Heme–Hemopexin Scavenging Is Active in the Brain and Associates With Outcome After Subarachnoid Hemorrhage
Background and Purpose—Long-term outcome after subarachnoid hemorrhage (SAH) is potentially linked to cytotoxic heme. Free heme is bound by hemopexin and rapidly scavenged by CD91. We hypothesized that heme scavenging in the brain would be associated with outcome after hemorrhage.
Methods—Using cerebrospinal fluid and tissue from patients with SAH and control individuals, the activity of the intracranial CD91–hemopexin system was examined using ELISA, ultrahigh performance liquid chromatography, and immunohistochemistry.
Results—In control individuals, cerebrospinal fluid hemopexin was mainly synthesized intrathecally. After SAH, cerebrospinal fluid hemopexin was high in one third of cases, and these patients had a higher probability of delayed cerebral ischemia and poorer neurological outcome. The intracranial CD91–hemopexin system was active after SAH because CD91 positively correlated with iron deposition in brain tissue. Heme–hemopexin uptake saturated rapidly after SAH because bound heme accumulated early in the cerebrospinal fluid. When the blood–brain barrier was compromised after SAH, serum hemopexin level was lower, suggesting heme transfer to the circulation for peripheral CD91 scavenging.
Conclusions—The CD91–heme–hemopexin scavenging system is important after SAH and merits further study as a potential prognostic marker and therapeutic target.
Subarachnoid hemorrhage (SAH) is an example of extravascular hemolysis with a high mortality and morbidity and associated economic cost. After hemolysis, cell-free hemoglobin undergoes oxidation to methemoglobin, which through several hemichrome intermediates, finally degrades into a denatured globin protein and the redox-active heme moiety (iron–protoporphyrin IX). Because of a combination of redox activity and lipophilicity, free heme is toxic in many ways, including covalent modification of substrates, intercalation in the lipid bilayer, and lipid peroxidation, which perturbs membrane homeostasis to cause cellular dysfunction and cell death.1 Hemopexin neutralizes the redox toxicity of heme by formation of the heme–hemopexin complex, which prevents heme from generating free radical reactions,2 and leads to its uptake by CD91.3 The CD163–haptoglobin system is the body’s first line of defense during hemolysis, but this system is saturated after SAH, and free hemoglobin is detectable in the cerebrospinal fluid (CSF).4 In this situation, the CD91–hemopexin system is likely to be important.
Hemopexin is an abundant plasma protein and it is also expressed by neurons and glia.5 The relative contribution of these 2 sources to human CSF hemopexin is unknown. Also, the response of the human hemopexin–CD91 scavenging system to SAH has not been studied. This study addresses these questions by analyzing CSF and brain tissue from patients after SAH and controls.
Materials and Methods
Participants were recruited after referral to tertiary centers in Manchester, Birmingham, Southampton, and Cambridge, with respective Research Ethical Committee approvals. The characteristics of control participants (n=20) and patients with SAH (n=30) in the main study are listed in Table I in the online-only Data Supplement. CSF in patients with SAH was obtained from external ventricular drains. Control participants were patients with noninflammatory/nonhemorrhagic conditions who underwent lumbar puncture and were subsequently found to have normal CSF with respect to protein, glucose, cell count, cytology, albumin CSF/serum quotient, and isoelectric focusing for oligoclonal bands. Seven patients with SAH were recruited to clinical study 2, to enable further analysis of heme, which required more CSF (Table II in the online-only Data Supplement). Heme quantitation was performed using an in-house validated ultrahigh performance liquid chromatography technique; established immunoassays were used for hemopexin and albumin (online-only Data Supplement).
Postmortem brain tissue from SAH (n=7) and matched control (n=5) cases was obtained from the University Hospital Southampton National Health Service Foundation Trust as part of the UK Brain Archive Information Network. Tissue sections were selected close to the bleed; so, anatomic region varied in individual cases. The mean SAH-to-death interval was 16 days (range, 11–25 days); further characteristics are available in the online-only Data Supplement. CD91 and iron were analyzed by immunohistochemistry and Perls staining, respectively (more details are available in the online-only Data Supplement).
The distribution of each data set was assessed, and parametric or nonparametric tests were used accordingly, as indicated in the text. Statistical tests were conducted at the 5% 2-sided significance level using SPSS version 21.
Hemopexin Is Mainly Produced Intrathecally in Control CSF
The CSF hemopexin reference range as determined by ELISA was 12.3 to 32.6 μg/mL; the mean concentration was 22.4 μg/mL. There was no sex difference (mean, 21.5 and 24.6 μg/mL in women and men, respectively; P=0.23, unpaired t test). A well-established and accepted technique to determine intrathecal synthesis of blood-derived proteins similar to albumin is the intrathecal index, defined as the CSF/serum ratio of hemopexin (Qhemopexin) divided by the CSF/serum ratio of albumin (Qalbumin).6 Albumin is a plasma protein that is not synthesized in the brain and is wholly derived from plasma via diffusion across the blood–brain barrier (BBB). With a molecular weight of 69 kDa and a molecular size of 25.6 Å, albumin has the appropriate biophysical parameters to act as a reference protein for the diffusion of hemopexin across the BBB (hemopexin: molecular weight, 68 kDa and molecular size, 36 Å). Qhemopexin was significantly greater than Qalbumin (P<0.0001, Wilcoxon signed-rank test; Figure 1A); the intrathecal index (Qhemopexin/Qalbumin) was 10.5. Thus, the vast majority of hemopexin is produced intrathecally in control individuals, with only about one tenth being derived from the circulation.
CD91–Hemopexin–Heme Scavenging System Is Present and Active in Human Brain After SAH
CD91 immunohistochemistry on human brain tissue revealed expression in neurons and glia (Figure 1B). Uptake of heme–hemopexin complexes leads to intracellular deposition of heme’s iron moiety.7 Perls staining to quantify iron deposition revealed a significantly greater deposition of iron in SAH versus control cases (P=0.028, Mann–Whitney test; Figure 1C). Regardless of whether there was blood clot in the sections, iron deposition significantly correlated with CD91 (Spearman r=0.79; P=0.0025; Figure 1D), indicating that the CD91–hemopexin system actively scavenges heme after SAH.
Increased CSF Hemopexin After SAH Is Associated With Poor Outcome
After SAH, CSF hemopexin had a bimodal distribution; in 30% of patients (n=9/30), CSF hemopexin was above the reference range (Figure 2A). In high versus normal CSF hemopexin patients, delayed cerebral ischemia occurred more frequently and neurological outcome 6 months after SAH, as assessed by the modified Rankin Scale (mRS), was poorer (delayed cerebral ischemia: 57% versus 11%; P=0.028, Fisher exact test and mean mRS: 5.0 versus 2.4; P=0.025, unpaired t test). The difference in Glasgow Outcome Scale was not statistically significant (mean, 2.5 versus 3.9; P=0.122, unpaired t test). The difference in outcome between high and normal CSF hemopexin groups could not be explained by several other factors related to bleed size or severity (Table III in the online-only Data Supplement). Although more women than men had a high CSF hemopexin, this was not significant, and CSF hemopexin was not significantly different between the sexes (25.2 and 23.3 μg/mL in women and men, respectively; P=0.699, Mann–Whitney test). Overall, this indicates that CSF hemopexin level is associated with outcome after SAH and may be a potentially useful prognostic marker, independent of bleed size.
Source of Increased CSF Hemopexin After SAH
The 7-fold increase in CSF hemopexin (Table III in the online-only Data Supplement) could be blood or brain derived. The hemopexin intrathecal indexwas significantly raised in high CSF hemopexin patients (4-fold; P<0.001, unpaired t test), indicating partial intracranial origin (Figure 2B). The blood-derived fraction could have 2 sources: the initial bleed itself and increased transfer from the circulation via a more permeable BBB. If the initial bleed was the predominant source of plasma proteins in the CSF, one would expect a significant negative correlation between sampling time and Qalbumin; however, this was not present (Pearson correlation coefficient=−0.045; P=0.826). In support of increased transfer from the circulation across a compromised BBB, there was a significant increase in BBB permeability in the high CSF hemopexin group as evidenced by a raised Qalbumin (2-fold; P=0.033, unpaired t test; Figure 2C). Hence, the predominant source of increased CSF hemopexin associated with poor outcome was intrathecal, with some derived from the circulation.
Saturation of Heme–Hemopexin Uptake After SAH
The high CSF hemopexin suggested saturation of heme–hemopexin uptake. To look for this, SAH CSF was examined for the presence of heme–hemopexin complexes. Sufficient CSF from 1 patient in clinical study 1 was available for ultrahigh performance liquid chromatography analysis to detect bound heme; in this patient, who had a high CSF hemopexin level (98.5 μg/mL), a substantial amount of heme was bound to hemopexin and albumin (Figure 3A, inset; Figure I in the online-only Data Supplement, peak at 4.5 minutes). In 7 other patients with high-grade SAH, with repeated CSF sampling (clinical study 2), bound heme was found at all time points, confirming rapid saturation of heme–hemopexin uptake, up to at least day 13 after SAH (Figure 3A).
Free Heme After SAH
In the same 7 patients from clinical study 2, unbound heme was assayed by performing ultrahigh performance liquid chromatography before and after adding recombinant hemopexin to the CSF sample, the difference in area-under-the-curve representing unbound heme. All samples tested showed an increase in the peak for bound heme when preincubated with recombinant hemopexin, indicating the presence of free, unbound heme in the CSF, up to at least day 13 after SAH (Figure 3B).
Evidence for Efflux of Free Heme From the Brain
After SAH, a small and lipophilic molecule, such as heme, can theoretically diffuse out of the brain into the bloodstream down a steep concentration gradient across the BBB. There was a significant drop in serum hemopexin after SAH (mean, 0.72 mg/mL in controls versus 0.55 mg/mL after SAH; P=0.002, unpaired t test; Figure 3C). This decrease was confined to patients with the highest Qalbumin, in whom BBB permeability to heme would be the highest (P=0.04, unpaired t test; Figure 3D).
This is the first study to characterize the intrathecal CD91–hemopexin system in control individuals and after SAH. Intrathecal production is likely to be the main source of CSF hemopexin, with only about one tenth being derived from the circulation. Hemopexin is known to be produced within the brain, and CD91 is not expressed by cerebral endothelium5,8; so, receptor-mediated transport of hemopexin into the brain is unlikely but cannot be excluded. Still, CSF hemopexin was 10-fold lower compared with serum, suggesting that the brain has a comparatively lower heme-binding capacity.
After SAH, the CD91–hemopexin scavenging system is active in the brain because CD91 significantly correlated with iron deposition. CSF hemopexin had a bimodal pattern after SAH. Despite similar bleed size and severity, high CSF hemopexin patients had a poorer outcome, with higher delayed cerebral ischemia rates and higher mRS scores. This needs replication because CSF hemopexin may be of clinical utility as a prognostic marker. The Glasgow Outcome Scale was not significantly different between individuals with normal and high CSF hemopexin; however, such a discrepancy between Glasgow Outcome Scale and mRS has been noted after SAH before9 and may be related to the higher number of scoring categories of the mRS being able to differentiate neurological sequelae with greater subtlety. In addition, mRS is a stroke outcome scale, more akin to SAH, whereas Glasgow Outcome Scale was developed as a scale for use in traumatic head injury.
The elevation of CSF hemopexin, and its relationship with outcome, is intriguing, and the mechanism is not clear. Hemopexin has been reported to be neuroprotective,10 and therefore, CSF hemopexin is not assumed to be toxic. However, it is also possible that when CSF hemopexin is too high, it becomes deleterious by binding heme, preventing its efflux from the brain and resulting in intracellular heme/iron overload, which may be toxic to neurons/glia. Another potential explanation is that events after SAH, such as saturation of heme–hemopexin uptake or inflammation, result in high CSF hemopexin and poor outcome, which are, therefore, indirectly associated. Clearly, further mechanistic studies are needed.
CSF hemopexin had 2 sources: (1) accumulation from an intrathecal origin (as indicated by a 3-fold rise in the hemopexin intrathecal index) and (2) increased transfer from the circulation (supported by a 2-fold higher Qalbumin). Accumulation of brain-derived hemopexin after SAH may occur as a result of increased synthesis or decreased CD91-mediated scavenging. Increased synthesis may be secondary to the host inflammatory response because the hemopexin promoter is interleukin-6 responsive,11 and interleukin-6 levels are elevated in SAH CSF.12 Decreased CD91-mediated scavenging may be because of plateauing in heme–hemopexin uptake because heme and hemopexin were detected simultaneously in the CSF. Also, because CD91 has multiple ligands, it is possible that there is competition for CD91 from other ligands, such as ApoE.13 Interestingly, CSF ApoE is known to decrease after SAH,13 and low CSF ApoE associates with a poor outcome,14 similar to patients with high CSF hemopexin in this study.
We thank Ardalan Zolnourian, Southampton General Hospital, for the help with Table II in the online-only Data Supplement.
Sources of Funding
This study was supported by Peel Medical Research Trust (I. Galea), Engineering and Physical Sciences Research Council (I. Galea), Royal College of Surgeons of Edinburgh (D. O. Bulters, I. Galea, and P. Garland), Smile for Wessex (D. O. Bulters, I. Galea), Medical Research Council (J. Galea), and National Institute for Health Research (A.J. Durnford).
Galea and D. Bulters were supported to study hemopexin, and A.I. Okemefuna is employed by Bio Products Laboratory Limited.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.115.011956/-/DC1.
- Received October 24, 2015.
- Revision received October 24, 2015.
- Accepted November 9, 2015.
- © 2016 American Heart Association, Inc.
- Gutteridge JM,
- Smith A.
- Hvidberg V,
- Maniecki MB,
- Jacobsen C,
- Højrup P,
- Møller HJ,
- Moestrup SK.
- Wildemann B,
- Oschmann B,
- Reiber H
- Reiber H.
- Immenschuh S,
- Nagae Y,
- Satoh H,
- Baumann H,
- Muller-Eberhard U.
- Kay A,
- Petzold A,
- Kerr M,
- Keir G,
- Thompson E,
- Nicoll J.
- Kay A,
- Petzold A,
- Kerr M,
- Keir G,
- Thompson E,
- Nicoll J.