Method development for enrichment of autoantibodies from human plasma

(1)

IN

DEGREE PROJECT BIOTECHNOLOGY, SECOND CYCLE, 30 CREDITS

STOCKHOLM SWEDEN 2020,

Method development for

enrichment of autoantibodies from human plasma

LOVISA SKOGLUND

KTH ROYAL INSTITUTE OF TECHNOLOGY

SCHOOL OF ENGINEERING SCIENCES IN CHEMISTRY, BIOTECHNOLOGY AND HEALTH

(2)

Abstract

Antibodies are naturally occurring in humans, with the function to protect the body from pathogens.

Occasionally, antibodies towards the body’s own proteins are produced. These so called autoantibodies are present in healthy individuals but are also highly associated with diseases with autoimmune involvement.

Research on autoantibodies in healthy individuals as well as in patients is important to gain knowledge and facilitate prognostics, diagnostics and treatment. However, a method for purification of these antibodies has not previously been described.

In the present project, an enrichment procedure of circulating autoantibodies found in human plasma is described. Twenty protein fragments previously known to be highly reactive were attached to magnetic microbeads, enabling autoantibodies from eight human plasma sample pools to be captured. The six antigens with highest shown reactivity were chosen for elution procedure. Using pH alterations and heat treatments, a successful elution and enrichment procedure was developed.

With analysis of the eluted autoantibodies, it can be established that the enrichment was successful on multiple sample pools. In the scaled-up procedure, autoantibodies could be enriched in all positive antigen- sample combinations. Concentration measurements indicated amounts of up to 0.23 mg antibodies per ml eluate. This implies sufficient concentrations for further applications of the enriched autoantibodies.

Keywords autoantibodies, enrichment, autoimmunity, plasma, suspension bead array

Sammanfattning

Antikroppar förekommer naturligt i människor, med syftet att skydda kroppen från patogen. I vissa fall skapas av misstag antikroppar som angriper kroppens egna proteiner. Dessa autoantikroppar förekommer hos alla människor, såväl friska som sjuka, men de är också starkt förknippade med autoimmuna sjukdomar.

Kunskapen om autoantikroppar hos friska personer och hos patienter är idag begränsad, men fortsatt forskning inom området förväntas i framtiden underlätta prognostik, diagnostik och behandling. Hittills har ingen metod för anrikning av autoantikroppar ur blodplasma beskrivits.

I detta projekt beskrivs en anrikningsmetod för autoantikroppar ur blodplasma från människa. Tjugo tidigare kända högreaktiva proteinfragment fästes på magnetiska mikrokulor. Dessa antigen-täckta mikrokulor användes för att fånga in autoantikroppar från åtta plasmaprover. De sex proteinfragment som hade högst reaktivitet i dessa prover valdes ut för elueringsförsök. Eluering genomfördes under basiska följt av sura förhållanden, tillsammans med värmebehandling.

Denna elueringsmetod fungerade för anrikning av några autoantikroppar från flera av plasmaproverna. I ett utökat experiment kunde autoantikroppar anrikas ur alla kombinationer av antigen och plasmaprov som förväntades ge signal. Koncentrationen av autoantikroppar i eluaten uppskattades till högst 0.23 mg/ml.

Denna koncentration är tillräcklig för flera vanliga metoder där antikroppar används.

Nyckelord autoantikroppar, anrikning, autoimmunitet, plasma, suspension bead array

(3)

1. Introduction

Antibodies are produced by the human immune system in order to protect the body from pathogens. These antibodies and their affinity are produced through multiple stochastic steps. In addition, they undergo a selection to avoid affinity towards the body’s own proteins. In some cases, this process fails, yielding autoantibodies that target self-antigens [1,2].

The human immune system is made up by several types of lymphocytes with different functions.

T-lymphocytes target foreign and deviant cells through their T-cell receptors, while B- lymphocytes secrete antibodies, such as immunoglobulin G (IgG) in order to protect the body from pathogens [1–3].

The mammalian IgG has a structure consisting of two light chain subunits connected to two heavy chain subunits. Assembled, the antibody consists of three regions; two fragment antigen-binding (Fab) regions, comprising both heavy chain constant and variable regions, and one fragment crystallisable (Fb) region made of solely constant heavy chains. At the edge of the Fab regions, there are six complementary-determining region (CDR) loops, three from each of the light chains and three from each of the heavy chains. The CDR loops make up the vast majority of the specific recognition and affinity towards antigen epitopes [4]. The antibody-antigen affinity interactions are non-covalent and are therefore based on chemical bonds through van der Waals forces, electrochemical interactions, hydrogen bonds and hydrophobic interactions [5].

The affinity of the CDR loops is generated in several steps and refined through a maturation process. In this process, negative selection is an important step. This should prevent antibodies with affinity towards epitopes present in the host from forming. However, this process sometimes fails, resulting in autoantibodies. Autoantibodies are produced by B-1 cells that reacts with one or more self-antigens [1,2]. The presence of autoantibodies is therefore a result of faulty B-cell response where these fail to distinguish self from nonself [2,6].

Autoantibodies are known to be present not only in patients with autoimmune diseases, but also in healthy individuals [1]. Disease-related autoantibodies may serve as indicators of current disease presence or future disease onset [7–11]. However, autoantibodies are not only harmful, as they may also have neutral or even beneficial effects on the body [12,13].

There are many techniques available to characterise autoantibodies and understand autoantibody responses. A commonly used characterisation method for autoantibodies is epitope mapping. In brief, epitope mapping identifies the length location and sequence of the epitope. Using peptides representing parts of the antigen, it is possible to map which fragments of the whole protein that constitute the epitope [14,15]. However, only linear epitopes may be interrogated in this manner.

(5)

2

A method for enriching autoantibodies from human samples could improve autoantibody characterisation by reducing confounding protein interactions. Enriched autoantibodies could be applied in multiple analyses examining antibody-antigen interactions, such as immunohistochemistry (IHC), epitope mapping and confirmation of autoantibody presence.

To be able to enrich autoantibodies from human plasma, a reliable strategy to elute antibodies from antigens is needed. Agaton et al. [16] has previously reported a technique for purification of antibodies from antisera obtained from immunised animals using protein fragments. The procedure was conducted in columns with large amounts of serum. This proves that antibodies can be purified, but it remains to develop a method where antibodies are enriched from small volumes of human plasma or serum. Ayoglu et al. [17] and Birgersson et al. [18] have described a method for elution of serum and plasma proteins from immobilised antibodies in a bead-based assay. There, a sequential alkaline-acidic elution with heat treatment at 56 °C was applied. As these protocols successfully have been used to break antibody-antigen interactions in a bead assay format, they formed the starting point of the present project.

In the present project, suspension bead array (SBA) technology was used to enrich autoantibodies from human blood plasma. In such assays, capture reagents are immobilised on up to 500 different magnetic colour-coded silica microspheres, or beads. Each bead ID possesses a unique internal dye from three spectral fluorophores, making it possible to distinguish them from each other [9].

Each bead ID is chemically coupled to an affinity reagent, enabling the identification of each reagent even in a mixture of hundreds of bead IDs. Thereby, SBAs can be highly multiplexed [9].

In 96-well or 384-well microtiter plates, the autoimmune landscape of up to 384 samples can be simultaneously investigated using an SBA. The SBA can be applied for either protein profiling or autoimmunity profiling with antibodies and antigens covalently bound to the beads, respectively.

When bound to the capture reagents, the affinity reagents on the microspheres may be detected and analysed using a flow cytometry based system [9].

The affinity reagents used in the present project were recombinant human protein fragments generated within the Human Protein Atlas (HPA) project. These protein fragments were immobilised on magnetic silica microspheres with the aim of developing a general protocol for the enrichment of autoantibodies from human plasma samples. An autoantibody enrichment procedure is of interest to facilitate the research on circulating human autoantibodies. The procedure may be applied to validate results from antibody and protein screenings, in order to verify the affinity and specificity of the antibody of interest. Furthermore, the ability to enrich autoantibodies from human samples may enable additional characterisation of their affinity. This may enable research which improves knowledge on the mechanisms of autoimmune disorders.

(6)

3

2. Materials and method

2. 1. Antigens and samples

In this project, protein fragments denoted Protein Epitope Signature Tags, or PrESTs, generated within the Human Protein Atlas (HPA) project were used to capture autoantibodies from human plasma samples. A total of 20 previously identified highly reactive PrESTs were chosen for the experiments, presented in Table 1.

In this work, we defined highly reactive PrESTs as giving high signals in a majority of healthy individuals [19], or in multiple in-house studies in different disease contexts (unpublished).

Table 1. Antigens chosen for experimental procedure.

Gene name Protein Uniprot ID

CAPRIN2 Caprin-2 Q6IMN6

ARFGAP1 ADP-ribosylation factor GTP-ase-activating protein 1 Q8N6T3

RNF185 E3 ubiquitin-protein ligase RNF185 Q96GF1

ZNF688 Zinc finger protein 688 P0C7X2

TNNT2 Troponin T, cardiac muscle P45379

ATF3 Cyclic AMP-dependent transcription factor ATF-3 P18847

PAPOLA Poly(A) polymerase alpha P51003

RIN3 Ras and Rab interactor 3 Q8TB24

FTCD Formimidoyltransferase-cyclodeaminase O95954

FAXDC2 Fatty acid hydroxylase domain-containing protein 2 Q96IV6

IPO13 Importin-13 O94829

CENPF Centromere protein F P49454

MYH10 Myosin-10 P35580

PRPH Peripherin P41219

KCNJ10 ATP-sensitive inward rectifier potassium channel 10 P78508

NCOA2 Nuclear receptor coactivator 2 Q15596

SIAH2 E3 ubiquitin-protein ligase SIAH2 O43255

DST Dystonin Q03001

ICA1 Islet cell autoantigen 1 Q05084

TDRD6 Tudor domain-containing protein 6 O60522

Eight sample pools were used for the experimental procedure. Pooled sample was utilised in the present project since it was readily available with high volumes. Three sample pools were of commercial origin (sample 6: lot. BRH1176239, sample 7: lot. BRH1176238, sample 8: lot.

BRH1176237, Seralab). The remaining five plasma pools originated from plasma donors for

(7)

4

technical use, with only gender and plasma preparation (lithium heparin or EDTA (ethylenediaminetetraacetic acid)) provided as information. Four pools contained plasma from both sexes, two pools were from females only, and two were from males only. The disease status of the samples was unknown.

2. 2. Experimental procedure

The experimental procedure followed a general workflow. First, sample pools were screened for autoreactivities. These results were applied in the selection of antigen-sample combinations in the elution experiments. Second, a series of elution experiments were performed to optimise protocol parameters. The enrichment beads were analysed before and after elution. These results were compared with eluate screening.

2. 2. 1. Construction of the SBA

Construction of the SBA was carried out as previously described [9]. In summary, 250 000 beads per bead ID were coupled to 2 µg of each PrEST. Each antigen was diluted to 100 µl 2-(N- morpholino)ethanesulfonic acid (MES) (M2933-100G, Sigma). Four control reagents were included and also diluted in MES to 100 µl; 0.9 µg rabbit anti-human IgG (309-005-082, Jackson) as a positive control, 0.675 µg EBNA1 (ab138345, abcam) as a semi-positive control, and 2 µg hexahistidine-albumine binding protein (His6ABP, Immunotech KTH) as a negative control. All PrESTs contain a His6ABP-tag, which may give false signals in human derived samples, as the ABP (albumin-binding protein) part is derived from streptococcal Protein G. Therefore, any affinity to His6ABP is blocked by a pre-incubation step using a buffer containing free His6ABP and the reagent can be coupled to beads as a negative control reagent. In addition, another negative control is included by not coupling anything to a bead ID. This is referred to as the bare-bead control.

Colour-coded magnetic silica microspheres (Luminex corp.) with 24 different IDs were transferred to 24 separate wells in a 96-well Greiner plate. These beads were manually washed with 80 µl activation buffer (S3139-250G, Sigma) and the whole volume then aspirated using BioTek plate washer (EL406, BioTek), followed by addition of 50 µl activation buffer to each well. A volume of 50 µl activation solution (5 % N-Hydroxysuccinimide (24510, Thermo Scientific), 5 % 1-Ethyl- 3-(3-dimethylaminopropyl)carbodiimide (77149, Thermo Scientific), in activation buffer) was added to each well and the plate was incubated on plate shaker at 650 rpm at RT for 20 minutes.

To preserve bead fluorophore, all incubations were made in dark.

The activated beads were washed twice with MES buffer using the BioTek plate washer, before addition of the diluted antigens and controls to assigned wells. To assure coupling, the plate was incubated on plate shaker at 650 rpm in RT for 2 hours. The plate with coupled beads was washed

(8)

5

twice with phosphate buffered saline with 0.05 % Tween20 (PBS-T) using BioTek plate washer.

Beads were resuspended in 50 µl storage buffer (10 % Blocking reagent for ELISA (11 112 589 001, Roche), 0.1 % ProClin (48912-U, Sigma-Aldrich), in Milli-Q water (Millipore)) per well and incubated at 4 °C over night before pooling the beads to form the SBA.

2. 2. 2. Autoantibody profiling of plasma samples

Prior to combining the antigen-coupled beads with the plasma samples, the samples were diluted in assay buffer (3 % bovine serum albumin (BSA), 5 % skim milk powder, 0.2 mg/ml His6ABP in 0.05 % PBS-T) to dilutions of 1:62.5, 1:125 and 1:250. Three dilution rates facilitated the decisions for further analysis and experiments. The diluted samples were incubated on a plate shaker at 650 rpm at RT for one hour before combining 45 µl of each sample dilution in separate wells with SBA (500 beads per ID), followed by incubation on plate shaker at RT for 2 hours. The beads were washed three times with 0.05 % PBS-T using plate washer BioTek, re-suspended with 0.2 % PFA (paraformaldehyde in 0.05 % PBS-T), followed by incubation on plate shaker at RT for 10 minutes.

The beads were again washed and then re-suspended in 0.4 µg/ml R-phycoerythrin (R-PE, H10104, Invitrogen) conjugated anti-human IgG in 0.05 % PBS-T and incubated on plate shaker at RT for 30 minutes. The beads were then washed three times with 0.05 % PBS-T and 100 µl 0.05

% PBS-T was dispensed. The plate was analysed in FlexMap 3D instrument (Luminex corp.), yielding median fluorescence index (MFI) signals and bead count for all bead IDs in each well.

2. 3. Elutions

A series of elution experiments was performed to develop and optimise a general autoantibody enrichment procedure.

2. 3. 1. Coupling of beads for elution

Based on data analysis performed with the standard protocol for autoantibody profiling, eight antigens, including EBNA1 as semi-positive control and His6ABP as negative control, were initially chosen for further experiments and elutions. In order to use large amounts, beads without ID was used for the elution steps of this project. The eight selected antigens were coupled to beads as previously described. As the beads lack ID, they were not pooled, but kept separate.

2. 3. 2. Elution of autoantibodies - proof of concept

The applied elution steps were used as described by Birgersson et al. [18]. These steps utilise gentle temperature and pH treatment.

For the first elution, four samples and four antigens coupled to beads without ID were selected to evaluate the efficiency. The selection was based on high reactivity in analysis of plasma samples

(9)

6

with SBA. This selection was based on the high MFI values from the SBA-run. The samples were diluted 1:125 in assay buffer and incubated on plate shaker at 650 rpm, at RT for one hour. A total of 6000 beads per well was transferred to assigned wells on a 96-well Greiner plate, followed by transfer of 94 µl of the diluted samples. All combinations of the four samples and four antigens were included. The plate was incubated on plate shaker, at RT for 2 hours. The beads were washed with 0.05 % PBS-T and re-suspended in 100 µl 0.05 % PBS-T.

To analyse the amount of antibodies captured on the beads, 8 µl of the bead mixture was transferred from each well to new wells in another 96-well Greiner plate prior to elution. These beads were re-suspended in 0.4 µg/ml R-PE conjugated anti-human IgG in 0.05 % PBS-T, and incubated on plate shaker at RT for 30 minutes. The beads were washed with 0.05 % PBS-T and re-suspended in 100 µl 0.05 % PBS-T, and analysed in a FlexMap 3D instrument. Wells with bead count lower than 32 were excluded. This procedure was also performed with the beads after elution to determine whether antibodies had successfully been eluted from the beads.

In the elution plate, the remaining PBS-T was aspirated using plate washer BioTek. Beads were resuspended in 15 µl elution buffer 1 (0.1 M glycine-NaOH, 0.05 % (v/v) Tween20 in Milli-Q water, pH 10.0,) followed by incubation at 56 °C in water bath for 20 minutes and at RT for additional 10 minutes. The eluates were transferred from the beads (on magnet) to a new plate with wells containing 30 µl neutralization buffer (2 M Tris with 0.2 % (w/v) casein, 1.0 % (w/v) polyvinyl alcohol, 1.6 % (w/v) polyvinylpyrrolidone, 0.05 % (v/v) ProClin, in 1xPBS, pH 7.95).

Beads were resuspended in15 µl of elution buffer 2 (2.5 % acetic acid, 0.05 % (v/v) Tween20 in Milli-Q water, pH 3.0), and the plate was incubated at 56 °C in water bath for 20 minutes followed by 10 minutes at RT. The supernatant was transferred to the corresponding wells in the plate containing neutralization buffer, followed by an incubation at RT for 40 minutes.

The beads were washed with 0.05 % PBS-T using plate washer BioTek and re-suspended in 100 µl 0.05 % PBS-T. Beads were tested once again in FlexMap 3D according to protocol previously described. The results from this analysis was compared to the results from the run prior to the elution in order to determine if antibodies have successfully been eluted from the antigens.

2. 3. 3. Antibody elution in all samples

A second elution experiment was performed according to the same protocol as used in the proof of concept setup in order to test the generality of the strategy. All eight samples and the eight antigen-coupled beads without ID were used in the procedure, generating 64 combinations. The sample concentration and sample amount was increased compared to the first elution to increase the concentration of interacting antibodies. The sample concentrations used in this run was set to 1:62.5, which corresponds to 0.78 µl crude sample. Again, 6000 beads per well were used.

(10)

7 2. 3. 4. Optimising sample dilution

A third elution was performed, evaluating how sample concentration, sample volumes and bead amount used would affect the enrichment of autoantibodies. Sample concentrations of 1:31.25, 1:62.5 and 1:125, diluted sample volumes of 50 µl, 75 µl and 100 µl were combined with bead amounts 3000 and 6000 beads per well, to a total of 18 combinations.

2. 3. 5. Scaling up the elution procedure

For the fourth elution, the experiment was scaled up in terms of sample amount, sample concentration and bead amount. A total of eight combinations from four samples and four antigen- coupled beads were used. A volume of 20 µl crude sample were diluted 1:60 in assay buffer in 1.5 ml Eppendorf tubes. After incubation for one hour on a shake arm at RT, 10 µl of bead stock with concentration 10 000 beads/µl was added to the samples according to pre-assigned combinations.

The tubes were incubated on shake arm at 650 rpm at RT for 2 hours. Supernatant was removed from the tubes on magnet. 0.05 % PBS-T was added to each tube off magnet to transfer the beads to a 96-well Greiner plate. The beads were washed twice with 0.05 % PBS-T using plate washer BioTek and re-suspended in 100 µl 0.05 % PBS-T.

Analysis of amount of affinity reagents on beads and elution was performed according to the protocol described above. However, due to the increased bead amount, 1 µl of beads was used for the analysis steps of beads pre and post elution.

2. 3. 6. SBA analysis of eluates

The eluates derived from all elution procedures were all analysed with the suspension bead array using the following protocol. A volume of 1.5 µl 4x SBA bead stock was distributed in wells corresponding to the number of eluates in a 96-well Greiner plate. The beads were washed twice with PBS to remove storage buffer before 20 µl of each eluate was added to the wells and the plate was incubated on plate shaker, at RT for 2 hours. Washing and readout was performed as described above. IgG concentrations in eluate was measured in NanoDrop (ND-1000, NanoDrop).

2. 4. Stability test of antigen-coupled beads

As a part of this project, the stability of protein fragment-coupled beads over time in buffer was investigated. It is of interest to establish the stability of the coupled beads in order to facilitate future experiments, and to assure that the generated results are reliable.

Eight weeks subsequent to the construction of the SBA, another autoantibody profiling procedure was conducted with crude samples diluted 1:125 according to protocol previously described.

(11)

8

Results were correlated with those from first assay screen to determine SBA stability over time, when stored in storage buffer at 4 °C.

2. 5. Data analysis and visualisation

Data from the experimental procedures was visualised in Rstudio [20] using packages tidyverse [21], ggplot2 [22] and rlang [23].

3. Results

In the present study, autoantibodies were enriched from combinatorial elution steps using a bead- based array. A protocol was adapted based on previous protein elution studies and the procedure was scaled up to be investigated as a general enrichment method. The final experiment was scaled up to examine whether the concentration of eluted antibodies could be high enough for further research. Assay performance was evaluated using the included control beads, with all assays passing.

3. 1. Selection of antigens for elution

A first analysis of eight human plasma sample pools and the 20 chosen highly reactive antigens (PrESTs) coupled to beads in a suspension bead array (SBA) was performed to investigate the presence of autoantibodies in sample pools. From these screening results, six of the bead-coupled antigens were selected for enrichment; RNF185, ZNF688, ATF3, RIN3, FAXDC2 and SIAH2 (supplementary Figure S1). The selection was based on high MFI-signals of the antigen in at least one plasma sample pool.

3. 2. Elution procedure

The development of an elution protocol was performed in three experiments, all constructed in a similar way. Beads without ID were used for the elution process. In each elution experiment, three tests were performed to investigate the efficiency of the elution protocol. First, an aliquot of the beads was analysed in the FlexMap 3D before elution to investigate whether any autoantibodies had bound to the antigens. Similarly, a bead aliquot was analysed after elution to evaluate the amount of antibodies still bound to the antigens. Lastly, eluates were analysed with the SBA to determine if the enrichment procedure was successful. This analysis made it possible to establish the specificity of the eluted autoantibodies to the PrESTs that were used for elution, and whether the autoantibodies retained their ability to recognise the antigen throughout the elution process.

(12)

9

3. 3. Proof of concept

To evaluate the chosen procedure, four antigens and four samples were used in the first elution experiment. The amount of affinity reagents bound to the antigens was measured before and after elution to evaluate the elution process. The two plots are presented in Figure 1. As shown, antibodies were concluded to have released from the antigens to a large extent, leaving only a fraction of measurable affinity reagent still attached.

Figure 1. Determination of amount antibodies attached to the beads (a) before, and (b) after first elution. Based on the decrease in MFI signals, the elution process has enabled antibodies to release from the beads. The antigen RNF185 did not reach the bead count threshold in sample 6 and 7 in measurements pre elution, and no signals were therefore recorded.

Furthermore, eluates were analysed to investigate presence and specificity of any eluted autoantibodies, i.e. if they were able to again bind the same antigen. As shown in Figure 2, the eluted autoantibodies were largely specific to the antigens with which they had been eluted.

However, eluates from sample 8 enriched using RIN3 or FAXDC2 give signals with analysis beads FAXDC2 and RIN3, respectively, in addition to the expected bead (Figure 2b). This indicates cross-reactivity or contamination in sample 8. Analysis of eluates with all beads in the SBA included is shown in supplementary Figure S2.

(13)

10

Figure 2. Results from screening of (a) plasma pools, and (b) eluates from first elution experiment. The samples from which autoantibodies are eluted are represented by different shapes. Antigens used for elution are indicated by colour.

Antigen-coupled beads are separated along the x-axis. As can be observed, antibodies eluted using antigen RIN3 gives signals in FAXDC2-coupled bead in SBA and vice versa in one of the samples.

3. 4. Antibody elution in all samples

64 bead-sample combinations were included in the second elution experiment to evaluate the general applicability of the protocol to different samples and antigens (Figure 3). Comparing the amount of antibodies on the beads before and after elution, the results were similar to that in the first elution experiment. All antibody signal was removed after elution, with the notable of the high-affinity antigen EBNA1. Signals are reported for almost all combinations of samples and antigens. For some wells analysed, bead count was lower than the threshold, resulting in missing data. Interestingly, RNF185 retains some signal after elution (Figure 3b).

(14)

11

Figure 3. Antibody signals of enrichment beads (a) before, and (b) after the second elution experiment. Based on the decrease in MFI signal, the elution strategy has enabled antibodies to release from the beads. Minimum bead count threshold was not reached in some wells, resulting in missing data.

Analysis of eluates with the SBA is presented in Figure 4. Eluted autoantibodies show affinity towards the antigens they were eluted by, indicating preserved specificity. The cross-reactivity of RIN3 and FAXDC2 can still be observed for sample 8. MFI is decreased in eluates compared to plasma samples. Analysis of eluates with all beads in SBA included are presented in supplementary Figure S3.

(15)

12

Figure 4. MFI signals from analysis of autoantibodies present in (a) plasma pools and (b) eluates generated in the second elution experiment. Signals show that eluted antibodies are specific to the antigens they were eluted by and that no cross-reactivity can be detected. Results for control beads EBNA1 and His6ABP is presented in supplementary Figure S4.

3. 5. Evaluation of optimum sample dilution, sample amount and bead amount A test evaluating the effect of three variables on the elution as well as the amount of antibodies yielded was included in the experimental procedure. All 18 combinations of three different sample concentrations, three different sample amounts, and two different bead amounts were included (results presented in Figure 5). An optimum was found at a dilution rate of 1:62.5 and with a bead amount of 5000 beads per µl crude sample. This was implemented in the final elution run, where volumes and amounts were scaled up, to examine whether an increased quantity of plasma sample and antigen-coupled beads would yield a concentration of enriched autoantibodies in the eluates sufficient for further analysis.

(16)

13

Figure 5. Investigation of sample and bead parameters for optimisation of protocol. Sub-plots denote, left to right, 50, 75, and 100 µl diluted sample used, respectively. Rabbit anti-human IgG is used as positive control, and FAXDC2 is the target antigen. Bead number indicates the number of beads used for elution. Autoantibodies were eluted from sample 8 with antigen FAXDC2 with different combinations of parameters. The highest control signals are obtained with a sample dilution of 1:31.25. Interestingly, the target antigen yields lower signals than using the dilution ratio 1:62.5. This ratio demonstrates the highest signals for each combination of sample volume and bead amount. In the target antigen, the higher bead amount (6000 beads) consistently gives higher signals. However, in the control bead, the lower bead amount (3000 beads) sometimes gives higher signals, although the signal difference is then small.

3. 6. Scaling up the elution procedure

The protocol was scaled up for the fourth elution experiment with the aim of yielding enriched autoantibodies with higher concentration. Based on previous results, a total of eight combinations with four samples and four antigens were used. The combinations were motivated with previously shown high MFI signals, which would indicate a greater possibility of high autoantibody concentrations in the eluates. Dilution rate, amount of sample, and bead amount were chosen based on the optimisation experiment.

(17)

14

Comparing MFI signals from the beads pre and post elution (Figure 6), autoantibodies were thought to largely have released from the beads. As seen in Figure 6, samples eluted from beads coupled with antigens RNF185 and EBNA1 still show somewhat elevated signals even post elution, indicating that antibodies still remain attached to the antigens.

Figure 6. Antigen-bound antibodies (a) before and (b) after elution using the scaled-up protocol.Based on the decrease in MFI signals, elution has released antibodies from the beads. However, EBNA1 and RNF185 have not eluted completely. Negative controls included (sample 7 with FAXDC2 and sample 6 with RNF185) gives low signals pre elution.

Comparing the results from the scaled-up elution eluates in Figure 7 to the eluate analysis from the second elution run in Figure 4, a lower elution efficiency was observed compared to previous elution experiments (presented in Figure 2b and Figure 4b). Furthermore, signals were again detected for FAXDC2-eluted antibodies by capture antigen RIN3 in the SBA.

IgG concentrations were measured in NanoDrop. The approximate concentrations are presented in Table 2.

(18)

15

Figure 7. Analysis of eluates from the scaled-up protocol. Signals were clearly detectable for all antibodies eluted.

The two negative controls included in this scaled-up procedure (sample 6 with antigen RNF185 and sample 7 with antigen FAXDC2) did not yield detectable levels of eluted autoantibodies. (a) Enriched antibodies analysed with all capture reagents in SBA except EBNA1. (b) Enriched antibodies analysed with EBNA1 in SBA. The plot was divided to illustrate signals more clearly.

Table 2. Eight combinations of antigens and samples used in the scaled-up fourth elution. Approximate concentrations for the eluted antibodies generated for each of the combinations were measured with NanoDrop. The median concentrations of triplicate measurements are presented.

Antigen Sample Concentration in eluates (mg/ml)

RNF185 1 0.23

RNF185 6 0.09

RNF185 8 0.09

ZNF688 7 0.10

FAXDC2 6 0.09

FAXDC2 7 0.05

FAXDC2 8 0.08

EBNA1 7 0.07

(19)

16

3. 7. Stability evaluation of antigen-coupled beads in SBA

As a part of the presented study, a stability test of PrEST-coupled beads was performed with plasma samples diluted 1:125 to evaluate the stability of the SBA in buffer over time. The correlation of the first screen and the second screen conducted eight weeks subsequent to the first analysis with the SBA is presented in Figure 8. A slight decrease in MFI signals can be observed in the second analysis compared to the first. The cluster found at a signal of approximately 10 000 is generated by the positive control rabbit anti-human IgG. For the semi-positive control EBNA1, signals are high and dispersed.

Figure 8. Correlation of first (denoted MFI 1) and second screen (denoted MFI 2) with crude sample and SBA. A decrease in MFI can be observed for the second screen, indicating a decline in reactivity between samples and antigens.

Each point represents a combination of sample (not indicated) and antigen.

4. Discussion

The aim of the presented project was to develop a protocol for the enrichment of autoantibodies from human plasma using antigen-coupled magnetic microspheres. Selection of antigens was based on a previously determined high reactivity in plasma and serum samples. This was primarily based on research by Neiman et al. [19], as well as in-house unpublished results. As the health status of the used samples was unknown, antigens shown to be reactive in healthy subjects [19]

were prioritised. In addition, antigens with recurring high reactivity in multiple in-house studies were chosen. Antigens with low or no previously observed reactivity were included as negative controls. However, several of the presumed low-reactive antigens captured antibodies from the plasma samples analysed in this study. For instance, FAXDC2, an antigen giving remarkably high signals throughout all experiments had not previously been found to be reactive in healthy subjects.

(20)

17

However, reactivity had been found in a psychiatric disease context. Though autoreactivity towards some antigens is more common than towards others, it is difficult to predict a general occurrence of single antibodies, as seen in our results. The combinations of antigens and sample pools show the diversity of reactivity towards self-antigens that may be found in the human immune system. Neiman et al. [19] have previously reported a wide repertoire of autoantibodies in healthy individuals. High reactivity has been of importance in this project since the aim has been to evaluate whether it is possible to enrich autoantibodies from human plasma samples.

Optimisation of this procedure is facilitated with higher amounts of autoantibodies to elute from the protein fragments.

To evaluate the efficiency of how well the autoantibodies are eluted from the antigens, analyses of MFI signals were made before and after elution procedure in all experiments. Signals were decreased for all beads in every elution, however, some more than others. This may be interpreted as some autoantibodies being more easily eluted than others with the applied elution strategy.

However, it is problematic to compare the success of elution by different antigens. Even though signal intensities provide an indication of eluate concentration, the relationship is not necessarily linear. Many factors affect the relationship of signal and concentration. For instance, affinity constants control on and off rates, which influences elutability.

It is important to consider that antibody clones likely are different in different individuals and thereby may react slightly differently to the elution conditions. Since the blood plasma utilised in the experiments was combined from multiple individuals to form sample pools, the concentration of each autoantibody may be elevated or decreased in samples depending on the autoantibody repertoires of the individuals. In sample pools composed of plasma from more individuals, autoreactivities originating in a single individual are diluted, causing a lower signal for those antigens. Conversely, signals remain constant if autoantibodies with affinity towards the same antigen are present in multiple constituent samples. This provides an explanation to how the reactivity is distributed in different samples. The sample pools 6 and 7 are both assembled from two individual patients respectively. Antibodies from these samples are therefore not very diluted.

These two samples generally generated higher signals for many of the chosen antigens. Since the number of individual patient samples in the other plasma pools is unknown, it is difficult to draw any conclusions about the relation of observed intensities to the number of individual samples in the pools. In addition, autoantibodies in different samples may have affinity towards the same antigens. However, this does not guarantee equal affinity constants, nor can any conclusions be drawn about the characteristics of autoantibodies eluted from the same antigen in a pool of samples until further analysis is made. We cannot exclude the possibility that different autoantibodies found in separate individuals possess affinity towards different epitopes on the same antigen, generating higher signals without necessarily having higher concentrations of specific antibodies. Analyses on this go beyond the extent of the present project but is a very interesting topic for further exploration.

(21)

18

In a successful enrichment procedure, it is imperative that the autoantibodies do not denature during the elution process. Irreversible disruption of IgG structure would the enrichment useless, as it likely would destroy affinity. Therefore, eluents and heat treatments included in the elution process must not damage the antibodies. It has been reported that IgG treated with temperatures higher than 65 °C or pH below 3.0 or above 10.0 risk irreversible denaturation [24]. To avoid the risk of damaging autoantibodies during elution experiments performed in the present study, a temperature of 56 °C was utilised for heat treatment and alterations in pH were set to 3.0 and 10.0.

Even though validation of elution was shown by post-elution analysis of beads, we needed to demonstrate that antibodies also had been enriched and could be re-captured on the same antigen.

The affinity of the eluted autoantibodies is evaluated on the SBA. Observed signals indicate intact antibodies. Since paratopes are sensitive loop domains of the antibody, these could be denatured in harsh conditions. Because strong signals were observed in the eluates, both the paratope and most of the antibody backbone were likely intact, as both regions are required for signal generation in the assay. However, it is still possible that some antibodies were affected by the elution process.

The observed signals did decrease in eluates compared to crude samples. This is expected, as the enrichment includes multiple washes and transfer, where antibodies may be lost. Furthermore, concentrations of autoantibodies in eluates may be lower than in crude samples. By further analysing the eluates, we could show that antibodies were present in the majority of the eluates, and that they generally remain specific to the antigens used for elution. However, as shown in Figure 2, antibodies enriched using the antigen RIN3 gives a signal for FAXDC2-bead using the SBA. The opposite is also true, which could indicate contamination or cross-reactivity. No significant sequential similarities explaining the cross-reactivity were found between the two antigens using BLAST [25]. However, epitope folding or other similarities between highly reactive PrEST cannot be excluded as a reason for cross-reactivity. This would need to be examined with epitope mapping.

From optimisation test, it could be determined that a sample dilution of 1:62.5 with 75 µl diluted sample and 6000 beads per well (i.e. 5000 beads per µl crude sample) gave the highest signals for the eluted target autoantibodies. As results for 50 µl diluted sample were similar to results for 75 µl diluted sample, the bead to sample ratio was calculated using the 75 µl results, as this allowed the use of fewer beads in relation to sample amount. The relatively low signals expressed in the less diluted samples could be explained by that an excess of plasma in relation to bead amount could result in unspecific interaction within the samples and aggregation. In addition, an excess of beads in relation to applied plasma suggests that fewer autoantibodies would bind to the antigen- coupled beads and lower signals are thus generated.

Approximate concentrations on the eluted autoantibodies indicated amounts of antibody available for further research. Concentrations of one enriched antibody generated somewhat conflicting results. Autoantibodies eluted from sample 1 by antigen RNF185 indicated a concentration of 0.23

(22)

19

mg/ml after concentration determination. MFI signals from eluate analysis from third elution with the SBA (shown in Figure 6) can be observed at around 5000. For many of the other sample- antigen combinations, signal intensities are considerably higher but with lower measured concentrations in concentration measurement. The low signals generated for the eluates in the SBA analysis may be an indication of low affinity towards the antigen, but concentration of autoantibodies and epitope accessibility are also contributing factors. The availability for antibodies to bind to the epitope is dependent on how the antigen is folded, but also the amount of available epitopes on each bead, i.e. how many of the antigens that successfully have attached to the beads. Furthermore, if the antigen contains multiple primary amines it may bind to numerous spots on the bead, affecting the epitope accessibility. This could provide an explanation to the fact that signals are low while the suspended antibody concentration may still be high. However, analysis of signals generated from the beads from which autoantibodies were eluted indicate a less successful elution compared to other combinations (see Figure 5). Conclusions from the low amount of antibodies eluted from the beads could imply strong interaction with the antigen or could perhaps also be an indication of a lower concentration of antibodies in the eluate. The conclusion regarding low affinity would thus be contradicted.

Furthermore, antibodies eluted from samples 6 and 8 using FAXDC2-coupled beads show high signals as well as a distinct decrease in observed intensity on beads post compared to pre elution, indicating a high degree of eluted IgG. The concentration determination resulted in 0.09 mg/ml and 0.08 mg/ml for the two samples respectively. These results were further compared with tested eluates on the negative control sample 7 from FAXDC2-beads. Potential eluted antibodies from sample 7 with FAXDC2 present considerably lower signals than eluates from other samples with the same antigen. However, the approximate IgG concentration eluted from sample 7 is measured to 0.05 mg/ml. Since elution from sample 7 with FAXDC2-coupled beads is implemented as a negative control as established in prior experiments, the approximate autoantibody concentration was assumed to be lower than that of other combinations of samples and antigens. Furthermore, measurements by NanoDrop provide an indication of protein concentration and return unstable results at low levels. Conclusions can be made that further analysis is required to explain this result.

The enrichment method developed in the present study has been successful in enriching autoantibodies from plasma samples. However, it cannot be described as a fully general or optimised method for enrichment. A total of 64 combinations were applied in the elution strategy.

Of these, only a few clearly demonstrated elution.

(23)

20

5. Future perspectives

A stable procedure for enrichment of autoantibodies from human blood samples would be valuable to gain deeper knowledge of circulating autoantibodies and antigen interaction. To verify that autoantibodies, and not some other affinity complex, have been enriched using the herein developed procedure, Western blot, SDS-PAGE, or Mass Spectrometry could be used. In SDS- PAGE (sodium dodecyl sulfate–polyacrylamide gel electrophoresis) and Western blot, the protein size can be determined, indicating or dismissing the presence of IgG. Additionally, in Western blot, the affinity or species of antibody could be tested.

Epitope mapping is sometimes performed on antibodies. In autoimmune disorders, this may be of medical importance. With a stable and general enrichment procedure, isolated autoantibodies may be investigated to analyse the specific affinity towards epitopes. This could improve reliability and reduce confounding in epitope mapping, which may be of importance in autoimmune disorders.

Immunohistochemistry (IHC) analysis and results could be greatly improved using enriched autoantibodies rather than whole antisera, which is normally used. Using antisera, many autoantibodies are present, which may result in undesired antigen binding. The risk of ambiguous signals might be reduced using enriched autoantibodies. Signals could be considered more precise, and higher reliability of the spatial presence of antigens in tissue may thus be attained.

Concentration measurements of eluates in the scaled-up elution procedure indicated antibody concentrations of 0.05 to 0.23 mg eluted antibody per ml eluate, which lies well within the range of antibody applications. For instance, antibody concentrations of 0.5 µg/ml are needed for application in IHC [26]. This implies that concentrations of enriched autoantibodies from the procedure applied in this project by far is enough for further research and applications.

Acknowledgements

I would like to express my gratitude to Peter Nilsson for providing me with the opportunity to do my Master’s thesis at his lab at Science for Life Laboratory. Thank you for believing in me.

Thank you Anna Månberg, for your support and for always being excited about the progress in this project. A big thanks to August Jernbom-Falk, for your patience and dedication for teaching me everything I could possibly need to know to be able to perform this project. Your help has been invaluable. Finally, I would like to acknowledge everyone in PAPP for your unbelievable kindness and generosity. You are always so happy to help, and that has been indescribably appreciated over the course of these 20 weeks. Thank you all.

(24)

21

References

1. Lacroix-Desmazes S, Kaveri SV, Mouthon L, Ayouba A, Malanchère E, Coutinho A, et al. Self- reactive antibodies (natural autoantibodies) in healthy individuals. J Immunol Methods. 1998 Jul 1;216(1):117–37.

2. Lleo A, Invernizzi P, Gao B, Podda M, Gershwin ME. Definition of human autoimmunity — autoantibodies versus autoimmune disease. Autoimmun Rev. 2010 Mar 1;9(5):A259–66.

3. Arnaout R, Lee W, Cahill P, Honan T, Sparrow T, Weiand M, et al. High-Resolution Description of Antibody Heavy-Chain Repertoires in Humans. Reindl M, editor. PLoS ONE. 2011 Aug

4;6(8):e22365.

4. Stanfield RL, Wilson IA. Antibody Structure. Microbiol Spectr [Internet]. 2014 Mar 21 [cited 2020 Apr 3];2(2). Available from:

https://www.asmscience.org/content/journal/microbiolspec/10.1128/microbiolspec.AID-0012-2013 5. Charles A Janeway J, Travers P, Walport M, Shlomchik MJ. The interaction of the antibody

molecule with specific antigen. Immunobiol Immune Syst Health Dis 5th Ed [Internet]. 2001 [cited 2020 May 25]; Available from: https://www.ncbi.nlm.nih.gov/books/NBK27160/

6. Wang L, Wang F-S, Gershwin ME. Human autoimmune diseases: a comprehensive update. J Intern Med. 2015;278(4):369–95.

7. Elkon K, Casali P. Nature and functions of autoantibodies. Nat Clin Pract Rheumatol. 2008 Sep;4(9):491–8.

8. Leslie D, Lipsky P, Notkins AL. Autoantibodies as predictors of disease. J Clin Invest. 2001 Nov 15;108(10):1417–22.

9. Ayoglu B, Schwenk JM, Nilsson P. Antigen arrays for profiling autoantibody repertoires.

Bioanalysis. 2016 Apr 21;8(10):1105–26.

10. Jonsson R, Theander E, Sjöström B, Brokstad K, Henriksson G. Autoantibodies Present Before Symptom Onset in Primary Sjögren Syndrome. JAMA. 2013 Nov 6;310(17):1854–5.

11. Eriksson C, Kokkonen H, Johansson M, Hallmans G, Wadell G, Rantapää-Dahlqvist S.

Autoantibodies predate the onset of systemic lupus erythematosus in northern Sweden. Arthritis Res Ther. 2011;13(1):R30.

12. Avrameas S, Ternynck T. Natural autoantibodies: The other side of the immune system. Res Immunol. 1995 May 1;146(4):235–48.

13. Wildbaum G, Nahir MA, Karin N. Beneficial Autoimmunity to Proinflammatory Mediators Restrains the Consequences of Self-Destructive Immunity. Immunity. 2003 Nov 1;19(5):679–88.

14. Zandian A, Forsström B, Häggmark-Månberg A, Schwenk JM, Uhlén M, Nilsson P, et al. Whole- Proteome Peptide Microarrays for Profiling Autoantibody Repertoires within Multiple Sclerosis and Narcolepsy. J Proteome Res. 2017 Mar 3;16(3):1300–14.

(25)

22

15. Just D. On the profiling of autoantibodies in psychiatric disorders. Stockholm: KTH Royal Institute of Sweden; 2019.

16. Agaton C, Falk R, Höidén Guthenberg I, Göstring L, Uhlén M, Hober S. Selective enrichment of monospecific polyclonal antibodies for antibody-based proteomics efforts. J Chromatogr A. 2004 Jul 16;1043(1):33–40.

17. Ayoglu B, Birgersson E, Mezger A, Nilsson M, Uhlén M, Nilsson P, et al. Multiplexed protein profiling by sequential affinity capture. Proteomics. 2016 Apr;16(8):1251–6.

18. Birgersson E, Schwenk JM, Ayoglu B. Bead-Based and Multiplexed Immunoassays for Protein Profiling via Sequential Affinity Capture. Methods Mol Biol Clifton NJ. 2017;1619:45–54.

19. Neiman M, Hellström C, Just D, Mattsson C, Fagerberg L, Schuppe-Koistinen I, et al. Individual and stable autoantibody repertoires in healthy individuals. Autoimmunity. 2019;52(1):1–11.

20. R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.

21. Wickham et al. Welcome to the tidyverse. Journal of Open Source Software. 2019;4(43):1686.

22. Wickham H, Chang W, Henry L, Pedersen TL, Takahashi K, Wilke C, et al. ggplot2: Create Elegant Data Visualisations Using the Grammar of Graphics [Internet]. Available from: https://CRAN.R- project.org/package=ggplot2

23. Henry L, Wickham H, RStudio. rlang: Functions for Base Types and Core R and ‘Tidyverse’

Features [Internet]. 2020. Available from: https://CRAN.R-project.org/package=rlang 24. Indyk HE, Williams JW, Patel HA. Analysis of denaturation of bovine IgG by heat and high

pressure using an optical biosensor. Int Dairy J. 2008 Apr 1;18(4):359–66.

25. BLAST: Basic Local Alignment Search Tool [Internet]. [cited 2020 May 22]. Available from:

https://blast.ncbi.nlm.nih.gov/Blast.cgi

26. Antibody dilutions and titer | Abcam [Internet]. [cited 2020 May 24]. Available from:

https://www.abcam.com/protocols/antibody-dilutions-and-titer

(26)

23

Supplementary information

Figure S1. Result of plasma sample pools diluted 1:125 with suspension bead array and analysed with the standard autoantibody profiling protocol. Output signals were used for determination of antigens for elution procedure.

(27)

24

Figure S2. Analysis of eluted autoantibodies from first elution with all beads included from suspension bead array.

For SBA analysis of plasma samples with same dilution as used for this elution (1:125), see Figure S1.

(28)

25

Figure S3. Analysis of (a) plasma samples, and (b) eluates from second elution experiment with suspension bead array. All capture antigens in the SBA and their signals are displayed. In both experiments, samples were diluted 1:62.5.

(29)

26

Figure S4. MFI signals generated from analysis of eluates from second elution with suspension bead array.

Autoantibodies eluted by semi-positive control EBNA1 display high signals, while no autoantibodies eluted by negative control His6ABP can be detected in the eluates.

(30)

www.kth.se