Anti‐tissue Plasminogen Activator (tPA) as an Effective Therapy of Neonatal Hypoxia–Ischemia with and without Inflammation

Summary

Hypoxic–ischemic brain injury is an important cause of neurodevelopmental deficits in neonates. Intrauterine infection and the ensuing fetal inflammatory responses augment hypoxic–ischemic brain injury and attenuate the efficacy of therapeutic hypothermia. Here, we review evidences from preclinical studies suggesting that the induction of brain parenchymal tissue‐type plasminogen activator (tPA) plays an important pathogenic role in these conditions. Moreover, administration of a stable‐mutant form of plasminogen activator inhibitor‐1 called CPAI confers potent protection against hypoxic–ischemic injury with and without inflammation via different mechanisms. Besides intracerebroventricular injection, CPAI can also be administered into the brain using a noninvasive intranasal delivery strategy, adding to its applicability in clinical use. In sum, the therapeutic potential of CPAI in neonatal care merits further investigation with large‐animal models of hypoxia–ischemia and cerebral palsy.

Introduction

Advances of neonatal care in the past four decades have markedly increased the survival rate of preterm births and infants suffering from asphyxia or hypoxic–ischemic (HI) brain injury 1. Yet, as high as 37% of infants with intrauterine or neonatal insults still develop long‐term neurological deficits, making brain protection in HI injury an unfulfilled goal 2. Further, preterm infants have a higher risk of intrauterine and neonatal infection, which is a compounding factor that reduces the efficacy of therapeutic hypothermia and worsens neurological outcomes 3, 4, 5. In short, while hypothermia protects against moderate HI, there is an unmet need for more effective treatments of severe HI brain injury in neonates.

The premises of this review, primarily based on our recent studies 6, 7, 8, 9, are that neonatal brains have a unique response to HI by the induction of tissue‐type plasminogen activator (tPA), which is a double‐edged sword because it prevents HI‐induced thrombosis in blood vessels, but unleashes a multitude of deleterious effects in the parenchyma of immature brains (Figure 1A). These deleterious effects include tissue proteolysis, edema secondary to the blood–brain barrier (BBB) damage, microglia activation, glutamate excitotoxicity to neurons and oligodendrocytes and are mediated through tPA or its downstream mediators. Yet, tPA‐antagonism with CPAI, a stable‐mutant form of plasminogen activator inhibitor‐1 (PAI‐1) 10, is a promising therapy of neonatal HI brain injury with and without inflammation.

Figure 1.

Induction of tPA as a double‐edged sword following neonatal hypoxia–ischemia (HI). (A) Overall, hypothesis depicting that neonatal HI induces tPA both inside and outside cerebral blood vessels. The up‐regulated tPA activity inside blood vessels contributes to fibrinolysis and rapid reperfusion, while the extravascular tPA synthesized by astrocytes and microglia degrades neurons, oligodendrocytes, BBB, and extracellular matrix (ECM) and mediates inflammation‐sensitized microglia activation. (B–D) Intracardiac FITC‐dextran perfusion revealed pockets of low blood flow in the cerebral cortex (Ctx) and the hippocampus (Hip), but not in the thalamus (Th), on the ipsilateral hemisphere at 1 h after the Vannucci model. These pockets of perfusion deficits disappeared by 4 h recovery. (E–G) Comparison of in‐situ plasminogen activator (PA) zymogram (E, 4 h after HI), tPA mRNA (F, 4 h after HI), and H & E stain (G, 7 d recovery) showed spatial correlation of tPA induction and HI‐triggered brain atrophy. R (Right) indicates the HI‐injured hemisphere. (H–J) Double‐labeling of tPA mRNA and antiglial fibrillary acidic protein (GFAP) indicates the expression of tPA by astrocytes that surround cerebral blood vessels (V) after HI injury. (K–M) Anti‐tPA and anti‐OX42 double‐labeling also indicates the expression of tPA by active microglia/monocytes near lateral ventricles (LV) following neonatal HI injury (Modified with permission from Ref #6).

tPA Induction is a Unique Response to Hypoxia–Ischemia in Neonatal Brains

A large body of experimental evidence indicates that tPA has complex functions in the nervous system 11, 12. By converting plasminogen to fibrinolytic plasmin, tPA is the mainstay of acute ischemic stroke therapy. Yet, tPA may also induce matrix metalloproteinases (MMPs), protease‐activated receptor (PAR‐1), and platelet‐derived growth factor (PDGF)‐CC to damage BBB and trigger hemorrhagic transformation during thrombolytic therapy 13, 14, 15. The brain parenchyma has a low basal level of tPA and plasminogen expression that is markedly up‐regulated by seizures and excitotoxins 16, 17. tPA also has plasmin‐independent mechanisms to stimulate microglia and the N‐methyl‐D‐aspartate (NMDA) receptor‐mediated excitotoxicity 18, 19, 20. While low level of tPA is neuroprotective 21, high levels of stress‐induced tPA likely exert an overall harmful effect in the brain.

In preclinical studies, the Vannucci HI model is the most commonly used paradigm to mimic neonatal hypoxic–ischemic encephalopathy (HIE). In the Vannucci model, the unilateral carotid artery of 7‐day‐old rats or 8‐ to 10‐day‐old mice were permanently ligated followed by a period of hypoxic challenge (typically 90 to 120 min) in temperature‐controlled chambers 22. Although neither carotid ligation nor systemic hypoxia triggers discernible brain injury, the combination of both insults produces unilateral brain tissue loss and white‐matter injury in the Vannucci model, but sparing the contralateral hemisphere as internal controls. The Vannucci HI model has contributed greatly to our understanding of the pathophysiology of HIE and remains instrumental for testing new therapies, including combinational hypothermia with adjuvants 23.

A salient feature of the Vannucci HI model is that it only suppresses cerebral blood flow (CBF) transiently, while the same HI insult induces blood coagulation and lasting reperfusion deficits in adult animals 24, 25. As shown in Figure 1B–D, intracardial injection of fluorescein isocyanate (FITC)‐conjugated dextran revealed patches of hypoperfusion in the carotid‐ligated hemisphere at 1 h after HI in P7 rat brain, but not on the contralateral side or at 4‐h recovery. Additional experiments showed the transient phase of CBF reduction was coupled to deposition of fibrin and platelets in cerebral blood vessels 6. These results suggest rapid induction of the plasminogen activator (PA) system for fibrinolysis after neonatal HI insult. Consistent with this notion, biochemical and histological analyses indicated the induction of tPA from 1 h and urine‐type plasminogen activator (uPA) from 4 h after HI, both of which remained up‐regulated at 24‐h recovery 6. This pattern contrasts those in adult brains showing decreased tPA activity after HI or ischemic insults 26, 27.

Further, in‐situ gelation/plasminogen zymography showed acute induction of PA activity correlated with tPA mRNA in the future degenerative areas at 4 h after the Vannucci HI insult (Figure 1E–G). Much of the tPA mRNA was distributed along the penetrating arteries in the HI‐injured cerebral cortex, consistent with its expected source in endothelial cells during fibrinolysis (Figure 1F). tPA was also produced by perivascular astrocytes and subventricular microglia or macrophages, as shown by double‐labeling (Figure 1H–M). The extravascular tPA induction may lead to tissue proteolysis and microglial activation underlying the HI brain injury in neonates.

Stable‐Mutant form of PAI‐1 (CPAI) is Potent Anti‐tPA Therapeutics

To test the pathogenic role of tPA during neonatal HI brain injury, we first examined the effects of intracerebroventricular (ICV) injection of α2‐antiplasmin, which decreased HI‐induced brain atrophy to half, but the therapeutic window was short for within 2 h postinsult 6. As tPA has plasmin‐independent harmful effects that will not be blocked by antiplasmin, we then tested a stable‐mutant form of PAI‐1 called CPAI 10 and found more robust protection.

PAI‐1, a 50 kDa glycoprotein, belongs to the serine protease inhibitor (serpin) family that includes many of the protease inhibitors found in the blood and tissues. The association between a serpin and its target protease occurs at a reactive cell loop (labeled as R in Figure 2A) located on the surface of serpin. Upon association with the target protease, cleavage of the reactive cell loop is initiated, which leads to rapid insertion of the cleaved loop into β‐sheets in the center of serpin and flipping of the target protease form one side to the other end of the serpin 28. This large conformational change, commonly described by a mouse‐trap model but also resembles a “Judo throw”, could distort and inactivate the target protease. In addition, serpin and the target protease are often frozen in a highly stable covalent acyl‐enzyme complex 28. PAI‐1 inhibits tPA and uPA with a second‐order rate constant at 10⁷/M/second, which is 10–1000 times faster than other PA inhibitors. PAI‐1 also inhibits plasmin and is considered the key regulator of plasmin generation in the blood. In contrast, basal expression of PAI‐1 in the brain is very low and not involved in the regulation of parenchymal tPA activity 6, 7.

Figure 2.

Inhibition of plasminogen activators with CPAI blocks HI‐induced MMP‐9 activation and brain damage. (A) The structure and conformational change of PAI‐1 before and after its interaction with tPA. R indicates the reactive cell loop, which is cleaved upon interaction with tPA and inserted into the β‐sheets of PAI‐1 in a “mouse‐trap” model. This conformational change may lead to target deformation and inactivation, in analogy to a “Judo throw”. (B) Plasminogen zymogram using recombinant tPA (rt‐PA) or the kidney extracts incubated with CPAI prior to gel electrophoresis showed attenuated tPA/uPA activity even when they were dissociated from the inhibitor. Shown are inverse image of a Coomassie blue‐stained gel. (C, D) Comparison of zymogram (tPA/uPA and MMP‐9/2) and immunoblots (against PAI‐1 and actin) with the brain extracts collected at 4 or 24 h after HI injury and saline‐versus‐CPAI treatment. Note that (a) post‐HI induction of PA occurred faster than the activation of MMP‐9/2, (b) CPAI treatment attenuated PA induction at both 4 and 24 h, and (c) inhibition of acute tPA/uPA induction by CPAI led to absence of MMP‐9/2 activation at 24‐h recovery. (E–M) Comparison of the effects of saline‐versus‐CPAI treatment on T2‐relaxation (E–G), apparent diffusion coefficient (H–J), and brain degeneration (K–M) showed significant protection by the CPAI treatment. Shown are representative images and quantification. *P < 0.001 compared to the saline‐injected animals (Modified with permission from Ref #7).

Native PAI‐1 exists in either an active form or an inactive latent conformation with the reactive cell loop imbedded in the β‐sheets. The half‐life of wild‐type PAI‐1 before converting to the latent form is ~2 h, but a four amino acid mutation (N150H, K154T, Q319L, M354I) extends its half‐life beyond 145 h 10. Further, as shown by plasminogen zymogram in Figure 2B, when recombinant or endogenous (kidney) tPA were incubated with CPAI and then separated by gel electrophoresis, the dissociated tPA had reduced proteolytic activity. The combination of >70‐fold longer half‐life and the ability to deform tPA makes CPAI a promising therapeutics against diseases associated with tPA neurotoxicity.

Robust Protection against Hypoxia–Ischemia with the CPAI Treatment

We first examined the effects CPAI in pure‐HI insults 7. When the HI‐injured rat pups were killed at 4 or 24 h after HI to compare the PA and MMP activity following ICV injection of saline or CPAI, it was evident that (a) the induction of tPA and uPA precedes MMP‐9 activation (b) the CPAI treatment not only attenuates acute PA induction, but also abolishes the subsequent MMP‐9 activity at 24‐h recovery (Figure 2C,D). These results suggest that activation of the PA system is needed for MMP‐9 induction, which is a critical mechanism for the demise of cerebral blood vessels in neonatal HI brain injury 29, 30, 31.

Besides attenuation of PA and MMP‐9 activation, therapeutic application of CPAI also reduced brain edema (indicated by increased T2‐relaxation time in magnetic resonance imaging), and brain damage (shown by reduction of apparent diffusion coefficient) at 24 h recovery (Figure 2E–J). The CPAI treatment also showed dose‐dependent reduction of brain atrophy with a ~4 h therapeutic window, when measured at 7 day after HI injury (Figure 2K–M). These results suggest an important role of tPA in neonatal HI brain injury as well as the therapeutic potential of CPAI.

Accelerated Microglia Activation in Inflammation‐Sensitized Hypoxia–Ischemia

Besides pure‐HI, inflammation‐compounded HI is another critical issue in neonatal care that deserves more study for several reasons. First, intrauterine infection (chorioamnionitis) is a major risk factor for cerebral palsy in term and near‐term infants 4. Clinical and animal studies suggested that the combination of HI and systemic inflammation produces greater brain damage and poorer response to hypothermia 32, 33, 34. A recent workshop on hypothermia sponsored by the NICHD cautioned, “cooling in the presence of infection may be deleterious” 5. Thus, better understanding the unique pathogenesis of infection‐HI brain injury is needed to develop more successful therapies.

In preclinical research, the most versatile model of infection‐HI brain injury is to expose rodent neonates to low‐dose lipopolysaccharide (LPS), the endotoxin of Gram‐negative bacteria that are common pathogens in chorioamnionitis, to mimic subclinical infection and then apply the Vannucci HI insult 4–72 h later 35, 36. In the standard Vannucci HI model, microglia are activated secondary to tissue damage in a delayed “sterile inflammation” manner 37, while in LPS‐sensitized HI, microglia activation comes out faster and stronger (Figure 3A). As shown in Figure 3B, the nuclear NFκB activity was intensely up‐regulated on the ipsilateral hemisphere at 4 h after LPS‐sensitized HI (either after 4 or 72 h exposure), but not by pure‐HI or low‐dose LPS stimulation alone. This pattern supports the hypothesis that “microglia are the convergence point for upstream (HI and infection) and downstream mechanisms in pathogenesis” of this condition 34. The combined LPS/HI insult also accelerated microglia activation in a Toll‐like receptor 4 and NFκB/MyD88‐dependent manner 36, 38, 39. Our recent study further indicated that the LPS endotoxin suppresses the proteolytic activity of tPA, but the nonprotease tPA activity is needed for microglia activation, and thus amenable to CPAI treatment (Figure 3C) 8.

Figure 3.

Inflammation‐sensitized HI accelerates neuroinflammation in neonatal brains, while therapeutic administration of CPAI mitigates inflammation‐sensitized neonatal HI brain injury. (A) Schematic drawing of key pathogenic mechanisms with different time scale after the onset of stroke or HI brain injury. This diagram emphasizes accelerated and amplified immune responses in inflammation‐sensitized HI brain injury of neonates. (B) Electrophoresis mobility shift assay (EMSA) showed acute NFκB activity on the carotid artery‐ligated hemisphere (R*), but not the contralateral side (L), at 4 h after HI injury that was sensitized by 4 or 72 h preexposure to low‐dose LPS (LPS4 h/HI or LPS72 h/HI). Pure‐HI injury and exposure to low‐dose LPS alone did not activate the NFκB activity. (C) Schematic diagram showing that LPS preexposure converts the mechanisms of HI brain injury from tPA‐mediated tissue proteolysis to microglia activation, in which the nonproteolytic tPA activity plays a critical role. (D–Q) Composite figures showing the evidence of CPAI‐mediated protection against LPS/HI injury in various assays, including acute NFκB activity at 4 h (D), tPA and MMP‐9 activity at 4 and 24 h (E), permeability to cardially injected sodium fluorescein (NaF) at 24 h (F), staining of Iba1‐positive microglia and OX42‐positive monocytes at 24 h (G–M), brain atrophy at 7 days recovery following CPAI treatment that was initiated at the indicated time‐points (P), and diffusion tensor imaging (DTI)‐based evaluation of white‐matter injury at 2 months of age (Q). Also shown are the effects of CPAI (N) and tPA‐versus‐uPA knockout mice (O) on the induction of TNFα after ICV injection of LPS (Modified with permission from Ref #8).

According to the literature, tPA may regulate microglial activation by two mechanisms. First, the seminal work by Dr. Tsirka et al. suggested that tPA mediates microglial activation by interaction with Annexin A2 18, 19. This notion is supported by later findings that Annexin A2 (an intracellular protein in normal circumstances) emerges to the cell surface in association with S100A10 to become a receptor for tPA and plasmin(ogen) 40, 41. The Annexin A2/S100 complex promotes tPA‐mediated fibrinolysis in the vascular wall and amplifies LPS/endotoxin‐TLR4 signaling in macrophages 42, 43, 44, 45. Whether this mechanism applies to microglia is yet to be tested. Second, Dr. Yepes proposed that LRP1 (the LDL receptor‐related protein 1) mediates tPA‐induced microglial activation 46, 47. While the notion of tPA‐LRP1‐mediated microglial activation is seemingly at odds with recent data showing that LPS‐induces LRP1‐intracellular domain (ICD) to dampen the NFκB activity in macrophages 48, 49, there is a way to reconcile current findings. We propose that, because LRP1 internalizes tPA and the tPA‐PAI1 complex for recycling and degradation 50, 51, 52, tPA may divert LRP1 from its otherwise inhibitory effects to up‐regulate the LPS/TLR4 signaling in a dis‐inhibition mode. Of note, these two scenarios of tPA‐modulated microglial activation are not mutually exclusive, and both remain to be tested.

CPAI Treatment Effectively Mitigates Inflammation‐Sensitized Hypoxia–Ischemia

We have evaluated the effects of ICV injection of CPAI against inflammation‐sensitized HI with the Vannucci model preceded by 4 h exposure to low‐dose LPS 8. The dual LPS/HI model provoked acute NFκB signaling and inhibited the tPA protease activity without depleting the tPA protein at 4 h recovery (Figure 3D,E). At 24 h, LPS/HI‐triggered MMP‐9 activity and the BBB permeability to sodium fluorescein dye (Figure 3F). In addition, the LPS/HI insult caused greater influx of OX42+ monocytes/macrophages into the subcortical white‐matter than pure‐HI (Figure 3G,H,J,K). In contrast, therapeutic administration of CPAI markedly reduced the acute NFκB induction, MMP‐9 activity, the BBB leakage, and the influx of OX42+ monocytes (Figure 3D–M). The CPAI treatment also reduced LPS/HI‐triggered brain atrophy with a >4 h therapeutic window (Figure 3P) and safeguarded near‐normal development of WM (Figure 3Q).

Moreover, CPAI administration mitigated the induction of TNFα following ICV injection of LPS, a paradigm for direct microglial activation (Figure 3N). Similarly, mice lacking tPA, but not uPA, showed attenuated TNFα production after LPS injection (Figure 3O). Together, these results suggest that tPA has a critical role in microglial activation and LPS/endotoxin‐sensitized HI brain injury in neonates.

Intranasal CPAI Delivery as an Effective Therapeutic Strategy in Neonates

We understand that ICV‐administration of CPAI has limited clinical appeal due to the invasive nature of this procedure. However, the CPAI protein at ~45 kDa molecular weight is too big to cross BBB in intravenous administration, but may interfere with the fibrinolytic activity in blood. Thus, to devise a better strategy in clinical use, we have examined the effects of intranasal delivery of CPAI in both pure‐ and LPS‐sensitized HI injury 9.

Intranasal drug delivery, proven effective to apply insulin into the brains of patients with Alzheimer's disease, uses the leaky olfactory nerve sheath to transport peptides and proteins first into the olfactory bulbs and in turn to the cerebral cortex 53, 54. As shown in Figure 4A and B, we readily detected PAI‐1 protein in the olfactory bulbs (30–60 min) and then in the cerebral cortex (60–120 min) following intranasal administration. This treatment also reduced LPS/HI‐triggered acute NFκB activity, pro‐inflammatory IL‐6 production, and brain tissue loss (Figure 4C,D) 9. These results suggest that intranasal delivery is an effective, noninvasive treatment of neonatal HI brain injury with and without infection sensitization.

Figure 4.

Intranasal CPAI delivery bypasses blood–brain barrier to confer therapeutic benefits. (A) Schematic diagram showing the anticipated transport path following intranasal delivery of CPAI (~45 kDa). (B) Immunoblotting analysis supported the expected transport of CPAI from the olfactory bulbs (OB, 30–60 min) to the cerebral cortex (60–120 min) after intranasal delivery. (C, D) Intranasal CPAI administration mitigated LPS/HI‐induced acute NFκB activity at 4 h, as well as induction of the pro‐inflammatory IL‐6 mRNA at 24 h. (Modified with permission from Ref #9).

Conclusions

Our recent studies as reviewed here add to the growing awareness that extravascular tPA, when expressed at an elevated level, has deleterious effects to the brain. In particular, neonatal brains have a unique response to HI by up‐regulation of the parenchymal tPA activity, which plays a critical role in pathogenesis of this condition. Therapeutic administration of the stable‐mutant form of PAI‐1 called CPAI confers impressive protection against neonatal HI injury with and without inflammation‐sensitization. Given these results in rodent models and the possibility to administer CPAI by the noninvasive intranasal strategy, it is warranted to test its therapeutic effects in nonrodent, large‐animal models of neonatal HI and cerebral palsy 55, 56. Currently, CPAI is available from Molecular Innovations Inc., but only for laboratory use. Future studies to produce Good Manufacturing Practice (GMP)‐grade CPAI is needed for clinical research of this powerful neuroprotective agent.

Conflict of Interest

The authors declare no conflict of interest.