIV-l seropositive sera is the reduction of non-specific background binding. False positive results due to non-specific

background binding have been reported in relation to

other chronic diseases. This high non-specific background is particularly exacerbated when the sera is

from a donor with a chronic disease, for example SLE (Kumar et al., 1991) and HIV infection, where

there are high levels of antibody and also where immune complexes and rheumatoid factor may be

present (Jarvis et al., 1993; Medina-Rodriguez et al.,

19931. Kumar et al. (1991) used ‘no-antigen’ control wells to demonstrate the variable non-specific bind-

ing of sera from systemic lupus erythematosus (SLE) patients in the ELISA to detect anti-cardiolipin antibodies. They also noted that levels of non-specific

binding were higher in sera from the SLE donors

compared with healthy control donors and attributed

this change to higher concentrations of IgG in the SLE sera. High titres of HIV-l specific antibodies

may contribute towards elevated levels of non- specific binding of HIV-l seropositive sera, compared with sera from HIV seronegative controls.

The reduction of non-specific binding in ELISA procedures has been shown to rely upon optimising the blocking agent (Spinola and Cannon, 1985; Bjer- eke et al., 1986; Sarma et al., 1986; Wirtz et al., 1989). We chose casein as a blocking agent as it has been demonstrated to inhibit non-specific binding by

86 L. Brown et al. / Journal of immunological Methods 200 (1997) 79-8~5’

up to 94% at low concentrations (Vogt et al., 1987). We also heat denatured the blocking solution before use since denaturing the proteins in the blocking solution reduces non-specific binding of serum IgG to microtitre plates (Mauracher et al., 1991). We have made further reductions in background binding

of both HIV-l seropositive and seronegative sera by using a higher grade of casein. Even here, the techni-

cal grades of casein varied between companies. The

concentrations of denatured casein in the blocking and wash/diluent solutions were tested for reduction

of non-specific binding without loss of assay sensitivity. We would recommend using J.T. Baker ‘Baker’ grade casein at a concentration of l%, in the blocking solution, and 0.25%, in the wash/diluent solution.

After modifying the ELISA method to reduce the level of non-specific binding, the technique was further optimised to maximise assay sensitivity. The

dilution factor of HIV-l seropositive sera had a distinct effect upon the assay sensitivity. which was

greatest when serum was used at a dilution of 1 in 1600. However, a dilution of 1 in 1600 may be too

high for sera which contain low titres of anti-HIV-l gp 120 antibodies, for example from seroconverting

donors or HIV-l subtype 0 infected donors (Lous- sert-Ajaka et al., 1994). Two out of five sera from recent seroconverters assayed for anti-ARP7022 antibodies in this study did not contain specific antibod-

ies, indicating that the use of this peptide as a diagnostic peptide is limited, particularly when an early post-exposure confirmatory result is required.

Also, since these sera were used a dilution of 1 in 1600, non-detection of anti-ARP7022 antibodies may

reflect very low levels of specific antibody in the sera of recent seroconverters falling below the limit of detection of the assay. When analysing sera from

these donors a range of serum dilutions up to 1 in 1600 should be tested. The detection of antibodies in sera from HIV-l subtype 0 infected donors using peptide-based indirect ELISA have been previously described by Gurtler et al. (1995). The optimum dilution factor is reached when a balance is struck between diluting out high levels of non-specific anti-

bodies, for example rheumatoid factor (Jarvis et al., 1993; Medina-Rodriguez et al., 1993) whilst main- taining a sufficient concentration of specific antibody.

The effect of heat inactivation upon the reactivity of sera was investigated by Layton et al. (1991) in

the ELISA to detect anti-cardiolipin antibodies. These authors found that heating the sera to 56°C for 30 min, as is sometimes carried out on HIV-l seropositive sera prior to ELISA, significantly increases its

capacity to bind non-specifically. For this reason, we

omitted any pretreatment of HIV-l seropositive sera which involved heat inactivation.

We found that the best microtitre plate type for

our system was Immulon 4. This is consistent with the manufacturer’s claims which say that this mi-

croplate is designed specifically for the binding of short peptides. The other Dynatech microplate which we analysed, Immulon 2, is designed for the binding of longer peptides and short proteins. Wirtz et al. ( 1989) found Immulon 2 to be best in terms of high

positive to low background absorbance value ratios in an ELISA to detect Plasmodium falciparum cir-

cumsporozoite antibody, using a 128 amino acid residue as antigen. These authors also tested Falcon;

Dynatech flexible and Immulon 1; Costar (Cam- bridge, MA. USA) and Linbro (Flow Laboratories, VA, USA) microtitre plates in their ELISA system.

The optimum concentration of peptide may de- pend upon a number of variables, for example number of residues, molecular charge. structure and affinity for the surface of the microtitre plate. In our system, the optimum concentration of the positive control peptide was 16 pg/ml. If a positive control peptide is available, we would recommend that the

optimum concentration be ascertained experimen- tally, using a range of peptide concentrations from 1

to 32 pg/ml. If a positive control peptide is not available, we would choose 8 pg/ml peptide as a

compromise. An important consideration in the design of the

peptide-based indirect ELISA is the length of the peptide antigen. From our results, none of the sera from HIV- 1 seropositive or HIV seronegative donors bound in a specific manner to wells in which the short eight amino acid residue peptide CV-FAS was incubated. However, antibodies specific for the 19 amino acid residue peptide ARP740-27 which en- compasses the CV-FAS sequence were found in six out of fourteen HIV-l seropositive donors. Whilst the antibodies from these donors may not bind to the part of ARP740-27 that contains the CV-FAS se-

L. Brow~z et al. /Journal of lmmunologicul Methods 200 (19971 79-88 87

quence, these results suggest that peptide antigens of

eight amino acids or less are insufficient for binding human serum antibodies. In our system, peptide antigens of between nineteen (ARP740-27) and 24 (ARP7022) amino acids in length were able to bind

specific antibodies. The modifications to the peptide-based indirect

ELISA technique which we have described reduce

non-specific binding of human sera but results can still be inaccurate if a control peptide is not included in the assay. By correcting absorbance values for

non-specific binding of sera to the irrelevant peptide pl’ we have shown that the incidence of false

positive and false negative results can be greatly reduced. However. it is a matter of debate as to what constitutes a control peptide. Originally we chose

p12 as a negative control peptide. The control peptide is a randomly scrambled sequence from the carboxy-terminal region of gp120 which we have studied in the past (Clerici et al., 1993). The amino

acid sequence of pl2 was screened using a protein database to highlight any sequence homologies and found to have no similarities with known protein

sequences. To date, we have not found any specific

binding to p 12 by any human sera tested (n > 100, OD value < 0. I5 at I/ 1600 serum dilution). We feel that the length of the scrambled peptide is not a

critical parameter in the peptide-based indirect ELISA described here. However we recognise that ideally negative control peptides of the same length as the test peptide may further improve the sensitivity and specificity of the method. Such customisation of the control peptide may be particularly critical for diag-

nostic purposes although the peptide-ELISA described in this report should prove to be a reliable research tool.

Acknowledgements

We would like to thank Dr. H. Holmes (MRC AIDS Reagent Project, NIBSC, UK) for supplying the ADP7022 peptide; Dr. E.F. Hounsell (Depart- ment of Biochemistry, UCL, UK) for supplying ~12; Dr. B. Austen (Department of Surgery, St. George’s Hospital Medical School, UK) for synthesizing HSP65; Dr. D. Lewis (Department of Communicable Diseases. St. George’s Hospital Medical School, UK)

for advice about blocking solution and Dr. C. Duck-

ett (Chicago, IL, USA) for supplying the J.T. Baker casein. Louise Brown was funded by a Medical

Research Council studentship, reference number G78/ 1290; Dr. Mike Westby is supported by Glaxo Research and Development and Dr. Bernard Souber- bielle is supported by the Cancer Research Cam-

paign.

References

Bjercke, R.J.. Cook, G.. Rychlik. N.. Gjika, H.B., Van Vunakis.

H. and Langone, J.J. (19861 Stereospecific monoclonal anti-

bodies to nicotine and cotinine and their use in enzyme-linked

immunosorbent assays. J. Immunol. Methods 90, 203.

Clerici, M., Shearer, G.. Hounsell E.F.. Jameson, B.. Habeshaw. J.

and Dalgleish. A.G. (1993) Alloactivated cytotoxic T-cells

recognise the carboxy-terminal domain of human immunodefi-

ciency virus-l gp 120 envelope glycoprotein. Eur. J. Immunol.

23, 2022.

Dalgleish, A.G. (1995) Autoimmune mechanisms of depletion of

CD4 cells in HIV infection. Br. J. Haematol. 91, 525.

Gurtler. L.G.. Zekeng, L., Simon, F., Eberle. J., Tsague J.M..

Kaptue. L., Brust. S. and Knapp, S. (19951 Reactivity of five

anti-HIV-l subtype 0 specimens with six different anti-HIV

screening ELISAs and three different immunoblots. J. Virol.

Methods S 1, 177.

Jarvis, J.N.. Taylor. H.. Iobidze. M.. Dejonge, J.. Chang. S. and

Cohen. F.( 19931 Rheumatoid factor expression and comple-

ment activation in children congenitally infected with human

immunodeficiency virus. Clin. Immunol. Immunopathol. 67,

50.

Kumar, V.. Bartholomew. W.. Shieh, M.T.. Pierce, C. and Kaul.

N. (1991) Standardization of ELISA for the detection of

anti-cardiolipin antibodies - effect of non-specific IgG bind-

ing. Immunol. Invest. 20. 583.

Lange, J.M., Teeuwsen. V.J., Boucher, CA.. Vahlne. A.. Barin.

F., Tjong-A-Hung, S., Dekker, J.. Parkhede. U.. de Wolf. F.

and Goudsmit, J. (1993) Antigen variation of the dominant

gp41 in Africa. AIDS 7, 461.

Layton, G.T., Johnston. S.C. and Bertwistle. N.G. (19911 Serum

factors affecting the specificity of anticardiolipin antibodies.

Immunol. Cell Biol. 69. 17.

Loussert-Ajaka, I.. Ly. T.D.. Chaix, M.L., Ingrand. D.. Saragosti,

S., Courouce. A.M., Brun-Vezinet, F. and Simon, F. (1993)

HIV-l/HIV-2 seronegativity in HIV-l subtype 0 infected

patients. Lancet 343 (89101, 1393.

Mauracher. C.A.. Mitchell, L.A. and Tingle, A.J. (1991) Reduc-

tion of rubella ELISA background using heat denatured sam-

ple buffer. J. Immunol. Methods 145. 2.51.

Medina-Rodriguez, F.. Guzman, C.. Jara, L.J.. Hermida. C.. Al-

boukrek, D.. Cervera, H., Miranda, J.M. and Fraga. A. (19931

Rheumatic manifestations in human immunodeficiency virus

positive and negative individuals: a study of 2 populations

with similar risk factors. J. Rheumatol. 20. 1880.

88 L. Brown et al. /Journal of Immunological Methods 200 (1997) 79-88

Sarma, J.K., Hoffman, S.R. and Houghten, R.A. (1986) Enzyme

linked immunosorbent assay @LISA) for dorphin and its

antibodies. Life Sci. 38, 1723

Scheidtmann KH (1989) Immunological detection of proteins of

known sequence. In: T.E. Creighton (Ed.), Protein Structure,

A Practical Approach. IRL Press. Oxford University Press,

Oxford, pp. 93-l 15.

Vogt, R.F., Phillips, D.L., Henderson, 0.. Whitfield, W. and

Spierto, F.W. (1987) Quantitative differences among various

proteins as blocking agents for ELISA microtitre plates. J.

Immunol. Methods 101, 43.

Webster, A., Hay, R.T. and Kemp, G. (1993) The adenovirus

protease is activated by a virus-coded disulphide-linked pep-

tide. Cell 72. 97-104.

Spinola, S.M. and Cannon, J.G. (1985) Different blocking agents Wirtz, R.A., Duncan, J.F., Njelesani, E.K., Schneider, I., Brown,

cause variation in the immunologic detection of proteins trans- A.E., Oster, C.N., Were, J.B.O. and Webster, H.K. (1989)

ferred to nitrocellulose membranes. J. Immunol. Methods 81. ELISA method for detecting Plasmodiumfalciparum circum-

161. sporozoite antibody. Bull. WHO 67, 535.

optimisation of a peptide-based indirect elisa for the detection of antibody in the serum of hiv-1...

Documents