Unravelling the twists and turns of the serpinopathies

Authors

  • Benoit D. Roussel,

    1.  Department of Medicine, University of Cambridge, Cambridge Institute for Medical Research, Wellcome Trust/MRC Building, UK
    Search for more papers by this author
  • James A. Irving,

    1.  Department of Medicine, University of Cambridge, Cambridge Institute for Medical Research, Wellcome Trust/MRC Building, UK
    Search for more papers by this author
  • Ugo I. Ekeowa,

    1.  Department of Medicine, University of Cambridge, Cambridge Institute for Medical Research, Wellcome Trust/MRC Building, UK
    Search for more papers by this author
  • Didier Belorgey,

    1.  Department of Medicine, University of Cambridge, Cambridge Institute for Medical Research, Wellcome Trust/MRC Building, UK
    Search for more papers by this author
  • Imran Haq,

    1.  Department of Medicine, University of Cambridge, Cambridge Institute for Medical Research, Wellcome Trust/MRC Building, UK
    Search for more papers by this author
  • Adriana Ordóñez,

    1.  Department of Medicine, University of Cambridge, Cambridge Institute for Medical Research, Wellcome Trust/MRC Building, UK
    Search for more papers by this author
  • Antonina J. Kruppa,

    1.  Department of Medicine, University of Cambridge, Cambridge Institute for Medical Research, Wellcome Trust/MRC Building, UK
    Search for more papers by this author
  • Annelyse Duvoix,

    1.  Department of Medicine, University of Cambridge, Cambridge Institute for Medical Research, Wellcome Trust/MRC Building, UK
    Search for more papers by this author
  • Sheikh Tamir Rashid,

    1.  Department of Medicine, University of Cambridge, Cambridge Institute for Medical Research, Wellcome Trust/MRC Building, UK
    2.  Laboratory for Regenerative Medicine, West Forvie Building, University of Cambridge, UK
    Search for more papers by this author
  • Damian C. Crowther,

    1.  Department of Medicine, University of Cambridge, Cambridge Institute for Medical Research, Wellcome Trust/MRC Building, UK
    2.  Department of Genetics, University of Cambridge, UK
    Search for more papers by this author
  • Stefan J. Marciniak,

    1.  Department of Medicine, University of Cambridge, Cambridge Institute for Medical Research, Wellcome Trust/MRC Building, UK
    Search for more papers by this author
  • David A. Lomas

    1.  Department of Medicine, University of Cambridge, Cambridge Institute for Medical Research, Wellcome Trust/MRC Building, UK
    Search for more papers by this author

D. Lomas, Department of Medicine, University of Cambridge, Cambridge Institute for Medical Research, Wellcome Trust/MRC Building, Hills Road, Cambridge CB2 0XY, UK
Fax: +44 (0)1223 336827
Tel: +44 (0)1223 762818
E-mail: dal16@cam.ac.uk

Abstract

Members of the serine protease inhibitor (serpin) superfamily are found in all branches of life and play an important role in the regulation of enzymes involved in proteolytic cascades. Mutants of the serpins result in a delay in folding, with unstable intermediates being cleared by endoplasmic reticulum-associated degradation. The remaining protein is either fully folded and secreted or retained as ordered polymers within the endoplasmic reticulum of the cell of synthesis. This results in a group of diseases termed the serpinopathies, which are typified by mutations of α1-antitrypsin and neuroserpin in association with cirrhosis and the dementia familial encephalopathy with neuroserpin inclusion bodies, respectively. Current evidence strongly suggests that polymers of mutants of α1-antitrypsin and neuroserpin are linked by the sequential insertion of the reactive loop of one molecule into β-sheet A of another. The ordered structure of the polymers within the endoplasmic reticulum stimulates nuclear factor-kappa B by a pathway that is independent of the unfolded protein response. This chronic activation of nuclear factor-kappa B may contribute to the cell toxicity associated with mutations of the serpins. We review the pathobiology of the serpinopathies and the development of novel therapeutic strategies for treating the inclusions that cause disease. These include the use of small molecules to block polymerization, stimulation of autophagy to clear inclusions and stem cell technology to correct the underlying molecular defect.

Abbreviations
ER

endoplasmic reticulum

ERAD

endoplasmic reticulum-associated degradation

FENIB

familial encephalopathy with neuroserpin inclusion bodies

hIPSC

human-induced pluripotent stem cell

NF-κB

nuclear factor-kappa B

UPR

unfolded protein response

Serpins and the mechanism of protease inhibition

Members of the serine protease inhibitors (serpin) superfamily are found in all branches of life and play an important role in the regulation of enzymes involved in proteolytic cascades. The superfamily is characterized by sharing more than 25% identity with the archetypal serpin α1-antitrypsin and conservation of tertiary structure. Serpins adopt a metastable conformation, composed in most cases of nine α-helices, three β-sheets (A to C) and an exposed, mobile, reactive centre loop. This flexible reactive centre loop typically contains 20 residues that act as a pseudosubstrate for the target protease [1,2]. After formation of a Michaelis complex, the enzyme cleaves the P1-P1′ bond of the serpin, releasing the P1′ residue and forming an ester bond between the protease and the serpin. This is then followed by a dramatic transition from a ‘stressed to relaxed’ (S→R) conformation, which flips the enzyme from the upper to the lower pole of the serpin as the reactive loop inserts as an extra strand in β-sheet A [3,4]. This conformational transition is subverted by mutations to form chains of ordered polymers associated with a group of diseases that we have termed the serpinopathies [5,6]. In the present review, we describe the pathophysiology of the serpinopathies and consider new strategies for therapies.

α1-Antitrypin deficiency is the archetypal serpinopathy

α1-antitrypsin is an acute phase glycoprotein that is synthesized by the liver and is present in the plasma at a concentration of 1.5–3.5 g·L−1. It functions primarily as an inhibitor of the enzyme neutrophil elastase. Most individuals are homozygous for the M allele, with the most common clinically relevant variants being the severe Z (Glu342Lys) and the mild S (Glu264Val) deficiency alleles. The retention of Z α1-antitrypsin within hepatocytes causes protein overload that manifests as periodic acid-Schiff positive inclusions associated with neonatal hepatitis, cirrhosis and hepatocellular carcinoma [7,8] (Fig. 1). Only 10–15% of Z α1-antitrypsin is released into the circulation; this leaves the lungs exposed to enzymatic damage by neutrophil elastase and so predisposes the Z homozygote to early onset panlobular emphysema. The S allele results in less retention of protein within hepatocytes, with plasma levels that are 60% of those of the M allele. This does not result in any clinical sequelae.

Figure 1.

 (A) The classical pathway of polymerization. The Z mutation (Glu342Lys, arrow head) or shutter domain mutations (red circle) destabilize β-sheet A (blue) to form an activated intermediate species. Intermolecular linkage occurs via donation of the reactive loop (red) from one molecule to the open lower portion of the central β-sheet A channel of a second molecule. (B) The polymers have a ‘beads on a string appearance’ on electron microscopy. Reproduced from with permission [15,47].

Mutations in neuroserpin form intracellular polymers that cause the dementia familial encephalopathy with neuroserpin inclusion bodies (FENIB)

The process of disease-related polymerization is most strikingly displayed by mutations in the neurone-specific serpin neuroserpin. This protein is expressed during the late stage of development in neurones of the central and peripheral nervous system and in the adult brain [9]. The target protease of neuroserpin is tissue plasminogen activator and thus it is likely to be important in the control of synaptic plasticity, as well as in learning and memory [10]. Mutations in neuroserpin result in the autosomal dominant inclusion body dementia FENIB [11–14]. This is characterized by eosinophilic neuronal inclusions of neuroserpin (Collins’ bodies) in the deeper layers of the cerebral cortex and the substantia nigra. The inclusions are periodic acid-Schiff positive and diastase-resistant and bear a striking resemblance to those of Z α1-antitrypsin that form within the liver. The observation that FENIB was associated with mutations Ser49Pro and His338Arg in the neuroserpin gene homologous to those in α1-antitrypsin causing liver disease (Ser53Phe and His334Asp, respectively) [15,16] strongly indicates a common molecular mechanism. This was confirmed by the finding that the neuronal inclusion bodies of FENIB were formed by entangled polymers of neuroserpin with identical morphology to the polymers of mutant α1-antitrypsin present in hepatocytes from a child with α1-antitrypsin deficiency-related cirrhosis [12]. Moreover, mutants of neuroserpin that cause FENIB have greatly accelerated rates of polymerization compared to the wild-type protein when assessed at the protein level [17, 18] or within cell models of disease [19, 20].

The direct relationship between the magnitude of the intracellular accumulation of neuroserpin and the severity of disease is clearly shown by the identification of other mutations of neuroserpin in families with FENIB [21]. Individuals with Ser49Pro neuroserpin (neuroserpin Syracuse) have diffuse small intraneuronal inclusions of neuroserpin with an onset of dementia between the ages of 45 and 60 years [11–13]. Those with a conformationally more severe mutation (neuroserpin Portland; Ser52Arg) have larger inclusions and an onset of dementia in early adulthood, whereas individuals with His338Arg neuroserpin display even more inclusions with an onset of dementia in adolescence. The most striking examples are in individuals with the most ‘polymerogenic’ mutations of neuroserpin, Gly392Glu and Gly392Arg. These individuals have an even earlier onset of disease, with the Gly392Arg mutation being associated with a profound intellectual decline in an 8-year-old girl, seizures and electrical brain activity in keeping with ‘epilepsy of slow-wave sleep’ [22]. Thus, FENIB can cause a spectrum of disease, from dementia to epilepsy, with variable electrical status.

Other serpins, polymers and disease

The phenomenon of loop-sheet polymerization has been reported in mutants of the plasma proteins C1-inhibitor [23,24], antithrombin [25,26] and α1-antichymotrypsin [27–29]. These mutants are retained in the liver, with the lack of circulating protein resulting in uncontrolled activity of proteolytic cascades and, hence, angio-oedema, thrombosis and chronic obstructive pulmonary disease, respectively [6,30]. A mutation in heparin cofactor II (Glu428Lys) has also been associated with plasma deficiency but, as yet, this has not been shown to cause disease [31]. This mutation is of particular interest because it is the same as the Z allele that causes polymerization and deficiency of α1-antitrypsin. We have shown that this same mutation also causes temperature-dependent polymerization and inactivation of the Drosophila serpin Necrotic [32]. The Necrotic protein plays an important role in the control of innate immunity in the fly.

Pathways to polymers

Biochemical, biophysical and crystallographic studies have been used to dissect the molecular basis by which monomeric Z α1-antitrypsin forms polymers. The Glu342Lys mutation associated with the Z allele is located at the head of strand 5 of β-sheet A and the base of the mobile reactive loop (Fig. 1). This mutation causes a conformational transition and the formation of an unstable intermediate termed M* [33,34] in which β-sheet A opens [33–37] and the upper part of helix F unwinds [29,38,39]. The patent β-sheet A can then accept the loop of another molecule to form a loop-sheet dimer, which extends to form longer chains of loop-sheet polymers [33–37]. There is evidence that the dimer initiates and propagates polymerization of the serpins [40,41].

Although many α1-antitrypsin deficiency variants have been described, only three other mutants of α1-antitrypsin have similarly been associated with profound plasma deficiency and hepatic inclusions: α1-antitrypsin Siiyama (Ser53Phe) [15], Mmalton (ΔPhe52) [42] and King’s (His334Asp) [16]. All of these mutations are within the shutter region and so directly destabilize the closure of a five-stranded β-sheet A (Fig. 1). This predisposes to the accumulation of intermediates that have an expanded β-sheet A, both during and subsequent to folding, and hence favours polymerization [43–45] [16].

Polymers can be induced to form in vitro by heating or incubation with denaturants (urea or guanidine) or at low pH [33,46,47]. This has resulted in the description of linkages between the reactive centre loop and strand 1C [24,25,42,48–51] and between the reactive centre loop and strand 7A [52–54]. A more recent suggestion is that polymers are formed by a larger domain swap involving the exchange of the reactive centre loop and strand 5A. This is based on the crystal structure of a dimer of antithrombin [55]. It was proposed that polymers form when the protein is folding rather than from protein with a near-native structure. Indeed, there is strong evidence that mutants of both α1-antitrypsin and neuroserpin are associated with a significant delay in folding [43–45,56]. This folding defect can be studied by treating the folded protein with urea or guanidine. However, in the case of α1-antitrypsin, the polymers that form as a consequence of treatment with these agents or at low pH are not recognized by a monoclonal antibody that is able to recognize the pathological polymers from the livers of individuals with α1-antitrypsin deficiency [47]. This implies that the long-lived folding intermediate is efficiently degraded by the proteasome via the pathway of endoplasmic reticulum-associated degradation (ERAD) [57]. Moreover, experiments performed in vitro with guanidine or urea produce conformers that are unlikely to be relevant to disease [55].

The quality control pathway is able to fold some of the material to a near native conformation. Some of this will continue to fold to a native conformation, traffic through the Golgi apparatus and then be secreted. However, a proportion will form polymers. Current evidence strongly supports a linkage in which the reactive centre loop of one molecule is inserted into β-sheet A of another (Fig. 1) [17,47]. However concerns about modelling A β-sheet linked polymers will need to be addressed before this model is accepted universally [58]. The pathologically relevant polymers can be recapitulated by heating the folded monomeric protein [47]. This approach should be used to define the structural characteristics of the polymers that form in disease.

Cellular processing of polymers

The accumulation of misfolded proteins within the lumen of the endoplasmic reticulum (ER) activates the RNA-activated protein kinase-like ER kinase, the inositol-requiring kinase 1 and the activating transcription factor 6 limbs of the unfolded protein response (UPR). The UPR serves to attenuate protein translation, increase levels of ER chaperones and enhance the degradation of misfolded ER proteins (a process termed ER-associated degradation or ERAD). Unexpectedly, mutant Z α1-antitrypsin is predominantly degraded by ERAD but does not activate the UPR [59,60]. This striking property results from the ordered structure of serpin polymers.

When ERAD fails to remove sufficient mutant α1-antitrypsin, the remaining protein can form ordered polymers. These polymers are retained within the ER (Fig. 2). There is evidence that ER polymers can be degraded by a process termed autophagy (self-eating) [61,62]. In this process, intracellular membranes envelope organelles and then fuse with lysosomes, allowing turnover of the degraded organelles. It has been suggested that autophagy is actively stimulated by the accumulation of serpin polymers in the ER [61]. However, many studies have failed to compare the degradation of mutant serpin with that of the wild-type control protein. We have shown that, although ERAD is selective for Z α1-antitrypsin, autophagy appears to be a nonselective mechanism for the turnover of all ER proteins [63]. Nevertheless, this does contribute to the defence against cellular accumulation of serpin polymers. Defects in autophagy genes render cells vulnerable to the toxic effects of aggregation-prone proteins, suggesting that liver disease in α1-antitrypsin deficiency may ensue when the autophagic pathway is overwhelmed [62]. The susceptibility of the degradative mechanisms to overload increases with age and may explain the late development of liver disease in some individuals with Z α1-antitrypsin deficiency.

Figure 2.

 Augmenting autophagy to clear intracellular inclusions. Mutant Z α1-antitrypsin is degraded by the proteasome or forms polymers that are retained within the endoplasmic reticulum of hepatocytes. Rapamycin has been used to stimulate the mammalian target of rapamycin (mTOR) pathway and hence the autophagic clearance of polymers. Carbamazepine lowers free inositol within the hepatocyte and thus stimulates the autophagy of Z α1-antitrypsin polymers by a pathway that is independent of mTOR [73]. IP1, inositol 1-phosphate; IP2, inositol 2-phosphate; IP3, inositol 3-phosphate; PIP2, phosphatidylinositol 4,5-bisphosphate. Reproduced with permission [74].

Recent data provide strong support for a signalling pathway that directly links the accumulation of ordered polymerized protein within the ER to the activation of nuclear factor-kappa B (NF-κB) [64]. The accumulation of polymers of mutants of neuroserpin activates NF-κB by a calcium-dependent pathway that is independent of the canonical UPR, causing the dementia FENIB. This ‘ordered protein response’ (also known as the ER overload response) was first described as a cellular response to the accumulation of virally encoded proteins in the ER [65]. It is this pathway that is considered to be activated by the accumulation of ordered polymers of mutant α1-antitrypsin. This UPR-independent NF-κB signal transduction in response to ER-ordered polymers may regulate key pathways that initiate the cellular toxicity of the serpinopathies.

Therapeutic strategies to treat the serpinopathies

Small molecules to block polymerization

The understanding that Z α1-antitrypsin polymerization underlies the liver accumulation and subsequent plasma deficiency has facilitated the development of novel strategies to attenuate polymerization and thereby treat the associated disease. Trials of chemical chaperones have identified agents that stabilize intermediates on the Z α1-antitrypsin folding pathway and are effective in cell and animal models of disease [66], although human trials have been disappointing [67]. Similarly, glycerol, the sugar alcohol erythritol, the disaccharide trehalose and its breakdown product, glucose, reduce the rate of polymerization of wild-type neuroserpin and the Ser49Pro mutant that causes the dementia FENIB [68]. They also attenuate the polymerization of Z α1-antitrypsin. However, these agents have yet to be assessed in cell and animal models of disease. A second approach has utilized the in silico discovery of drug-like small molecules to target a surface cavity for allosteric blockage of the conformational transition that underlies polymer formation [69]. This cavity is patent in the native protein but is filled during the formation of the β-sheet linkages that underlie polymerization. The lead candidate blocks α1-antitrypsin inhibitory function, inhibits the polymerization of Z α1-antitrypsin in vitro and significantly increases the clearance of Z α1-antitrypsin in a cell model of disease [69]. It will be important to identify small molecules that can block polymerization of Z α1-antitrypsin without affecting inhibitory activity.

Accelerating the degradation of polymers

The finding that polymers of Z α1-antitrypsin are cleared by autophagy has been used as a novel strategy to treat the inclusions of Z α1-antitrypsin associated with liver disease [70,71]. Mice have been treated with the pro-autophagic agents rapamycin and carbamazepine [70,71] that stimulate autophagy by pathways dependent and independent of the mammalian target of rapamycin, respectively (Fig. 2). They were effective in clearing soluble and insoluble aggregates of Z α1-antitrypsin in cell lines and in mice that express the Z mutant of α1-antitrypsin [70,71]. It remains to be seen whether the human liver responds to pro-autophagic agents as effectively as that of the mouse. Moreover, upregulating clearance via autophagy may have multiple off-target affects. Nevertheless, these findings provide strong support for carrying out a clinical trial aiming to assess the efficacy of pro-autophagic agents (e.g. carbamezepine) in individuals with liver disease as a result of inclusions formed by Z α1-antitrypsin.

Stem cell technology

Human-induced pluripotent stem cells (hIPSCs) promise great opportunities for the advancement of developmental biology, cell-based therapy and the modelling of human disease. Dermal fibroblasts have been isolated from patients with α1-antitrypsin deficiency and used to generate patient-specific hIPSC lines. Each of the hIPSC lines was differentiated into ‘hepatocyte-like cells’ using a novel and simple three-step differentiation protocol under chemically-defined conditions. The patient-specific, hIPSC-derived hepatocytes show the key pathological features of α1-antitrypsin deficiency: protein misfolding, the formation of pathological polymers and the retention of polymers in the ER [72]. The next stage is to correct the genetic defect that underlies α1-antitrypsin deficiency in induced pluripotent stem cells and then to differentiate these cells to hepatocytes that can be used as replacement therapy in individuals with α1-antitrypsin deficiency.

Concluding remarks

Polymerization of the serpins provides a tractable model for the conformational diseases that are characterized by the aberrant deposition of misfolded proteins. There is still controversy regarding the structure of the pathological polymer, although this is likely to be resolved by further structural studies. The current data suggest a single-strand linkage between the reactive centre loop and β-sheet A. However, the precise details may differ with respect to different mutations and the different members of the serpin superfamily The long-term aim must be to exploit our understanding of the pathobiology of the disease with the aim of developing novel therapeutic strategies. There are likely to be small molecules that stabilize the folding, or accelerate the degradation, of the mutant protein. Stem cell technology provides an exciting alternative approach, although our understanding of the reprogramming of stem cells is still in its infancy.

Acknowledgements

D.A.L. is supported by grants from the Medical Research Council (UK), the Engineering and Physical Sciences Research Council (UK), the British Lung Foundation, GlaxoSmithKline and Papworth NHS Trust. A.O. is the recipient of an eALTA Award and A.J.K. is a Wellcome Trust Student. U.I.E. is an MRC Clinical Research Training Fellow and S.T.R. is a Wellcome Trust Clinical Training Fellow. D.C.C. is an Alzheimer’s Research Trust Senior Research Fellow and is supported by the Medical Research Council (UK) and the Engineering and Physical Sciences Research Council (UK). S.J.M. is a Medical Research Council Clinical Scientist (G0601840) and is supported by Diabetes UK and the British Lung Foundation.

Ancillary