Integrated cross-linking by TG2 and FXIII generates hepatoprotective fibrin(ogen) deposits in injured liver Free

Activation of the blood coagulation cascade leads to the generation of the protease thrombin, which converts plasma fibrinogen to fibrin monomers that polymerize to form fibrin clots.¹ Thrombin also activates the transglutaminase factor XIII (FXIII), which circulates in plasma bound to ∼2% of fibrinogen.² Activated FXIII (FXIIIa) catalyzes intermolecular ε-N-(γ-glutamyl)-lysyl cross-links between residues in the γ and α chains of polymerized fibrin, ultimately forming γ dimers and cross-linked α polymers.^2,3 FXIII-dependent fibrin cross-linking plays a key role in determining fibrin structure, including resistance to lysis and clot function and inflammatory cell engagement.³ 

Tissue transglutaminase (TG2) also cross-links fibrin(ogen), driving both inter- and intramolecular cross-links in fibrinogen and fibrin, generating primarily Fibα-Fibγ cross-links without a requirement for thrombin-induced fibrin formation.^4,5 Studies examining the framework of intravascular fibrin cross-linking do not typically consider TG2, given that TG2 plasma concentration is several orders of magnitude lower than that of FXIII. However, TG2 protein levels are high in a wide range of tissues,^6,7 and for this reason, extravascular fibrinogen deposits within injured tissues may be uniquely subject to cross-linking reactions catalyzed by both FXIII and TG2. Surprisingly, few studies have defined the precise effect of TG2 on fibrin formed in vitro by traditional coagulation reactions (and FXIII). Moreover, the precise role of TG2 in fibrin(ogen) cross-linking in vivo is not well understood.

The liver expresses high levels of TG2⁶ and acute liver injury (ALI) is associated with robust extravascular fibrin(ogen) deposition.⁸ For example, fibrin(ogen) deposits are extensively cross-linked in the liver of mice challenged with an overdose of acetaminophen (APAP), a widely accepted experimental setting of ALI/acute liver failure (ALF).^9-12 Multiple studies suggest that fibrin(ogen) exerts hepatoprotective and prorepair effects in the APAP-injured liver,^10,12 highlighting a need to define the mechanisms driving fibrin(ogen) accumulation and cross-linking in the injured liver. Notably, neither hepatic fibrin(ogen) deposition nor FXIIIa-directed fibrin(ogen) cross-linking in the APAP-injured liver requires fibrin polymerization,¹⁰ a clear distinction from intravascular clot formation. We posited that high hepatic levels of TG2 may explain this uncharacteristic mechanism and pursued studies to define the precise impact of TG2 on fibrin(ogen) cross-linking in vitro and in the APAP-injured liver.

Wild-type C57Bl/6J mice (strain 000664) were purchased from the Jackson Laboratory (Bar Harbor, ME). Male mice expressing a flox-flanked TG2 allele (TG2^flox/flox) on a C57BL/6J background (Jackson Laboratory, strain 024694)¹³ were crossed with female mice on a C57Bl/6J background expressing a Cre recombinase driven by the cytomegalovirus (CMV) promoter (TG2^+/+/CMVCre^pos; strain 006054). These mice were intercrossed to recombine both alleles, CMVCre was removed by crossing to C57BL/6J mice, and heterozygous mice were crossed to generate wild-type and TG2^−/− mice, which were mated as homozygous pairs. Mice used for experiments were the first litters from these pairs. Mice lacking FXIII catalytic A subunit (FXIII-A^−/−)¹⁴ and wild-type mice (9-13 weeks old) on an identical C57Bl/6J background have been described previously. Mice were maintained under a 12-hour light/dark cycle, and standard rodent diet and purified drinking water were provided ad libitum. The Institutional Animal Care and Use Committee of Michigan State University approved all animal procedures.

Mice were fasted overnight (12-16 hours) before challenge with APAP (300 mg/kg, intraperitoneal; Sigma, A7085) at 30 μL/g body weight dissolved in sterile saline (10 mg/mL) or with sterile saline as vehicle. Approximately 24 hours after APAP challenge, liver and blood samples were collected under deep surgical isoflurane anesthesia. Blood samples were collected from the inferior vena cava into sodium citrate (final concentration ∼0.38%). Blood samples were centrifuged at 4000g for 10 minutes to obtain plasma and were stored at −80°C. The liver was rinsed in phosphate-buffered saline, snap frozen in liquid nitrogen, or fixed in 10% neutral-buffered formalin for 4 days before routine histopathological processing.

Paraffin-embedded liver sections were sectioned and stained for hematoxylin and eosin and for fibrin(ogen) by the Michigan State University Investigative Histopathology Laboratory, digitized using Virtual Slide System VS200 as described previously.¹³ The total area of positive fibrinogen staining was quantified using QuPath version 0.4.3.¹⁵ Detailed methods are available in the supplemental Methods (available on the Blood website).

Alanine aminotransferase activity in plasma was determined using commercial reagents (Thermo Fisher Scientific, Waltham, MA, or Pointe Scientific Inc, Canton, MI). Fibrinogen concentration was measured using an in-house enzyme-linked immunosorbent assay (ELISA) described previously.⁹ Pooled normal plasma (citrate, final 0.38%) from C57Bl/6J mice was used to generate a standard curve. Prothrombin concentration was measured using a commercial ELISA (IMSFIIKT, Innovative Research, Novi, MI). Plasmin-antiplasmin (PAP) complexes were measured using a commercial ELISA (MBS762337, MyBiosource, San Diego, CA). Fibrin degradation products (FDPs) were measured using a commercial ELISA (Asserachrom D-DI, Stago, Parsippany, NJ). Plasma FXIII-A levels were measured by capillary western blotting (12-230 kDa separation module, Wes instrument, ProteinSimple, San Jose, CA) in diluted (1:150) plasma using a sheep antihuman FXIII-A antibody (Enzyme Research Laboratories, South Bend, IN) shown previously to detect mouse FXIII-A.¹⁶ Peak area was analyzed using Compass Software for Simple Western (version 6.2.0).

Cross-linked fibrin(ogen) was measured in the insoluble protein fractions from snap-frozen mouse liver using fibrin(ogen) chain-selective antibodies¹⁷ and capillary western blotting, as described previously.¹¹ TG2 levels were measured in soluble protein samples using a sheep anti-TG2 antibody. Detailed methods are available in the supplement.

Notably, 50 to 100 mg snap-frozen naïve liver from C57Bl/6J mice was homogenized at 55 mg/mL in lysis buffer (15 mM n-octyl-β-d-glucopyranoside, 20 μg/mL bivalirudin, 15 μg/mL tranexamic acid in HEPES [N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid]-buffered saline). Each sample was then diluted in HEPES-buffered saline (+bivalirudin/tranexamic acid) to obtain a final concentration of 5 mM n-Octyl. Samples were then incubated with 10 μg/mL human fibrinogen (FIB1; Enzyme Research Laboratories) and CaCl₂ (5 mM) for 10 minutes. EDTA (final conc, 6 mM) was added, and then each reaction was reduced (95°C for 10 minutes) in lithium dodecyl sulfate buffer (Bio-Rad) and β-mercaptoethanol (1× and 2.5% final, respectively). Fibrin(ogen) chains were detected by capillary western blotting (see earlier section).

Human fibrinogen was reacted with TG2 as described previously⁴ with modification. Purified human fibrinogen (2.3 μM), either FIB1 or FXIII-free peak 1 fibrinogen (Enzyme Research Laboratories), was incubated for 10 or 30 minutes with human α thrombin (2 U/mL; Enzyme Research Laboratories) in the presence or absence of recombinant human TG2 (4376-TG-050, 335 nM, R&D Systems) in buffer containing 5 mM CaCl₂, 50 mM Tris-HCl, and 274 mM NaCl at 37°C. Cross-linking reactions were stopped by the addition of excess EDTA and then subjected to centrifugation for 4500g for 2 minutes. Clots were washed once with 1 mL of 500 mM NaCl + 0.1% SDS, vortexed, and then spun at 4500 × g for 2 minutes; 0.5 mL of solution containing 6 M guanidine-HCl and 100 mM ammonium bicarbonate (pH 9.0) was added to each pellet and vortexed at room temperature for 5 minutes and then stored on ice before proteomic analysis.

Cross-linked fibrin(ogen) was analyzed in 38 liver explant samples collected from patients with severe ALI or ALF who were recruited from the US Acute Liver Failure Study Group Registry between 2011 and 2018. Details of this cohort and analysis of hemostatic parameters have been described previously.^18,19 The etiology of the liver samples used was adjudicated as described previously.²⁰ Eleven samples were from patients with APAP-induced ALF along with 29 samples from patients with other etiologies (autoimmune, drug-induced liver injury, viral hepatitis, shock, Wilson disease, or indeterminate cause). All patients, or in the case of incapacity their consultee, gave informed consent for participation in this study. Ethical approval was obtained from the study-wide institutional review board of the University of Texas Southwestern Medical Center and by the institutional review board of each participating center. Approval for analysis of fibrin deposits in these samples was granted by the Ancillary Studies Committee of the Acute Liver Failure Study Group. Analyses of these samples and control livers were performed at University Medical Centre Groningen (UMCG). Control livers were small wedges of human liver (∼10 grams) collected from 5 human donors after partial resection and donor liver resizing or from livers unsuitable for transplantation at UMCG. Liver collection was approved by the Medical Ethical Committee of UMCG, according to Dutch legislation and the Code of Conduct for dealing responsibly with human tissue in the context of health research, refraining from the need of a written consent for “further use” of coded-anonymous human tissue. The procedures were performed in accordance with the experimental protocols approved by the Medical Ethical Committee of the UMCG. Control liver tissue was stored in University of Wisconsin preservation solution (= 4°C) before storage at −80°C. Frozen liver was sectioned at 5 μm and homogenized in 8 M urea containing 40 mM dithiothreitol and 12.5 mM EDTA and incubated overnight at 60°C, a digestion approach used for venous thrombi²¹ previously adapted for injured liver samples.¹² Hepatic levels of cross-linked fibrin(ogen) were measured by capillary western blotting (Wes Simple Western instrument) as described earlier.

Tissue samples were prepared for mass spectrometry (MS) analysis as previously described.²² Protein-protein cross-linking analyses in vitro were performed using methods as previously reported.²³ Detailed methods are available in the supplement.

Statistical analyses were performed using GraphPad Prism v.10.1.2 (GraphPad Software Inc, San Diego, CA). Results from individual mice are plotted and bars represent mean ± standard error of the mean. Comparison of 2 groups was performed using the Student t test. Correlations in proteomic data were evaluated using the Pearson correlation coefficient. Comparison of >2 groups was performed using 1- or 2-way analysis of variance, as appropriate, with the Tukey post hoc test. Levels of cross-linked fibrin(ogen) in livers from patients with ALF were analyzed using the Kruskal-Wallis test and the Dunn post hoc test. A P value <.05 was considered statistically significant.

Wild-type mice were challenged with a hepatotoxic dose of APAP and livers were collected 24 hours later. APAP challenge increased the hepatic levels of 257 proteins (P < .05; 1.5-fold; supplemental Tables 1 and 2). Analysis of these proteins for pathway enrichment (ShinyGO 0.80) identified coagulation and complement as the most significant pathway (supplemental Table 3). As anticipated, hepatic levels of all 3 fibrinogen chains increased significantly in livers of APAP-challenged mice (Figure 1A-C), as did hepatic levels of FXIII-A (Figure 1D). No clear association between Fibγ and FXIII-A (F13a1) was evident in vehicle-treated mice (r = 0.1354; P = .86), in which hepatic levels likely reflect cellular stores of FXIII-A (Figure 1E).⁹ In contrast, a strong association (r = 0.96; P = .008) between these proteins was observed in APAP-challenged mice (Figure 1E), likely reflecting the accumulation of fibrinogen-bound FXIII within the injured liver. Indeed, relative hepatic FXIII-A levels were ∼2% of fibrinogen (Figure 1G), a close approximation for the concentration of FXIII-A bound to fibrinogen in plasma. TG2 tended to increase in livers of APAP-challenged mice (Figure 1F). We also observed an apparent increase in plasma TG2 concentration in the plasma of APAP-challenged mice (supplemental Figure 1). Notably, hepatic TG2 levels were ∼20× and 7× higher than FXIII-A in livers of vehicle- and APAP-challenged mice, respectively (Figure 1D-E,G). The results suggest that the injured liver microenvironment includes fibrin(ogen) deposits and both FXIII-A and TG2.

Next, we crafted in vitro clots that reflected relative levels of TG2 and FXIII-A within the injured liver. The ratio of TG2:FXIII-A was ∼19:1 and 6:1 in livers from vehicle- and APAP-challenged mice. Reactions of fibrinogen with TG2 were performed as described previously,^4,24 with purified human fibrinogen (FIB1, 2.3 μM). FXIII circulates bound to ∼1% to 2% of circulating fibrinogen. Thus, we estimated the reaction to contain ∼46 nM FXIII-A. Recombinant human TG2 was added at 330 nM to produce a ratio of TG2:FXIII-A observed in the liver, and the reaction incubated in the presence of thrombin (2 U/mL) for 10 or 30 minutes. Similar experiments were performed using FXIII-free fibrinogen (peak 1 fibrinogen). For each reaction, we used MS-based cross-linked mapping to identify the amount of Fibγ-Fibγ, Fibα-Fibγ, and Fibα-Fibα cross-links in each reaction. Reactions containing only FXIIIa displayed a robust and sustained increase in Fibγ-Fibγ cross-links (Figure 2A). Substituting TG2 as the sole transglutaminase produced remarkably less Fibγ-Fibγ cross-linking (Figure 2A). Interestingly, the addition of TG2 reduced FXIIIa-directed Fibγ-Fibγ cross-linking (Figure 2A). Compared with FXIIIa, reactions with TG2 favored the formation of Fibα-Fibγ cross-links (Figure 2B). Interestingly, TG2-directed Fibα-Fibγ cross-linking was reduced in reactions that contained FXIII. Both FXIIIa and TG2 were separately able to catalyze Fibα-Fibα cross-links, particularly with time (Figure 2C). Interestingly, a synergistic increase in α-α cross-linking was evident after 10 minutes when both transglutaminases were present in the reaction (Figure 2C).

TG2 was detected by western blot in livers of wild-type mice but not TG2^−/− mice (see “Materials and methods”), validating the mice and polyclonal antibody (Figure 3A). Importantly, plasma levels of FXIII-A (Figure 3B) and fibrinogen (Figure 3C) were similar in naïve wild-type and TG2^−/− mice. As a proof of concept that liver-derived TG2 can cross-link fibrinogen, liver homogenates from wild-type or TG2^−/− mice were incubated with purified human fibrinogen in the presence of the thrombin inhibitor bivalirudin and tranexamic acid to limit plasmin-mediated lysis. Liver homogenate from wild-type mice, but not TG2^−/− mice, cross-linked fibrin(ogen) detected by both Fibα and Fibγ antibodies (Figure 3D), an observation repeatable across multiple experiments at a different dilution of liver homogenate (Figure 3E).

Hepatic fibrin(ogen) deposition in APAP-challenged wild-type mice resembled previous studies,¹¹ with fibrin(ogen) labeling aligned closely with areas of necrosis (Figure 4A). Frequent brighter labeling consistent with a punctate hepatocellular pattern was observed in livers of APAP-challenged wild-type mice (Figure 4A; arrows). The pattern of fibrinogen labeling was more varied in APAP-challenged TG2^−/− mice, an observation made by individuals not masked to genotype. Observations perceived as consistent included a reduced labeling intensity within areas of necrosis along with the emergence of stringy and fibril-like staining both within areas of necrosis and in the sinusoids of neighboring healthy parenchyma (Figure 4A; arrowheads). Insoluble fibrinogen β chain levels, reflecting total hepatic insoluble fibrin(ogen), were reduced in livers of APAP-challenged TG2^−/− mice (Figure 4B). The results suggest that TG2 deficiency has quantitative and qualitative impacts on fibrin(ogen) deposition in the APAP-injured liver, highlighted by a distinct shift in labeling pattern within necrotic areas. These changes were not explained by changes in biomarkers of coagulation (ie, reductions in plasma prothrombin and fibrinogen) or in biomarkers of fibrinolysis (ie, increases in plasma PAP complexes or FDPs; Figure 4C-F).

Strong evidence indicates that cross-linked fibrin(ogen) accumulates in the liver of APAP-challenged mice.^11,12 Notably, we also found that hepatic levels of cross-linked fibrin(ogen) are also increased in liver explants from patients with APAP-induced ALF and patients with non-APAP-induced ALF compared with control livers (supplemental Figure 2A-B). In agreement with previous studies,^10,11 APAP challenge increased hepatic levels of cross-linked fibrinogen, including Fibγ-Fibγ dimer (Figure 5A). Notably, Fibγ-Fibγ dimer was absent in livers of APAP-challenged FXIII^−/− mice (Figure 5C), validating previous studies.¹¹ In contrast, robust Fibγ-Fibγ dimer was apparent in livers of APAP-challenged TG2^−/− mice (Figure 5D). Accounting for lower overall fibrin(ogen) deposition in livers of APAP-challenged TG2^−/− mice (ie, indicated by less insoluble Fibβ; Figure 4A-B), TG2 deficiency was associated with a relative increase in hepatic Fibγ-Fibγ (Figure 5E). The results suggest that FXIII directs Fibγ-Fibγ dimer formation in the APAP-injured liver and that this cross-link is suppressed by TG2. Validating previous studies, APAP challenge increased hepatic levels of high-molecular-weight (HMW) cross-linked Fibα species (Figure 6A). Hepatic levels of cross-linked Fibα were dramatically reduced in APAP-challenged FXIII^−/− mice (Figure 6C-D). Interestingly, hepatic levels of HMW cross-linked Fibα were significantly reduced in livers of APAP-challenged TG2^−/− mice (Figure 6B,E), and this difference remained significant even when adjusting for the overall decrease in hepatic insoluble fibrin(ogen) (Figures 6F and 4B). Importantly, this reduction was not a simple consequence of altered plasma FXIII-A levels (Figure 3B). The results indicate that deficiency in either FXIII or TG2 significantly reduces HMW Fibα cross-linking in the APAP-injured liver.

Strong evidence suggests that fibrinogen deficiency or dysfunction leads to exaggerated hepatic injury and delayed liver repair after APAP challenge.^10,12 These changes are typified by vascular damage and congestion. As anticipated, plasma alanine aminotransferase activity was substantially increased (normal ∼20-80 U/L) and centrilobular hepatic necrosis was evident in wild-type mice challenged with APAP (Figure 7A-B,D). TG2 deficiency tended to increase hepatic necrosis (Figure 7B) and hepatic congestion/hemorrhage was significantly increased in livers of APAP-challenged TG2^−/− mice (Figure 7C-D).

As the predominant plasma transglutaminase that cross-links fibrin, FXIII circulates in plasma in complex with fibrinogen, whereupon coagulation activation drives cross-linking of both the α and γ chains of fibrin polymer.^2,3 This confers critical structural stability and effector functions to intravascular fibrin-containing thrombi. These reactions likely occur alongside coagulation-independent modification of fibrin(ogen) in the context of complex microenvironments such as tissue necrosis. How these mechanisms interact to affect extravascular fibrin(ogen) formation is not known. Using a robust experimental setting of liver necrosis, our results uncover marked effects of TG2 deficiency on fibrin(ogen) deposition and cross-linking. We document relative levels of TG2 and FXIII in the injured liver, that liver-derived TG2 is sufficient to cross-link fibrinogen, and that TG2 shifts fibrin cross-linking in vitro in a manner reflective of changes observed within the injured liver. Overall, these proof-of-concept studies suggest that TG2 may have fundamental impacts on fibrin(ogen) cross-linking within injured tissues.

Estimated TG2 concentration in human plasma (Peptide Atlas, https://www.proteinatlas.org) is at least 3 orders of magnitude lower than that of FXIII,²⁵ making plasma TG2 an unlikely contributor to intravascular clot formation in health. Notably, we did identify an obvious increase in plasma TG2 concentration in APAP-challenged mice (supplemental Figure 1), suggesting that, in conditions of acute tissue damage, TG2 may emerge as a relevant plasma transglutaminase. Notably, hepatic TG2 can effectively cross-link exogenous fibrinogen, even in the absence of thrombin activity, as documented by our studies. This is consistent with previous experiments showing that TG2 cross-links fibrinogen to the surface of hepatocytes.^26,27 Indeed, plasma fibrinogen and FXIII seem to accumulate in a tissue containing high levels of TG2 deployed to remodel matrix proteins after injury.^28,29 It seems most plausible for fibrinogen (and FXIII) to encounter TG2 within the injured liver and at levels where TG2 exceeds that of FXIII. The cellular sources of TG2 contributing to fibrin(ogen) cross-linking in the APAP-injured liver are not known. Notably, such knowledge would be required for next-level experiments seeking to restore TG2 expression in specific cell types before APAP challenge. Previous studies have shown that isolated hepatocytes can cross-link fibrinogen^26,27 and hepatocytes are the primary cellular target of APAP toxicity.³⁰ Indeed, this could explain the unique pattern of fibrin(ogen) labeling in the APAP-injured liver. However, multiple cell types in the liver express TG2, and additional studies would be required to define the precise role of hepatocyte-derived TG2.

APAP-induced liver damage is associated with robust coagulation activation.^9,31 However, deposition of fibrin(ogen) within the injured liver occurs by atypical mechanisms. For example, hepatic fibrin(ogen) deposition was not significantly reduced in low tissue factor mice or in mice expressing a mutant fibrinogen for which fibrinopeptide A cannot be removed by thrombin.^11,31 Notably, TG2 deficiency moderately reduced hepatic fibrin(ogen) accumulation, one of few examples of an intervention that reduces hepatic fibrin(ogen) deposition in this context. TG2 deficiency also altered the apparent pattern of fibrin(ogen) deposition, with APAP-challenged TG2^−/− mice displaying a more discontinuous and even fiber-like labeling within areas of necrosis. The immunolabeling used is knockout validated,^32,33 and this change in pattern may reflect an altered fibrin(ogen) structure driven by changes in fibrin(ogen) cross-linking or altered degradation of fibrin(ogen) deposits. For example, previous studies have shown that TG2 addition to established fibrin clots delays plasmin-mediated lysis.³⁴ Although increased fibrin(ogen) degradation cannot be excluded, we did not detect an impact of TG2 deficiency on plasma PAP complexes or FDPs.

FXIII generates Fibγ-Fibγ cross-links, among others, in clots formed by thrombin in vitro. TG2 forms Fibα-Fibγ hybrid cross-links of increasing size even without thrombin addition.^4,5 Using both sodium dodecyl sulfate–polyacrylamide gel electrophoresis and MS-based proteomics,²³ our studies corroborate the unique cross-links imposed by each transglutaminase. At levels of FXIII and TG2 approximating those in the injured liver, we observed an apparent reciprocal inhibition wherein the presence of TG2 inhibits FXIII-directed Fibγ-Fibγ cross-linking. This aligns with the relative increase in Fibγ-Fibγ dimer in livers of APAP-challenged TG2^−/− mice and may be driven by hepatic TG2 levels outcompeting FXIII for gamma chain cross-links. Interestingly, the presence of FXIII also diminished TG2-dependent Fibα-Fibγ cross-linking, implying cross-competition between the 2 transglutaminases. Concentrations of each transglutaminase and cross-linking sequence likely define the penultimate pattern of cross-links imposed, and additional experiments seeking to explore these concentration-dependent effects should be performed in the future. Indeed, the relative concentration of TG2 may explain reductions in Fibγ-Fibγ cross-linking. Our proteomic studies also explain a mysterious aspect of our studies in the APAP-injured liver. In particular, how are hepatic levels of HMW cross-linked Fibα reduced in both FXIII^−/− mice and TG2^−/− mice? This was not a consequence of reciprocal dysregulation of expression, given that plasma FXIII-A levels were unaffected by TG2 deficiency, and hepatic TG2 expression was similar in FXIII^−/− mice (not shown). Rather, we report, to our knowledge, the first evidence of synergistic Fibα-Fibα fibrin(ogen) cross-linking imposed by the combination of both TG2 and FXIII, documenting a mechanistic explanation for our observations in the APAP-injured liver in a reductionist system. In these conditions, TG2 may favor α-α cross-linking because it is not loaded on the Fibγ chain or its preference for turns between secondary structural elements (alpha helixes and beta sheets) and intrinsically disordered regions of proteins.³⁵ The present studies establish a framework and method in which these detailed questions can be addressed.

Although fibrinogen is occasionally assumed to drive liver pathology, multiple studies indicate that fibrinogen deficiency or dysfunction increases hepatic injury or inhibits liver repair. It is possible that TG2 plays a key role in cross-linking fibrin(ogen) in necrotic liver to support a unique role of this hemostatic protein in tissue repair. Consistent with this, TG2 deficiency increased liver damage in another model of chemical-induced liver damage.³⁶ Previous studies have suggested that TG2-cross-linked fibrinogen supports leukocyte activation,²⁴ a mechanism identified as critical in APAP-challenged mice.¹² Indeed, TG2^−/− mice challenged with APAP displayed histological features reported in APAP-challenged Fibγ^390-396A mice, which express fibrin(ogen) that does not engage β2 integrins.¹² However, it is worth noting that TG2 is a multifunctional molecule and alternative transcriptional roles of TG2 in hepatocytes are but one additional mechanism or function whereby TG2 deficiency could affect liver damage in this and other contexts.^7,37,38 With this limitation noted, the precise role of TG2-mediated fibrinogen cross-linking in ALI could potentially be determined using mice expressing mutant fibrinogen resistant to TG2 cross-linking, as have been generated for FXIII-specific cross-links.³⁹ If a complete map of TG2-cross-linked fibrin(ogen) emerges, these studies are plausible and could be quite exciting.

In summary, we uncovered a novel mechanism whereby TG2 tailors the pattern of FXIII-directed fibrin(ogen) cross-linking in experimental liver injury. Observed changes in fibrin(ogen) cross-linking within the injured liver were largely mirrored by cross-links imposed in vitro by pathologically relevant ratios of TG2 and FXIII. The results have substantial importance because they document that TG2, traditionally viewed as a “nonhemostatic” transglutaminase, contributes significantly to fibrin(ogen) clot cross-linking. Implications of these results extend beyond the toxicological roots of APAP overdose. Synergistic cross-linking imposed by TG2 and FXIII may have valuable repurposing in the treatment of FXIII deficiency. Likewise, given the ubiquitous cross-tissue expression of TG2, a role for this mechanism in defining outcomes in traumatic or chemical injury to multiple tissues seems likely.

The authors thank Ryan Lewandowski and the digital slide scanning core for their dedicated assistance with digital histopathology, and Anna Kopec for inspiration and for contributions to initial experiments. The Acute Liver Failure Study Group was central to analysis of fibrin(ogen) in human liver samples and the authors acknowledge the contributions of Jody Rule and Nahid Attar to this effort.

This research was supported by grants from the National Institutes of Health (NIH), National Institute of Diabetes and Digestive and Kidney Diseases (R01 DK120289 and R01 DK136733 [J.P.L.]), and research support from the US Department of Agriculture (USDA) National Institute of Food and Agriculture and the Albert C. and Lois E. Dehn Endowment to Michigan State University for Veterinary Medicine (Pathobiology and Diagnostic Investigation) to J.P.L. Z.W. and J.P.L. thank the Michigan State University College of Natural Sciences and the Family of John A. Penner for support from the John A. Penner Endowed Research Fellowship.

The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH or the USDA.

Contribution: Z.W., N.K.K.B., L.R.S., M.T.F.F., A.N., H.C., A.T., L.G.P., K.C.H., J.A., T.L., and J.P.L. contributed to the design of the studies; K.C.H. and L.R.S. performed proteomic analyses; J.A. and T.L. analyzed fibrin levels in human ALF samples provided by R.T.S. and W.M.L.; Z.W. and J.P.L. drafted the manuscript; and all authors performed the experiments, collected and interpreted the data, and reviewed and approved the submitted version of the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests (beyond research support from the National Institutes of Health).

Correspondence: James P. Luyendyk, Department of Pathobiology and Diagnostic Investigation, Michigan State University, 1129 Farm Lane, 253 FST, East Lansing, MI 48824; email: luyendyk@msu.edu.