We performed a secondary analysis of prospectively collected data and blood specimens from a clinical study performed between August 2018 and December 2019 (Uhlich et al., 2020). The original study included adult trauma patients (>18 years) presenting to the University of Alabama at Birmingham Hospital as a level I trauma with suspected hemorrhagic shock, as determined by an Assessment of Blood Consumption score ≥2 (Nunez et al., 2009). Prisoners and pregnant patients were excluded. Consent was obtained from the patient or a legally authorized representative within 24 h of hospital arrival. Healthy adult volunteers with no history of bleeding disorders or use of aspirin or non-steroidal anti-inflammatory drugs within 48 h of sampling were included as control subjects. The University of Alabama at Birmingham Institutional Review Board approved this study (protocol # 300001642 for healthy controls and 300001292 for trauma).

Demography (e.g., age, sex), injury characteristics [e.g., injury severity score (Baker et al., 1974); severe head injury as defined by an abbreviated injury score of ≥3 for the head and neck body region (Gennarelli and Wodzin, 2008); penetrating versus blunt mechanism; shock index as defined by the initial heart rate divided by the systolic blood pressure value (Allgower and Burri, 1967)] laboratory values [e.g., serum base excess, serum lactate, plasma international normalized ratio (INR), and plasma activated partial thromboplastin time (aPTT)], and clinical outcome data were recorded from the electronic medical record. Blood samples collected upon hospital arrival and 24 h after hospital arrival were used for analysis. Blood specimens were collected into sodium-citrated tubes and immediately centrifuged at 1,500 rcf for 15 min (room temperature) twice to deplete plasma of platelets. Platelet-poor plasma was then aliquoted and stored at −70°C for future analysis.

Total levels of HS, Angpt-2, syndecan-1 and thrombomodulin were measured in available plasma using commercially available enzyme-linked immunosorbent assays (ELISA) according to manufacturer specifications (human heparan sulfate ELISA, E-EL-H2364, Elabscience, Houston, TX, United States; human angiopoietin-2 DuoSet ELISA, DY623, R&D Systems, Minneapolis, MN, United States; human syndecan-1 ELISA, Amsbio, E01S0301, Cambridge, MA, United States; human thrombomodulin ELISA, DTHBD0, R&D Systems).

Heparanase activity was measured in randomly selected plasma samples collected from 10 trauma subjects at hospital arrival and from 10 controls using a homogeneous time resolved fluorescence (HTRF) assay according to manufacturer specifications (Cisbio, Bedford, MA, United States) and as previously described (Enomoto et al., 2006). See Supplementary Methods for further detail.

Two hundred microliters of platelet-depleted plasma was processed as previously described (Wang et al., 2022) to isolate heparin lyase I,II-digested HS disaccharides and tetrasaccharides for analysis (Table 1). HS di-/tetrasaccharides were then analyzed against ¹³C-labeled HS di-/tetrasaccharide standards using liquid chromatography-tandem mass spectrometry (LC-MS/MS) as described previously (Wang et al., 2022). See Supplementary Methods for further detail.

Primary human lung microvascular ECs (HLMVEC) (S540-05a, Cell Applications, San Diego, CA, United States) were conditioned under shear stress (15 dyn/cm²) for 48 h as described previously (Richter et al., 2022a). After 48 h of flow conditioning, HLMVECs were exposed to vehicle (1× phosphate buffered saline) or heparinase III (500 mU/mL) (P0737L, New England BioLabs, Ipswich, MA, United States) for 6 h while 15 dyn/cm² shear stress was maintained. Primary HLMVEC were chosen as a representative microvascular EC population that demonstrates flow responsive gene expression and mimics a physiologic microvascular environment (Richter et al., 2022a). Heparinase III is a Bacteroides-derived heparin lyase that functions similarly to human heparanase by selectively hydrolyzing the β1,4 glycosidic bond between glucosamine and hexuronic acid residues in regions of N- and 6-O-sulfation within HS (Desai et al., 1993; Peterson and Liu, 2010; Zhu et al., 2020) and, unlike human heparanase, retains enzymatic activity at the neutral pH of cell growth medium (Pikas et al., 1998; Sotnikov et al., 2004; Peterson and Liu, 2012). Heparinase III concentration of 500 mU/mL was informed by our previous work demonstrating near-complete loss of HS 10E4 staining within the eGC at this concentration (Richter et al., 2022a). After the 6-h treatment, flow was discontinued, and cell monolayers were washed thrice with 1× phosphate buffered saline prior to RNA harvest. See Supplementary Methods for further detail.

Total RNA was isolated from HLMVEC lysate using an RNAspin mini RNA isolation kit (25-0500-72, Cytiva, Marlborough, MA, United States) according to manufacturer standard procedure. Cell lysate from two slides was combined into a single sample to ensure robust RNA quantity, which generated 2 RNA samples per treatment group for bulk sequencing. RNA libraries were developed using the NEBNext Ultra II Directional RNA-Seq library prep kit (E7760S, New England Biolabs, Inc., Ipswich, MA, United States) per the manufacturer’s instructions. The resulting libraries were quantitated using the KapaBiosystems qPCR quantitation kit (MilleporeSigma, Burlington, MA, United States) using standard techniques. Sequencing was performed on the Illumina NextSeq 500 (Illumina, Inc., San Diego, CA, United States) following the manufacturer’s protocols.

STAR (version 2.7.7a) was used to align raw RNA-Seq FASTQ reads to the human reference genome (GRCh38 p13, release 36) from Gencode (parameters used in STAR: –outReadsUnmapped Fastx –outSAMtype BAM SortedByCoordinate –outSAMattributes All) (Dobin et al., 2013). Following alignment, HTSeq-count (version 0.11.1) was used to estimate the transcript abundances for each gene from the STAR alignment files (parameters used in HTSeq-count: -m union -r pos -t exon -i gene_id -a 10 -s no -f bam) (Anders et al., 2015). Normalization and differential expression was then applied to the count files using DESeq2, following the default settings in the DESeq2 case vignette (Love et al., 2014).

RNA-Seq analysis was performed using RStudio in the R environment (version 4.2.1). An unbiased approach was used at the outset to report the top 40 differentially expressed genes by the false discovery rate (FDR)-adjusted probability (p) value (FDR q value). pheatmap was used to generate the resulting heatmap, and EnhancedVolcano was used to generate the volcano plot. Untargeted differential gene expression pathway analysis was performed using Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) datasets. Gene set enrichment analysis was performed using the Database for Annotation, Visualization, and Integrated Discovery (DAVID) 2021 (version 2023q4) (Huang da et al., 2009; Sherman et al., 2022). Inclusion criteria for pathway analysis included gene overlap counts ≥10 and FDR q value <0.05. We also took a targeted approach to characterize genes associated with EC function and eGC expression that demonstrated an FDR q value <0.05.

Given the nonparametric distribution of clinical and biomarker data with sample sizes below the threshold for eliciting the Central Limit Theorem, descriptive statistics are presented as median (interquartile range) or number (percentage). Differences between two groups were determined using the Mann-Whitney U test, and differences between three groups were determined using the Kruskal-Wallis 1-way ANOVA followed by Dunn’s multiple comparisons test. Associations between two continuous variables were determined with the Spearman’s rank correlation, and the Fieller, Hartley and Pearson method was used to estimate variance (reported as the 95% confidence interval, or 95% CI). A 2-sided alpha level of 0.05 was used to determine significance. Statistical analyses were performed using IBM^® SPSS Statistics (version 28.0, Chicago, IL, United States) or GraphPad Prism (version 9.3.1, Boston, MA, United States).

For the RNA-Seq analysis, adjusted p values (q values) for individual gene expression compared between treatment groups were determined using a Benjamini-Hochberg correction to account for the FDR. Statistical comparisons were performed using DESeq2.

Forty-nine trauma subjects were enrolled in the original study, and 25 healthy controls were included in the current analysis. There were no differences in age or sex between controls and trauma subjects (Table 2). As previously reported (Uhlich et al., 2020), 37% of injuries resulted from blunt mechanisms, and 5 (10%) trauma subjects suffered severe head injury. Twenty-eight (57%) patients presented in hemorrhagic shock (defined as a serum base excess ≤−6 mMol/L or lactate ≥2.5 mMol/L in a patient who required transfusion with ≥2 units of packed red blood cells within 4 h of hospital admission), and 16 (33%) were coagulopathic (defined as INR >1.2 or aPTT ≥36.5 s) (Uhlich et al., 2020). Other injury characteristics not previously reported are presented in Table 2.

Total HS levels and heparanase activity were increased in trauma plasma at hospital arrival relative to controls (Figures 1A,B), suggesting enzymatic shedding of HS resulting from traumatic injury. Cleavage of HS was dependent on injury severity as indicated by a moderate direct correlation between circulating levels of HS at arrival and injury severity score [Spearman’s rho = 0.428 (n = 49; 95% CI = 0.158, 0.638), p = 0.002]. Twenty-four hours after injury, total levels of HS in trauma plasma were comparable to controls.

Ten controls and 20 trauma subjects had available plasma for LC-MS/MS evaluation (Supplementary Table S1). Sulfation modifications were detected in circulating HS collected from trauma samples relative to controls at both hospital arrival and at 24 h (Figure 1C). Specifically, expression of tetra-2 and tetra-4 sulfation declined immediately following trauma concomitantly with a rise in ΔIIS and ΔIIIS disaccharides. By 24 h following trauma, ΔIIIS levels remained elevated and tetra-2 and tetra-4 remained low in trauma subjects. Furthermore, we observed a significant increase in the expression of tetra-1 24 h following trauma relative to hospital arrival.

Modifications to HS observed in trauma samples moderately correlated with injury characteristics. We observed that the molar percentage of ∆IVA HS disaccharides present in trauma plasma at hospital arrival correlated with shock index, a physiologic measure of tissue perfusion [rho = 0.478 (n = 20; 95% CI = 0.033, 0.765), p = 0.033], whereas the level of tetra-4 inversely correlated with shock severity as indicated by serum lactate levels [rho = −0.587 (n = 20; 95% CI = −0.822, −0.182), p = 0.006]. Interestingly, increased level of tetra-1 at 24 h was specific to trauma subjects who suffered penetrating injuries as this tetrasaccharide was not elevated at 24 h in subjects who suffered blunt injuries (Supplementary Figure S1). No other mechanism-specific differences in HS di-/tetrasaccharide levels were observed (Supplementary Figure S1). In subjects who suffered penetrating injuries, 24-h tetra-1 levels inversely correlated with 24-h INR [rho = −0.879 (n = 10; 95% CI = −0.972, −0.543), p < 0.001] and lactate levels [rho = −0.799 (n = 10; 95% CI = −0.952, −0.321), p = 0.006].

HS has been reported as a measure of eGC injury and vascular dysfunction in many pathologic conditions, including trauma (Rahbar et al., 2015; Torres Filho et al., 2016; Halbgebauer et al., 2018). To expand upon this prior work, we evaluated associations between HS and other established markers of eGC damage following trauma that could support HS levels as a marker of eGC injury. In our trauma population, total HS levels at hospital arrival moderately correlated with eGC levels of syndecan-1 [rho = 0.339 (n = 48; 95% CI = 0.053, 0.575), p = 0.018] and soluble thrombomodulin [rho = 0.352 (n = 49; 95% CI = 0.069, 0.581), p = 0.013] (Supplementary Figure S2).

We have previously observed in two separate trauma populations that the vascular endothelium is pathologically activated after traumatic injury, as indicated by elevations in circulating levels of Angpt-2 (Uhlich et al., 2020; Richter et al., 2022b). Moreover, we found that circulating levels of syndecan-1, an established marker of eGC injury following trauma (Gonzalez Rodriguez et al., 2017), rise before significant elevations in Angpt-2 (Richter et al., 2022b). Building upon our prior findings that HS shedding from the endothelial surface promotes Angpt-2 production (Richter et al., 2022a), we next evaluated associations between circulating HS (total levels and relative levels of di-/tetrasaccharides) at arrival and Angpt-2 levels after 24 h. Data from control subjects were included in these analyses. Consistent with our previous observations in children with sepsis in which systemic inflammation and EC activation are manifest (Richter et al., 2022a), total HS levels at arrival weakly correlated with Angpt-2 levels at 24 h [rho = 0.299 (n = 73; 95% CI = 0.066, 0.499), p = 0.011] (Figure 2A). We also observed a moderate correlation between heparanase activity levels at arrival and 24-h Angpt-2 levels [rho = 0.605 (95% CI = 0.193, 0.835), p = 0.006] in 19 subjects for whom both heparanase activity and Angpt-2 levels were available (Supplementary Figure S3). Moreover, we observed a significant, though moderate, correlation between 24-h Angpt-2 levels and arrival levels of ΔIIIS disaccharides [rho = 0.477 (n = 30; 95% CI = 0.129, 0.720), p = 0.008], and significant, though moderate, inverse correlations between 24-h Angpt-2 levels and arrival levels of tetra-2 [rho = −0.597 (n = 30; 95% CI = −0.792, −0.291), p < 0.001] and tetra-4 [rho = −0.400 (n = 30; 95% CI = −0.671, −0.036), p = 0.028] (Figures 2B–D). While correlative, these findings may point to early HS structural modifications and/or cleavage in the eGC as a contributing process in pathological EC activation in traumatically injured subjects.

Relative to other organ systems, HS expression is enriched within the lungs and, as an integral constituent of the eGC, serves key functions in regulating vascular homeostasis a (LaRiviere and Schmidt, 2018). Systemic inflammatory insults resulting from trauma and hemorrhagic shock are known to cause pulmonary eGC damage and microvascular injury in association with coagulation abnormalities and organ dysfunction (Abdullah et al., 2021). To gain mechanistic insight into the downstream consequences of eGC damage, we investigated how cleavage of HS from the eGC affects the transcriptional landscape of MVEC isolated from human lungs. HS plays a pivotal role in maintaining shear stress-mediated vascular EC homeostasis (Richter et al., 2022a), and therefore, we conditioned HLMVEC to physiologic shear stress for 48 h prior to treatment with heparinase III to enzymatically cleave HS from the cell surface. As previously observed (Richter et al., 2022a), HLMVEC exposure to laminar flow at 15 dyn/cm² resulted in cytoskeletal reorganization and cellular alignment relative to statically cultured cells (Supplementary Figure S4). We identified 13,712 unique protein-encoding genes (13,742 unique EnsembID values) within HLMVEC following treatment with either vehicle or heparinase III [bulk RNA-Seq data are available on the Gene Expression Omnibus (GEO) database, accession number GSE260628]. One thousand three hundred seventy-five genes were upregulated and 1,286 genes were downregulated [FDR-adjusted p-value (FDR q value) <0.05] in HLMVEC following heparinase III treatment relative to vehicle-treated HLMVEC.

Among the top 40 genes transcribed that were most significantly different between treatment groups (Figures 3A,B), we found expression of claudin-5 (CLDN5), thrombomodulin (THBD), and endothelin-1 (EDN1), more commonly expressed in ECs, was significantly impacted by heparinase III treatment. More specifically, we observed that CLDN5 and THBD were downregulated while EDN1 was upregulated following heparinase III treatment. Moreover, we observed downregulation in KLF2, KLF4, NOS3, and SLC9A3R2 (Figure 3C), flow-responsive genes that promote EC homeostasis (Ajami et al., 2017; Sangwung et al., 2017; Helle et al., 2020), thus confirming a disruption in homeostatic signaling by heparinase III. Consistent with our prior work (Richter et al., 2022a), we confirmed that heparinase III treatment promoted enrichment of ANGPT2 in addition to other markers of EC activation (ESM1, THBS1) (Figure 3D) (Helle et al., 2020). There were no changes in gene expression of canonical EC surface markers (KDR, CDH5, PECAM1, TEK) with heparinase III treatment (Supplementary Figure S5) (Helle et al., 2020).

Targeted analysis identified 33 GO: Biologic Process and 25 KEGG pathways enriched in heparinase III-treated HLMVEC and 50 GO: Biologic Process and 56 KEGG pathways enriched in vehicle-treated HLMVEC that met the pre-defined thresholds (Supplementary Table S2). Of these, 43 GO: Biologic Process and 38 KEGG pathway terms either unrelated to EC biology or represent pathway redundancy were removed from analysis (Supplementary Table S2). Remaining pathways were categorized based on their relevance to 1) cell maintenance and bioenergetics, 2) cell organization and adhesion, 3) intracellular signaling, 4) angiogenesis and wound healing, or 5) response to biophysical cues (Figures 4A,B).

Heparinase III treatment led to the enrichment in pathways regulating cell division and cellular response to DNA damage stimulus with downregulation of pathways suppressing cell proliferation and pathways regulating efferocytosis, apoptosis, and cell senescence (Figures 4A,B). Consistent with these changes, we observed reductions in the expression for the TNF and p53 signaling pathways following heparinase III treatment, pathways known to regulate cell proliferation and cell cycle arrest respectively (Frater-Schroder et al., 1987; Mijit et al., 2020) (Figure 4B), with a concomitant enrichment in the epidermal growth factor (ErbB) family of receptor tyrosine kinases signaling pathway that is known to support cell differentiation and proliferation (Wieduwilt and Moasser, 2008).

We observed downregulation in GO: Biological Process pathways that maintain cell-cell junctional integrity and enrichment for pathways that promote endocytosis following heparinase III application (Figures 4A,B). In line with these findings, we found that the KEGG pathway encoding Rap1 signaling, an established promoter of cell adhesion and cell-cell junctional stability (Pannekoek et al., 2014), was significantly less enriched in ECs treated with heparinase III (Figure 4B). Moreover, there appeared to be a reduction in the expression of pathways directing cell polarity and lamellipodium assembly that may regulate EC morphology under shear stress (Figure 4A) (Breslin et al., 2015).

Not surprisingly, we observed a reduction in cellular response to fibroblast growth factor (FGF) stimulus following heparinase III treatment given the role HS plays in mediating FGF-cognate receptor interactions at the EC surface (Figure 4A) (Wang et al., 2004; Ferreras et al., 2012). We also found that heparinase III treatment reduced the expression of genes regulating PI3K/Akt, sphingolipid, and apelin signaling while enriching for forkhead box O (FoxO), platelet-derived growth factor receptor (PDGFR)/Ras, and tissue growth factor (TGF)-β signaling pathways (Figures 4A,B). Genes in pathways regulating angiogenesis were found to be differentially upregulated in both treatment groups without a clear signature pointing toward a shift from one predominant signaling axis toward another (Figure 4A). Consistent with the reduced expression in flow-responsive genes discussed above, heparinase III treatment resulted in downregulation in the fluid shear stress response of ECs (Figure 4B). Together, these findings support that importance of HS expression for regulating mechanosensitive signaling pathways that control homeostatic EC functions.

Finally, we interrogated the impact of HS cleavage on the expression of genes related to HS proteoglycan and glycosaminoglycan synthesis in effort to gain insight into how ECs may act to regenerate the eGC following injury. In an unbiased approach, we assessed various isoforms of genes associated with the biosynthesis of the most abundantly expressed HS proteoglycans and glycosaminoglycans within the eGC, including syndecan 1-4 (SDC1-4), glypican 1 (GPC1), hyaluronan (HAS1-3), HS/chondroitin sulfate linkage region tetrasaccharide (XYLT1/2; β4GALT7; FAM20B; β3GALT6; β3GAT3), chondroitin sulfate disaccharide polymerization (CSGALNACT1-2; CHSY1-3; CHPF), and HS disaccharide polymerization (EXTL3; EXT1-2) (Merry et al., 2022). Additionally, we evaluated transcriptional levels of hyaluronidases (HYAL1-5), glucuronic acid C5 epimerase (GLCE), preferential chondroitin sulfotransferases (CHST3,7,11-13,15; UST), preferential HS sulfotransferases (NDST1-4; HS2ST1; HS3ST1-6; HS6ST1-3), sulfatases (SULF1-2), and heparanase (HPSE) that are regulators of glycosaminoglycan structural modifications and sulfation expressed within the eGC (Mizumoto and Yamada, 2021; Merry et al., 2022). No transcripts were detected for CHSY2, CHST13, NDST4, HS3ST4-6, HS6ST2, HYAL4, or HYAL5, indicating that these isoforms were not abundantly expressed in the ECs used in our studies. Thirteen genes associated with HS proteoglycan or glycosaminoglycan synthesis and sulfation in the eGC were significantly impacted by heparinase III treatment (Figure 5). The remaining 33 genes were not impacted by HS cleavage (data not shown). Notably, we observed downregulation of SDC3 and SDC4 following heparinase III treatment (Figure 5A), whereas genes involved in the synthesis of chondroitin sulfate (CHSY3 and CSGALNACT1) and hyaluronan (HAS2) were upregulated (Figures 5B,C). Moreover, we found that heparinase III treatment resulted in a downregulation in NDST1 and GLCE expression with upregulation in HS3ST1 and HS6ST3 expression (Figure 5D). Finally, we observed that heparinase III treatment resulted in a reduction in HYAL1, HYAL2, and HPSE expression (Figures 5B,D) without any change in SULF1 or SULF2 expression (data not shown). Overall, these data reflect changes in the eGC biosynthetic machinery in response to HS cleavage that may indicate compensatory eGC repair mechanisms following injury.