Human blood derivatives

Surplus blood from healthy blood donor donations was obtained from the Blood Bank of the Department of Clinical Immunology, Aarhus University Hospital, Denmark.

Plasma pools for affinity isolation of isohemagglutinins

Whole blood from healthy blood donors was collected in tubes containing ethylenediaminetetraacetic acid (EDTA) and centrifuged at 2,000 × g for 10 min to separate plasma. Four 100 mL plasma pools were prepared, each containing 1 mL of plasma from each of 100 different donors. One pool consisted of plasma from ABO type A donors, containing anti-B isohemagglutinins; a second pool consisted of plasma from ABO type B donors, containing anti-A isohemagglutinins; the two remaining pools contained plasma from ABO type O donors, who produce both anti-A and anti-B isohemagglutinins. These were used separately for the isolation of anti-A and anti-B isohemagglutinins, respectively.

Each 100 mL plasma pool was diluted with 400 mL Tris-buffered saline (TBS; 10 mM TRIS, 140 mM NaCl, 0.1% (v/v) NaN3, pH 7.4) containing Tween-20 (TBS/Tw, 0.05% (v/v) Tween-20) to a final volume of 500 mL. The diluted plasma was centrifuged at 10,000 × g for 10 min, and the resulting supernatants were filtered through gauze to produce the starting materials for isohemagglutinin affinity purification.

Preparation of fixed Red Blood Cells (RBCs)

EDTA-stabilized donor blood of ABO type A or B was pooled separately for each blood type, with each pool containing 1 mL of blood from ten individuals. The following procedures were performed at ambient temperature, with centrifugation at 200 × g for 5 min between each step.

1.

Washing: Pooled blood (10 mL) was diluted with 40 mL phosphate-buffered saline (PBS, pH 7.4) and washed twice by centrifugation.

2.

Fixation: The cell pellets were resuspended in 50 mL PBS containing 2% (w/v) glucose and 0.25% (v/v) glutaraldehyde, followed by end-over-end rotation for 30 min.

3.

Quenching: After centrifugation, the cells were resuspended in 50 mL PBS, then pelleted and resuspended in 50 mL PBS containing 50 mM ethanolamine, followed by end-over-end rotation for 15 min.

4.

Final washes: The cells were pelleted by centrifugation and washed twice in 50 mL TBS.

5.

Resuspension: The final cell suspension was adjusted to 5 mL in TBS with human serum albumin (TBS/HSA, 1 g/L).

The hematocrit, i.e., the percentage of the sample volume occupied by erythrocytes, was measured using a Sysmex XT-1800i (Sysmex Corporation, Japan).

Affinity purification of isohemagglutinins

The starting materials (diluted plasma pools) were processed using columns containing 20 mL of beads conjugated with the blood group A antigen (A column) or blood group B antigen (B column) (Glycosorb®-ABO, Glycorex Transplantation AB, Sweden). The conjugated ABO antigens included a mixture of tri-, tetra-, and longer saccharides. The purification was performed at ambient temperature with a flow rate of approximately 40 mL per hour.

After loading the plasma, the columns were washed with 100 mL TBS/Tw, followed by an additional 200 mL TBS to remove unbound components. Isohemagglutinins were eluted using 0.1 M glycine buffer (pH 2.5) in 1 mL fractions. Eluates were collected into Tween-blocked tubes containing 85 µL of 1 M Tris-HCl buffer (pH 8.5) to neutralize the pH in the sample.

For each purification run, the following samples were collected and stored: the starting material, the entire flow-through (identified by its yellowish color, in 10 batches of 50 mL), the first 100 mL of the wash-out (in two 50-mL batches), and 40 sequential elution fractions.

TRIFMAs (Time-Resolved ImmunoFluorometric Assays; solid-phase immunoassays)

The following antigens were used for coating of plates: HSA (CSL Behring, King of Prussia, PA, USA), blood group A antigen conjugate (Blood group A-HSA, NGP9305, Dextra Laboratories, Reading, UK), blood group B antigen conjugate (Blood group B-HSA, NGP9323, Dextra Laboratories), mannan (prepared as described by Nakajima and Ballou [23]; selected as a high-mannose structure to control for contaminating anti-polysaccharide antibodies), and tetanus toxoid (TT, product number 2674, Statens Serum Institut, Copenhagen, Denmark; selected as a control for contaminating anti-protein antibodies; anti-TT antibodies are anticipated to be abundant in the plasma pools due to widespread vaccination). For coating, the antigens were diluted in carbonate buffer (0.1 M, pH 9.4) to final concentrations of 1 µg/mL for the blood group A and B antigen conjugates and mannan, and 0.01% (v/v) for TT. Antigen solutions (100 µL per well) were applied to microtiter plates (FluoroNunc, Thermo Scientific Nunc A/S, Denmark) and incubated overnight at 4 °C for coating. After incubation, wells were emptied and blocked with 200 µL of TBS/Tw containing 1 g/L HSA (TBS/Tw/HSA) for 1 h at ambient temperature. Wells were then washed three times with 250 µL of TBS/Tw using an automated plate washer. Samples diluted in TBS/Tw/HSA containing 10 mM EDTA were added to the wells (100 µL per well, in duplicates) and incubated overnight at 4 °C. Subsequent steps were performed at ambient temperature. After washing, wells were incubated for 1 h with biotinylated secondary antibodies (rabbit polyclonal anti-human IgA [Dako, Denmark], IgG [Dako], or IgM [Rockland, PA, USA]) diluted to 1 mg/L in TBS/Tw/HSA.

After washing, 100 µL of europium-streptavidin complex (product 1244 − 360, PerkinElmer, MA, USA, 0.1% [v/v] in TBS/Tw containing 25 µM EDTA) was added to each well and incubated for 1 h. Wells were washed again before adding 200 µL of enhancement solution (PerkinElmer). Plates were shaken, and time-resolved fluorescence was measured after 5 min using a plate reader (VICTOR X5, PerkinElmer).

Data analysis was conducted as previously described [24]. Briefly, signals attributable to the carbohydrate moieties of glycoconjugates were estimated by subtracting signals obtained on an HSA-coated surface from those on a glycoconjugate-coated surface using the same sample. Antibody reactivity was quantified after log10 transformation of all data, with standard curves approximated by third-degree polynomials in Microsoft Excel 2016 (Microsoft Corporation, WA, USA). Data was retransformed to a linear scale for presentation.

SDS-PAGE and western blottingSDS-PAGE analysis

The concentration of affinity-purified isohemagglutinins, along with normal human (nh) IgA (I4036, Sigma-Aldrich, St. Louis, MO, USA), nhIgG (Privigen, CSL Behring, King of Prussia, PA, USA), and nhIgM (I8260, Sigma-Aldrich, St. Louis, MO, USA), was determined by spectrophotometry (NanoDrop 1000, Thermo Fisher Scientific, Waltham, MA, USA) and adjusted with TBS/Tw. Samples were then mixed with loading buffer (final concentrations: 6.5% (v/v) glycerol, 1.96% (w/v) SDS, 40 mM Tris, 5.25 M urea, 6.5 ppm (w/v) bromophenol blue, pH 6.7) and dithiothreitol (60 mM). The mixtures were boiled for 3 min. Following denaturation, samples and a protein ladder marker (Precision Plus Protein™ All Blue Prestained Protein Standards, Bio-Rad, Hercules, CA, USA) were loaded onto 4–15% gradient gels (Criterion™ TGX™, Bio-Rad, cat#5671084) and subjected to SDS-PAGE. Electrophoresis was performed at 80 V until the dye front exited the stacking gel, followed by 150 V for approximately 1 h.

Coomassie staining

Gels were stained with Coomassie Brilliant Blue by heating in a microwave oven for 5 min. Subsequently, gels were destained in a solution containing 9.6% (v/v) ethanol and 7.5% (v/v) acetic acid prepared in demineralized water.

Western blotting

Proteins were transferred onto nitrocellulose membranes using the Trans-Blot® Turbo™ Transfer System (Bio-Rad) with the 7-minute transfer protocol provided by the manufacturer. Membranes were blocked in TBS with 0.1% Tween-20 for 45 min and then incubated overnight with gentle rocking in TBS/Tw/HSA containing 1 mM EDTA and biotinylated secondary antibody. The secondary antibodies against hIgG, hIgA, and hIgM were those described previously for the TRIFMAs. Biotinylated normal rabbit immunoglobulin was included as a control for nonspecific binding of biotinylated antibodies. The detection antibodies were used at a final concentration of 1 mg/L. After six washes in TBS/Tw (without azide), membranes were incubated with streptavidin-HRP conjugate (Dako, Denmark) diluted 1:3,000 in TBS/Tw (without azide) containing 1 mM EDTA and nhIgG at 0.1 g/L for 1.5 h. Signal detection was performed using SuperSignal™ West Pico chemiluminescence substrate (Thermo Fisher Scientific, Waltham, MA, USA), and images were acquired on an ImageQuant LAS 400 mini (GE Healthcare, Chicago, IL, USA).

Bacterial strains and preparation

S. pneumoniae strains were obtained from the Kilian Collection, the bacterial culture collection at the Department of Biomedicine, Aarhus University, Denmark. A total of 30 serotypes were included: 1, 3, 4, 5, 6 A, 6B, 6 C, 7 F, 8, 9 N, 9 V, 10 A, 11 A, 12 F, 14, 15 A, 15B, 16 F, 18 C, 19 F, 19 A, 20, 22 F, 23 F, 23 A, 24 F, 31, 33 F, 35 F, and 38. Additionally, the unencapsulated strain CSR SCS-2 clone I, referred to as the “C-mutant” [25, 26], was included.

Individual bacterial strains were cultured overnight at 35 °C in 50 mL Todd-Hewitt broth in a 5% CO₂ heating cabinet. Cultures were harvested by centrifugation (2,000 × g, 30 min), and bacterial pellets were resuspended in PBS containing 1% (v/v) formaldehyde for fixation. After overnight incubation, the fixed bacteria were washed twice by centrifugation in 50 mL PBS, followed by two additional washes in 50 mL TBS to neutralize residual aldehyde groups. The final bacterial pellets were resuspended in TBS and stored at 4 °C until use.

Flow cytometry

Isohemagglutinin reactivity was assessed by flow cytometry (standard configuration NovoCyte 3000, ACEA Biosciences, CA, USA). Fixed bacterial suspensions were adjusted in TBS/HSA to achieve approximately 15,000 bacterial events per µL, as determined by flow cytometry. A forward-scatter threshold was set to ensure that background events in TBS/HSA alone remained below 50 events per µL. Each recorded flow cytometry event corresponded to an average of 2.7 bacteria, based on linear regression analysis of the relationship between optical density (OD) measurements of pneumococcal suspensions (where OD 1.0 corresponds to 1.8 × 10⁹ bacteria/mL [27]) and the measured event count per µL (R² = 0.73).

Fixed bacterial suspensions diluted in TBS/HSA were incubated with an equal volume of PBS/HSA containing one of the following reagents:

No added antibodies (buffer control).

nhIgG, 200 mg/L (positive control).

Recombinant humanized monoclonal IgG antibody against complement C5 (eculizumab, 20 mg/L; negative control).

Anti-A isohemagglutinin (αA-IH, 10 mg/L).

Anti-B isohemagglutinin (αB-IH, 20 mg/L).

Anti-A,B isohemagglutinin (αA,B-IH, 10 mg/L).

IgG anti-αGal antibody (affinity-purified as previously described [28], 5.0 mg/L; included as a reference for previous findings)

Samples were incubated at 37 °C for 60 min, followed by two washes in 1 mL PBS/HSA (2000 × g, 10 min). The bacterial pellets were resuspended in 20 µL PBS/HSA containing 0.33% (v/v) fluorescein-isothiocyanate (FITC)-coupled polyclonal rabbit F(ab’)₂ anti-human IgG (F0315, DAKO, Denmark) and incubated in the dark at room temperature for 30 min. After incubation, 80 µL PBS was added, mixed, and 20 µL of each sample was analyzed by flow cytometry at a flow rate of 35 µL/min. Forward-scatter, side-scatter, and FITC fluorescence (excitation at 488 nm, emission detected through a 530/30 nm bandpass filter) were recorded.

To account for potential differences in bacterial background fluorescence, the median fluorescence intensity (MFI) of F(ab’)₂ anti-human IgG binding for each strain and antibody condition was normalized to the MFI of the same strain incubated without primary antibody. This ratio, MFInorm, has a theoretical value of 1.0 in the absence of primary antibody binding.

Inhibition experiments

To evaluate the inhibition of isohemagglutinin binding to solid-phase ABO antigens by RBCs, isohemagglutinin preparations were diluted in sample buffer (αA-IH, 1:10,000; αB-IH, 1:7,500) containing glutaraldehyde-fixed type A or B RBCs at varying hematocrit levels. The mixtures were incubated for 30 min at ambient temperature, followed by 4 h at 4 °C with end-over-end rotation. After incubation, samples were centrifuged at 200 × g for 5 min, and the supernatants were analyzed for residual reactivity with blood group A or B antigens using TRIFMA.

For inhibition experiments analyzed by flow cytometry, the isohemagglutinin concentrations were adjusted for each bacterial strain to achieve an MFInorm of approximately 5 in the absence of an inhibitor. As soluble inhibitors, we used pneumococcal capsule polysaccharides of vaccine quality from serotypes 6B, 7 F, and 9 V (LCG Standards, Teddington, UK), as well as pneumococcal Cell Wall Polysaccharide, CWP (Statens Serum Institut, Copenhagen, Denmark). Soluble polysaccharide inhibitors were incubated with isohemagglutinins for one hour at 37 °C before being added to fixed bacterial cells. The subsequent handling and flow cytometry analysis proceeded as described above.

The percentage of inhibition was determined using standard curves generated from serial dilutions of the tested isohemagglutinin preparation, analyzed alongside the inhibition experiments for each bacterial strain. Quantitative calculations were performed as described for the TRIFMA assay.

Cohort establishment and data sources

To investigate any association between ABO blood group and invasive pneumococcal disease (IPD), we established a nationwide cohort comprising the entire Danish population (approximately 6 million inhabitants as of 2025). ABO blood group data together with RhD antigen status (included as a control) were obtained from all Danish blood bank computer systems, where each record was linked to the unique personal identification number (CPR) assigned to all Danish residents since 1968. Duplicate entries were removed, and only individuals with a unique CPR number were included in the final cohort.

Data on IPD were obtained from the Statens Serum Institut (Copenhagen, Denmark), which performs nationwide surveillance of IPD and serotyping of clinical isolates. The dataset spanned the period from 1966 to June 26, 2014; however, only IPD cases with a documented CPR number were included in the analysis. Furthermore, IPD events occurring in individuals under 2 years of age were excluded, as anti-A and anti-B antibodies are typically not expected to be fully formed in young children. The IPD and ABO blood group datasets were subsequently merged via the CPR identifier for downstream analyses.

Statistics

Effect sizes were estimated with 95% confidence intervals (CIs) using bootstrapping with 5,000 iterations, as implemented in Estimation Statistics [29].

Data on ABO blood group types and IPD events were imported and joined in an Oracle SQL database, using the CPR identifier as the primary key. The number of IPD cases for specific pneumococcal serotypes was stratified by ABO blood group type using STATA version 12 (StataCorp, College Station, TX, USA).

GraphPad prism version 10 for Windows (GraphPad Software, Boston, MA, USA) was used for regression analyses, curve fitting (including Hill-slope calculations), odds ratio estimations, and figure preparations.

The standard deviation (\( _\)) of a product (\( z=x\cdot y\)) was calculated using the error propagation formula for multiplication:

$$ _=z\cdot \sqrt_}\right)}^+_}\right)}^}$$

where \( _\)​ and \( _\)​ are the standard deviations of \( x\) and \( y\), respectively.

Ethics

Anonymized blood samples used in this study consisted of surplus material from voluntary blood donations collected at the blood bank of Aarhus University Hospital. The use of these samples complied with current Danish legislation. Approval for the registry study was obtained from the Danish Data Protection Agency (Ref. no. 2014-41-3065).