caADAMTS13 reduces neutrophil-mediated vascular inflammation without exacerbating bleeding in experimental intracerebral haemorrhage

Abstract

Targeting thromboinflammation through the von Willebrand factor (VWF)/ADAMTS13 axis is a promising therapeutic target for acute ischaemic stroke (AIS). Constitutively active ADAMTS13 (caADAMTS13) is a novel thrombolytic that improves outcomes in experimental AIS, with enhanced thrombolytic efficacy compared to current licensed AIS treatments. Current thrombolytics are limited by a narrow therapeutic window and the requirement for definitive imaging to exclude intracerebral haemorrhage (ICH). This study aimed to assess whether caADAMTS13 is safe in the context of ICH, evaluating its potential use when the stroke subtype is unknown, including pre-hospital settings. In zebrafish larval and murine ICH models, both hyperacute (1 h) and delayed (12 or 24 h) caADAMTS13 administration did not exacerbate haemorrhage. Early cerebrovascular VWF accumulation (4 h) and acute neutrophil recruitment (24 h) were unchanged, while vascular neutrophil deposition was reduced by 80% in mice 7 days post-ICH. Delayed caADAMTS13 treatment reduces brain neutrophil recruitment by 37% and injury by 32%, improving ICH outcomes in zebrafish larvae. Together, these findings support the development of caADAMTS13 as a thrombolytic suitable for use in suspected stroke prior to brain imaging to maximise benefit from ultra-early treatment or in regions where rapid access to brain imaging is not possible.

Introduction

Stroke is a major global health burden with approximately 12.2 million people affected each year and accounting for substantial mortality and morbidity worldwide. ¹ Reperfusion-based interventions, namely recombinant tissue-type plasminogen activator; alteplase (rt-PA) and Tenecteplase (TNK) are used for ischaemic stroke (AIS) treatment.^2,3 Despite their effectiveness, these thrombolytic treatments are limited by a narrow therapeutic window, partly due to increased risk of intracerebral haemorrhage (ICH).^2,4 Currently, administration of AIS therapies requires brain imaging to exclude ICH, resulting in delays to treatment. Therefore, the development of therapies that could be safely administered in the pre-hospital setting, prior to definitive brain imaging, has the potential to improve clinical outcomes following stroke.

Thromboinflammation is a central driver of AIS and vascular pathology. von Willebrand factor (VWF) orchestrates the interaction between thrombosis and inflammation ⁵ and in response to vascular injury VWF drives platelet adhesion and thrombus formation.^6–9 Beyond its haemostatic role, VWF is a central mediator of vascular inflammation by promoting neutrophil recruitment, endothelial activation and the amplification of inflammatory pathways.^10–14 Since the size of VWF multimers is strongly correlated with its thromboinflammatory potential, VWF cleavage by the metalloproteinase ADAMTS13 is a central regulatory mechanism.^5,15,16 With thrombosis and inflammation being key drivers of AIS pathology, targeting the VWF/ADAMTS13 axis is considered a promising therapeutic approach for stroke. ¹⁷ However, it is unclear whether targeting this pathway could disrupt haemostasis in active intracerebral bleeding as occurs in ICH.

Preclinical studies indicate that wild-type (wt)ADAMTS13 is protective in models of AIS, however, its efficacy is limited by its conformational quiescence, which restricts substrate accessibility. ¹⁸ A novel constitutively active variant of ADAMTS13 (caADAMTS13) was generated to maintain the enzyme in an open, active conformation. ¹⁹ In preclinical models of AIS, caADAMTS13 was found to be protective by dissolving microvascular platelet aggregates and reducing neutrophil infiltration into the ischaemic tissue with greater efficacy compared to wtADAMTS13.^19,20 Additionally, a recent direct comparison against r-tPA and TNK, showed that caADAMTS13 exhibits greater thrombolytic efficacy against platelet-rich thrombi while maintaining a lower incidence of haemorrhagic transformation in experimental stroke. ²¹ With caADAMTS13 exhibiting enhanced thrombolytic activity, it is important to determine whether it may exacerbate bleeding if administered during active intracerebral bleeding.

While ADAMTS13 variants have been primarily developed for AIS to date, demonstrating the haemostatic safety of caADAMTS13, would support its use prior to brain imaging, thereby maximising the benefits of hyperacute AIS treatment. In addition, preclinical studies suggest that wtADAMTS13 may even mitigate pathological neutrophil-driven inflammation in both ICH and subarachnoid haemorrhage,^22,23 therefore it is possible caADAMTS13 may improve inflammatory outcomes across all stroke subtypes. caADAMTS13 has never been assessed in the context of active ICH. In this study, we evaluated the safety of both hyperacute and delayed administration of caADAMTS13 in two complementary preclinical models of ICH, spontaneous ICH in zebrafish larvae and collagenase induced ICH in mice, with the primary objective of determining safety and the secondary objective of examining the potential for caADAMTS13 to mitigate inflammation-driven vascular injury.

Materials and methods

Animals

Animal studies were conducted in accordance with the Animal Scientific Procedures Act (1986) and were approved by the Animal Welfare and Ethical Review Body, University of Manchester, UK.

C57BL/6 male (11–16 weeks old, 25–31 g) and female (11–16 weeks old, 17–24 g) mice were obtained from Charles River Laboratories. Mice had free access to food and water and were housed under standard conditions. Across experiments, a total of eight animals were excluded due to declining post-ICH health issues.

Zebrafish (Danio rerio) were raised and maintained at The University of Manchester Biological Services Unit under standard conditions. ²⁴ The bubblehead^m292 (bbh) model of spontaneous ICH was utilised for this study, crossed onto ubiq:secAnnexin V-mVenus and mpo:GFP transgenic reporter backgrounds to visualise cell death and neutrophils, respectively. ²⁵ Fertilised eggs were collected following natural breeding and incubated at 28.5 °C in standard embryo E3 water following standard guidelines. ²⁶ Following each experiment, zebrafish larvae were terminated using a lethal dose of 4% MS222 prior to reaching protected status at 5 days post-fertilisation (dpf).

Intracerebral haemorrhage models

Mice were anaesthetised using 4% isoflurane in a gas mixture of 70% N₂O and 30% O₂ and were immobilised in a stereotactic frame. Topical analgesia and iodine were applied at the incision site. A burr-hole was drilled into the skull and 0.5 μL of type VII-S collagenase (0.09 units/µL) was infused into the right striatum at a constant rate of 1 µL/min. Core body temperature was maintained at 37 °C. caADAMTS13 (5.5 mg/kg) or vehicle (150 mM NaCl, 20 mM histidine, 2% sucrose, 0.05% Tween 80, pH 7.4) were administered intravenously (i.v.) following induction of ICH. For the MRI study, treatment was only administered 1 h post-ICH. Grip strength tests were performed on an independent cohort of mice treated at either 1 or 24 h post-ICH. Blinding and randomisation were performed prior to starting the experiment and were maintained until data analysis and histology was concluded.

Experimental zebrafish larvae were generated from in-crosses of adult heterozygous bbh mutant breeder fish. In this model, ICH occurs spontaneously from ~36 h post-fertilisation (hpf). ²⁵ Larvae were monitored closely for ICH onset and embryos with spontaneous brain haemorrhages (ICH+) were separated from non-haemorrhage (ICH−) siblings which served as sibling controls. caADAMTS13 or vehicle were randomly administered through i.v. microinjections into the Duct of Cuvier. bbh; ubiq:secAnnexin V-mVenus larvae were treated at either 1 or 12 h post-ICH whereas bbh; mpo:GFP larvae were only treated 12 h post-ICH. Delayed treatment in both zebrafish larval and murine model was chosen to investigate the potential of caADAMTS13 to mitigate vascular inflammation once haemostasis is achieved following ICH.

Antibody conjugation to microparticles of iron oxide (MPIO)

Dynabeads MyOne were washed in coating buffer (0.1 M sodium borate buffer, pH 9.5) to remove the storage buffer. Anti-VWF IgG (rabbit polyclonal; Dako, A0082), anti-Ly6G IgG (rat monoclonal; BioLegend, 127602), rabbit isotype control IgG (rabbit polyclonal; BioLegend, 910901), rat isotype control IgG (rat monoclonal; BioLegend, 402301) were diluted in coating buffer and 3 M ammonium sulphate at optimised concentration for coupling. Washed beads were resuspended in diluted antibodies and left rotating at 37 °C for 24 h.

MRI

MRI data (T2-weighted, T2*-weighted and MPIO-based molecular MRI) were acquired at 4, 24 h and 7 days post-ICH using an Agilent 7-Tesla, 16 cm horizontal-bore magnet (Agilent Technologies, Oxford, UK) interfaced with a Bruker Biospec Avance III console (Bruker Biospin, Coventry, UK). A 16 cm volume coil was used for transmission, and a mouse brain surface coil (Rapid Biomedical, Rimpar, Germany) was used for reception.

Mice were anaesthetised with 4% isoflurane in a 70% N₂O/30% O₂ mixture and maintained at 1.5%–2% isoflurane during scanning. A 27 G tail vein catheter was placed for administration of microparticles of iron oxide (MPIO). A three-plane localiser scan was acquired to verify head positioning. T2-weighted RARE scans were acquired to assess the total lesion volume. This was later split into hyperintense (oedema) and hypointense (non-oedematous) tissue. The scanning parameters were: repetition time/echo time (TR/TE) = 3600/35 ms; NEX = 6; FOV = 16 × 16 mm²; matrix size = 256 × 256 mm; slice thickness = 0.5 mm; number of slices = 32, rare factor = 8. 2D T2*-weighted FLASH scans were acquired to measure iron load present within the lesion (either stored within red blood cells or free iron) due to bleeding. ²⁷ Scan parameters were: TR/TE = 1500/10 ms, flip angle = 60°, NEX = 2, FOV = 16 × 16 mm², matrix size 256 × 256 and slice thickness = 0.375 mm.

In vivo MPIO-based molecular MRI was used to determine the vascular distribution of VWF at 4 h post-ICH and Ly6G at 24 h and 7 days post-ICH. Molecular MRI scans consisted of a multi-gradient echo 3D FLASH with scan parameters: TR/TE = 200/4.8 ms, ΔTE = 7.1 ms, number of echoes = 4, flip angle = 30°, number of repetitions = 15, zero filling factor = 2 along phase and slice direction, FOV = 16 × 16 × 2 mm³, matrix size = 256 × 256 × 10 and total scan duration = 38 min. Prior to injection, antibody-conjugated MPIOs were washed twice in PBS and resuspended in sterile saline. The multi-echo scan was initiated and 100 µL of MPIO solution injected via the tail vein catheter on the sixth image volume. Imaging then continued for another 10 image volumes. The effect of MPIO on multi-echo signals was quantified by calculating the difference in R₂* between baseline and post-injection volumes (more detail provided in MRI analysis section below).

Light sheet microscopy

Light sheet microscopy was used to quantify cell death (ubiq:secAnnexin V-mVenus) and neutrophils (mpo:GFP) in zebrafish larval brains, as previously described. ²⁸ Imaging was performed at 36 h post-ICH for bbh; ubiq:secAnnexin V-mVenus larvae as we have previously observed this time point aligns to maximal brain cell death ²⁹ bbh; mpo:GFP larvae were imaged at 18 and 24 h post-ICH as we have previously shown neutrophil recruitment occurs between these time points. ²⁵ Larvae were anaesthetised using 0.02% MS222 and mounted in 1.5% low-melt agarose (Promega) in E3 embryo water without methylene blue. A W Plan-Apochromat 20× magnification/1.0 UV-VIS objective was used to acquire images. Images were processed using ZEN software. Maximum intensity projection (MIP) composites were made from z-stack images. ImageJ software was then used to select regions of interest around the brain (excluding eyes) and for quantification of the fluorescent signal.

For brain cell death, the average intensity fluorescence of mVenus signal within a brain region of interest was calculated using a previously described macro. ²⁵ For neutrophils, the number of GFP-positive cells with the brain region was analysed using ImageJ, and this was verified by blinded manual count.

Haemoglobin staining

O-dianisidine was used to stain haemoglobin and subsequently as a readout of brain haematoma size in the zebrafish ICH model. At 3 dpf larvae, were incubated in the dark for 30 min at room temperature within a solution containing 0.6 mg/mL o-dianisidine (Sigma, D9143-5g) as previously described. ³⁰ Once staining was completed, samples were washed with deionised H₂O and fixed in 4% PFA for 1 h at room temperature. Melanocytes were bleached using 10% hydrogen peroxide (5% formamide, 0.5 × SSC) to aid visualisation. Blind imaging was then performed within 2 days of staining using a brightfield stereomicroscope (Leica S9e). O-dianisidine staining was quantified as percentage of total brain area using ImageJ.

Grip strength test

A grip strength metre was used to measure forelimb grip strength. Mice were briefly suspended by the tail and positioned to grasp a horizontally oriented metal bar with their forepaws. The tail was pulled back in a horizontal plane until the mice released their forepaws. Grip strength was recorded both bilaterally and unilaterally with three trials performed for each. The order in which the right and left forepaws were tested was randomised to avoid bias. Measurements were excluded if mice used their hindlimbs to pull on the bar rather than relying on grip strength or if the mouse refused to engage with the task. A total of four grip strength measurements were excluded.

MRI analysis

T2-weighted and T2*-weighted MRI scans were analysed using MRIcron. For each animal, masks were manually drawn to delineate the visible lesion. For T2 scans at 24 h, individual masks for hyper- and hypo-intensities were also generated to capture volumes associated with oedema and non-oedematous lesioned tissue, respectively. Masks were drawn on consecutive slices and while blinded to the experimental groups. For T2* scans, masks were drawn on visible areas of hypointensity, providing an estimate of iron distribution due to the initial bleed. Volumes were calculated using MRIcron based on the total number of voxels within each mask and voxel dimensions.

Molecular MRI scans were analysed by fitting a monoexponential model to multi-echo signals to estimate the R₂* relaxation rate. Lesion, pseudo lesion, ipsilateral, contralateral and whole-brain masks were drawn and applied to the R₂* maps, with morphological erosion (two-pixel cube) to reduce partial volume effects. An R₂* threshold (R₂* > 0.02/s) was applied to identify MPIO-associated signal. For each region, the volume of MPIO-positive voxels were quantified. Animals were excluded from MPIO analysis if MPIO administration during MRI was unsuccessful. Injections were unsuccessful in two animals in the α-VWF-MPIO group and two animals in the α-Ly6G-MPIO at 24 h post-ICH; these animals were therefore excluded. Quantification of IgG isotype control is available in Supplemental Table 1.

Tissue processing

Mice were transcardially perfused with saline followed by 4% paraformaldehyde (PFA). Brains were removed, fixed in 4% PFA overnight at 4 °C and transferred to a 30% sucrose solution at 4 °C for up to 3 days. Brains were snap-frozen in isopentane and stored at −70 °C. Thirty micrometre coronal sections were cut using a freezing sledge microtome (Bright Instruments, UK). Cut sections were stored in a cryoprotectant (0.05 M Na₂HPO₄ × 2H₂O, 5 mM NaH₂PO₄, 30% anhydrous ethylene glycol, 20% glycerol) at −20 °C until mounted on slides for histology.

Histology

Haematoxylin and eosin (H&E) staining was performed to assess haemorrhages in mouse brain tissue. Slides were rehydrated through decreasing ethanol concentrations, stained with haematoxylin (4 min), differentiated in acid alcohol (10 s), stained with eosin (30 s), dehydrated through ascending ethanol and cleared in xylene. Images were collected using a 3D Histech Panoramic P250 microscope slide-scanner using a ×20/0.30 Plan Achromat objective (Zeiss, Germany). Image analysis was performed using QuPath 0.6 software. Haematoma lesion volume was calculated by manually delineating the lesion on H&E-stained brains. Total haematoma volume was estimated using the trapezoidal method to calculate the area under the curve with x-axis being the inter-slice distance and y-axis the haematoma area per slice. Measurements were performed while blinded to experimental groups.

Statistical analysis

Data are presented as mean ± SD. p Values, individual data points and group sizes are shown in all figures. Shapiro–Wilk test was used to determine normality of all data in each experiment. Statistical comparison between two groups (sham vs ICH or vehicle vs caADAMTS13) was performed using Welch’s t-test if data were normally distributed. Mann–Whitney U test was used when data were shown to be non-parametric. Two-way ANOVA was used to compare ICH+ and ICH− larvae across vehicle and caADAMTS13 groups. Tukey’s multiple comparisons test was used to assess pairwise comparisons. Individual statistical tests used for each comparison are available in the figure legend and in Supplemental Table 2. Power calculations were performed prior to the study to confirm group sizes based on pilot data, assuming a two-sided α of 0.05 and 80% power. Significance was taken at the 5% level.

Ethics approval for the use of animals

All animal procedures were carried out in accordance with the Animals (Scientific Procedures) Act (1986), under Home Office UK project licenses, approved by the local Animal Welfare Ethical Review Board, and experiments were performed in accordance with Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines, ³¹ with researchers blinded to experimental groups.

Results

Hyperacute administration of caADAMTS13 does not alter haemorrhage size or brain cell death in zebrafish larvae

To investigate the effects of caADAMTS13 in the context of active brain bleeding, we first used a zebrafish larval model of spontaneous ICH. bbh; ubiq:secAnnexin V-mVenus zebrafish larvae were injected with vehicle or caADAMTS13 1 h following ICH (Figure 1(a)). With the haematoma beginning to resolve at 72 hpf in this model, we decided to examine the effects of caADAMTS13 at 36 hpf post-treatment to capture any potential early effects before the haematoma resolves. ²⁹ Bright-field microscopy of haemoglobin-stained larvae revealed comparable haemorrhage lesion sizes between caADAMTS13, and vehicle treated groups (p = 0.1279; Figure 1(b) and (c)). As previously reported, ICH induced an increase in brain Annexin V fluorescent intensity, used as a readout of brain injury in both vehicle (p < 0.001) and caADAMTS13 treated larvae (p = 0.0028). ²⁵ caADAMTS13 did not alter Annexin V signal intensity compared to vehicle treated larvae (p = 0.7862; Figure 1(d) and (e)).

Figure 1.

Hyperacute caADAMTS13 administration does not alter haemorrhage size or brain cell death in zebrafish larvae: (a) schematic of the experimental design. Zebrafish larvae were treated with vehicle or caADAMTS13 (caAD13) 1 h after ICH, (b) representative images of ICH-positive bubblehead larvae stained with o-dianisidine 24 h after caAD13 or vehicle treatment to visualise haemoglobin, (c) quantification of haematoma area (µm²) at 3 dpf (n = 17 for vehicle and n = 20 for caADAMTS13). Normality is assessed using the Shapiro–Wilk test, and statistical analysis was performed using the Mann–Whitney U test, (d) representative maximum-intensity projections of brain cell death in bubblehead zebrafish larvae expressing ubiq:secAnnexin V-mVenus at 3 dpf. Larvae that did not develop a brain haemorrhage are used as controls (ICH−), while ICH+ larvae developed a brain haemorrhage. Eye lenses are indicated by green circles, and hatched lines outline the brain region used for Annexin V quantification. Images are acquired at 20× magnification and (e) quantification of mean Annexin V fluorescence intensity from three independent experiments (n = 35 per group). Statistical analysis was performed using two-way ANOVA with Tukey’s post-hoc test. Data are presented as mean ± SD.

dpf: days post-fertilisation; ICH: intracerebral haemorrhage; ns: non-significant comparisons.

Statistical significance is indicated as **p < 0.01. ****p < 0.0001.

Early administration of caADAMTS13 does not affect the total lesion and iron-associated volume in mice

Building on the zebrafish findings, we next investigated the effects of caADAMTS13 on the total lesion and the iron-associated volume in a collagenase-induced mouse model of ICH (Figure 2(a)). Clinically, the risk of rebleeding is highest within the first 24 h after ICH onset, suggesting that 4 and 24 h post-ICH may coincide with the period of greatest haemostatic vulnerability. ³² We also assessed whether caADAMTS13 might increase the risk of rebleeding at 7 days post-ICH. The total lesion volume was assessed by T2-weighted and T2*-weighted MRI. T2-weighted MRI showed no significant differences in the total lesion volume between vehicle and caADAMTS13 treated animals at 4 h (p = 0.5116; Figure 2(d)), 24 h (p = 0.2833; Figure 2(e)) and 7 days post-ICH (p = 0.9278; Figure 2(h)). At 24 h post-ICH, the areas of hyperintense (oedema, p = 0.3189; Figure 2(f)) and hypointense (non-oedematous, p = 0.5928; Figure 2(g)) tissues were unchanged between caADAMTS13 and vehicle groups. No significant differences in iron-associated volume were observed between vehicle and caADAMTS13 treated animals on T2* weighted MRI at 4 h (p = 0.5405; Figure 2(j)), 24 h (p = 0.2681; Figure 2(k)) and 7 days post-ICH (p = 0.8638; Figure 2(l)). These results suggest that acute caADAMTS13 treatment does not alter the total lesion and iron-associated volume and that, at 24 h post-ICH, the extent of oedema and non-oedematous tissue remains unchanged.

Figure 2.

Early caADAMTS13 administration does not alter haemorrhagic lesion volume or composition in a collagenase mouse model of ICH: (a) schematic of the experimental design. Mice were treated with vehicle or caADAMTS13 (caAD13) 1 h after ICH induction, (b) representative T2-weighted MRI images at 4, 24 h and 7 days post-ICH, (c) illustration of the quantification method for hypo- and hyperintense regions on T2-weighted MRI. Quantification of total T2 lesion volume at 4 h (d; n = 10 for vehicle and n = 11 for caADAMTS13), 24 h (e; n = 9 for vehicle and n = 11 for caADAMTS13) and 7 days (h; n = 3 for vehicle and n = 3 for caADAMTS13) post-ICH. Quantification of hyperintense (f) and hypointense (g) regions on T2-weighted MRI at 24 h post-ICH and (i) representative T2*-weighted MRI images at 4, 24 h and 7 days post-ICH in vehicle- and caADAMTS13-treated mice. Quantification of iron-associated T2* volume at 4 h (j), 24 h (k) and 7 days (l). All volumes are expressed in mm³. Normality is assessed using the Shapiro–Wilk test, and statistical comparisons are performed using Welch’s t-test. Data are presented as mean ± SD.

ns: non-significant comparisons.

ICH increases cerebrovascular VWF and neutrophil deposition in mice

While previous studies have reported increased VWF expression locally within the haematoma following ICH, whether ICH induces cerebrovascular deposition of VWF and neutrophils more widely across the brain remains unclear. To address this, we first compared vascular VWF (4 h post-ICH) and neutrophil (24 h post-ICH) deposition between sham and ICH animals using molecular MRI with MPIO (Figure 3(a)). To confirm binding specificity, isotype control MPIOs were injected in parallel and showed only minimal R₂* signal (Supplemental Figure S1 and Supplemental Table 1). α-VWF-conjugated MPIOs were administered and detected by T2* MRI 4 h post-ICH (Figure 3(b)). Quantification of α-VWF-MPIO signal showed a significant increase in the whole brain vascular VWF deposition in haemorrhaged animals compared to sham animals (p = 0.0485; Figure 3(d)), primarily driven by VWF accumulation within the haematoma (p = 0.0013; Figure 3(e)). No significant differences in the extent of VWF deposition between sham and ICH animals were observed in the ipsilateral (p = 0.1037; Figure 3(f)), contralateral (p = 0.2606; Figure 3(g)) and pseudo lesion (p = 0.8559; Figure 3(h)).

Figure 3.

Intracerebral haemorrhage increases cerebrovascular VWF and neutrophil deposition detected by molecular MRI: (a) summary of the experimental design. Mice were subjected to sham or ICH surgery and injected with α-VWF- or α-Ly6G-conjugated MPIOs, (b) representative echo-averaged MRI images showing α-VWF- and α-Ly6G-MPIO binding in sham and ICH mice at 4 and 24 h post-ICH, (c) illustration of the quantification method for α-VWF- and α-Ly6G-MPIO signal regions on R₂* maps, (d–h) quantification of α-VWF-MPIO signal at 4 h post-ICH (n = 3 for sham and n = 8 for ICH): whole brain (d), lesion (e), ipsilateral hemisphere (f), contralateral hemisphere (g) and pseudo lesion (h) and (i–m) quantification of α-Ly6G-MPIO signal at 24 h post-ICH (n = 3 for sham and n = 6 for ICH): whole brain (i), lesion region (j), ipsilateral hemisphere (k), contralateral hemisphere (l) and pseudo lesion (m). Statistical analysis is performed using Welch’s t-test after confirming normality with the Shapiro–Wilk test. Data are presented as mean ± SD.

ns: non-significant comparisons.

Significant differences are indicated by *p < 0.05. **p < 0.01.

Similarly, ICH resulted in increased vascular neutrophil deposition across the brain 24 h following ICH compared to sham animals, assessed by α-Ly6G-conjugated MPIO (p = 0.0476; Figure 3(i)). Neutrophil-vascular interactions were elevated in the ipsilateral hemisphere (p = 0.0238, Figure 3(k)). No significant differences in α-Ly6G deposition were found within the haematoma (p = 0.0952; Figure 3(j)), in the contralateral hemisphere (p = 0.2619; Figure 3(l)) and in the pseudo lesion (p = 0.7024; Figure 3(m)). These findings confirm that ICH triggers quantifiable cerebrovascular inflammation both within and surrounding the haematoma that can be detected using molecular MRI.

caADAMTS13 has a delayed effect on vascular neutrophil deposition 7 days after ICH in mice

caADAMTS13 has previously been shown to modulate neutrophil migration both in vitro and in vivo.^19,20 To evaluate the effect of caADAMTS13 on VWF and neutrophils following ICH, MPIOs were used as described above (Figure 4(a)). Studies increasingly document a delayed, distinct wave of neutrophil recruitment to the brain.^33,34 In post-mortem tissue, neutrophils and NETs are present around fibrin fibres within the ICH lesion and the surrounding vasculature, indicating that neutrophils remain active and upregulated during the later wave of infiltration. ³⁴ We therefore investigated whether caADAMTS13 could modulate this later wave of neutrophil infiltration. At 7 days post-ICH, caADAMTS13 significantly reduced the extent of Ly6G-MPIO vascular distribution across the brain (p = 0.0325; Figure 4(b) and (c)), within the lesion (p = 0.0068; Figure 4(d)) and the ipsilateral hemisphere (p = 0.0024; Figure 4(d)). The extent of Ly6G-MPIO remained unchanged in the contralateral (p = 0.1414; Figure 4(f)) and pseudolateral region (p = 0.3828; Figure 4(g)). Ly6G and CD31 immunostaining showed that in the vehicle-treated mice, 67% of total neutrophils colocalised with CD31, compared to caADAMTS13 animals, where only 50% of neutrophils colocalised with CD31 (p = 0.0876; Supplemental Figure S3). After ICH, neutrophils are among the earliest immune cells recruited to the brain; therefore, we examined α-Ly6G-MPIOs at 24 h post-ICH. ³⁵ Quantification of Ly6G-MPIO showed no significant effect of caADAMTS13 on overall vascular neutrophil deposition (p = 0.7564; Figure 4(b) and (h)). Ly6G-MPIO signal within the lesion (p = 0.6607; Figure 4(i)), the ipsilateral (p = 0.8421; Figure 4(j)) and contralateral hemispheres (p = 0.7802; Figure 4(k)), as well as pseudo lesion (p = 0.6527; Figure 4(l)) remained unchanged between treatment groups 24 h post-ICH. At 4 h post-ICH, whole-brain VWF-MPIO signal did not change between vehicle and caADAMTS13 treated animals (p = 0.5490; Figure 4(b) and (m)), with no differences observed when quantified across the lesion (p = 0.5768; Figure 4(n)), nor the ipsilateral (p = 0.32180; Figure 4(o)) or contralateral hemispheres (p = 0.56190; Figure 4(p)), or pseudo lesion (p = 0.1426; Figure 4(q)) in the contralateral hemisphere.

Figure 4.

caADAMTS13 reduces vascular neutrophil deposition 7 days after ICH in mice: (a) summary of the experimental design. Mice were subjected to ICH and treated with vehicle or caADAMTS13 (caAD13) 1 h post-ICH. α-VWF- and α-Ly6G-conjugated MPIOs were used to assess cerebrovascular VWF and neutrophil deposition, (b) representative echo-averaged MRI images showing α-VWF- and α-Ly6G-MPIO binding in vehicle- and caADAMTS13-treated mice at 4, 24 h and 7 days post-ICH, (c–g) quantification of α-Ly6G-MPIO signal at 7 days post-ICH (n = 3 for vehicle and n = 3 for caADAMTS13): whole brain (c), lesion (d), ipsilateral hemisphere (e), contralateral hemisphere (f) and pseudo lesion (g), (h–l) quantification of α-Ly6G-MPIO signal at 24 h post-ICH (n = 9 for caADAMTS13 and n = 10 for caADAMTS13): whole brain (h), lesion (i), ipsilateral hemisphere (j), contralateral hemisphere (k) and pseudo lesion (l) and (m–q) quantification of α-VWF-MPIO signal at 4 h post-ICH (n = 10 for vehicle and n = 11 for caADAMTS13): whole brain (m), lesion (n), ipsilateral hemisphere (o), contralateral hemisphere (p) and pseudo lesion (q). Statistical analysis was performed using Welch’s t-test after confirming normality with the Shapiro–Wilk test. Data are presented as mean ± SD.

ns: non-significant comparisons.

Significant differences are indicated by *p < 0.05. **p < 0.01.

Delayed administration of caADAMTS13 reduced brain cell death and neutrophil recruitment after spontaneous ICH in zebrafish larvae

Many AIS therapies are limited by a narrow therapeutic window due to the bleeding risk associated with delayed treatment. ³ To directly assess the haemostatic safety of late intervention, we administered caADAMTS13 to bbh; ubiq:secAnnexin V-mVenus zebrafish larvae 12 h post-ICH (Figure 5(a)). Quantification of brain haemoglobin indicated that delayed caADAMTS13 treatment was associated with a significantly reduced area of o-dianisidine staining, indicating less blood in the brain (p = 0.0129; Figure 5(b) and (c)). In addition, the transgenic bbh; ubiq:secAnnexin V-mVenus zebrafish larvae develop clusters of Annexin V-positive cells after approximately 36 h of spontaneous ICH, providing a readout of ICH-induced brain cell death at this time point. ²⁵ caADAMTS13-treated larvae showed a 32% decrease in Annexin V signal intensity (p = 0.0050; Figure 5(d) and (e)), suggesting that delayed administration is safe and may mitigate brain cell death in the zebrafish larval model of ICH.

Figure 5.

Delayed administration of caADAMTS13 reduces cell death and brain neutrophil recruitment after spontaneous ICH in zebrafish larvae: (a) summary of the experimental design, (b) representative images of ICH+ bubblehead larvae stained with o-dianisidine 24 h after vehicle or caADAMTS13 (caAD13) treatment to visualise haemoglobin, (c) quantification of haematoma area (µm²) at 3 dpf (n = 32). Normality was assessed using the Shapiro–Wilk test, and statistical comparisons were performed using Welch’s t-test, (d) representative images of brain cell death in bubblehead larvae expressing ubiq:secAnnexin V-mVenus at 3 dpf. Larvae that did not develop a brain haemorrhage are shown as ICH− controls. Hatched lines outline the brain region used for Annexin V quantification. Images are acquired at 20× magnification, (e) quantification of mean Annexin V fluorescence from three independent replicates (n = 34 per group). Statistical analysis was performed using two-way ANOVA with Tukey’s post-hoc test, (f) representative images of mpo:GFP zebrafish larvae (ICH+ and ICH−) treated with vehicle or caADAMTS13, and (g) quantification of mpo+ cells in the brain from three independent replicates (vehicle ICH− n = 10, vehicle/caADAMTS13 ICH+ and ICH− n = 12). Statistical analysis was performed using two-way ANOVA with Tukey’s post-hoc test. Data are presented as mean ± SD.

dpf: days post-fertilisation; ns: non-significant comparisons.

Significant differences are indicated as *p < 0.05. **p < 0.01. ****p < 0.0001.

Our previous results show that ICH drives cerebrovascular neutrophil deposition in the murine model of ICH. We therefore hypothesised that caADAMTS13’s protective effects could involve modulation of neutrophil recruitment. To test this in zebrafish, bbh; mpo:GFP larvae were treated with caADAMTS13 12 h post-ICH, and MPO-positive neutrophils were quantified 12 h later (24 h post-ICH; Figure 5(a)). Delayed caADAMTS13 administration significantly reduced neutrophil numbers by 37% in the brain compared with vehicle-treated controls (p = 0.0087; Figure 5(f) and (g)).

Delayed caADAMTS13 treatment does not worsen grip strength in mice

Given that the striatum is the primary region affected by ICH in the mouse model and based on prior evidence that delayed caADAMTS13 administration may be protective in zebrafish, motor impairment and recovery were assessed using the grip strength test following caADAMTS13 treatment at 1 or 24 h post-ICH in mice (Figure 6(a)). H&E staining at 7 days post-ICH confirmed haematoma volumes were similar across early-caADAMTS13, delayed-caADAMTS13 and vehicle-treated groups (vehicle vs early caADAMTS13, p = 0.9129; vehicle vs delayed caADAMTS13, p = 0.6075; early vs delayed, p = 0.3673; Figure 6(b) and (c)). Forelimb grip strength did not differ between vehicle- and caADAMTS13-treated groups at 72 h post-ICH in the left paw (vehicle vs early p = 0.9787, vehicle vs delayed p = 0.9303; Figure 6(e)), while right paw and bilateral grip strength remained similar across groups (Figure 6(f) and (g)). We previously demonstrated that caADAMTS13 may reduce the extent of vascular neutrophil deposition at 7 days post-ICH. We therefore asked whether, in a separate cohort of animals, this reduction would affect motor outcomes. At 7 days post-ICH, delayed-caADAMTS13-treated animals showed no significant differences in the left paw grip strength (vehicle vs delayed p = 0.0516, vehicle vs early p = 0.9710; Figure 6(h)), with right paw and bilateral grip strength remaining similar across groups (Figure 6(i) and (j)). Importantly, caADAMTS13 did not alter the VWF plasma levels at 7 days post-ICH (Supplemental Figure S2).

Figure 6.

Delayed caADAMTS13 treatment does not worsen forelimb grip strength in mice: (a) summary of experimental design. Mice were treated with vehicle, early-caADAMTS13 (1 h post-ICH) or delayed-caADAMTS13 (24 h post-ICH) following ICH induction, (b) representative H&E-stained coronal brain sections at 7 days post-ICH, (c) quantification of haematoma volume (mm³), (d) illustration of the method used to quantify haematoma volume, (e–g) forelimb grip strength at 72 h post-ICH (n = 9 for vehicle, n = 8 for early caADAMTS13 and n = 10 for delayed caADAMTS13). Left paw (e), right paw (f) and bilateral grip strength (g) and (h–j) forelimb grip strength at 7 days post-ICH (n = 9 for vehicle, n = 8 for early caADAMTS13 and n = 10 for delayed caADAMTS13). Left (h), right (i) and bilateral paw grip strength (j). Ordinary one-way ANOVA was used to analyse the data. Data are presented as mean ± SD.

ns: non-significant comparisons.

Discussion

By cleaving pathogenic VWF, caADAMTS13 shows efficacy in rodent model of AIS.^19,21 Establishing that caADAMTS13 is also safe in ICH would offer a major therapeutic advantage over current AIS treatments, particularly by supporting its potential use in ultra-early stroke management prior to brain imaging. In this study, we showed that hyperacute administration of caADAMTS13 is not detrimental in the context of active bleeding in two preclinical models of ICH. We also showed that delayed administration of caADAMTS13 may have beneficial effects after ICH. Importantly, we were also able to show that caADAMTS13 may influence cerebrovascular neutrophil interactions following ICH.^19–21

In mice, we demonstrated that ICH induces a significant accumulation of VWF within the haematoma compared to sham animals, suggesting that VWF may be a key driver of clot formation soon after ICH. ³⁶ The mouse collagenase model directly damages the vascular basement membrane and may therefore exaggerate VWF deposition compared with the single-vessel rupture observed clinically. ³⁷ In the autologous blood model of ICH, parenchymal injection of femoral blood also led to local upregulation of VWF; however, this does not reflect a physiological increase directly resulting from vessel rupture. ³⁶ Therefore, although VWF expression may be exaggerated in this model, it is a more relevant platform than the autologous blood model for evaluating whether caADAMTS13 interferes with early haemostatic processes. Consistent with the literature, we demonstrated that neutrophils quickly interact with the vasculature following ICH, both within and around the haematoma, in the ipsilateral hemisphere. The detection of these cerebrovascular changes between sham and ICH animals allowed us to confirm the sensitivity of molecular imaging MRI to investigate the effects of caADAMTS13 on vascular thromboinflammatory markers in ICH. ³⁸

Unlike autologous blood injection models, which produce static haematomas, the models chosen here are characterised by active bleeding, allowing direct evaluation of caADAMTS13 on haematoma expansion. ²³ Acute administration of caADAMTS13 did not alter haematoma volume in the murine and zebrafish model of ICH. After ICH, erythrocyte lysis releases free haemoglobin, which has been shown to inhibit ADAMTS13. Haemoglobin binds to the ADAMTS13 cleavage site on VWF, thereby blocking ADAMTS13 access and reducing VWF proteolysis (Supplemental Figure S4). ³⁹ We therefore hypothesise that caADAMTS13 activity is likely to be inhibited within the haematoma core, allowing ultra-large VWF multimers to drive clot formation. ⁴⁰ This proposed mechanism remains speculative, and future studies will need to directly assess haemoglobin-mediated inhibition of caADAMTS13 within the haematoma. However, this interpretation aligns with our α-VWF-MPIO data, which showed no difference in vascular VWF deposition within the haematoma core 4 h following treatment. caADAMTS13 was administered intravenously, so it is immediately proteolytically active. However, to determine whether caADAMTS13 regains enzymatic activity as free haemoglobin concentrations decrease over time, α-VWF-MPIO measurements should be repeated at later time points. Although VWF blockade has been shown to reduce perihaematomal oedema in the autologous blood ICH model, ³⁶ no differences in T2-weighted hyperintensity were observed 24 h following hyperacute caADAMTS13 treatment. This discrepancy likely reflects model-specific injury mechanisms, as early oedema in the collagenase model is more pronounced than the autologous blood model due to direct collagenase-induced damage to multiple cerebral vessels, a process that might be independent of VWF-mediated pathways.^41,42

While prior studies reported reduced parenchymal neutrophils with VWF blockade or recombinant ADAMTS13,^19,20,23,36 our α-Ly6G MPIO data detect interactions of neutrophils within the vascular compartment rather than parenchymal infiltration. At 24 h post-ICH, the α-Ly6G MPIO signal is observed across the ipsilateral hemisphere in both vehicle and caADAMTS13 animals, suggesting widespread cerebrovascular endothelial activation following ICH. Importantly, although based on a small sample size, caADAMTS13 significantly reduced these vascular neutrophil interactions by 7 days post-ICH, predominantly within the haematoma and ipsilateral hemisphere. Ex vivo Ly6G staining showed that up to 60% of total neutrophils colocalised with CD31, suggesting that these are vascular rather than parenchymal neutrophils. However, we did not observe the same significant difference between caADAMTS13- and vehicle-treated animals ex vivo as in vivo. This could reflect that MPIOs may still be bound to Ly6G, as they can take up to 24 h to clear, ⁴³ thereby reducing the availability of the epitope for binding. Immunohistochemical analysis is limited to the sampled brain sections, whereas MPIOs can be quantified across the brain. Although speculative, this delayed effect of caADAMTS13 may suggest that neutrophil-endothelial interactions may evolve dynamically in the days following ICH, potentially increasing their reliance on VWF-mediated adhesion at later stages following vascular injury. This may coincide with decreasing haemoglobin concentrations within the haematoma that allow caADAMTS13 to regain proteolytic activity between 24 h and 7 days after ICH. ⁴⁴ Future work using single-cell RNAseq and flow cytometry will be required to define neutrophil subset dynamics and their dependence on VWF for vasculature interactions after ICH.

The risk of haemorrhagic complications of current AIS treatment significantly increases with time after stroke onset. We therefore tested whether, in the context of ICH, delayed administration of caADAMTS13 would exacerbate ICH-induced brain injury. ²³ When caADAMTS13 was administered 12 h post-ICH in the zebrafish model, it did not exacerbate bleeding. Remarkably, it reduced the amount of blood in the brain though whether this was due to a reduction in haematoma expansion or increased red blood cell clearance was not assessed. Furthermore, delayed treatment in zebrafish also reduced ICH-induced brain cell death and the extent of neutrophil recruitment to the brain. Delayed caADAMTS13 treatment was then tested in the murine ICH model; however, it did not improve motor outcomes and showed no significant histological change in haematoma size 7 days after ICH. These findings are limited by the use of only the grip strength test to evaluate motor outcomes. Therefore, future studies should include additional behavioural assays, such as rotarod, adhesive tape removal or cylinder test, to evaluate motor outcomes. Importantly, from a translational stroke perspective, delayed caADAMTS13 treatment did not worsen motor recovery or haematoma volume in mice.

While these findings are encouraging, they must be interpreted carefully, and limitations in study design should be considered. VWF function in haemostasis is largely conserved between zebrafish larvae and mammals; however, evidence regarding the precise mechanisms of thrombosis and haematoma clearance in the zebrafish bbh model is still limited. ²⁹ Moreover, mouse safety data are limited to behavioural outcomes and histological data without a parallel in vivo assessment of haematoma volume. Nevertheless, our data suggest that delayed administration of caADAMTS13 may not inherently aggravate ICH-related injury. Although we did not specifically assess the haemostatic safety of caADAMTS13 in peripheral organs, previous evidence showed that caADAMTS13 did not prolong tail bleeding time in a mouse AIS model. ¹⁹ Here, we observed no reduction in systemic VWF. However, future studies will need to address the peripheral haemostatic safety of caADAMTS13, particularly in aged animals.

It is important to note that these studies were conducted in young healthy animals, which may not fully recapitulate the cerebrovascular instability or comorbidities of older stroke patients. In addition, although we included both male and female mice, this study was not powered to stratify the analysis by sex. Appropriately powered studies will need to be conducted to investigate the influence of both sex and age. However, this study provides initial preclinical evidence that caADAMTS13 can be safely administered beyond the established therapeutic window for AIS, and even in the context of active ICH.

While administration of a thrombolytic agent in the context of ICH may seem counterintuitive, our results demonstrate that caADAMTS13 is safe in two complementary preclinical models of ICH, despite its thrombolytic properties in AIS. Furthermore, preliminary observations in the zebrafish larval model of ICH suggest that delayed administration of caADAMTS13 may be beneficial, but mechanisms remain unclear. These findings support further investigation of caADAMTS13 as a therapy that may enable more rapid, and potentially pre-hospital, intervention in AIS, even before stroke subtype is confirmed by imaging, and may help extend the therapeutic window, which currently reduces treatment eligibility. Importantly, our work also highlights a novel temporal aspect of caADAMTS13’s ability to modulate vascular-neutrophil interactions following ICH, which may reflect aspects of post-ICH vascular physiology that require further investigation.