The antibodies to CD4 FITC and CD14 BV711 were obtained from Biolegend (San Diego, CA, United States). The antibodies to integrin β1 were obtained from BD Bioscience (San Jose, CA, United States). GILZ monoclonal antibody (CFMKG15) PE, rat IgG2a isotype control PE were from eBioscience (San Diego, CA, United States). Natalizumab was obtained from Novus Biologicals (Centennial, CO, United States). ELISA kits to TNF-α, IL-6, IL-1β, IL-8, IFN-γ were from R&D SYSTEMS (Minneapolis, MN, United States). ELISA kits to C-X-C motif chemokine 12 (CXCL12), chemokine (C-C motif) ligand 5 (CCL5), chemokine (C-C motif) ligand 7 (CCL7), and chemokine (C-C motif) ligands 4 (CCL4) were from LSBio (Seattle, WA, United States). Human CD14 cell isolation Kit was from Miltenyi Biotec Inc. (Auburn, CA, United States). EasySep Human naïve CD4 T cell isolation kit was from STEMCELL Technologies Inc. (Seattle, WA, United States). The recombinant cytokines of IL-1β, IL-6, IL-8, IFN-γ, IL-4, GMCSF, and TNF-α were obtained from Sino Biological (Wayne, PA, United States). The cyclic peptide inhibitor of integrin α4β1 (azide modified peptide: MePhe-Leu-Asp-Val-Aib-Lys) (Yu et al., 2020) control peptides (cyclic MePhe-Ala-Ala-Ala-Aib-Lys), and peptide drug conjugates (PDC), cyclic peptide inhibitor of integrin α4β1 conjugated methylprednisolone (α4β1/Methrol), and control peptide/methrol were synthesized using solid-supported chemistry by InnoPep Inc (San Diego, CA, United States). Click-iT Cell Reaction Buffer Kit, Alexa 488 alkyne, Alexa 555 conjugate of wheat germ agglutinin (WGA-555), and Alexa 488 conjugate of wheat germ agglutinin (WGA-488) were purchased from Invitrogen (San Diego, CA, United States). LPS and all other chemical reagents were obtained from Sigma (St. Louis, MO, United States).

Peripheral blood samples were collected from healthy adult donors. Ethical approval was obtained from the Hunan Children Hospital, China (Ethical approval number: HCHLL-2021-30). All donors signed a written informed consent. Whole blood of healthy donors was withdrawn via puncture of a medial cubital vein. To prepare RBC samples, blood from two donors were combined as a sample, isolated red cells were washed twice in phosphate-buffered saline (PBS) containing 2% fetal bovine serum (FBS) with buffy coat removal, and leukocyte depletion filter (BioR, Fresenius Kabi, Bad Homburg, Germany). Subsequently, cells were resuspended in PBS, counted, and then kept on ice for 1 h. For preparation of RBCs conditioned medium samples, RBCs were exposed to different concentrations of LPS for various periods of time. To prepare large volume hemolytic samples, 5 ml of 1 × 10⁹ RBCs/ml were used in assay. After being exposed to LPS, RBCs suspensions were centrifuged at 1,500×g for 10 min. The supernatants were harvested and filtered, designated here as control sup (C-sup, no treatment) or LPS-treated RBCs sup (L-sup) for downstream experiments.

Peripheral blood mononuclear cells were isolated with Ficoll-Pague gradient centrifugation from the donors’ peripheral blood. CD14 cells were purified with the CD14 cell isolation kit (negative selection) following the instruction manual provided by Miltenyi Biotec Inc. The naïve CD4 T cells were purified with EasySep Human Naïve CD4⁺ T Cell Isolation Kit following the manufacturer’s protocol (STEMCELL Technologies Inc.). The purity of CD14 or CD4⁺ T cells was confirmed by flow cytometric analysis.

Primary human lung microvascular endothelial cells were purchased from Lonza (HMVEC-L, CC-2527, Basel, Switzerland). The HMVEC-L cells were maintained in endothelial basal medium-2 (EBM-2, CC-3156) supplemented with endothelial growth medium (EGM-2) MV BulletKit from Lonza Clonetics (CC-3202) at 37°C and 5% CO₂, until achieving a 70–90% confluence. The cells were then treated with Trypsin-EDTA (Lonza) and seeded in assay plates. The cells at 3–8th passages were used for the experiments.

Ex vivo hemolysis assay was performed essentially as described by Karsten et al. (Brittain et al., 2004). Human RBCs were washed twice in PBS and resuspended in RPMI-1640 medium (Sigma) at a density of 1 × 10⁹ RBCs/mL at 4°C. One hundred microlitre of RBCs exposed to LPS (0, 5, and 50 μg/ml) in RPMI-1640 medium supplemented with 1% penicillin-streptomycin (HyClone Laboratories, Logan, UT, United States) and incubated at 37°C for 24–48 h. The supernatants were collected or stored at −80°C until measurements for free hemoglobin protein (HGP) concentration, plasma activity of lactate dehydrogenase (LDH), and the levels of cytokines and chemokines. Hemolytic activity was assessed using levels of HGP by blood cell analyzer (Sysmex 800i, Japan) and LDH by automatic biochemical analyzer (Beckmancounter AU5800, United States) at the Laboratory of Hunan Children Hospital Center, China.

6 × 10⁶ isolated CD14⁺ monocytes were cultured in 3 ml/well of 6-well plate in complete medium (RPMI 1640, 2 mM L-glutamine, 1% autologous plasma, 1% penicillin-streptomycin) supplemented with 250 IU/ml IL-4 and 800 IU/ml GM-CSF. Cells were incubated at 37°C and 5% CO₂ for 2 days. 1.5 ml/well of cells were resuspended in complete medium with the 2-fold concentration of IL-4 and GM-CSF and then put back to the original culture. Cells were incubated at 37°C and 5% CO₂ for 3 days. At day 6, cells were resuspended in 1.5 ml (per well) medium supplemented with 2,000 IU/ml IL-6, 400 IU/ml IL-1β, 2,000 IU/ml TNF-α, and 2 μg/ml PGE₂ and placed the supplemented resuspension back into the original culture and incubated for another 3 days. The phenotypes of mo-DCs were assessed by flow cytometry on day 10. The morphology of mo-DCs was defined by a light microscope image (Keyence microscope BZ-X800).

We recently reported a single cell suspension self-aggregation assay (Yu et al., 2020). In this study, the protocol was modified to test multiple cell suspensions for aggregation in 96-well plate. Briefly, different dyes of Hoechst, WGA-555, and WGA-488 were used to label the different live cells. Each cell suspension was immediately mixed into wells of 96-well plate with a total volume of 100 µl per well. Cell densities, medium volumes, incubation times, and treatment schedules were optimized for different experiments. Cell-cell suspensions in plates were placed on a microplate shaker to shake orbitally for 30–60 min followed by 10 s at 500 rpm to ensure that the cell aggregates and existing single cells within the wells were processed uniformly. This procedure also eliminated the non-specific cell clusters that would mechanically be dissociated into single cells while cell-cell aggregates remained in wells. Afterward, the plates were left on the bench for 20 min and subjected to imaging captures by fluorescence microscopy (Keyence BZ-X800) or cells were fixed by addition of 50 µl 3× prefer fixative for further image capture (Anatech Ltd. sop 410, Battler Creek, MI, United States). Values calculated by ImageJ (National Institutes of Health (NIH), Bethesda, MD, United States) represented the mean ± standard deviation (SD) of at least six independent experiments.

Monocyte derived dendritic cell migration assay was performed in a Boyden chamber system (Millipore QCM 24-well, Temecula, CA, United States). mo-DCs were suspended in assay medium (serum-free RPMI 1640 containing 0.05% BSA). 800 μl of indicated sample was placed in the lower chamber. To prepare the testing samples in the lower chamber, RBC supernatants were filtered using 0.45 µm filter (Millipore) to remove RBCs after being pre-treated with LPS (50 μg/ml) for 48 h in culture (L-sup) or medium as control supernatant (C-sup). 400 µl of each supernatant was diluted in 800 µl assay medium in well of 24 well plate in the lower chamber. 200 µl of 2 × 10⁵ mo-DCs labeled with 0.1 μg/ml of Hoechst in assay medium was added into the upper chamber and then incubated for 6 h. Afterwards, the plate was put in a microplate shaker to shake orbitally for 10 s at 500 rpm to ensure that the cells in wells were processed uniformly. 200 μl of cells from the low chamber well was transferred to 96 well plate and shaken for 10 s at 500 rpm and then Hoechst positive cells were counted under a Keyence fluorescence microscope (BZ-X800 Series, Itasca, IL, United States). Values calculated by ImageJ represented the mean ± SD of at least three independent experiments.

One millilitre of 1 × 10⁹/ml RBCs were exposed to LPS or PBS for 48 h. Cells were washed once using PBS and resuspended in PBS. Cells were stained with FITC anti-human β1 antibody and FITC mouse IgG1 isotype control for 60 min on ice. Cells were washed twice with 100 µl of PBS and then resuspended in 200 µl PBS for flow cytometric analysis. Values calculated by mean fluorescence intensity (MFI) represented the mean ± SD of at least three independent experiments.

Levels of cytokines (TNF-α, IL-6, IL-1β, IL-8, IFN-γ) in RBCs supernatants were quantified using a commercial ELISA Kit according to the manufacturer’s guidelines (R&D system, Minneapolis, MN, United States). Chemokines (CXCL12, CCL5, CCL7, and CCL4 were quantified in RBCs supernatants using commercial ELISA Kit following the manufacturer’s recommendations (LSBio, Seattle, WA, United States). Human VCAM-1 was quantified in RBCs supernatants using commercial ELISA Kit following the manufacturer’s recommendations (LSBio, Seattle, WA, United States). Red blood cells were kept at a density of 5 × 10⁸ cells per well in 96-well plate in a total volume of 200 µl PBS in the presence of LPS (0, 50 μg/ml) in 37°C, 5% CO₂ incubator for 48 h. The supernatants were harvested after centrifugation at 1,000 × g for 15 min and stored at −80°C until use for the cytokines ELISA assay. To determine cytokine concentration, each sample was diluted in PBS at different ratios for optimization. Concentration of each cytokine or chemokine was determined from a linear regression standard curve using the standard protein controls provided in the kit. Data represented the mean ± SD of three independent experiments.

To prepare RBCs/mo-DCs culture, 1 ml of 1 × 10⁹ RBCs were exposed to LPS (0, 50 μg/ml) in well of 24-well plate in PBS at 37°C and 5% CO₂ for 48 h. Afterwards, 50 µl of RBCs suspension was added into well of 96-well plate containing 2 × 10⁴ mo-DCs in 100 µl complete medium and co-cultured at 37°C and 5% CO₂ overnight. The next day, 5 × 10⁶ purified naive CD4⁺ T cells were suspended in 400 µl PBS, and labeled with 100 µL of a 10 µM CellTrace Violet solution (Life Technologies) for 5 min at room temperature (RT). Subsequently, the cells were washed three times with 2 ml MLR medium (RPMI 1640, 2 mM L-glutamine, nonessential amino acids, 0.1 mM sodium pyruvate, 5% human AB serum). Finally, the CD4⁺ T cells were suspended in MLR medium at a density of 5 × 10⁵ cells/ml. Subsequently, the labeled T cells (100 µl per well) were added and cultured for 5 days at 37°C, 5% CO₂. Proliferation of CD4⁺ T cells was determined by measuring the fluorescence of CellTrace Violet by flow cytometry. Cell debris and dead cells were excluded from the analysis by scatter signals and propidium iodide (PI) fluorescence.

CD4 naïve T cells were seeded in six-well plates at 2 × 10⁶ cells per well and then treated with medium only, RBCs supernatant only, mo-DCs supernatant only, LPS only, and LPS pre-treated RBCs/mo-DCs supernatant, respectively. The pretreated CD4 naïve T cells were transfected for 48 h with NF-κB luciferase reporter vector and a negative control vector provided from the NF-κB reporter kit (BPS Bioscience, San Diego, CA, United States). Cells were harvested, lysed, and prepared before subjected to the dual luciferase reporter assay (Promega, Madison, WI, United States). Luciferase activity was measured using a luminometer (Tecan, Mannedorf, Switzerland) and relative luciferase activity was calculated as the fold change over the unstimulated vehicle control.

One millilitre of 1 × 10⁹ RBCs was exposed to LPS (0, 50 μg/ml) in well of 24-well plate in PBS at 37°C and 5% CO₂ for 48 h. Afterward, 50 µl of RBCs were added into well of 96-well plate containing 2 × 10⁴ mo-DCs in 100 µl complete medium and co-cultured at 37°C and 5% CO₂ overnight. Subsequently, the α4β1/Methrol (10 µM) and controls (100 µl per well) were added and incubated on ice for 60 min, then washed with complete medium and cultured overnight at 37°C, 5% CO₂. Afterwards, non-nucleated erythrocytes were lysed with 1-step Fix/Lyse solution at 37°C for 20 min while preserving leukocyte population. Cells were washed twice with 100 μl of Intracellular Staining Perm Wash Buffer (Biolegend), centrifuged at 350 × g for 5 min, permeabilized by exposure to 100% methanol for 90 min on ice, and washed twice with PBS. Samples were stained with anti-GILZ PE (1:200) at RT for 60 min. Samples were washed with PBS and 100 μL PBS added to each well for flow cytometric analysis.

We previously reported that the cyclic peptide inhibitor, peptide C (MePhe-Leu-Asp-Val-Aib-Lys) could block integrin α4β1 mediated CD4 T cell self-aggregation by HMGB1 or LPS stimulation (Yu et al., 2020). This α4β1 inhibitor binds VCAM-1 via the “Leu-Asp-Val” tripeptide (LDV) (Wayner and Kovach, 1992; Kaneda et al., 1997; Cringoli et al., 2020). The cyclic peptide inhibitor of integrin α4β1 conjugated to Methrol (α4β1/Methrol) was synthesized using solid-supported chemistry by InnoPep Inc (San Diego, CA, United States). This conjugate had been used to target integrin α4β1 for blocking RBCs mediated cell-cell aggregation in our previous study (Yu et al., 2020). In this study, we further explored whether the molecule could reverse the immune cell disorder in hemolysis.

All data were presented as mean ± SD. Comparison among groups was performed by Kruskal–Wallis one-way analysis of variance. Significant differences were noted at p < 0.05.

To investigate LPS-induced hemolysis, we used red cell suspension only, which was absent of plasmatic blood components to exclude the other effectors such as complement systems and leukocytes that may contribute to hemolysis. As shown in Figure 1A,B, a dark red color was observed in the wells of RBCs treated with LPS at 50 μg/ml for 24 h, 5 μg/ml for 48 h or 50 μg/ml for 48 h compared to PBS-treated control cells. After LPS exposure, free HGP concentrations were significantly increased at both 5 and 50 μg/ml of LPS for 48 h (p < 0.001) (Figure 1C), and free LDH activities were dramatically increased for all the three treatment schedules (p < 0.001) (Figure 1D). These data provided the clear evidence that LPS evoked hemolysis in vitro. Our data were consistent with a previous study (Brauckmann et al., 2016).

To prepare cell suspension, mo-DCs were generated from CD14⁺ monocytes after the addition of a cytokine cocktail for 10 days. Monocyte-derived dendritic cells were identified by phenotypic maturation markers such as CD83, CD11c, HLA-DR, and CD86 (all >95% purity) using flow cytometry (Supplementary Figure S2A) and by cell morphology (Supplementary Figure S2B). The purity of CD14 and CD4 naïve T cells isolated from PBMC was greater than 85 and 95%, respectively (Supplementary Figure S2C). To evaluate cell-cell aggregation in a quantitative manner, we developed imaging protocol to visualize and analyze their morphologies by using the Keyence fluorescence microcopy assisted with ImageJ software. After pre-labelling RBCs with WGA-555 and mo-DCs nuclei with Hoechst, the clustered area was determined by ImageJ software and RBCs/mo-DCs aggregates were defined and counted when the area was greater than 350 μm² (Figure 2A). RBCs mixed with mo-DCs could form RBCs/mo-DCs aggregation, whereas LPS pre-treated RBCs formed larger cell aggregates with mo-DCs, which could be reduced by Integrin α4β1 peptide inhibitor. Figure 2B shows the strategy for dynamic imaging analysis for mixture of RBCs, mo-DCs, and HLMVEC. RBCs, mo-DCs, and HLMVEC were pre-labeled with WGA-555, WGA-488, and Hoechst dye, respectively. Based on cell area and combination of each color intensity, imaged objects in wells were characterized to define the diversity of cell aggregates by Image-Pro Plus software. Based on image three color fluorescence, cell aggregate objects were gated and counted as different aggregate events (event-1: red and blue color threshold for RBCs/mo-DCs; event-2: red and blue colors threshold for RBCs/HLMVEC aggregates; and event-3: red, green, and blue colors threshold for RBCs/mo-DCs/HLMVEC aggregate). Figure 2C left two panels shows image data calculation and analysis for cell aggregation. The mo-DCs per se were able to form cell self-aggregation. Compared to mo-DCs self-aggregation baseline, LPS alone or untreated RBCs did not significantly induce mo-DCs aggregation (p > 0.05). To our surprise, LPS treated RBCs dramatically increased RBCs/mo-DCs cell aggregation compared to the controls (mo-DC alone, LPS + mo-DCs, and RBCs/mo-DCs) (p < 0.05). In addition, cyclic α4β1 peptide inhibitor significantly blocked RBCs/mo-DCs aggregation induced by LPS. To further evaluate multiple cell interaction, we identified three types of cell aggregation in suspension. To test whether LPS pre-treated RBCs could affect aggregation to mo-DCs and/or HLMVEC, a multiple-cell aggregation study was performed. As expected, LPS pre-treated RBCs facilitated aggregation of mo-DCs, HLMVEC or mo-DCs/HLMVEC as compared to RBCs controls (p < 0.05). In the same well, the total single cell counts dropped dramatically (data not shown). The results revealed that the LPS-induced hemolysis could significantly facilitate both mo-DCs and HLMVEC aggregation. Since integrins mediate immune cell interaction and migration and up-regulation of integrin α4β1 induced aggregation of mononuclear cells and RBCs in sickle cell disease, we sought to determine whether the integrin α4β1 on RBCs is a mediator of cell-cell aggregation in hemolytic condition. As shown in Figure 2C middle panel flow cytometry picture, % of integrin α4β1 positive RBCs was increased dramatically on LPS treated RBCs compared with the controls (0.01, Figure 2C indicated panel) by the flow cytometric analysis. Of note, natalizumab, a humanized monoclonal antibody against α4 integrin that is FDA-approved medication used to treated multiple sclerosis and Crohn’s disease (Sellner and Rommer, 2019) had an activity in blocking RBCs/mo-DCs aggregation to an extent comparable to that of the cyclic peptide inhibitor of integrin α4β1 (Figure 2C last right panel). Taken together, these results confirmed that the integrin α4β1 on RBCs in LPS-induced hemolysis facilitated cell-cell aggregation.

Chemokines are essential for mo-DCs migration. Several studies have implicated the role of DCs in progression and destabilization of the atherosclerotic plaque (Yao et al., 2011; Rai et al., 2016). A variety of factors including LPS, inflammatory cytokines, ligation of selected cell surface receptors and viral products can induce DC maturation toward a DC profile capable of inciting primary T cell responses (Yao et al., 2011). Having demonstrated that LPS could evoke hemolysis, we next sought to determine whether the hemolysis would affect mo-DCs migration. As shown in Figure 3A, the filtered L-sup induced mo-DCs migration in a concentration-dependent fashion. Figure 3B shows that a panel of cytokines that include TNF-α, IL-1β, IL-6, IL-8, IFN-r, chemokine CXCL12, CCL5, CCL7, and CCL4 were significantly increased compared to the control groups (*p < 0.05, **p < 0.01, ***p < 0.001). Therefore, RBCs lysis by LPS could release chemokines or cytokines to induce mo-DCs migration. Intriguingly, vascular cell adhesion molecule 1 (VCAM-1), known to be an endothelial ligand for integrin α4β1 that could mediate the adhesion of lymphocytes, monocytes, eosinophils, and basophils to vascular endothelium, was also increased in the filtered L-sup.

Given that the RBCs are the reservoirs of cytokines and chemokines, we further investigate whether RBCs in hemolysis would affect immune T cell proliferation and signaling activation. In MLR assay, naïve CD4 T cells isolated from healthy PBMC donors were co-cultured with mo-DCs in the presence of RBCs treated with LPS for five days. Figure 4A shows the representative flow cytometry data in which a new generation of CD4 T cells were stained relatively weak by Celltrace Violet. The representative result demonstrated that the culture of mo-DCs with LPS pre-treated RBCs stimulated CD4 T cell proliferation (35.56% vs. 21% in the untreated control group). Consistent with a previous study (Zinser et al., 2019), mo-DCs with LPS pre-treated RBCs significantly induced CD4 T cell proliferation (Figure 4B). To elucidate a mechanism responsible for the CD4 T cell activation, the NF-κB Reporter kit designed for monitoring the activity of the NF-κB signaling pathway in the cultured cells was employed. Figure 4C shows that LPS pre-treated RBCs co-cultured with mo-DCs significantly increased the NF-κB-stimulated luciferase activity (p < 0.01 compared to medium, RBCs alone and mo-DCs alone; p < 0.05 compared to LPS alone). These results suggest that RBCs in hemolysis may affect mo-DCs function and stimulate naïve T cell proliferation via activation of the NF-κB signaling pathway.

Based on the observation that RBCs in hemolysis could evoke mo-DCs dysfunction, we try to formulate a strategy by using the conjugate of α4β1 integrin inhibitor and methylprednisolone to reverse the hemolysis-induced abnormal cell-cell aggregation, migration, T cell immunity and NF-κB signaling. Figure 5A shows that LPS pre-treated RBCs significantly promoted cell-cell aggregation between RBCs/mo-DCs, RBCs/HLMVEC, and RBCs/mo-DCs/HLMVEC, all of which could be blocked by α4β1/Methrol. As expected, the α4β1 inhibitor/Methrol significantly inhibited mo-DCs migration induced by the hemolytic supernatant (Figure 5B). In the mo-DCs migration assay, while LPS alone could induce mo-DCs migration the supernatant from LPS treated RBCs produced a larger effect on the cell migration (*p < 0.05). Moreover, it also dose-dependently inhibited mo-DCs migration initially promoted by L-sup (Figure 5C). Notably, the flow cytometric analysis showed that the α4β1 inhibitor/Methrol dramatically upregulated GILZ expression in the LPS pre-treated RBCs and mo-DCs cultures (Figure 5D,E, p < 0.05). Thus, the conjugate of cyclic peptide inhibitor of integrin α4β1 and Methrol could reverse mo-DCs dysfunctions caused by the hemolysis.