Proteins at the cell surface are heavily modified with elongated carbohydrate chains though little is known about how modification impacts protein structure and function. Unlike protein sequence that is encoded in the genome, these carbohydrates are synthesized during secretion in the endoplasmic reticulum and the Golgi without a template, endowing cells an ability to rapidly modulate ligand affinity, receptor expression, spacing, clustering, and accessibility. Most secreted proteins contain at least one asparagine(N)-linked carbohydrate chain (glycan).
Our laboratory determined that N-glycosylation of a circulating immunoglobulin was required for receptor binding because long-range contacts within the antibody molecule stabilize key residues to form an interface with the receptor. We believe, based on these results and the high frequency of N-glycan addition to secreted proteins, that comparable mechanisms are likely found throughout the body but remain undescribed due to technological limitations. We developed a multi-technique strategy utilizing solution nuclear magnetic resonance spectroscopy, mass spectrometry, recombinant protein expression and stable isotope labeling as well as isolation and high-resolution characterization of proteins from primary human tissue to surmount this obstacle. We are currently applying and expanding this platform to identify new N-glycans that modulate protein function by focusing on the immune system where we expect to find countless more examples. A high-resolution definition of novel cell-mediated variables that affect function will provide insight into cell function and will form the foundation for future therapies to promote health and alleviate disease.
1. Identifying intramolecular interactions
We surmounted a fundamental challenge by developing a strategy to identify individual N-glycans that form stabilizing contacts with the polypeptide. Many proteins have multiple N-glycans and some are heavily glycosylated. For example, FcgRIIIa / CD16a has a 20 kDa extracellular domain with five N-glycans that increase the mass by ~40%. Not all form stabilizing contacts, in fact, it is likely most don’t. We use two primary approaches to identify N-glycans that form stabilizing contacts: solution NMR and glycoproteomics mass spectrometry.
2. Identify compositional changes of primary human NK cells and monocytes resulting from disease
We were surprised to determine that limiting CD16a N-glycan processing increased affinity for IgG1 Fc by 40-50x in vitro. This range is comparable to, if not greater than, the effect of IgG1 Fc modifications. Though IgG N-glycan composition is routinely analyzed because differences correlate to disease severity and susceptibility, nothing was known about processing of CD16a, the receptor responsible for antibody-based natural killer cell activation that contains five N-glycans and is present in the periphery at a level ~500,000x lower than IgG. Despite the obvious importance, tools to probe CD16a did not exist.
Receptor modifications can be rapidly recycled and are thus a logical choice for tuning a system in contrast to antibodies that circulate for three weeks. We wanted to know if high-affinity CD16a glycoforms are present in the human body. This question posed an exceptional technological challenge. We partnered with local two blood donation services to recover filters ordinarily discarded following plasma or platelet donation (apheresis). From a single filter we could recover enough NK cells to isolate CD16a N-glycans and found a preponderance of glycans associated with high affinity binding. These high affinity forms are not utilized to study antibody binding in drug development. Furthermore, we showed that receptor glycosylation cannot be predicted from recombinant expression and N-glycan composition impacted CD16a structure.
Beyond developing unprecedented tools to study the processing of a single integral membrane protein present at vanishingly small levels from a single cell type of a single donor, we demonstrated that NK cells express CD16a forms that bind with high affinity. By determining CD16a glycosylation in situ we provide appropriate CD16a forms for optimizing antibody-based drugs in addition to a high-resolution view of CD16a at the NK cell surface in our bodies. We also provided data that N-glycans from another protein impact structure and function.
3. Targeted improvement of IgG-based drugs
The first immunoglobulin G1 (IgG1) crystallizable fragment (Fc) structure determined by X-ray crystallography in 1976 revealed a high-resolution model of both the protein and carbohydrate components in a single pose with the carbohydrate packed tightly against the protein surface. However, this model and that of the Fc:CD16 complex 24 years later failed to explain 1) why covalent carbohydrate modification (N-glycosylation) was required for cell activation or 2) why increasing the length of the N-glycan increased receptor binding affinity or 3) how the N-glycan is enzymatically modified if stuck to the Fc surface. Though X-ray crystallography provides high resolution details of a molecule’s low-energy state, it provides little detail regarding molecular motions at physiological temperature. Insight into these questions can as a major breakthrough in 2011 when my mentor and I developed novel protein labeling methods and solution NMR spectroscopy strategies to study glycoproteins. We determined that the IgG1 Fc N-glycan sampled multiple conformations in contrast to the single pose observed with X-ray crystallography. Furthermore, this discovery led to a powerful hypothesis: IgG1 N-glycan motion is integral to receptor binding affinity and cell activation. This finding is important because it provided basic insight into the structure/function relationship of immunoglobulins, a widely used pharmaceutical, biological probe, and essential component of the immune system. Furthermore, this also provides an answer to question 3: because the N-glycan is mobile it is transiently accessible to modifying enzyme. Answers to the first two questions were revealed in the studies presented below.
We developed methods to define the relationship between N-glycan motion (described under manuscript #1) and receptor binding affinity. The existence of a positive relationship would explain why N-glycosylation at a comparable site is conserved in four of the five human antibody classes. We sought to mutate amino acid residues that form stabilizing contacts with the N-glycan to measure the impact of intramolecular contacts on motion and receptor binding affinity. The targeted residues contact the N-glycan but are 10-15 Å from the receptor binding interface. To achieve this, we developed an expression system that would allow stable isotope labeling (13C, 15N), an ability to express mutants, and provide appropriate N-glycan modifications. Common prokaryotic expression systems do not N-glycosylate (E. coli) or do so with dramatically different glycoforms (fungi). We surmounted limitations of human cell-based expression systems to achieve these goals. Another fundamental challenge existed: removing contacts that stabilize the N-glycan also promotes more extensive elaboration by processing enzymes in the Golgi and it is likely that changing the glycan composition changes motion. To eliminate this effect, we developed highly sophisticated methods to remodel the Fc N-glycans in vitro following purification, generating a single glycoform on all Fc variants for the measurements of motion and receptor binding affinity.
IgG Fc mutations at the N-glycan/polypeptide interface increased N-glycan motion and decreased CD16a binding affinity, revealing a strong linear correlation between glycan stabilization and receptor binding affinity (R2=0.82). These results were the first of their kind and demonstrated the N-glycan stabilizes receptor binding through a conserved allosteric mechanism. This provided the first direction for engineering antibody-based drugs: improve receptor binding and drug efficacy by reducing N-glycan motion. These data also explain why increasing N-glycan length increases receptor binding affinity: distal residues form additional stabilizing contacts.
We leveraged our powerful technologies to determine why antibody N-glycosylation is required for cell activation and how N-glycosylation affects antibody structure. This effort required novel approaches to define structure and mobility in large glycoproteins, a challenge consciously avoided by structural biology labs. We first rejected the longstanding hypothesis that N-glycosylation stabilized Fc quaternary structure. We determined that the relative orientations of the Cg2 and Cg3 domains in the glycosylated (wt) and non-glycosylated (T299A) Fc were indistinguishable using solution NMR. However, by incorporating 15N labels in the protein backbone, we identified structural changes in a single region: the C' strand and the C'E loop that contains the N-glycosylation site. In addition to structural differences we determined that glycan removal increased ms-ms motions in this same region, indicating that the IgG1 Fc N-glycan stabilizes receptor binding by reducing slow micros-ms motions of the C'E loop. Later reports from other groups using orthogonal techniques confirmed our results (Sci Rep 2017. 7:12659, MAbs 2019. 11:453, Antibodies 2019 8: pii: E39).
This singular result built upon our previous efforts, showing with atomic-level detail how N-glycosylation affects protein structure and that IgG N-glycosylation reduces micros-ms motions of a critical protein loop at the receptor-binding interface. I strongly believe that intramolecular interactions formed between carbohydrates and amino acid residues are prevalent in the immune system and that we are the only group currently applying the tools to identify new examples (for an example on a ion channel see Structure 2019 27:55-65). These results highlight a clear route to improving antibody-based drugs through protein engineering: stabilize the receptor-binding interface. Our definition avoids the pitfalls of prior false prophets (Fc quaternary structure), providing a strategy to identify similar crucial intramolecular interactions in other immune system proteins and drugs.