Characterisation of a type II functionally-deficient variant of alpha-1-antitrypsin discovered in the general population

Lung disease in alpha-1-antitrypsin deficiency (AATD) results from dysregulated proteolytic activity, mainly by neutrophil elastase (HNE), in the lung parenchyma. This is the result of a substantial reduction of circulating alpha-1-antitrypsin (AAT) and the presence in the plasma of inactive polymers of AAT. Moreover, some AAT mutants have reduced intrinsic activity toward HNE, as demonstrated for the common Z mutant, as well as for other rarer variants. Here we report the identification and characterisation of the novel AAT reactive centre loop variant Gly349Arg (p.G373R) present in the ExAC database. This AAT variant is secreted at normal levels in cellular models of AATD but shows a severe reduction in anti-HNE activity. Biochemical and molecular dynamics studies suggest it exhibits unfavourable RCL presentation to cognate proteases and compromised insertion of the RCL into β-sheet A. Identification of a fully dysfunctional AAT mutant that does not show a secretory defect underlines the importance of accurate genotyping of patients with pulmonary AATD manifestations regardless of the presence of normal levels of AAT in the circulation. This subtype of disease is reminiscent of dysfunctional phenotypes in anti-thrombin and C1-inibitor deficiencies so, accordingly, we classify this variant as the first pure functionally-deficient (type II) AATD mutant.


Background
Severe alpha-1-antitrypsin deficiency (AATD, MIM #613490) affects approximately 1 in 2000 of the Northern European population. It is associated with pathogenic variants of the SER-PINA1 gene (MIM #107400) which encodes alpha-1-antitrypsin (AAT). AAT is the archetypal member of the serpin superfamily of serine-protease inhibitors [1]; its primary physiological role is to protect the lung parenchyma from attack by the serine proteases neutrophil elastase (HNE), cathepsin G [2] and proteinase 3 [3]. Mutations that reduce AAT plasma levels alter the balance between inhibitory and proteolytic activity leading to early onset emphysema and COPD [4]. A subset of SERPINA1 pathogenic alleles, well-represented by the common severe deficiency Z allele (E342K, p.E366K) can lead to accumulation of the protein as ordered a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 residues 344 and 362 in the canonical nomenclature). We identified a Gly!Arg mutation at the P10 (349) position that violates the pattern noted for residues in the RCL hinge region [28], and therefore putatively could interfere with inhibitory activity (Fig 2A). In addition, Gly349 is relatively conserved among SERPINA1 orthologues, while within the human serpins the 349 position shows a preference for small aliphatic residues (Fig 2B and S1 File). There was sufficient information in the ExAC database to establish that this variant was present in three separate carriers of 60,706 individuals, and they were not first-or second-degree relatives. A female and a male were present in the Non-Finnish ExAC population, and a male in the Finnish population. We received age data for two of the three individuals: one 40-45 years old, and another 65-70 years old.

Characterization of the G349R alpha-1-antitrypsin variant in cell models
Several cellular models have been utilized successfully to study accumulation-prone variants of AAT and other serpins, and have been shown to represent excellent predictors of in vivo behaviour [7,[29][30][31][32][33][34][35]. To characterize the efficiency of secretion of AAT G349R, we transiently expressed it in HEK293T and Hepa 1-6 cell models in comparison with the wild-type M AAT and the common deficient and polymerogenic Z mutant of AAT (Fig 3).
AAT levels in the cell media from HEK293T and Hepa 1-6 transfected cells were determined by ELISA (Fig 3A) showing that in both cell lines, AAT G349R appears to be secreted at similar levels to wild-type M AAT (85.6% ± 12.5 MEAN±SD for HEK293T cells, 87.3 ± 6.6 for Hepa 1-6 cells). Analysis of extracellular AAT by non-denaturing PAGE also shows that AAT G349R is exclusively monomeric in the cell media (Fig 3B), while the Z AAT polymerogenic and deficient variant is present in the media primarily as high molecular weight oligomers.
In addition, we investigated the intracellular distribution of AAT between the NP40-soluble and -insoluble fractions. As previously reported [33], the tendency of an AAT variant to accumulate in the NP-40 insoluble fraction is a symptom of severe polymerogenic tendency. In Fig  3C the only AAT variant identified within the insoluble fraction is Z AAT.
In conclusion, the G349R variant does not exhibit a "classical" deficiency phenotype.

The G349R alpha-1-antitrypsin variant is non-functional
A survey of the literature identified a protein engineering study in which the P10 alanine residue of α 1 -antichymotrypsin was arbitrarily mutated to arginine to evaluate its ability to become incorporated into β-sheet A upon cleavage. It was found that due to compensatory molecular rearrangements of the α 1 -antichymotrypsin molecule, there was an increase in substrate behaviour rather than an elimination of inhibitory activity [36]. Correspondingly, we undertook experiments to assess the functional capability of G349R AAT, with a view to ascertaining the clinical impact of this variant. One of the characteristics of the serpin-enzyme complex is that it is stable in the presence of SDS [13,27]. We incubated the media of HEK293T cells expressing M or G349R with an equimolar or 2:1 excess over HNE ( Fig 4A) and analysed complexes by SDS-PAGE and The Met358 (P1) and Ser359 (P1') residues, critical for the anti-protease activity of AAT, are shown as black sticks on the native AAT structure (PDB: 1QLP), while the residue Gly349 (P10) is indicated by a blue stick. β-sheet A is blue, β-sheet B is green and β-sheet C is yellow; the RCL is coloured in red. The figure was prepared with PyMOL. (B) Conservation of residues in the RCL (top sequence, from residue G344 to P362) is represented using WebLogo [56], calculated from a sequence alignment of the SERPINA1 orthologues or from human serpins paralogues. immunoblot. Instead of forming the 68 kDa SDS-resistant complex as the wild-type M ( Fig  3A, top arrow), the G349R variant was preferentially cleaved by the HNE (Fig 4A, bottom  arrow).
To further confirm this dysfunctional behaviour, we measured the activity against porcine pancreatic elastase (PPE), a tool serine protease sometimes used as a surrogate for HNE. After incubation with an increasing ratio of AAT variants in the media of HEK293T cells, compared to the inhibitory activity of the wild-type M (dashed line), the new variant (dotted line) showed no signs of inhibitory activity ( Fig 4B).
The variant was also expressed in bacteria and purified to homogeneity. It was found that this material required an 8-fold greater concentration to fully inhibit HNE than M AAT, representing an over 80% decrease in activity ( Fig 4C, upper panel), and was entirely inactive against the tool protease chymotrypsin (CHT) (Fig 4C, lower panel). The exclusive presence of the cleaved form of G349R AAT when tested with HNE ( Fig 4A) shows that the loop is still recognised by the protease, but that the branched inhibitory pathway is skewed in favour of nonproductive turnover of AAT and premature release of active protease.
AAT Pittsburgh represents a variant in which an amino acid substitution in the specificitydetermining site of the RCL introduces a novel cleavage site for other human proteases, including thrombin [25]. While the G349R mutation does not prevent an interaction with HNE, it may similarly lead to off-target recognition by other proteases. By querying the MER-OPS database [37] for human proteases that could potentially cut the sequence introduced by the G349R substitution (EAARAMFL), we identified plasma kallikrein (KLK), which recognises the P1-P1' sequence R-X, as a candidate. Using the recombinant protein, it was found that indeed this variant is susceptible to cleavage by KLK when incubated at equimolar ratio ( Fig 4D).
In conclusion, AAT G349R is a novel dysfunctional variant lacking effective inhibitory activity. Many AAT variants are named according to the birthplace of the proband or the site of the diagnosing centre; as this information is unknown we decided to name the novel variant AAT Iners, from the latin word meaning "inactive".

Comparative structural analysis and molecular dynamics simulations suggest an impaired presentation and insertion of the RCL domain
The experimental data suggest two possible structural causes of dysfunction: the first is that the conformation of the RCL is altered, which can affect protease recognition and docking; the second is that the bulky charged group that replaces Gly349 impedes a timely conformational change of the RCL after cleavage.
An Ala!Arg mutation has been introduced at the equivalent site in a protein engineering study of a related serpin, α 1 -antichymotrypsin [36]. A comparison between the crystal structures of the RCL-cleaved forms of Ala349Arg alpha-1-antichymotrypsin (PDB ID: 1AS4) and the wild-type protein (PDB ID: 2ACH) showed that the introduced arginine side-chain can be accommodated into β-sheet A, but only by rearrangement of the local hydrophobic packing AAT levels in cell media from transfected HEK293T or Hepa 1-6 cells were quantified by sandwich ELISA and represented as percentages of wild-type M levels (mean ± SD, n = 3; one-way ANOVA, p < 0.0001; two-tailed unpaired t-test between each variant and M AAT, n.s non-statistically significant, �� p < 0.001, ��� p < 0.0001). (B) Immunoblots with anti-total AAT pAb loaded with equal volume of cell media from HEK293T (left) or Hepa 1-6 (right) cells expressing the indicated variants, resolved by 7.5% w/v acrylamide SDS-PAGE (top) and 8% Native-PAGE (bottom). (C) Immunoblots with anti-total AAT pAb of NP40-soluble (SOL) or insoluble (INS) cellular fractions from HEK293T (left) or Hepa 1-6 (right) cells expressing the indicated variants, resolved by 7.5% SDS-PAGE. https://doi.org/10.1371/journal.pone.0206955.g003 interactions and incorporation of a counter-ion [36]. This is sufficient to greatly reduce inhibitory efficiency, as measured by the non-productive turnover of inhibitor [36]. A comparison with the equivalent residues in cleaved AAT (PDB ID: 1EZX) shows that in order to accommodate the G349R mutation, similar shifts would be required in Phe384 and Ala336, and one residue, Phe51, provides a more marked occlusion in AAT with respect to Ile51 in alpha-1-antichymotrypsin ( Fig 5A).
As well as interference with β-sheet incorporation during inhibition, a glycine-to-arginine mutation would be expected to result in angular restriction of the local protein backbone in the RCL-exposed conformation. To test this possibility, molecular dynamics (MD) simulations were performed for the wild-type protein (PDB ID: 1QLP) and AAT Iners in the uncleaved form. In each case, four 70-ns simulations were performed, and the last 50 ns of each trajectory were combined to form an aggregated 200-ns trajectory for subsequent analysis. Careful visual analysis of the simulations showed that the orientation of the side chain of Met358 in the AAT Iners variant has a different behaviour with respect to the wild-type (S1 Video). To quantify this phenomenon, an observable was defined as follows: a vector v 1 was defined, joining atom Cα of residue Phe370 to atom Cα of residue Met358, and a vector v 2 joining atom Cα to atom S of residue Met358. The angle φ between vectors v 1 and v 2 was then calculated at each time during the simulations. The resulting probability density distribution, in the range from 0˚to 180˚, is shown in Fig 5B. Values of φ near 0˚correspond to the side chain of Met358 pointing outwards with respect to the bulk of the protein, while values near 180˚correspond to the side chain pointing inwards. The probability of finding an inward side chain (90˚< φ < 180˚) was 2% for the wild type and 11% for AAT Iners. To further study the atypical behaviour of the RCL region with the G349R substitution, we performed simulations of the protein immediately following cleavage, and calculated the average number of hydrogen bonds between Arg349 and selected residues (Fig 5C). Arg349 acquired novel interactions with the proximal residues Ala347 and Ala348, but also established additional bonds with the distal amino acids Asp107 and Glu195. Thus, AAT Iners is predicted from the simulations to have an altered RCL presentation. Coupled with the observation that compensatory movements are required during insertion into β-sheet A, possibly including negation of the buried charge, this substitution leads to malfunction of early steps in the inhibitory mechanism.

Discussion
The reactive centre loop is an exposed loop that exhibits relatively few interactions with the serpin body and plays a lesser role in stability and folding of the native molecule than other functional components of the serpin structure. As the primary determinant of serpin specificity, this element is by definition permissive of sequence variability, with the exception of marked deviations in length [38].
However, inhibitory serpins exist as two conformations and must present a sequence that is recognised by the target protease, and these facts impose evolutionary constraints on some positions in the loop [1,28]. Thus, mutations at certain positions can impair the interaction with a target protease, permit non-productive degradation by other proteases, or perturb the inhibitory mechanism by interfering with the accommodation of the RCL by β-sheet A.
An increasing number of gene variants are being discovered by exome or genome-wide sequencing of large cohorts and made available in annotated databases such as ExAC or LOVD [26,27,39]. In the present study, we have shown that one such variant, G349R, is secreted normally in the native form by cells, but would provide no protection against proteolytic activity due to its near-inability to form a stable inhibitory complex with target proteinases. Notably, this variant was predicted by the pathogenicity predictor REVEL to be benign (score 0.448 with pathogenicity threshold at 0.477 [27]), suggesting that mutations affecting the RCL and the activity may be misinterpreted by commonly-used predictors.
Upon cleavage of the P1-P1' bond by a protease, the RCL of AAT begins its insertion into β-sheet A via a zipper-like motion involving residues 342 to 350 [13]. To achieve the full insertion of the RCL into β-sheet A, the AAT molecule must rearrange helix-F (hF) [14,15], a step predicted by accelerated molecular dynamics to critically involve Ala348 and Gly349 [14].
Ultimately, the RCL is accommodated into the β-sheet and the protease is trapped. Interference with the timely progression of this change is deleterious to inhibitory activity. One member of the serpin superfamily unable to spontaneously undergo the loop-to-sheet conformational change is ovalbumin [40], which has an arginine residue at the P14 position of the RCL. When incorporated into inhibitory serpins, this residue abrogates activity [41]. Conversely, replacement at this site by a serine or threonine in ovalbumin greatly improves the insertion rate of the RCL upon cleavage by a protease [40,41]. Mutations at other positions violating the hinge region motif at 'P-even' residues in AAT have similarly led to compromised inhibitory activity [28,42,43], and studies using its closest homologue, α 1 -antichymotrypsin [1], have shown loss of activity from introduction of arginines at P14, P12, and P10 positions [36]. In all three cases, structure determination by protein crystallography revealed that the RCL could be accommodated, in the case of P14 by introducing a twist in the backbone orienting the side-chain towards the solvent, and in the latter two by compensatory changes in the hydrophobic residues underlying β-sheet A. The structural evidence that such unfavourable substitutions can be accommodated highlights that the partition between a productive and non-productive interaction is contingent on the balance between the kinetics of insertion and hydrolysis [19].
The Pittsburgh AAT variant represents a case in which substitution within the specificitydetermining residues results in the ability to interact with a novel target protease. Changes outside of this region can also lead to aberrant interactions but are much more likely to lead to non-productive cleavage. Our data show that AAT Iners can act as a novel substrate of KLK, which liberates bradykinin (BK), a vasodilator hormone, from the high molecular weight kininogen (HMWK) [44] and activates plasminogen [45]. Given the expected high circulating AAT concentrations for a heterozygote in vivo (around 10 to 25μM) this could conceivably lead it to act as a competitive substrate for this enzyme.
Two other genetic disorders have been reported that are associated with dysfunctional plasma inhibitory serpins produced at normal levels. Hereditary angioedema type I and II (HAE, #MIM106100), an autosomal dominant disorder, is caused by mutations in C1 inhibitor (C1INH, SERPING1). In type I HAE, found in 85% of patients, plasma levels of C1INH are less than 35% of normal, leading to a loss of function of the C1INH [46][47][48]. In HAE type II, the C1INH serum levels are normal or elevated, but the protein is non-functional due to a mutation within the RCL domain that also causes the inefficient inhibition of the target protease [48]. Antithrombin deficiency (AT3D, #MIM613118), which is the result of variants of the inhibitory serpin antithrombin III (AT3, SERPINC1), leads to venous thromboembolic disease. As for the HAE, two categories of AT3D have been defined based on AT3 plasma concentrations and inhibitory activity [49]. Most AT3D manifestation belong to the type I deficiency group with a severe plasma deficiency of AT3; in type II (functional) deficiency, these subjects possess normal AT3 serum levels, but mutations in functional domains of this anticoagulantincluding the RCL, the heparin-binding site (HBS), or the A-and C-sheet domain-impair or abolish inhibitory activity [50][51][52][53].
We now report AAT Iners as the first-described, pure type II AATD variant. We have shown that this variant is secreted at wild-type levels and would therefore not be identified in a patient by conventionally used diagnostic protocols. However, as it is non-functional, a carrier is likely to present the same susceptibility to lung disease as individuals with a recognised deficiency mutant. AATD type II mutants are likely to contribute to the under-diagnosed burden of disease in the general population.

Reagents and antibodies
Product details are listed in S1 Table.

Expression vectors
The mammalian expression vectors for expression of AAT variants are based on pcDNA3.1/ Zeo (+) [29]. Bacterial expression of hexahistidine-tagged protein was undertaken using pQE-30 (Qiagen). The AAT Iners mutation was obtained using the QuikChange II Site-Directed Mutagenesis Kit (Agilent) and the oligonucleotide 5'-taaaaacatggccctagcagcttcagtccctttct (and reverse complement thereof).

Bacterial expression and recombinant protein purification
Recombinant proteins were expressed in the XL1-Blue strain of E. coli and purified by nickelaffinity chromatography and ion-exchange chromatography as described previously [38]. Purity was assessed using 4-12% w/v acrylamide SDS-PAGE and 3-12% w/v acrylamide nondenaturing PAGE (Life Technologies) and the resulting proteins were exchanged into 20 mM Tris-HCl pH 7.4, 100 mM NaCl and stored at −80˚C.

Cell culture and transfection
HEK 293T/17 (ATCC, CRL-11268) and Hepa 1-6 (ATCC, CRL-1830) cells were maintained in DMEM/10% v/v FBS. Transfections with vectors encoding M1V, Z or AAT Iners were performed with PEI "Max" or with FuGENE HD as described previously [31,57,58]. To analyse AAT in the cell media, transfected cells were incubated in serum-free Optimem for 24h at 37˚C. Cell media were collected and centrifuged at 800g for 5'. Soluble and insoluble cellular fractions were obtained by lysing cells in 10 mM Tris-HCl pH 7.4, 150 mM NaCl, 1% v/v NP-40 and a protease inhibitor cocktail (BLA) with subsequent centrifugation at 16000g to separate the soluble to the insoluble fraction.

SDS-PAGE, non-denaturating PAGE and immunoblot
Cell media and intracellular fractions were resolved by 7.5% w/v acrylamide SDS-PAGE or by 8% w/v acrylamide non-denaturing as previously described [30,32]. The resolved proteins were blotted onto PVDF 0.45 μm membranes by wet transfer, probed with the indicated primary antibodies, revealed with HRP-conjugated secondary antibodies and detected by ECL Clarity and exposure to Hyperfilm ECL.

Sandwich ELISA
Quantification of AAT in cell media was performed by sandwich ELISA as previously described [32], using rabbit anti-AAT polyclonal antibody (pAb) for capture and HRP-conjugated sheep anti-AAT pAb for detection. AAT concentrations were calculated for each experiment using a standard curve of commercial purified AAT (Millipore) and expressed as percentages of M AAT concentration.

Formation of the inhibitory complex between AAT and serine-proteases
Culture media of HEK293T cells containing 10 ng of M1V or AAT Iners variants were incubated at 37˚C for 20 min with 1:1 and 1:2 molar ratios of AAT:HNE in 10 mM phosphate buffer pH 7.4/50 mM NaCl, before separation on 7.5% w/v acrylamide SDS-PAGE and immunoblot with anti-AAT pAb. Recombinant wild-type M1V and AAT Iners produced in bacteria were incubated at 37˚C for 30 min with equimolar purified human plasma kallikrein (KLK) in a buffer containing 10 mM Tris-HCl pH 8.0, 50 mM NaCl, 0.02% w/v PEG8000. Samples were then separated by 4-12% w/v acrylamide SDS-PAGE and revealed with Coomassie brilliant blue.

Assessment of proteinase inhibition
The stoichiometry of inhibition (SI) was assessed at 25˚C using bovine α-chymotrypsin or HNE as described previously [42].

Molecular dynamics analysis
Molecular dynamics (MD) simulations of the wild-type and G349R mutant protein were performed using the GROMACS software package [59,60]. The initial structure for the wild-type protein was taken from PDB 1QLP. The mutant structure was built from the wild-type, by replacing residue 349 with an arginine. The force field amber99-sb was used, with the PME method for Coulomb interactions and a Lennard-Jones potential with a cut-off of 10 Å for the short-range interactions. The initial structure was completed by addition of hydrogens and solvated with TIP3P water in a simulation box with a minimum distance of 10 Å between solute and box boundaries. Na + and Cl − ions were added to reproduce a salt concentration of 150 mM and to neutralize the system. All simulations were conducted at 310 K. The system was first subjected to energy minimization, then equilibration at constant volume for 100 ps, and at constant pressure for 100 ps was performed before the production run at constant temperature and pressure. The temperature was kept constant by velocity rescaling with a characteristic time of 0.1 ps. The pressure was controlled using the Parrinello-Rahman method with a time constant of 1 ps and a compressibility of 4.5�10 −5 bar −1 .

Statistical analysis
All the statistical analyses were performed by software Prism5 (GraphPad software Inc, San Diego, USA) as detailed in the figure legends.
Supporting information S1 Video. Movie showing one nanosecond of an MD trajectory of G349R AAT, during which the side chain of M358 changes orientation. Residues M358 and R349 are highlighted in ball-and-stick representation, while the secondary structure is shown for the rest of the protein. The movie was generated by means of VMD (http://www.ks.uiuc.edu/Research/vmd/). (MPG) S1 Table. Reagents and antibodies. (DOCX) S1 File. Alignments by Clustal Omega of the human alpha-1-antitrypsin (Uniprot P01009) with SERPINA1 orthologues or inhibitory and non-inhibitory human serpins. (DOCX)