Engineered IgM and IgG cleaving enzymes for mitigating antibody neutralization and complement activation in AAV gene transfer

Abstract

Systemic dosing of adeno-associated viral (AAV) vectors poses potential risk of adverse side effects including complement activation triggered by anti-capsid immunity. Due to the multifactorial nature of toxicities observed in this setting, a wide spectrum of immune modulatory regimens are being investigated in the clinic. Here, we discover an IgM cleaving enzyme (IceM) that degrades human IgM, a key trigger in the anti-AAV immune cascade. We then engineer a fusion enzyme (IceMG) with dual proteolytic activity against human IgM and IgG. IceMG cleaves B cell surface antigen receptors and inactivates phospholipase gamma signaling in vitro. Importantly, IceMG is more effective at inhibiting complement activation compared with an IgG cleaving enzyme alone. Upon IV dosing, IceMG rapidly and reversibly clears circulating IgM and IgG in macaques. Antisera from these animals treated with IceMG shows decreased ability to neutralize AAV and activate complement. Consistently, pre-conditioning with IceMG restores AAV transduction in mice passively immunized with human antisera. Thus, IgM cleaving enzymes show promise in simultaneously addressing multiple aspects of anti-AAV immunity mediated by B cells, circulating antibodies and complement. These studies have implications for improving safety of AAV gene therapies and possibly broader applications including organ transplantation and autoimmune diseases.

Graphical abstract

Smith and colleagues develop newly engineered immunoglobulin cleaving enzymes to address immune challenges in clinical gene therapy. A dual activity enzyme, dubbed IceMG, clears neutralizing antibodies and blocks complement activation in nonhuman primates. Potential translational applications include improving safety of AAV gene therapies and possibly in organ transplantation and autoimmune disease.

Introduction

Recombinant adeno-associated viral (AAV) vectors have emerged as the leading platform for therapeutic gene transfer in the past decade.^1,2 Four AAV-based products are approved by the FDA for systemic dosing in patients—Zolgensma for spinal muscular atrophy, Hemgenix for hemophilia B, Roctavian for treating hemophilia A, and Elevidys for treating Duchenne muscular dystrophy in children ages 4–5 years.^3,4,5,6 Furthermore, over 50 AAV-based gene therapy clinical trials requiring systemic dosing are registered at ClinicalTrials.gov as of November 2023. Despite the progress, immune responses arising from de novo exposure to high AAV vector levels and pre-existing immunity to AAV capsids have posed significant toxicity risks to patients.^7,8,9,10 Of particular note, activation of the complement cascade triggered by AAV capsids complexed to anti-capsid IgM and IgG antibodies has been linked to severe and life-threatening inflammatory responses leading to cardiovascular, hepatic and renal damage and in some cases, death. Specifically, high intravenous AAV vector doses in the clinic have been associated with thrombotic microangiopathy (TMA), acute kidney injury, thrombocytopenia and immune-mediated myocardial injury; all of which have been linked to complement activation via anti-AAV capsid antibodies.^11,12,13,14

Mitigation of AAV-mediated immunotoxicity in the clinic has involved immune modulatory regimens (IMRs) based on glucocorticoids, Rituximab for depleting B cell subsets to attenuate antibody production over time, Sirolimus for T and B cell activation, plasmapheresis and IgG clearing agents.^15,16,17 Acute and life-threatening clinical sequelae have been managed through hemodialysis, platelet transfusion and blocking downstream elements of the complement cascade (e.g., eculizumab for C5 inhibition).¹⁸ However, with growing understanding of the mechanisms underlying complement-mediated AAV vector toxicity, there is an urgent need for more effective and targeted immunomodulatory agents to improve the safety of clinical AAV gene transfer.

Within this framework, monomeric IgM is the first antibody to emerge as the B cell antigen receptor (BCR) on the surface of B cells followed by secreted pentameric IgM from plasma cells. Class switching upon B cell activation generates different isotypes or subtypes of IgG, which constitute the most abundant, secreted class of antibodies.¹⁹ In addition to the risks outlined above leading to AAV immunotoxicity, recognition of self-antigens can sometimes lead to the generation of harmful IgG and IgM autoantibodies leading to autoimmune disease.²⁰ Similarly, antibodies directed against human leukocyte antigens expressed on the surface of cells in a transplanted organ can lead to immune rejection.²¹ Importantly, activation of the complement cascade leading to significant tissue damage is a shared hallmark in these distinct clinical scenarios.^22,23 However, exactly how mitigation of complement-mediated toxicity can be effectively achieved by addressing the relative contributions of IgM and IgG is not well understood.

Recent efforts have focused on developing human IgG degrading enzymes (e.g., IdeS, IdeZ) for immune modulation, but largely from a pre-existing immunity standpoint.^24,25,26,27 Clinical applications for these reagents to date includes desensitization of kidney transplant patients with high levels of anti-donor antibodies, while preclinical development for AAV gene transfer applications is currently being explored. In contrast, selective modulation of IgM has been intractable due to lack of appropriate agents. We reasoned that the advent of microbial genomics efforts, and the abundance of sequenced genomes provides an opportunity to mine bacterial enzymes through phylogenetic analyses and structural modeling. Using robust molecular docking tools, we further refined prospective enzyme candidates that may recognize IgM constant domains as substrates. Altogether, we were able to systematically construct and engineer IgM cleaving enzymes that enabled direct interrogation of the relative contribution(s) of IgM vs. IgG in complement activation. Using human antisera, rhesus macaques, and passive immunization mouse models, we demonstrate that IgM cleaving enzymes can robustly (and reversibly) decrease circulating antibody levels, block the complement cascade and reduce BCR expression. In addition, an engineered dual activity, enzyme fusion protein confirms that the impact of IgG cleavage is more moderate compared with IgM, albeit cumulative in helping mitigate complement activation. Our study emphasizes the untapped potential of IgM cleaving enzymes in expanding the class of immunomodulatory agents beyond IgG specific agents for addressing AAV vector-mediated immunotoxicity and more broadly for mitigation of autoantibody-driven autoimmune diseases and antibody-mediated rejection of organ transplants.

Results

Identifying and characterizing a human IgM protease

Bacteria have evolved specific enzymes such as the IgG cleaving enzyme, IdeS from S. pyogenes, to evade opsonization and immune recognition.²⁸ These secreted enzymes belong to the papain-like cysteine protease superfamily and selectively cleave IgG at the Fc hinge region.²⁹ Another member of this cysteine protease family, Ide_Ssuis, selectively cleaves porcine IgM (Figure S1A) as an endopeptidase with subsequent impact on complement activation and B cell signaling.^30,31,32 To identify a human IgM protease, we utilized NCBI BLAST to phylogenetically analyze and select bacteria found within the human microbiome that express novel papain-like proteases. The core protease domain (Figure 1A) was identified and engineered to exclude non-essential protein elements such as cell wall binding and excretion motifs to enhance recombinant protein expression in E. coli. Of the candidates tested, one protein from Lachnoanaerobaculum saburreum (Protein 2; NCBI Taxonomy ID: 467210) was identified as a human IgM cleaving enzyme (referred to herein as IceM) that produced cleavage products consistent with separation of Cμ3 and Cμ4 domains from the remainder of IgM heavy chain (Figures 1B and S1B). To further characterize enzymatic activity, recombinant IceM expression in E. coli and purification by Nickel column affinity resin was confirmed via SDS-PAGE (Figure S1C). Treatment of purified human IgM with IceM generated two cleavage fragments at 41 and 32 kDa indicating hydrolysis of the IgM heavy chain at the interdomain region between constant domains Cμ2 and Cμ3, consistent with predictive structural modeling (Figures 1B–1D and S1D). IceM efficiently and specifically cleaved human IgM with an EC₅₀ of ∼0.16 nM, but did not cleave isotypes IgG, IgA, IgE, or IgD (Figures 1E and 1F). Furthermore, IceM cleaves human and rhesus macaque IgM and shows minimal cross-reactivity with dog IgM in vitro. We did not observe any cross-reactivity for IgM from other species such as pig or mouse (Figure 1G).

Figure 1.

Structural model and in vitro characterization of a human IgM cleaving enzyme

(A) Representative structural model of the core domain of cysteine protease candidates with active site in orange. (B) Cartoon schematic illustrating IgM and the molecular weights of full-length heavy chain (73 kDa) and predicted F(ab) (41 kDa) and Fc (32 kDa) cleavage fragments outlined in red. (C) SDS-PAGE gel of purified human IgM, IceM, or IgM treated with 20 μg/mL IceM for 1 h at 37°C under reducing conditions. (D) Structural docking analysis of IceM in complex with human IgM heavy chain with active site in orange. (E) Concentration curve for IceM enzymatic activity in human sera. (F) Immunoblot analyses showing the banding patterns of IgM, IgA, IgD, IgE, or IgG from human sera. The detection antibody recognizes an epitope on the Fc of each antibody and leads to banding patterns that occur at either full-length heavy chain or the Fc fragment after cleavage. (G) Immunoblot analyses showing the banding patterns IgM from sera across species treated with PBS or 20 μg/mL IceM for 1 h at 37°C. ∗ indicates full-length heavy chain (IgM, ∼73 kDa; IgG, ∼53 kDa; IgA, ∼55 kDa; IgE, ∼73 kDa; IgD, ∼63 kDa) and arrows indicate Fc cleavage fragments. Error bars denote mean ± standard deviation. All experiments were completed in at least biological triplicate. Analyses were performed with GraphPad Prism v.9.5.

Recently the structure of IdeS, another highly specific IgG cleaving papain-like protease superfamily member, was determined in complex with human IgG and demonstrated that human IgG binds to a major pocket adjacent to the enzyme active site.³³ We next sought to elucidate the high specificity of IceM for IgM. Utilizing AlphaFold2, we predicted structures of IceM in complex with a single human IgM heavy chain. Notably, 17/20 folding attempts resulted in the IgM Cμ3 domain docking IceM within a major binding pocket homologous to the binding pocket described previously for IdeS/IgG (Figure S1E). Further analysis of two IgM heavy chains complexed to IceM confirmed overlap with similarly positioned contact residues within the IdeS/IgG binding pocket (Figures S1F and S1G). Thus, the specificity of IceM for human IgM is likely due to differences in amino acid composition of the major binding pocket adjacent to the cleavage site for specific immunoglobulins.

Engineering a dual function IgM and IgG degrading enzyme

Next, we sought to engineer a dual IgM and IgG degrading enzyme through the fusion of IceM and IdeZ proteolytic domains. To link the catalytic domains, we designed multiple linkers varying in length and flexibility to ensure optimal protein expression and catalytic activity (Figure 2A). Plasmids for three different fusion protein linkers (rigid linker, flexible linker 1, and flexible linker 2) were each constructed and expressed using E. coli. After incubation, bacteria were collected, lysed, and pre- and post-induction samples were analyzed by western blot for expression. Significantly higher protein expression and stability for the rigid linker compared with the flexible linkers was observed (Figures 2A and 2B). Enzyme-substrate docking analyses indicated the rigid fusion protein, hereinafter termed IceMG (IgM and IgG cleaving enzyme), maintained the ability to tightly bind human IgM and IgG heavy chain within their respective catalytic pockets (Figure 2C). Recombinant IceMG expression in E. coli and purification by Nickel column affinity resin was confirmed via SDS-PAGE (Figure S1H). Treatment of purified human IgM with IceMG generated two cleavage fragments at 41 and 32 kDa. These cleavage fragments are consistent with hydrolysis of the IgM heavy chain at the interdomain region between constant domains Cμ2 and Cμ3 and with predictive structural modeling of previous IceM cleavage analysis (Figures S1I and S1J). Treatment of purified human IgG with IceMG generated cleavage fragments at ∼25 kDa. These fragments run slightly above light chain and indicates hydrolysis at a single site below the hinge region, yielding F(abʹ)₂ and Fc fragments, consistent with prior reports of IdeZ cleavage of IgG (Figures 2D, 2E, and S1K).²⁴ Treatment of IceMG efficiently and specifically cleaved IgM and IgG with an EC₅₀ of ∼1.8 and 35 nM, respectively, but did not cleave isotypes IgA, IgE, IgD (Figures 2F and 2G). Furthermore, IceMG efficiently cleaves human and rhesus macaque IgM and IgG and has minimal cross-reactivity with dog IgM and mouse IgG in vitro. No cross-reactivity was observed for pig IgM or IgG (Figure 2H).

Figure 2.

Engineering a dual function IgM and IgG degrading enzyme

(A) (Top) Structural modeling of IceM fusion proteins with three different linkers (rigid linker, flexible linker 1, and flexible linker 2 with corresponding linker amino acid sequence beneath the structure. Models are colored by pLDDT. (Bottom) Immunoblot analyses for his-tagged fusion protein expression with a theoretical molecular weight of 70.4 kDa. The fusion protein with the rigid linker, hereinafter termed IceMG, showed the superior expression and was chosen for further evaluation. (B) Structural model of the dual cleaving enzyme IceMG with IgM (blue) and IgG (red) catalytic domains flanking the rigid helical linker (cyan). (C) Structural docking analysis of IceMG in complex with human IgM and human IgG heavy chain colored as described in (B). (D) Cartoon schematic illustrating IgG and the molecular weights of full-length heavy chain (50 kDa) and predicted F(ab) (25 kDa) and Fc (25 kDa) cleavage fragments outlined in red. The cleavage products run slightly above light chain (23 kDa). (E) SDS-PAGE gel of purified human IgG, IceMG, or IgG treated with 20 μg/mL IceMG for 1 h at 37°C under reducing conditions. (F) Concentration curve for IceMG enzymatic activity on IgM and IgG in human sera. (G and H) Immunoblot analyses showing the banding patterns of IgM, IgG, IgA, IgE, or IgD from human sera (G) and IgM or IgG from sera across species (H) treated with PBS or 20 μg/mL IceMG for 1 h at 37°C. The detection antibody recognizes epitopes on the Fc of each antibody. ∗ indicates full-length heavy chain (IgM, ∼73 kDa; IgG, ∼53 kDa; IgA, ∼55 kDa; IgE, ∼73 kDa; IgD, ∼63 kDa) and arrows indicate Fc cleavage fragments. Error bars denote mean ± standard deviation. All experiments were completed in at least biological triplicate. Analyses were performed with GraphPad Prism v.9.5.

IceM and IceMG cleave the IgM B cell receptor and inactivate phospholipase-γ2 signaling in vitro

We assessed the ability of IceM and IceMG to remove surface-bound monomeric IgM (BCR) from B cells. IceM and IceMG robustly remove IgM BCR from the surface of the CD19⁺ sub-population in human peripheral blood mononuclear cells (PBMCs) at 1 h post-treatment (Figures 3A, 3B, and S2A). Importantly, surface IgM baseline levels were restored between 12 and 24 h post-treatment indicating that BCR inactivation is transient and reversible (Figures 3A and 3B) with no adverse impact on cell viability after 24 h (97.9%, 97.8%, 92.5%) for PBS, IceM, and IceMG treated PBMCs, respectively (Figure 3C). Next, we investigated if IceM or IceMG could impact IgM BCR signaling in the human B lymphoblast Daudi cell line. Antigen binding to the BCR triggers a signaling cascade and activation of signaling intermediates such as PLC-γ2 that influence B cell fate including B cell activation, proliferation, and differentiation. We assessed the activation of PLC-γ2 following stimulation with anti-Ig as a surrogate antigen. We confirmed the removal of surface-bound IgM from Daudi cells by flow cytometric analyses, immunoblot, and fluorescent microscopy (Figures 3D, 3E, and S2B–S2D). Daudi cells were unable to signal through the IgM BCR following treatment with IceM or IceMG as demonstrated by the absence PLC-γ2 phosphorylation (Figures 3F and 3G).

Figure 3.

In vitro assessment of recombinant IceM and IceMG on the IgM B cell receptor

(A) Representative plots for flow cytometric analysis of IgM surface expression on CD19⁺ B cells (BCRs) following treatment with PBS, 20 μg/mL IceM or IceMG at 1, 12, and 24 h post-treatment. Data are gated on live cells and then CD19⁺ (APC⁺) to analyze the IgM⁺ (FITC⁺) cohort. (B) Quantitation of percent IgM⁺ B cells from (A). (C) Quantitation of percent viability for the total lymphocyte population 1, 12, and 24 h post-treatment with IceM or IceMG. (D) Representative plots for flow cytometric analysis of IgM surface expression on Daudi cells following treatment with PBS, 20 μg/mL IceM or IceMG for 1 h. Data are gated to analyze the IgM⁺ (FITC⁺) cohort. (E) Quantitation of percent IgM⁺ Daudi cells. (F) Immunoblot analyses of Daudi cells stimulated with anti-Ig for 10 min following pre-treatment with PBS, 20 μg/mL IceM or IceMG for 1 h. Antibodies against phospho-PLC-γ2/PLC-γ2 (140 kDa) were used to assess B cell activation. Antibodies against IgM (73 kDa) and Tubulin (50 kDa) were used to confirm IgM BCR cleavage and equal loading, respectively. (G) Quantitation of phospho-PLC-γ2/PLC-γ2 immunoblots from (F). Data were normalized to untreated, unstimulated control. Error bars denote mean ± standard deviation. Flow experiments were completed in biological triplicate. B cell activation experiments were completed in duplicate. Statistical significance was determined using two-way ANOVA with Šidák’s correction for multiple comparisons (B, C, and G) or ordinary one-way ANOVA with Dunnett’s correction for multiple comparisons (E). Only p values less than 0.05 were considered significant and shown in each graph. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

IceM and IceMG are potent inhibitors of complement activation in human serum

IgM is known to be a potent activator of the classical complement pathway that is triggered by the binding of C1q followed by activation of the C1 complex.²³ Therefore, we assessed the ability of IceM and IceMG to block and/or mitigate complement activation. Using an in vitro platelet factor-based assay³⁴ mimicking heparin-induced thrombocytopenia to trigger IgM-mediated classical complement pathway activation (Figure 4A), we determined that pre-treatment of human sera with IceM and IceMG blocked the complement cascade at levels comparable to baseline (Figure 4B). We then used AAV9 in conjunction with human serum samples containing anti-AAV9 neutralizing antibodies (NAbs) to trigger complement (Figure 4A). As expected, human donor serum containing anti-AAV9 NAbs exhibited significantly higher complement activation by AAV9 capsids compared with the seronegative cohort (Figures 4C–4E) consistent with recent studies demonstrating clinical adverse events due to capsid antibody-mediated complement activation.¹⁴ Pre-treatment with IceM or IdeZ significantly reduced C3a levels whereas pre-treatment with IceMG completely blocked the complement cascade to levels comparable to baseline. Notably, the partial attenuation in C3a levels by the IgG degrading enzyme, IdeZ and IceM (relative to IceMG) underscores the concept that while IgM is generally a more potent trigger of complement activation, both anti-capsid IgM and IgG antibodies are major contributors to complement activation. These unique properties of IceM and IceMG corroborate the notion that selective IgM and IgG cleavage can potentially be leveraged to modulate classical complement pathway activation.

Figure 4.

In vitro assessment of recombinant IceM and IceMG on complement activation

(A) Overview schematic illustrating the platelet factor- or AAV9-based complement activation assay. (B) ELISA-based quantitation of C3a levels in human sera from 10 individual donors pre-treated with PBS, 20 μg/mL IdeZ, IceM, or IceMG for 1 h at 37°C to cleave IgM and/or IgG followed by incubation with PF4/heparin complexes to trigger complement activation. EDTA inactivates complement and gives a snapshot of complement protein levels prior to experimental incubations. (C) Neutralization profiles for AAV9 in human antisera. Serum was serially diluted and coincubated with AAV9 in vitro to generate sigmoidal neutralization curves that were normalized to controls containing no antisera. The dotted line represents neutralizing antibody-mediated inhibition of AAV9 transduction by 50% (ID50). Donors with neutralizing antibody titers ≥1:4 (1090947, 1090950, 1090944, 1090946) were considered seropositive. (D and E) ELISA-based quantitation of C3a levels in AAV9 NAb− (D) or AAV9 NAb+ (E) human sera pre-treated with PBS, 20 μg/mL IdeZ, IceM, or IceMG for 1 h at 37°C to cleave IgM and/or IgG followed by incubation with AAV9 to trigger complement activation. Error bars denote mean ± standard deviation. Statistical significance was determined using ordinary one-way ANOVA (B) or two-way ANOVA (D and E) with Tukey’s correction for multiple comparisons. EDTA negative control samples were omitted from the statistical analyses. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; ns, not significant.

IceMG enables robust and transient clearance of circulating IgM and IgG macaques in vivo

Next, we evaluated IceMG in rhesus macaques in vivo. Rhesus macaques (R1, R2, and R3) were dosed at 0.25, 1.0, or 2.5 mg/kg IceMG, respectively, followed by monitoring of the kinetics of IgM and IgG clearance at indicated time intervals (Figure 5A). Pharmacokinetics of IceMG in monkey sera revealed rapid clearance from blood circulation, with detectable levels up to 24 h post-administration (Figures S3A–S3C). Specifically, a majority of dosed IceMG was cleared between 6 and 24 h post-administration. Animals exhibited significant reduction in total circulating IgM across all doses with IgM rebounding to baseline levels by day 14 post-treatment (Figure 5B). Strikingly, near complete IgM cleavage by 1 h post-treatment was confirmed by western blot at all doses (Figures 5D and S3D–S3F). A dose-dependent reduction in total circulating IgG was observed with 2.5 mg/kg exhibiting the greatest reduction (Figure 5C). IgG cleavage was confirmed by western blot, however, a portion of uncleaved IgG remained detectable at all time points (Figures 5E and S3G–S3I). To investigate why only partial cleavage of IgG was observed in nonhuman primates, we compared the enzymatic activity of IceMG on human and nonhuman primate sera. IceMG was less efficient against nonhuman primate IgG consistent with previous studies that observed similar conclusions for the IgG cleaving enzyme IdeS (Figure S3J).^25,35

Figure 5.

IceMG clears circulating IgM and IgG in rhesus macaques

(A) Overview of experimental design for the NHP study. Three rhesus macaques were treated with 0.25 mg/kg (R1), 1.0 mg/kg (R2), or 2.5 mg/kg (R3) and serum samples were collected at the time points indicated. (B and C) ELISA-based quantitation of total IgM (B) or IgG (C) from day 0 (pre-treatment) to 14 days post-treatment. (D and E) Quantitation of circulating IgM (D) or IgG (E) cleavage assessed by immunoblot in each animal relative to serum samples collected prior to IceMG administration. The first 24 h post-treatment are shown in the insets. Data are presented as n = 1 animal per dose and standard deviation across technical replicates.

Antisera from IceMG-treated macaques shows reversible depletion of anti-AAV NAbs and attenuated complement activation in vitro

As natural hosts, nonhuman primates often have preexisting humoral immunity to AAV. We investigated whether the macaques in this study had antibodies against AAV and how IceMG treatment effected the ability of these antibodies to neutralize AAV transduction and activate complement in vitro. We confirmed the presence of anti-AAV9 IgM and IgG in antisera from all three animals that, upon treatment with IceMG, were significantly reduced in a dose-dependent fashion (Figures 6A and 6B). Animals exhibited anti-AAV9 neutralizing titers of 1:49, 1:21, and 1:17 for R1 (0.25 mg/kg), R2 (1.0 mg/kg), and R3 (2.5 mg/kg), respectively, that correlated with their anti-AAV9 IgG levels (Figures 6B, 6C, and S4A–S4E). We observed no change in neutralizing titers for the animal that received 0.25 mg/kg dose, 2-fold reduction in neutralization at 1.0 mg/kg, and 6-fold reduction in neutralization at 2.5 mg/kg that peaked between 24 and 72 h post-treatment and returned to pre-treatment levels between day 7 and 14 (Figures 6C and S4A–S4E).

Figure 6.

In vitro assessment of antisera from IceMG-treated macaques for AAV neutralizing antibodies and complement activation

(A and B) ELISA-based quantitation of circulating anti-AAV9 IgM (A) or IgG (B) in each animal from day 0 (pre-treatment) to 14 days post-treatment. (C) Neutralizing antibody titers for NHP serum samples. Serum samples at each time point were serially diluted from 1:2 to 1:2,048 and coincubated with AAV9 in vitro to generate sigmoidal neutralization curves that were normalized to controls containing no antisera. The serum concentration required to inhibit AAV9 transduction by 50% (ID50) was determined at each time point for each dose and plotted. (D) Overview of the complement activation assay. (E and F) ELISA-based quantitation of complement factor B (E) and C3a (F) levels in serum samples from each animal following in vitro treatment with AAV9 to elicit complement activation. EDTA was used at a negative control. Untreated serum samples (no AAV9) were incubated alongside treated samples and considered baseline. The first 24 h post-treatment are shown in the insets. Data are presented as n = 1 animal per dose and standard deviation across technical replicates.

Next, using antisera at each time point from animals R1, R2, and R3, we next evaluated the impact of IceMG treatment on complement activation by AAV9 capsids in vitro (Figure 6D). In agreement with recent studies, we observed moderate activation of the alternative pathway demonstrated by a 72%, 67%, and 64% decrease from baseline in complement factor B levels for animals R1, R2, and R3, respectively, that were unaffected by antibody clearance by IceMG (Figure 6E). Conversely, we observed a dose-dependent reduction in activation of the classical complement pathway by 65%, 85%, or 100% for animals R1, R2, and R3, respectively. Complement activation by AAV9 was reduced from 2 to 7 days post-treatment concomitant with IgM rebound (Figure 6F). While preliminary, these results suggest that 24–72 h prior to vector administration may be well suited for pre-treating seropositive subjects with IceMG (in conjunction with other immunomodulatory agents), followed by AAV vector dosing.

IceM and IceMG clear NAbs and restore AAV transduction in passively immunized mice

As initial proof-of-concept, we first evaluated Ig clearance in a mouse model in vivo, First, adult mice were passively immunized with pooled human sera and administered IceM or IceMG intravenously at doses of 0.25, 1.0, or 2.5 mg/kg (Figure S5A). Rapid (<24 h) and near-complete clearance of circulating human IgM or IgG at all doses compared with control animals for IceM (Figures S5B–S5E) and IceMG (Figures S5F–S5I) was observed. Based on these results, we determined whether pre-treatment with IceM or IceMG could rescue AAV transduction in mice passively immunized with human sera and IVIg containing anti-AAV NAbs in vivo (Figures 7A, 7B, and 7E). In animals treated with only sera/IVIg, an ∼1 log decrease in AAV8- (Figures 7C and 7D) and AAV9- (Figures 7F and 7G) mediated luciferase expression in heart and liver tissue was observed. In mice treated with IceM, a partial restoration of AAV transduction, and in mice treated with IceMG or IdeZ, a near-complete rescue of luciferase expression in heart and liver tissue, was observed. While these results suggest that high levels of neutralizing IgG can be rate limiting in AAV transduction, the previously under-appreciated impact of IgM is also exemplified. Furthermore, we observed serotype-specific contributions to anti-capsid IgM in these studies. Notably, AAV8 transduction was less impacted by IgM cleavage (0% and 28% rescue for heart and liver, respectively) compared with AAV9 vectors (72% and 47% rescue for heart and liver, respectively). These differences may indicate serotype-specific attributes of immune recognition and memory, which in turn could impact relative levels of neutralizing IgG vs. IgM. In addition, despite the inability to model complement inactivation in mice, the importance of addressing anti-AAV IgM is evident when these results are considered together with our earlier observation pertaining to complement inactivation upon cleaving IgM.

Figure 7.

IceM and IceMG rescue AAV transduction in passively immunized mice

(A) Overview of experimental design for AAV mouse study. In brief, mice were passively immunized with human sera and IVIg containing anti-AAV NAbs. Twenty-four hours later mice were treated with PBS, 1.0 mg/mL IdeZ, IceM, or IceMG followed by injection of AAV8- or AAV9-Luc 24 h later. (B and E) Human sera or IVIg were diluted 1:20 and evaluated for neutralizing antibodies against AAV8 (B) or AAV9 (E). (C, D, F, and G) Luciferase expression was evaluated 21 days post-injection in heart and liver tissue for AAV8 (C and D) and AAV9 (F and G). Luciferase expression levels were normalized for total tissue protein concentration and represented as log relative luminescence units per gram of tissue (log RLU/g tissue). Error bars denote mean ± standard deviation. Statistical significance was determined using one-way ANOVA with Tukey’s correction for multiple comparisons. Only p values less than 0.05 were considered significant and are shown in each graph. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Discussion

In this study, we describe new immunomodulatory agents derived from highly specific endoproteases that can rapidly cleave human and rhesus IgM or both IgM and IgG. Our data demonstrate the ability to enzymatically remove IgM BCR from human B lymphocytes in vitro and degrade circulating human IgM in vivo in mice as well as primate IgM in rhesus macaques. Notably, pre-treatment with IceM and IceMG significantly reduced complement activation in human sera, which may be of therapeutic benefit for multiple applications.

The primary conceptual advance described in the current study is the application of IgM degradation in AAV-mediated gene therapy. A major challenge that prevents the recruitment of patients for gene therapy clinical trials and poses acute immunotoxicity risk in AAV-based gene therapies is the prevalence of pre-existing neutralizing anti-AAV NAbs.³⁶ Patient screening in the clinic using enzyme-linked immunosorbent assay (ELISA) or NAb assays is commonplace with exclusion criteria that can render ∼50% or more patients ineligible for treatment.^37,38 More critically, recent clinical trials have shown that antibody response and complement activation can lead to serious adverse events including TMA, acute hemolytic uremia syndrome, renal failure and in some cases death, particularly in high dose cohorts.^7,12,14 The recombinant enzymes IceM and IceMG described in this study offer immunomodulatory solutions that can directly address these challenges by transiently removing circulating as well as surface IgM, thereby reducing the risk associated with complement activation. In our study, IceMG treatment reduced complement activation triggered by AAV9 capsids at all doses tested as early as 1 h post-treatment. This blockade extends up to 7–14 days accompanying the return of IgM/IgG to baseline levels with the 2.5 mg/kg dose enabling the longest duration. A recent analysis of patients receiving AAV9 vectors in different gene therapy trials shows that antibody-mediated (IgM/C1q) complement activation begins 2 days post-administration and peak levels at day 5.¹⁴ Based on the duration of complement reduction by IceMG (1–2 weeks), we postulate that IceMG pre-conditioning 1 or 2 days prior to AAV administration may provide a window to effectively dampen antibody-mediated complement activation. While further in vivo monkey studies are required to validate these hypotheses, our initial study supports that these aspects can be modeled and validated in rhesus monkeys. It is noteworthy to mention that several pre-conditioning regimens such as plasmapheresis, IgG targeting agents (imlifidase, Vyvgart), B cell targeted agents (e.g., rituximab) and/or complement inhibitors (e.g., eculizumab) are being actively evaluated in the clinic. Against this backdrop, IceM/IceMG may provide a complementary approach and a comprehensive solution to prophylactically address IgG and IgM and complement activation.

Some limitations of the current study are important to note. Despite the promising attributes of IceM/IceMG, we note that the protective role of IgM in autoimmune and infectious diseases is well established and careful assessment of these novel agents is essential prior to incorporation in clinical IMRs. The current study provides proof-of-concept data in rhesus monkeys validating the engineered enzymes as potential immunomodulatory agents, but additional studies with dosing regimens, safety assessment and routes of administration (e.g., subcutaneous) is required prior to advancement to the clinic. In addition, testing of different pre-conditioning regimens prior to AAV vector dosing in macaque models as well as potential applications of vector re-dosing are warranted in the future. Nevertheless, degradation of AAV NAbs and inactivation of the complement cascade with IceM and IceMG expand patient recruitment and help decrease acute immunotoxicity risks associated with AAV-based gene therapies. In addition to human gene therapy, another potential application of IceM/IceMG is in the fields of autoimmune disease and organ transplantation. For instance, kidney transplant recipients often have donor-specific antibodies that are known to trigger antibody-mediated rejection through the activation of the complement cascade.³⁹ Further corroborating our discussion, the IgG degrading enzyme IdeS or Imlifidase (marketed as Idefirix in Europe), has successfully enabled kidney transplantation in patients harboring donor-specific antibodies.⁴⁰ Taken together, IceM and IceMG are promising new agents for modulating humoral and complement pathways in autoimmune disorders, organ transplantation and AAV gene therapy.

Materials and methods

Structural modeling and enzyme-immunoglobulin complex analyses

Structural models for IceM and IceMG were generated using AlphaFold2-ptm.⁴¹ Molecular graphics and analyses were performed with UCSF ChimeraX, developed by the Resource for Biocomputing, Visualization, and Informatics at the University of California, San Francisco, with support from National Institutes of Health R01-GM129325 and the Office of Cyber Infrastructure and Computational Biology, National Institute of Allergy and Infectious Diseases.⁴² Structural models of IceM-IgM and IceMG-IgM/IgG complexes were generated using AlphaFold2-multimer v.2 using the existing structure of IdeS (PBD: 8A47) as a template.²⁹ Contact residues analyses were performed using the select contacts command in ChimeraX with atomic distance ≤5 Ang. For image rendering of complexes, superposition of AlphaFold2 prediction to PDB: 2RCJ was performed using the matchmaker command in ChimeraX.⁴³ AAV capsid structural model used for the graphical abstract were generated using ChimeraX using PDB: 3NG9 originally deposited on the Protein Data Bank by the remarkable Dr. Mavis Agbandje-McKenna.

Plasmid constructs and recombinant protein expression

For expression of recombinant His-tagged Ide_Ssuis, IceM, or IceMG, coding sequences were synthesized by Twist Bioscience and cloned into the pET-21a expression vector. E. coli strain BL21 star was transformed with recombinant IceM-pET-21a or IceMG-pET-21a. A single colony was inoculated into Terrific Broth medium containing ampicillin and incubated in 37 °C at 200 rpm. When the OD₆₀₀ reached ∼4, the bacterial culture was induced with 1 mM isopropyl-β-d-thiogalactoside at 37°C for 4 h. Cells were harvested by centrifugation and lysed with lysis buffer (50 mM Tris-HCl, 150 mM NaCl, 1% Triton X-100, cocktail [pH 8.0]). The target protein was obtained by two-step purification. Proteins were purified from cell lysate supernatant using Ni-NTA resin followed by Superdex 200 column purification. Recombinant protein was stored in 50 mM Tris-HCl, 150 mM NaCl, and 10% glycerol (pH 8.0) and sterilized by 0.22 μm filter before being stored in aliquots. Endotoxin was removed from recombinant protein using High-Capacity Endotoxin Removal Spin Columns following the manufacturer’s instructions (Thermo Fisher Scientific, 88274). The concentration was determined by Bradford protein assay with BSA as standard. The protein purity and molecular weight of IceM (41.6 kDa) and IceMG (70.4 kDa) were determined by standard SDS-PAGE followed by Coomassie staining with 2 μg BSA used as a standard.

Tissue culture

Cell lines were handled and maintained according to instructions from the American Type Culture Collection (ATCC). Human PBMCs (ATCC, PCS-800-011) and Daudi cells (ATCC, CCL-213) were maintained at 37°C and 5% CO₂ in RPMI 1640 medium (Gibco, 11875119) supplemented with 10% fetal bovine serum (MilliporeSigma, F0926). Human embryonic kidney (HEK293) cell lines were maintained at 37°C and 5% CO₂ in Dulbecco’s modified Eagle’s minimal medium (DMEM) (Gibco, 11995073) supplemented with 10% fetal bovine serum and 100 U/mL penicillin and 100 μg/mL streptomycin (Gibco, 1514014).

SDS-PAGE and western blot analysis

Samples were incubated at 95°C for 5 min in LDS loading buffer supplemented with 2.5% β-mercaptoethanol, separated by SDS-PAGE, and either stained using Coomassie blue (Themo Fisher Scientific, 24590) or transferred to PVDF membrane. Membranes were blocked with 5% milk powder in Tris-buffered saline with 0.1% Tween 20 (TBST) and incubated for 1 h at room temperature with horseradish peroxidase-conjugated primary antibodies for human IgM Fc (1:5,000; Proteintech, HRP-66484), porcine IgM Fc (1:5,000; Novus, NBP2-42699H), mouse IgM Fc (1:10,000; Invitrogen, 62-6820), IgG Fc (1:10,000; Invitrogen, 31423), IgA Fc (1:1,000; Invitrogen, A18781), IgE Fc (1:1,000; Invitrogen, A18793), IgD Fc (1:5,000; Proteintech, 16489-1-AP), 6xHistidine (1:10,000; Proteintech, HRP-66005), Phospho-PLC-γ2 (1:1,000; Cell Signaling Technology, 3871), PLC-γ2 (1:1,000; Cell Signaling Technology, 3872), or alpha-Tubulin (1:50,000; Proteintech, 66031). Membranes were then washed 3× with TBST and incubated for 1 min with enhanced chemiluminescence substrate (Thermo Fisher Scientific, 34049). Coomassie-stained gels and immunoblots were visualized on a Bio-Rad ChemiDoc imaging system, and where applicable, quantitated using Bio-Rad Image Lab v.6.1.0.

In vitro cleavage assays

For the initial cleavage assay to assess protein candidates, IgM cleavage was assessed by treating 25 μL of human serum or 10 μg of purified human IgM (MilliporeSigma, I8260) with 75 μL bacterial lysate at 37°C for 2 h. For assays with purified recombinant enzyme, IgM cleavage was assessed by treating 10 μg of purified human IgM with 20 μg/mL IceM, IceMG, or PBS and incubated at 37°C for 1 h. Enzymatic reactions were halted by adding LDS loading buffer, separated by SDS-PAGE, and stained with Coomassie blue as described above. For serum samples, 25 μL of serum were treated with IceM at the concentration indicated for 1 h at 37°C. Serum samples were diluted 1:10 or 1:50 for IgM or IgG immunoblots, respectively. Western blots were completed as described above and probed using Fc-specific anti-IgM, -IgG, -IgA, -IgE, or -IgD antibodies. Human serum samples were purchased from BioIVT; pig, dog, and mouse serum samples were obtained from in-house lab stocks. Cleavage assays were performed in biological triplicate using sera from different animals or individual human samples.

Microscopy and immunofluorescence

Daudi cells incubated with PBS, 20 μg/mL IceM or IceMG for 1 h at 37°C. Cells were washed 3× with ice-cold PBS and then cytocentrifuged onto Lab-Tek II Chamber Slides (Thermo Fisher Scientific, 154453) coated with poly-L-lysine. Cells were washed with PBS supplemented with 1 mM CaCl₂ and 1 mM MgCl₂ and then fixed using 10% formalin for 20 min. Cells were blocked with 5% normal goal serum in PBS with for 30 min and then incubated with F(ab)-specific anti-human IgM antibody conjugated to FITC (LSBio, LS-C536848), for 1 h at 4°C. Following three washes with ice-cold PBS, slides were mounted with Prolong Gold Antifade Reagent with DAPI (Invitrogen, P36930) and immunofluorescence was visualized using 10× or 20× Olympus objectives on a Zeiss 880 Airyscan Fast Inverted Confocal microscope.

Flow cytometry analysis

For surface bound IgM cleavage analysis, 3 × 10⁵ PBMCs or 5 × 10⁴ Daudi cells were incubated with PBS, recombinant IceM, or recombinant IceMG at the concentration indicated for 1 h at 37°C. Cells were washed 3× with ice-cold PBS and then incubated with F(ab)-specific anti-human IgM antibody conjugated to FITC (LSBio, LS-C536848), anti-CD19 conjugated to APC (Abcam, AB18224), and Zombie Violet (BioLegend, 77477) to assess cell viability per manufacturer recommendations in PBS supplemented with 3% BSA on ice for 30 min. Cells were then washed 3× with ice-cold PBS and analyzed using a BD Biosciences Fortessa X-20 flow cytometer. For Daudi cells, flow data was assessed for the IgM⁺ (FITC⁺) cohort only. For PBMCs, flow data were gated on live cells and then CD19⁺ (APC⁺) to analyze the IgM⁺ (FITC⁺) cohort. Experiments were performed in triplicate. Data were analyzed using FlowJo v.10.8.1 (BD Biosciences) and graphs were generated using GraphPad Prism v.9.5.

Recombinant virus production, purification, and quantification

Recombinant AAV8 and AAV9 vectors packaging a single-stranded genome encoding firefly luciferase driven by a chicken beta actin (CBA-Luc) promoter were generated by triple plasmid transfection. In brief, HEK293 cells were seeded around 70% confluence in DMEM with 5% FBS and allowed to adhere overnight. Cells were transfected with adenoviral helper plasmid (pXX680; Aldevron), AAV rep-cap plasmid (pLH8 or pLH9) and the luciferase transgene plasmid (pTR-CBA-Luc) using polyethylenimine (PolySciences, 24765-1). Media and/or cell pellets were collected 6 days post-transfection and viral vectors were harvested by polyethylene glycol precipitation followed by purification by iodixanol density gradient ultracentrifugation. Viral vectors were further subjected to desalting and buffer exchange (Thermo Fisher Scientific, 87770) into PBS supplemented with 0.001% pluronic F68 and 1 mM MgCl₂. To quantify viral titers, purified viral vectors were treated with 10 mg/mL DNase to degrade unencapsidated DNA and subjected to 5% Tween to release viral genomes from capsids. Vector genomes (vg) and were then quantified by quantitative PCR using a Lightcycler 480 (Roche Applied Sciences). SYBR Green I Master Mix (Roche Applied Sciences, 4887352001) and primers specific for the luciferase transgene were used for amplification (IDT; forward 5′-AAAAGCACTCTGATTGACAAATAC-3ʹ; reverse 5′-CCTTCGCTTCAAAAAATGGAAC-3′).

Complement assays

Complement activation assays were carried out as described previously.³⁴ To assess the effect of these enzymes on complement activation in human samples in vitro, 25 μL of complement preserved human serum (BioIVT, HUMANSRM-01017110) were incubated with PBS, 20 μg/mL of recombinant IdeZ, IceM, or IceMG for 1 h at 37°C. Samples were then treated with PBS, 2 × 10¹¹ vg AAV9, 25 μg/mL:0.25 U/mL platelet factor 4 (PF4):heparin, or 10 mM EDTA for 45 min at 37°C. Each data point represents a serum sample from an individual human donor. For the rhesus macaque samples, 25 μL of serum from animals R1, R2, or R3 at each time point were incubated with AAV9 as described above. Untreated serum samples (no AAV9 or PF4/heparin) were incubated alongside treated samples and considered baseline. Baseline for the nonhuman primate study represent an average across all three doses. EDTA inactivates complement and was used as a negative control. Complement activation was evaluated by assessing Complement Factor B (LSBio, LS-F42697) or C3a (Thermo Fisher Scientific, BMS2089) levels by a commercial ELISA kit and absorbance was measured at 450 nm using a Varioskan LUX Multimode Microplate Reader. Factor B and C3a levels were determined using a four-parameter logistic (4PL) regression and expressed as μg/mL (factor B) or ng/mL (C3a).

NAb assays

AAVs packaging a CBA-Luc reporter were made by triple plasmid transfection described above. NAb assays were carried out as described previously.²⁴ Serum samples were serially diluted and coincubated with equal volumes (25μL of each) of AAV at an MOI of 1 × 10⁵ for 30 min at 37°C in black 96-well assay plates. HEK293 cells (1 × 10⁴) in DMEM with 10% FBS were then added to each well and incubated at 37°C in 5% CO₂ overnight. Cells were then lysed using 25 μL Passive Lysis Buffer (Promega, E1941), and transduction was assessed by measuring luciferase transgene expression with 25 μL Luciferase Assay Reagent (Promega, E1483). Luminescence was measured using a Varioskan LUX Multimode Microplate Reader (Thermo Fisher Scientific) per manufacturer specifications. Sigmoidal neutralization curves were generated and were normalized to controls with no antisera. The neutralizing titer was calculated as the serum dilution required to inhibit AAV transduction ≥50% compared with no serum controls.

Total and anti-AAV9 IgM and IgG ELISAs

Total anti-monkey IgM and IgG levels were measured by ELISA. In brief, serum samples were diluted 10,000-fold (IgM) or 100,000-fold (IgG) and incubated on monkey IgM- or IgG-coated assay plates (Life Diagnostics, IGG-3, IGM-3) and processed per manufacturer specifications. Anti-AAV9 IgM and IgG levels were evaluated by capsid capture antibody ELISA. AAV9 capsids were diluted 1 × 10¹⁰ vg/mL in coating buffer (50 mM carbonate buffer [pH 9.4]) and 60 μL were added to each well in Medisorb Immunoplates (Thermo Fisher Scientific, 467320) and incubated overnight. Plates were then washed five times in PBS with 0.1% Tween 20 (PBST) followed by blocking in 6% BSA in PBS for 2 h at room temperature. Serum samples were serially diluted and allowed to bind the plate at 37°C for 1 h with gentle agitation. Standard curves were generated using purified IgM (Rockland, 017-0107) or IgG (Rockland, 017-0102). Plates were washed with PBST and incubated with horseradish peroxidase-conjugated antibodies for monkey IgM (Rockland, 617-103-007) or IgG (Rockland, 617-103-012) at 37°C for 1 h with gentle agitation. Plates were washed and incubated with 3,3′5,5′-tetramethylbenzidine high sensitivity substrate (BioLegend, 423001) for 20 min before adding stop solution (BioLegend, 423001). Absorbance was measured at 450 nm using a Varioskan LUX Multimode Microplate Reader (Thermo Fisher Scientific). Anti-AAV9 IgM or IgG levels were determined using a 4PL regression and expressed as μg/mL.

In vivo mouse studies

All mouse care and procedures were conducted in accordance with National Institutes of Health (NIH) guidelines and approved by the Duke University Institutional Animal Care and Use Committee. For mouse studies, 12-week-old female C57BL/6J mice were purchased from Jackson Laboratory and injected intraperitoneally with 200 μL of pooled human sera (MilliporeSigma, H4522). After 24 h, 200 μL of PBS, IceM or IceMG were administered intravenously by tail vein injection at the dose indicated. Blood samples were collected by submental bleeds at 24 h post-injection, allowed to coagulate for 30 min, and serum was separated by centrifugation. Mouse serum samples were diluted 1:10 for IgM blots or 1:50 for IgG blots in LDS loading buffer with 2.5% β-mercaptoethanol, incubated at 95°C for 5 min prior to SDS-PAGE, and western blots for anti-IgM or anti-IgG were performed as described above. Five mice (numbered 1–5) were tested per experimental condition. Human serum positive control was included with the PBS samples. For the AAV study, mice were injected intraperitoneally with pooled human sera supplemented with 8 mg human IVIg followed by 200 μL of PBS, IdeZ, IceM, IceMG administered intravenously by tail vein injection (left tail vein) at the dose indicated. Twenty-four hours later, mice were injected with 1 × 10¹³ vg/kg AAV8- or AAV9-CBA-Luc intravenously by tail vein injection (right tail vein). To quantify luciferase expression, mice were euthanized 21 days post-treatment, tissues were harvested, weighed, and lysed by adding 200 μL of 1× passive lysis buffer (Promega) before mechanical lysis using a Fastprep-24 tissue homogenizer (MP Biomedicals). Lysate was cleared by centrifugation (21,100 × g, 3 min, 4°C) to remove any remaining tissue debris. To measure luciferase transgene expression, 25 μL of supernatant from each lysate was then loaded onto an assay plate along with 25 μL of Luciferase Assay Reagent, and luminometric analysis was performed using a Victor X3 microplate reader (PerkinElmer) per manufacturer specifications. The relative luminescence units obtained for each sample were then normalized to the input tissue weight for each sample, measured in grams, followed by log transformation.

In vivo rhesus macaque studies

All procedures conformed to the requirements of the Animal Welfare Act, and protocols were approved prior to implementation by the IACUC at the University of California, Davis. Three rhesus macaques (R1, R2, R3; ∼2 years of age, ∼ 3–3.5 kg, 2 males and 1 female) were confirmed for preexisting AAV9 NAbs using the assay described above and designated seropositive. Animals were sedated with telazol (5–8 mg/kg) (for IceMG administration) or ketamine intramuscular (∼5–10 mg/kg) according to established IACUC-approved standard operating procedures for select blood sample collection (complete blood counts [CBCs], clinical chemistry, serum, plasma; all CBCs and clinical chemistry panels remained within normal limits during the study period). Animals were dosed intravenously at 0.25, 1.0, or 2.5 mg/kg IceMG respectively. Blood samples were collected prior to administration (0 time point) then at 1, 3, 6, and 24 h post-administration, followed by 2, 3, 7, and 14 days post-treatment (light ketamine sedation for sample collection and monitoring body weights). Serum was separated by centrifugation and diluted 1:10 for IgM blots or 1:50 for IgG blots in LDS loading buffer with 2.5% β-mercaptoethanol. Samples were incubated at 95°C for 5 min prior to SDS-PAGE and western blots for Fc-specific anti-IgM, Fc-specific anti-IgG, and anti-6xHistidine were performed as described above. Serum samples from each animal were assessed for in vitro complement activation and the presence of anti-AAV9 NAbs as described above.

Statistical analyses

Error bars denote mean ± standard deviation. All experiments were completed in at least biological triplicate with exception of Daudi B cell signaling (n = 2), and macaque studies where n = 1 were included for each of the three doses. Statistical significance was determined as indicated in each figure and analyses were performed with GraphPad Prism v.9.5 with p values less than 0.05 being significant. Statistical analyses were performed comparing all samples within each study. In some figures, only p values that were significant (p < 0.05) were noted in the graphs whereas p values greater than 0.05 were omitted from the graphs. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; ns, not significant.

Data and code availability

Bacterial genome sequences are available on NCBI (txid: 467210). Enzymes described in this study are available for sharing upon request through Duke University Material Transfer Agreement.