Recent thymic emigrants that fail postpositive selection maturation are targeted by complement proteins. T cells likely acquire complement resistance during maturation in the thymus, a complement-privileged organ. To test this, thymocytes and fresh serum were separately obtained and incubated together in vitro to assess complement deposition. Complement binding decreased with development and maturation. Complement binding decreased from the double-positive thymocyte to the single-positive stage, and within single-positive thymocytes, complement binding gradually decreased with increasing intrathymic maturation. Binding of the central complement protein C3 to wild-type immature thymocytes required the lectin but not the classical pathway. Specifically, MBL2 but not MBL1 was required, demonstrating a unique function for MBL2. Previous studies demonstrated that the loss of NKAP, a transcriptional regulator of T cell maturation, caused peripheral T cell lymphopenia and enhanced complement susceptibility. To determine whether complement causes NKAP-deficient T cell disappearance, both the lectin and classical pathways were genetically ablated. This blocked C3 deposition on NKAP-deficient T cells but failed to restore normal cellularity, indicating that complement contributes to clearance but is not the primary cause of peripheral T cell lymphopenia. Rather, the accumulation of lipid peroxides in NKAP-deficient T cells was observed. Lipid peroxidation is a salient feature of ferroptosis, an iron-dependent nonapoptotic cell death. Thus, wild-type thymocytes naturally acquire the ability to protect themselves from complement targeting by MBL2 with maturation. However, NKAP-deficient immature peripheral T cells remain scarce in complement-deficient mice likely due to ferroptosis.

The complement system is an integral component of innate immunity and consists of soluble protein fragments and complexes found in the circulation (14). Although initially thought to be predominantly important for host defense by aiding in Ab-mediated clearance of pathogens, it is now well appreciated that functions of the complement system extend to tissue homeostasis, immune response modulation, and elimination of unwanted or dying cells (5, 6). Complement activation relies on interplay between multiple soluble proteins and receptors via three main mechanisms: the classical, alternative, and lectin pathways (2). These pathways converge at the complement component 3 (C3) convertase, which catalyzes formation of effector molecules, opsonins, and anaphylatoxins by enzymatic fragmentation of C3. This leads to outcomes such as phagocytosis of microbial targets, modulation of immune cell activation, and clearance of host Ags generated via apoptotic or necrotic processes (2, 4). Classical complement pathway can be initiated by Ab-dependent or Ab-independent C1q-mediated mechanisms. In contrast, lectin pathway activation depends on the collectin family of molecules, including mannose-binding lectins (MBL) and ficolins. These molecules recognize and bind specific glycosylation patterns such as mannose residues, which are found on bacterial cell walls and in injury such as ischemia (2, 3, 7, 8). Lastly, the alternative pathway is spontaneously activated leading to membrane attack complex formation in the absence of regulatory molecules (4). Multiple factors determine the activity of complement proteins, including inhibitory molecules, glycosylation patterns on the cell surface, and accessibility of tissues to complement proteins (9, 10).

The thymus is a complement-privileged organ, as the deletion of a complement inhibitor–encoding gene, complement receptor 1–related (Crry) gene, leads to peripheral T cell lymphopenia mediated by complement attack while thymic cellularity is maintained (10). There is no evidence of complement deposition on thymocytes when analyzed directly ex vivo (10). Glycosylation patterns change as thymocytes progress through development in the thymus (11). There is a gradual increase in cell surface sialylation due to increasing expression of sialic acid transferases (1215). Increasing sialylation masks exposed mannose residues on developing thymocytes (11, 16). Sialylation is required to prevent complement activation as demonstrated by enzymatic stripping of sialic acid by neuraminidase resulting in complement activation and cell death (17). Taken together, these observations suggest that T cells gain resistance to complement prior to egress into blood, which contains complement proteins. Previously, we have shown that the transcriptional regulator NKAP is required for T cell maturation (15, 18). NKAP-deficient thymocytes fail to upregulate α-2,8 sialyltransferases (15). Peripheral T cells lacking NKAP are opsonized by complement proteins and eliminated at the recent thymic emigrant (RTE) stage, further indicating that complement resistance is gained intrathymically as part of T cell maturation (15, 19).

In this study, we directly tested whether resistance to complement is gained during thymic T cell maturation. Susceptibility of thymocytes to complement as a function of maturation and contribution of complement proteins in mediating disappearance of NKAP-deficient T cells was investigated. To circumvent restricted access of the thymus to complement (10), freshly harvested wild-type (WT) thymocytes were incubated with freshly isolated serum in vitro and assessed for complement binding. C3 and C4 deposition on thymocytes was inversely proportional to development and maturation. Deposition required the lectin pathway, whereas the classical pathway was dispensable. Specifically, MBL2 was required for C3 and C4 deposition, whereas MBL1 was not. Finally, ablation of both the classical and lectin pathways (C1q knockout [KO] MBL1 MBL2 double KO [dKO]) was needed to prevent C3 deposition on NKAP-deficient mature naive T cells (MNTs) in the periphery but failed to restore normal naive T cell percentages and absolute numbers in CD4-cre NKAP conditional KO (cKO) mice, suggesting another mode of cell death as the primary cause of T cell lymphopenia. Increased lipid peroxidation in NKAP-deficient cells indicated ferroptosis, a form of regulated cell death driven by reactive oxygen species derived from iron metabolism. Overall, this study is the first, to our knowledge, to provide evidence that thymocytes gain resistance to the lectin pathway of complement deposition as a function of increasing thymic T cell maturation and that the death of NKAP-deficient T cells is driven by ferroptosis, followed by complement-mediated clearance.

C57BL6, C1q KO, C3 KO, MBL1/MBL2 dKO mice were obtained from The Jackson Laboratory. NKAP fl mice (20) were interbred with MBL1/MBL2 dKO mice or CD4-cre NKAP cKO mice (18) to create MBL1/MBL2 dKO CD4-cre NKAP cKO mice. CD4-cre NKAP cKO mice were crossed to C1q KO mice to generate C1q KO CD4-cre NKAP cKO mice. C1q KO CD4-cre NKAP cKO were interbred with MBL1/MBL2 dKO CD4-cre-NKAP cKO mice to generate C1q KO MBL1/2 dKO CD4-cre NKAP cKO mice. MBL1/MBL2 dKO mice were outbred to generate MBL1 KO and MBL2 KO mice. Mice between 6 and 10 wk of age were used for all the experiments.

For complement deposition experiments, freshly harvested WT (6–10 wk old) thymocytes were incubated in freshly isolated sera in GVB++ buffer (Complement Technology) at a 1:5 (serum:GVB++) ratio for 1 h at room temperature. Serum was isolated by bleeding mice (6–10 wk old) retro-orbitally, incubating blood at room temperature to allow coagulation followed by centrifugation at 8000 rpm. Cells were washed with FACS buffer and stained with fluorescent-conjugated cell surface Abs against CD4 (1:500 dilution, clone RM4-5 or GK1.5), CD8 (1:500 clone 53-6.7 or 2.43), CD24 (1:1000, clone mature 1 [M1]/69), and CCR7 (1:100, clone 4B12) (BioLegend, eBioscience, and Tonbo) and biotinylated Abs for C1q (1:100), C3 (1:200), and C4 (1:500) (Cedarlane). Surface stain of CCR7 was performed for 30 min at 37°C followed by washing and staining of other surface Ags on ice for 30 min and washing to remove unbound Abs. Secondary stain was performed with PE (1:1000) or allophycocyanin (1:1000)–conjugated streptavidin (BioLegend and eBioscience), and unbound Abs were washed. All experiments included fixable viability dye (Tonbo) for analysis of live cells. Stained cells were analyzed using an Attune NxT Flow Cytometer (Thermo Fisher Scientific). Data were analyzed using FlowJo (Tree Star) 9.8 or 10. For analysis of all experiments, doublets were excluded based on forward and side scatter, and dead cells positive for fixable viability dye were also excluded. For mannose inhibition of C3 deposition, freshly isolated WT sera was first treated for 5 min with mannose (Sigma-Aldrich) at various concentrations and incubated with WT thymocytes as described above. A large range of dilutions of mannose was used to determine whether low levels of mannose could impose a block on C3 deposition. These concentrations did not alter osmotic potential of cells as determined using forward and side scatter of total thymocytes.

Freshly harvested splenocytes were cell surface stained on ice followed by incubation with fluorescent-conjugated Annexin-V in Annexin-V binding buffer for 10 min at room temperature (BioLegend) immediately or following a 1-h incubation at 37°C in RPMI 1640 media (Corning) supplemented with 10% FBS and 1 μM BODIPY-C11 (Thermo Fisher Scientific). For examination of Liperfluo (Dojindo Molecular Technologies) positivity, freshly harvested splenocytes were incubated with 10 μM Liperfluo (resuspended in 1% DMSO) in HBSS for 30 min at 37°C. Cells were then washed and surface stained. To examine the impact of ferroptosis inhibitors on lipid peroxidation, total splenocytes were incubated with 100 μM α-tocopherol (αToc) (Millipore Sigma) and/or 10 μM Ferrostatin-1 (Millipore Sigma) for 15 min followed by staining with 1 μM BODIPY-C11 for 1 h. Cells were then washed and surface stained. For intracellular protein detection, cells were fixed and permeabilized (TNB-0607-KIT; Tonbo) and stained with fluorescent-conjugated Abs against Bcl2 and cleaved caspase-3 (BioLegend and Cell Signaling Technology). For determining intracellular glutathione (GSH) peroxidase 4 (GPX4) expression, cells were surface stained, fixed, permeabilized (558049 and 558050; BD Biosciences), and stained with primary Ab against GPX4 (Abcam) followed by anti-rabbit secondary Ab (Southern Biotech). For cellular dyes, 1 μM YO-PRO1 iodide (Thermo Fisher Scientific) was added, and splenocytes were stained on ice for 30 min followed by cell surface staining. MitoTracker, MitoSOX, TMRM, CellROX, and ThiolTracker dyes (all from Thermo Fisher Scientific) were added at 50 nM, 5 μM, 100 nM, 5 μM, and 10 μM, respectively. Data were analyzed as above using flow cytometry methods.

Mouse sera were diluted 500-fold in sample buffer and boiled before loading on gels. Blotting was performed overnight at 4° with the following Abs: 0.4 μg/ml rabbit anti-C3 (ab200999; Abcam), 0.1 μg/ml goat anti-MBL1 (AF2077; R&D Systems), 0.1 μg/ml goat anti-MBL2 (AF2208; R&D Systems), and 0.4 μg/ml rabbit anti-Transferrin (ab82411; Abcam). Proteins were then detected by incubation for 1 h at room temperature with HRP-coupled secondary Abs at 0.1 μg/ml, using either goat anti-rabbit IgG (4050; Southern Biotech) or donkey anti-goat IgG (6420; Southern Biotech). All Abs were diluted in tris-buffered saline with 3% BSA. Bands were detected on film with ECL Pro reagent (PerkinElmer).

All statistical analysis involved enumeration of absolute cell counts or mean fluorescence intensity, calculation of SEM, and statistical significance using one-way ANOVA (GraphPad Prism software) for multiple comparisons.

NKAP-deficient T cells have a defect in postpositive selection maturation, as peripheral T cells lacking NKAP are both phenotypically and functionally immature (15, 18). Furthermore, NKAP-deficient T cells are opsonized by complement proteins in the periphery and eliminated at the RTE stage, suggesting that complement resistance is gained as part of T cell maturation. In contrast, complement factors do not bind to NKAP-deficient thymocytes when directly assessed ex vivo (15). This suggests that the thymus is a complement-privileged organ with restricted access to complement proteins. To circumvent restricted access of the thymus to complement (10) and to directly assess whether complement susceptibility of thymocytes decreases with maturation, freshly harvested WT thymocytes were incubated with serum in vitro and assessed for complement binding. The binding of C1q, C4, and C3 to double-positive (DP), CD8 single-positive (SP), CD4 SP, CD4 semimature (SM), CD4 M1, and CD4 mature 2 (M2) WT thymocytes were assessed by incubating with serum from WT mice in vitro (Fig. 1). Deposition of C1q, C3, and C4 decreased as T cell development and maturation progressed. In particular, there was almost no complement deposition on M2 thymocytes, the most mature SP cells that are licensed to egress. Thus, thymocytes become resistant to complement prior to egress from the thymus.

FIGURE 1.

Susceptibility to complement binding decreases during thymic development and maturation. (A) Examination of C1q, C3, and C4 binding on DP, CD4 SP, CD8 SP thymocytes after incubation with WT serum. DP thymocytes incubated without serum were used as a negative control. (B) Examination of C1q, C3, and C4 complement binding on CD4 SP thymocytes divided into CD24+CCR7 SM, CD24+CCR7+ M1, and CD24CCR7+ M2 populations. CD4 SP thymocytes incubated without serum were used as a negative control. Data are representative of four WT mice in total from four independent experiments.

FIGURE 1.

Susceptibility to complement binding decreases during thymic development and maturation. (A) Examination of C1q, C3, and C4 binding on DP, CD4 SP, CD8 SP thymocytes after incubation with WT serum. DP thymocytes incubated without serum were used as a negative control. (B) Examination of C1q, C3, and C4 complement binding on CD4 SP thymocytes divided into CD24+CCR7 SM, CD24+CCR7+ M1, and CD24CCR7+ M2 populations. CD4 SP thymocytes incubated without serum were used as a negative control. Data are representative of four WT mice in total from four independent experiments.

Close modal

Relative contributions of the lectin and the classical complement pathways to C3 deposition on thymocytes were examined. These two pathways converge at C4 (Fig. 2A). The C3 convertase, C4bC2a, formed via activation of either the lectin or the classical pathway enables C3 deposition. Sera from C1q KO, MBL1/MBL2 dKO, and C3 KO mice were examined for their ability to deposit C1q, C3, and C4 on thymocytes. Absence of C3, MBL1, and MBL2 in sera was confirmed by Western blot (Supplemental Fig. 1). Sera were isolated from WT, C1q KO, MBL1/MBL2 dKO, as well as C3 KO mice and individually incubated with WT thymocytes (Fig. 2B–E, gating strategy in Supplemental Fig. 2). Compared with WT thymocytes incubated with sera from WT mice, WT thymocytes incubated with sera from MBL1/MBL2 dKO mice lacked C3 and C4 deposition, whereas C1q deposition was intact. WT thymocytes incubated with sera from C3 KO mice were used as a negative control (Fig. 2B). The extent of C3 binding on WT thymocytes incubated with sera from C3 KO mice was comparable to WT thymocytes incubated with sera from MBL1/MBL2 dKO mice ruling out the alternative pathway and demonstrating that the lectin pathway is necessary for complement deposition on thymocytes (Fig. 2B, 2E). Loss of C1q from sera did not prevent C3 and C4 deposition on WT thymocytes (Fig. 2C), demonstrating that the classical pathway is not required. Further, preincubation of sera from WT mice with mannose abrogated C3 deposition, indicating that soluble mannose binds mannose-binding molecules in sera were responsible for C3 deposition (Fig. 2F).

FIGURE 2.

The lectin pathway is required for C3 deposition. (A) Simplified schema of the classical and lectin pathways of complement activation; both pathways converge at C4 deposition. (BD) Examination of C3, C1q, and C4 binding to WT DP, CD4 SM, CD4 M1, CD4 M2 cells after incubation with either WT, C1Q KO, MBL1 MBL2 dKO, or C3 KO sera. Data are representative of four WT mice in total from four independent experiments. (E) Quantification of percent DP and CD4 SP thymocytes bound by C3, C1Q, and C4. Data are averaged from four mice in total from four independent experiments, and error bars represent SEM. The p values were calculated by one-way ANOVA. (F) Preincubation of sera with various concentrations of d-mannose blocks C3 deposition on thymocytes. Data were averaged from two independent experiments with three mice in total, and error bars indicate SEM.

FIGURE 2.

The lectin pathway is required for C3 deposition. (A) Simplified schema of the classical and lectin pathways of complement activation; both pathways converge at C4 deposition. (BD) Examination of C3, C1q, and C4 binding to WT DP, CD4 SM, CD4 M1, CD4 M2 cells after incubation with either WT, C1Q KO, MBL1 MBL2 dKO, or C3 KO sera. Data are representative of four WT mice in total from four independent experiments. (E) Quantification of percent DP and CD4 SP thymocytes bound by C3, C1Q, and C4. Data are averaged from four mice in total from four independent experiments, and error bars represent SEM. The p values were calculated by one-way ANOVA. (F) Preincubation of sera with various concentrations of d-mannose blocks C3 deposition on thymocytes. Data were averaged from two independent experiments with three mice in total, and error bars indicate SEM.

Close modal

Mice express two highly homologous MBL genes (MBL1 and MBL2), which are thought to be redundant in function (21). Sera from either MBL1 KO or MBL2 KO mice were incubated with WT thymocytes (Fig. 3, gating strategy in Supplemental Fig. 3). Sera from MBL1 KO mice led to similar levels of C3, C1q, and C4 deposition on WT thymocytes as WT sera (Fig. 3A–D). In contrast, WT thymocytes incubated with sera from MBL2 KO mice exhibited negligible C3 and C4 deposition, similar to WT thymocytes incubated with C3 KO sera, whereas C1q binding was unaffected (Fig. 3). Thus, although MBL1 and MBL2 are thought to be redundant (21), MBL2 is required for C3 deposition on thymocytes, whereas MBL1 is not.

FIGURE 3.

MBL2 but not MBL1 is required for C3 deposition. (AC) Examination of C3, C1q, or C4 binding on WT DP, CD4 SM, CD4 M1, CD4 M2 stages incubated with WT, C1Q KO, MBL1 KO, MBL2 KO, or C3 KO sera. Data are representative of at least four independent experiments with at least four mice in total. (D) Quantification of percentage of DP and CD4 SP thymocytes bound by C3, C1Q, or C4 in the presence of WT, C1Q KO, MBL1 KO, MBL2 KO, or C3 KO sera. Data are averaged from four mice in total from four independent experiments, and error bars represent (SEM). The p values were calculated by one-way ANOVA comparing percent bound by C3, C1q, and C4 between WT and complement KO sera in either DP or CD4 SP thymocytes.

FIGURE 3.

MBL2 but not MBL1 is required for C3 deposition. (AC) Examination of C3, C1q, or C4 binding on WT DP, CD4 SM, CD4 M1, CD4 M2 stages incubated with WT, C1Q KO, MBL1 KO, MBL2 KO, or C3 KO sera. Data are representative of at least four independent experiments with at least four mice in total. (D) Quantification of percentage of DP and CD4 SP thymocytes bound by C3, C1Q, or C4 in the presence of WT, C1Q KO, MBL1 KO, MBL2 KO, or C3 KO sera. Data are averaged from four mice in total from four independent experiments, and error bars represent (SEM). The p values were calculated by one-way ANOVA comparing percent bound by C3, C1q, and C4 between WT and complement KO sera in either DP or CD4 SP thymocytes.

Close modal

The transcriptional regulator NKAP is critical for T cell maturation (15, 18). The majority of NKAP-deficient naive peripheral T cells are RTEs. Very few NKAP-deficient MNTs remain and exhibit significant deposition of complement proteins C1q, C3, and C4 on their cell surface. Deletion of the classical pathway by crossing C1q KO and CD4-cre NKAP cKO mice did not restore the number or percent of naive CD4 T cells to normal (Fig. 4A–C) or prevent deposition of C3 on CD4+CD62L+CD44CD45RB+ MNTs (Fig. 4D, gating strategy shown in Supplemental Fig. 4). Given the observation that MBL2 was required for complement deposition on immature thymocytes incubated with serum, we next examined whether deficiency of the lectin pathway could rescue naive T cell numbers in CD4-cre NKAP cKO mice by preventing C3 deposition on MNTs. MBL1/MBL2 dKO mice were crossed to CD4-cre NKAP cKO mice to ablate the lectin pathway. In the absence of both MBL proteins, C3 deposition still occurred on NKAP-deficient MNTs in stark contrast to the observations made with developing thymocytes (Fig. 4D). The elimination of both classical and lectin pathways by crossing CD4-cre NKAP cKO mice to C1q KO MBL1/MBL2 dKO mice was required to prevent C3 deposition on NKAP-deficient MNTs (Fig. 4D). Despite prevention of C3 binding, there were no enhancement of naive peripheral NKAP-deficient T cell proportions or numbers (Fig. 4A–C). In addition, peripheral T cell numbers and proportions were not rescued when C3 KO mice were crossed to CD4-cre NKAP cKO mice. Thus, complement proteins likely act as a clearance mechanism of NKAP-deficient peripheral T cells rather than the primary cause of cell death.

FIGURE 4.

C3 deposition on NKAP-deficient naive T cells requires both the lectin and classical pathways. (A) Representative frequencies of CD4+ and CD4+CD62L+CD44 splenocytes in WT, CD4-cre NKAP cKO, MBL1 MBL2 dKO CD4-cre NKAP cKO, C1q KO CD4-cre NKAP cKO, C1q KO MBL1 MBL2 dKO CD4-cre NKAP cKO, and C3 KO CD4-cre NKAP cKO mice. (B and C) Enumeration of absolute cell counts and frequencies of CD4+CD62L+CD44 splenocytes in WT, CD4-cre NKAP cKO, MBL1 MBL2 dKO CD4-cre NKAP cKO, C1q KO CD4-cre NKAP cKO, C1q KO MBL1 MBL2 dKO CD4-cre NKAP cKO, and C3 KO CD4-cre NKAP cKO mice. (D) Enumeration of frequencies of CD4+CD62L+CD44CD45RB+ mature naive splenocytes bound by C3 from WT, CD4-cre NKAP cKO, MBL1 MBL2 dKO CD4-cre NKAP cKO, C1q KO CD4-cre NKAP cKO, C1q KO MBL1 MBL2 dKO CD4-cre NKAP cKO, and C3 KO CD4-cre NKAP cKO mice. All data are representative of at least three independent experiments with at least six mice per genotype in total. The p values were calculated by one-way ANOVA with multiple comparisons. Error bars for all figures represent SEM.

FIGURE 4.

C3 deposition on NKAP-deficient naive T cells requires both the lectin and classical pathways. (A) Representative frequencies of CD4+ and CD4+CD62L+CD44 splenocytes in WT, CD4-cre NKAP cKO, MBL1 MBL2 dKO CD4-cre NKAP cKO, C1q KO CD4-cre NKAP cKO, C1q KO MBL1 MBL2 dKO CD4-cre NKAP cKO, and C3 KO CD4-cre NKAP cKO mice. (B and C) Enumeration of absolute cell counts and frequencies of CD4+CD62L+CD44 splenocytes in WT, CD4-cre NKAP cKO, MBL1 MBL2 dKO CD4-cre NKAP cKO, C1q KO CD4-cre NKAP cKO, C1q KO MBL1 MBL2 dKO CD4-cre NKAP cKO, and C3 KO CD4-cre NKAP cKO mice. (D) Enumeration of frequencies of CD4+CD62L+CD44CD45RB+ mature naive splenocytes bound by C3 from WT, CD4-cre NKAP cKO, MBL1 MBL2 dKO CD4-cre NKAP cKO, C1q KO CD4-cre NKAP cKO, C1q KO MBL1 MBL2 dKO CD4-cre NKAP cKO, and C3 KO CD4-cre NKAP cKO mice. All data are representative of at least three independent experiments with at least six mice per genotype in total. The p values were calculated by one-way ANOVA with multiple comparisons. Error bars for all figures represent SEM.

Close modal

To determine the underlying cause for peripheral T cell deficiency in CD4-cre NKAP cKO mice, additional mechanisms of cell death and mitochondrial health were examined in CD4+CD62L+CD44 naive T cells from WT, CD4-cre NKAP cKO, and C1q KO MBL1/MBL2 dKO CD4-cre NKAP cKO mice (Fig. 5). Levels of Bcl2 expression and cleaved caspase-3 were equivalent in CD4+CD62L+CD44 naive T cells from mice of all three genotypes (Fig. 5A). Binding of YO-PRO-1, a large dye that indicates presence of pores in the cell membrane (22), was also unchanged for all three genotypes (Fig. 5B). Parameters of mitochondrial and cellular health, including presence of mitochondrial superoxide (MitoSOX), intact mitochondrial membrane potential (TMRM), cellular oxidative stress (CellROX), and mitochondrial content (MitoTracker) were examined. However, there were no notable differences between naive T cells from WT, CD4-cre NKAP cKO, or C1q KO MBL1/MBL2 dKO CD4-cre NKAP cKO mice (Fig. 5C). Although there is no difference in percentage of apoptotic (Annexin-V+) or dead (FVD+) cells within freshly harvested and stained naive T cells from CD4-cre NKAP cKO mice compared with WT mice (18), after 1 h in culture at 37°C, live naive T cells from CD4-cre NKAP cKO and C1q KO MBL1/2 KO CD4-cre NKAP cKO mice were slightly decreased in frequency, but this was NS (Fig. 5D). Finally, ferroptosis, a form of regulated cell death caused by dysregulated iron metabolism was analyzed. Iron overload results in lipid peroxidation, the defining characteristic of ferroptosis (2329). Live naive T cells from CD4-cre NKAP cKO and C1q KO MBL1/MBL2 dKO CD4-cre NKAP cKO mice had significantly increased lipid peroxidation as measured using C11-BODIPY 581/591, a lipophilic fluorescent molecule that shifts fluorescence when oxidized after reaction with peroxyl radicals (Fig. 5E). Similarly, Liperfluo, another lipophilic reactive oxygen species sensor, indicated increased hydroxyl peroxides present in NKAP-deficient T cells from complement-sufficient and -deficient backgrounds (Fig. 5F). To examine whether lipid peroxidation was reversible, WT, CD4-cre NKAP cKO, and C1q KO MBL1/MBL2 dKO CD4-cre NKAP cKO naive CD4 T cells were treated with αToc, the most abundant form of vitamin E and a lipid soluble antioxidant (27, 28). Naive CD4 T cells from CD4-cre NKAP cKO and C1q KO MBL1/MBL2 dKO CD4-cre NKAP cKO mice showed a significant decrease in oxidation of C11-BODIPY compared with untreated cells, indicating a reduction in lipid peroxidation (Fig. 5G). Next, the ferroptosis inhibitor Ferrostatin-1 was used to assess mitigation of lipid peroxidation (29). Similar to treatment with αToc, Ferrostatin-1–treated naive CD4 T cells from CD4-cre NKAP cKO, and C1q KO MBL1/MBL2 dKO CD4-cre NKAP cKO mice showed a significant decrease in oxidation of C11-BODIPY compared with untreated cells, indicating decreased lipid peroxidation. Treatment of naive CD4 T cells from WT, CD4-cre NKAP cKO, and C1q KO MBL1/MBL2 dKO CD4-cre NKAP cKO with αToc and Ferrostatin-1 further reduced accumulation of lipid peroxides, indicating synergistic antioxidant effects of both inhibitors of ferroptosis. Previously, αToc and Ferrostatin-1 were shown to reverse ferroptosis stemming from genetic ablation of GPX4 in CD4 T cells (28). GPX4 limits ferroptosis, and T cell–specific GPX4 deletion does not hinder T cell development but results in increased lipid peroxidation and peripheral T cell lymphopenia as a result of ferroptosis (28). However, GPX4 expression and intracellular GSH levels, a substrate of GPX4, did not differ between naive T cells from WT, CD4-cre NKAP cKO, and C1q KO MBL1/MBL2 dKO CD4-cre NKAP cKO mice, suggesting ferroptosis independent of GPX4 (Fig. 5H).

FIGURE 5.

NKAP-deficient naive T cells have increased lipid peroxidation. (A) Bcl2 and cleaved caspase-3 expression. (B) Levels of YO-PRO1. (C) Levels of MitoSOX, TMRM, CellROX, and MitoTracker. (D) Examination of frequencies of Annexin-VFVD (live), Annexin-V+FVD (dying), and Annexin-V+FVD+ (dead) CD4+CD62L+CD44 splenocytes from WT, CD4-cre NKAP cKO, and C1q KO MBL1 MBL2 dKO CD4-cre NKAP cKO mice after 1 h in culture. (E) Frequencies of oxidized BODIPY C-11 (Ox.BODIPY) CD4+CD62L+CD44Annexin-VFVD splenocytes (nonapoptotic and nonnecrotic naive T cells) after 1 h in culture. (F) Frequencies of Liperfluo-positive CD4+CD62L+CD44 splenocytes after 30 min in culture. (G) Frequencies of oxidized BODIPY C-11 (Ox.BODIPY) CD4+CD62L+CD44FVD splenocytes treated without and with ferroptosis inhibitors αToc (vitamin E), Ferrostatin-1, and a combination of αToc and Ferrostatin-1. (H) GPX4 expression and levels of GSH. Flow cytometry plots are representative of at least three independent experiments with at least three mice. Bar graphs show the mean of three independent experiments with at least three mice per group in total. The p values were calculated by one-way ANOVA with multiple comparisons. Error bars represent SEM.

FIGURE 5.

NKAP-deficient naive T cells have increased lipid peroxidation. (A) Bcl2 and cleaved caspase-3 expression. (B) Levels of YO-PRO1. (C) Levels of MitoSOX, TMRM, CellROX, and MitoTracker. (D) Examination of frequencies of Annexin-VFVD (live), Annexin-V+FVD (dying), and Annexin-V+FVD+ (dead) CD4+CD62L+CD44 splenocytes from WT, CD4-cre NKAP cKO, and C1q KO MBL1 MBL2 dKO CD4-cre NKAP cKO mice after 1 h in culture. (E) Frequencies of oxidized BODIPY C-11 (Ox.BODIPY) CD4+CD62L+CD44Annexin-VFVD splenocytes (nonapoptotic and nonnecrotic naive T cells) after 1 h in culture. (F) Frequencies of Liperfluo-positive CD4+CD62L+CD44 splenocytes after 30 min in culture. (G) Frequencies of oxidized BODIPY C-11 (Ox.BODIPY) CD4+CD62L+CD44FVD splenocytes treated without and with ferroptosis inhibitors αToc (vitamin E), Ferrostatin-1, and a combination of αToc and Ferrostatin-1. (H) GPX4 expression and levels of GSH. Flow cytometry plots are representative of at least three independent experiments with at least three mice. Bar graphs show the mean of three independent experiments with at least three mice per group in total. The p values were calculated by one-way ANOVA with multiple comparisons. Error bars represent SEM.

Close modal

Our data demonstrate that murine thymocytes gradually gain resistance to complement proteins as C3, C4, and C1q deposition decreases with increasing T cell maturation (Fig. 1). Cumulatively, these results provide evidence that complement resistance is achieved during maturation in the complement-privileged thymus and prior to T cell egress into blood, which harbors complement proteins. Mechanistically, the lectin pathway is required for C3 and C4 deposition on thymocytes, whereas the classical pathway is dispensable, suggesting that exposed mannose residues of developing T cells underlies complement susceptibility (Fig. 2). The lectin pathway consists of two key initiating molecules, MBL1 and MBL2. It is known that serum MBL binds to immature BALB/c thymocytes and this binding decreases during development (30), but whether MBL protein binding is the primary determinant of C3 and C4 deposition has not been previously demonstrated. Further, C3 and C4 binding are dependent on MBL2 but not MBL1 (Fig. 3). Although murine MBL1 and MBL2 are structurally homologous (21), their ability to recognize or bind to specific glycosylation patterns on thymocytes’ surface may vary as indicated by our data. Murine and primate MBL1 and MBL2 are thought to be products of a gene duplication event (21). Although the murine MBL1 and MBL2 genes encode functional proteins that are thought to be redundant, in humans only MBL2 encodes a protein product and MBL1 is silenced (21). Whether MBL2 may interact with human thymocytes is unknown. In contrast, the observation that MBL1 cannot substitute MBL2 in mice for C3 deposition on thymocytes is a novel finding and challenges the existing paradigm (21) that MBL1 and MBL2 are likely functionally interchangeable and redundant. The exclusion of complement proteins from the thymus is likely important to ensure proper T cell development and maturation, as immature thymocytes are extremely susceptible to complement-mediated attack. However, it remains unclear how the thymus achieves complement exclusion, and it is unknown why the immature thymocyte surface remains largely desialylated until the SP stage (11, 15, 16).

Our data also raise the question of functional significance of complement susceptibility of immature T cells that egress into the periphery. One possible explanation is that complement proteins selectively target T cells that fail maturation. NKAP-deficient T cells exhibit a maturation block characterized by elimination in the periphery at the RTE and MNT stages and opsonization by complement proteins (15, 18). It has previously been shown that stripping of sialic acids using neuraminidase renders erythrocytes susceptible to complement attack and consequent death, providing direct evidence of the importance of sialylation in complement protection (31). NKAP-deficient T cells fail to upregulate sialic acid transferases, suggesting faulty glycosylation as the initiator of complement deposition (15). Previously, to determine whether absence of C3 could restore long-term survival of NKAP-deficient T cells, Rag GFP WT and Rag GFP CD4-cre NKAP cKO–derived bone marrow were transferred into irradiated C3 KO or WT recipients to assess kinetics of maturation in C3-deficient and -sufficient backgrounds, respectively. Some NKAP-deficient RTEs (GFP+) were able to transition to the MNT (GFP) stage, as higher frequencies of NKAP-deficient MNTs (GFP) were found in the spleen of C3 KO recipients compared with WT recipients. NKAP-deficient RTEs also expressed increased levels of Qa2 and CD55 in C3 KO recipients compared with WT recipients, suggesting survival was prolonged (15). However, NKAP-deficient RTEs were present at a higher frequency compared with WT RTEs in C3 KO recipients, suggesting that maturation was not completely restored and transition to MNT stage was incomplete. Unforeseen effects of bone marrow transplantation on maturation or possible contribution of C3 production by hematopoietic cells could potentially halt transition of NKAP-deficient RTEs to MNTs. Crossing CD4-cre NKAP cKO mice to C3 KO mice failed to restore peripheral T cell lymphopenia and therefore prompted us to examine additional mechanisms that could explain NKAP-deficient T cell paucity.

By crossing CD4-cre NKAP cKO mice to C1q KO mice, MBL1/2 dKO mice, or C1qKO MBL1/2 dKO mice, we investigated the relative contributions of the lectin and classical pathways to C3 deposition on peripheral NKAP-deficient T cells. NKAP-deficient T cells were not restored to normal frequency or numbers in any of these crosses (Figs. 3, 4). In contrast to thymocytes incubated with serum in vitro, ablation of the lectin pathway is not sufficient to prevent C3 deposition on NKAP-deficient T cells in vivo. Rather, loss of both the classical and lectin pathway was required. Thus, combined deficiencies of C1q, MBL1, and MBL2 prevented C3 deposition but failed to rescue naive T cell persistence. This suggested that another mechanism of cell death must be the primary cause of CD4-cre NKAP cKO RTE and MNT clearance. However, the possibility of substitution of thrombin as a C5 convertase (32) or the recognition of the desialylated cell surface and engulfment by resident macrophages (10) as a mechanism of NKAP-deficient T cell clearance cannot be ruled out.

In this study, we used a number of assays to assess the metabolic health of NKAP-deficient T cells, along with alternative modes of cell death, including ferroptosis. Although parameters of mitochondrial health and apoptosis were comparable between CD4-cre NKAP cKO, C1q KO MBL1 MBL2 dKO CD4-cre NKAP cKO, and WT naive T cells, levels of lipid peroxidation were significantly increased as assessed by two different fluorescently labeled lipophilic dyes. The use of BODIPY-C11 is a well-established indicator of ferroptosis, and Liperfluo, which becomes fluorescent due to the presence of hydroxyl peroxides, confirmed that NKAP-deficient T cells have significantly increased levels of lipid peroxidation (2628). Lipid peroxidation could be reversed by the ferroptosis inhibitors αToc and Ferrostatin-1 providing further evidence that NKAP deficiency leads to oxidative stress causing enhanced lipid peroxidation, an essential feature of ferroptosis. However, the levels of neither GPX4 nor GSH, known inhibitors of ferroptosis (24), were altered in NKAP-deficient cells. Although levels of lipid peroxidation can increase due to decreased GPX4 expression or decreased levels of its cofactor GSH, multiple alternative genetic and environmental factors, such as oxidative stress, independent of GPX4 and GSH, although not overt regulators of ferroptosis, can also influence sensitivity to ferroptosis (33). For example, DP thymocytes express high levels of the cysteine/glutamate transporter, or system Xc- (SLC7A11), but this decreases to low expression in SP thymocytes and peripheral T cells (ImmGen database). System Xc- imports cystine into cells, which is then reduced to cysteine and acts in conjunction with glutamate to produce GSH, the cofactor for GPX4 important for maintaining low levels of lipid peroxides and prevention of ferroptosis (24, 26). A low level of system Xc- alone in peripheral T cells is not enough to induce ferroptosis in peripheral WT T cells, but it may synergize with NKAP deficiency to favor a ferroptotic fate. Although GPX4 deletion causes peripheral T cells to undergo ferroptosis, rate-limited enzymatic activity of GPX4 or limited GSH availability could feasibly be overwhelmed by other genes or environments that increase susceptibility to ferroptosis, including but not limited to the regulation of mitochondrial or cellular-free iron abundance by iron transporters or abundance of lipid peroxides by arachidonate lipoxygenases (ALOXs) (27). Taken together, ferroptosis is a highly complex process involving amino acid, GSH, and lipid metabolic pathways that are affected by several genes and environments (27). We propose that NKAP deficiency may prompt sensitivity to ferroptosis independent of the GSH/GPX4 pathway, as our data suggest.

Although it is known that complement proteins mediate clearance of apoptotic and necrotic cells (3), it has not been shown whether they also mediate clearance of cells undergoing ferroptosis. As NKAP-deficient T cells in complement-deficient (combined classical and lectin pathways) mice undergo ferroptosis similar to complement-sufficient mice, it is likely that complement-mediated clearance is secondary to ferroptosis. Although a causal link between ferroptosis and complement activation has not been established, both pathways have been implicated in disease settings such as cardiac and renal ischemic injuries as well as neurodegenerative disorders, suggesting co-occurrence (23, 25, 34). Interestingly, a recent study demonstrated heme-mediated complement activation on endothelial cells, raising the possibility that alterations in iron metabolism may precipitate complement attack (35).

Overall, this study presents an intriguing link between glycosylation, complement activation, and ferroptosis. Our data provide evidence that T cells gain resistance to complement as a function of maturation in the thymus, a complement-privileged organ, and that complement binding to immature thymocytes is completely dependent on MBL2. After egress, WT thymocytes are protected from complement by their sialylated cell surface. Previous studies by our laboratory demonstrated that NKAP-deficient T cells, which are less sialylated and immature after egress, are opsonized by complement in the periphery. Our present data expand on this observation and conclude that this is due to the underlying cell death process of ferroptosis, with complement aiding T cell clearance. The functional significance of MBL2’s potential interaction with developing T cells and the mechanism by which NKAP increases susceptibility to ferroptosis will be investigated in the future.

We thank Derek Glaser for assisting with Western blots for confirmation of cKO models and Michael Lehrke for critical reading of the manuscript.

This work was supported by National Institutes of Health R01 AI083279 (to V.S.S.), Mayo Graduate School funds (to B.D. and P.J.B.), and the Mayo Clinic Wilbur T. and Grace C. Pobanz Predoctoral Research Fellowship (to P.J.B.).

The online version of this article contains supplemental material.

Abbreviations used in this article:

C3

complement component 3

cKO

conditional KO

dKO

double KO

DP

double-positive

GPX4

GSH peroxidase 4

GSH

glutathione

KO

knockout

M1

mature 1

M2

mature 2

MBL

mannose-binding lectin

MNT

mature naive T cell

RTE

recent thymic emigrant

SM

semimature

SP

single-positive

αToc

α-tocopherol

WT

wild-type.

1
Delves
,
P. J.
,
I. M.
Roitt
.
2000
.
The immune system. First of two parts.
N. Engl. J. Med.
343
:
37
49
.
2
Ricklin
,
D.
,
E. S.
Reis
,
J. D.
Lambris
.
2016
.
Complement in disease: a defence system turning offensive.
Nat. Rev. Nephrol.
12
:
383
401
.
3
Sarma
,
J. V.
,
P. A.
Ward
.
2011
.
The complement system.
Cell Tissue Res.
343
:
227
235
.
4
Elvington
,
M.
,
M. K.
Liszewski
,
J. P.
Atkinson
.
2016
.
Evolution of the complement system: from defense of the single cell to guardian of the intravascular space.
Immunol. Rev.
274
:
9
15
.
5
Stephan
,
A. H.
,
B. A.
Barres
,
B.
Stevens
.
2012
.
The complement system: an unexpected role in synaptic pruning during development and disease.
Annu. Rev. Neurosci.
35
:
369
389
.
6
Stevens
,
B.
,
N. J.
Allen
,
L. E.
Vazquez
,
G. R.
Howell
,
K. S.
Christopherson
,
N.
Nouri
,
K. D.
Micheva
,
A. K.
Mehalow
,
A. D.
Huberman
,
B.
Stafford
, et al
.
2007
.
The classical complement cascade mediates CNS synapse elimination.
Cell
131
:
1164
1178
.
7
La Bonte
,
L. R.
,
B.
Dokken
,
G.
Davis-Gorman
,
G. L.
Stahl
,
P. F.
McDonagh
.
2009
.
The mannose-binding lectin pathway is a significant contributor to reperfusion injury in the type 2 diabetic heart.
Diab. Vasc. Dis. Res.
6
:
172
180
.
8
Osthoff
,
M.
,
G.
Trendelenburg
,
D. P.
Eisen
,
M.
Trendelenburg
.
2011
.
Mannose-binding lectin—the forgotten molecule?
Nat Med.
17
:
1547
1548
.
9
Ritchie
,
G. E.
,
B. E.
Moffatt
,
R. B.
Sim
,
B. P.
Morgan
,
R. A.
Dwek
,
P. M.
Rudd
.
2002
.
Glycosylation and the complement system.
Chem. Rev.
102
:
305
320, 19
.
10
Miwa
,
T.
,
L.
Zhou
,
Y.
Kimura
,
D.
Kim
,
A.
Bhandoola
,
W.-C.
Song
.
2009
.
Complement-dependent T-cell lymphopenia caused by thymocyte deletion of the membrane complement regulator Crry.
Blood
113
:
2684
2694
.
11
Daniels
,
M. A.
,
K. A.
Hogquist
,
S. C.
Jameson
.
2002
.
Sweet ‘n’ sour: the impact of differential glycosylation on T cell responses.
Nat. Immunol.
3
:
903
910
.
12
Hsu
,
F. C.
,
M. J.
Shapiro
,
B.
Dash
,
C. C.
Chen
,
M. M.
Constans
,
J. Y.
Chung
,
S. R.
Romero Arocha
,
P. J.
Belmonte
,
M. W.
Chen
,
D. C.
McWilliams
,
V. S.
Shapiro
.
2016
.
An essential role for the transcription factor Runx1 in T cell maturation.
Sci. Rep.
6
:
23533
.
13
Despont
,
J. P.
,
C. A.
Abel
,
H. M.
Grey
.
1975
.
Sialic acids and sialyltransferases in murine lymphoid cells: indicators of T cell maturation.
Cell. Immunol.
17
:
487
494
.
14
Hsu
,
F. C.
,
P. J.
Belmonte
,
M. M.
Constans
,
M. W.
Chen
,
D. C.
McWilliams
,
S. W.
Hiebert
,
V. S.
Shapiro
.
2015
.
Histone deacetylase 3 is required for T cell maturation.
J. Immunol.
195
:
1578
1590
.
15
Hsu
,
F. C.
,
M. J.
Shapiro
,
M. W.
Chen
,
D. C.
McWilliams
,
L. M.
Seaburg
,
S. N.
Tangen
,
V. S.
Shapiro
.
2014
.
Immature recent thymic emigrants are eliminated by complement.
J. Immunol.
193
:
6005
6015
.
16
Clark
,
M. C.
,
L. G.
Baum
.
2012
.
T cells modulate glycans on CD43 and CD45 during development and activation, signal regulation, and survival.
Ann. N. Y. Acad. Sci.
1253
:
58
67
.
17
Lambre
,
C. R.
,
M.
Thibon
,
S.
Le Maho
,
G.
Di Bella
.
1983
.
Auto-antibody dependent activation of the autologous classical complement pathway by Guinea-pig red cells treated with influenza virus or neuraminidase: in vitro and in vivo study.
Immunology
49
:
311
319
.
18
Hsu
,
F.-C.
,
A. G.
Pajerowski
,
M.
Nelson-Holte
,
R.
Sundsbak
,
V. S.
Shapiro
.
2011
.
NKAP is required for T cell maturation and acquisition of functional competency.
J. Exp. Med.
208
:
1291
1304
.
19
Dash
,
B.
,
M. J.
Shapiro
,
J. Y.
Chung
,
S.
Romero Arocha
,
V. S.
Shapiro
.
2018
.
Treg-specific deletion of NKAP results in severe, systemic autoimmunity due to peripheral loss of Tregs.
J. Autoimmun.
89
:
139
148
.
20
Pajerowski
,
A. G.
,
C.
Nguyen
,
H.
Aghajanian
,
M. J.
Shapiro
,
V. S.
Shapiro
.
2009
.
NKAP is a transcriptional repressor of Notch signaling and is required for T cell development.
Immunity
30
:
696
707
.
21
Seyfarth
,
J.
,
P.
Garred
,
H. O.
Madsen
.
2005
.
The “involution” of mannose-binding lectin.
Hum. Mol. Genet.
14
:
2859
2869
.
22
Fujisawa
,
S.
,
Y.
Romin
,
A.
Barlas
,
L. M.
Petrovic
,
M.
Turkekul
,
N.
Fan
,
K.
Xu
,
A. R.
Garcia
,
S.
Monette
,
D. S.
Klimstra
, et al
.
2014
.
Evaluation of YO-PRO-1 as an early marker of apoptosis following radiofrequency ablation of colon cancer liver metastases.
Cytotechnology
66
:
259
273
.
23
Miyajima
,
H.
,
S.
Kono
,
Y.
Takahashi
,
M.
Sugimoto
.
2002
.
Increased lipid peroxidation and mitochondrial dysfunction in aceruloplasminemia brains.
Blood Cells Mol. Dis.
29
:
433
438
.
24
Xie
,
Y.
,
W.
Hou
,
X.
Song
,
Y.
Yu
,
J.
Huang
,
X.
Sun
,
R.
Kang
,
D.
Tang
.
2016
.
Ferroptosis: process and function.
Cell Death Differ.
23
:
369
379
.
25
Wu
,
Y.
,
J.
Song
,
Y.
Wang
,
X.
Wang
,
C.
Culmsee
,
C.
Zhu
.
2019
.
The potential role of ferroptosis in neonatal brain injury.
Front. Neurosci.
13
:
115
.
26
Dixon
,
S. J.
,
K. M.
Lemberg
,
M. R.
Lamprecht
,
R.
Skouta
,
E. M.
Zaitsev
,
C. E.
Gleason
,
D. N.
Patel
,
A. J.
Bauer
,
A. M.
Cantley
,
W. S.
Yang
, et al
.
2012
.
Ferroptosis: an iron-dependent form of nonapoptotic cell death.
Cell
149
:
1060
1072
.
27
Stockwell
,
B. R.
,
J. P.
Friedmann Angeli
,
H.
Bayir
,
A. I.
Bush
,
M.
Conrad
,
S. J.
Dixon
,
S.
Fulda
,
S.
Gascón
,
S. K.
Hatzios
,
V. E.
Kagan
, et al
.
2017
.
Ferroptosis: a regulated cell death nexus linking metabolism, redox biology, and disease.
Cell
171
:
273
285
.
28
Matsushita
,
M.
,
S.
Freigang
,
C.
Schneider
,
M.
Conrad
,
G. W.
Bornkamm
,
M.
Kopf
.
2015
.
T cell lipid peroxidation induces ferroptosis and prevents immunity to infection.
J. Exp. Med.
212
:
555
568
.
29
Zilka
,
O.
,
R.
Shah
,
B.
Li
,
J. P.
Friedmann Angeli
,
M.
Greisser
,
M.
Conrad
,
D. A.
Pratt
.
2017
.
On the mechanism of cytoprotection by ferrostatin-1 and liproxstatin-1 and the role of lipid peroxidation in ferroptotic cell death.
ACS Cent. Sci.
3
:
232
243
.
30
Uemura
,
K.
,
Y.
Yokota
,
Y.
Kozutsumi
,
T.
Kawasaki
.
1996
.
A unique CD45 glycoform recognized by the serum mannan-binding protein in immature thymocytes.
J. Biol. Chem.
271
:
4581
4584
.
31
Durocher
,
J. R.
,
R. C.
Payne
,
M. E.
Conrad
.
1975
.
Role of sialic acid in erythrocyte survival.
Blood
45
:
11
20
.
32
Huber-Lang
,
M.
,
J. V.
Sarma
,
F. S.
Zetoune
,
D.
Rittirsch
,
T. A.
Neff
,
S. R.
McGuire
,
J. D.
Lambris
,
R. L.
Warner
,
M. A.
Flierl
,
L. M.
Hoesel
, et al
.
2006
.
Generation of C5a in the absence of C3: a new complement activation pathway.
Nat. Med.
12
:
682
687
.
33
Finkel
,
T.
,
N. J.
Holbrook
.
2000
.
Oxidants, oxidative stress and the biology of ageing.
Nature
408
:
239
247
.
34
Rocha-Ferreira
,
E.
,
M.
Hristova
.
2015
.
Antimicrobial peptides and complement in neonatal hypoxia-ischemia induced brain damage.
Front. Immunol.
6
:
56
.
35
Merle
,
N. S.
,
R.
Paule
,
J.
Leon
,
M.
Daugan
,
T.
Robe-Rybkine
,
V.
Poillerat
,
C.
Torset
,
V.
Frémeaux-Bacchi
,
J. D.
Dimitrov
,
L. T.
Roumenina
.
2019
.
P-selectin drives complement attack on endothelium during intravascular hemolysis in TLR-4/heme-dependent manner.
Proc. Natl. Acad. Sci. USA
116
:
6280
6285
.

The authors have no financial conflicts of interest.

Supplementary data