Journal of Immunological Methods 364 (2011) 83–93

Contents lists available at ScienceDirect

Journal of Immunological Methods

j ourna l homepage: www.e lsev ie r.com/ locate / j im

Research paper

Development of a multiplex microsphere immunoassay for the quantitationof salivary antibody responses to selected waterborne pathogens

Shannon M. Griffin a,b, Ing M. Chen c, G. Shay Fout b, Timothy J. Wade d, Andrey I. Egorov d,⁎a Oak Ridge Institute for Science and Education, Oak Ridge, TN, USAb National Exposure Research Laboratory, USEPA, Cincinnati, OH, USAc Self-employed student contractor with USEPA, Cincinnati, OH, USAd National Health and Environmental Effects Research Laboratory, USEPA, 109 T.W. Alexander Drive, MD 58C, Research Triangle Park, NC 27711, USA

a r t i c l e i n f o

Abbreviations: USEPA, US Environmental Protcontaminant candidate list; SAPE, streptavidin R-phyMFI, median fluorescence intensity; OD, optical densittransferase; VLP, virus-like particle; CI, confidence intof variation.⁎ Corresponding author. Tel.: +49 228 815 0416.

E-mail address: andegorov@gmail.com (A.I. Egorov

0022-1759/$ – see front matter. Published by Elseviedoi:10.1016/j.jim.2010.11.005

a b s t r a c t

Article history:Received 5 August 2010Received in revised form 1 November 2010Accepted 10 November 2010Available online 18 November 2010

Saliva has an important advantage over serum as a medium for antibody detection due to non-invasive sampling, which is critical for community-based epidemiological surveys. Thedevelopment of a Luminex multiplex immunoassay for measurement of salivary IgG and IgAresponses to potentially waterborne pathogens, Helicobacter pylori, Toxoplasma gondii,Cryptosporidium, and four noroviruses, involved selection of antigens and optimization ofantigen coupling to Luminex microspheres. Coupling confirmation was conducted usingantigen specific antibody or control sera at serial dilutions. Dose–response curvescorresponding to different coupling conditions were compared using statistical tests. Controlproteins in the specific antibody assay and a separate duplex assay for total immunoglobulins Gand A were employed to assess antibody cross-reactivity and variability in saliva composition.200 saliva samples prospectively collected from 20 adult volunteers and 10 paired sera from asubset of these volunteers were used to test this method. For chronic infections, H. pylori andT. gondii, individuals who tested IgG seropositive using commercial diagnostic ELISA also hadthe strongest salivary antibody responses in salivary antibody tests. A steep increase in anti-norovirus salivary antibody response (immunoconversion) was observed after an episode ofacute diarrhea and vomiting in a volunteer. The Luminex assay also detected seroconversionsto Cryptosporidium using control sera from infected children. Ongoing efforts involve furtherverification of salivary antibody tests and their application in larger pilot community studies.

Published by Elsevier B.V.

Keywords:Luminex immunoassaySalivary antibodyWaterborneHelicobacter pyloriToxoplasma gondiiNorovirusesCryptosporidium

1. Introduction

Waterborne infections can be caused by a number ofbacterial, protozoan, and viral pathogens (Reynolds et al.,2008). Outbreaks of waterborne infections in the US aretypically associated with treatment failures, contamination ofuntreated groundwater, or secondary contamination of

ection Agency; CCL,coerythrin conjugate;y; GST, glutathione S-erval; CoV, coefficient

).

r B.V.

drinking water in the distribution system (Liang et al.,2006; USEPA, 2007). Low levels of pathogens can also bepresent in drinking water during normal operation of publicwater systems. It has been estimated that millions of sporadicwaterborne infections occur annually in the US (Messneret al., 2006; Colford et al., 2006). In order to reduce publichealth burden of waterborne infections, USEPA promulgatednew regulations that imposed more stringent treatment andmonitoring requirements on public water systems (USEPA,2006a,b).

Risk assessments provide theoretical estimates of publichealth effects of drinking water pollution that are used insupport of USEPA's rulemaking. The Safe Drinking Water Actalso mandates USEPA to conduct epidemiological studies of

84 S.M. Griffin et al. / Journal of Immunological Methods 364 (2011) 83–93

waterborne infections to verify these risk assessments (104thCongress, 1996). However, a randomized prospective inter-vention study in a US community did not demonstrate anassociation between tap water consumption and sporadicself-reported gastroenteritis (Colford et al., 2005), whileearlier studies in other countries produced inconsistentresults (Payment et al., 1991, 1997; Hellard et al., 2001).Using fecal sampling in such prospective studies proved to beproblematic due to low rates of compliance and pathogendetection. The invasive nature of blood sampling also limitedthe application of serology in prospective surveys, especiallyin children.

Oral fluid (hereafter referred to as “saliva”) is a promisingalternative to serum in evaluating immune responses tocommon waterborne infections (Gammie et al., 2002; McKieet al., 2002). Because it is easy and safe to collect, salivahas beenused in large-scale, inexpensive, cross-sectional populationsurveys of several infections (Morris-Cunnington et al., 2004a,b; Quoilin et al., 2007). These previous studies used monoplexsalivary immunoassays. A multiplex assay capable of simulta-neous quantitation of salivary antibody responses to severalpathogens would be highly advantageous. In this project, amultiplex immunoassay was developed using Luminex xMAPmicrosphere-based technology (Luminex Corp., Austin, TX).Several pathogens transmissible through drinking water werechosen based on their importance for public health, priorresearch on salivary antibody responses, and availability ofpurified or recombinant immunogenic proteins. Selectionincluded causative agents of chronic or latent infection(Helicobacter pylori and Toxoplasma gondii), and transientacute infection (noroviruses and Cryptosporidium).

2. Materials and methods

The multiplex assay to measure specific salivary antibodyresponses employed a mixture of different sets of microbeadswith unique fluorescent characteristics, each set coupled to aspecific antigen. Two versions of this assay using isotype-specific detection antibodies were developed for measuringspecific IgG and IgA responses. A supplemental duplex assaywith two sets of beads coupled to anti-human IgG and IgAcapture antibodies was also developed to measure totalsalivary immunoglobulins and quantify specific antibodyresponses in relation to the total isotype-specific immuno-globulin content of the sample.

2.1. Saliva and serum samples

Twenty USEPA scientists donated 200 saliva samples ona weekly or monthly basis over two weeks to nine months.Saliva was collected by rubbing the gums with an Oracolsampler sponge (Malvern Medical Developments, Worcester,UK) for 1 min, separated from sponges by centrifugation,and stored at −80 °C until use. Ten volunteers also donatedfinger tip blood samples. Human study protocols wereapproved by the Institutional Review Board at University ofNorth Carolina, Chapel Hill, operating under an agreementwith USEPA. Written informed consent was acquired from allvolunteers.

This project also employed de-identified human sera fromindividuals seropositive toH. pylori (gift ofDr.HarryKleanthous,

Acambis, Cambridge, MA) and sera from children diagnosedwith Cryptosporidium, collected before and after infection (giftof Dr. Honorine Ward, Tufts Medical Center, Boston, MA).

2.2. Coupling proteins to microspheres

Proteins were conjugated to unique sets of fluorescentLuminex microspheres according to manufacturer's recom-mendations. Briefly, 5×106 carboxylated microspheres werepelleted by centrifugation and washed in 100 μL of reagentgrade water. For surface activation, beads were resuspendedin 80 μL of 100 mM monobasic sodium phosphate, pH 7.4with subsequent addition of 10 μL of 50 mg/mL N-hydroxy-sulfosuccinimide (NHS; Pierce Biotechnology, Rockford, IL)and 10 μL of 50 mg/mL 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDC, Pierce), followed by20 min incubation at room temperature.

Four buffers were tested for coupling proteins to micro-spheres: (i) 50 mM 2-(N-morpholino) ethanesulfonic acid(MES), pH 5.0; (ii) 100 mM MES, pH 6.1; (iii) phosphate-buffered saline (PBS), pH 7.4; and (iv) 100 mMNaCl, 100 mMNaHCO3, pH 9.5. Activated beadswerewashed twice in 250 μLof coupling buffer and resuspended in 100 μL of couplingbuffer. Varying amounts of each protein were added todetermine an optimal coupling concentration. Total volumewas brought to 500 μL with coupling buffer, followed byrotational mixing for 2 h at room temperature. Beads werethen pelleted and resuspended in blocking buffer (PBS, 0.1%BSA, 0.05% sodium azide, 0.02% Tween 20, pH 7.4), washedtwice and resuspended in 1 mL of the same buffer, and storedat 4 °C.

2.3. Luminex analysis

MultiScreenBV1.2 μmfiltermicroplates (Millipore, Billerica,MA) were washed with 200 μL of wash buffer (PBS-0.05%Tween 20) and aspirated by vacuum manifold. Microsphereswere diluted to a concentration of 100 microspheres ofeach type per μL in dilution buffer (PBS-1% BSA). Then, 50 μLof bead mixture and 50 μL of sample (primary antibody,diluted serum or saliva) were added to wells. Plates wereincubated on a shaker for 1 h at room temperature at 500 rpm,washed 3 times, and beads resuspended in 50 μL of dilutionbuffer. Next, 50 μL of biotinylated detection antibody indilution buffer was added to wells, followed by incubationfor 30 min at room temperature at 500 rpm and threewashes.Beadswere resuspended in 50 μL of dilution buffer, then 50 μLof dilution buffer with streptavidin R-phycoerythrin conju-gate (SAPE; Invitrogen, Carlsbad, CA)was added towells. Afterincubation for 30 min at the previous conditions and threewashes, beads resuspended in 100 μL of dilution buffer wereanalyzed using a Luminex-100 flow cytometer at defaultsettings. Median Fluorescence Intensity (MFI) of the reportersignal estimated from at least 100 beads of each typewas usedin data analysis.

2.4. Data analysis

Luminex output data were analyzed using SAS 9.2 (SASInstitute, Cary, NC). Four-parameter S-shaped weightedlogistic dose–response curves (De Lean et al., 1978; Plikaytis

85S.M. Griffin et al. / Journal of Immunological Methods 364 (2011) 83–93

et al., 1991; Findlay and Dillard, 2007) were fitted using PROCNLIN. The basic four parameter logistic equation is as follows:

Y = a +b−a

1 +Xc

� �d

where Y is Median Fluorescence Intensity (MFI), X isconcentration of primary antibody or sample dilution factor,a and b are the lower and upper asymptotes respectively, c isthe location parameter (same units as X), and d is thedimensionless shape parameter. The weighting function waschosen based on variability in the response and set equal to 1/(Y)1.8 (Marchese et al., 2009).

Dose–response curves for each protein were compared toidentify the coupling conditions producing the highestreactivity of primary antibody at the lowest concentrations.Regression models that allowed individual parameters a, b, c,and d for each coupling condition were initially fitted to thedata. These full models were compared with reduced(constrained) models that forced one or more parameter(s)to be shared by all curves using standard F tests as previouslydescribed (De Lean et al., 1978).

Below is an example of a curve equation with commonparameters a, b, and d for all k coupling conditions, andcondition-specific parameter c:

Y = a +b−a

1 +X

c0 + m1Z1 + … + mk−1Zk−1

� �d

where c0 is the location parameter for the reference category,Z1 to Zk−1 are binary indicators for k coupling conditioncategories and m1 to mk−1 are parameters reflecting theeffect of changing coupling conditions on the locationparameter c.

The reference category was always defined as couplingbuffer pH 5.0 and the highest protein coupling concentrationtested. If a higher pH was not associated with a significantly(αb0.05) better result, the default pH 5.0 was selected as“optimal”. The highest concentration of protein tested wasassumed “optimal” if the second highest concentrationproduced significantly worse results. If the difference wasnot significant, the next highest concentration was selecteduntil the effect of reducing the protein concentration becamesignificant.

Four-parameter logistic dose–response curves were alsofitted to purified human antibody data obtained from thetotal immunoglobulin assay. The resulting equations werethen rearranged and applied to the results of saliva analysesto estimate concentrations of salivary immunoglobulins fromthe MFI data.

2.5. Development of a multiplex specific antibody assay

2.5.1. Antigens and primary detection antibodies

2.5.1.1. Helicobacter pylori. Detergent extract of whole H.pylori (strain 43504) cell sonicate (R92101, Meridian LifeScience, Saco, ME), purified recombinant urease expressed inE. coli as described by Lee et al. (1995) (gift of Dr. Harry

Kleanthous, Acambis), recombinant flagellin, Vac toxin, andsmall subunit urease (HPA-5040-5, HPA-5010-5 and HPA-5020-5 respectively, Austral Biologicals, San Ramon, CA) werecoupled to different sets of microbeads. Coupling confirmationtests were conducted using de-identified positive control sera(gift of H. Kleanthous), rabbit anti-H. pylori polyclonal antibody(B92310R, Meridian Life Science), rabbit anti-Vac polyclonalantibody, and mouse anti-small subunit urease monoclonalantibody (HPP-5013-9 and HPM-5021-5 respectively, AustralBiologicals).

2.5.1.2. Toxoplasma gondii. Protein coupling experimentsinvolved soluble protein extract of strain RH tachyzoites(R9A350, Meridian Life Science) and three sources ofimmunogenic p30 membrane protein (all from Meridian):(i) recombinant from E. coli (R18426), (ii) purified frominfected HeLa cells (R9A300), and (iii) purified from infectedmice (R86300). Recombinant p30 and p30 from mice weredialyzed at 4 °C overnight against PBS (1 change) using Slide-A-Lyzer 10 K molecular weight cut-off (MWCO) dialysiscassettes (Pierce) to remove storage buffers containingprimary amines, which can bind to carboxyl groups on themicrosphere surface. Mouse monoclonal antibody to T. gondiip30 protein (C86319M, Meridian Life Science) and goatpolyclonal antibody raised against tachyzoites of T. gondiistrain RH (OBT0969, AbD Serotec, Raleigh, NC) were used forcoupling confirmation.

2.5.1.3. Noroviruses. Purified recombinant P domains of majorcapsid proteins from three noroviruses, Norwalk (genogroupI, subgroup 1, GI.1), VA387 (GII.4), and VA207 (GII.9), and apolyclonal antibody raised in guinea pigs against noroviruseswere gifts of Dr. Xi Jiang of Cincinnati Children's HospitalMedical Center (Cincinnati, OH). Purification of these GST-tagged proteins from E. coli and removal of GST tags wereperformed as previously described; purified proteins formed5 nm P-particles comprised of 12 P dimers (Tan and Jiang,2005; Tan et al., 2008). Norovirus GII.4 Minerva virus-likeparticles (VLPs) and two different polyclonal sera raisedagainst Minerva VLPs, one from guinea pig and another fromrabbit, were gifts of Dr. Alec Sutherland (Arizona StateUniversity). These VLPs were prepared using a baculovirusexpression system as previously described (Jiang et al., 1992;Jiang et al., 1995).

2.5.1.4. Cryptosporidium. Purified recombinant gp15 (15 kDa)protein of C. hominis TU502 isolate and mouse anti-C. hominisgp15 monoclonal antibody were gifts of Dr. Honorine Ward(Tufts Medical Center). The protein containing thioredoxin,His, and S tags, was cloned and purified as describedpreviously (Cevallos et al., 2000; Preidis et al., 2007).Transformed E. coli strain BL21 culture expressing therecombinant C. parvum 27 kDa protein with N-terminalglutathione S-transferase (GST) tag and ascites fluid contain-ing mouse monoclonal antibody to this protein were gifts ofDr. Jeffrey Priest (Centers for Disease Control and Prevention,Atlanta, GA). The protein was expressed and purified asdescribed previously (Priest et al., 1999; Moss et al., 2004).Goat anti-C. parvum polyclonal IgG antibody (ab20787,Abcam, Cambridge, MA) was also used for couplingconfirmation.

86 S.M. Griffin et al. / Journal of Immunological Methods 364 (2011) 83–93

2.5.1.5. Internal controls. A set of beads coupled to GST (20237,Pierce) was included in the multiplex assays to control fornon-specific reactivity, particularly with GST tags of certainrecombinant proteins. Rabbit anti-GST antibody (A-5800,Invitrogen, Carlsbad, CA) was used in coupling confirmationexperiments. Beads coupled to bovine serum albumin (BSA,A7030, Sigma-Aldrich, St. Louis, MO) were also included inmultiplex assays to assess potential non-specific reactivity ofsalivary antibodies with BSA-blocked microbeads.

2.5.2. Secondary antibodies and reporterThe following biotinylated secondary antibodies were

used in coupling confirmation tests: donkey anti-goat IgG(ab6884) from Abcam; goat anti-mouse IgG Fc-specific (115-065-071), donkey anti-guinea pig IgG (706-065-148), donkeyanti-rabbit IgG (H+L) (711-065-152), and donkey anti-human IgG Fc-specific (709-065-098), all from JacksonImmunoResearch Laboratories, West Grove, PA. Concentra-tions were prepared at 8 μg/mL for all secondary detectionantibodies and 10 μg/mL for SAPE.

Secondary detection antibodies for analysis of human salivawere selected from among five biotinylated anti-human IgGand three biotinylatedanti-human IgAantibodies: donkey anti-human IgG Fc-specific (709-065-098), donkey anti-human IgGH+L (709-065-149), goat anti-human IgG H+L (109-065-003), all from Jackson ImmunoResearch, goat anti-human IgGH+L (81-7140, Invitrogen), goat anti-human IgGγ (16-10-02,KPL, Gaithersburg, MD), goat anti-human IgAα (109-065-011,Jackson ImmunoResearch), goat anti-human IgA F(ab)′2(AHI1109, Invitrogen), and goat anti-human IgA (16-10-01,KPL). To determineanoptimal antibody concentration, selecteddetection antibodies were assayed at 7.5 to 37.5 μg/mL withseveral serially diluted saliva samples. Next, a concentration ofSAPE was selected from a 2.5 to 17.5 μg/mL range at a constantconcentration of detection antibody.

Table 1Protein coupling conditions for specific antibody and total antibody assays.

Protein Protein concentration(μg per 500 μL)

Coubuf

Specific antibody assayH. pylori whole cell sonicate 25 5.0H. pylori recombinant urease 5 5.0T. gondii soluble protein extract 100 5.0T. gondii p30 from HeLa cells 5 5.0

Norwalk virus P-particles 10 5.0Norovirus VA387 P-particles 10 5.0Norovirus VA207 P-particles 10 5.0Norovirus Minerva VLPs 90 5.0

C. hominis recombinant gp15 protein 5 5.0C. parvum recombinant 27 kDa protein 25 5.0

Glutathione-S-transferase (GST) 5 5.0BSA 25 5.0

Total antibody assayGoat anti-human IgG γ 25 5.0Goat anti-human IgA α 25 5.0

2.6. Serological ELISA assays

IgG responses to H. pylori and T. gondii in serum samplescollected from EPA volunteers were measured using commer-cial diagnostic ELISA kits (IB79236, IBL-America, Minneapolis,MN and EG 127, Viro-Immun Labor-Diagnostika GmbH,Oberursel, Germany, respectively) according to the manufac-turers' instructions. Plates were read at 450 nm on a BioTekSynergy HT (Winooski, VT) microplate reader.

2.7. Development of a duplex assay for total immunoglobulinsG and A

2.7.1. Anti-human capture antibodiesAffinity purified goat anti-human antibodies (capture

antibodies), specific to human IgGγ and IgAα (01-10-02and 01-10-01, KPL) were used in this assay. Couplingconfirmation tests were conducted using purified humanIgG and IgA (009-000-001, 009-000-012, Jackson ImmunoR-esearch) assayed at serial concentrations.

2.7.2. Secondary antibodies and reporterAnti-human IgG and IgA detection antibodies had to be

from the same species (goat) as both capture antibodies inorder to avoid cross-reactivity. Biotinylated goat anti-humanIgG and IgA detection antibodies (16-10-02 and 16-10-01,KPL) were used in this assay. To determine optimal secondaryantibody concentrations, saliva samples were tested with arange of concentrations of each detection antibody, whileSAPE concentration was held constant at 8 μg/mL. At the nextstep, saliva samples were analyzed with a constant concen-tration of detection antibodies and a range of SAPE concen-trations from 1 to 10 μg/mL in order to select the optimalSAPE concentration.

plingfer pH

Confirmation tests (primary and secondary antibodies)

Rabbit anti-H. pylori+biotin donkey anti-rabbitPositive control sera+biotin donkey anti-human IgGGoat anti-T. gondii+biotin donkey anti-goat1) Mouse anti-p30+biotin goat anti-mouse;2) Goat anti-T. gondii+biotin donkey anti-goatGuinea pig anti-norovirus+biotin donkey anti-guinea pigGuinea pig anti-norovirus+biotin donkey anti-guinea pigGuinea pig anti-norovirus+biotin donkey anti-guinea pig1) Guinea pig anti-Minerva+biotin donkey anti-guinea pig;2) Rabbit anti-Minerva + biotin donkey anti-rabbitMouse anti-gp15+biotin goat anti-mouse;1) Mouse anti-27 kDa+biotin goat anti-mouse;2) Goat anti-C. parvum+biotin donkey anti-goatRabbit anti-GST+biotin donkey anti-rabbitNot applicable

Purified human IgG+biotin goat anti-human IgGPurified human IgA+biotin goat anti-human IgA

Fig. 1. Confirmation of coupling H. pylori whole cell sonicate to microbeads using anti-H. pylori antibody at serial dilutions: observed fluorescence data and fittedfour-parameter logistic dose–response curves. These curves shared lower asymptote a and upper asymptote b. Values of the location parameters c are curve-specific. Two values of the shape parameter d were used: one for pH 5.0 to 7.4 and another for pH 9.5.

87S.M. Griffin et al. / Journal of Immunological Methods 364 (2011) 83–93

3. Results

3.1. Multiplex assay for IgG and IgA responses to microbialantigens

The final assay composition and coupling conditions foreach protein are described below and presented in Table 1.Results pertaining to H. pylori sonicate are presented in greaterdetail to demonstrate the methodology, while results for otherproteins are described more succinctly with additionalinformation available in on-line Supplementary materials.

3.1.1. H. pyloriFor whole cell sonicate, coupling experiments were con-

ducted at 25 and 100 μg/500 μL, and pH values 5.0, 6.1, 7.4, and9.5. Coupling confirmation tests using rabbit anti-H. pyloripolyclonal antibody demonstrated that optimal couplingconditions were pH 5.0 and 100 μg/500 μL (Fig. 1, Table 2). Areduced model with a common upper asymptote parameter b

Table 2Parameter estimates for four-parameter logistic dose–response curves fitted to the ccommon lower asymptote a and upper asymptote b parameters. Two shape parametunique location parameter c. P values are provided for contrasts with the reference

Protein couplingconditions

Dose–response curve parameters

pH Concentration,μg/500 μL

Lower asymptotea (95% CI), MFI

Upper asymptoteb (95% CI), MFI

5.0 100 50.7 (20.8, 80.5) 31,519 (27,910, 35,128)5.0 25 Same as above Same as above6.1 100 Same as above Same as above6.1 25 Same as above Same as above7.4 100 Same as above Same as above7.4 25 Same as above Same as above9.5 100 Same as above Same as above9.5 25 Same as above Same as above

did not differ significantly from a full model. The 31,519 MFIupper asymptote value approximates the maximum fluores-cence intensity for Luminex assays using biotinylated detectionantibodies and SAPE (manufacturer's data). Two dose–response curves for experiments conducted at pH 9.5 weresignificantly less steep (smaller absolute value of d) than theother six curves. A common value of d for these two curves wasjustifiable (pN0.05 for the difference). The smallest estimatedvalue of the location parameter c, suggesting themost efficientprotein coupling, corresponded to pH 5.0 and 100 μg/500 μL(reference category). Since values of c for higher pH bufferswere greater indicating less efficient coupling, the default pH5.0 was selected as “optimal”. The effect of using 25 μg/500 μLof protein at pH 5.0 on c was short of being significant,suggesting that this lower concentration can be used to reducethe expenditure of proteins (Table 1). However, since H. pyloridetergent extract is inexpensive and readily available from acommercial source, using a higher protein concentration mayalso be justifiable.

oupling confirmation test data for H. pyloriwhole cell sonicate. All curves haders dwere used: one for pH 5.0 to 7.4 and another for pH 9.5. Each curve had acategory (coupling buffer pH 5.0, concentration of protein 100 μg/500 μL).

Location parameter c Shape parameter d

Parameter estimate(95% CI), μg/mL

P value forcontrast

Parameterestimate (95% CI)

P value forcontrast

0.045 (0.031, 0.058) Reference −1.38 (−1.52, −1.24) Reference0.058 (0.041, 0.075) 0.11 Same as above0.076 (0.053, 0.098) 0.002 Same as above0.057 (0.040, 0.074) 0.14 Same as above1.55 (0.74, 2.36) 0.0006 Same as above2.36 (1.41, 3.30) b0.0001 Same as above19.4 (9.27, 29.4) 0.0005 −0.91 (−1.18, −0.65) 0.00338.5 (15.8, 61.2) 0.002 same as above

0

5000

10000

15000

20000

25000

30000

0 0.5 1 1.5 2

Sal

iva

Lu

min

ex, M

FI

Serum ELISA, OD at 450 nm

SeropositiveSero-nega-tive

Fig. 2. Verification of Luminex test for salivary IgG responses to H. pyloriwhole cell sonicate using diagnostic IgG ELISA of paired serum samples.Saliva was analyzed at 1:4 dilution in the Luminex assay; serum wasanalyzed at 1:100 dilution in ELISA.

88 S.M. Griffin et al. / Journal of Immunological Methods 364 (2011) 83–93

Two EPA volunteers who tested seropositive usingdiagnostic ELISA likewise had the strongest salivary IgGresponses to H. pylori sonicate (Fig. 2). Analysis of seriallydiluted saliva samples from both seropositive (“infected”)and selected seronegative (“not infected”) individuals usingthe multiplex assay which included antigens of otherpathogens and internal controls demonstrated that salivaryIgG responses remained much stronger in seropositiveindividuals, even at a 1:1024 dilution of saliva (Fig. 3).

For H. pylori urease, nine coupling experiments wereconducted at concentrations of 1, 5, and 25 μg/500 μL and pHvalues of 5.0, 6.1, and 7.4. Coupling confirmation testsconducted using positive control sera demonstrated that thebest results were produced using pH 5.0 and 25 μg/500 μL(reference category). However, results of coupling at pH 5.0and 5 μg/500 μL were not significantly different from thereference category (Supplementary Table 1 and Supplemen-tary Fig. 1). Therefore, use of pH 5.0 and 5 μg/500 μL of ureaseat coupling can minimize use of this protein without

Fig. 3. Salivary IgG responses to H. pylori whole cell sonicate in two seropositive ancurves were fitted using four-parameter logistic functions.

compromising the assay performance (Table 1). Optimalcoupling conditions for small subunit urease, Vac, andflagellin were pH 5.0 and 25 μg/500 μL of protein (notshown).

Serial dilutions of saliva from seropositive individualswere analyzed using a multiplex assay that included all ofthe above antigens (Fig. 4). Salivary antibody responses toH. pylori whole cell sonicate and recombinant urease werestrongest; therefore, these two were included in the finalassay (Table 1).

3.1.2. T. gondiiFor soluble protein extract of tachyzoites, optimal cou-

pling conditions occurred at a protein concentration of100 μg/500 μL and pH 5.0 (Supplementary Fig. 2). Couplingconfirmation tests indicated that only p30 protein from HeLacells produced satisfactory results compared to other sourcesof this protein. Optimal coupling conditions for the p30protein were 5 μg/500 μL at pH 5.0 (Table 1); increasingprotein concentration did not improve coupling efficiency(not shown).

Two of 10 EPA volunteers who tested seropositive toT. gondii using diagnostic serological ELISA, likewise had thestrongest salivary IgG responses to this parasite (Supplemen-tary Fig. 3). Using the multiplex assay, salivary IgG responsesto soluble T. gondii proteins remained markedly higher inboth seropositive individuals compared to seronegativecontrols at 1:64 dilution; in one seropositive individual, IgGresponse remained elevated at 1:1,024 dilution of saliva(Supplementary Fig. 4).

3.1.3. NorovirusesRecombinant P-particles of Norwalk, VA387, and VA207

viruses, and VLPs of Minerva viruswere coupled tomicrobeadsat pH 5.0 only. Concentrations at coupling varied from 1 to10 μg/500 μL for P-particles and from 5 to 90 μg/500 μL forMinerva VLPs. Analysis of dose–response curves (results forVA207 are shown on Supplementary Fig. 5) indicated that

d three seronegative individuals at serial dilutions of saliva samples. Smooth

Fig. 4. Comparison of IgG responses to H. pylori whole cell sonicate and various recombinant proteins in saliva from a seropositive volunteer at serial dilutions.Smooth curves were fitted using the spline method.

89S.M. Griffin et al. / Journal of Immunological Methods 364 (2011) 83–93

coupling at 10 μg/500 μL produced significantly (pb0.05)betterresults than at 5 μg/500 μL, although these improvementswere not substantial. While 10 μg/500 μL, is optimal, couplingat 5 μg/500 μL can minimize the use of valuable P-particles.Anti-GST antibodies reacted strongly with beads coupled tothese norovirus proteins suggesting incomplete removal ofthe GST tag during purification. In further tests, salivaryantibody reactivity with GST-coupled beads was used as assaybackground.

For Minerva VLPs, results of coupling at 5 μg/500 μL wereunsatisfactory, probably due to the much larger size of VLPscompared to P-particles (30 nm vs. 5 nm). A substantiallyhigher concentration of VLPs at coupling, 90 μg/500 μL, wasselected for the final assay (Table 1). Tests using themultiplexassay showed that anti-Minerva antibodies cross-reactedstrongly with VA387 (both GII.4 noroviruses), but not withNorwalk and VA207 antigens (Supplementary Fig. 6).

Fig. 5. Temporal patterns of salivary antibody responses to norovirus VA387 P-partdiarrhea and vomiting (1:4 dilution of saliva samples).

For analysis of immunoconversions, all saliva samplesfrom each EPA volunteer were assayed on one plate after theend of follow-up. Two of 20 volunteers reported episodes ofacute diarrhea and vomiting, typical symptoms of norovirusinfection. One individual exhibited a characteristic steepincrease and subsequent gradual decline in salivary IgG andIgA responses to VA387 norovirus after an episode ofsymptoms (Fig. 5). Antibody responses to all other pathogensin the assay, including Norwalk virus, remained practicallyunchanged or slightly declined.

Convalescent phase paired serum and saliva samples fromthis individual were analyzed for IgG and IgA responses at serialdilutionsusing themultiplex assay. Ratios of antibody responsestoVA387P-particles andGSTwere thegreatest at 1:4 dilutionofsaliva suggesting that this dilution was optimal in this case.Salivary IgA exhibited a stronger non-specific reactivity thanIgG. For serum, optimal sample dilution was approximately

icles and selected other antigens in an EPA volunteer who had an episode of

90 S.M. Griffin et al. / Journal of Immunological Methods 364 (2011) 83–93

1:30,000 for IgA tests and N60,000 for IgG, demonstrating ahighintensity antibody response (Supplementary Fig. 7).

3.1.4. CryptosporidiumTwo recombinant proteins of Cryptosporidium spp.,

C. parvum 27 kDa protein and C. hominis gp15 (15 kDa),were each coupled at pH 5.0, 6.1, and 7.4 and proteinconcentrations 1, 5, and 25 μg/500 μL. For gp15 protein, bestcoupling results were produced at pH 5.0 and 25 μg/500 μL.However, results at pH 5.0 and 5 μg/500 μL were notsignificantly worse (Supplementary Fig. 8). Therefore, theseconditions are recommended when protein availability islimited (Table 1). For the 27 kDa protein, optimal couplingconditions were pH 5.0 and 25 μg/500 μL (not shown).

Control pre-infection and post-infection serum samplesfrom two Cryptosporidium-diagnosed children were analyzedat serial dilutions from 1:100 to 1:409,600 using themultiplex assay. Since the recombinant protein of C. parvumcontained a GST tag, the results for this protein and, forconsistency, the C. hominis gp15 protein, were expressed asratios of anti-Cryptosporidium and anti-GST antibody reac-tions. Both children seroconverted to both proteins followinginfection (Fig. 6), but did not experience increases in antibodyreactivity with other antigens in the multiplex assay (notshown). Pearson correlation between log-transformed spe-cific antibody responses to Cryptosporidium in paired sera andsaliva from EPA volunteers was approximately 0.9 for IgA andIgG antibodies.

3.1.5. Selection and optimization of secondary antibodiesTo select best performing anti-human IgG and IgA

detection antibodies for the multiplex assay, saliva samplesfrom volunteers with strong salivary anti-H. pylori, anti-T. gondii, and anti-norovirus VA387 responses were assayedat 1:4 to 1:1,024 dilutions and analyzed using five differentanti-human IgG antibodies and three anti-human IgA anti-bodies (see Section 2.5.2). All five anti-human IgG antibodiesproduced similar MFI for antibody responses to the targetantigens in all positive individuals. However, biotinylateddonkey anti-human IgG Fc-specific antibody (JacksonImmunoResearch) consistently produced two to three times

0

10

20

30

40

50

60

70

80

90

Cp 27 kDa child 1

Ch gp15 child 1

Cp 27 kDa child 2

Ch gp15 child 2

Rat

io o

f re

spo

nse

s to

Cry

pto

an

tig

en a

nd

GS

T

IgA before

IgA after

IgG before

IgG after

Fig. 6. Serum antibody responses to C. parvum (Cp) 27 kDa and C. hominis(Ch) gp15 recombinant proteins in two children before and after diagnosedCryptosporidium infections at 400x (for IgA) and 1600x (for IgG) serumdilutions.

lower reactivity with GST and BSA, as well as Cryptosporidium27 kDa protein compared with other four anti-human IgGantibodies. All three anti-human IgA antibodies producedsimilar results; goat anti-IgAα (Jackson ImmunoResearch)was selected due to its slightly lower cross-reactivity andlower cost. Optimal results were produced using 30 μg/mL ofthese anti-IgG and anti-IgA antibodies; improvement due tohigher antibody concentrations was not significant. The effectof increasing SAPE concentration leveled off at 10 μg/mL (notshown). All tests of saliva and serum samples described abovewere conducted using the optimized multiplex assay.

3.1.6. Effects of multiplexingAliquots of saliva samples from individuals with strong

antibody responses toH. pylori, T. gondii, and VA387 noroviruswere analyzed using monoplex assays for these pathogensand the full optimized multiplex assay. The effects ofmultiplexing on antibody responses to H. pylori and T. gondiiwere not detectable; however, there was approximately 15%reduction in measured anti-VA387 antibody responses.

3.2. Supplementary duplex assay for total immunoglobulinsG and A

Goat anti-human capture antibodies were coupled tomicrobeads at pH 5.0 (Luminex-recommended pH forimmunoglobulins) and concentrations from 5 to 125 μg/500 μL. Coupling confirmation tests using purified humanimmunoglobulins indicated that optimal concentration ofboth antibodies at coupling was 25 μg/500 μL (Table 1).Higher concentrations did not result in significant improve-ment (pN0.05). The optimal concentration for each anti-human IgG and anti-human IgA secondary detection antibodywas 4 μg/mL, while optimal SAPE concentration was 8 μg/mL.

Cross-reactivity tests using serial dilutions of purifiedhuman IgG or IgA demonstrated that, for each antibodyisotype, cross-reactivity with another isotype accounted forless than 0.1% of the estimated antibody concentration.Average intra-plate and inter-plate coefficients of variation(CoV) were 5.4% and 11.4%, respectively, for IgG, and 4.5% and11.7% for IgA. The quantitation ranges for purified human IgGwere from 0.2 ng/mL to 10 ng/mL and from 0.2 ng/mL to80 ng/mL for IgA (Fig. 7). Total salivary IgG and IgAconcentrations were quantifiable in the dilution range from1:5,000 to 1:200,000 (Supplementary Fig. 9). An optimaldilution of saliva for the duplex total IgG and IgA assay wasbetween 1:20,000 and 1:60,000.

Volumes of saliva samples collected from EPA volunteersusing Oracol samplers were approximately normally distrib-uted with mean 0.71 mL, standard deviation 0.33 mL, andrange 0.01–1.50 ml. Considering losses during sample prep-aration, a minimum initial saliva volume of 0.2 mL wasnecessary to conduct all three tests (specific IgG and IgA, andtotal immunoglobulins). Five percent of samples had volumesless than 0.2 mL.

Concentrations of total IgG and IgA in saliva samplesapproximated a log-normal distribution with geometricmean (geometric standard deviation) 21 (2.6) and 50 (2.2)μg/mL, respectively. Random intercept regression analysisdemonstrated that larger sample volumes were associatedwith lower antibody concentrations: IgG and IgA were 43%

Fig. 7. Duplex assay for total IgG and IgA: four-parameter logistic dose–response curves fitted to the purified human immunoglobulins G and A data.

91S.M. Griffin et al. / Journal of Immunological Methods 364 (2011) 83–93

(95% CI 34%, 51%) and 34% (22%, 44%) lower at the 75thpercentile of sample volume distribution compared to the25th percentile.

4. Discussion

This manuscript presents the first stage of a pilot, proof-of-concept project to develop a non-invasive salivary anti-body technique for surveillance of waterborne infections. TheLuminex fluorescent microsphere suspension method wasselected due to its multiplexing capabilities, which cangreatly reduce analysis costs and the necessary volume ofsaliva.

Previous research demonstrated that pH 5.0 is optimal forbinding immunoglobulins to Luminex beads because it allowsfor lysine residues in the antibody's constant region to bindcarboxyl groups on the bead surface. This orientates theantibody binding region away from the surface (Martins et al.,2006). Different pH values may be optimal for coupling otherproteins to beads (Martins et al., 2006). In this study, severalunrelated purified proteins were coupled to beads at pHvalues from 5.0 to 9.5; pH 5.0 was selected as an optimalcondition for all proteins.

Salivary and serological IgG tests for H. pylori tend to havesimilarly high sensitivity and specificity values (Luzza et al.,1997; Reilly et al., 1997; Bode et al., 2002). Limited existingdata suggest that salivary IgG responses to several antigens ofT. gondii strongly correlate with systemic antibody responses(Stroehle et al., 2005). In this study, results of Luminex testsfor salivary IgG antibodies to H. pylori and T. gondii matchedthose of diagnostic serological ELISA tests. However, furthervalidation using a larger set of samples is necessary toevaluate sensitivity and specificity of the Luminex assay.

Diagnostic serological tests for acute short-term infec-tions, norovirus and Cryptosporidium, are not available.Incidence of these infections cannot be reliably elucidatedfrom cross-sectional antibody data. While seroconversion canserve as an indicator of incident infections, prospectiveserological studies are few due to the invasive nature of

blood sampling. The non-invasive salivary antibody methodlends itself to prospective studies.

Serum and salivary anti-norovirus IgA antibodies increasesteeply within several days after infection and start to declinewithin two weeks, while IgG antibodies typically peakbetween two and three weeks post-infection, and thengradually decline (Erdman et al., 1989; Lindesmith et al.,2003, 2005). Previous studies defined immunoconversion asat least a fourfold increase in serum or salivary anti-norovirusantibody responses (Monroe et al., 1993; Moe et al., 2004). Inthis project, one volunteer experienced an immunoconver-sion to norovirus VA387 after an episode of symptoms typicalof norovirus infection. The large magnitude of an increase inanti-VA387 antibody with no concurrent increases inresponses to Norwalk virus or other pathogens in the assayprovides strong evidence of infection with a genogroup IInorovirus (these samples were not tested for responses toVA207 and Minerva noroviruses, which were added to themultiplex assay at a later stage).

Two Cryptosporidium proteins that were included in thisassay have been shown to induce strong antibody responsesin humans (Moss et al., 1994, 1998a). The median half-life ofserum IgG to these proteins can vary from 12 weeks (Priestet al., 2001) to almost 15 months (Sandhu et al., 2006), whileserum IgA response can decay to a non-detectable levelwithin 10 weeks post-infection (Moss et al., 1998a,b). In thisstudy, the multiplex immunoassay detected concurrent ser-oconversions to both C. parvum 27 kDa and C. hominis 15 kDaproteins in two children with confirmed Cryptosporidiuminfections (species not determined). These results are consis-tent with previous research indicating the lack of speciesspecificity of antibody responses to these proteins (Sandhuet al., 2006; Priest et al., 2006).

This salivary antibody assay remains to be validated usingsaliva from individuals with diagnosed Cryptosporidiuminfections. Previous research demonstrated elevated anti-Cryptosporidium salivary IgG and IgA responses in recentlyinfected individuals (Cozon et al., 1994; Moss et al., 2004).Elevated salivary IgG responses to the gp15 protein have also

92 S.M. Griffin et al. / Journal of Immunological Methods 364 (2011) 83–93

been statistically associated with recent episodes of diarrhea(Egorov et al., 2010).

The Oracol samplers, designed to collect crevicular fluidenrichedwith serumantibodies (Nishanian et al., 1998;McKieet al., 2002), are well accepted across age groups (Nokes et al.,1998). This method produces higher titers of total IgGcompared with other saliva/oral fluid sampling methods(Vyse et al., 2001). The present study using Oracol samplersdemonstrated an inverse association between sample volume(a function of sampling technique and individual salivaoutput) and concentration of total salivary immunoglobulins.Quantifying specific antibody responses in relation to the totalimmunoglobulin content helped to account for sample-to-sample variability in saliva composition and demonstrate astrong association between episodes of diagnosed infectionsor acute gastrointestinal symptoms and salivary antibodyimmunoconversions to specific pathogens in subsequent pilotsurveys (manuscripts in preparation).

In summary, this project demonstrates that salivaryantibody has the potential to serve as a useful bioindicatorof selected infections. After further validation, this non-invasive method may facilitate relatively low cost epidemi-ological studies involving adults and children.

Supplementary materials related to this article can befound online at doi:10.1016/j.jim.2010.11.005.

Acknowledgements

Shannon Griffin was supported through an appointmentto the Research Participation Program at the U.S. Environ-mental Protection Agency administered by the Oak RidgeInstitute for Science and Education through an interagencyagreement between the U.S. Department of Energy andUSEPA. The authors are grateful to Drs. Jeffrey Priest, JanVinje, Harry Kleanthous, Honorine Ward, Xi Jiang, and AlecSutherland for kindly providing proteins, antibodies, orcontrol sera for this study, and Eric Rhodes, Jeffrey Swartout,and Swinburne Augustine for critical review of this manu-script. Although this work was reviewed by USEPA andapproved for publication, it represents views of its authorsand does not reflect official Agency policy. The mention oftrade names or commercial products does not constituteendorsement or recommendation for use.

References

Bode, G., Marchildon, P., Peacock, J., Brenner, H., Rothenbacher, D., 2002.Diagnosis of Helicobacter pylori infection in children: comparison of asalivary immunoglobulin G antibody test with the [13C]urea breath test.Clin. Diagn. Lab. Immunol. 9, 493.

Cevallos, A.M., Zhang, X., Waldor, M.K., Jaison, S., Zhou, X., Tzipori, S., Neutra,M.R., Ward, H.D., 2000. Molecular cloning and expression of a geneencoding Cryptosporidium parvum glycoproteins gp40 and gp15. Infect.Immun. 68, 4108.

Colford Jr., J.M., Roy, S., Beach, M.J., Hightower, A., Shaw, S.E., Wade, T.J., 2006.A review of household drinking water intervention trials and anapproach to the estimation of endemic waterborne gastroenteritis inthe United States. J. Water Health 4 (Suppl 2), 71.

Colford Jr., J.M., Wade, T.J., Sandhu, S.K., Wright, C.C., Lee, S., Shaw, S., Fox, K.,Burns, S., Benker, A., Brookhart, M.A., van der Laan, M., Levy, D.A., 2005. Arandomized, controlled trial of in-home drinking water intervention toreduce gastrointestinal illness. Am. J. Epidemiol. 161, 472.

Cozon, G., Biron, F., Jeannin, M., Cannella, D., Revillard, J.P., 1994. SecretoryIgA antibodies to Cryptosporidium parvum in AIDS patients with chroniccryptosporidiosis. J. Infect. Dis. 169, 696.

DeLean, A., Munson, P.J., Rodbard, D., 1978. Simultaneous analysis of familiesof sigmoidal curves: application to bioassay, radioligand assay, andphysiological dose–response curves. Am. J. Physiol. 235, E97.

Egorov, A.I., Montuori Trimble, L.M., Ascolillo, L., Ward, H.D., Levy, D.A.,Morris, R.D., Naumova, E.N., Griffiths, J.K., 2010. Recent diarrhea isassociated with elevated salivary IgG responses to Cryptosporidiumin residents of an Eastern Massachusetts community. Infection 38,117.

Erdman, D.D., Gary, G.W., Anderson, L.J., 1989. Serum immunoglobulin Aresponse to Norwalk virus infection. J. Clin. Microbiol. 27, 1417.

Findlay, J.W., Dillard, R.F., 2007. Appropriate calibration curve fitting in ligandbinding assays. AAPS J. 9, E260.

Gammie, A., Morris, R., Wyn-Jones, A.P., 2002. Antibodies in crevicular fluid:an epidemiological tool for investigation of waterborne disease.Epidemiol. Infect. 128, 245.

Hellard, M.E., Sinclair, M.I., Forbes, A.B., Fairley, C.K., 2001. A randomized,blinded, controlled trial investigating the gastrointestinal health effectsof drinking water quality. Environ. Health Perspect. 109, 773.

Jiang, X., Matson, D.O., Ruiz-Palacios, G.M., Hu, J., Treanor, J., Pickering, L.K.,1995. Expression, self-assembly, and antigenicity of a Snow Mountainagent-like Calicivirus capsid protein. J. Clin. Microbiol. 33, 1452.

Jiang, X., Wang, M., Graham, D.Y., Estes, M.K., 1992. Expression, self-assembly, and antigenicity of the Norwalk virus capsid protein. J. Virol.66, 6527.

Lee, C.K., Weltzin, R., Thomas Jr., W.D., Kleanthous, H., Ermak, T.H., Soman, G.,Hill, J.E., Ackerman, S.K., Monath, T.P., 1995. Oral immunization withrecombinant Helicobacter pylori urease induces secretory IgA antibodiesand protects mice from challenge with Helicobacter felis. J. Infect. Dis.172, 161.

Liang, J.L., Dziuban, E.J., Craun, G.F., Hill, V., Moore, M.R., Gelting, R.J.,Calderon, R.L., Beach, M.J., Roy, S.L., Centers for Disease Control andPrevention (CDC), 2006. Surveillance for waterborne disease andoutbreaks associated with drinking water and water not intended fordrinking — United States, 2003–2004. MMWR Surveill. Summ. 55, 31.

Lindesmith, L., Moe, C., Lependu, J., Frelinger, J.A., Treanor, J., Baric, R.S., 2005.Cellular and humoral immunity following Snow Mountain viruschallenge. J. Virol. 79, 2900.

Lindesmith, L., Moe, C., Marionneau, S., Ruvoen, N., Jiang, X., Lindblad, L.,Stewart, P., LePendu, J., Baric, R., 2003. Human susceptibility andresistance to Norwalk virus infection. Nat. Med. 9, 548.

Luzza, F., Oderda, G., Maletta, M., Imeneo, M., Mesuraca, L., Chioboli, E., Lerro,P., Guandalini, S., Pallone, F., 1997. Salivary immunoglobulin G assay todiagnose Helicobacter pylori infection in children. J. Clin. Microbiol. 35,3358.

Marchese, R.D., Puchalski, D., Miller, P., Antonello, J., Hammond, O., Green, T.,Rubinstein, L.J., Caulfield, M.J., Sikkema, D., 2009. Optimization andvalidation of a multiplex, electrochemiluminescence-based detectionassay for the quantitation of immunoglobulin G serotype-specificantipneumococcal antibodies in human serum. Clin. Vaccine Immunol.16, 387.

Martins, T.B., Augustine, N.H., Hill, H.R., 2006. Development of a multiplexedfluorescent immunoassay for the quantitation of antibody responses togroup A streptococci. J. Immunol. Methods 316, 97.

McKie, A., Vyse, A., Maple, C., 2002. Novel methods for the detection ofmicrobial antibodies in oral fluid. Lancet Infect. Dis. 2, 18.

Messner, M., Shaw, S., Regli, S., Rotert, K., Blank, V., Soller, J., 2006. Anapproach for developing a national estimate of waterborne disease dueto drinking water and a national estimate model application. J. WaterHealth 4 (Suppl 2), 201.

Moe, C.L., Sair, A., Lindesmith, L., Estes, M.K., Jaykus, L.A., 2004. Diagnosis ofNorwalk virus infection by indirect enzyme immunoassay detection ofsalivary antibodies to recombinant Norwalk virus antigen. Clin. Diagn.Lab. Immunol. 11, 1028.

Monroe, S.S., Stine, S.E., Jiang, X., Estes, M.K., Glass, R.I., 1993. Detection ofantibody to recombinant Norwalk virus antigen in specimens fromoutbreaks of gastroenteritis. J. Clin. Microbiol. 31, 2866.

Morris-Cunnington, M.C., Edmunds, W.J., Miller, E., Brown, D.W., 2004a. Anovel method of oral fluid collection to monitor immunity to commonviral infections. Epidemiol. Infect. 132, 35.

Morris-Cunnington, M.C., Edmunds, W.J., Miller, E., Brown, D.W., 2004b. Apopulation-based seroprevalence study of hepatitis A virus using oralfluid in England and Wales. Am. J. Epidemiol. 159, 786.

Moss, D.M., Bennett, S.N., Arrowood, M.J., Hurd, M.R., Lammie, P.J., Wahlquist,S.P., Addiss, D.G., 1994. Kinetic and isotypic analysis of specificimmunoglobulins from crew members with cryptosporidiosis on a U.S.Coast Guard cutter. J. Eukaryot. Microbiol. 41, 52S.

Moss, D.M., Bennett, S.N., Arrowood, M.J., Wahlquist, S.P., Lammie, P.J., 1998a.Enzyme-linked immunoelectrotransfer blot analysis of a cryptosporidiosisoutbreak on a United States Coast Guard cutter. Am. J. Trop. Med. Hyg. 58,110.

93S.M. Griffin et al. / Journal of Immunological Methods 364 (2011) 83–93

Moss, D.M., Chappell, C.L., Okhuysen, P.C., DuPont, H.L., Arrowood, M.J.,Hightower, A.W., Lammie, P.J., 1998b. The antibody response to 27-, 17-,and 15-kDa Cryptosporidium antigens following experimental infectionin humans. J. Infect. Dis. 178, 827.

Moss, D.M., Montgomery, J.M., Newland, S.V., Priest, J.W., Lammie, P.J., 2004.Detection of Cryptosporidium antibodies in sera and oral fluids usingmultiplex bead assay. J. Parasitol. 90, 397.

Nishanian, P., Aziz, N., Chung, J., Detels, R., Fahey, J.L., 1998. Oral fluids as analternative to serum for measurement of markers of immune activation.Clin. Diagn. Lab. Immunol. 5, 507.

Nokes, D.J., Enquselassie, F., Vyse, A., Nigatu, W., Cutts, F.T., Brown, D.W.,1998. An evaluation of oral-fluid collection devices for the determinationof rubella antibody status in a rural Ethiopian community. Trans. R. Soc.Trop. Med. Hyg. 92, 679.

Payment, P., Richardson, L., Siemiatycki, J., Dewar, R., Edwardes, M., Franco,E., 1991. A randomized trial to evaluate the risk of gastrointestinaldisease due to consumption of drinking water meeting currentmicrobiological standards. Am. J. Public Health 81, 703.

Payment, P., Siemiatycki, J., Richardson, L., Renaud, G., Franco, E., Prevost, M.,1997. A prospective epidemiological study of gastrointestinal healtheffects due to the consumption of drinking water. Int. J. Environ. HealthRes. 7, 5.

Plikaytis, B.D., Turner, S.H., Gheesling, L.L., Carlone, G.M., 1991. Comparisonsof standard curve-fitting methods to quantitate Neisseria meningitidisgroup A polysaccharide antibody levels by enzyme-linked immunosor-bent assay. J. Clin. Microbiol. 29, 1439.

Preidis, G.A., Wang, H.-C., Lewis, D.E., Castellanos-Gonzalez, A., Rogers, K.A.,Graviss, E.A., Ward, H.D., White Jr., A.C., 2007. Seropositive humansubjects produce interferon gamma after stimulation with recombinantCryptosporidium hominis gp15. Am. J. Trop. Med. Hyg. 77, 583.

Priest, J.W., Bern, C., Xiao, L., Roberts, J.M., Kwon, J.P., Lescano, A.G., Checkley,W., Cabrera, L., Moss, D.M., Arrowood, M.J., Sterling, C.R., Gilman, R.H.,Lammie, P.J., 2006. Longitudinal analysis of Cryptosporidium species-specific immunoglobulin G antibody responses in Peruvian children.Clin. Vaccine Immunol. 13, 123.

Priest, J.W., Kwon, J.P., Moss, D.M., Roberts, J.M., Arrowood, M.J., Dworkin, M.S.,Juranek, D.D., Lammie, P.J., 1999. Detection by enzyme immunoassay ofserum immunoglobulinGantibodies that recognize specificCryptosporidiumparvum antigens. J. Clin. Microbiol. 37, 1385.

Priest, J.W., Li, A., Khan, M., Arrowood, M.J., Lammie, P.J., Ong, C.S., Roberts, J.M.,Isaac-Renton, J., 2001. Enzyme immunoassay detection of antigen-specificimmunoglobulin G antibodies in longitudinal serum samples from patientswith cryptosporidiosis. Clin. Diagn. Lab. Immunol. 8, 415.

Quoilin, S., Hutse, V., Vandenberghe, H., Claeys, F., Verhaegen, E., De Cock, L.,Van Loock, F., Top, G., Van Damme, P., Vranckx, R., Van Oyen, H., 2007. Apopulation-based prevalence study of hepatitis A, B and C virus usingoral fluid in Flanders, Belgium. Eur. J. Epidemiol. 22, 195.

Reilly, T.G., Poxon, V., Sanders, D.S., Elliott, T.S., Walt, R.P., 1997. Comparisonof serum, salivary, and rapid whole blood diagnostic tests forHelicobacter pylori and their validation against endoscopy based tests.Gut 40, 454.

Reynolds, K.A., Mena, K.D., Gerba, C.P., 2008. Risk of waterborne illness viadrinking water in the United States. Rev. Environ. Contam. Toxicol. 192,117.

Sandhu, S.K., Priest, J.W., Lammie, P.J., Hubbard, A., Colford Jr, J.M., EisenbergJr., J.N., 2006. The natural history of antibody responses to Cryptosporid-ium parasites in men at high risk of HIV infection. J. Infect. Dis. 194, 1428.

Stroehle, A., Schmid, K., Heinzer, I., Naguleswaran, A., Hemphill, A., 2005.Performance of a Western immunoblot assay to detect specific anti-Toxoplasma gondii IgG antibodies in human saliva. J. Parasitol. 91, 561.

Tan, M., Fang, P., Chachiyo, T., Xia, M., Huang, P., Fang, Z., Jiang, W., Jiang, X.,2008. Noroviral P particle: structure, function and applications in virus-host interaction. Virology 382, 115.

Tan, M., Jiang, X., 2005. The p domain of norovirus capsid protein forms asubviral particle that binds to histo-blood group antigen receptors.J. Virol. 79, 14017.

USEPA, 2006a. GWR 40 CFR Parts 9, 141, and 142 national primary drinkingwater regulations: ground water rule; final rule. Federal Register/Vol. 71,No. 216.

USEPA, 2006b. LT2 40 CFR Parts 9, 141, and 142 national primary drinkingwater regulations: long term 2 enhanced surface water treatment rule;final rule. Federal Register/Vol. 71, No. 3.

USEPA, 2007. Estimating the burden of disease associated with outbreaksreported to the U.S. waterborne disease outbreak surveillance system:identifying limitations and improvements. U.S. Environmental Protec-tion Agency, National Center for Environmental Assessment, Cincinnati,OH. EPA/600/R-06/069. http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=156410.

Vyse, A.J., Cohen, B.J., Ramsay, M.E., 2001. A comparison of oral fluidcollection devices for use in the surveillance of virus diseases in children.Public Health 115, 201.

104th Congress, August 6, 1996. Safe Drinking Water Act Amendments of1996. Public Law 104–182. Summary at http://www.epa.gov/OGWDW/sdwa/summ.html#3D.