Exacerbation of hemorrhagic stroke and impaired hemostasis in mice deficient of the integrated stress response kinase heme-regulated inhibitor

Abstract

BACKGROUND:

Heme-regulated inhibitor (HRI/EIF2AK1) is a protein kinase that initiates the integrated stress response (ISR) following mitochondrial damage and/or oxidative stress. Dependent on a context, ISR may promote death or recovery of cells under stress. This study was initiated to determine HRI contributions to brain damage after experimental intracerebral hemorrhage (ICH) where oxidative toxicity of hemoglobin is a major trigger of neural cell loss.

MATERIALS AND METHODS:

Publicly available transcriptomic data from a mouse model of ICH were analyzed using standard bioinformatic tools. Hri^−/− and wild-type control mice were subjected to unilateral collagenase-induced ICH followed by evaluation of forelimb functional deficits (adhesive tape removal), brain histology (Fluoro-Jade B staining of degenerating neurons and immunofluorescence for GFAP and CD36), and extravasated brain hemoglobin assays. Tail bleeding time and platelet number were determined to assess hemostasis.

RESULTS:

ISR transcripts were robustly overrepresented in the transcriptome of the peri-hematoma region of wild-type mice at 1-day post-ICH induction, suggesting acute ISR activation. After ICH, Hri^−/− mice showed greater impairment of forelimb function and exacerbated brain damage as indicated by increases of (1) GFAP-negative lesion area, (2) number of degenerating neurons, and (3) signal area of the activated macrophage marker CD36. These effects correlated with acute increases of hemoglobin extravasation, lower platelet counts, and prolonged tail bleeding.

CONCLUSIONS:

While ISR is activated by ICH and may play a role in secondary injury, impaired hemostasis is a likely driver of enhanced ICH severity in Hri^−/− mice.

Introduction

Intracerebral hemorrhage (ICH) is associated with a severe damage of brain tissue that follows intraparenchymal bleeding.[1] After ICH, primary damage is caused by locoregional hematoma mass effect, clot expansion, and perihematomal edema.[1] ICH is followed by secondary injury to the perihematoma tissue that is driven by oxidative toxicity of hemoglobin- and iron-related neuroinflammation.[1] Unfortunately, apart from risk factor control and early surgical evacuation of select intracerebral hematomas, there are no clinically effective treatments to reduce ICH-associated neurological impairment and mortality.[1]

The integrated stress response (ISR) is activated by a variety of pathological or physiological stimuli that perturb proteostasis and/or disturb mitochondria.[2] The activated ISR attempts to restore cellular homeostasis. However, when the homeostasis restoration is unsuccessful, ISR may initiate cell death or cause persistent cell dysfunction. In addition, beyond its role in response to tissue and cell damage, ISR regulates various physiological processes, including cell differentiation, immunity, and synaptic plasticity.

Transient phosphorylation of the eukaryotic translation initiation factor 2α (eIF2α) is the key component of ISR activation.[2] As a result, general protein synthesis is inhibited while translation of stress-regulated mRNAs is increased. Those include the ISR transcription factors ATF4 and CHOP/DDIT3, which initiate the ISR-associated gene expression program. There are four major ISR kinases that phosphorylate eIF2α, including HRI/EIF2AK1, PKR/EIF2AK2, PERK/EIF2AK3, and GCN2/EIF2AK4. Each ISR kinase responds to distinct stress stimuli as exemplified by ER stress or mitochondrial/oxidative stress responsiveness of PERK or HRI, respectively.

ISR activation has been reported in a rat ICH model of autologous blood injection.[3] Moreover, reducing post-ICH ISR by a treatment with a pharmacological PERK inhibitor attenuated neuronal degeneration. Opposite effects were reported when post-ICH ISR was enhanced by treatment with a drug inhibitor of the peIF2a phosphatase. Likewise, activation of the ISR transcription factor ATF4 has been proposed to enhance neuronal ferroptosis in a mouse ICH model of intrastriatal collagenase injection.[4] However, the pathogenic role of ISR kinases other than PERK has not been tested in ICH.

HRI activates ISR in response to oxidative- or mitochondrial stress as well as cytosolic protein aggregation, proteasome inhibition, and heme depletion.[2,5] Mice with a germline Hri knock out (Hri^−/−) display no major phenotype under physiological conditions.[6] However, the Hri^−/− genotype exacerbates various pathologies of erythropoiesis,[7] increases the lethality of pro-oxidative toxins and at least some bacterial pathogens,[8,9] and is associated with age-dependent α-synuclein aggregation in the spinal cord.[10] In addition, Hri^−/− macrophages display attenuated inflammatory potential.[11,12] Consistently, reduced inflammation and lung tissue protection were shown in Hri^−/− mice that were infected with the influenza virus.[12] However, there are no published reports of any Hri^−/− phenotypes in mouse models of stroke, including ICH. The current study was initiated to determine the role of HRI in a mouse model of collagenase-induced ICH.

Materials and Methods

Animals

Constitutive Hri^−/− (B6.Cg-Eif2ak1^tm1Jjch/MdfJ, Jackson Lab Cat# 035666, C57Bl/6J background) and C57Bl/6J wild type (WT) mice were bred in house. Animals were maintained in a standardized light/dark cycle (12 h/12 h) with free access to food and water. Mice of both sexes were used at the age of 10–13 weeks.

Intracerebral hemorrhage surgery

ICH induction surgeries followed the published methodology.[13] Briefly, mice were anesthetized with 100 mg/kg of ketamine + 10 mg/kg of xylazine (intraperitoneal, i.p.) followed by 1 mL of 0.9% NaCl (subcutaneous, s.c.) and topical application of eye ointment. Animal heads were placed in the stereotaxic frame with ear bars and nose cone. A half-inch-long midline scalp incision was made, and a 10 μL Hamilton syringe with a 26 gauge needle was positioned directly over the bregma to adjust all x, y, and z coordinates to 0. A burr hole was drilled through the skull 0.2 mm posterior and 2.0 mm lateral from bregma on the left side. The needle was inserted to a depth of 4 mm, and 0.075 U of Clostridium histolyticum collagenase VIIs (Sigma, C2399) in 1 μL of phosphate-buffered saline (PBS) was injected by an UltraMicroPump 3 (World Precision Instruments, Sarasota, FL) at 2 μL/min. The needle was left in place for an additional 5 min to avoid backflow. The burr hole was filled with bone wax (Ethicon, Somerville, NJ), and the skin incision was closed with cyanoacrylate glue (Henkel Consumer Adhesive Inc., Scottsdale, AR). One mg/mL buprenorphine SR-LAB (ZooPharm) was administrated subcutaneously to provide postsurgery analgesia. During surgery and recovery from anesthesia, animals were placed on heating pads set to 37°C. Of the 36 mice that underwent ICH surgery, one animal was lost shortly after anesthesia administration; no ICH animals reached humane endpoints before planned euthanasia for tissue collection. Brain tissue was collected at day 1 or 7 post-ICH (dpi, 5 WT and 6 Hri^−/− mice or 13 WT and 11 Hri^−/−, respectively). Animal numbers used for specific assessments are provided in the respective figures.

Adhesive tape removal test

The adhesive tape removal test was performed following the previously described methodology.[14] Briefly, a 2.5 mm × 2.5 mm adhesive tape strip was placed on each forelimb paw using equal force. Strips consistently covered the hairless part of the forepaws. The time to remove the tape from each paw was determined by two observers after placing animals in an experimental cage that contained home cage bedding. Two consecutive 5-min trials were performed to obtain average tape removal time. Mice were habituated to the assessment room, handling, and test for 3 consecutive days before the injury. Test results from the 3^rd day were used as baselines. Injured mice were subjected to the test at dpi 1 and 5. If the tape was not removed during the trial period, the tape removal time was recorded as 5 min.

Tissue collection for histological analyses

At dpi 7, deeply anesthetized mice were transcardially perfused with PBS and then 4% paraformaldehyde in PBS. After 48 h of 4% paraformaldehyde postfixation and 3 days of dehydration (30% sucrose in PBS), the brains were mounted in TFM tissue freezing medium (GeneralData, Cincinnati, OH). Twenty-micrometer coronal cryostat sections were cut, mounted on slides, and stored at −20°C until further use.

Fluoro-Jade B staining and quantification

Staining with Fluoro-Jade B (FJB, Millipore, AG310, Temecula, CA) was performed following the manufacturer’s recommendations. The number of FJB-positive neurons was counted for the hematomal and perihematomal area, excluding cortex, and in the cerebral cortex separately. Six sections were analyzed from each animal spanning the center of the hematoma (as defined by the maximal area of hemosiderin stain) as well as the region located 240 μm anterior and posterior from the lesion epicenter (two sections/region).

Immunofluorescence

A standard immunofluorescence procedure[15] was used with primary antibodies to GFAP (Millipore AB5541, 1:500 in 5% goat serum, 0.1% Triton X-100 in PBS, 4 C, overnight) or CD36 (BD Pharmingen 552544, 1:200 in 5% goat serum, 0.1% Triton X-100 in PBS, 4°C, overnight). Fluorescently-labeled secondary antibodies were applied at room temperature for 1 h. Stained sections were examined with Nikon Eclipse Ti inverted fluorescent microscope or C2 Nikon confocal microscope (Nikon Instruments Inc., Melville, NY). Quantitative analyses of positive (CD36) or negative staining areas (GFAP) were performed using digitalized images and the NIH ImageJ freeware with background corrections. At least two sections/animals cut through the ICH epicenter (as determined by the greatest area of hemosiderin stain) were analyzed to determine average area values for each mouse.

Extravasated brain hemoglobin assay

An established procedure was followed.[16] Deeply anesthetized mice were transcardially perfused with PBS to remove intravascular blood. Brain tissue was collected, and hemispheres were carefully divided. Each hemisphere was homogenized with Dounce homogenizer and then with Tissue-Tearor (BioSpec Products Inc., Bartlesville, OK) in ice-cold PBS supplemented with 0.16 mg/mL heparin. After centrifugation (10,000 g, 15 min at 4°C) hemoglobin assay was performed using the supernatant and the Hemoglobin Colorimetric Assay Kit (Cayman Chemical, Ann Arbor, MI).

Tail bleeding assay

Tail bleeding assay was performed using naïve, uninjured mice (WT, four females and six males; Hri^−/−, four females and eight males) following previously published methodology.[17] After anesthesia (100 mg/kg of ketamine + 10 mg/kg of xylazine i.p.), animals were placed in the prone position, and a distal 5-mm tip of the tail was amputated with a sharp blade to avoid crushing the tissue. The tail was then immersed in a 50-mL conical tube with prewarmed 0.9% NaCl (37℃). The tail tip was positioned 2 cm below the body level. Each animal was observed for 20 min. The initial bleeding time (to the first bleeding termination) and total bleeding time were recorded. After the test, blood was collected from the orbital sinus for blood cell count analysis, and mice were euthanized.

Meta-analysis of transcriptomic data from intracerebral hemorrhage mice

The NCBI GEO data set GSE200575 that represents RNAseq analysis of brain perihematomal tissue from the collagenase ICH model, was used (Li et al., 2022).[18] The data set includes transcriptomes of ICH and sham control mice (C57Bl/6 males at 8–10/young/or 20–22/old/weeks of age, n = 3 per condition). Raw sequencing counts were retrieved from GEO, filtered to remove genes with fewer than 10 counts, and input to DESeq2 (v1.40.2) for differential expression analysis. DESeq2 normalizes the raw counts using relative log expression (RLE) prior to performing differential expression. ICH upregulated transcripts (log₂ fold change > 1, q < 0.05) were identified for each age. Then, gene ontology enrichment analysis was performed for those transcripts using NCBI’s DAVID and the Pathway Browser (reactome.org). Overlaps between ICH upregulated genes and CHIPSeq-confirmed target genes of the ISR transcription factors ATF4/CHOP[19] were analyzed using hypergeometric testing (http://nemates.org/MA/progs/overlap_stats.html). In those calculations, the number of all identified transcripts (24,546 or 24,918 for young or old mice, respectively) was used to define the size of the whole perihematomal transcriptome.

Statistical analyses

Repeated measures ANOVA followed by Bonferroni post hoc t-tests was used for analyzing the results of the adhesive tape removal test. The binominal proportion test was used to compare percentages of animals which were able to remove adhesive tape during a trial period versus nonperformers. The parametric t-test was used to analyze the number of FJB-positive cells, bleeding time, hemoglobin concentration, and blood cell counts. Ratiometric data, including fold changes of the staining area, were analyzed using the nonparametric Mann–Whitney U-test.

Results

Acute integrated stress response activation in a mouse intracerebral hemorrhage model

To test for ISR activation in a mouse collagenase injection model of ICH, an analysis of a public RNASeq dataset from mouse perihematomal tissue was performed.[18] At 24 h post-ICH, 658 and 477 transcripts were highly upregulated in young (8–10 week old) and old (20–22 week old) mice, respectively (log₂ fold change > 1, q < 0.05). Intriguingly, response to stress was among the top five enriched gene ontology: biological process (GO: BP) terms among those mRNAs [Figure 1a]. In addition, the highly ICH upregulated transcripts showed significant overlaps with CHIPSeq-validated target genes for the key ISR transcription factors, ATF4 and CHOP/DDIT3[17] [Figure 1b]. Examples of canonical ISR genes that are acutely upregulated after ICH in both young and old mice include Atf3, Atf5, Chop/Ddit3, Ppp1r15a/Gadd34, Tnfrsf10b, and Trib3 [Figure 1c]. Therefore, ICH is associated with a strong activation of the ISR-associated transcription.

Figure 1.

The integrated stress response (ISR) induced mRNAs are upregulated after mouse intracerebral hemorrhage (ICH). The bioinformatic meta-analysis used public transcriptomic data from peri-hematoma region at 1 day post collagenase-induced ICH in 8–10 (young) or 20–22-week-old (old) male C57Bl/6 mice (GEO GSE200575,[18] n = 3/group, sham surgery controls). (a) The top five most enriched gene ontology terms (GO: BP) among ICH-upregulated mRNAs (fold sham control > 1, q < 0.05) include “response to stress.” (b) ICH-upregulated transcriptome shows overrepresentation of the ISR-induced genes whose transcription is directly regulated by ATF4 and/or CHOP/DDIT3 (ATF4/CHOP CHIPSeq dataset GSE35681 from ER stressed MEF cells;[17] p values from a hypergeometric test of significance are shown). (c) Examples of the ISR-induced transcripts that are upregulated by ICH

Increased functional impairment and greater brain damage in Hri^−/− mice after intracerebral hemorrhage

As the HRI-ISR activating stimuli, including oxidative stress, mitochondrial damage, and neuroinflammation are all implicated in the pathogenesis of ICH, the ICH phenotype of Hri^−/− mice was investigated. One day post-ICH injury (1 dpi), 10/11 Hri^−/− (90.9%) vs. 1/13 WT (7.7%) mice were inactive with no signs of fur grooming, P < 0.001 [Figure 2a]. The sensory-motor function of the forelimbs was assessed using an adhesive tape removal test. At baseline, each genotype showed similar performance proficiency with no signs of forelimb asymmetry (Hri^−/− vs. WT mice tape removal time of 10.9 ± 1.5 vs. 15.6 ± 2.3 s or 13.2 vs. 17.8 from the contralateral or ipsilateral side, respectively, P > 0.05) [Figure 2b and Tables 1 and 2]. As compared to baseline, contralateral forelimb tape removal time of Hri^−/− mice increased at 1 and 5 dpi while no significant effects were observed in the WT group [Figure 2b and Tables 1 and 2]. Moreover, a significant delay of tape removal from the contralateral paw was found in Hri^−/− versus WT during either post-ICH assessment (158.2 ± 96.3 s vs. 72.7 ± 90.4 s in Hri^−/− versus WT group on dpi 5, P < 0.05) [Figure 2b and Tables 1 and 2]. A transient ipsilateral impairment was observed in Hri^−/− versus WT mice on dpi 1 [Figure 2b and Tables 1 and 2]. These results indicate worsened acute severity and greater functional deficits in Hri^−/− mice.

Figure 2.

Worsened functional impairment after intracerebral hemorrhage (ICH) in Hri^−/− mice. ICH was induced by unilateral intracerebral collagenase injection. (a) At 1-day post-ICH induction (day postinjury, dpi), % mice with ungroomed coat was higher in Hri^−/− versus WT group (***P < 0.001, binominal proportion test).(b) At the indicated times after ICH induction motor skills were evaluated using front paw adhesive tape removal test. At the contralateral site, Hri^−/− mice showed increased tape removal time at dpi 1 and dpi 5 (two-way repeated measures ANOVA showed significant effects of genotype (F_1,77 = 25.0, P < 0.001), time after ICH induction (F_2,75 = 41.2, P < 0.001), and tested side (F_1,77 = 16.8, P < 0.001). Tukey post hoc test P values for Hri^−/− versus WT comparisons are indicated [^nsP > 0.05, *P < 0.05, ***P < 0.001, Tables 1 and 2 for more results]. Averages ± standard error of the mean are shown. Data were collected from three independent experiments. ICH: Intracerebral hemorrhage

Table 1.

Statistical analysis of adhesive tape removal test, repeated measures ANOVA

	F	Degrees of freedom	P
Effect of “genotype”	25.0	1.77	< 0.001
Effect of “time point”	41.2	2.75	< 0.001
Effect of “forelimb side”	16.8	1.77	< 0.001
Combined effect of “genotype” and “time point”	10.6	2.75	< 0.001
Combined effect of “genotype” and “forelimb side”	3.1	1.77	0.0001
Combined effect of “time point” and “forelimb side”	10.0	2.75	< 0.001

Table 2.

Statistical analysis of adhesive tape removal test, Bonferroni post hoc t-tests

	Time point; Genotype; Forelimb side
	Baseline				Day 1 post-ICH				Day 5 post-ICH
	WT		Hri −/−		WT		Hri −/−		WT		Hri −/−
	Ipsilateral	Contralateral	Ipsilateral	Contralateral	Ipsilateral	Contralateral	Ipsilateral	Contralateral	Ipsilateral	Contralateral	Ipsilateral	Contralateral
Adhesive tape removal time (s)	13.2±2.7	10.9±1.5	17.8±4.3	15.6±2.3	38.8±12.8	72.4±25.4	119.5±39.9	217.2±33.6	22.9±10.2	72.7±26.1	50.3±18.7	158.2±30.4
Baseline
WT
Ipsilateral
Contralateral	NS
Hri−/−
Ipsilateral	NS	NS
Contralateral	NS	NS	NS
Day 1 post-ICH			< 0.005		< 0.05
WT				< 0.001		= 0.001	< 0.05
Ipsilateral	NS	NS	NS	NS
Contralateral	NS	NS	NS	NS	NS
Hri−/−
Ipsilateral	NS	NS	< 0.005	NS	< 0.05	NS
Contralateral	NS	NS	NS	< 0.001	NS	0.001	< 0.05
Day 5 post-ICH
WT
Ipsilateral	NS	NS	NS	NS	NS	NS	NS	NS
Contralateral	NS	NS	NS	NS	NS	NS	NS	NS	NS
Hri−/−
Ipsilateral	NS	NS	NS	NS	NS	NS	NS	NS	NS	NS
Contralateral	NS	NS	NS	< 0.001	NS	NS	NS	NS	NS	< 0.05	< 0.005

WT: Wild type, ICH: Intracerebral hemorrhage, NS: Not significant, Hri^-/-: Hri knock-out

Consistently with enhanced forelimb dysfunction in Hri^−/− mice, their brains showed greater damage. The seven dpi brain sections that were stained for the reactive astrocyte marker GFAP revealed a greater maximal lesion area as defined by the absence of the GFAP signal [Figure 3a, a’ and b]. For both genotypes, the maximum damage was consistently observed on coronal sections cut through the beginning of the dorsal hippocampus. Lesions were localized to the subcortical structures, including the striatum, with potential involvement in the lateral thalamus and medial amygdala [Figure 3a and a’]. As compared to WT mice, the Hri^−/− maximal lesion area increased by 35% [Figure 3b]. A higher number of degenerating, FJB ⁺ neurons was also observed in the ipsilateral cortex of Hri^−/− mice [Figure 3c and d]. Specifically, in the preserved tissue at and around the injury epicenter (240 mm rostral or caudal), the average count of FJB ⁺ neurons was 269 ± 114 or 404 ± 58 cells/section in WT or Hri^−/− mice, respectively, P < 0.05 [Figure 3d]. Compared to WT mice, that represents a 50% increase in delayed neurodegeneration after ICH. Likewise, a 76% increase in a positive staining area for the activated macrophage marker CD36 was observed in sections across the lesion epicenter of Hri^−/− versus WT mice, P < 0.05 [Figure 3e and f].

Figure 3.

Greater brain tissue damage in Hri^−/− mice 7 days postintracerebral hemorrhage. Representative images of GFAP immunostaining (a and a’), Fluoro-Jade B (FJB) staining (c), and CD36 immunostaining (e) on coronal brain sections cut through the lesion epicenter ipsilateral to the collagenase injection site. In a, low power images of the ipsilateral hemispheres. In a’, high power images of areas outlined by the squares in a. White lines delineate the lesion boarder as defined by the lack of GFAP-positive reactive astrocytes (arrows). In c, degenerating, FJB-positive neurons are indicated by white arrowheads. (b) Increased lesion area fraction of the ipsilateral hemisphere in Hri^−/− mice. These animals also showed a higher number of FJB-positive neurons (d) and the greater area of CD36 immunostaining (f) in the ipsilateral hemisphere. The graphs depict averages ± standard deviation (*P < 0.05, U-test)

Increased intracranial bleeding and disrupted hemostasis in Hri^−/− mice

At 7 days post-ICH, brain sections from Hri^−/− mice showed increased hemosiderin stain in the ipsilateral hemisphere, suggesting greater hematoma (52% or 125% increase over WT in hemosiderin + area or integrated density of hemosiderin signal, P < 0.05 or P < 0.01, respectively) [Figure 4a, b and b’, data from individual animals are in Supplementary Table 1]. To determine whether such a hematoma enlargement is due to greater initial hemorrhage, content of hemoglobin was measured in ipsilateral and contralateral hemispheres 24 h post-ICH induction. Both genotypes showed higher hemoglobin in the ipsilateral versus contralateral hemisphere (P < 0.05 or < 0.01 for WT or Hri^−/− mice, respectively) [Figure 4c]. However, ipsilateral hemoglobin levels were 63% higher in Hri^−/− versus WT mice, P < 0.05 [Figure 4c]. Contralateral hemoglobin levels were unaffected by genotype, P > 0.05 [Figure 4c]. These data suggest that the greater severity of ICH in Hri^−/− mice is driven primarily by increased bleeding following collagenase injection.

Figure 4.

Increased intracerebral bleeding and disrupted hemostasis in Hri^−/− mice. (a, b and b) Increased hemosiderin signal in the lesion area of Hri^−/− mice 7 days postintracerebral hemorrhage (ICH). (a) Representative bright field images of fixed, unstained coronal brain sections; hemosiderin signal is indicated by Black triangles are black arrowheads and indicate hemosiderin signal in the lesion. (b and b’) Quantifications of hemosiderin signal; hemosiderin-positive area (% total section area), or integrated density of hemosiderin stain (fold WT) are shown in b or b’, respectively. (c) Increased hemoglobin content in the ipsilateral hemisphere of Hri^−/− versus WT mice 24 h post-ICH induction. (d) Tail bleeding test revealed increases of total bleeding time in Hri^−/− vs. WT mice. (e) Lower blood platelet count in Hri^−/− vs. WT mice. Additional blood cell count data are presented in Table 3. Data represent group averages ± standard deviation (b, c and e) or individual animals (d); ^nsP > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001, U-test. For b, two independent experiments were performed and combined for analysis

Supplementary Table 1.

Quantification of hemosiderin signal

	Area of hemosiderin/brain section area (%)	Hemosiderin integrated density
WT mouse 1	7.80	1.06945
WT mouse 2	6.39	0.8821
WT mouse 3	4.71	0.74065
WT mouse 4	4.71	0.58832
WT mouse 5	9.91	1.62862
WT mouse 6	7.95	1.09087
Hri−/− mouse 1	11.92	2.51915
Hri−/− mouse 2	10.54	2.01923
Hri−/−- mouse 3	8.92	2.26269
Hri−/− mouse 4	6.51	1.25777
Hri−/− mouse 5	13.68	3.83772
Hri−/− mouse 6	11.28	1.64086

WT: Wild type, Hri^−/−: Hri knock out

Importantly, prolonged tail bleeding was observed in Hri^−/− versus WT mice (600 ± 26.8 vs. 200 ± 423.4s, respectively, P < 0.01) [Figure 4d] suggesting disruption of physiological hemostasis. Moreover, Hri^−/− mice showed a 41.4% decline in blood platelet count (PLT, Hri^−/− vs. WT, 490.2 ± 122.3 vs. 835.8 ± 87 × 10³/μΛ, P < 0.001) [Figure 4e]. Additional hematologic abnormalities included moderate but significant decreases in red blood cell (RBC) count and total blood hemoglobin (HGB) together with increases in RBC volume (MCV) and cellular hemoglobin content (MCH) [Table 3]. These effects are consistent with moderate hyperchromic anemia, as previously reported in Hri^−/− mice.[6] However, the Hri^−/− -associated PLT reduction is unlikely to explain defective hemostasis as > 90% PLT drop is needed to significantly extend tail bleeding time in mice.[20] Therefore, the requirement(s) of HRI for platelet function and/or other aspects of hemostasis may underlie the anti-hemostatic phenotype of Hri^−/− mice, including exacerbated ICH.

Table 3.

Complete blood count analysis of uninjured Hri^−/− and wild-type mice

Parameter	WT	Hri −/−	P	Hri−/− effect
PLT (k/μL)	835.8±87.04	490.17±122.26	0.00051	Decreased
RBC (m/μL)	10.6±0.53	9.46±0.4	0.0027	Decreased
HGB (g/dL)	15.22±0.75	14.38±0.44	0.046	Decreased
HCT (%)	53.28±2.95	51.37±2.04	0.24	-
MCV (fL)	50.24±0.7	54.33±1.12	0.00006	Increased
MCH (pg)	14.38±0.05	15.22±0.45	0.0025	Increased
MCHC (g/dL)	28.58±0.38	28.03±0.53	0.087	-
RDW-SD (fL)	32.4±1.54	34.3±1.65	0.082	-
RDW-CV (%)	25.54±0.81	23.85±0.45	0.0018	Decreased
RET (k/μL)	378.22±48.54	506.52±124.46	0.059	-
PDW (fL)	7.28±0.26	7.35±0.15	0.59	-
MPV (fL)	6.0±0.16	6.33±0.29	0.0501	-
P-LCR (%)	3.16±0.73	4.62±1.46	0.074	-
Procalcitonin (%)	0.5±0.058	0.31±0.072	0.00097	Decreased
WBC (k/μL)	7.45±0.72	5.91±1.93	0.13	-
Neutrophils (k/μL)	1.57±0.73	0.98±0.39	0.12	-
Lymphocytes (k/μL)	5.65±1.13	4.75±1.55	0.31	-
Monocytes (k/μL)	0.15±0.095	0.08±0.033	0.13	-
Eosinophils (k/μL)	0.082±0.03	0.088±0.038	0.77	-
Basophils (k/μL)	0.006±0.0055	0.005±0.0084	0.82	-

Average±SD, P values of two-tailed Student’s t-test. WBC: White blood cell, MPV: Mean platelet volume, P-LCR: Platelet-large cell ratio, PDW: Platelet distribution width, RET: Reticulocytes, RBC: Red blood cell, MCHC: MCH concentration, MCH: Mean corpuscular hemoglobin, MCV: Mean corpuscular volume, PLT: Platelets, HGB: Hemoglobin, HCT: Hematocrit, WT: Wild type, RDW-SD: RBC distribution width-SD, RDW-CV: RBC distribution width-coefficient of variation, SD: Standard deviation

Discussion

Using a collagenase ICH model, we show increased acute ICH severity and greater intracranial bleeding in Hri^−/− mice. The latter phenomenon is the most likely driver of greater functional deficits and increased tissue damage, as ICH pathogenesis is mediated directly by blood extravasation. Moreover, impairment of physiological hemostasis may play a role in increased intracranial hemorrhage in Hri^−/− animals.

Genetic and pharmacological data implicate another ISR kinase, PERK, in the regulation of hemostasis. In a mouse model of carotid artery injury, megakaryocyte/platelet-specific KO of PERK increased thrombus formation.[21] Conversely, activation of the PERK-mediated ISR in tumor cells promotes expression and/or secretion of the tissue factor that activates the plasma coagulation cascade.[22] Moreover, PERK inhibition is anti-thrombotic following carotid artery injury.[23] Therefore, PERK may play a complex role in hemostatic regulation. In platelets, functional PERK reduces injury-induced thrombosis. However, higher levels of PERK activity in platelets and/or other cells may be pro-thrombotic. While PERK has been identified as a component of the human or mouse platelet proteome under physiological conditions, HRI has not.[21,24] Therefore, the hemostasis supported by HRI may involve nonplatelet cells.

HRI is activated by mitochondrial stress to promote the functional integrity of mitochondria.[25,26,27] In the erythroid lineage, HRI is activated under conditions of low iron/heme supply, reducing globin translation, mitochondrial metabolism, and erythroid differentiation.[25] In macrophages, HRI supports erythrophagocytosis and recycling of iron back to erythropoiesis.[11] One can speculate that one or more of those various aspects of HRI-mediated regulation may be involved in hemostasis. Thus, megakaryocyte selective KOs of several key regulators of mitochondrial gene expression result in an extended tail bleeding time and moderate thrombocytopenia.[28] Moreover, RBCs, whose efficient generation is supported by HRI,[6] are significant contributors to hemostasis.[29] Their mechanical and biochemical characteristics modulate the activation of the coagulation cascade and platelets. Therefore, in various forms of anemias, bleeding time is extended.[29] Of note, the erythroid lineage cells, which share a common progenitor with megakaryocytes,[30] are the top expressors of HRI.[31] While Hri^−/− mice show only moderate changes in RBC count/HGB concentration under physiological conditions, MCV and MCH are both increased[6] [Table 3]. Such an unusual hyperchromia may reduce RBC stiffness, thereby lowering their pro-hemostatic activities.[29] It is also possible that the hemostatic phenotype of Hri^-/- mice has several causes including abnormal RBCs, reduced thrombopoiesis, platelet dysfunction as well as other, yet unidentified, defects of the hemostatic system. Future studies, including direct ex vivo analysis of platelet activation/aggregation and blood clotting activity, will be needed to address the mechanism of hemostatic disruption in Hri^−/− mice.

Importantly, our unexpected observations suggesting reduced hemostasis as a mechanism of increased ICH severity in Hri^−/− mice do not exclude HRI’s role in blood extravasation-induced secondary injury. The autologous blood injection model of ICH may be suitable for future studies to address this issue separately from the consequences of altered hemostasis.

The current dataset has several limitations that should be acknowledged. First, the collagenase-induced ICH that was used here produces a diffuse hematoma originating from many small vessels as opposed to focal, arterial bleeding in most clinical cases.[32] Therefore, the presumed pro-hemorrhagic consequences of Hri deficiency should also be tested in other intracranial hemorrhage models, including a subarachnoid hemorrhage induced by a direct perforation of large cerebral arteries.[33] Second, a phenotype in germline Hri^−/− line may have developmental origins and/or be a result of compensation for lack of HRI-driven ISR. Therefore, potential pro-hemorrhagic effects of HRI deficiency should re-examined once specific pharmacological inhibitors and/or conditional knock-out mice become available. Finally, there are gender differences in ICH severity, with female mice usually showing improved survival and recovery.[34] While the current study includes mice of both genders, it is statistically underpowered to conclusively analyze interactions between sex and HRI status that may affect ICH severity and outcome.

Taken together, we uncovered an anti-hemostatic phenotype in Hri^−/− mice that exacerbates ICH. Pharmacological modulation of HRI is an emerging therapeutic approach in cancer.[35,36] Therefore, an important question to ask is whether transient or acute inhibition/stimulation of HRI activity will impact hemostasis. Since the risk of bleeding and/or thrombosis is increased in various cancers,[37] anti/pro-hemostatic effects of HRI therapeutics need to be determined in a relevant cancer context. Similar scrutiny may apply to other potential indications, including neurotrauma.

Conclusions

Increased ICH severity and greater ICH-induced brain damage are observed in mice deficient in the ISR kinase HRI. These effects are associated with increased hematoma and reduced physiological hemostasis. Therefore, the lack of pro-hemostatic activity of HRI is a likely contributor to the worsened ICH outcome in Hri^−/− mice.

Author contributions

Concept (LPS, SSO, DD, MH), design (LPS, RLO, DD, MH), definition of intellectual content (MH, DD), literature search (LPS, MH, DD), experimental studies (LPS, RLO, ERH, JM), data acquisition (LPS, JM, RLO, MH), data analysis (LPS, MH, JHC, ECR), statistical analysis (DAB, MH, LPS, JHC, ECR), manuscript preparation (LPS, MH), manuscript editing and reviewing (MH, DD), supervisor (MH).

Ethical policy and institutional review board statement

All procedures were approved by the Institutional Animal Care and Use Committee and Institutional Biosafety Committee (ethical approval number: IACUC 21966, approval date: April 27, 2022, University of Louisville, Kentucky, USA), and strictly adhered to NIH guidelines on the use of experimental animals. ARRIVE guidelines were followed.

Data availability statement

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Conflicts of interest

There are no conflicts of interest.

Funding Statement

This work was supported by NS103433 (Michal Hetman), Jewish Heritage Fund for Excellence Research Enhancement Grant (Michal Hetman), the Commonwealth of Kentucky Challenge for Excellence (Michal Hetman). Bioinformatics support for this work was provided by P20 GM103436 and P20 GM106396 (Eric C. Rouchka).