Healthy donor serum samples were collected as part of a previous study from anonymous, healthy volunteers with informed, written consent in accordance with Dutch regulations and this study was approved by the Sanquin Ethical Advisory Board in accordance with the Declaration of Helsinki (6). Normal human serum (NHS) pool comprises serum from 400 healthy donors. FHR-3-deficient serum pool comprises serum of four healthy donors previously genotyped by multiplex ligation-dependent probe amplification to carry no CFHR3 gene copies (6). Samples of patients with confirmed bacterial infections were collected as part of the EUCLIDS project (van Beek et al., manuscript in preparation). CRP levels were determined in these samples as part of routine testing.

Rat anti-mouse kappa (RM-19) mAb was from Sanquin Reagents (Sanquin, Amsterdam, the Netherlands). High-performance ELISA buffer (HPE) was provided by Sanquin. Proteins were biotinylated according to the manufacturer’s instructions using EZ-Link Sulfo-NHS-LC-Biotin, No-Weigh Format (Thermo Scientific, Waltham, MA, USA), when indicated. Polyclonal anti-FHR-3 antibodies and mAb clone anti-FHR-3.3 were obtained and characterized as part of a previous study (6). Recombinant human (rh) FHR proteins, containing a C-terminal 6×-histidine tag, were produced and purified as previously described (6). In short, proteins were expressed by transient transfection of pcDNA3.1 expression vectors in HEK293F cells, after which proteins were purified from the supernatant by Ni²⁺ affinity chromatography using HisTrap™ High Performance 1 mL columns (GE Healthcare Life Sciences, Freiburg, Germany). rhFHRs were filtered and concentrated using Amicon^® Ultra Centrifugal Filter Devices (Merck Millipore, Darmstadt, Germany) according to the manufacturer’s instructions using appropriate molecular weight cut-offs. For rhFHR-4A, a 100 kDa cut-off Amicon^® Filter was used to further purify rhFHR-4A from any high molecular weight aggregates observed after Ni²⁺ affinity chromatography purification (Figures S1A,B in Supplemental Material).

Anti-FHR-4A mAbs were generated, screened, and purified as previously described, using rhFHR-4A as immunogen (6). Isotypes of mAbs were determined by ELISA or with the use of the Mouse mAb Isotyping Kit (Hycult Biotech, Uden, the Netherlands) according to the manufacturer’s instructions. Cross-reactivity against other FH protein family members was determined as previously described (6).

The epitope location of each of the anti-FHR-4A mAbs was determined using rhFHR-4A, rhFHR-4B, and rhFHR-4A fragments consisting of SCR domains 1–3, 2–4, 4–9, 5–7, or 8–9 as previously described (15). Proteins and fragments were coated (4 µg/mL in PBS) onto Nunc Maxisorp 96-well microtiter plates (Invitrogen, Life Technologies, Carlsbad, CA, USA) by overnight (O/N) incubation at 4°C. Plates were washed three times with PBS + 0.02% (w/v) Tween-20 (PT) and blocked with 4% (w/v) BSA in PBS by incubation for 1 h. After another wash with PT, wells were incubated with the anti-FHR-4A mAbs diluted in PT for 1 h. Unbound anti-FHR-4A mAbs were washed away by washing the wells five times with PT, followed by incubating with 0.05% (v/v) HRP-conjugated rabbit anti-mouse IgG (Dako Agilent, Santa Clara, CA, USA) for 1 h and subsequently washing the plates five times. The ELISA was developed using 3,5,3′,5′-tetramethylbenzidine (TMB) solution and the reaction was stopped using 50 µL 2 M H₂SO₄. As a positive control, FHR-4A fragments were detected using polyclonal rabbit anti-FHR-4A, followed by HRP-conjugated goat anti-rabbit IgG (Dako Agilent). All ELISA steps were performed with a final volume of 50 µL per well and incubated at room temperature while shaking unless stated otherwise. Absorption was measured at 450 nm and corrected for background absorbance at 620 nm.

To determine whether the mAbs competed for the binding of rhFHR-4A, Nunc Maxisorp 96-well microtiter plates (Invitrogen) were coated with 100 µL of 2 µg/mL anti-FHR-4A mAbs, in PBS, by incubating O/N at room temperature. Next, the plates were washed with PT. Biotinylated rhFHR-4A (0.1 µg/mL, in HPE) was incubated with 10 µg/mL of each mAb for 20 min, followed by incubation on the washed plate for 1 h. Next, unbound biotinylated rhFHR-4A was washed away and the wells were incubated with 0.01% (v/v) strep-poly-HRP (Sanquin), diluted in HPE, for 20 min. After washing, the assay was developed by addition of 100 µL of 100 µg/mL TMB in 0.11 M sodium acetate containing 0.003% (v/v) H₂O₂, pH 5.5. Substrate conversion was stopped after approximately 10 min by addition of 100 µL 2 M H₂SO₄. Absorbance was measured at 450 nm and corrected for the absorbance at 540 nm with a Synergy 2 Multi-Mode plate reader (BioTek Instruments, Winooski, VT, USA). All ELISA steps were performed with a volume of 100 µL per well and incubated at room temperature while shaking unless stated otherwise.

Sucrose gradients were described previously (18). In short, NHS or FHR-3-deficient serum (150 µL of 50%, v/v, pooled serum, diluted in PBS) were loaded on 5–32.9% (w/v) sucrose (Merck 1.07654) gradients. Gradients were centrifuged for 20 h at 160,000 × g after which they were fractionated in 24 fractions of 500 µL. Proteins were immunoprecipitated and visualized on Western Blots as described below.

Factor H-related-4A and FHR-4B were immunoprecipitated from human healthy donor serum or sucrose gradient fractions using indicated mAbs. IP was performed by incubating 200 µL serum or 250 µL sucrose gradient fraction with 500 µL of 5 mg/mL CNBr-activated sepharose (GE Healthcare, Little Chalfont, UK) to which RM-19 was coupled (25 mg mAb per 1 g sepharose), and 50 µL of 100 µg/mL mAb, diluted in PBS supplemented with 0.1% Tween-20, 0.1% BSA, and 10 mM EDTA. Following overnight incubation at 4°C, while rotating, the sepharose was washed three times with 1 mL PT and two times with 1 mL PBS. Precipitated proteins were eluted by addition of 50 µL 1× NuPAGE Sample buffer solution (Invitrogen) and incubation at 70°C for 10 min. After spinning down the sepharose, SDS-PAGE under non-reducing conditions was performed using a Novex NuPAGE 10 or 4–12% Bis–Tris gel followed by Western Blot onto a nitrocellulose membrane (Novex iBlot Gel Transfer kit, Invitrogen). Membranes were blocked with 1% (v/v) Western Blocking Reagent (WBR) (Roche, Basel, Switzerland) in PBS for 30 min and incubated with 1 µg/mL biotinylated polyclonal rabbit anti-FHR-3 in PBS + 0.5% (v/v) WBR, O/N. After washing three times with PT, membranes were incubated with 0.1% (v/v) Strep-HRP in PBS + 0.5% (v/v) WBR. After 1 h, the membranes were washed three times with PT followed by two washes with PBS. Western Blots were developed with the Pierce ECL 2 Western Blotting substrate kit (Thermo Scientific) according to the manufacturer’s instructions and analyzed using the ChemiDoc™ MP System (BioRad, Hercules, CA, USA).

To measure FHR-4A in serum, RM-19 was coated (3 µg/mL in PBS) onto Nunc Maxisorp 96-well microtiter plates (Invitrogen) by O/N incubation at room temperature. After coating, plates were washed five times with PT, followed by incubation with 1 µg/mL anti-FHR-4A.04 in HPE, for 1 h. After washing, samples, diluted in HPE, were added and incubated for 1 h. Following washing with PT, 0.5 µg/mL biotinylated polyclonal anti-FHR-3 (in HPE) was incubated on the plate for 1 h. Next, unbound conjugate was washed away and the wells were incubated with 0.01% (v/v) strep-poly-HRP (Sanquin), diluted in HPE, for 20 min. After washing, the assay was developed as described above using TMB and measuring absorbance at 450 nm and correcting for the absorbance at 540 nm. All ELISA steps were performed with a volume of 100 µL per well and incubated at room temperature while shaking unless stated otherwise. For the calibration of the FHR-4A ELISA, highly pure rhFHR-4A was used (Figure S1B in Supplemental Material) of which the concentration was determined by measuring the absorbance at 280 nm and using an absorbance coefficient of 2.134 (0.1% w/v solution).

Analysis and statistical tests were performed using GraphPad Prism, version 7.03 (GraphPad Software, La Jolla, CA, USA).

We characterized 13 mouse mAbs raised against rhFHR-4A. All mAbs were first tested for cross-reactivity against all other members of the FH protein family (Figure 1A). The mAbs were named in order of increasing cross-reactivity: anti-FHR-4A.01 to 4A.06 being mono-specific for rhFHR-4A, anti-FHR-4A.07, and 4A.08 recognizing both rhFHR-4A and -4B, anti-FHR-4A.09 to 4A.11 recognizing rhFHR-3, 4A, and -4B, anti-FHR-4A.12 binding rhFHR-2, -3, -4A, and -4B, whereas anti-FHR-4A.13 recognizes all rhFHRs except rhFHR-5. None of the anti-FHR-4 mAbs showed cross-reactivity with either FHR-5 or FH.

With the use of recombinant fragments comprising different FHR-4A domains (Figure 1B), the epitope location of almost all anti-FHR-4A mAbs were mapped. As expected, all FHR-4A-specific mAbs (anti-FHR-4A.01 to 4A.06) bound to an epitope located in domains 2–5, which are unique for rhFHR-4A and not present in rhFHR-4B (Figures 1C,D). Of these FHR-4A specific mAbs, only anti-FHR-4A.04 recognized an epitope located in domain 5, whereas all other mAbs bound to an epitope in domain 2 or 3. Anti-FHR-4A.01, 4A.02, 4A.03, 4A.05, and 4A.06, competed with each other for the binding of rhFHR-4A, indicating identical or partially overlapping epitopes (data not shown).

Of the mAbs that cross-reacted with rhFHR-4B and other FHR protein family members, most appeared to bind to either of two epitopes in rhFHR-4A, reflecting the high degree of similarity between these SCR domains 2–5 in FHR-4A and the domains 1, 6, 7, and 8 in FHR-4A and FHR-4B.

Only anti-FHR-4A.09 appeared to have one epitope in domain 8 or 9, recognizing both rhFHR-4A and rhFHR-4B, while also cross-reacting with FHR-3.

The cross-reactive mAbs anti-FHR-4A.08 and anti-FHR-4A.11 were able to block binding of rhFHR-4A to the monospecific FHR-4A mAbs; however, in the reverse setting, this cross-blocking was not achieved (Figure 1E). This is in line with the presence of two epitopes for anti-FHR-4A.08 and 4A.11 in rhFHR-4A. It indicates that one of the two epitopes for anti-FHR-4A.08 and anti-FHR-4A.11 is partially overlapping or sterically hindering the binding site of the monospecific mAbs. The epitope location of anti-FHR-4A.13 could not be mapped due to a very low and ambiguous binding signal when the fragments were used. However, binding of rhFHR-4A to anti-FHR-4A.13 could be blocked with anti-FHR-4A.09, suggesting an epitope location in either domain 8 or 9 for anti-FHR-4A.13. An overview of the mAbs characteristics, the mapped epitope location, and the cross-reactivity is given in Table 1.

Next, we tested whether the anti-FHR-4A mAbs capture FHR-4A and FHR-4B from NHS. To this end, an IP followed by Western Blot was performed, which was developed with cross-reactive biotinylated polyclonal anti-FHR-3. Of the specific anti-FHR-4A mAbs, only anti-FHR-4A.04 seemed to efficiently capture FHR-4A from NHS, resulting in a clear protein band corresponding to the molecular weight of FHR-4A (86 kDa) (Figure 2A). Only a very faint FHR-4A band was visible in the precipitates of anti-FHR-4A.01, 4A.02, 4A.03, 4A.05, and 4A.06. In addition, both anti-FHR-4A.07 and anti-FHR-4A.13 were neither able to efficiently immunoprecipitate FHR-4A from serum nor any of the other FHR proteins with which these mAbs cross-react. A band corresponding to FHR-4B (43 kDa) in the IP with anti-FHR-4A.08 also appeared to be missing, while the mAb was found to be cross-reactive with rhFHR-4B in the ELISA. A protein band corresponding to FHR-4B could not be distinguished in the IP of anti-FHR-4A.09 to anti-FHR-4A.12, as these mAbs also precipitated FHR-3 (43–50 kDa), which migrates at the same expected molecular weight and is also recognized by the polyclonal antibody. Therefore, to better visualize the IP of FHR-4B, an IP from a FHR-3-deficient serum pool was performed, using the cross-reactive mAbs. However, also in these conditions, no band corresponding to FHR-4B was detected in the IP of anti-FHR-4A.08 to anti-FHR-4A.12, which was unexpected given the cross-reactivity we observed with rhFHR-4B in ELISA (Figure 2B). On the other hand, anti-FHR-4A.12 did immunoprecipitate FHR-2 from NHS, confirming the cross-reactivity results obtained with rhFHR-2. As a control, the IP was repeated with the cross-reactive polyclonal anti-FHR-3 using serum of two healthy donors, one with two CFHR3 gene copies, and one with no CFHR3 gene copies. This resulted in the precipitation of FH, FHL-1, FHR-3, and FHR-4A (Figure 2C). However, also with this set-up, no FHR-4B was detected.

Serum fractionated by sucrose gradients was previously used to investigate the molecular size of FHR proteins in their native state (18). Using a recently characterized anti-FHR-3 mAb that cross-reacts with FH, rhFHR-4A, rhFHR-4B, and FHL-1 (clone anti-FHR-3.3) (6), we investigated the circulating molecular size of FHR-4A relative to FH and FHR-3, by IP of these proteins from sucrose gradient fractions. As expected, FH was precipitated from the fractions also containing IgG, while FHR-3 was mainly present in the fractions containing albumin, corresponding with their respective molecular weights (155 and 50 kDa) (Figure 3A). FHR-4A was found in the fractions corresponding with its molecular weight (86 kDa), between the main fractions containing FH and FHR-3. Again, no bands corresponding with FHR-4B were found in the sucrose gradient fractions of neither the FHR-3 sufficient nor the FHR-3-deficient serum pool (Figure 3B).

As we could not detect FHR-4B in human serum, we next focused on specifically measuring FHR-4A by ELISA. FHR-4A was measured using the monospecific mAb anti-FHR-4A.04, captured on immobilized RM-19, and using biotinylated polyclonal anti-FHR-3 antibody as a conjugate. The specificity of the FHR-4A ELISA was confirmed using rhFHR proteins and plasma purified FH, giving a clear signal for rhFHR-4A (Figure 4A). Both rhFHR-3 and rhFHR-4B were detected at the highest concentrations we tested, but the FHR-4A ELISA was found to be 843-fold and 1,490-fold more sensitive for rhFHR-4A compared to rhFHR-4B or rhFHR-3, respectively. Next, using purified rhFHR-4A, we calibrated a NHS pool to contain 2.33 µg/mL FHR-4A, which was subsequently used as a calibration curve in the FHR-4A ELISA (Figure 4B). FHR-4A levels in serum and plasma did not differ (Figure 4C, P > 0.05), nor were the levels affected by repeated freeze/thaw cycles (Figure 4D). Next, we measured the levels of FHR-4A in serum from 129 healthy donors. The average concentration of FHR-4A was found to be close to the concentration found in the NHS pool, being 2.55 µg/mL or 29.65 nM. Moreover, FHR-4A levels were found to be highly variable with a SD of 1.46 µg/mL and ranging from 0.26 up to 6.20 µg/mL (Figure 4E). We subsequently investigated whether FHR-4A behaves as an acute phase protein. To this end, we measured FHR-4A levels in 78 patients during an acute bacterial infection. No correlation was found for FHR-4A with CRP levels (r = 0.024, P = 0.834, Figure 4F). Of note, no total deficiency for FHR-4A was found in any healthy donor or patient measured during this study.