Journal of Immunological Methods 405 (2014) 67–73

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 luciferase-based system for the detection ofZnT8 autoantibodies

Julia Ustinova a,c, Eva Zusinaite b, Meeme Utt a,c, Kaja Metsküla a, Koit Reimand a,c,Vimala Huchaiah a, Andres Merits b, Raivo Uibo a,c,d,⁎a Department of Immunology, Institute of Biomedicine and Translational Medicine, Faculty of Medicine, University of Tartu, Estoniab Institute of Technology, Faculty of Science and Technology, University of Tartu, Estoniac Centre of Excellence for Translational Medicine, University of Tartu, Estoniad Immunotron OÜ, Estonia

a r t i c l e i n f o

⁎ Corresponding author at: Department of ImmBiomedicine and Translational Medicine, Faculty of MTartu, Ravila 19, 51014 Tartu, Estonia. Tel.: +372 7374232.

E-mail address: raivo.uibo@ut.ee (R. Uibo).

0022-1759/$ – see front matter © 2014 Elsevier B.V. Ahttp://dx.doi.org/10.1016/j.jim.2014.01.009

a b s t r a c t

Article history:Received 11 October 2013Received in revised form 19 December 2013Accepted 14 January 2014Available online 22 January 2014

Appearance of autoantibodies represents the first detectable sign of autoimmune destructionof beta cells in type 1 diabetes (T1D). In addition, autoantibody levels represent an importantpredictive marker regarding the development of an autoimmune process. Recently, the zinctransporter (ZnT8) protein was identified as an autoimmune target in T1D; therefore, there is aneed for reliable and simple methods for detection of ZnT8 autoantibodies.This report describes an assay designed to detect anti-ZnT8 autoantibodies in the serum ofpatients with T1D. This was carried out by generating a ZnT8 C-terminal dimer fused to theN-terminus of the Gaussia princeps luciferase that was overexpressed in insect cells using thebaculovirus expression system. Recombinant protein was semi-purified and used as the targetantigen in the liquid-phase luciferase immunoprecipitation system assay (LIPS), and results werecompared to data obtained using a commercial ELISA designed to detect ZnT8 autoantibodies inT1D patient sera, particularly among adults. LIPS was less effective in detecting antibodies inchildren probably due to the relatively high prevalence of IgM anti-ZnT8 antibodies in childrenwith T1D.

© 2014 Elsevier B.V. All rights reserved.

Keywords:Type 1 diabetesZinc transporter 8 (ZnT8)AutoantibodiesLuciferase immunoprecipitation system(LIPS)

1. Introduction

Type 1 diabetes (T1D) is mediated by CD4+/CD8+ T cell-dependent autoimmune responses characterized by destructionof pancreatic β cells in the islet of Langerhans (Anderson andBluestone, 2005). Monitoring of autoantibodies specific to isletantigens facilitates identifying individuals with a predispositionto developing T1D and is therefore important to the preventionof disease onset in these individuals (Achenbach et al., 2005). Thepersistence of autoantibodies reactive against different islet cellproteins predicts T1D progression (Riley et al., 1990; Knip and

unology, Institute ofedicine, University of4231; fax: +372 737

ll rights reserved.

Siljander, 2008). The first identified disease-related autoanti-bodies linked to clinical T1D manifestation were islet cellautoantibodies (ICA) followed by the identification of additionalautoantibodies with specificity to other islet cell antigens. Theseautoantibodies are specific to single tissue antigens and includeantibodies to insulin (IAA), glutamic acid decarboxylase 65(GAD65), or to the tyrosine phosphatase-related IA-2 molecule(IA2A) (Knip and Siljander, 2008). The number of circulatingautoantibodies represents an important predictive marker forthe progression of the autoimmune process. Apparently, theseautoantibodies are not the only ones found in T1D patients' sera.

The cation efflux transporter zinc transporter 8 (ZnT8) is anautoantigen recently identified in patients with new onset T1D(Wenzlau et al., 2007). ZnT8 is a pancreas-specific protein thatbelongs to the ZnT protein family. ZnT8 is localized in insulinsecretory granules and provides zinc for insulinmaturation andstorage in pancreatic β cells (Chimienti et al., 2004). ZnT8 is a

68 J. Ustinova et al. / Journal of Immunological Methods 405 (2014) 67–73

protein with six transmembrane helixes (Chimienti et al.,2004) flanked by cytosolic N- and C-termini (Wenzlau et al.,2007). ZnT8 autoantibodies are directed primarily against theC-terminal domain. A single nucleotide polymorphism in thegene encoding for ZnT8 (rs13266634) results in an amino acidsubstitution at the residue 325 from arginine to tryptophan(R325W) in the C-terminus of ZnT8; this polymorphism hasbeen hypothesized to have an influence on β-cell function(Wenzlau et al., 2008a). ZnT8 autoantibodies (ZnT8A) arediscovered in about 60–80% of patients with T1D. Moreover,several patients who were positive for a single islet autoanti-body were also positive for ZnT8A suggesting that ZnT8A maybe a marker for identifying patients at risk for T1D (Wenzlauet al., 2007).

The C-terminal domain of human ZnT8 contains twowell-defined epitopes (Wenzlau et al., 2008b) recognized bythe majority of ZnT8A that can be exploited in the screeningof autoantibodies present in patient sera. The liquid-phaseluciferase immunoprecipitation system (LIPS) assay is a highlysensitive and specificmethod for detecting antibodies to proteintargets using Renilla or Gaussia luciferase (RLuc or Gluc)-taggedantigens (Burbelo et al., 2011). LIPS can detect many conforma-tional epitopes, has a highly dynamic range of detection, and isinexpensive (Burbelo et al., 2010a). Moreover, LIPS decreasesincubation time that is beneficial to point-of-care testingapproaches (Burbelo et al., 2008a; Burbelo et al., 2009). It hasbeen shown that T1D autoantigens contain many conforma-tional epitopes. Currently, radioactive liquid-phase assays arethe preferred method for detecting autoantibodies to theseautoantigens (Liu and Eisenbarth, 2007); however, LIPS can beused as a nonradioactive alternative for detecting the presenceof T1D-associated autoantibodies with a high diagnostic po-tential (Burbelo et al., 2008b; Burbelo et al., 2010b). Comparedto Renilla luciferase, Gaussia luciferase has much brighterluminescence significantly increasing sensitivity of the assay(Tannous et al., 2005). Moreover, the secreting nature of thisenzyme allows measurement of its activity in the growthmedia without cell lysis, which considerably simplifies antigenpurification procedure and the assay itself.

The present study describes a novel LIPS approach fordetecting the presence of ZnT8 autoantibodies in human serausing Gaussia luciferase as a luminescent marker. LIPS is asensitive and precise assay that can be used to predict theonset of autoimmune diabetes.

2. Materials and methods

2.1. Patient and control sera

This studywas preformedwith blood samples obtained fromthe biobank of the Department of Immunology, Institute of Bio-and Translational Medicine, University of Tartu. Blood samples(n = 232) were obtained from 73 adults and 159 childrenrecruited from the Tartu University Hospital, Estonia, between2009 and 2011. TID patients (n = 109) and age- and sex-matched controls (n = 123) were enrolled. Diabetes diagnosiswas based on criteria defined by the Expert Committee onDiagnosis and Classification of Diabetes Mellitus (Diabetes Care2003). In all T1D patients, blood samples have been taken atthe onset of the disease. In 55 T1D patients, GADA and IA2Adata was available showing that all but 4 had antibodies.

Corresponding antibody titers for children were 5–N2000 U/mland 20–N3000 U/ml, and for adults 17–550 U/ml and 850–N4000 U/ml. The Ethics Review Committee on Human Researchof the University of Tartu approved the study, and writteninformed consent was obtained from each participant. Serawere delivered coded for testing with LIPS. Sera were storedat −80 °C (long storage), aliquoted, and stored at −20 °C(frequent usage). Sera were then tested using a commercialELISA kit (ElisaRSR™ZnT8 Ab™, RSR Limited, UK) for ZnT8seropositive or seronegative status. The sensitivity and speci-ficity of the LIPS assay were determined based on diagnosis(presence or absence of T1D) and compared to data obtainedusing the commercial ELISA kit.

2.2. Cell lines and media

DH10Bac™ cells (Invitrogen, Germany) were used for thetransposition of the expression cassette into the baculovirusbackbone plasmid. The SF9 insect cell line (Invitrogen, Germany)derived from pupal ovarian tissue of the Fall armywormSpodoptera frugiperda was grown in serum-free Ex-Cell 420medium (Sigma, Germany) at 28 °C without CO2. The Tn5insect cell line (Invitrogen, Germany)was grown in serum-freeEx-Cell 405 medium (Sigma, Germany) at 28 °C without CO2.

2.3. Construction of the recombinant protein expression cassetteand the baculovirus expression vector

The coding sequences of the two ZnT8 C-terminal mono-mers (amino acid residues 268 to 369, GeneBank accessionnumber AY117411) linked together with a hinge peptideGSGGSGSGGS coding sequence were ordered as synthetic DNAfragment from GeneArt (Life Technologies, Germany). Forexpression in secreted form in insect cells, a sequence codingfor a secretion signal of the baculovirus EGT (Ecdysteroid UDPglycosyltransferase, Chen et al., 1997) proteinwas added to the5′ end of the fragment. The construct contained Trp325 andArg325 polymorphic substitutions in the first and secondmonomers, respectively. PCR was used to add the Gaussiaprinceps luciferase coding sequence from which its secretionsignal was deleted. This construct encoded for the antigenwithan estimated molecular weight of 45 kDa (Fig. 1).

Construction of the recombinant baculovirus Baculo-2CGlucwas done using the Bac-to-Bac Baculovirus Expression System(Invitrogen, Germany) as described by the manufacturer.The coding sequence for 2CGluc (Fig. 1) was cloned into thepFastBac™ donor plasmid and transposed into the baculovirusgenome backbone under the polyhedrin promoter usingDH10Bac E. coli cells. Recombinant Baculo-2CGluc DNA waspurified by minipreparation and transfected into Sf9 insectcells by lipofection (Cellfectin, Invitrogen, Germany). Primaryvirus stock (PI) was collected 72 h post transfection, and thepresence of the 2CGluc coding sequence was verified usingPCR and the expression of the desired protein confirmed byWestern blot analysis (data not shown). The PI low-titer virusstock was used to amplify the virus. SF9 cells were infected byPI, and the secondary high-titer viral stock (PII) was collected72 h post-infection. PII virus titer was determined by immu-nofluorescence (IF) analysis using an in-house developedZnT8-specific mouse monoclonal antibody and an Alexa568-conjugated secondary antibody that was verified using an

Fig. 1. Schematic of the 2CGluc protein construct. The fusion protein contains two C-termini (positions 268–369) of ZnT8 fused to the G. princeps luciferase (Gluc).Position 325 in the first C-terminus is occupied by tryptophan (Trp) and by arginine (Arg) in the second. A secretion signal was added to the 5′ end. C-terminusmonomers are connected via a linker.

69J. Ustinova et al. / Journal of Immunological Methods 405 (2014) 67–73

end-point titrationmethod (detection of Gluc activitywas usedas indication of infection). The mean value was accepted asthe virus titer, and the working virus stock (PII) had a titer5.2 × 107 infectious units per 1 ml (iu/ml).

2.4. Protein production and purification

Tn5 cell linewas used for production of recombinant proteinas follows: cells were infected with the PII viral stock followedby incubation at 28 °C for 55 h. Media, containing secreted2CGlucwas collected, and cell debris removed by centrifugationat 200 ×g for 10 min. A protease Inhibitor Cocktail (PIC, Roche,France) was added to the obtained supernatant, which wasthen ultrafiltrated using an Amicon column (Millipore, USA)with a 100-kDa cutoff to remove baculovirus virions from therecombinant protein (45 kDa) preparation. The flow-throughcontaining recombinant protein was then concentrated usingan Amicon column with a 30-kDa cutoff, and the cell culturemedium substituted with buffer containing 50 mM Tris pH 7.5and 100 mM NaCl. Protein aliquots were subsequently flash-frozen in liquid nitrogen and stored at−80 °C.

40

60

80

100

120

140

160

180

200MOI 0.02

MOI 0.1

MOI 0.5

MOI 2.5

RL

U x

106

2.5. LIPS analysis

Serum samples were diluted 1:10 in buffer A (50 mM TrispH 7.5; 100 mM NaCl; 5 mM MgCl2; 1% Triton X-100) andexperiments performed on two parallel 96-well plates. To eachwell, the following components were added: 40 μl of buffer A,10 μl diluted serum, and 50 μl antigen solution (10–15 × 106

RLU [Relative Light Units] of 2CGluc diluted in buffer A). Theplate was then incubated for 1 h at room temperature on arotary shaker (290 rpm). Next, 5 μl of a 30% suspension ofDynaBeads Protein G (Life Technologies, Oslo) in buffer A wasadded and the plate incubated for 1 h on a rotary shaker atroom temperature. Suspensions containing Protein G beadswere then transferred into centrifuge tubes, washed 4 timeswith 1-ml buffer A andoncewith 1-ml PBS (phosphate bufferedsaline). After the final wash, Gluc activity was measured witha luminometer (Promega) using a coelenterazine substrate(Promega, USA).

0

20

0 20 40 60 80

Incubation time (hours)

Fig. 2. Infection of Tn5 cells with Baculo-2CGluc. Secretion of 2CGluc to thegrowth media of Tn5 cells infected with Baculo-2CGluc at different MOI.Gluc activity was measured from Tn5 cell media at different time points.Data are expressed as RLU (relative light units).

2.6. ELISA

The commercial ZnT8 Ab ELISA test ElisaRSR™ZnT8 Ab™was used to identify the presence of anti-ZnT8 autoanti-bodies in patient sera. This kit's performance was previouslyevaluated on standard sera using an Islet AutoantibodyStandardization Program (IASP) 2012 with the following

outcome: sensitivity 74%, specificity 95.5%, and a ROC-AUCof 86.1%.

2.7. Statistical methods

Receiver operating characteristic (ROC) analysis was per-formed using the MedCalc statistical software (Version 9.6.0.0;Mariakerke).

3. Results

3.1. Optimization of recombinant protein production

Preliminary experiments were carried out to determinethe optimal viral MOI (multiplicity of infection) and theincubation conditions required to obtain high and consistentrecombinant protein yields. Briefly, Tn5 cells were infectedwith different MOIs (0.02, 0.1, 0.5, or 2.5) for 72 h and Glucactivity in the growth media measured at 2, 6, 20, 30, 44, 55,and 72 h post-infection (Fig. 2). These experiments deter-mined that the optimal MOI for infection was 2.5 iu/cell for55 h. Longer incubation times did not increase recombinantprotein production, presumably due to degradation of thesecreted protein in the growth media. Recombinant proteinexpression and purity were determined by Western blotanalysis using ZnT8-specific mouse monoclonal antibodies(data not shown).

70 J. Ustinova et al. / Journal of Immunological Methods 405 (2014) 67–73

3.2. Protein purification

In addition to the secreted 2CGluc recombinant protein, Tn5growthmedia contain numerous other proteins in addition to alarge amount of baculovirus particles. Due to their large size,baculovirus-derived proteins can be easily removed by ultrafil-tration. The flow-through was then concentrated and trans-ferred to storage buffer as described in the Materials andMethods. The efficacy of purification was analyzed using SDS-PAGE followed by Coomassie-blue staining of the gel. The final(concentrated) 2CGluc preparation was not completely free ofother proteins (at least one additional protein with highermolecular mass than ZnT8 was observed by Coomassie-bluestaining), but it was subsequently demonstrated that thepresence of these proteins did not have a negative effect onthe specificity of the test.

The stability of the purified and concentrated 2CGlucantigen at different storage conditions (room temperature,+4 °C, −20 °C, and −80 °C) over time was also evaluated.Analysis of stability at different temperatures revealed that2CGluc was relatively unstable at +4 °C and−20 °C. Further-more, Gluc activity decreased rapidly if the protein wassubjected to repeated freeze/thaw cycles (at least half activityloss after each cycle). Therefore, in order to preserve antigenintegrity for a longer period of time, small aliquots ofthe protein were flash-frozen in liquid nitrogen and storedat −80 °C. Under these conditions, antigen was stable forapproximately 2 months.

3.3. Determination of antigen amount for the LIPS assay

A key parameter affecting the accuracy of the LIPS assay isthe use of minimal but sufficient amounts of recombinantantigen that allows for the discrimination between positiveand negative signals. That is, if the amount of antigen in theimmunoprecipitation reaction was too small, it could lead tofalse-negative results. The opposite scenario could result infalse-positive results due to nonspecific binding.

Since the ZnT8 2C dimer was expressed as a Gluc fusion, itwas expected that Gluc activity would be directly proportionalto the 2CGluc protein amount. In order to determine theoptimal antigen amount required, the LIPS assay was carriedout using different 2CGluc activity levels ranging from1 million to 500 million RLU. This analysis demonstrated thatthe most effective protein amount needed for the LIPS reactionwas 10–15 million RLU allowing for the clear discriminationbetween positive and negative sera (data not shown).

3.4. LIPS assay data analysis

Sensitivity and specificity of the LIPS assay were deter-mined by screening of 232 serum samples and analyzing thedata using the ROC analysis. The ROC curve represents anarea that describes the sensitivity versus 1-specificity of arespective test and provides an opportunity to take intoaccount each value of the predictor as a potential cutpoint(Mandrekar, 2010). In our test, the cutpoints of 1432 RLUfor adults and 1164 for children were chosen based oncalculations (MedCalc software) that identified the mostappropriate cutoff between positive and negative samples.That is, samples with signals N1432 RLU (adults) or N1164

RLU (children) were considered positive. Values below thesecutoffs were considered negative. Under these conditions, thesensitivity and specificity of the developed LIPS assay for T1Din adults were 78.6% and 78.0%, respectively (p b 0.001), andin children the sensitivity and specificity were 87.3% and68.7%, respectively (p b 0.001) (Fig. 3). LIPS intraassay CVswere 8.8% and interassay CVs for negative control 8.0% andfor positive control 16.6 %.

The calculated area under the ROC curve (AUC) for LIPS was0.789 (adults) and 0.803 (children) meaning that this testshould be considered acceptable for the screening of adultserum samples and excellent for the analysis of children serumsamples (an AUC N0.8 is considered excellent) (Mandrekar,2010). Data presented in this report demonstrated that 78%and 80% of adults and children, respectively, with the presenceof ZnT8 autoantibodies would be distinguished from thepatients without autoantibodies. ZnT8 autoantibodies wererevealed in 3 out of 4 patients without GADA or IA2A (fromtotal 55 T1D patients tested for all these antibodies).

3.5. Comparison of the LIPS assay with the commercial ELISA test

Seventy-two adult serum samples (14 patients with T1Dand 58 controls) and 62 children serum samples (41 patientswith T1D and 21 controls) were previously tested using acommercially available ELISA kit; therefore, their status regard-ing the presence of anti-ZnT8 autoantibodies had been defined.

To compare these two autoantibody detection systems(LIPS assay vs. the commercial ELISA test), sera previouslyscreened by ELISA were screened using the LIPS assay followedby ROC analysis (Fig. 4). Differences in the performancebetween two tests were not identical between adults andchildren. Adult serum samples screened by LIPS had an AUC of0.789 (95% CI 0.677 to 0.877) compared to the AUC obtained bythe ELISA of 0.781 (95% CI 0.668 to 0.870). LIPS analysis ofserum obtained from children resulted in an AUC of 0.754 (95%CI 0.628 to 0.854) that was significantly different from theELISA AUC of 0.901 (95% CI 0.799 to 0.962). These resultsdemonstrated that LIPS and the commercial ELISA results areless concordant among children than among adults.

Both assay results gave concordant results when com-pared in 12 T1D children during 1–28 months follow-up—noseroconversion was revealed in either assay.

4. Discussion

Various tests are available for the detection of auto-antibodies in clinical samples, including ELISA, RIP(radioimmunoprecipitation), RBA (Radio-Binding assay), anddirect or indirect IF. An ideal assay should be 100% sensitive and100% specific for a particular autoantibody; however, this is notthe case, and each assay has its advantages and shortcomings. Inaddition, some tests are time-consuming,may involve the use ofradioactivity, or require the use of highly purified antigens.

Since ZnT8was identified as an autoantigen associatedwithT1D progression, it has been a challenge to screen for ZnT8-specific autoantibodies. Expression and purification of full-length ZnT8 using classical E. coli expression systems aretechnically impossible due to the presence of transmembranedomains. Efforts to produce a truncated form of ZnT8 (missingthese transmembrane domains) in E. coli resulted in the

Fig. 3. ROC analysis of the LIPS assay. ROC analysis showing all possible variants of sensitivity and specificity of the LIPS assay. Analysis was done using adult (leftpanel) and children serum samples (right panel).

71J. Ustinova et al. / Journal of Immunological Methods 405 (2014) 67–73

expression of a protein trapped within inclusion bodies orformation large insoluble aggregates (data not shown). Incontrast, production of a radio-labeled protein in an in vitrotranscription/translation systemwas partially successful. How-ever, despite the use of codon-optimized sequences, theresulting antigen yield was modest, and the protein tendedto form aggregates. Therefore, an alternative approach forexpression of reasonably high ZnT8 antigen levels that could berapidly purified for use in an immunoassay was needed. In thisreport, we have described a novel assay for the detection ofZnT8 autoantibodies in patient sera using antigen expressedusing a baculovirus expression system and the LIPS assay.

It was previously demonstrated that the majority of ZnT8autoantibodies were directed against the ZnT8 C-terminus(Wenzlau et al., 2009); therefore, most studies have focused onusing the ZnT8 C-terminus in diagnostic immunoassays. Todate, different ZnT8 C-terminus variants have been constructed

Fig. 4. Comparison of the LIPS and the commercial ELISA tests using ROC curve analysis0.781 and 0.789, respectively (left panel). AUC from ELISA and LIPS analysis of children

for the detection of ZnT8 autoantibodies in patient sera(Vaziri-Sani et al., 2011; Yu et al., 2010). Vaziri-Sani et al., forexample, developed three different constructs with arginine,tryptophan, or glutamine at position 325, expressed themusing an in vitro translation system separately and mixed in anRBA assay. Yu et al. demonstrated expression of a chimericIA2-ZnT8WR (IA-2, ZnT8-R and ZnT8-W) molecule that was invitro translated and used in a radioassay. In contrast, our assaywas based on the use of a recombinant ZnT8 C-terminal dimer(with tryptophan and arginine at positions 325) expressed as aGluc fusion (Fig. 1). The 2CGluc fusion represents an efficienttool since Gluc is highly active and easy to detect. Furthermore,recombinant 2CGluc can be produced in much larger amounts(compared to an in vitro translation protein expression system).For example, using the Quick TnT Sp6 kit (Promega) for in vitrotranslation of 2CGluc, we obtained 25,000 RLU in 1 μl oftranslation mixture at most. Conversely, 1 μl of 2CGluc protein

. AUC from ELISA and LIPS analysis of adult serum samples was calculated to beserum samples was calculated to be 0.901 and 0.754, respectively (right panel).

72 J. Ustinova et al. / Journal of Immunological Methods 405 (2014) 67–73

expressed using the baculovirus expression system had anactivity ofmore than 1 billion RLU. In addition, compared to theproduct obtained using in vitro translation, which is alwayscontaminated with very large amounts of proteins present inthe translation mixture, 2CGluc was expressed using mediawhich did not contain fetal calf serum or any other proteins.Therefore, the only contaminating proteins were those associat-edwith the insect cells that can for themost part be removed bya simple and fast ultrafiltration procedure. Thus, the substrate forthe LIPS assay was much easier and cheaper to produce thanantigens required for RIP.

Employment of luminescent marker fused with antigenallowed to avoid the use of radioactivity that greatly increasedsafety of the assay. High brightening properties of Gaussialuciferase made the assay very sensitive allowing to minimizethe antigen amounts used for immunoprecipitation.

Moreover, compared to the ELISA, the LIPS testwas easier tocarry out and took less time. The LIPS assay was sufficientlysensitive and specific allowing it to be used in different seramonitoring experiments. The specificity and sensitivity of teststhat measure autoantibodies can never be 100% since not allpatients who develop diabetes have autoantibodies and viceversa. That is, some healthy controls can be positive forautoantibodies that may serve as markers for progressiontowards diabetes. For example, in a Japanese population,autoantibodies to ZnT8 were identified in 20% patients withslow-onset and in 58%with acute-onset of T1D (Kawasaki et al.,2011). Other studies demonstrated the presence of autoanti-bodies to ZnT8 in 60–80% of Caucasian patients with T1D(Wenzlau et al., 2007). This showed that these tests consis-tently had sensitivities and specificities under 100%.

Head-to-head comparison between a commercial ELISAand the LIPS assay described in this report showed that theELISA test was more sensitive and specific for analyzing serumfrom children (AUC of LIPS is 0.754 and AUC of ELISA is 0.901)but not differentwith regard to analysis of adult serumsamples(AUC of LIPS is 0.789 and AUC of ELISA is 0.781). A possiblereason could be the fact that the commercial ELISA screened forall antibody isotypes (i.e., IgG, IgM, IgA) whereas LIPS capturedonly IgG autoantibodies. In general, at the initial stages ofthe disease, the immune system produces mostly IgM typeautoantibodies (Pihoker et al., 2005). Therefore, it could be areason for why LIPS was less sensitive than ELISA in detectingautoantibodies in children since the response in this populationwas still developing. However, LIPS could be easily used instudying changes in the same individual in time. Investigatingthe serum from one individual during longer period of timecan give us information, for example, about when exactly theswitch from IgM to IgG occurs.

In conclusion, the developed LIPS assay represents asimple and precise method for detecting autoantibodies toZnT8 that can be used to screen patients with diabetesor patients with the potential to progress towards T1D.In perspective, this assay could be a subject for commerciali-zation and successfully used as a complementary or alternativeof ELISA test.

Acknowledgments

Authors would like to thank Dr. Kaire Heilman, AleksanderPeet, Karin Varik, for clinical investigations and Kaupo Teesalu

for helpful discussion of statistical part. Financial support ofEstonian Science Foundation (grant 7749) and Ministry ofEducation and Research (grant Sf0180035s08), EU RegionalDevelopment Fund and European Union through the EuropeanSocial Fund (Mobilitas grant MJD239) is greatly appreciated.The partial support from EU DIABIMMUNE project F2-2008-202036 is also greatly appreciated.

References

Achenbach, P., Bonifacio, E., Ziegler, A.G., 2005. Predicting type 1 diabetes.Curr. Diab. Rep. 5 (2), 98 (Apr).

Anderson, M.S., Bluestone, J.A., 2005. The NOD mouse: a model of immunedysregulation. Annu. Rev. Immunol. 23, 447.

Burbelo, P.D., Ramanathan, R., Klion, A.D., Iadarola, M.J., Nutman, T.B., 2008a.Rapid, novel, specific, high-throughput assay for diagnosis of Loa loainfection. J. Clin. Microbiol. 46 (7), 2298.

Burbelo, P.D., Hirai, H., Leahy, H., Lernmark, A., Ivarsson, S.A., Iadarola, M.J.,Notkins, A.L., 2008b. A new luminescence assay for autoantibodies tomammalian cell prepared insulinoma-associated protein 2. DiabetesCare 31 (9), 1824.

Burbelo, P.D., Leahy, H.P., Iadarola, M.J., Nutman, T.B., 2009. A four-antigenmixture for rapid assessment of Onchocerca volvulus infection. PLoSNegl. Trop. Dis. 3 (5), e438.

Burbelo, P.D., Ching, K.H., Bush, E.R., Han, B.L., Iadarola, M.J., 2010a.Antibody-profiling technologies for studying humoral responses toinfectious agents. Expert Rev. Vaccines 9 (6), 567.

Burbelo, P.D., Hirai, H., Issa, A.T., Kingman, A., Lernmark, A., Ivarsson, S.A., Notkins,A.L., Iadarola, M.J., 2010b. Comparison of radioimmunoprecipitation withluciferase immunoprecipitation for autoantibodies to GAD65 and IA-2b.Diabetes Care 33 (4), 754.

Burbelo, P.D., Ching, K.H., Bren, K.E., Iadarola, M.J., 2011. Searching forbiomarkers: humoral response profiling with luciferase immunoprecip-itation systems. Expert Rev. Proteomics 8 (3), 309.

Chen, X., Hu, J., Jehle, J.A., Zhang, Y., Vlak, J.M., 1997. Analysis of theecdysteroid UDP-glucosyltransferase gene of Heliothis armigera single-nucleocapsid baculovirus. Virus Genes 15 (3), 219.

Chimienti, F., Devergnas, S., Favier, A., Seve, M., 2004. Identification andcloning of a beta-cell-specific zinc transporter, ZnT-8, localized intoinsulin secretory granules. Diabetes 53 (9), 2330 (Sep).

Kawasaki, E., Nakamura, K., Kurija, G., Satoh, T., Kobayashi, M., Kuwahara, H.,Abiru, N., Yamasaki, H., Matsuura, N., Miura, J., Uchigata, Y., Eguchi, K.,2011. Differences in the humoral autoreactivity to zinc transporter 8between childhood- and adult-onset type 1 diabetes in Japanesepatients. Clin. Immunol. 138, 146.

Knip, M., Siljander, H., 2008. Autoimmune mechanisms in type 1 diabetes.Autoimmun. Rev. 7 (7), 550.

Liu, E., Eisenbarth, G.S., 2007. Accepting clocks that tell time poorly: fluid-phase versus standard ELISA autoantibody assays. Clin. Immunol. 125(2), 120.

Mandrekar, J.N., 2010. Receiver operating characteristic curve in diagnostictest assessment. J. Thorac. Oncol. 5, 1315.

Mellitus ECotDaCoD: report of the expert committee on the diagnosis andclassification of diabetes mellitus, 2003. Diabetes Care 26 (Suppl. 1),S5.

Pihoker, C., Gilliam, L.K., Hampe, C.S., Lernmark, A., 2005. Autoantibodies indiabetes. Diabetes 54 (Suppl. 2), S52.

Riley, W.J., Maclaren, N.K., Krischer, J., Spillar, R.P., Silverstein, J.H.,Schatz, D.A., Schwartz, S., Malone, J., Shah, S., Vadheim, C., Rotter, J.I.,1990. A prospective study of the development of diabetes in relativesof patients with insulin-dependent diabetes. N. Engl. J. Med. 323,1167.

Tannous, B.A., Kim, D.E., Fernandez, J.L., Weissleder, R., Breakefield, X.O.,2005. Codon-optimized Gaussia luciferase cDNA for mammalian geneexpression in culture and in vivo. Mol. Ther. 11 (3), 435.

Vaziri-Sani, F., Delli, A.J., Elding-Larsson, H., Lindblad, B., Carlsson, A.,Forsander, G., Ivarsson, S.A., Ludvigsson, J., Marcus, C., Lernmark, Å.,2011. A novel triple mix radiobinding assay for the three ZnT8 (ZnT8-RWQ) autoantibody variants in children with newly diagnosed diabetes.J. Immunol. Methods 371, 25.

Wenzlau, J.M., Juhl, K., Yu, L., Moua, O., Sarkar, S.A., Gottlieb, P., Rewers, M.,Eisenbarth, G.S., Jensen, J., Davidson, H.W., Hutton, J.C., 2007. The cationefflux transporter ZnT8 (Slc30A8) is a major autoantigen in human type1 diabetes. Proc. Natl. Acad. Sci. U. S. A. 104, 17040.

Wenzlau, J.M., Liu, Y., Yu, L., Moua, O., Fowler, K.T., Rangasamy, S., Walters, J.,Eisenbarth, G.S., Davidson, H.W., Hutton, J.C., 2008a. A common

73J. Ustinova et al. / Journal of Immunological Methods 405 (2014) 67–73

nonsynonymous single nucleotide polymorphism in the SLC30A8 genedetermines ZnT8 autoantibody specificity in type 1 diabetes. Diabetes 57(10), 2693 (Oct).

Wenzlau, J.M., Ong, M., Yu, L., Eisenbarth, G.S., Hutton, J.C., Davidson, H.W.,2008b. Identification of a major humoral epitope in Slc30A8 (ZnT8).Ann. N. Y. Acad. Sci. 1150, 252.

Wenzlau, J.M., Frisch, L.M., Gardner, T.J., Sarkar, S., Hutton, J.C., Davidson,H.W., 2009. Novel antigens in type 1 diabetes: the importance of ZnT8.Curr. Diab. Rep. 9, 105.

Yu, L., Liu, Y., Miao, D., Wenzlau, J., Davidson, H., Hutton, J., Eisenbarth, G.S.,2010. Triple chimeric islet autoantigen IA2–ZnT8WR to facilitate isletautoantibody determination. J. Immunol. Methods 353, 20.