D. Belorgey, Cambridge Institute for Medical Research, Wellcome Trust/MRC Building, Hills Road, Cambridge, CB2 2XY, UK. Fax: +44 1223 336827, Tel.: +44 1223 336825, E-mail: email@example.com
The dementia familial encephalopathy with neuroserpin inclusion bodies (FENIB) is caused by point mutations in the neuroserpin gene. We have shown a correlation between the predicted effect of the mutation and the number of intracerebral inclusions, and an inverse relationship with the age of onset of disease. Our previous work has shown that the intraneuronal inclusions in FENIB result from the sequential interaction between the reactive centre loop of one neuroserpin molecule with β-sheet A of the next. We show here that neuroserpin Portland (Ser52Arg), which causes a severe form of FENIB, also forms loop-sheet polymers but at a faster rate, in keeping with the more severe clinical phenotype. The Portland mutant has a normal unfolding transition in urea and a normal melting temperature but is inactive as a proteinase inhibitor. This results in part from the reactive loop being in a less accessible conformation to bind to the target enzyme, tissue plasminogen activator. These results, with those of the CD analysis, are in keeping with the reactive centre loop of neuroserpin Portland being partially inserted into β-sheet A to adopt a conformation similar to an intermediate on the polymerization pathway. Our data provide an explanation for the number of inclusions and the severity of dementia in FENIB associated with neuroserpin Portland. Moreover the inactivity of the mutant may result in uncontrolled activity of tissue plasminogen activator, and so explain the epileptic seizures seen in individuals with more severe forms of the disease.
familial encephalopathy with neuroserpin inclusion bodies
tissue plasminogen activator
The autosomal dominant dementia familial encephalopathy with neuroserpin inclusion bodies (FENIB) results from point mutations in the neuroserpin gene and is characterized by inclusions of neuroserpin within cortical and subcortical neurons [1–3]. Neuroserpin is a member of the serine proteinase inhibitor or serpin superfamily. It inhibits the enzyme tissue plasminogen activator (tPA) and may be important in regulating neuronal plasticity and memory [4–7]. We have recently expressed, purified, and characterized wild-type neuroserpin and neuroserpin with the Ser49Pro mutation, which was identified in the first reported family with FENIB . The mutation reduced the inhibitory activity of neuroserpin by ≈ 100-fold and increased the formation of polymeric protein under physiological conditions. Neuroserpin polymers result from the sequential insertion of the reactive centre loop of one molecule into β-sheet A of another [1,6]. The resulting species is inactive as a proteinase inhibitor and accumulates in the endoplasmic reticulum in cell models of disease  and in vivo.
Three other mutants of neuroserpin are now recognized to cause FENIB: Ser52Arg, His338Arg and Gly392Glu . This condition is unusual among neurodegenerative disorders in that there is a striking correlation between the number of inclusions within the cerebral cortex and an inverse relationship with the age of onset of disease . For example, individuals with the Ser52Arg and Gly392Glu neuroserpin mutation have 3 and 9.5 times more inclusions within the cerebral cortex than individuals with the Ser49Pro mutant. This corresponds to an age of onset of symptoms in individuals with Ser49Pro, Ser52Arg and Gly392Glu neuroserpin of 48, 24, and 13 years, respectively. There is also a change in phenotype, with the Ser49Pro mutation causing predominantly dementia whereas the Ser52Arg, His338Arg and Gly392Glu mutants cause both dementia and severe progressive epilepsy. In addition to the striking genotype–phenotype correlation, FENIB is also unusual in that the mutant neuroserpin forms ordered polymers within the endoplasmic reticulum [1,8]. This contrasts with other conditions such as Parkinson's and Huntington's disease in which the mutant proteins form disordered aggregates within the cytoplasm . We have expressed and characterized the Ser52Arg variant of neuroserpin (neuroserpin Portland) to determine if the rate of polymer formation can explain the correlation between the mutation, the number of intraneuronal inclusions, and the clinical phenotype. Our data show that the Ser52Arg mutation favours the rapid formation of polymers as the protein is locked as an inactive folding intermediate. These polymers explain the increased number of inclusions in individuals with Ser52Arg compared with those with the Ser49Pro mutation. Moreover the inactivity of the mutant may result in uncontrolled activity of tPA, and so explain the epileptic seizures seen in individuals with more severe forms of the disease.
Materials and Methods
Oligonucleotides were synthesized by MWG-Biotech AG (Ebersberg, Germany). The expression vectors pQE81L and Ni-nitrilotriacetate agarose were from Qiagen (Crawley, Sussex, UK), HiTrap Q Sepharose was from Amersham Biosciences (Chalfont St Giles, Bucks., UK), and the tPA substrate S-2288 (H-d-Ile-Pro-Arg-p-nitroanilide) was from Chromogenix (Quadratech, Epsom, Surrey, UK). 1,5-Dansyl-Glu-Gly-Arg-chloromethylketone and tissue plasminogen activator (tPA) were from Calbiochem (Merck Biosciences, Nottingham, UK). Mineral oil was from either Sigma Chemical Co. (catalogue number M-3516, M-8410, M-5904, M-1180, M-5310) or from Fluka (Buchs, Switzerland; catalogue number 69808).
Expression and purification of recombinant proteins
The Ser52Arg mutation was introduced into the cDNA of human neuroserpin in the pQE81L expression vector  by a two-step PCR. The gene was fully sequenced to ensure that there were no PCR errors. Recombinant wild-type, Ser49Pro and Ser52Arg neuroserpin were expressed with a six-histidine tag at the N-terminus and purified as described previously , except that the HiTrap chelating column was replaced by Ni-nitrilotriacetate agarose. The resulting proteins were assessed by SDS, nondenaturing and transverse urea gradient PAGE, and activity was assessed against tPA .
Complex formation assays
Wild-type and Ser52Arg neuroserpin were incubated in various ratios with tPA at 25 °C as described previously . Samples were taken at different time intervals, and the reaction was stopped by the addition of 1 mm 1,5-dansyl-Glu-Gly-Arg-chloromethylketone (final concentration) to inhibit any free tPA . The samples were then mixed with SDS/PAGE loading buffer, snap-frozen in liquid nitrogen, and stored until the completion of the experiment. They were then thawed and boiled for 3 min. Proteins were separated by SDS/PAGE [10% (w/v) gel] and visualized by staining with Coomassie Blue.
Determination of the reaction parameters describing tPA inhibition
Inhibition rate constants for the inhibition of tPA by wild-type or mutant neuroserpin were determined under pseudo-first-order conditions, i.e. [I] = 10[E]0, using the progress-curve method [11,12]. Rate constants of inhibition were measured at 25 °C in inhibition buffer [50 mm Hepes, 150 mm NaCl, 0.01% (w/v) dodecyl maltoside, pH 7.4] by adding tPA (20 nm) to a mixture of wild-type (from 200 nm to 1000 nm) or Ser52Arg (6600 nm) neuroserpin and the substrate S-2288 (1 mm) and recording the release of product as a function of time. The progress curves were analyzed as described previously [12,13].
CD experiments were performed using a Jasco J-810 spectropolarimeter in 100 mm sodium phosphate buffer, pH 7.4. Polymers were formed by heating wild-type or mutant neuroserpin at 0.5 mg·mL−1 and 45 °C for 24 h. Changes in the secondary structure of wild-type or Ser52Arg neuroserpin with time and temperature were measured by monitoring the CD signal at 216 nm for 24 h with protein at a concentration of 0.5 mg·mL−1. When possible, the data were fitted to a single exponential function. Thermal unfolding experiments were performed by monitoring the CD signal at 216 nm in a 150-µL cuvette between 25 °C and 95 °C using a heating rate of 1 °C·min−1 at a concentration of 0.7 mg·mL−1. The second derivative of the resulting data was used to calculate the inflection point of the transition and hence the Tm.
Assessment of the polymerization of wild-type and Ser52Arg neuroserpin
Polymerization of wild-type, Ser49Pro and Ser52Arg neuroserpin was assessed by incubating the protein at concentrations of 0.1 or 0.4 mg·mL−1 in NaCl/Pi, pH 7.4, at 37 °C or 45 °C. Aliquots were taken over time, and 2 µg protein was loaded on a 7.5% (w/v) nondenaturing gel. To avoid evaporation during the experiment, the different samples were covered with mineral oil. The proteins were visualized by staining with GelCode® Blue Stain Reagent (Pierce, Tattenhall, Cheshire, UK) or by silver staining.
Unfolding of wild-type and Ser52Arg neuroserpin in urea
Neuroserpin at 25 µg·mL−1 was incubated at 20 °C with various concentrations of urea (from 0 to 9 m) in 50 mm sodium phosphate buffer, pH 7.4, and unfolding was monitored by measuring the intrinsic tryptophan fluorescence by excitation at 295 nm. The fluorescence spectra were measured with a PerkinElmer LS50B fluorimeter with both the excitation and emission slit widths set to 10 nm. The spectrum data were obtained as the average of five traces, and the wavelength at the emission maximum was determined by PerkinElmer FL WinLab software. The unfolding of wild-type neuroserpin was also monitored by CD ellipticity at 222 nm with a Jasco J-810 spectropolarimeter. The path length and slit width were 1.0 cm and 2 nm, respectively. The fluorescence and CD measurements were performed at the incubation times of 3, 6, 12, and 24 h to confirm equilibrium in urea, with no difference being observed between the 12 h and 24 h data. The transition midpoint of unfolding was determined by fitting of the triplicate experimental data to a theoretical sigmoidal equation at a urea concentration of 2–9 m.
Results and Discussion
The expression of Ser52Arg neuroserpin resulted in a poor yield, with only 0.1–0.5 mg pure monomeric protein being obtained from 3 L culture medium. This compares with an average of 5 and 1 mg for wild-type and Ser49Pro neuroserpin, respectively, when expressed under the same conditions. Wild-type, Ser49Pro and Ser52Arg neuroserpin migrated as single bands on SDS, nondenaturing, 8 m urea and isoelectrofocusing PAGE.
Ser52Arg neuroserpin is inactive as an inhibitor of tPA
Wild-type neuroserpin forms complexes with tPA with a stoichiometry of inhibition of 1 and an association rate constant (kass) of 1.2 × 104m−1·s−1. At higher ratios of enzyme to inhibitor (i.e. [tPA] ≫ [neuroserpin]), there was cleavage of the reactive centre loop  and loss of the 4-kDa C-terminal fragment. In contrast, it was not possible to determine a rate of inhibition of tPA by Ser52Arg neuroserpin. Indeed there was no inhibition of tPA even at concentrations as high as 6.6 µm Ser52Arg neuroserpin (Fig. 1). To determine if the formation of a complex was possible between Ser52Arg and tPA, higher concentrations (in the micromolar range) of both species were used to favour complex formation. On SDS/PAGE, there is only a transient band corresponding to the complex between Ser52Arg neuroserpin and tPA (Fig. 2A).
Incubation of Ser52Arg neuroserpin with an excess of tPA resulted in the formation of a transient complex, which represented only a small fraction of the total amount of Ser52Arg neuroserpin, consistent with the lack of inhibition observed previously (Fig. 2B). The addition of tPA to wild-type neuroserpin resulted in complete cleavage of the inhibitor after a 1-h incubation at an enzyme to inhibitor ratio of 1 : 1 (Fig. 2B). The reactive loop of Ser52Arg neuroserpin was more resistant to cleavage, as there was always a significant proportion of Ser52Arg neuroserpin that remained uncleaved even after incubation for 1 h at a tPA to Ser52Arg neuroserpin ratio of 10 : 1. These data show that the reactive centre loop is not as readily accessible in Ser52Arg neuroserpin as it is in the wild-type protein.
Ser52Arg neuroserpin forms polymers more rapidly than wild-type or Ser49Pro neuroserpin
Polymerization was assessed by incubating wild-type, Ser49Pro and Ser52Arg neuroserpin at 37 °C or 45 °C and separating the resulting mixture by nondenaturing PAGE. Ser52Arg neuroserpin readily formed polymers at 0.4 mg·mL−1 and 37 °C, which were apparent as a reduction in the intensity of the monomeric band after 6 h of incubation (Fig. 3). After 52 h, this mutant had formed higher-order aggregates which were stacked at the top of the gel. The rate of polymerization was determined by measuring the reduction in density of the monomeric band. Wild-type neuroserpin had not formed polymers at a measurable rate after 24 days at 0.4 mg·mL−1 and 37 °C compared with a rate of 5.3 × 10−6 s−1 for Ser49Pro neuroserpin and 7.9 × 10−5 s−1 for Ser52Arg neuroserpin (Table 1 and Fig. 3). The same effect was apparent if the polymerization experiments were conducted at 45 °C. Both mutants formed polymers more rapidly than wild-type neuroserpin but there was no difference between the rates of the two mutants (Fig. 4 and Table 1). The rates of polymerization for wild-type and Ser49Pro neuroserpin are slower than those that we reported previously . The difference was due to the mineral oil used to overlay the protein solution. We had previously used mineral oil (Sigma Chemical Co.; M-3516) that was more than 3 years old. Repeating the experiment with newer batches of oil confirmed the difference between wild-type and Ser49Pro but the rates were 10-fold slower.
Table 1. Rate of polymerization of neuroserpin at 0.4 mg·mL−1as measured by densitometry from nondenaturing PAGE. The results are the mean of at least three experiments.
5.3 (± 0.3) × 10−6
7.9 (± 0.4) × 10−5
3.3 (± 0.9) × 10−5
2.7 (± 0.8) × 10−4
2.2 (± 0.2) × 10−4
It was not possible to follow the change in secondary structure of Ser52Arg neuroserpin during polymerization with CD because the signal at 216 nm did not change during the course of the experiment, i.e. after incubation of Ser52Arg neuroserpin at either 37 °C or 45 °C for 24 h.
Assessment of the conformation of Ser52Arg neuroserpin
The most likely cause for the inactivity of Ser52Arg neuroserpin and inaccessibility of the reactive loop is that the mutant had adopted an aberrant conformation. One possibility is that the reactive loop had fully inserted into its own β-sheet A to form a latent conformer . However, this is unlikely as the latent conformer of the serpins is unable to form polymers [17,18]. Other characteristics of the latent conformer are a failure to unfold in denaturants and enhanced thermal stability [17,18]. The conformation adopted by Ser52Arg neuroserpin was therefore assessed by electrophoresis on transverse urea gradient gels. Ser52Arg neuroserpin unfolded with a profile that was similar to wild-type neuroserpin (Fig. 5A), indicating that there was no gross distortion of structure. The melting point temperature was determined by monitoring the change in CD signal at 216 nm while increasing the temperature at 1 °C·min−1 (Fig. 5B). In the case of Ser52Arg neuroserpin, the signal magnitude only allows us to calculate an approximation of the melting temperature (Tm). This gave a Tm of ≈ 55 °C. This Tm was surprising as it is close to the value for wild-type neuroserpin (56.6 °C) and significantly higher than that for Ser49Pro neuroserpin (49.9 °C) . Previous studies have shown an inverse relationship between rate of polymer formation and Tm, and thus it was unusual to find that the Tm was higher than that of the less severe Ser49Pro neuroserpin. The overall structure of Ser52Arg neuroserpin was therefore assessed by CD spectroscopy. There were marked differences in the profiles of native wild-type and Ser52Arg neuroserpin (Fig. 5C). The profile for wild-type neuroserpin was comparable to that obtained for other serpins, including α1-antitrypsin and α1-antichymotrypsin [19,20]. In comparison, the spectrum of Ser52Arg neuroserpin shows an increase in both β-sheet and α-helical structure content as determined by the large increase in magnitude of the signal at 216 nm and the small increase at 222 nm. This profile is comparable to that obtained for monomeric Ser49Pro neuroserpin and the polymers of both wild-type and Ser49Pro neuroserpin . Spectra taken after incubation of Ser52Arg neuroserpin for 24 h at 0.5 mg·mL−1 and 45 °C (i.e. after the protein was 100% polymers on nondenaturing PAGE) showed a profile that was similar to monomeric Ser52Arg neuroserpin and polymers of wild-type and Ser49Pro neuroserpin.
More detailed unfolding experiments were then performed to further assess the conformation of wild-type and mutant neuroserpin. The proteins were added to increasing concentrations of urea, and the change in fluorescence profile was followed by exciting the protein at 295 nm and measuring the shift in maximum fluorescence after a 12 or 24 h incubation time (Fig. 6A). No differences were observed between the two incubation times. The profiles obtained for wild-type neuroserpin, Ser49Pro and Ser52Arg were consistent with the results obtained from both transverse urea gradient gels and assessment of the melting temperature (Fig. 6B). The transition points calculated for wild-type neuroserpin and Ser52Arg were very similar, at 6.4 and 6.3 m urea, respectively. The calculated transition point for Ser49Pro was lower, at 5.3 m urea. The CD ellipticity of wild-type neuroserpin was also assessed at 222 nm; the data were identical with those obtained from urea unfolding (Fig. 6). The transition midpoint calculated from the CD data for wild-type neuroserpin was also 6.4 m.
Correlation of biochemical characteristics with the dementia and epilepsy found in individuals with the neuroserpin Portland (Ser52Arg) mutation
The neuroserpin Portland (Ser52Arg) mutation is associated with three times the number of intracellular inclusion bodies (or Collin's bodies) in neurons compared with dementia associated with the Syracuse mutation (Ser49Pro). The clinical manifestations are more severe, with an age of onset of disease ≈ 20 years earlier . This earlier age of onset is in keeping with the faster rate of polymerization of Ser52Arg neuroserpin compared with Ser49Pro neuroserpin. It was surprising that Ser52Arg neuroserpin was almost inert when incubated with tPA. There was only transient complex formation, and 50% of Ser52Arg neuroserpin remained uncleaved even after incubation with a 10-fold excess of tPA for 1 h. Moreover the melting temperature and unfolding in urea were similar to that of wild-type neuroserpin, which is unusual for a serpin that spontaneously forms polymers in vitro and in vivo.
The polymers of mutant neuroserpin that form in FENIB are analogous to polymers that form with mutants of other members of the serpin superfamily such as α1-antitrypsin , antithrombin , C1 inhibitor [23,24] and α1-antichymotrypsin  in association with cirrhosis, thrombosis, angio-oedema and emphysema, respectively. Indeed we have recently grouped these conditions together as the serpinopathies as they have a common underlying mechanism [26,27]. A mutation in the shutter region of α1-antichymotrypsin (Leu55Pro) resulted in a similar conformer to Ser52Arg neuroserpin in that it was inactive as a proteinase inhibitor, had enhanced thermal stability but still rapidly formed polymers . We were able to solve the crystal structure of this δ conformer of α1-antichymotrypsin and showed that the reactive loop was partially inserted into β-sheet A . The F-helix was unfolded and inserted into the lower part of β-sheet A, which explains the increased stability. However, this helix must be readily displaced by the reactive loop of another molecule to form the chains of polymers.
This conformation would also explain the data obtained with Ser52Arg neuroserpin. The Ser52Arg mutation is in the shutter domain of the molecule which controls opening of β-sheet A . The arginine mutation would cause a significant disruption in this area, thereby forcing β-sheet A into an ‘open’ or acceptor configuration. This in turn would allow partial insertion of the reactive loop into β-sheet A, with the lower part of β-sheet A being filled by unfolding and insertion of the F-helix (Fig. 7). The reactive loop must be inserted further than in Ser49Pro neuroserpin (which is also a shutter domain mutation) because Ser49Pro neuroserpin remains partly active as a proteinase inhibitor . In keeping with this, the CD profile of Ser52Arg neuroserpin is similar to that of the polymeric conformation in which β-sheet A is filled with the reactive loop of another neuroserpin molecule. Moreover the emission maxima of the native wild-type and mutant proteins (Fig. 6) are also different, in keeping with a different conformation induced by the shutter domain mutants. The F-helix must be readily displaced from the lower portion of β-sheet A during polymerization of Ser52Arg neuroserpin. This would allow acceptance of the reactive loop of a second neuroserpin molecule and the formation of a dimer. Extension of this process forms the characteristic loop–β-sheet A polymers.
Epilepsy is far more common with Ser52Arg neuroserpin than Ser49Pro neuroserpin. This may be explained by the increased number of inclusions. However, it may also be explained by the lack of inhibitory activity caused by the Ser52Arg mutation. There is growing evidence from animal models that epilepsy results from an imbalance between tPA and neuroserpin . The inactivity of Ser52Arg neuroserpin will contribute to this imbalance in individuals who carry this mutation and may exacerbate the intrinsic ability of the intracerebral inclusions to cause epilepsy.
We are grateful to Tim Dafforn for help in preparing Fig. 7. We are also grateful to Kerstin Nordling and Ingemar Björk for helpful comments. This work was supported by the Medical Research Council (UK), the Wellcome Trust (UK) and Papworth NHS Trust (UK). D.C.C. is a Wellcome Trust Intermediate Clinical Fellow.