Probing of the reactive center loop region of alpha-1-antitrypsin by mutagenesis predicts new type-2 dysfunctional variants

Abstract

Lung disease in alpha-1-antitrypsin deficiency (AATD) mainly results from insufficient control of the serine proteases neutrophil elastase (NE) and proteinase-3 due to reduced plasma levels of alpha-1-antitrypsin (AAT) variants. Mutations in the specificity-determining reactive center loop (RCL) of AAT would be predicted to minimally affect protein folding and secretion by hepatocytes but can impair anti-protease activity or alter the target protease. These properly secreted but dysfunctional ‘type-2’ variants would not be identified by common diagnostic protocols that are predicated on a reduction in circulating AAT. This has potential clinical relevance: in addition to the dysfunctional Pittsburgh and Iners variants reported previously, several uncharacterized RCL variants are present in genome variation databases. To prospectively evaluate the impact of RCL variations on secretion and anti-protease activity, here we performed a systematic screening of amino acid substitutions occurring at the AAT–NE interface. Twenty-three AAT variants that can result from single nucleotide polymorphisms in this region, including 11 present in sequence variation databases, were expressed in a mammalian cell model. All demonstrated unaltered protein folding and secretion. However, when their ability to form stable complexes with NE was evaluated by western blot, enzymatic assays, and a novel ELISA developed to quantify AAT–NE complexes, substrate-like and NE-binding deficient dysfunctional variants were identified. This emphasizes the ability of the RCL to accommodate inactivating substitutions without impacting the integrity of the native molecule and demonstrates that this class of molecule violates a generally accepted paradigm that equates circulating levels with functional protection of lung tissue.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00018-023-05059-1.

Introduction

Alpha-1-antitrypsin (AAT) is produced by hepatocytes and represents the most abundant serpin in plasma [1], with a basal concentration of 100–220 mg/dL (19–42 μM) that becomes elevated during an acute phase response [2]. AAT is a crucial contributor to proteostasis in the lung, and modulates inflammatory responses mainly by inhibiting two target chymotrypsin-like serine proteases, neutrophil elastase (NE) and proteinase-3 [3, 4], which are released by activated neutrophils. A crucial structural element involved in protease inhibition by serpins is the reactive center loop (RCL), an exposed stretch of amino acids that acts as a bait for target proteases and thereby mediates serpin-protease specificity. Inhibition proceeds via a “suicide-substrate” mechanism [5] that has been structurally elucidated using AAT complexed with trypsin or porcine pancreatic elastase (PPE) [6–8]. Protease inhibition occurs through well-defined steps: (a) formation of a non-covalent Michaelis complex by docking of the protease to the distal portion of the RCL in the vicinity of AAT residues M358-S359, respectively, named P1 and P1′ according to the conventional annotation of protease substrates [9]; (b) nucleophilic attack by the catalytic site serine on the carbonyl group of the P1 residue, forming an acyl-protease intermediate with the N-terminal fragment of the newly cleaved serpin; (c) prior to hydrolysis of this covalent intermediate, a thermodynamically-favoured conformational change occurs whereby the N-terminal fragment of the RCL is inserted as an extra stand in the serpin β-sheet A, the bound protease is translocated to the opposite pole of the serpin and the catalytic triad is distorted. Hydrolysis of the resulting complex (d) is significantly delayed and the protease irreversibly inactivated [8, 10]. One representation of this process is:

$$E+S\underset{a.Binding}{⟷}E·S\text{(Michaelis)}\underset{b.Cleavage}{⟶}E-S\text{*(Acyl-intermediate)}\begin{array}{ccc}\stackrel{\text{c. Translocation}}{⟶}& \text{E}-\text{S}& \text{(final inactive complex)}\\ \underset{\text{d. Hydrolysis}}{⟶}& \text{E + S*}& \text{(cleaved serpin, active enzyme)}\end{array}$$

where E and S denote the enzyme and serpin, respectively.

The hereditary disorder alpha-1-antitrypsin deficiency (AATD) highlights the pivotal role of AAT in regulating the innate inflammatory response in the lungs [11, 12]. In severe cases of AATD, such as those arising in homozygote carriers of the Z-AAT mutant (E342K), levels of AAT in the circulation decrease below a 'protective threshold’ concentration leading to the uncontrolled activity of elastolytic proteases in the lungs, progressive damage of the alveolar walls and development of early-onset emphysema [13]. The decreased plasma levels in ZZ homozygotes are the consequence of misfolding accompanied by degradation and polymerization of the Z-AAT mutant within the endoplasmic reticulum of hepatocytes [14, 15], which also predispose to liver disease manifestations such as hepatitis in newborns and cirrhosis in adults [16]. The occurrence of Z-AAT in compound heterozygosity with the relatively common S-AAT variant (E264V) also leads to an increased risk of lung and liver dysfunction [17], which could be explained by formation of SZ heteropolymers [18, 19]. Moreover, several ultra-rare AAT alleles have been shown to cause AATD, generally in association with the more common S and Z alleles. These include null mutations [20] with undetectable plasma AAT, as well as missense mutations associated with different degrees of plasma deficiency and variable intrahepatic accumulation of AAT polymers [21, 22].

However, mutations in the RCL of AAT have revealed additional mechanisms of AAT dysfunctionality that arise from impairment in one of the steps of the inhibitory pathway mentioned above, rather than AAT retention within hepatocytes. To date, the only RCL mutant described in patients is the Pittsburgh (M358R) variant. Substitution of methionine P1 with an arginine residue does not affect AAT secretion by hepatocytes, but abrogates its interaction with NE and changes AAT specificity, which becomes a potent thrombin inhibitor, thus leading to a bleeding disorder [23–25]. In our previous work, we showed that exome databases can reveal dysfunctional variants that have not yet been described clinically, such as Iners-AAT (G373R), which displays normal folding and secretion when expressed in cellular models but has a severely impaired anti-elastase activity. This has allowed us to extend the concept of ‘type-2’ functional deficiency seen with homologous serpin proteins to AAT [26].

The observation of the relative propensity with which people heterozygous and homozygous for the Z allele develop lung disease has led to the general acceptance of a ‘protective concentration threshold’ paradigm in AATD. It is generally observed that levels above ~ 11 μM are protective for the lung disease [27]. This reference value has proved to be useful in the context of the more common pathogenic genotypes, but the implicit correspondence between circulating levels and functional anti-protease activity would not hold for rare AAT variants that are well-secreted but functionally inactive. Notably, as rare deficient mutations associated with AATD are increasingly identified [28], dysfunctional type-2 mutants may be overlooked by standard diagnostic protocols that triage based on circulating AAT levels.

Here we sought to establish the pathogenic potential of a panel of mutations in the protease specificity-determining region of the RCL. AAT variants that could arise from single nucleotide polymorphisms affecting the RCL region were expressed in a mammalian cell line to evaluate their effect on protein folding and cellular secretion in a system that models the physiological context. The use of such a system, rather than one based on expression in E. coli—which lacks the mammalian folding, glycosylation and secretion machinery, and expresses a non-mammalian proteolytic repertoire—allowed us to more directly draw conclusions regarding the likely overall effect of the mutations in vivo. We characterized the anti-protease activity of the secreted, glycosylated variants in the cell media by multiple biochemical approaches. We also report here the development and characterization of a novel ELISA method to quantify AAT–NE complexes, named protease complex detection ELISA (PCD-ELISA), which has potential as the basis for a quantitative diagnostic tool for evaluating the AAT inhibitory activity in clinical samples. Our results demonstrate the high tolerance of the RCL to mutations, as our panel of variants perturbed neither protein folding nor the efficiency of cellular secretion. However, several of these variants were found to impair anti-elastase activity, reflecting the potential pathogenic relevance of this class of type-2 mutations to AATD.

Results

Structural analysis of the RCL distal region (P5-P4′)

Crystal structures of native AAT show that the RCL (G344-P362) is an exposed loop protruding from the protein core [29, 30] (Fig. 1a, b). The conformation of the distal C-terminal portion of the RCL, encompassing residues from A355 to P362—P4-P4′ by substrate nomenclature [9]—is crucial for AAT activity as it mediates AAT docking to target proteases. Salt bridges between the glutamate in the P5 position and arginine residues in the AAT main body help to maintain the canonical substrate-like conformation of the protease docking region [31]. Following the proteolytic cleavage-induced conformational transition, the P14-P1 portion of the RCL (G344-M358) becomes inserted as an extra strand into β-sheet A in the final inhibitory complex (Fig. 1c). Consequently, the P14-P1 residues exist in two very different structural contexts; substitutions within the exposed RCL that are compatible with folding to the native conformation may not be tolerated by the loop-buried cleaved conformation. The pivotal role of the P5-P4′ region to protease binding and the requirement for compatibility with insertion into β-sheet A is reflected by the substantial conservation of its amino acid sequence among mammalian AAT orthologues, with a minor exception of I360 (P2′) showing tolerance among mammalian AAT for other hydrophobic residues (Fig. 1d).

Fig. 1.

Structural context of the residues mutated in this study (P5-P4′). a Atomic model of the native AAT conformation (PDB 3NE4) [30] is shown as cartoon, coloring α-helices in green, β-sheets in purple and loops in grey, except for the RCL that is highlighted in light blue. Residues of the P5-P4′ positions of the RCL are shown as sticks and labelled. b Close-up of the P5-P4′ region shown in panel (A): salt-bridges between E354 (P5) and arginine residues of the protein core (R196-R223-R281) and a coordinating water molecule (red sphere) are represented as red dashed lines. c Atomic model of the complex between AAT and trypsin (PDB 1EZX) [8]. AAT representation is the same as in panel (a), while trypsin is reported as a black cartoon. d WebLogo representation [32] of an amino acid multiple alignment among 100 mammalian AAT orthologues. Proteins were identified using BLAST [33] and aligned with COBALT [34]. Colors refer to chemical properties of the residues: hydrophobic, polar, and acidic residues are in black, green, and red, respectively. Residues are numbered according to the conventional AAT sequence and the P-substrate notation

To predictively assess the pathological implications of variations within the P5-P4′ region, we identified all 55 possible single nucleotide missense variations in this region and among these we selected a panel of non-conservative amino acid substitutions for characterization in a mammalian cell model. This selection was pruned to incorporate substitutions that exemplified a change in physicochemical properties (charge, polarity and size) while avoiding duplicates of a given physicochemical class, where these arose. We also included uncharacterized single-nucleotide variations annotated in the GnomAD and ClinVar exome and genome sequence variation databases within the P5-P4′ region and the previously characterized Pittsburgh mutant. The resulting selection is shown in Table 1.

Table 1.

Amino acid substitutions within the RCL P5-P4′ region selected for characterization

Reference AAT residues (Uniprot ref #P01009-1) are reported according to the sequence numbering recommended by the human genome variation society (HGVS), the conventional AAT numbering that excludes the N-terminal leader sequence, and the P-substrate notation. Reference amino acids and substitutions are colored according to their chemical properties: grey for glycine, black for hydrophobic, green for polar, red for acidic, and blue for basic residues. Asterisks indicate AAT variants identified in the GnomAD [35] (*) and ClinVar [36] (**) databases

P5-P4′ variants are normally secreted and do not polymerize when expressed in mammalian cells

The secretion efficiency of AAT variants by mammalian cells was investigated by transiently transfecting HEK293T/17 cells with expression vectors encoding the P5-P4′ variants reported in Table 1. As controls, we included the wild-type M-AAT, the polymerogenic Z-AAT, and the dysfunctional Iners-AAT variants. We measured AAT content in cell supernatants by sandwich ELISA with polyclonal antibodies both for AAT capture and detection (Fig. 2a). Transfection with a mock control plasmid showed an absence of cross-reactivity of the antibodies with endogenous components of the HEK293T/17 cells present in the cell media. As previously reported, the secretion-deficient reference Z-AAT showed 23% secretion relative to M-AAT, reflecting its intracellular degradation and accumulation as polymers [19, 37]. Conversely, all P5-P4′ and Iners variants did not show evidence of a secretory defect and secreted levels were comparable to the wild-type protein. In parallel, we analyzed equal volumes of the same supernatants by PAGE and western blot, both under non-denaturing and denaturing/reducing conditions. This experiment confirmed that Z-AAT is secreted less than M-AAT, while P5-P4′ and Iners-AAT variants were secreted at the same extent as the wild-type protein (Fig. 2b). Under non-denaturing conditions, secreted Z-AAT showed higher molecular weight bands, likely representing polymers of Z-AAT [19, 37], which were better revealed by a non-denaturing-PAGE in which loading of cell supernatants was normalized to their AAT content (Fig. S1). The secreted RCL variants mainly migrated as monomers, with a low abundance of higher molecular weight bands observed only in the P357S (P2), M358R and M358K (P1) samples. To further assess the tendency of the P5-P4′ variants to polymerize, we analyzed the supernatants by a sandwich ELISA that specifically captures polymeric AAT by the conformation-specific monoclonal antibody (mAb) 2C1 [38]. We observed that 80% of the secreted Z-AAT consisted in polymers, while AAT polymers secreted by the other variants in study were either undetectable or below 2.6% of total AAT (Fig. 2c). Together, these results show that AAT variants with substitutions by amino acids with substantially different physicochemical properties within the P5-P4′ region were not polymerogenic and were normally secreted by mammalian cells.

Fig. 2.

P5-P4′ variants are normally secreted by HEK293T/17 cells and do not polymerize. Supernatants of HEK293T/17 cells transiently transfected with expression vectors for M, Z, P5-P4′, and Iners AAT variants were collected 24 h post-transfection. Empty vector (mock) was included as negative control. a Sandwich ELISA quantification of total AAT in cell supernatants utilizing polyclonal antibodies for AAT capture and detection, normalized to the respective total protein content. Values are expressed relative to M-AAT (n = 3). b Western blot analysis of AAT content in equal volumes of cell supernatants resolved by either non-denaturing-PAGE (top panels) or reducing SDS-PAGE (bottom panels) and detected by a polyclonal anti-AAT antibody. Black and white arrowheads indicate the monomeric AAT form resolved under non-denaturing and denaturing/reducing conditions, respectively. Differences in the migration of the monomers of different RCL variants under non-denaturing conditions is due to altered charge. The blots are representative of three independent experiments. c Sandwich ELISA quantification of AAT polymers in cell supernatants, utilizing an AAT polymer-specific monoclonal antibody (2C1 mAb) as capture and polyclonal anti-AAT antibody for detection. Polymeric AAT is expressed as percentage of total AAT measured in panel (a) (n = 2). Data in a and c are reported as mean ± SD. Significant differences with M-AAT were identified with one-way ANOVA analysis corrected by Dunnett’s test (no asterisk: non-significant; ***p ≤ 0.01). The supernatants whose total AAT or polymeric AAT were below detection limit of the ELISA assays are reported as N.D

P5-P4′ variants include complex-forming, substrate-like and binding-deficient AAT molecules

Substitutions within the P5-P4′ region can potentially interfere with protease recognition (preventing RCL cleavage and insertion) or with the translocation mechanism (allowing cleavage but preventing stable complex formation). An accurate quantification of inhibitor concentration is a key parameter required for inhibitory assays and the levels determined in immunoassays such as those presented above were used in all subsequent experiments for this purpose.

The ability of the secreted AAT variants to form a stable complex with NE was evaluated. A hallmark of the serpin-enzyme complex is that it remains intact during SDS-PAGE. To account for a possible inactive component of the NE used in our experiments, the molar concentration of active protease was determined by titration against plasma-derived pure AAT of known concentration, prior to performing NE-inhibition assays (Fig. S2). We incubated each secreted AAT variant with increasing amounts of human NE, separated by electrophoresis and detected AAT by western blot. Based on their inhibitory behavior, three classes of variants were evident in this experiment: complex-forming, substrate-like and binding-deficient.

Representing a fully functional inhibitor, secreted M-AAT generated a complex with NE in a dose-dependent manner and, at the equimolar point, only a ~ 75 kDa complex with the protease could be detected. In the presence of 1.5 molar excess of NE, we observed a band of intermediate molecular weight between AAT and AAT–NE complex, consistent with secondary proteolytic cleavage of the complex as observed previously [39] (Fig. 3a, M panel). A set of P5-P4′ variants (E354K, E354G, A355T, A355V, I356R, M358I, M358T, S359A, S359Y, I360N, P361R, P362R, P362H, P362S, P362T, and P362A) likewise complexed with NE under equimolar conditions (Fig. 3a).

Fig. 3.

Complex formation between NE and secreted AAT variants. Each secreted AAT variant (20 nM) was incubated with increasing quantities of human NE for 1 h at room temperature, generating reactions at different NE:AAT molar ratios as shown. Reaction mixtures were resolved by SDS-PAGE under reducing conditions and AAT was detected with an anti-AAT polyclonal antibody. The AAT variants are grouped according to their behavior as a complex-forming, b substrate-like and c binding-deficient variants. Bands corresponding to AAT and AAT–NE complex are indicated by white and black arrowheads, respectively. Black arrows indicate the cleaved form of AAT that lacks the C-terminal fragment (S359-K394). Blots are representative of three independent experiments

Conversely, we have shown the Iners-AAT mutation to impair the protease translocation that occurs during RCL insertion into β-sheet A and to exhibit a corresponding substrate-like behavior. This results in a cleaved Iners AAT form that migrates in SDS-PAGE as a band lower than the 52 kDa native AAT, due to dissociation of the C-terminal fragment (S359-K394) under denaturing conditions. Like the Iners-AAT variant, several RCL variants showed a cleaved form when incubated with NE, and they completely (A355D and A355P) or partially (P357R and P357S) failed to generate complexes with NE (Fig. 3b).

By contrast, the previously described dysfunctional Pittsburgh variant (M358R) did not interact with NE, as shown by marked reduction of AAT–NE complexes and absence of an AAT cleaved form (Fig. 3c). Instead, the excess of active protease caused proteolytic cleavage at multiple sites revealed by decreasing AAT band intensity, and produced faster-migrating AAT fragments in the immunoblots (not shown). M358K and S359P variants showed similar behavior to Pittsburgh, reflecting a disrupted NE docking site, leading us to classify them as binding-deficient RCL variants (Fig. 3c).

Stoichiometry of NE inhibition by the P5-P4′ variants

Further insight into the degree of dysfunctionality of P5-P4′ variants was obtained by determining the stoichiometry of NE inhibition, defined as the number of AAT molecules necessary to inhibit one molecule of NE, and reported in Table 2 and figure S3. A completely functional inhibitor has a stoichiometry of inhibition (SI) value of 1, reflecting the 1:1 nature of the interaction between M-AAT and NE, the irreversible NE inhibition and the stability of the so-formed AAT–NE covalent complex. In these experiments, we verified that no NE inhibition was observed for supernatants of HEK293T/17 cells transfected with mock plasmid, nor was conversion of the substrate observed in the absence of exogenous NE (not shown), thus excluding interference by other inhibitors or proteases of cellular origin. All but one of the P5-P4′ complex-forming variants identified above (E354K, E354G, A355T, A355V, I356R, M358I, M358T, S359A, S359Y, I360N, P361R, P362R, P362H, P362S, P362T, and P362A) were found to not be significantly different from M-AAT, with an SI value close to 1. The exception was I356R whose SI was 1.40 ± 0.24 (± SD, n = 3) (Fig. S3a). For the substrate-like class, consistent with a defect of protease translocation, the A355D and A355P variants showed no ability to inactivate NE, while Iners was severely compromised and showed an SI value of 6.74 ± 0.42 (Fig. S3b). Variants P357R and P357S were partially affected with SI values of 1.52 ± 0.24 and 3.20 ± 0.21, respectively. The binding-deficient M358K, M358R (Pittsburgh) and S359P variants were completely ineffective in inhibiting NE (Fig. S3c).

Table 2.

Stoichiometry of NE inhibition by the AAT variants

	AAT variant	SI	SD
Controls	M	1.00	0.06
	Mock	ND	–
Complex-forming	E354K	1.06	0.08
	E354G	1.12	0.26
	A355T	0.95	0.06
	A355V	1.04	0.07
	I356R	1.40*	0.24
	M358T	1.10	0.06
	M358I	1.06	0.04
	S359A	1.17	0.11
	S359Y	1.23	0.30
	I360N	1.18	0.17
	P361R	1.06	0.04
	P362R	0.99	0.05
	P362S	0.88	0.12
	P362T	0.98	0.04
	P362H	0.94	0.04
	P362A	0.99	0.03
Substrate-like	Iners	6.74*	0.42
	A355D	ND	–
	A355P	ND	–
	P357R	1.52*	0.24
	P357S	3.20*	0.21
Binding-deficient	M358R	ND	–
	M358K	ND	–
	S359P	ND	–

For each secreted AAT variant, increasing volumes of supernatant containing 0.4-0.7-1.0-1.5-2.2-3.3-5.0-7.5 nM AAT were incubated with human NE (5 nM) for 1 h at room temperature. Residual NE activity was measured by hydrolysis of 0.4 mM MeO-Suc-Ala-Ala-Pro-Val-pNa. SI values were determined, normalized to M-AAT, and reported as mean ± SD (n = 3). For those supernatants that did not inhibit NE activity, SI values could not be determined (ND). Significant differences from M-AAT were calculated with one-way ANOVA analysis corrected by Dunnett’s test and those that had p values < 0.05 are denoted by an asterisk*

Quantification of AAT–NE complexes by a novel ELISA method

The elastase complex formation immunosorbent assay (ECFISA) is a recently reported method that measures AAT inhibitory activity by its binding to PPE-coated polystyrene plates [40]. Since cell media supernatants comprise complex heterogeneous protein-rich mixtures, and the proportion of secreted AAT is relatively low, we considered this approach might be useful for evaluation of the mutants. However, a correlation matrix between fractional activity (the reciprocal of SI values determined above) and the ECFISA binding signal relative to M-AAT (Fig. S4a) showed a poor correlation (r = 0.41): most of P5-P4′ variants exhibited a reduced activity when analyzed by ECFISA compared to that determined by their stoichiometry of NE inhibition. Most strikingly the M358I variant, whose SI was comparable to the wild-type protein (1.06 ± 0.04) showed a twofold higher PPE binding than M-AAT (Fig. S4b). This discrepancy is likely due to the fact that while NE and PPE both are ‘elastases’, they are not orthologous proteases and exhibit different substrate specificity and AAT interaction kinetics [3].

We were not able to substitute PPE with NE during the ECFISA plate coating step due to the latter’s susceptibility to autoproteolysis. Consequently, we developed a novel method, PCD-ELISA, based on formation of AAT–NE complexes in solution followed by their capture and detection by a sandwich ELISA. A monoclonal anti-AAT antibody (3C11 mAb) [41, 42], was used for the capture and then complexed NE was detected by an anti-NE polyclonal antibody. This assay detected complexes between purified plasma-derived M-AAT and human NE in a dose-dependent manner, with a lower detection limit in the pM range, confirming that the 3C11 mAb epitope is accessible once AAT is complexed with NE. Moreover, when the experiment was performed with the same purified M-AAT inactivated either by heat-induced polymerization or oxidation, the assay showed undetectable NE binding signal, reflecting specificity toward correctly formed AAT–NE complexes (Fig. S5).

The AAT variants were incubated with equimolar human NE and evaluated by PCD-ELISA. In contrast to the ECFISA approach, the P5-P4′ variants previously shown to generate stable serpin-elastase complexes bound NE to a similar extent as secreted M-AAT. The only exception was the I356R variant that showed a 30% lower signal than M-AAT (Fig. 4a), in agreement with its slightly elevated SI. The M358R, M358K and S359P variants showed negligible binding signals indistinguishable from the mock control, consistent with their lack of interaction evident by SDS-PAGE and SI.

Fig. 4.

Quantitative ELISA analysis of AAT–NE complexes (PCD-ELISA). The supernatant of each AAT variant (2 nM AAT) was incubated with equimolar human NE in solution for 1 h. AAT molecules (unbound and NE-complexed) were captured in plates coated with a mouse monoclonal anti-AAT antibody (3C11 mAb) and bound NE was detected by an anti-NE antibody. Secreted M-AAT, complex-forming, substrate-like and binding-deficient AAT variants are reported in black, blue, green and red, respectively; mock control is reported in white. a Binding signal of M-AAT was utilized for normalization and the data are reported in bars as mean ± SD (n = 2). Significant differences with M-AAT were calculated by one-way ANOVA analysis corrected with Dunnett’s test (no asterisk: non-significant, *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001). b Plot of correlation matrix between the stoichiometry of NE inhibition (reciprocal of SI value) and the detection of AAT–NE complexes by PCD-ELISA (NE binding signal relative to M-AAT). Pearson correlation coefficient (r) was calculated with 95% confidence interval (dotted lines) and reported in the graphs as inset

Among the substrate-like mutants, the A355D, A355P, P357S and Iners variants were also reported as dysfunctional by PCD-ELISA; conversely the mildest of this class of mutants, the P357R variant, showed binding to NE not significantly different from the wild-type protein. The correlation between the stoichiometry of NE inhibition and detection of AAT-NE complexes by PCD-ELISA revealed that the latter is a robust discriminator between complex-forming and binding-deficient variants. Interestingly, for the substrate-like variants A355P and Iners, there was a lesser impact on binding by PCD-ELISA than observed in SDS-PAGE and SI experiments (Fig. 4b). Translocation of the protease occurs in three distinct phases, with different degrees of active site distortion and susceptibility to dissociation [43]; our working hypothesis is that these substrate-like variants generate complexes that are partially inactivated and more readily disrupted by the conditions of the SDS-PAGE and SI experiments. To corroborate this, we analyzed the substrate-like variants by electrophoresis under non-denaturing conditions and revealed the AAT-NE complexes by immunoblot (Fig. S6). This analysis confirmed that, in addition to the P357R and P357S variants, complexes with NE are detectable also for the A355P and Iners variants under non-denaturing conditions. Conversely, the complex of NE with the A355D variant, that showed no binding in PCD-ELISA, was undetectable. Interestingly therefore, for this class of mutants, it appears that dysfunctionality can be more nuanced, and the standard use of enzymatic assays and SDS-PAGE may under-quantify protease engagement relative to approaches that provide direct measures of binding. 

Kinetics of NE inhibition of the P5-P4′ variants

Whether or not a mutation has an impact on the stoichiometry of the interaction with NE, it may slow down the rate at which inhibition occurs relative to the wild-type protein. Consequently, rates of NE inhibition (k_inh) by secreted P5-P4′ variants were evaluated by the progress curve method [44, 45]. Monitoring of the turnover of substrate by neutrophil elastase in the absence of inhibitor showed the enzyme to remain stable and active in the cell media over the course of the experiment. The resulting k_inh values are reported in Table S1. In our experimental system, secreted M-AAT revealed a k_inh of 8.6⋅10⁶ ± 0.3⋅10⁶ M⁻¹ s⁻¹, in good agreement with previously published data [3, 46, 47].

All P5-P3′ AAT variants that complexed with NE were found to do so between 1.5- to 12.5-fold more slowly than M-AAT, with the exception of I356R that showed an even more substantial impairment with a 10⁴-fold lower k_inh value (Fig. 5a). In contrast, variants bearing amino acid changes in P4′ showed comparable rates to the wild-type protein.

Fig. 5.

NE inhibition rate of P5-P4′ variants. a Human NE (0.25 nM) was incubated with MeO-Suc-Ala-Ala-Pro-Val-AMC (0.75 mM) and an appropriate volume of supernatant of cells expressing each AAT variant (yielding a 7.5-fold molar excess normalized for the respective SI). NE activity was measured by the accumulation of fluorescence due to substrate hydrolysis (excitation 367 nm, emission 460 nm). k_inh values were calculated as described in the methods section, expressed relative to the M-AAT control, and reported in bars as mean ± SD (n = 2). Those secreted variants whose progress curves did not fit the pseudo-first order equation are indicated as ND. Secreted M-AAT, complex-forming, translocation-deficient and binding-deficient AAT variants are reported in black, blue, green, and red, respectively. Significant differences with M-AAT supernatant were calculated with one-way ANOVA analysis corrected by Dunnett’s test (no asterisk: non-significant, *p ≤ 0.05, **p ≤ 0.01, ****p ≤ 0.0001). b Progress curves of the indicated AAT variants. For those variants whose SI value could not be determined (A355D, A355P, M358R, M358K, and S359P) and the mock control, equal volumes of cell supernatants were analyzed. As reference, the progress curves in the presence and absence of a 7.5-fold molar excess of secreted M-AAT are reported

Even though the progress curves of the severely deficient variants could not be fitted to a pseudo-first order equation due to incomplete NE inhibition at the concentrations achievable in the experiment, their profiles provided information about the mechanistic basis for their dysfunction. In the case of A355D and A355P, a gradual regain of NE activity was observed during the course of the reaction (Fig. 5b), indicating an impact on complex stability likely due to a defect in protease translocation. In comparison, the binding deficient variants M358R, M358K, and S359P showed either a very slow rate of interaction or undetectable inhibition resembling the AAT-free mock control.

Discussion

The prevalent pathogenic AAT variants Z-AAT (E342K) and S-AAT (E264V) were initially noted for their reduction of protein levels in electrophoretic analysis of plasma components. The reason for this is that they, along with many ultra-rare AAT variants that have been described since, are prone to misfolding in the ER of hepatocytes and exhibit a correspondingly reduced secretion into the bloodstream. As the primary function of AAT is as a protease inhibitor, low circulating levels are implicitly equated with a reduced protection of lung tissue from elastolytic proteases. Consequently, reduced circulating AAT is the main triage parameter considered in a typical AATD clinical diagnostic pipeline.

However, AAT activity can be perturbed not only by folding defects hindering secretion of the protein, but also by mutations that target regions crucial for inhibitory activity. The Pittsburgh variant (M358R) represents a striking demonstration of how a single polymorphism can entirely alter the protease target despite exhibiting normal levels of circulating AAT, with a mutation at the crucial protease-bait P1 position resulting in loss of anti-NE activity and gain of new specificity against thrombin. The RCL represents an attractive target for engineering of bespoke inhibitory activity, and many such efforts have been reported in the literature and in patents, from single substitutions to exchange of the entire region. Examples include the elimination of inhibitory activity by introducing the T345R mutation seen in ovalbumin [48], substitution of the oxidation-susceptible P1 methionine with another hydrophobic residue [49] and the production of recombinant Pittsburgh-AAT variants with enhanced specificity for factor XIa [50].

The understanding that has driven these rational engineering strategies has relevance to the AATD clinical context. Recent advances in large-scale human population screening by genome and exome sequencing has revealed the presence of variants encoding mutations in the RCL. One such example is Iners AAT (G373R) that we characterized in cellular models as normally secreted but severely dysfunctional [26]. We have proposed that this is classified as a ‘type-2’ AAT variant, in common with the designation given to similar mutations in other serpins, C1-inhibitor [51] and antithrombin [52]. Type-2 dysfunctional variants would be expected to be associated with an AATD lung phenotype due to lack of NE inhibition, but they would escape detection by a typical diagnostic flowchart based on the quantification of AAT levels in circulation without taking account of activity. Consistent with the observation that mutations in the RCL do not generally impact folding, the Iners AAT variant was predicted to be ‘likely benign’ by bioinformatic tools designed to identify pathogenic mutations [21]. Indeed, two pathogenicity predictors validated for SERPINA1 non-synonymous polymorphisms, Polyphen-2 and REVEL (Table S2), failed to correctly classify the majority of dysfunctional variants found in our study, revealing that dysfunctional mutations affecting the RCL are mis-interpreted by these commonly used algorithms.

In this work, we characterized the non-conservative missense variants that can arise from single-nucleotide polymorphisms within the P5-P4′ specificity-determining region of the RCL, including some uncharacterized variants that are present in exome/genome sequence variation databases. We used the mammalian cell line HEK293T/17 to assess effects on secretion in the context of mammalian cell post-translational modifications. We demonstrated that all the investigated RCL variants were secreted at levels comparable to wild-type M-AAT and exclusively in the monomeric form, thus excluding major conformational alterations of the protein fold.

Anti-elastase activity of the secreted P5-P4′ variants was assessed by multiple assays directly performed on the conditioned, FBS-free media of transfected cells. These were successfully used without additional purification procedures, supporting their potential application to complex mixtures such as human plasma. Our experiments revealed that the mutants, despite their common location within the specificity-determining region of the molecule, impacted the inhibitory mechanism in different ways. Analysis of complex formation by immunoblot analysis and SI assays led us to classify the P5-P4′ variants into three main subtypes (Fig. 6): the complex-forming group showed a stoichiometry of complex formation similar to wild-type M-AAT (Fig. 6a); the substrate-like group in which the variants were inactivated by NE to varying extents through non-productive cleavage (Fig. 6b); and the binding-deficient group that failed to interact with NE (Fig. 6c).

Fig. 6.

Schematic representation of the proposed mechanisms of AAT dysfunctionality. The serpin AAT is represented in black, while its target protease NE is in cyan. Based on the anti-NE activity of the RCL variants of AAT we observed distinct behaviors: a the complex-forming group of mutants showed a stoichiometry of complex formation similar to wild-type M-AAT, thus reflecting AAT cleavage, release of the C-terminal fragment (showed in red), and protease translocation, eventually leading to an irreversible covalent complex with the inactivated NE; b the substrate-like group of mutants failed to properly inhibit NE because of defective protease translocation, resulting in hydrolysis of the complex, thus releasing the inactive cleaved AAT and active NE; and c the binding-deficient group of mutants did not bind the target protease

The disruptive effect of methionine 358 substitutions with basic residues (M358R and M358K), but not with the hydrophobic isoleucine (M358I) and threonine (M358T), were consistent with the preference of NE for a hydrophobic amino acid in position P1 [49, 50, 53, 54]. The NE interaction was also abrogated by substitution of serine 359 in the P1′ position with a proline, which may introduce a steric distortion not compatible with NE docking, while replacement by a small non-polar alanine (S359A) or bulky tyrosine (S359Y) sidechain had no effect. The A355D and A355P mutants, situated at the P4 position that in the cleaved conformation is incorporated into a cluster of hydrophobic residues at the base of β-sheet A, showed a severely dysfunctional behavior very similar to the translocation-deficient Iners AAT. This is consistent with previous results that demonstrated an impact of A355D on inhibitory activity and polymerization [55], a variant that has since been reported in the ClinVar database indicating that it has been found in an individual. Conversely, the P4 site proved to be tolerant of the small residues threonine (A355T) and valine (A355V). An intermediate defective phenotype was found in the P357R and P357S variants which were partially ineffective against NE with non-productive cleavage evident by electrophoresis, an elevated SI value and a modest reduction in the rate of interaction. Following RCL insertion into β-sheet A, the P2 position, unlike P4, does not become part of a hydrophobic cluster but instead remains partially solvent-exposed, suggesting that the hydrophilic nature of the substituted side chains should remain compatible with this insertion. These data, therefore, suggest that the inhibitory mechanism relies on the conformational restriction of the backbone by the proline to ensure orderly translocation of the protease.

One discrepancy with the previous literature was found with the E354G and E354K variants. While the SI found in our study was at wild-type levels, disruption of the salt-bridge with R196 and R223 was previously noted to severely impair the stoichiometry of NE inhibition [31]. One key difference is that the P5 variants in our study have been secreted by mammalian cells and thus in the presence of endogenous chaperones and native glycosylation patterns, while the previous work used protein refolded from E. coli inclusion bodies. Our results do not preclude the importance for the glutamic acid in the P5 position, but suggest, based on the two to fourfold decrease in the rate of NE inhibition, that the P5 interactions with AAT core arginines provide a more nuanced optimization of RCL presentation.

For those variants that exhibited some degree of inhibitory activity, kinetic analysis by the progress curve method revealed that all non-conservative changes in the P5-P3′ region had an impact on the rate of NE inhibition. Consequently, some of these complex-forming variants, despite being able to bind and inhibit the same amount of NE molecules as the M-AAT at the endpoint, may nevertheless exert an impact by delaying quenching of NE activity in the lungs and therefore timely protection of the tissue from proteolytic damage.

The experimental setup required for enzymatic stoichiometric and kinetic assays would be expected to limit their routine implementation for measuring AAT activity in plasma samples in most clinical laboratories. Recently, a novel plate-based ELISA assay, ECFISA, was proposed to evaluate AAT binding to PPE-coated plates [40]. Since in our experimental model AAT variants are secreted by mammalian cells at relatively low concentrations, mixed with a plethora of other proteins, the low detection limit and AAT enrichment achieved by a capture-based approach such as ECFISA are advantageous properties. However, we observed fundamental divergences between AAT activity measured by ECFISA and inhibitory assays. Importantly, these discrepancies may be explained by the fact that PPE is not the physiological target of AAT, with differences in substrate preference and fine details of enzymatic and structural characteristics relative to NE. To overcome this issue, our laboratory has developed a novel, sensitive method to detect the AAT–NE complex, namely PCD-ELISA. This assay was applied to the RCL variants and showed a good correlation with the results obtained by the enzymatic approach, thus allowing the identification of dysfunctional variants. Nevertheless, a few variants displaying translocation defects were reported to have a higher degree of complex formation by PCD-ELISA compared to that observed in the SI assay, apparently due to increased perturbation of partially inactivated protease by the conditions of the SI experiment. We propose therefore that for clinical evaluation, PCD-ELISA is to be preferred due to its ability to maintain a native-like environment even within complex mixtures. In a research setting, our results also reveal the potential benefit of integrating multiple approaches to quantify inhibitory activity and to obtain mechanistic information on uncharacterized variants.

This study provides the first systematic characterization of AAT variants that can feasibly result from single nucleotide polymorphisms, encompassing the P5-P4′ region of the RCL. The use of mammalian cells has allowed us to evaluate secretion of proteins that have undergone cellular quality control processes. This cell-centric approach has been previously shown to allow characterization of AAT variants of clinical relevance with a high degree of confidence [28]. Our results reveal the likely clinical impact of dysfunctional mutations within a region crucial for NE recognition and inactivation; it also highlights a selection bias in diagnostic assays that mainly consider AAT circulating levels and may thus overlook type-2 functional deficiency. It is worth noting that even the predominant disease-associated Z allele causes a loss of ~ 30–40% activity and a decreased rate of association. We propose an approach that quantifies anti-protease activity, in a similar fashion as the PCD-ELISA developed here, to determine a protective functional activity threshold, rather than a protective concentration threshold, as a central concept in the evaluation of AATD.

Materials and methods

Expression vectors encoding AAT variants

Expression vectors based on pcDNA3.1/Zeo ( +) and encoding M1V (M-AAT), Z-AAT and Iners AAT were described previously [26, 37, 56]. Expression vectors encoding the P5-P4′ variants were obtained from pcDNA3.1/Zeo ( +)/M1V using the QuikChange II site-directed mutagenesis kit (Agilent, 200,523), according to the manufacturer’s instructions, and using the mutagenic oligonucleotide primers reported in Table S3. Each AAT variant cDNA was checked by Sanger sequencing (Eurofins Genomics).

Cell culture and transient transfection of HEK293T/17 cells

HEK293T/17 cells (ATCC CRL-11268) were cultured at 37 °C with 5% CO₂ in Dulbecco′s Modified Eagle′s Medium (DMEM) containing 4.5 g/L glucose, 2 mM L-glutamine, 1 mM sodium pyruvate (Sigma-Aldrich, D6429), supplemented with 100 I.U./mL Penicillin–Streptomycin (Sigma-Aldrich, P0781) and 10% endotoxin-free fetal bovine serum (FBS; Sigma-Aldrich, F7524). Cells were seeded at 1·10⁵ cells/cm² and cultured overnight. Transfection mixtures consisted of serum-free DMEM, 40 μg/mL PEI MAX® (Polysciences, 24,765), and 6 μg/mL of plasmid expression vectors, incubated at room temperature for 20 min. When cells reached 80% confluency, the medium was replaced with fresh culture medium and transfection mixture in 2:3 volume ratio [57]. Five hours post-transfection the medium was removed, cells were washed twice with pre-warmed serum-free DMEM and once with OptiMEM (Gibco, 11058021), and further cultured in OptiMEM in the absence of FBS. Supernatants of transfected cells were harvested 24 h post-transfection and cells debris removed by centrifuging at 3000 g for 5 min at 4 °C. Cell supernatants were aliquoted and stored at −20 °C for subsequent analyses.

Measurement of secreted AAT by sandwich ELISA

High-binding 96-well microplates (Costar, 3690) were coated with 2 μg/mL rabbit anti-AAT polyclonal antibody (Dako, A001202) in PBS, overnight at 4 °C. The plate was washed with PBS + 0.05% Tween-20 (PBS-T), non-specific binding blocked with 0.25% bovine serum albumin (BSA; Sigma, A9647) in PBS-T, for 1 h at room temperature, and washed again with PBS-T. M-AAT purified from plasma (Athens Research & Technology, 16–16-011609) was used to generate 11-point calibration curves by 1:2 serial dilution starting from 500 ng/mL in PBS-T + 0.1% BSA (dilution buffer). Calibration samples and dilutions of cell supernatants (1:50 and 1:100 prepared in dilution buffer) were applied to the coated plates, incubated for 1 h at room temperature and subjected to three rounds of washing with PBS-T. Detection of AAT was carried out by incubation with 67 ng/mL horseradish peroxidase (HRP)-conjugated goat polyclonal antibody anti-AAT (Abcam, ab191350) in dilution buffer, for 1 h at room temperature. After three rounds of washing with PBS-T, the signal was developed by the chromogenic HRP-substrate 3,3′,5,5′-tetramethylbenzidine (TMB) (Sigma-Aldrich, T8665), stopped by 3 N HCl and the absorbance detected at 450 nm emission wavelength using the Ensight® multimode plate reader (PerkinElmer). AAT concentrations were determined by fitting the absorbance values of the AAT calibration curve to a sigmoidal equation and interpolating the absorbance signal of the samples. The resulting secreted AAT was normalized for the total protein content of the respective cell supernatant, determined by QuantiPro BCA assay (Sigma-Aldrich, QPBCA).

Measurement of secreted AAT polymers by sandwich ELISA

Quantification of AAT polymers in cell supernatants was performed similarly to what reported before [18, 28, 58]. We adapted the same methodology reported for total AAT quantification by sandwich ELISA, except for the following modifications. Coating of the plates was performed with 2 μg/mL mouse monoclonal antibody anti-AAT polymer (2C1 mAb) [38] in PBS, overnight at 4 °C. A polymeric AAT standard was generated by heat treatment of purified plasma M-AAT (Athens Research & Technology, 16-16-011609), according to previously published methods [59], and was utilized to generate a 11-points calibration curve in dilution buffer, starting from 500 ng/mL and performing 1:2 serial dilutions. Dilutions of supernatants were prepared in dilution buffer (1:3 and 1:6).

Detection of secreted AAT by western blot

Equal volumes of cell supernatants were separated by either SDS-PAGE in reducing conditions, utilizing 4–12% Bolt™ Bis–Tris Mini Protein Gels (Thermo Fisher Scientific, NW04125) in Bolt™ MES running buffer (Thermo Fisher Scientific, B0002), or by non-denaturing-PAGE, utilizing Novex™ Value™ 4–12% Tris–Glycine Mini Protein Gel (Thermo Fisher Scientific, XV04125) resolved in anode buffer (0.1 M Tris–HCl, pH 7.8) and cathode buffer (0.1 M Tris–HCl, 68 mM Glycine pH 8.9). Proteins were transferred to Hybond-P 0.45 PVDF membrane (GE Healthcare, 10,600,023), and incubated with 5% defatted milk in TBS containing 0.05% Tween-20 (TBS-T), for 1 h at room temperature. AAT was detected by incubating with 2 μg/mL rabbit polyclonal antibody anti-AAT (Dako, A001202), overnight at room temperature, followed by incubation with 100 ng/mL HRP-conjugated donkey antibody anti-rabbit IgG (Cytiva, NA9340) in TBS-T, for 2 h at room temperature. After three rounds of washing with TBS-T, membranes were developed with SuperSignal™ West Pico PLUS Chemiluminescent Substrate (Thermo Fisher Scientific, 34,578) and recorded with the Odyssey Fc imager (LI-COR Biosciences).

Detection of AAT–NE complexes by western blot

For each secreted AAT variant, cell supernatant aliquots containing 20 nM AAT were incubated with increasing concentrations of human NE (Athens Research & Technology, 16-14-051200) (0-10-20–30 nM) in 0.05% NP-40, 0.5 M NaCl (PBS-N), for 1 h at room temperature. Reaction mixtures were resolved by SDS-PAGE in reducing conditions and AAT was detected by western blot, as described above.

Determination of stoichiometry of NE inhibition

For each secreted AAT variant, increasing volumes of supernatant (0.4-0.7-1.0-1.5-2.2-3.3-5.0-7.5 nM AAT) were mixed with human NE (5 nM) in 150 µL/well of the reaction buffer PBS-N, utilizing a non-binding 96-well microplate (Greiner Bio-One, 655901), incubated for 1 h at room temperature. The chromogenic substrate MeO-Suc-Ala-Ala-Pro-Val-pNA (Fisher Scientific, 50260400) was diluted to 1.2 mM in PBS-N and 50 µL/well added to a final 400 μM concentration. Substrate hydrolysis was performed at 25 °C and it was monitored by measuring absorbance at 410 nm once every minute for 30 min, using the Ensight® multimode plate reader (PerkinElmer). For each AAT:NE molar ratio the initial rate of substrate hydrolysis (V₀) was determined by fitting the absorbance at 410 nm versus time curves and by calculating the respective slope value. Plots of V₀ versus AAT:NE molar ratios were then fitted to a linear regression and the X-axis intercept extrapolated to obtain the SI value of each AAT variant.

Elastase complex formation Immunosorbent assay (ECFISA)

The ECFISA method was adapted from Weber and colleagues [40]. High-binding half-area 96-well microplates (Costar, 3690) were coated with 20 μg/mL PPE (Sigma, E7885) in PBS, overnight at 4 °C, washed three times with PBS-T, blocked with 0.25% BSA in PBS-T, for 1 h at room temperature, and washed with PBS-T. Supernatants in BSA-containing dilution buffer (2 nM AAT) were applied to the PPE-coated plates, for 1 h at room temperature, for AAT capture. After three rounds of washing with PBS-T, detection of AAT was carried out by incubation with 67 ng/mL HRP-conjugated goat polyclonal antibody anti-AAT (Abcam, ab191350) in BSA-containing dilution buffer, for 1 h at room temperature. Unbound detection antibody was removed by three rounds of washing with PBS-T. The signal was developed by TMB incubation, stopped by 3 N HCl and absorbance measured at 450 nm emission wavelength using Ensight® multimode plate reader (PerkinElmer).

Detection of AAT–NE complexes by sandwich protease complex detection ELISA (PCD-ELISA)

High-binding 96-well microplates (Costar, 3690) were coated with 4 μg/mL mouse monoclonal antibody anti-AAT (3C11 mAb) [41, 42] in PBS and incubated overnight at 4 °C, washed with PBS-T, blocked with 0.25% BSA in PBS-T, for 1 h at room temperature, and washed again with PBS-T. In parallel, the supernatant of each AAT variant (2 nM AAT) was incubated with equimolar human NE in solution for 1 h at room temperature. Reaction mixtures were applied to the 3C11 mAb-coated plate and incubated for 1 h at 37 °C for AAT (unbound and NE-complexed) capture. After three rounds of washing with PBS-T, detection of NE was carried out by incubation with 200 ng/mL rabbit polyclonal antibody anti-NE (Athens Research & Technology, 01-14-051200) in BSA-containing dilution buffer, for 1 h at 37 °C. HRP-conjugated donkey antibody anti-rabbit IgG (Cytiva, NA9340) was applied at a concentration of 67 ng/mL in dilution buffer. Unbound antibodies were removed by three rounds of washing with PBS-T. The signal was developed by TMB incubation (Sigma-Aldrich, T8665), stopped by 3 N HCl and absorbance measured at 450 nm emission wavelength using Ensight® multimode plate reader (PerkinElmer).

Determination of NE inhibition rate constants

The supernatant of each AAT variant was mixed concomitantly with human NE and its fluorogenic substrate, while immediately monitoring the generated fluorescent signal over time. A pseudo-first order condition was achieved using a 7.5-fold molar excess of active secreted AAT over NE adjusted according to the SI value of each variant, ensuring complete NE inhibition during the reaction. Importantly, the substrate of NE was utilized in large excess, corresponding to 2.5-fold the Michaelis–Menten constant of NE determined in our experimental conditions (K_M = 293 µM), aiming to slow down the reaction and allow its measurement [45]. The enzymatic reaction was set up in black non-binding 96-well microplates (Greiner Bio-One, 655076) by mixing human NE (0.25 nM), fluorogenic substrate of NE (0.75 mM) Meo-Suc-Ala-Ala-Pro-Val-AMC (Cayman chemicals, 14907), and the supernatant of each AAT variant (7.5-fold AAT molar excess normalized for the respective SI) in 150 μL/well PBS-N. Hydrolysis of the substrate was performed at 25 °C, and the fluorescence signal was monitored by excitation 367 nm and emission 460 nm wavelengths, every 5 s for 30 min, using the Ensight® multimode plate reader (PerkinElmer). For each AAT variant, the k_obs of inhibition under pseudo-first order conditions was extrapolated by plotting progress curves of fluorescence signal versus time, and fitting data to Eq. (1). Consequently, k_inh values were calculated with Eq. (2) according to [45].

$$F=V_{s}t+\frac{{(V}_i-V_s)\left(1,-,e^{-k_{\mathrm{obs}}t}\right)}{k_{\mathrm{obs}}}+\mathrm{baseline}$$ 1

F fluorescence units; V_i initial rate; V_s final rate; t time; k_obs observed rate constant.

$$k_{\mathrm{inh}}=k_{\mathrm{obs}}\bullet \frac{\left(1,+,\frac{\left[S\right]}{k_M}\right)}{\left[\mathrm{AAT}\right]}$$ 2

k_inh second-order inhibition rate constant; k_obs observed rate constant under pseudo-first order conditions; [S] molar concentration of substrate; k_M Michaelis–Menten constant of NE substrate hydrolysis; [AAT] molar concentration of AAT.

Statistics

As indicated in the text, data are reported as mean ± SD of replicates from independent transfection experiments and analysis of significant differences was performed by one-way ANOVA corrected with Dunnett’s post hoc test. Fittings to linear or non-linear equations and interpolations were performed with 95% confidence interval, and they were considered correlated if R² ≥ 0.9. Statistical analyses were performed using GraphPad Prism 8 software.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

AAT

Alpha-1-antitrypsin

AATD

Alpha-1-antitrypsin deficiency

ELISA

Enzyme-linked immunosorbent assay

NE

Neutrophil elastase

PPE

Porcine pancreatic elastase

RCL

Reactive center loop

SI

Stoichiometry of inhibition

Author contributions

AD and AF designed the research project; AD, EBK, MB, RG, EM, ED, and AF performed experiments; AD and JAI performed structural analysis; AD, EBK, MB, EM, JAI, and AF analyzed data; AD, JAI, and AF wrote the paper. All authors read and approved the manuscript.

Funding

This work was mainly supported by a grant from the Alpha-1 Foundation USA (ID:829920) to AF. JAI is supported by the Medical Research Council UK (MR/V034243/1) and the Alpha-1 Foundation (ID: 1036784). EM was supported by the Alpha-1 Foundation (ID: 497841).

Data availability

Supporting data are included as electronic supplementary material.

Declarations

Conflict of interest

The authors have no relevant financial or non-financial interests to disclose.

Ethical approval

This study does not involve human subjects or animals.