JOURNAL OF CLINICAL MICROBIOLOGY, Aug. 1980, p. 185-1920095-1137/80/08-0185/08$02.00/0

Vol. 12, No. 2

Enzyme-Linked Immunosorbent Assay for Titration ofHaemophilus influenza Capsular and O Antigen Antibodies

T. DAHLBERG* AND P. BRANEFORS

Department of Bacteriology, Institute ofMedical Microbiology, University of Goteborg, Gôteborg, Sweden

The enzyme-linked immunosorbent assay (ELISA) was elaborated for thedetection of immunoglobulin M (IgM) and IgG antibodies against capsular and Oantigens of Haemophilus influenza. Purified capsular polysaccharide and lipo-polysaccharide were used as antigens, with optimal coating concentrations beingabout 50 and 100 jg/ml, respectively. The antibody content was expressed as thehighest serum dilution (-log1o) showing an absorbance of 0.2 above the back-ground level. The titers of hyperimmune sera (reference sera) ranged between 5and 7 -log1o. The sensitivity of the method was about 80 ng/ml with regard toanticapsular antibodies and 3 to 5 ng/ml with regard to anti-lipopolysaccharideantibodies. For detection of antibodies against capsular polysaccharide in seraobtained after primary immunization, ELISA was about 100-fold more sensitivethan the indirect hemagglutination assay, whereas in hyperimmune sera, ELISAwas about 10-fold more sensitive than the indirect hemagglutination assay. Thesensitivity of ELISA for detecting anticapsular antibodies after primary andbooster immunizations was 50-fold higher than that of the bactericidal assayusing capsulated bacteria, whereas the sensitivity of the two methods was thesame when hyperimmune sera were tested. ELISA performed with lipopolysac-charide as the antigen was about 50- and 150-fold more sensitive than thecomplement fixation and bactericidal assays tested with noncapsulated variantsafter primary injection and hyperimmunization, respectively.

The antibody response to the capsular poly-saccharide (CPS) of Haemophilus influenzatype b has been estimated by indirect hemagglu-tination assay (IHA) (1, 4) and, more recently,by radioimmunoassay (1, 11). Both methodshave certain disadvantages. IHA is compara-tively insensitive, whereas radioimmunoassay,being sensitive, requires certain precautions inhandling radioactive reagents as well as special-ized equipment for measurement. In addition tothe aformentioned methods, the complement-dependent bactericidal assay (BC) has been usedas a test mainly for measuring the amount ofantibodies against the CPS antigen (1, 5, 6a),although the bactericidal effect might, to a cer-tain extent, be due to antibodies against otherantigens as well (1, 6a). For the determinationof antibodies against the cell wall antigens ofnoncapsulated H. influenza, the BC assay (5,6a) and a complement fixation (CF) test havebeen available <4).

In contrast to the above-mentioned methods,which are dependent on secondary manifesta-tions for visualization of the antigen-antibodyreaction, the primary binding between antigenand antibody may be revealed by the enzyme-linked immunosorbent assay (ELISA) (7, 15).

By the use of immunoglobulin class-specific con-jugates, this technique also offers a simplemethod for measuring and differentiating be-tween immunoglobulin M (IgM), IgG, IgA, orsecretory IgA class antibodies without laboriousseparation of serum or secretions. The micro-ELISA variant (12) is useful when smallamounts of serum and secretion are available.The aim of the present investigation was to

elaborate suitable experimental conditions forthe application of ELISA to the determinationof antibodies against H. influenza CPS andlipopolysaccharide (LPS) antigens. The sensitiv-ity of ELISA for the detection of H. influenzaantibodies after primary and booster immuni-zations and hyperimmunization was comparedwith that of other methods such as IHA, CF,and BC.

MATERIALS AND METHODSStrains. H. influenza type a, strain Smith (Ma),

H. influenza type b, strain RAB (Mb), and theirrespective noncapsulated variants (Sa and Sb) werethe same as those used in previous studies (3-5, 6a).

Antisera. Antisera against Ma, Mb, Sa, and Sbbacteria, obtained as described previously (5, 6a), wereused in the study. The immunization schedule and thebleedings are shown in Fig. 1 and 2.

185

on July 13, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

186 DAHLBERG AND BRANEFORS

TITRE- 10

D.-8

o.-

IgM,.V

; t 4 X g~~~~~M 0-0

B-IHA

O 4 8 12 16 weeks£ A , , à injections

y' bleedings

FIG. 1. Antibody response of one rabbit duringimmunization course with Mb bacteria as tested withELISA, IHA, and BC. The immunization procedurewas as follows. The rabbit was given one primaryimmunization and, 4 weeks later, a booster dose withabout 1 x 108 bacteria. After another 3 weeks, ahyperimmunization course was started, and eightinjections were given. The bacterial amount at eachinjection was increased in twofold steps from I x 10'to 6 x 109 bacteria.

TITRE

- log10

ELISA IgG--g M o--O

BCF -

C F

O 4 8 12 16 weeksà à à à à à à à à à injectionsyy y y y v y y' y*y, ,y*bleedings

FIG. 2. Antibody response of one rabbit duringimmunization course with Sb bacteria as tested withELISA, CF, and BC. Immunization procedure is as

described in the legend to Fig. 1.

Additionally, a series of hyperimmune sera againstthe following H. influenza strains was employed tostudy the specificity of ELISA: H. influenzae types c

(NCTC 8469), d (Garf), e (NCTC 8472), and f (NCTC8473). Hyperimmune sera against Streptococcuspneu-moniae (types 5, 6, 21, and 27), Klebsiellapneumoniae(types O?:K1 and O?:K9), Escherichia coli (types 016:KI, 04:K12, and 075:K100), Proteus vulgaris (types03H1, 03H2, 01OH3, and 013H4), and Vibrio chol-

erae were also utilized. The antisera against E. coli075:K100 and the pneumococcal strains were obtainedfrom Statens Seruminstitut, Copenhagen, Denmark;antisera against the remaining strains were obtainedfrom the Institute for Medical Microbiology, Univer-sity of Goteborg, Goteborg, Sweden.

Antigenic preparations for ELISA. PurifiedCPS was prepared from the Ma and Mb supernatantsof shaken 10-h-old fluid cultures in antigen-free me-dium (2). These preparations were designated CPSMa and CPS Mb, respectively. A slight modificationof the medium was utilized; the medium materials,casein hydrolysate and yeast autolysate, were dialyzedagainst distilled water at 4°C for 1 week (dialysistubes; pore size, 240 nm; Kebo-Grave, Spànga, Swe-den), instead of being filtered through Sephadex G-25,for obtaining low molecular medium. Salts and growthfactors were added before sterilization as previouslydescribed (2). During the earlier part of the study,CPS (designated CPS I) was prepared from 10-fold-concentrated culture supernatants wilAch were filteredthrough Sephadex G-150 (3). Fractions tllat elutedwith the void volume, containing the capsular antigentogether with some contaminants (LPS and probablyadhering medium proteins), were pooled and furtherpurified by diethylaminoethyl-Sepharose chromatog-raphy using an NaCl gradient (O to 2 M). The Oantigen eluted before the main part of the capsularantigen as tested by immunodiffusion analyses (16).Capsular antigen-containing fractions, with no absor-bance or only absorbance below 0.03 at 260 and 280nm, were pooled and lyophilized, constituting the CPSI antigen. Its protein content was about 1% as esti-mated by the method of Lowry et al. (10).

During the latter part of the study, capsular antigen(CPS Mb II) was obtained by precipitation of dialyzedculture supernatant with 0.5% cetyltrimethylammo-nium bromide (final concentration) at 40C for 2 weeks.The capsular material was dissolved in 0.4 M NaCland further purified by three repeated precipitationsin 3 volumes of ethanol with solution in distilled water.Centrifugation and discarding of water-insoluble ma-terial were performed between each precipitation. Thematerial was finally purified by filtration on a Sepha-rose-2B column, and fractions containing the capsularantigen, as determined by immunodiffusion analysis,were collected as shown in Fig. 3. The pooled capsularmaterial was dialyzed and lyophilized. Its protein con-tent was estimated at 0.2% by the method of Lowry etal. The preparation did not contain detectableamounts of O antigen, as estimated by immunodiffu-sion analysis with 1 mg of CPS Mb II per ml. Theimmunodiffusion method indicated about 15 gg ofLPSper ml.LPS was prepared from two noncapsulated variants,

Sa and Sb, of the capsulated strains (5, 6a) by hotphenol-water extraction (17). The crude LPS extractswere treated with ribonuclease (type 1-A; SigmaChemical Co., St. Louis, Mo.) at a concentration of 0.1mg of enzyme per mg of LPS (pH 7.8) for 7 h and thencentrifuged at 105,000 x g for 4 h. These preparationswere designated LPS Sa and LPS Sb, respectively.One part of LPS was lyophilized to estimate the con-centration of antigen. The part of LPS used for ELISA

6

5.

4-

3-

2

o

6

5

4

3

2

o

J. CLIN. MICROBIOL.

on July 13, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

VOL. 12, 1980

o

OD(nm>

0I3 -

0.2-

ELISA FOR H. INFLUENZAE ANTIBODIES 187

8

f

FIG. 3. Sepharose-2B filtration of 30methylammonium bromide-precipitatecmaterial (column size, 2.6 by 100 cm; phM]-buffered saline [0.5 MI, pH 7.1,,elution; flow rate, 20 ml/h). Fractionstaining all detectable O antigen and s

material were discarded. Fractions 38by dotted line under figure) were collectas CPS Mb II antigen. It contained alpurified CPS Mb II per ml. A 30-ml arMb II material corresponded to abouculture fluid.

was not lyophilized because of incomplelH. influenza LPS after lyophilizationcontent was 2% as tested by the methoal.

Serological methods. (i) ELISA. Elformed essentially as described by Engmann (8).

(a) Antigen-coated tubes. Disposabtubes (Heger Plastic AB, Stallarholmen,coated with 0.5 or 1 ml of antigen solutiîtrations ranging from 1 to 1,000 jig/ml.antigen concentrations were determineof IgM as well as IgG antibodies. Thcoated overnight at 37°C and then keptsealed tubes for from a few days up tcevaluate the efficiency and stability of t

(b) Enzyme-conjugated anti-immSheep or swine antisera specific to rabbiimmunoglobulin (Nordic Immunologiries, Tilburg, The Netherlands) were co

alkaline phosphatase (calf intestinal mu(Sigma Chemical Co.) with the aid of g

as described by Holmgren and Svennersuitable concentration of the conjugatezyme-substrate reaction time were dethyperimmune reference sera. The conjulused generally in dilutions of about 1:30and 1:400 for anti-IgG, with the enz:

reaction time being about 50 min.(c) Standardization of ELISA. Ra

mune sera were used as references in a]for the different serological systems. Th.strate reactions were allowed to contitubes containing the reference serum

reached appropriate absorbance values,imately the same endpoint titers (see Tiseparate experiment. An antigen-coatbated with the enzyme substrate (blanh

tube incubated with conjugates and phosphate-260 nm buffered saline instead of serum (background absor-280 nm - bance) were included as controls in each set of exper-

iments.(d) Micro-ELISA. Micro-ELISA was performed on

microplates (Dynatech, Micro-ELISA; Nova KemiAB, Enskede, Sweden) using a total reactant volumeof 0.2 ml. Otherwise, experimental conditions were asdescribed above. Absorbances were registered by an

r-r----.- automatic spectrophotometer (Titertek Multiskan;o 90 100 Flow Laboratories, Solna, Sweden). The referenceraction number anti-Ma, anti-Mb, anti-Sa, and anti-Sb sera were5ml each) tested simultaneously by micro- and macro-ELISA for

comparison and standardization of experimental con-mi olceyMbtrII ditions so as to obtain comparable sensitivity.ioCPS'Mb05 (i) Quantitative precipitin determination.osp ated0.0 Quantitative precipitin determination (9) was per-was used for formed to estimate the amount of specific antibodies25 to 37 con- against CPS Ma, CPS Mb, LPS Sa, and LPS Sb inomre capsular homologous hyperimmune antisera. A 0.2- or 0.25-mlto 75 (marked amount of serum was mixed with equal volumes ofacted and used antigen dilutions with concentrations ranging from°out 160 ,gpof 0.002 to 1 mg incubated for 1 h at 37°C and was thent 1.7 IPtS kept about 1 week at 4°C. The maximum amount oft *. 7 lters of precipitated antibodies was determined by the method

ofLowry et al.te solubility of (iii) Absorption of serum. The anti-Mb serummThe protein diluted 1:1,000 was mixed with various dilutions ofd ofheLowrtent CPS Mb I or CPS Mb II antigens ranging in concen-

trations from 0.2 to 200 gg. After incubation for 1 h at

LISA was per 37°C, the tubes were stored at 2°C for about 1 week.vail and Perl- The supernatants were then tested for remaining anti-,vall and Perl- bodies by micro-ELISA using the IgG- conjugate. Theole polystyrene antigen concentration giving 50% reduction of theSweden) were extinction at 400 nm was used as a measure of theons in concen- inhibiting capacity.The optimal (iv) Other serological methods. BC, used for

d for titration determination of bactericidal antibodies against cap-ie tubes were sulated and noncapsulated bacteria, IHA, used fortat 4°C hi the measuring antibodies against capsular antigen, and3 months to CF, used for measuring antibodies against cell wall

;he coating. antigens, were performed as previously described (4,unoglobulin. 5)t IgG and IgM Statistical methods. Calculation of mean andical Laborato- standard deviation of titer values was performed ac-njugated with cording to standard methods.cosa, type VII;,lutaraldehyde RESULTSholm (8). The A series of experiments was performed tois and the en- study various factors having a possible influencegates cwuldbe on the results obtained by ELISA as used forO foruanti-IgM detection of antibodies against H. influenzayme-substrate CPS and LPS.

Antigen concentration and coating time.Lbbit hyperim- For CPS Ma I and CPS Mb 1, increasing absor-Il experiments bance was registered when increasing concentra-e enzyme-sub- tions of antigen were used for the coating of thenue until the tubes, with no leveling of absorbance beinggiving approx reached up to an antigen concentration of 1 mg/able 1) at each ml (Fig. 4). To obtain the highest titers of theed tube incu- homologous reference sera, coating with CPSo) and another Ma I preparations should be performed at 4°C

on July 13, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

188 DAHLBERG AND BRANEFORS

E400nm

O.---....o

f.. /~~~o

CPS Ma

CPS Mb I c

CPS Maill

CPS Mb Ilc.. w

E 400nm1,2-

1 -

0,8-

0.6-

0.4-

0.2 -

O -

*-e

o-o

......

o-....o

10 100 1000

ANTIGEN ug/ml

/"

e~~~~~~~~~~

1 110 100 1000

ANTIGEN ug/ml

FIG. 5. Determination of optimal amount of LPSSb antigen for coating of tubes. The anti-Sb serum

was diluted 1:1,000. The IgG conjugate was used inthe test.

FIG. 4. Determination of optimal amounts of CPSMa I, CPSMb I, CPSMa IH, and CPSMb II antigensfor coating of tubes. The anti-Ma and anti-Mb sera

were diluted 1:1,000. A second anti-Mb serum was

tested with CPS Mb II and was diluted 1:100. TheIgG conjugate was used in the test.

for 1 to 2 days, whereas coating with CPS Mb Ishould be performed for about 2 to 3 weeks.When the CPS I antigen was used for coating ata concentration of 100 to 200,ug/ml, the IgG andIgM titers of the reference sera were about thesame as those observed when higher concentra-tions of antigen were used for coating. In theearlier part of the study, 100 to 200 ,tg of CPS Iper ml was found to be suitable for determina-tion of both IgG and IgM antibodies.The CPS II antigen preparations adsorbed to

the tubes in lower concentration than did CPSI. Optimal coating was obtained by using about50,ug/ml as determined by IgG conjugate (Fig.4). In addition, the CPS II preparations adsorbedmore effectively (overnight incubation at 37°Cand 1 to 2 days at 4°C) than did the CPS I. Tosave material, a suboptimal concentration of 30,ug/ml was utilized for further assays.The inhibition study with CPS Mb I and CPS

Mb II antigen preparations and an anti-Mb se-

rum dilution of 1:1,000 revealed that the inhibi-tory concentrations causing 50% reduction ofextinction at 400 nm were about 250 ,g/ml forthe CPS Mb I and about 30 ,g/mI for the CPSMb II antigen preparations.Optimal coating of tubes with LPS was ob-

tained by using a solution containing 100 ,ug/mlas tested with 10-fold dilutions of LPS antigen(Fig. 5).

E 400 nm

1.6

1.44

1 2 3 4 5 1 2 3 4 5 6

SERUM DILUTION -bogl

FIG. 6. ELISA titration curves for IgG and IgMclass antibodies as obtained with rabbit antiseratested against CPS Mb II antigen. (a) After primaryinjection. IgG titer, <1 -loglo; IgM titer, 2.5 -log1o.

(b) After hyperimmunization (reference serum). IgGtiter, 4.5 -logio; IgM titer, 3.8 -log1o.

Principles for quantitation of antibodies.The steepness of the linear portions of the titra-tion curves increased during the course of theimmunization. This increase was pronounced forIgG antibodies. The IgG curves were steeperthan those obtained with anti-IgM conjugates.Figures 6a and 6b give representative examplesof the titration curves of sera obtained afterprimary injection and hyperimmunization usingCPS Mb II antigen. For ELISA performed withH. influenza CPS and LPS antigens, the anti-body content was expressed as the -loglo of theserum dilution titerr) which showed an absor-

1,6

1,4-

1,2-

1 -

0,8-

0,6-

0.4-

0,2-

J. CLIN. MICROBIOL.

1

on July 13, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

ELISA FOR H. INFLUENZAE ANTIBODIES 189

bance of 0.2 above the background level. Thebackground absorbance for the CPS as well asthe LPS antigens was generally low, about 0.01to 0.07.Precision of ELISA. The rabbit reference

serum sample was titrated 10 times on separatedays with two lots of conjugates and two CPSMb II antigen preparations. The standard devia-tion calculated from these experiments was 0.21-log1o.

Sensitivity and specificity of ELISA. Toestimate the sensitivity of ELISA, quantitationof the specific antibody content in rabbit hyper-immune sera was performed in parallel by meansof the quantitative precipitation test. The lowestamount of antibodies detectable with ELISA forH. influenza CPS and LPS was calculated fromthe maximum amount of precipitated antibodiesand the corresponding ELISA titer by usingIgG conjugate. The lowest amount of antibodiesdetectable with ELISA was about 80 ng/ml foranti-CPS Ma and Mb and about 3 to 5 ng/mlfor anti-LPS Sa and Sb (Table 1).To study the specificity of ELISA concerning

anti-capsular b antibodies, in addition to thehomologous antisera, a number of antiseraagainst various gram-negative and gram-positivebacteria, including five remaining types of H.influenzae, were tested for presence of IgG anti-bodies against CPS Mb II antigen by micro-ELISA (Table 2). Demonstrable titers were ob-tained in sera against H. influenza type a(about 2.4 -logwo), S. pneumoniae type 6 (about2.2 -logio), and E. coli 075:K100 (about 3.4-logio).As regards antibodies against LPS Sb antigen,

the aforementioned sera were employed in thestudy. No demonstrable titers were noted in anyof them.

TABLE 1. Comparative IgG antibody classdetermination of anti-Ma, anti-Mb, anti-Sa, and

anti-Sb hyperimmune sera tested againsthomologous CPS (I and II) and LPS antigens,

respectivelySpecific Lws eprecipi- ELISA end- Lowest de-

Antisera tated anti- point titer amt of anti-bodies (IgG) bodies (ng)(mng/mld)

Anti-Ma I 5.2-5.4 60,000 86-90"Anti-Mb I 3.1 30,000 103"Anti-Mb II 4 50,000 80"Anti-Sa I 4.4 1,000,000 4.4Anti-Sb I 3.6 1,000,000 3.6a CPS I used in ELISA and quantitative precipita-

tion.b CPS II used in ELISA and quantitative precipi-

tation.

TABLE 2. ELISA titers (-logîo) against CPS Mb IIantigen in antisera against various bacteria

Antisera against: Strain type Titer(-logo)

H. influenza b 4.5a 2.4c <1d <1e <1f <1

S. pneumoniae 5 <16 2.221 <127 <1

E. coli 016:K1 <104:K12 <1075:K100 3.4

K. pneumoniae O?:K1 <1O?:K9 <1

P. vulgaris 03H1 <103H2 <1010H3 <1013H4 <1

V. cholerae <1

Comparison of macro- and micro-ELISA.Parallel experiments were performed with ref-erence sera by using optimal concentrations ofthe corresponding antigens and the same con-jugate concentration and enzyme-substrate re-action time. The greatest differences recordedbetween results obtained by macro- and micro-ELISA techniques were limited to 0.2 -log1owhen CPS and LPS antigens were used in theassays. There was no tendency for any techniqueto give generally higher values.Antibody response in rabbits against

capsulated bacteria studied by means ofELISA, IHA, and BC assays. The four rabbitsimmunized with capsulated bacteria had no de-monstrable antibodies against CPS Ma and CPSMb in their preimmune serum samples as re-vealed by ELISA. The primary immunizationwith Ma or Mb bacteria caused an IgM antibodyresponse within 2 to 4 days, whereas an IgGantibody response was not demonstrable until 7to 14 days. The IgM titers rapidly decreased tolow levels before the booster dose, in contrast tothe IgG titers, which increased continuously.The booster dose caused an IgM antibody

response of the same magnitude as that obtainedafter the primary immunization, whereas theIgG titer levels increased only slightly. Duringthe hyperimmunization course, the IgM and IgGantibody levels increased gradually to titer val-ues of about 4 and 5 -logio, respectively. Figure1 gives the development of ELISA IgM and IgGantibody titers in a rabbit given Mb bacteria,the titer development being representative for

VOL. 12, 1980

on July 13, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

190 DAHLBERG AND BRANEFORS

the other rabbits as well.The antisera were also tested against LPS Sb

antigen. The development of IgM and IgG anti-bodies was similar to that described below forthe antisera against noncapsulated bacteria.However, the titers were about 10- to 100-foldlower in the hyperimmune sera against capsu-lated bacteria than in sera against the corre-sponding noncapsulated variant.

In addition, the IHA and BC titers are givenfor comparison of sensitivity. For the four sera,the IHA titers were 50- to 100-fold lower afterthe primary immunization than the correspond-ing ELISA titers. During the hyperimmuniza-tion course, the IHA titers increased to aboutthe same levels as ELISA IgM titers, being about70-fold lower than the ELISA IgG titers (Fig. 1).The BC titers were 10- to 50-fold lower than thecorresponding ELISA IgM titers after the pri-mary immunization. At the end of the hyper-immunization course, very high BC titers were

obtained, showing even higher levels than theELISA IgG in the corresponding serum samples(Fig. 1).Antibody response in rabbits against

noncapsulated bacteria studied by meansof ELISA, CF, and BC assays. The two rab-bits immunized with Sa bacteria had lowamounts of IgM antibodies, as demonstrated byELISA (about 1.2 -log1o), against LPS Sa beforeimmunization. In sera obtained after primaryimmunization, IgM and IgG antibody responseagainst Sa bacteria was obtained after 2 days(titer about 3 -logio for both classes of anti-bodies). The preimmune sera of two rabbitsimmunized with Sb bacteria did not reveal anyIgM antibodies against LPS Sb. The primaryimmunization caused an IgM antibody responseafter 2 days, whereas IgG antibodies were notrevealed until after about 3 weeks (Fig. 2). TheIgM titers against LPS did not increase anyfurther after the booster dose and during thehyperimmunization course, whereas the IgGantibodies increased rapidly to values between6 and 7 -logio (Fig. 2). In addition, the CF andBC titers are given for comparison. The sensitiv-ity of the CF and BC methods was about 50- and150-fold lower than that of ELISA in serum

samples taken after primary immunization andhyperimmunization, respectively.

DISCUSSIONThe present report evaluates the suitability of

ELISA for quantitation of antibodies against theCPS and LPS antigens ofH. influenza in rabbitantisera.The investigation showed that the capsular

substance of H. influenza types a and b (CPS

Ma and CPS Mb) adsorbed to the plastic tubesin sufficient amounts to allow determination ofantibodies. Two preparations of purified H. in-fluenzae capsular antigen were utilized, one ob-tained by Sephadex G-150 filtration and di-ethylaminoethyl Sepharose chromatography(method I) and the other obtained by cetyltri-methylammonium bromide precipitation andSepharose-2B filtration (method II). They werefound to differ in binding capacity to the plasticsurface, with CPS II being attached more easily;the optimal concentration was markedly lower,and for CPS Mb II, the necessary attachmenttime was considerably shorter than for CPS MbI. Different lots of CPS II preparations showedthe same high attachment ability. It was furtherobserved in inhibition studies that the immu-nological reactivity of CPS Mb II antigen wasabout 10-fold higher than that of CPS Mb I. Inaddition, its purity (e.g., low protein content) isa further advantage of method II for preparationof H. influenza capsular antigen for ELISA.

Results obtained by ELISA may be expressedin a variety of ways (15). In this study, theantibody content was expressed as -logio ofserum dilution representing an endpoint titer.Repeated titration of reference sera showed thatthere was good reproducibility of results ex-pressed in this way.For standardization of the assay, the use of

reference serum, with known amounts of specificantibodies in all sets of experiments, was neces-sary to compensate for variation in test param-eters such as antigen coating on different occa-sions, different lots of conjugates, etc. The stand-ard deviation of ELISA described in this studywas as low as 0.21 -logio.ELISA was found to be a sensitive method for

the demonstration of antibodies to H. influenzaantigens. By means of this technique, it waspossible to measure about 80 ng of IgG anti-bodies against CPS. However, since IgM anti-bodies present in the hyperimmune sera mayalso have contributed to the amount of precipi-tate obtained at the quantitative precipitationanalysis, the sensitivity ofELISA may have beensomewhat underestimated. Despite this possi-biity, the sensitivity was of the same magnitudeas that of radioimmunoassays as reported byNorden and Michaels (11). However, other stud-ies have reported higher sensitivity of radioim-munoassay techniques (1). The sensitivity ofELISA with regard to determination of anti-bodies against LPS was at least 10-fold higherthan that of antibodies against CPS. It may benotable that the sensitivity of the assay used inthe present study for demonstration of anti-bodies against H. influenza LPS was about

J. CLIN. MICROBIOL.

on July 13, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

ELISA FOR H. INFLUENZAE ANTIBODIES 191

fourfold higher than that registered in an assayfor determination of antibodies against Salmo-nella O antigen (6).To obtain high specificity ofELISA, a primary

requisite is to use highly purified antigen prep-arations at the coating of the tubes. The pres-ence of a trace amount of O antigen in the CPSpreparation used in ELISA (less than 0.5 ig/ml)should not have influenced the absorbance ob-tained in the tests performed with this antigenbecause an absorbance as low as 0.2 was ob-served with 1 ,g of CPS per ml.ELISA performed with hyperimmune sera

against other H. influenzae strains and otherspecies of bacteria revealed antibodies againstH. influenzae type a, S. pneumoniae type 6, andE. coli K100 cross-reacting with the CPS Mb IIantigen. The capsular antigens of these bacteriaare known to have some chemical similarities tothe capsular antigen of H. influenzae type b (13,14).Using the same experimental parameters in

micro-ELISA as in macro-ELISA, no differencesin titer values exceeding 0.2 -logio were re-corded. On this basis, the same sensitivity canbe assumed for the micro method and macromethod.Although the number of animals included in

the study was rather small, some general conclu-sions concerning the sensitivity of ELISA ascompared with that of other serological methodsmight be drawn.For estimation of the primary response (IgM

class antibodies), ELISA was found to be moresensitive (10- to 100-fold) than the other meth-ods, especially when compared to IHA, whichmeasures capsular antibodies, too.

After the booster dose, in addition to IgMclass antibodies, ELISA IgG titers were revealedwhich were higher than those obtained by othermethods.

In the hyperimmune sera to capsulated bac-teria, the differences between ELISA titers andthe titers of the other methods were less pro-nounced. In the antisera to noncapsulated bac-teria, the higher sensitivity of ELISA was moreevident. The discrepancy in sensitivity betweenBC and ELISA may partly be due to the veryhigh sensitivity of ELISA for the detection ofanti-LPS antibodies. Moreover, antibodiesagainst LPS and other cell wall antigens of non-capsulated bacteria are possibly not as effectivein the BC assay as are anticapsular antibodies(6a).ELISA was found to be a suitable method for

detection of anticapsular antibodies against H.influenza type b in humans, too (manuscript inpreparation). Especially, its high sensitivity in

estimation of primary response antibodies, asobserved for the rabbit sera, may be valuable fordetection of antibodies during the acute stagesof infectious diseases. The micro variant is easierto perform and is more economical with regardto consumption of antigen, conjugate, etc., andthis modification can, therefore, be recom-mended for routine work.

ACKNOWLEDGMENTSThis study was supported by grants from the Swedish

Medical Research Council (16X-04978), the Faculty of Medi-cine, University of Goteborg, and the Gôteborg Medical So-ciety.

LITERATURE CITED1. Anderson, P., R. B. Johnston, Jr., and D. H. Smith.

1973. Methodology in detection of human serum anti-bodies to Haemophilus influenza, type b, p. 87-94. InS. Sell and D. Karzon (ed.), Haemophilus influenza.Vanderbilt University Press, Nashville, Tenn.

2. Branefors-Helander, P. 1972. Antigen-free medium forcultivation of Haemophilus influenza, AFH-medium.Acta Pathol. Microbiol. Scand. Sect. B 80:211-220.

3. Branefors-Helander, P. 1972. Serological studies ofHaemophilus influenza. I. Extracellular antigens ofcapsulated strains, types a to f, and two non-capsulatedvariants. Int. Arch. Allergy Appl. Immunol. 43:533-547.

4. Branefors-Helander, P. 1973. Serological studies onHaemophilus influenza. IV. The antibody response inrabbits against capsular and O antigens of H. influ-enzae. Int. Arch. Allergy Appl. Immunol. 45:657-674.

5. Branefors, P., and T. Dahlberg. 1980. Serum bacteri-cidal effect on capsulated and non-capsulated Haemo-philus influenza. Acta Pathol. Microbiol. Scand. Sect.C 88:47-56.

6. Carlsson, H. E., A. A. Lindberg, and S. Hammar-strom. 1972. Titration of antibodies to Salmonella Oantigen by enzyme-linked immunosorbent assay. Infect.Immun. 6:703-708.

6a.Dahlberg, T., and P. Branefors. 1980. Characterizationof the bactericidal antibody response against Hae-mophilus influenza. Acta Pathol. Microbiol. Scand.Sect. C 88:115-120.

7. Engvall, E., and P. Perlmann. 1972. Enzyme-linkedimmunosorbent assay. III. Quantitation of specific anti-bodies by enzyme-labeled anti-immunoglobulin in an-tigen-coated tubes. J. Immunol. 109:129-135.

8. Holmgren, J., and A. M. Svennerholm. 1973. Enzyme-linked immunosorbent assay for cholera serology. In-fect. Immun. 7:759-763.

9. Kabat, E. A. 1971. Experimental immunochemistry, 2nded. Charles C Thomas, Publisher, Springfield, III.

10. Lowry, O. H., N. J. Rosebrough, A. L. Farr, and R. J.Randall. 1951. Protein measurement with the Folinphenol reagent. J. Biol. Chem. 193:265-275.

11. Norden, C. W., and R. Michaels. 1973. Immunologicresponse in patients with epiglottitis caused by Hae-mophilus influenza type b. J. Infect. Dis. 128:777-780.

12. Ruitenberg, E. J., and B. J. M. Brossi. 1978. Automa-tion in enzyme immunoassay. Scand. J. Immunol.8(Suppl. 7):63-72.

13. Schneerson, R., M. Bradshaw, J. K. Whisnant, R. L.Meyerowitz, J. C. Park, and J. B. Robbins. 1972.An Escherichia coli antigen cross-reactive with the cap-sular polysaccharide of Haemophilus influenza typeb: occurrence among known serotypes, and immuno-chemical and biological properties of E. coli antiseratoward H. influenza type b. J. Immunol. 108:1551-1562.

VOL. 12, 1980

on July 13, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

192 DAHLBERG AND BRANEFORS

14. Tunevall, G. 1952. Studies on Haemophilus influenzatype characteristics. Acta Pathol. Microbiol. Scand. 30:203-212.

15. Voller, A., D. E. Bidwell, and A. Barlett. 1976. Enzymeimmunoassays in diagnostic medicine. Bull. W. H. O.53:55-65.

16. Wadsworth, C. 1962. A microplate technique employing

J. CLIN. MICROBIOL.

a gel chamber compared with other micro- and macro-plate techniques for immune diffusion. Int. Arch. Ai-lergy Appl. Immunol. 21:131-137.

17. Westphal, O., O. Luderitz, and F. Bister. 1952. Uberdie Extraction von Bakterien mit Phenol/Wasser. Z.Naturforsch. Teil B 7:148-155.

on July 13, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from