Research Highlights

The Anti-inflammatory Activity of IgG Is Enhanced by Co-engagement of Type I and II Fc Receptors A.T. Jones et al., Science 2025;390: eadv2927. DOI: 10.1126/science.adv2927 Intravenous immunoglobulin (IVIG) is widely used to treat a variety of autoimmune diseases as well as antibody-mediated rejection in organ transplantation.1 However, IVIG requires extremely high doses (1–2 g/kg) and relies on IgG pooled from many human donors, therefore faces many challenges in supply, manufacturing capacity, infusion burden, and cost. These issues have motivated work exploring the development of recombinant IgG-based therapeutics with similar or better anti-inflammatory properties. It has previously been shown that terminal sialylation of the single N-linked glycan in the Fc domain of IgG enhances its anti-inflammatory properties by interacting with the inhibitory Fc receptor FcγRIIB. Prior mouse studies suggested that sialylated IgG (sIgG) or the Fc fragments (sFc) can recapitulate much of IVIG’s immunomodulatory activity.2 However, murine FcγRs differ significantly from human FcγRs, limiting the ability to draw translational conclusions from mouse systems, especially when Fc-engineering strategies are used to modulate specific human IgG(Fc)-FcγR interactions. To address these challenges, Jones et al3 sought to determine whether an engineered sialylated human IgG1 Fc fragment (sFc) with enhanced affinity for human type I inhibitory Fc receptor FcγRIIB (hFcγRIIB) could function as a highly potent anti-inflammatory therapeutic. To explore this, the authors used a humanized mouse model of arthritis in which the complete repertoire of human type I FcγR was expressed. Two major types of FcγRs exist: (1) type I FcγRs (classical Fc receptors), which are IgG-binding receptors, such as FcγRI, FcγRIIA, FcγRIIB, and FcγRIIIA, and which mediate activating or inhibitory signaling and (2) type II FcγRs (C-type lectin receptors) that include DC-SIGN and CD23, which bind IgG in a glycan-dependent manner and modulate anti-inflammatory pathways. The authors first established that sFc and its engagement of hFcγRIIB were necessary for the anti-inflammatory effects using a well-characterized K/BxN serum transfer-induced arthritis mouse model. In this model transferring K/BxN mouse serum to naive humanized FcγR mice results in the development of inflammatory arthritis. Mice pretreated with recombinant sFc before receiving K/BxN serum were protected against arthritis at ~25-fold lower doses than IVIG. This inhibitory effect was sialylation-dependent, as the removal of sialic acid residues completely abrogated this protective effect. Similarly, a mutant Fc incapable of binding to hFcγRIIB also failed to protect. Finally, the authors used rational Fc engineering to create the V11 sFc variant, with a ~37-fold higher affinity for hFcγRIIB. Remarkably, V11 sFc protected against K/BxN arthritis at just 10 mg/kg, a dose about 100-fold lower than IVIG. These findings demonstrate that enhanced hFcγRIIB engagement drives anti-inflammatory activity of sialylated IgG and establish hFcγRIIB as a compelling therapeutic target. Surprisingly, the authors demonstrated that type I and II FcγRs synergize to bind sFc to mediate the anti-inflammatory activity. To determine the requirement for type II FcγRs in mediating the protective effect of V11 sFc, V11 sFc was administered to mice also receiving anti-SIGN-R1 (a mouse homolog for DC-SIGN, a human type II FcγRs). Blocking SIGN-R1 eliminated the protective effect of V11 sFc in the K/BxN model. Using glycoengineered antihuman CD20 rituximab, they further demonstrated that only sialylated IgG exhibited enhanced binding to cells co-expressing both type I and type II FcγRs. This glycan-dependent physical interaction between type I and type II FcγRs challenges the traditional separation of type I and II FcγR pathways and suggests that receptor co-engagement may be essential for sialylated IgG–mediated immunomodulation. For transplant immunology, these data are highly pertinent. Many forms of allograft injury including antibody-mediated rejection and BK nephropathy have been effectively treated with IVIG. IVIG may further synergize with current immunosuppressive regimens or costimulation blockade approaches. A highly potent sFc with preferential FcγRIIB engagement could therefore serve as a more available, better tolerated, and more consistent alternative to conventional IVIG. Deep Generative Models Design mRNA Sequences With Enhanced Translational Capacity and Stability Zhang H, Liu H, Xu Y et al. Science 2025;390:eadr8470 The COVID-19 pandemic saw the first global impact of mRNA therapeutics as an effective vaccine against a pathogen, although applications of mRNA therapeutics may extend far beyond vaccine strategies. For example, protein replacement for protein deficient conditions is another common application. However, broader application of mRNA therapeutics is currently limited by several challenges, particularly in designing mRNA sequences that can achieve enhanced and durable protein expression for optimal therapeutic outcomes.1 To explore this, Zhang et al2 tested a deep generative model for mRNA design (called GEMORNA, for “GEnerative MOdels for RNA”) to achieve unprecedented protein expression capacity and durability. The model was developed in two stages. The first stage was to optimize an algorithm for coding sequences (CDSs), trained using a transformer architecture with an “encoder” and a “decoder” module, the former to process protein information and the latter to generate CDS sequences. More than 3000 GEMORNA-generated CDSs were then tested for known CDS optimization-related features, including codon adaptation index, GC content, U percentage, normalized minimum free energy, percentage of unwanted codon pairs, percentage of slippery sites, naturalness, and others. These comparisons demonstrated significant advantages of GEMORNA CDSs over other designs. Next, they used 2 cell lines, HEK293T and HepG2 cell lines, to test the expression of GEMORNA-CDS of firefly luciferase, and both cell lines exhibited up to 20-fold increase of protein expression and higher in-cell stability of GEMORNA-CDS in comparison with 5 other control designs. The second stage was to develop a model for the design of untranslated regions (UTRs) that accompany CDSs, as these sequences play major roles in the performance of mRNAs. The GEMORNA-UTR model was generated by first training a “decoder”-only transformer architecture on natural UTRs, followed by “fine-tuning” with (1) selected natural 5′UTRs that possess higher mean ribosome loads predicting higher translation efficiency and (2) 3′UTRs predicting higher stability. Combining GEMORNA-CDS and GEMORNA-UTR allowed the authors to test GEMORNA designed full-length mRNA for desired therapeutic effects. Here, the authors tested GEMORNA generated COVID-19 spike protein mRNA vaccine and human erythropoietin (EPO) in mice. Lipid nanoparticle-encapsulated COVID-19 spike protein mRNA induced higher antibody titers than the commercial BNT162b2 at multiple time points after immunization. Similarly, GEMORNA EPO-mRNA achieved longer durability and stronger expression levels in mice than benchmark comparisons. Finally, the authors advanced GEMORNA to circular RNA (circRNA) designs. They combined GEMORNA CDSs with mined internal ribosomal entry sites to generate circRNAs. GEMORNA circRNAs encoding EPO similarly demonstrated long-lasting high level protein expression in vitro and in vivo. Finally, they tested GEMORNA circRNA for the generation of CD19 CAR T cells from primary human T cells and showed substantial increases in CD19 gene expression (by 28-fold) for a much longer period (up to 120 h), with accompanying potent cytotoxicity against human leukemic B cells. Applying to transplantation, this tool could have several uses. For example, it could be used in the design of HLA-based negative vaccines for transplant tolerance induction; for HLA CAR Tregs production for controlling deleterious transplant immunity; or for the release of anti-inflammatory factors during organ procurement and organ preservation. Overall, it adds another powerful tool to our armamentarium for future therapeutic design.