Successful prenatal therapy for anti-CD36-mediated severe FNAIT by deglycosylated antibodies in a novel murine model

Fetal and neonatal alloimmune thrombocytopenia (FNAIT) is caused by maternal antibodies that cross through the placenta via neonatal Fc receptor (FcRn)-mediated transport during pregnancy. This action can lead to the clearance of fetal platelets and endothelial dysfunction, which results in bleeding complications.^1-3 In White populations, antibodies against human platelet antigen-1a (HPA-1a) are responsible for ∼80% of FNAIT cases.⁴ The incidence of severe FNAIT is ∼1:2500 newborns,⁵ in which intracranial hemorrhage (ICH) occurs in ∼20% of these cases, leading to fetal death or persistent neurologic sequelae.⁶ However, FNAIT caused by anti–HPA-1a antibodies has not been well recognized in other populations.

The data indicate that antibodies against CD36 represent an important risk factor for FNAIT in Asian and African populations.^7-10 CD36 is a receptor for thrombospondin-1, a free fatty acid, and a scavenger receptor for different danger signals. Accordingly, CD36 is implicated in inflammatory response, atherosclerosis, thrombosis, and metabolic disorders.¹¹ Individuals lacking CD36 on both platelets and monocytes (type I) are at risk of developing anti-CD36 antibodies (known as anti-Nak^a) after receiving a platelet transfusion or during pregnancy.^12,13 These antibodies may cause immune-mediated thrombocytopenia and repeated early fetal loss.^14-16 Immunization in type II individuals (lacking CD36 on platelets only but not on monocytes) has not been described. Type I CD36 deficiency is extremely rare in White subjects but more common in African subjects (∼2%) and relatively frequent in Asian populations (>0.5%).^7,9,17-19

The clinical manifestation of FNAIT caused by anti-CD36 antibodies is heterogeneous, ranging from widespread petechial hemorrhages to thrombocytopenia, miscarriages, hydrops fetalis, and ICH. Compared with anti–HPA-1a, anti-CD36 antibodies cause more common hydrops and recurrent miscarriages.^16,20,21 The reason for this phenomenon is unknown. Therefore, the current approach for treating FNAIT caused by anti–HPA-1a may not be adaptable for the disease caused by anti-CD36 antibodies.

The current clinical approach for managing FNAIT relies on preventing fetal bleeding complications through antenatal therapy because the prophylaxis is currently unavailable.²² Presently, noninvasive maternal treatment with intravenous immunoglobulin (IVIG) represents the antenatal therapy’s first line.¹ However, such strategies may inhibit the transfer of immune-protective maternal immunoglobulin G (IgG), which might cause an increased risk of infection during pregnancy and the first weeks following birth.²³ An attractive alternative strategy would be to administer nondestructive IgG that shares the pathogenic antibodies’ specificity and retains the ability to be transported across the placenta. Ghevaert et al^24,25 introduced such a strategy using human recombinant single-chain fragment variable anti–HPA-1a antibodies with a modified Fc part. These modified recombinant anti–HPA-1a antibodies could pass the placenta and ease FNAIT in immunized mothers.

The therapeutic potential of deglycosylated IgG (deg-IgG) antibodies for autoimmunity treatment has been widely recognized.²⁶ Removal of N-glycan (linked to asparagines 297), located on the Fc part, leads to a significant reduction of IgG binding with the FcγRs expressed on macrophages and its ability to activate complement C1q. Interestingly, these deg-IgG antibodies could still be transported from the maternal circulation to the fetus via FcRn.²⁷ Accordingly, our previous in vivo study in mice showed that deglycosylated–monoclonal antibody (deg-mAb) specific for HPA-1a could pass through the placenta and prevent the clearance of fetal platelets mediated by maternal anti–HPA-1a antibodies. This observation indicates that the use of epitope-specific antibodies for the antenatal therapy of severe FNAIT is feasible.²⁸ 

The current study established a novel mouse FNAIT model to evaluate the efficacy of deglycosylated anti-CD36 (deg-anti-CD36) to treat FNAIT caused by maternal anti-CD36 antibodies and compared this treatment with IVIG.

Human IVIG was purchased from Jiangxi Boya Bio-Pharmaceutical, Jiangxi, China.

Wild-type (WT) C57BL/6J mice were provided by the Animal Centre of Sun Yat-Sen University, China. Cd36^−/− mice were purchased from The Jackson Laboratory (B6.129S1-Cd36^tm1Mfe/J). The Animal Care Committee approved this study (IACUC-2014-0303).

Platelets were prepared as previously described.²⁹ Cd36^−/− female mice (6-8 weeks old) were immunized 3 times with 10⁸ WT platelets intraperitoneally at weekly intervals. After immunization, 5 µL of serum was collected and incubated with 100 µL of EDTA blood (1:100) from WT mice for 30 minutes and washed with phosphate-buffered saline/1% bovine serum albumin. Then, 50 µL of fluorescein isothiocyanate–conjugated anti-mouse IgG (1:200; Jackson ImmunoResearch Laboratories, West Grove, PA) was added for 30 minutes. After red blood cell lysis (BD Biosciences, Shanghai, China), cells were suspended in 0.5 mL phosphate-buffered saline/1% bovine serum albumin and then analyzed by flow cytometry (FACS Canto II; BD Biosciences).

Platelets from the pups were counted as previously described.²⁹ Briefly, 50 µL of counting beads (Life Technologies, Carlsbad, CA) was added to determine the platelet counts by using flow cytometry. Counting of 2,000 beads represented the standard.

Cd36^−/− mice were immunized as previously described.^3,29 Splenocytes were harvested and fused with SP2/0-Ag14 mouse myeloma cells (ATCC, Manassas, VA) with the help of polyethylene glycol using a standard protocol.³⁰ Hybridomas were screened by using mouse platelets according to flow cytometry. IgG subclasses were determined by a Pierce Rapid Isotyping Kit (Thermo Fisher Scientific, Waltham, MA). Seventeen hybridoma clones producing anti-CD36 could be generated. One clone (labeled 32-106) producing a high-affinity IgG antibody that effectively blocked maternal anti-CD36 binding was selected.

IgG was purified and digested as previously described.²⁸ Aliquots of 1 µL of native and deg-mAb (0.6 mg/mL) were analyzed by flow cytometry using mouse IgG2a as isotype control (eBioscience, San Diego, CA).

One milligram of IgG derived from immunized Cd36^−/− mice was labeled with fluorescein-EX (Invitrogen, Carlsbad, CA). Unlabeled deg-32-106 (5-50 ng) was added to 100 µL of EDTA- anticoagulated mice blood at room temperature for 30 minutes. After washing, 8 µL of fluorescein-conjugated IgG (1.73 mg/mL) was added for 30 minutes. Red blood cells were lysed and the samples were analyzed by flow cytometry.

Deg-32-106 (100 µg) was labeled with Alexa Fluor dye (AF-647; Invitrogen). Subsequently, Alexa Fluor–labeled deg-32-106 was intravenously injected into WT female mice (1 mg/kg). After 10 minutes, fluorescein-EX conjugated maternal IgG containing anti-CD36 was administered (200 mg/kg). The binding of both antibodies on platelets was analyzed by flow cytometry at 30 and 60 minutes. A single administration of labeled deg-32-106 or maternal anti-CD36 was run as a control.

Naive or immunized Cd36^−/− female mice were bred with WT male mice. Anti-CD36 developed in Cd36^−/− mothers was monitored by flow cytometry. During pregnancy, mothers were untreated or treated with human IVIG (1 g/kg), deg-anti-CD36 polyclonal IgG (5 mg/kg), or deg-anti-CD36 mAb (5 mg/kg) intravenously 3 times.^3,29 Dead pups were determined in utero or within 24 hours after delivery. Miscarriage during pregnancy (a drop of body weight >1 g), platelet counts (within 24 hours), and bleeding in the pups after delivery were analyzed, as previously described.³¹ 

Murine placentas were collected at 16.5 days’ postcoitum and embedded in paraffin. Placental sections (3 µm) were stained with hematoxylin/eosin to determine the labyrinth zone ratio of the whole placenta. For immunohistochemistry, placental sections were incubated with anti-CD31 at 4°C overnight and stained with enzyme-labeled secondary antibody (Servicebio, Wuhan, China). Two fields per placenta labyrinth zone were selected, and the CD31⁺ area was quantified by using Image J (National Institutes of Health, Bethesda, MD).

Human placental microvascular endothelial cells (HPVEC; 4 × 10⁵/mL; FuDan, IBS Cell Center, Shanghai, China) in 50 µL Dulbecco’s modified Eagle medium (serum free) were seeded onto the Matrigel (Corning, Bedford) and incubated for 30 minutes at 37°C. Thrombin (1 U; MilliporeSigma, Burlington, MA), 4 µg mouse IgG, mAb 32-106, or deg-32-106 in 50 µL Dulbecco’s modified Eagle medium containing 5% mouse serum was added. For the blocking experiment, 1 µg anti-FcγRIIa (clone IV.3; Stemcell Technologies, Vancouver, BC, Canada) was added with mAb 32-106. Furthermore, 50 µL anti-CD36 sera (1:10, 1:40) was transferred into the gel and incubated with HPVEC at 37°C for 6 hours in the absence or presence of deg-32-106. Images were taken from 2 to 3 selected areas per well. The tube formation assay was repeated 3 times. The total tube length was quantified by using Wimasis Image Analysis (Onimagin Technologies, Córdoba, Spain).

Data are presented as mean ± standard deviation and were analyzed by using Prism version 5.0 (GraphPad Software, La Jolla, CA). Comparisons of the 2 groups were assessed by using the two-tailed unpaired Student t test and χ² test. One-way analysis of variance was used for the multiple comparisons, and the Fisher exact test was used for small sample sizes. Correlations between anti-CD36 titers and mortality rates were analyzed by using the Pearson correlation method. P values <.05 were considered significant.

We initially investigated whether Cd36^−/− female mice could develop natural anti-CD36 antibodies during the first pregnancy when crossed with WT male mice. An analysis of maternal sera by flow cytometry showed that immunization rarely occurred. Only one-fifth of mothers developed anti-CD36 at low titers and only after the second and third pregnancy. Interestingly, one mother (#1) delivered a smaller littermate (litter size, 1.50 ± 0.71) compared with mothers (#2-5) that lacked anti-CD36 antibodies (litter size, 7.50 ± 1.93) (Figure 1A), indicating that miscarriages and fetal death occurred during pregnancy.

To establish a murine model of anti-CD36–mediated FNAIT at the first pregnancy, we immunized Cd36^−/− female mice 3 times with 10⁸ WT platelets before breeding with WT male mice. This study designed 8 cohorts (#1-8) (supplemental Table 1 [available on the Blood Web site]; Figure 1). Naive WT (#1) and Cd36^−/− female mice (#2) were conducted as control cohorts. Flow cytometry analysis showed that Cd36^−/− mice (#3) developed anti-CD36 antibodies after immunization (Figure 1B). Interestingly, these antibodies also reacted with human Cd36^+/+ platelets (supplemental Figure 2). These results also pointed out that besides anti-CD36, no other platelet-reactive antibodies are involved in our animal model. In contrast, anti-CD36 was undetectable in naive Cd36^−/− and naive Cd36^+/+ mice (cohorts #1-2). Notably, 31 (40.26%) of 77 pups that were delivered from 16 immunized Cd36^−/− mothers were found dead: 3 in utero and 28 pups after deliveries (bleeding, ICH, and hydrops) (Figure 1C-D). The mortality of pups in this cohort was significantly higher than in cohort #1 (χ² = 17.46; P < .0001) and cohort #2 (χ² = 11.52; P = .0007). In total, only 2.70% to 4.00% of dead pups were documented in the control cohorts (#1-2). Furthermore, litter size numbers from immunized Cd36^−/− mothers with detectable anti-CD36 antibodies (cohort #3) showed a trend to be smaller than control cohorts; however, it was not statistically significant compared with naive WT (P = .064) and Cd36^−/− (P = .323) mothers (Figure 1E). This smaller litter size was probably related to 2 miscarriages found in cohort #3, which were not found in the control cohorts.

Analysis of anti-CD36 antibodies in immunized Cd36^−/− mothers and their corresponding pups showed that maternal anti-CD36 antibodies could not only be detected in fetal sera (circulating IgG) but also on fetal platelets (platelet-binding IgG) (Figure 2A). In these mothers, the platelet counts in the surviving pups were significantly lower than in pups from naive Cd36^−/− mothers (377.12 ± 121.14 ×10⁹/L vs 607.76 ± 87.80 ×10⁹/L; P < .0001) (Figure 2B). A significant correlation (95% CI, 0.1161-0.8349; P = .0189) between anti-CD36 antibody titer and pups’ mortality rates was observed (Figure 2C).

These findings showed that maternal anti-CD36 antibodies could cross the placenta, bind to fetal platelets, and induce severe FNAIT with a high frequency of fetal death.

Subsequently, we queried whether IVIG could prevent the severe FNAIT found in our animal model. IVIG (1 g/kg) was intravenously administered to immunized Cd36^−/− mothers 3 times on days 10, 15, and 20 after breeding. In the control experiment, no fetal death was detected in naive Cd36^−/− mothers treated with IVIG (cohort #4) (Figure 3A). Surprisingly, this treatment did not prevent the severe clinical outcome of FNAIT. A high frequency of fetal death was still documented (40.00% vs 0.00%; P < .001) (cohort #5a). However, we found a significant reduction in fetal death (12.70% vs 40.00%; P < .01) when IVIG was administered earlier, on days 7, 12, and 17 (cohort #5b), although similar titers of maternal anti-CD36 antibodies before IVIG administration were observed in both cohorts (data not shown). In accordance, a reduced antibody titer was first observed on day 17 after the second antenatal administration of IVIG (P < .05) (Figure 3B).

These results show that IVIG is suitable for the antenatal treatment of anti-CD36–mediated FNAIT, but particular care is required due to the delayed response of this therapy management.

Based on our previous studies, we sought to investigate whether deg-anti-CD36 antibodies could ease thrombocytopenia and prevent the severe effects of maternal anti-CD36 antibodies on the fetus during pregnancy.²⁸ For this purpose, a pool of IgG purified from sera of immunized Cd36^−/− mice (n = 10) and nonimmunized (n = 10; as control) mice were deglycosylated with PNGase F enzyme. Because undigested anti-CD36 IgG antibodies can worsen the pups’ fate, only pure deg-anti-CD36 was used for this in vivo study (supplemental Figure 3). When deg-anti-CD36 (5 mg/kg) was injected into immunized Cd36^−/− mothers 3 times on days 10, 15, and 20 after breeding, the numbers of dead pups (cohort #7) were significantly lower than those of mothers treated with deg-mouse IgG (cohort #6) (5.26% vs 52.94%; P < .01) (Figure 4A).

Given the marked advantages of deg-anti-CD36 for the antenatal treatment of FNAIT, mouse mAbs against mouse CD36 were generated. One clone producing a high-affinity IgG2a antibody against mouse and human CD36, termed 32-106, was selected and deglycosylated. Flow cytometry analysis showed that both native and deg-32-106 IgG bound with similar affinity to WT platelets (Figure 4B). Furthermore, the competitive inhibition study in vitro (Figure 4C) showed that deg-32-106 could completely block the binding of maternal anti-CD36 antibodies into WT platelets. In vivo, administration of deg-32-106 significantly reduced the binding of maternal IgG containing anti-CD36 antibodies after 30 and 60 minutes (54.8% vs 34.9%; 35.5% vs 14.9%) (Figure 4D). However, the frequency of fluorescein isothiocyanate–labeled platelets (sensitized platelets) decreased after 60 minutes, most probably due to platelet clearance. Finally, we administered deg-32-106 (5 mg/kg) into immunized Cd36^−/− mothers by the same therapy protocol (cohort #8). Only 1 dead pup (1 of 46 [2.17%]) was found in this cohort (P < .0001) (Figure 4A).

In addition, treatment with deg-32-106 led to increased pup platelet counts compared with the nontreated cohort (468.24 ± 85.20 ×10⁹/L vs 377.12 ± 121.14 ×10⁹/L; P < .05) (Figure 5A). Increased platelet counts (463.30 ± 30.88 ×10⁹/L vs 377.12 ± 121.14 ×10⁹/L; P < .05) were also observed with IVIG therapy; however, this was noted only when IVIG was administered early. These results showed that deg-32-106 could prevent the fatal severe effect of maternal anti-CD36 antibodies and improve the thrombocytopenic status of the fetus. More importantly, this treatment’s positive effect occurred more rapidly and was more effective compared with IVIG therapy. A similar phenomenon was observed with the litter size (Figure 5B).

Analysis of the placenta of immunized Cd36^−/− mothers revealed a significant reduction in the placental labyrinth area compared with that of the naive cohort (55.75 ± 3.28% vs 62.88 ± 2.58%; P < .005) (Figure 6A-B). Furthermore, the quantification of vascular density in the placenta labyrinth by staining of endothelial cells with anti-CD31 antibodies showed lower fetal capillary numbers in the placenta labyrinth from immunized Cd36^−/− mothers compared with naive cohort (9.86 ± 1.40 vs 13.98 ± 1.48; P < .0005) (Figure 6C-D).

Subsequently, tube formation assay was performed by using the placenta microvascular endothelial cells (HPVEC) expressing human CD36 to study the influence of mouse sera containing anti-CD36 on the angiogenesis process. Our flow cytometry analysis showed that human CD36 could react with mAb 32-106 against mouse CD36 (supplemental Figure 4). As shown in Figure 7A to 7B, sera containing anti-CD36 antibodies significantly reduced the tube length compared with the control sera (P < .0005). A similar phenomenon was observed with 32-106. In contrast, deg-32-106 did not alter angiogenesis (Figure 7C-D), indicating the important role of FcγR. Indeed, blocking endothelial FcγRIIa with mAb IV.3 abolished the antiangiogenic effect of mAb 32-106 (supplemental Figure 5). Finally, preincubation of HPVEC with deg-32-106 restored the antiangiogenic effect caused by the anti-CD36 sera, both at 1:40 and 1:10 serum dilutions (P < .005). These results suggest that impaired angiogenesis of fetal placenta endothelial cells caused by maternal anti-CD36 antibodies could lead to the fetal death observed in our mouse model of FNAIT.

This study established the first animal model of anti-CD36–mediated FNAIT and compared the capability of IVIG and deg-anti-CD36 antibodies to lessen FNAIT. After immunization with WT platelets, Cd36^−/− female mice were bred with WT male mice. A high mortality rate of pups (40.26%) was detected. Furthermore, massive skin bleeding, hydrops fetalis, and ICH was identified. The surviving pups’ platelet counts (377.12 ± 121.14 ×10⁹/L) were lower than those of the control cohorts. However, more severe thrombocytopenia (platelet counts, 132.2 ± 10.5 ×10⁹/L) in pups caused by maternal anti-β3 antibodies was observed in a similar FNAIT model.²⁹ The reason for this mild thrombocytopenia in our cohorts may be the low level of CD36 surface expression on mouse platelets (<25,000 copies) and the broad cellular distribution of the CD36 antigen (platelets, monocytes, macrophages, and endothelial cells).³² In contrast, high copy numbers of the αIIbβ3 integrin (110,000 to 130,000 copies/platelets) and the more restricted cellular distribution of mouse β3 integrin (platelets and endothelial cells) have been documented.³³ 

In humans, variable clinical pictures of FNAIT caused by anti-CD36 antibodies have been observed, including widespread petechial hemorrhages, severe thrombocytopenia, ICH, and hydrops fetalis.^{10,15,16,20,34} All these clinical pictures were found in our FNAIT animal model, indicating that this model is suitable for studying the mechanism of anti-CD36–mediated FNAIT and proving certain therapy strategies.

Several treatment options have been conducted to prevent severe FNAIT caused by anti––HPA-1a antibodies, including serial fetal blood sampling, intrauterine platelet transfusions, and infusions of IVIG. A systematic review suggested that IVIG administration represents the first-line antenatal treatment, whereas fetal blood sampling and intrauterine platelet transfusion resulted in a high complication rate.¹ Nevertheless, refractory states under IVIG treatment were still observed in mothers with ICH, although the success rate is 98.7%.^35,36

Despite the increasing use of IVIG for treating FNAIT, the precise mechanism of IVIG action is still unclear.³⁷ Several mechanisms have been proposed, including decreasing maternal antibody production by inducing immune tolerance³⁸ and increasing pathogenic antibody clearance³⁹ or decreasing antibody transport by saturated FcRn.⁴⁰ In the β3^−/− mice model of FNAIT, administration of IVIG could downregulate anti-β3 antibodies in both maternal and fetal circulations through FcRn-dependent and FcRn-independent pathways.^3,29 However, such strategies may inhibit the transfer of immune-protective maternal IgG, which might cause an increased risk of infections by the local dominant micropathogens during pregnancy and in the first weeks following birth. Therefore, more specific and effective antenatal therapies are desirable.

Our previous studies have shown that deg-mAb SZ21 against HPA-1a could pass through the placenta, inhibit the binding of maternal anti–HPA-1a antibodies, and prevent the clearance of fetal platelets by macrophages.²⁸ Furthermore, we found that some anti–HPA-1a antibodies explicitly bound to αvβ3 expressed by endothelial cells could induce endothelial dysfunction responsible for developing ICH in the fetus with severe FNAIT.^41,42 More recently, we observed that deg-SZ21 could prevent not only thrombocytopenia but also inhibit endothelial dysfunction caused by anti-HPA-1a antibodies (S.S. manuscript in preparation). Based on this knowledge, an antenatal therapy for severe FNAIT based on epitope-specific competitive antibodies should be feasible.

Earlier studies have shown that most mAbs against CD36 and anti-CD36 sera recognized epitopes within amino acids 155-183, indicating the important role of this domain as an immune-dominant target for anti-CD36 antibodies.^43,44 In the current study, we selected one mAb, 32-106, from our panel. Our in vitro and in vivo data showed that deg-mAb 32-106 could inhibit the binding of maternal anti-CD36 antibodies, indicating that both antibodies also react with the immune-dominant region. Accordingly, we found that the administration of deg-32-106 (5 mg/kg body weight) to the immunized Cd36^−/− mothers 3 times (on days 10, 15, and 20) after breeding significantly increased not only fetal platelet counts (377.12 ± 121.14 ×10⁹/L to 468.24 ± 85.20 ×10⁹/L; P < .05) but also significantly reduced fetal death (40.26% to 2.17%; P < .005).

Surprisingly, similar antenatal treatment with IVIG administered in a dose of 1 g/kg body weight on days 10, 15, and 20 after breeding did not result in increased platelet counts and did not restore fetal death (40.00%). Based on risk stratification, weekly doses of IVIG (0.5 or 1 g/kg) are recommended to prevent bleeding complications in pregnancies complicated by FNAIT.³⁷ At the start of treatment, the gestational age is mainly based on the estimated onset of ICH, ranging from 20 until 28 weeks of gestation. In women with a previous child with ICH, IVIG is commonly introduced earlier, at 12 weeks of gestation. This management indicates that early antennal treatment with IVIG prevents severe FNAIT. Following this view, IVIG administration 3 days earlier to the immunized Cd36^−/− mothers (ie, days 7, 12, and 17) decreased fetal death (40.00% to 12.70%). However, fetal death frequency is still significantly higher than that of antenatal treatment with deg-32-106 (2.17%). The late response of IVIG may be attributed to the slow downregulation of maternal antibodies. In our animal model, downregulation of maternal anti-CD36 antibodies was observed only after the second administration of IVIG, which agrees with the current view mechanism that IVIG can prevent serious FNAIT by downregulation of maternal anti-CD36 titer due to clearance of IgG via FcRn.³⁹ However, other mechanisms, such as direct inhibition of placenta transport by IVIG, should be considered.⁴⁰ Nevertheless, the effect of IVIG seems to be delayed.

Furthermore, Leontyev et al⁴⁵ reported that C57BL/6 mice are much less sensitive than BALB/c mice to IVIG-mediated attenuation of autoimmune thrombocytopenia (ITP), requiring ∼2.5-fold more IVIG (2.5 g/kg) than BALB/c mice. In the previous FNAIT model, mice on the BALB/c background were used to study the effect of IVIG on anti-β3 antibody-mediated FNAIT.³ Here, our results were based on the experiments with C57BL/6 mice, which may explain the low sensitivity of the IVIG treatment in our FNAIT model.

Disturbance of placenta vascular development and function could dramatically alter fetal growth development and thereby neonatal survival. In this process, placental vascularization and angiogenesis play critical roles.^46,47 During the third trimester of pregnancy, placenta preferential transport of maternal plasma fatty acids is critical for fetal growth and development.⁴⁸ CD36 is found on placental membranes, microvillus, and basal membrane.⁴⁹ The central role of CD36 (also known as fatty acid translocase) as a high-affinity receptor for fatty acid uptake and lipid metabolism has been well documented.^50,51 It is conceivable that the inhibition of fatty acid uptake via CD36 receptor by anti-CD36 antibodies could lead to inadequate placental angiogenesis. Indeed, the placenta analysis from immunized Cd36^−/− mothers revealed a significant reduction of placental labyrinth area and decreased fetal capillary numbers compared with the naive cohort.

Previous studies reported on FcγRIIa (CD32) expression in the placenta microvascular endothelial cells.⁵² Accordingly, preincubation of HPVEC with mAb against FcγRIIa restored the antiangiogenic effect of anti-CD36 antibodies. Consequently, the decisive effect of anti-CD36 antibodies could be restored by the addition of deg-mAb 32-106. One study showed that antibodies bound to endothelial cells significantly increased polymorphonuclear leukocyte adhesion in an FcγRIIa-dependent manner in cooperation with CXCR1/2, a mechanism of tissue injury during the inflammatory response.⁵³ The use of deg-32-106 may prevent antibody-mediated endothelial activation and recruitment of other blood cells such as monocytes and platelets. Taken together, our results showed that despite thrombocytopenia, maternal anti-CD36 antibodies contribute to fetal death by affecting placental angiogenesis.

In summary, we established a mouse model of FNAIT that reproduced the symptoms of human FNAIT induced by anti-CD36 antibodies. Although only mild thrombocytopenia was observed in the fetus, maternal anti-CD36 antibodies can cause severe bleeding, miscarriage, and fetal death. These severe clinical symptoms could be prevented by antenatal treatment with IVIG and deg-mAb 32-106 against CD36. Notably, treatment with deg-mAb against CD36 seems more beneficial than IVIG for various reasons, including the use of a lower dose (at least 200-fold less), the later start of treatment, and therapy success, despite other general IVIG disadvantages. Because our mouse hybridoma, 32-106, also recognizes human CD36 and can inhibit the binding of maternal anti-CD36 from FNAIT cases (supplemental Figure 6), humanized deg-mAb 32-106 should be feasible for immunotherapy in the near future.

The authors gratefully thank Jean-Pierre Allain (Department of Hematology, University of Cambridge) for his valuable suggestions and careful revision of the manuscript. They also thank LetPub (www.letpub.com) for its linguistic assistance and scientific consultation during the manuscript preparation.

This work was supported by grants from the National Natural Science Foundation of China (81601451 and 81970169), the Natural Science Foundation of Guangdong Province of China (2016A030313124 and 2020A1515011348), the Science and Technology and Innovative Commission of Guangzhou City (201707010021), and the Canadian Institute of Health Research Foundation Grant (389035) for H.N.

X.X. and D.C. are PhD candidates at the Justus Liebig University. This work is submitted in partial fulfillment of the doctoral requirements of the PhD.

Contribution: X.X., X.Y., Y.F., and S.S. conceived and designed the research; X.X., D.C., Y.X., Y.C., Y.S., J.D., H.D., J.L., and J.W performed experiments; X.X., S.S., D.C., X.Y., and W.X. analyzed data; X.X., Y.F., and S.S. wrote the manuscript; and H.N. analyzed and revised the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Yongshui Fu, Guangzhou Blood Centre, 31 Lu Yuan Rd, Guangzhou, Guangdong, China 510095; e-mail: fuyongshui@sina.com; or Sentot Santoso, Institute for Clinical Immunology and Transfusion Medicine, Justus Liebig University Giessen, Langhansstr. 7, 35385 Giessen, Germany; e-mail: sentot.santoso@immunologie.med.uni-giessen.de.