Article Text
PDF
Abstract
Objectives Low-density granulocytes (LDGs) are a distinct subset of proinflammatory and vasculopathic neutrophils expanded in systemic lupus erythematosus (SLE). Neutrophil trafficking and immune function are intimately linked to cellular biophysical properties. This study used proteomic, biomechanical and functional analyses to further define neutrophil heterogeneity in the context of SLE.
Methods Proteomic/phosphoproteomic analyses were performed in healthy control (HC) normal density neutrophils (NDNs), SLE NDNs and autologous SLE LDGs. The biophysical properties of these neutrophil subsets were analysed by real-time deformability cytometry and lattice light-sheet microscopy. A two-dimensional endothelial flow system and a three-dimensional microfluidic microvasculature mimetic (MMM) were used to decouple the contributions of cell surface mediators and biophysical properties to neutrophil trafficking, respectively.
Results Proteomic and phosphoproteomic differences were detected between HC and SLE neutrophils and between SLE NDNs and LDGs. Increased abundance of type 1 interferon-regulated proteins and differential phosphorylation of proteins associated with cytoskeletal organisation were identified in SLE LDGs relative to SLE NDNs. The cell surface of SLE LDGs was rougher than in SLE and HC NDNs, suggesting membrane perturbances. While SLE LDGs did not display increased binding to endothelial cells in the two-dimensional assay, they were increasingly retained/trapped in the narrow channels of the lung MMM.
Conclusions Modulation of the neutrophil proteome and distinct changes in biophysical properties are observed alongside differences in neutrophil trafficking. SLE LDGs may be increasingly retained in microvasculature networks, which has important pathogenic implications in the context of lupus organ damage and small vessel vasculopathy.
- lupus erythematosus
- systemic
- inflammation
- cardiovascular diseases
Statistics from Altmetric.com
Key messages
What is already known about this subject?
Low-density granulocytes (LDGs) are a subset of neutrophils expanded in systemic lupus erythematosus (SLE). These cells have been shown to have a pathogenic role through their enhanced ability to form neutrophil extracellular traps, promote type I interferon responses and damage the vasculature. Their levels and gene signature associate with enhanced vasculopathy and atherosclerosis in patients with lupus.
What does this study add?
The findings from this study indicate that lupus LDGs display distinct proteomic and biomechanical properties that may impact their ability to travel through the vasculature, interact with the endothelium and enhance their trapping in the small vessels of various organs.
How might this impact on clinical practice or future developments?
Increased retention of lupus LDGs in microvasculature could have pathogenic implications in lung or kidney damage, and in development of small vessel vasculopathy. These results suggest that development of therapeutics modulating neutrophil biomechanical properties could modulate deleterious responses in lupus and other autoimmune diseases.
Introduction
Neutrophil dysregulation may play critical roles in systemic lupus erythematosus (SLE) pathogenesis.1 Enhanced release of neutrophil extracellular traps (NETs)—the externalisation of oxidised nucleic acids and granule proteins—promotes immune dysregulation, vasculopathy and organ damage associated with SLE.2–5
We previously identified a subset of SLE proinflammatory neutrophils (low-density granulocytes, LDGs), purified from the peripheral blood mononuclear cell (PBMC) layer.6 In contrast to normal dense neutrophils (NDNs), LDGs spontaneously form proinflammatory NETs7 8 induce endothelial damage,6 and associate with in vivo vascular inflammation, coronary atherosclerosis,5 9–11 and T cell activation,12 suggesting they play important roles in SLE pathogenesis.
Previous LDG studies focused on transcriptomic analysis, with little known about proteome modulation and protein function.8 9 13–16 Proteomic analyses comparing SLE LDGs to SLE and healthy control (HC) NDNs identified differential phosphorylation of proteins associated with cytoskeletal organisation. Using real-time deformability cytometry (RT-DC) and a polydimethylsiloxane (PDMS) device mimicking neutrophil trafficking through the pulmonary microvasculature, we determined that SLE LDGs are biophysically distinct from other neutrophil subsets, which may affect their ability to traffic through small blood vessels.
Results
Differential protein profiles of lupus and HC neutrophils
Proteomic/phosphoproteomic analyses were performed in SLE LDGs and NDNs, and HC NDNs (n=5/group; online supplemental tables 1 and 2). As controls, HC NDNs were also analysed following priming with N-formylmethionine leucyl-phenylalanine (fMLF), given that priming decreases HC NDN density.17 Neutrophil preparations used identical protocols optimised to minimise biophysical or functional disruption of cells from their unstimulated state in whole blood (online supplemental figure 1).
Neutrophil mass spectrometry analysis identified 4109 proteins (figure 1A), of which 601 (14.6%) and 685 (16.6%) were identified only in HC or SLE neutrophils, respectively (online supplemental figure 2A,B). This is comparable to the most robust neutrophil proteomic analysis previously reported.18 Results were aligned with SLE LDG, NDN, and HC NDN transcriptomics (GEO GSE139358)16 to identify proteins not present at the mRNA level that may be of exogenous source (online supplemental figure 3). SLE LDGs and NDNs showed complete proteome overlap, although with considerable variation in protein abundance. Indeed, 9.4% of proteins expressed by SLE neutrophils were differentially abundant in SLE LDGs versus NDNs, with 270 more abundant and 60 less abundant (ratio cut-off >1.5 or<0.5 in at least 4/5 matched samples; figure 1C). Of the 2823 proteins common to both SLE and HC NDNs, 304 (10.7%) showed differential abundance. FMLF-primed and unstimulated HC NDNs showed complete proteome overlap with little variation in protein abundance, except for decreased abundance of L-selectin in primed HC NDNs, suggesting protein shedding from in vitro activation19 (online supplemental figure 2D–H). Overall, many proteins were uniquely present in either HC or SLE neutrophils and protein abundances varied between subsets, indicating neutrophil proteome heterogeneity.
Systemic lupus erythematosus (SLE) normal dense neutrophils (NDNs) differ in their proteome and phosphoproteome compared with healthy control (HC) NDNs. (A) A total of 4109 proteins and (B) 875 phosphoproteins were identified by mass spectrometry in low-density granulocytes (LDGs) and NDNs from subjects with SLE (n=5) and unstimulated and primed NDNs from HC volunteers (n=5). Volcano plots depict differences between SLE NDN and LDG proteomes (C) and phosphoproteomes (D). The upregulated (red) and downregulated (green) proteomes are in LDGs, while NDNs are the reference proteome. (E) Gene ontology biological process analysis highlighting biological networks associated with proteins more abundant in at least 4/5 SLE NDN samples relative to HC NDNs. Proteins with abundance ratios greater than 1.5 were included and significance was established by false discovery rate. (F) Proteins responsible for upregulation of networks associated with neutrophil activation in SLE NDNs relative to HC NDNs in arbitrary units. Significance was established by Kruskal-Wallis test with post hoc Dunn’s tests for multiple comparisons (three comparisons) or by Mann-Whitney U test (two comparisons). (G) Relative abundance of interferon-inducible proteins in SLE NDNs and HC NDNs relative to SLE LDGs. SLE NDNs were compared with autologous SLE LDGs. HC NDNs were compared with the mean protein abundance in SLE LDGs. Open boxes in heatmaps indicate the given protein was not identified in the sample. (H) Abundance of cell integrins and adhesion-related proteins in SLE NDNs and HC NDNs relative to autologous SLE LDGs and autologous primed HC NDNs, respectively. (I) Abundance of phosphoproteins associated with regulation of neutrophil–endothelial interactions in all neutrophil subsets, in arbitrary units. Significance was established by Kruskal-Wallis test with post hoc Dunn’s tests for multiple comparisons. All results are mean±SEM and significance was set at *p≤0.05, **p≤0.01, ns=not significant.
We identified 875 proteins phosphorylated on serine, threonine, and/or tyrosine residues in neutrophils (figure 1B). Some proteins were phosphorylated at multiple sites (online supplemental table 3). Of these phosphoproteins, 48 (5.4%) and 366 (41.8%) were only identified in HC NDNs and SLE neutrophils, respectively. The same phosphoproteins were identified in HC unstimulated and fMLF-primed NDNs, and one phosphoprotein was uniquely identified in SLE LDGs (round spermatid basic protein 1-like protein, pRSBN1L; online supplemental figure 2C). When comparing SLE LDGs and NDNs, 95 phosphoproteins (11.5%) were differentially abundant, with 11 less and 84 more abundant in SLE LDGs (figure 1D). Of the 509 phosphoproteins coexpressed in fMLF-primed and unstimulated HC NDNs, 167 (32.8%) were differentially abundant (online supplemental figure 2E). Of the 460 phosphoproteins common to all neutrophils, 100 (21.7%) were differentially abundant between HC and SLE NDNs (online supplemental figure 2G). These data support neutrophil phosphoproteome heterogeneity.
The LDG proteome displays a distinct profile
Using ShinyGO20 and MetaScape,21 we mapped proteins differentially abundant in at least 4/5 samples to known gene-ontology biological processes. Proteins more abundant in SLE NDNs relative to HC NDNs mapped to neutrophil activation networks, including proteins facilitating migration to inflammatory sites and release from bone marrow.22–24 Some proteins associated with neutrophil activation were most abundant in SLE LDGs (figure 1E,F).
SLE subjects express elevated type 1 IFN-stimulated genes (ISGs) in various organs and cells, including NDNs and LDGs.16 25 While ISG-encoded proteins were not uniformly upregulated in SLE NDNs versus HC NDNs, many were upregulated in SLE LDGs relative to SLE or HC NDNs (figure 1G). ISG transcription is mediated by phosphorylation of signal transducer and activator of transcription (STAT) molecules26 but we did not detect phospho-STATs, possibly because pTyr residues are less abundant than pSer.27 Limited LDG numbers prevented immunoprecipitation of pTyr residues alongside phosphopeptide enrichment. Collectively, the SLE neutrophil proteome suggests an activated status, while the IFN-associated protein signature is distinct to SLE LDGs.
Neutrophil priming/activation facilitates interactions with the endothelium.28 There were no differences in adhesion molecule or integrin expression among neutrophil subsets. However, phosphoproteins regulating neutrophil–endothelial interactions were more abundant in fMLF-primed HC NDNs than other neutrophil subsets (figure 1H,I). This suggests differences between SLE LDGs/NDNs and fMLF-primed HC NDNs.
Proteins with differential phosphorylation in SLE NDNs versus HC NDNs were associated with organelle organisation and actin cytoskeletal organisation, including phospho-coronin 1A (pCORO1A) and phospho-heat shock protein 90AA1 (pHSP90AA1; online supplemental figure 2H). Some proteins less abundant in SLE LDGs versus SLE NDNs associated with neutrophil degranulation but key granule proteins, including myeloperoxidase and cathepsin-G, were not decreased (figure 2A,B). Rather, lower abundance of membrane proteins, particularly ficolin-1-rich granule membrane proteins, accounted for downregulated degranulation-associated networks in SLE LDGs. Differences in degranulation capabilities did not explain changes in the neutrophil proteome between neutrophil subsets.
Systemic lupus erythematosus (SLE) low-density granulocytes (LDGs) have a distinct proteomic profile characterised by enhanced pathways associated to translational activity, intracellular trafficking and type I interferon-induced protein pathways. (A) Gene ontology biological process analysis highlighting biological networks associated with proteins less abundant in SLE LDGs (n=5) relative to SLE normal dense neutrophils (NDNs; n=5). Proteins with abundance ratios less than 0.5 in at least 4/5 matched samples were included and significance was established by false discovery rate (FDR). (B) Relative abundance of degranulation network-associated proteins in SLE NDNs and healthy control (HC) NDNs relative to SLE LDGs. SLE NDNs were compared with autologous SLE LDGs. HC NDNs were compared with the mean protein abundance in SLE LDGs. (C) Gene ontology biological process analysis highlighting biological networks associated with proteins more abundant in at least 4/5 SLE LDG samples relative to SLE NDNs. Proteins with abundance ratios greater than 1.5 were included and significance was established by FDR. (D) Relative abundance of SLE network-associated proteins in SLE NDNs and HC NDNs relative to SLE LDGs. SLE NDNs were compared with autologous SLE LDGs. HC NDNs were compared with the mean protein abundance in SLE LDGs. (E) Relative abundance of eukaryotic translation network-associated proteins in SLE NDNs and HC NDNs relative to SLE LDGs. SLE NDNs were compared with autologous SLE LDGs. HC NDNs were compared with the mean protein abundance in SLE LDGs. Open boxes in heatmaps indicate the given protein was not identified in the sample. (F) Relative abundance of coagulation and platelet network-associated proteins in SLE NDNs and HC NDNs relative to SLE LDGs. SLE NDNs were compared with autologous SLE LDGs. HC NDNs were compared with the mean protein abundance in SLE LDGs. (G) Gene ontology biological process analysis highlighting biological networks associated with phosphoproteins differentially abundant in SLE LDGs and NDNs. Phosphoproteins with abundance ratios less than 0.5 or greater than 1.5 in at least 4/5 matched samples were included and significance was established by FDR. (H) Abundance of transcription factors CEBPD and SPI1 in arbitrary units. CEBPD not identified in HC NDNs. Results are mean±SEM, with comparisons between autologous SLE LDG and NDNs. Significance established by Kruskal-Wallis test with post hoc Dunn’s tests for multiple comparisons and set at *p≤0.05, ns=not significant. N.F.=not found.
Abundant proteins in SLE LDGs versus SLE NDNs clustered in neutrophil activation, coagulation, platelet and intracellular trafficking networks (figure 2C). The SLE biological network (false discovery rate=10−11.119) was upregulated in SLE LDGs versus autologous NDNs, primarily driven by complement proteins (figure 2D). Immunoglobin chains and apolipoproteins were more abundant in SLE LDGs versus other neutrophils. Differential phosphorylation in SLE LDGs versus NDNs also associated with neutrophil activation and intracellular trafficking. In addition, SLE LDGs expressed higher abundances of ribosomal proteins (figure 2E–G).
SLE LDGs are a heterogeneous group comprising CD10− (immature, less abundant) and CD10+ (intermediate-mature, most abundant) subsets. CD10− LDGs have decreased CEBPD and SPI1 transcripts relative to SLE NDNs and CD10+ LDGs.16 Proteomic analysis was completed on unfractionated SLE LDGs and displayed similar SPI1 protein abundance across neutrophil subsets. CEBPD was not identified in HC NDNs but was similar in SLE LDGs and NDNs (figure 2H). Most SLE LDGs had multilobulated nuclei (online supplemental figure 1C), confirming intermediate-mature cells represent the most abundant LDG subset.16
SLE LDGs and NDNs differ in expression of cytoskeleton-associated proteins
Consistent with evidence that cytoskeleton-associated transcriptional networks are enhanced in SLE LDGs,8 we found upregulation at the protein level when compared with autologous NDNs (figure 3A). Many proteins differentially expressed and/or phosphorylated in LDGs regulate intracellular trafficking (figure 3B,C).29–33 In addition, we assessed modulation of the HC NDN phosphoproteome by fMLF.17 34 35 Many proteins differentially phosphorylated in fMLF-primed HC NDNs were associated with cytoskeletal organisation (figure 3D,E). Upregulation of cytoskeleton-associated networks in the SLE LDG proteome, alongside phosphoproteomic findings suggestive of differential cytoskeletal reorganisation among neutrophil subsets, prompted investigation of neutrophil biomechanical properties.
Proteomic and phosphoproteomic analyses indicate differential expression of proteins associated to cytoskeleton function between systemic lupus erythematosus (SLE) normal dense neutrophils (NDNs) and low-density granulocytes (LDGs). (A) Gene ontology biological process analysis highlighting biological networks related to the cytoskeleton and associated with proteins more abundant in at least 4/5 SLE LDG samples relative to their matched SLE NDNs. Proteins with abundance ratios greater than 1.5 were included and significance was established by false discovery rate (FDR). (B) Abundance of cytoskeleton-associated proteins in SLE NDNs and HC NDNs relative to SLE LDGs. SLE NDNs were compared with autologous SLE LDGs. Healthy control (HC) NDNs were compared with the mean protein abundance in SLE LDGs. Open boxes in heatmaps indicate that the given protein was not identified in one of the two autologous samples being compared. (C) Abundance of cytoskeleton network-associated phosphoproteins in SLE LDGs relative to abundance in autologous SLE NDNs. (D) Gene ontology biological process analysis highlighting biological networks associated with phosphoproteins differentially abundant in primed HC NDNs (n=5) and unstimulated HC NDNs (n=5). Proteins with abundance ratios greater than 1.5 or less than 0.5 in at least 4/5 matched samples were included and significance was established by FDR. (E) Abundance of phosphoproteins regulating the cytoskeleton in primed HC NDNs relative to autologous HC NDNs.
Neutrophil biomechanical properties are altered in clinically active SLE
RT-DC is a high-throughput technique that analyses biomechanical properties of thousands of cells in suspension.36 An inverted microscope with a high-speed camera captures images of individual cells moving through a narrow constriction channel within a PDMS microfluidic chip, where cells are deformed by hydrodynamic shear stress. Cell tracing algorithms generate biomechanical profiles per cell, including cell size (cross-sectional area), roughness (cell surface perturbations quantified by dividing the convex hull area by the cross-sectional area), and deformability (one minus the value of circularity within a constriction channel). Cell populations are identified in blood by size and brightness (figure 4A).37 These measurements were obtained in peripheral blood from HC (n=11), clinically quiescent (n=11) or clinically active SLE (n=4). In some experiments, fMLF was added directly to HC peripheral blood to prime neutrophils prior to analysis. Neutrophils from HC and clinically quiescent SLE were biomechanically identical, while active SLE neutrophils had larger areas, enhanced deformability and roughness. FMLF-primed HC neutrophils were also larger and more deformable than unstimulated neutrophils and significantly rougher than any unstimulated neutrophil subsets (figure 4B). Overall, neutrophils from active SLE subjects displayed altered biomechanical properties and cell membrane perturbations.
Systemic lupus erythematosus (SLE) low-density granulocytes (LDGs) are biomechanically rougher than other neutrophil subsets by real-time deformability cytometry (RT-DC). (A) RT-DC was used to biomechanically characterise the shape, size and deformability of neutrophil subsets. (B) Biomechanical profiling of neutrophils in blood samples obtained from healthy volunteers (n=12), clinically quiescent patients with SLE (n=9), and active patients with SLE (n=4) by RT-DC. In some instances, 100 nM N-formylmethionine leucyl-phenylalanine (fMLF) was used to prime neutrophils within the healthy blood. (C) Percentage of LDGs identified in SLE (n=6) and healthy control (HC) peripheral blood mononuclear cells (PBMCs; n=6) by RT-DC. (D) LDGs as a percentage of total neutrophils in patients with SLE (n=6), HC volunteers (n=6), and fMLF-primed HC blood (n=6). (E) Biomechanical profiling of isolated NDNs and LDGs from HC volunteers (n=11) and clinically quiescent patients with SLE (n=11) by RT-DC. Some HC NDNs were primed with fMLF prior to isolation. For each sample analysed by RT-DC, the median measurement of over 500 neutrophils is graphed and the mean±SEM for each neutrophil subset is depicted. Autologous unstimulated/primed HC NDNs or autologous SLE NDN/LDGs were compared and significance was established by Wilcoxon matched-pairs signed rank tests. In other comparisons, significance was established by Mann-Whitney U tests. Significance was set at *p≤0.05, **p≤0.01, ns=not significant. (F) Images of NDNs and LDGs captured during RT-DC, representative of >500 images of each neutrophil subset from n=11 patients with SLE and n=11 HC volunteers, 10× objective. White arrows identify a concave cell surface feature common in primed HC NDNs and an irregular protrusion in the cell surface common in SLE LDGs. (G) Brightfield microscopy of NDNs and LDGs (n=3). White arrows identify rounded, protruded, membrane features observed in nearly 100% of primed HC NDNs and an irregular protrusion observed in ~30% of SLE LDGs. See online supplemental videos 1–4.
Supplementary video
Supplementary video
Supplementary video
Supplementary video
SLE LDGs and NDNs are biomechanically distinct
Biomechanical properties of purified SLE LDGs/NDNs from clinically quiescent subjects and HC NDNs were quantified, using optimised purification strategies to avoid disruption of biomechanical properties (online supplemental figure 1). Gating strategies allowed for identification of neutrophils, monocytes, lymphocytes and eosinophils in mixed cell fractions (figure 4A).37 Biomechanical properties did not differ in lymphocytes and monocytes between SLE and HC subjects (online supplemental figure 4). In contrast, SLE LDGs displayed distinct biomechanical features relative to other neutrophil subsets (figure 4E). While HC and SLE NDNs were round and smooth, SLE LDGs had significantly rougher cell surfaces that correlated with age but not with other clinical/demographic characteristics (online supplemental figure 5). HC NDNs incubated for various time-points with Sm/RNP immune complexes7 and/or recombinant IFN-α displayed no changes in neutrophil roughness (online supplemental figure 6). Overall, SLE LDGs display distinct biomechanical properties seemingly unrelated to exposure to immune complexes or type I IFNs.
Neutrophil percentages were higher in SLE than HC PBMC fractions (figure 4C,D), consistent with higher LDG numbers.9 It is unclear whether LDGs are present in small numbers in healthy individuals but expanded in SLE.38 We compared biomechanical properties of SLE LDGs to the small population of HC LDGs. Like autologous HC NDNs, and not consistent with the SLE LDG biomechanical phenotype, HC LDGs displayed smooth, non-polarised surfaces. HC LDGs were also more deformable than HC NDNs (figure 5C). These differences in biomechanical properties support SLE LDGs do not represent expansion of a minor LDG population found in HCs.
Systemic lupus erythematosus (SLE) low-density granulocytes (LDGs) are increasingly retained within a three-dimensional pulmonary microvasculature mimetic, but do not display enhanced adherence to endothelial cells in a two-dimensional system of flow. (A) A branching microfluidic mimetic of the pulmonary microvasculature was designed. Arrows indicate the inlet, outlet and a cell navigating the network, 10× objective. (B) Retention of normal dense neutrophils (NDNs) and LDGs in the microvasculature mimetic (n>150 neutrophils per subset from n=6 healthy control (HC) volunteers and n=6 patients with SLE). Seconds until release refers to the amount of time each neutrophil was retained within the mimetic, measured from entry at the inlet to exit at an outlet. Seconds until release was recorded as >120 s if cells did not exit the mimetic within the 2 min video. Times were determined manually with a timer superimposed on the video during data collection. Significance was determined by log-rank test. (C) Transit times through the microvasculature mimetic for all NDNs and LDGs navigating the entire device. Significance assessed by Mann-Whitney U test. (D) Retention of HC NDNs treated with dimethyl sulfoxide or cytochalasin D in the microvasculature mimetic (n>50 neutrophils from n=3 HC volunteers). Significance was determined by log-rank test. (E) Percentage of neutrophils interacting with endothelium under 0.4 mL/min flow. Significance was determined by Kruskal-Wallis test with post hoc Dunn’s tests for multiple comparisons. (F) Light microscopy of neutrophil binding to endothelium post-flow assay. Arrows show enhanced binding of primed NDNs. Images representative of n=4 images obtained for each neutrophil subset isolated from n=6 patients with SLE or n=6 HC volunteers. All results are mean±SEM with significance was set at *p≤0.05, **p≤0.01, ****p≤0.0001, ns=not significant.
While fMLF-primed HC NDNs are morphologically rougher17 and localise to the PBMC interphase on density gradients (figure 4D),17 they were consistently larger than autologous unprimed NDNs. This contrasts with SLE LDGs, which were similar in size to autologous SLE NDNs (figure 4E), supporting primed HC NDNs and SLE LDGs are biomechanically distinct. These biomechanical differences were confirmed by brightfield (figure 4F,G) and lattice light-sheet fluorescence microscopy (online supplemental file 1). The cell surface of fMLF-primed HC NDNs appeared to ruffle, with small membrane perturbations moving inwards and outwards. In contrast, SLE LDGs’ cell surface was smooth, except for sections of dramatic protrusions, which were irregularly shaped (figure 4F,G). This suggests SLE LDGs have distinct biophysical properties not consistent with acutely primed phenotypes.
SLE LDGs are retained in a microfluidic microvasculature mimetic (MMM)
Neutrophil biomechanical properties can modulate transit through the pulmonary microvasculature. Primed neutrophils are retained in pulmonary capillary beds,39 possibly due to enhanced cell stiffness and/or irregular cell shape.40 To mimic trafficking through the pulmonary microvasculature, we developed an MMM formed of a branched pyramidal network within a PDMS chip (figure 5A). Neutrophils flowed through this network at physiologically relevant pressures (10 and 50 mbar, or 10.2 and 51.0 cmH2O, respectively)40 without impact on viability (online supplemental figure 8). As previously reported,39 40 fMLF-primed HC NDNs were increasingly retained in the MMM, with >80% unable to fully navigate it. In contrast,>80% unprimed HC NDNs navigated the MMM within 3 s. Trafficking patterns of SLE LDGs resembled those of primed HC NDNs, with >75% SLE LDGs retained in the MMM versus approximately 50% SLE NDNs (figure 5B). Of neutrophils transiting the entire MMM, SLE and unprimed HC NDNs averaged a transit time of <0.9 s, while SLE LDGs and primed HC NDNs averaged transit times of 1.87 and 2.89 s, respectively (figure 5C).
HC NDNs treated with cytochalasin D, which disassembles filament actin and decreases neutrophil deformability,36 were increasingly retained in the MMM (>75% retention vs <20% in vehicle-treated HC NDNs (figure 5D)), supporting biomechanical modulation alters neutrophil trafficking. Like primed HC NDNs,39 SLE LDGs may be preferentially retained in microvasculature due to biomechanical property differences.
The MMM evaluated effects of cellular biomechanical properties on trafficking but not the putative role of neutrophil–endothelial interactions. By decoupling effects of biophysical properties and cell-surface markers on neutrophil trafficking, we evaluated their independent contributions. A two-dimensional assay evaluated neutrophil interactions—rolling alongside or adherence to microvascular endothelium—in a circulatory flow system mimicking physiological conditions (flow rate 0.4 mL/min). Over 3 min, 15% and 60% primed HC NDNs interacted with unstimulated or stimulated endothelium, respectively. In contrast,<10% and<20% HC NDNs, SLE LDGs and NDNs interacted with unstimulated and stimulated endothelium, respectively (p<0.01 compared with primed HC NDNs and p>0.05 comparing other neutrophil subsets; figure 5E,F; online supplemental file 1). These observations suggest that, while enhanced neutrophil–endothelium interactions may contribute to microvasculature retention of primed HC NDNs, they do not explain differences in microvasculature trafficking observed between SLE LDGs and NDNs. Overall, SLE LDGs may be retained in microvasculature networks,39 by intrinsic changes in cellular biomechanical properties rather than by specific neutrophil–endothelium interactions.
Discussion
We identified significant differences between the SLE and HC neutrophil proteomes as well as heterogeneity in the proteome of SLE neutrophils including proteins involved in formation/rearrangement of the cytoskeleton. In addition, we identified biomechanical differences in SLE LDGs with implications for neutrophil trafficking in the microvasculature.
Consistent with the proteomics data and previous transcriptomic analyses reporting differential gene expression associated with the actin cytoskeleton in SLE LDGs,8 we found SLE LDGs are biomechanically distinct and showed cell membrane perturbations differing from fMLF-primed neutrophils.17 41 While mechanisms promoting enhanced SLE LDG cytoskeletal changes remain unclear, differential abundance of proteins associated with extracellular structure organisation and cytolysis may be implicated. For example, profilin 1 (PFN1) modulates actin/microtubule dynamics30 42 and actin polymerisation,43 while histidine rich glycoprotein (HRG) induces neutrophil morphological changes29 implicated in neutrophil retention in microvasculature.29 44 Furthermore, the enhanced ability of LDGs to form NETs may contribute to cytoskeleton perturbations and disruptions in cell membrane integrity.15 45–47 Although SLE LDGs did not morphologically resemble HC neutrophils treated with Sm/RNP immune complexes or phorbol myristate acetate (PMA) to induce NETosis (online supplemental figures 6 and 7), differences in spontaneous LDG NET formation and PMA-induced NET formation have been reported.7 8 48 The potential link between NET formation, neutrophil proteome and biomechanical properties of LDGs should be studied further.
Previous studies indicate that rougher, primed neutrophils are retained in the lungs.29 40 Our MMM data suggest that LDG roughness may similarly hinder LDGs’ ability to traffic through narrow capillaries and biophysical properties, not enhanced binding to endothelium.40 Increased retention in microvasculature networks could have pathogenic implications in lung or kidney damage, and in development of small vessel vasculopathy. SLE lung manifestations are associated with blood vessel damage triggered by neutrophils.49–51 Circulating immune complexes can activate neutrophils, promote endothelial cell barrier dysfunction and perturbed vascular permeability.52 53 While distinct biomechanical properties of SLE LDGs did not align with preferential binding to microvascular endothelial cells, LDGs have potent deleterious effects on endothelium through NET formation.7 54 55 Accordingly, we propose a model where slow LDG microvasculature transit, coupled to enhanced NETosis, promotes vasculopathy. Future studies should assess mechanisms of enhanced SLE LDG roughness and in vivo significance of its effect on LDG trafficking.
In contrast to SLE LDGs, fMLF-primed NDNs showed enhanced adherence to endothelium and higher abundance of phosphoproteins linked to cell adhesion.56–59 Actin-regulatory proteins are dephosphorylated in LDGs but phosphorylated in primed neutrophils, suggesting both actin depolymerisation and polymerisation may induce biophysical changes affecting trafficking. Indeed, imaging showed primed NDNs with contracted cortical actin rings while LDGs appeared irregularly shaped with incomplete actin rings (online supplemental figure 7, online supplemental videos 1–4). Overall, SLE LDGs differ from acutely primed neutrophils and interact with the vasculature differently.
The type I IFN pathway is linked to SLE pathogenesis and neutrophils responding to these cytokines exhibit proinflammatory responses.4 ISG-encoded proteins were higher in SLE LDGs, consistent with transcriptome reports.16 Why SLE LDGs express higher ISG-encoded proteins than autologous SLE NDNs, exposed to similar levels of cytokines in vivo, could be related to differences in JAK-STAT activity or to differences in activation status.16 SLE LDGs have enhanced ISG hypomethylation relative to HC neutrophils, perhaps modulating the protein response.60 Future studies should address how enhanced IFN responses modulate pathogenic differences linked to LDGs’ ability to NET and damage vasculature.
The SLE LDG proteome contained increased acute phase response proteins associated with complement and coagulation.61 Corroborating our findings, LDGs display significantly enhanced transcription of several complement components (online supplemental figure 3). Some complement proteins identified by proteomics were not identified by transcriptomics. These proteins, including C6–C9, may bind to circulating neutrophils. This aligns with findings of C6–C9 contributing to formation of MAC-induced lytic pores in rheumatoid arthritis neutrophils.62 Additionally, activated HC NDNs upregulate C3 transcription (online supplemental figure 10), suggesting activated LDGs may behave in a similar manner. This LDG–complement relationship should be investigated further.
Some proteins associated to platelet biology were more abundant in SLE LDGs, similar to descriptions in psoriasis LDGs.63 This was confirmed by fluorescence microscopy (online supplemental figure 1) and suggests commonalities in the proteome of LDGs across inflammatory diseases associated with enhanced vascular damage. Platelet–neutrophil interactions can drive inflammation and thrombosis64; thus, increased platelet presence in LDG samples may contribute to their upregulation of coagulation and some neutrophil activation-associated proteins relative to NDNs. Ultimately, LDG–platelet interactions may play distinct pathophysiological roles in vasculopathy development.
Alongside the proteomics, the biomechanical profile and trafficking pattern of SLE LDGs support reports that LDGs represent a distinct neutrophil subset rather than expansion of immature/primed neutrophils present in healthy subjects.15 65 66 SLE LDGs have a distinct proteomic signature and specific biomechanical features impacting transit through the microvasculature. This study adds to the understanding of neutrophil heterogeneity in the context of blood vessel trafficking, with important implications for development of small vessel vasculopathy and organ damage and development of therapeutics modulating neutrophil biomechanical properties.67
Acknowledgments
The authors thank Jane Hollis, Cecelia Matara and Frederica Mescla for their assistance in drawing clinical blood samples.
References
Footnotes
Handling editor Josef S Smolen
Twitter @ajtwood
ERC and CS contributed equally.
Funding This research was supported by the Intramural Research programs at NIAMS (ZIA-AR041199), NHLBI and NIBIB. ACM is supported by a Clinical Research Career Development Fellowship from the Wellcome Trust (WT 2055214/Z/16/Z). CS is funded by Medical Research Council, Wellcome Trust, British Heart Foundation, Glaxo Smith Kline, Astra Zeneca and NIHR Cambridge Biomedical Research Centre. ERC receives funding from MRC, Wellcome Trust, GlaxoSmithKline, the NHLI Foundation and the NIHR Imperial Biomedical Research Centre. AJTW was supported by a Gates Cambridge Scholarship. KRB was also supported by a National Institutes of Health OxCam Scholarship.
Competing interests None declared.
Patient consent for publication Not required.
Ethics approval All studies were approved by site-specific IRBs: the Cambridge Local Research Ethics Committee (REC reference 06/Q0108/281) and NIAMS/NIDDK IRB (NIH 94-AR-0066). All subjects signed informed consent.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD021096 and 10.6019/PXD021096. Transcriptomics data are in GEO database GSE139358