Samuel H. Gellman

Photo of Samuel H. Gellman
Ralph F. Hirschmann Professor (also Department of Chemistry)
A.B., Harvard University
Ph.D., Columbia University
Postdoctoral, Caltech
Phone: (608) 262-3303

Proteins and protein-like molecules:
design, synthesis, structure and function

We are interested in relationships between biopolymer structure and function, with an emphasis on proteins and protein-inspired molecules. Our group combines tools from biochemistry, biophysics, chemical biology and organic chemistry in pursuit of experimental goals. Many projects involve collaboration with other laboratories at UW-Madison and around the world.

A major theme in our research emerges from a fundamental question: how do structural and functional properties of polypeptides change when the backbone is modified? Proteins evolve naturally via side chain modification (mutation of one α-amino acid residue to another), and engineered side chain modification has become a standard strategy in protein science. In contrast, very little is known about the impact of polypeptide backbone modification, perhaps because this type of variation does not seem to have been explored in biological systems. Evaluating backbone modification currently requires the application of chemical synthesis methods (standard solid-phase techniques).

Figure 1. A generic α-amino acid (left) compared with a generic β amino acid (right).

Figure 1. A generic α-amino acid (left) compared with a generic β amino acid (right).
The dotted line on the latter drawing indicates that the two side chains can be incorporated into a ring.

Much of our work so far has involved extending the amino acid backbone by one carbon, i.e., examination of β-amino acids as an alternative to α-amino acids (Figure 1). We are interested further subunit extension as well, to γ-amino acids and beyond, but such efforts require considerable investment in organic synthesis.

We have asked how backbone modification affects the informational properties of polypeptides, i.e., how this type of modification affects interactions with partner proteins. Oligomers that contain mixtures of α and β residues (α/β-peptides) have proven to be particularly interesting. α/β-Peptides containing 25-30% β residues, with α→β replacement sites distributed evenly, can adopt an α-helix-like conformation and bind to protein partners that naturally associate with α-helical ligands. In contrast to authentic α-peptides, however, the α/β-peptides are poorly recognized by protease enzymes. Thus, these α/β-peptides persist in biological environments that cause rapid destruction of analogous conventional peptides (α residues only).

Figure 2. An α/β-peptide that inhibits HIV infection.

Figure 2. An α/β-peptide that inhibits HIV infection.
Sequence shown at top. Lower left shows a co-crystal structure of the α/β-peptide (multicolor) bound to a protein partner (gray) (PDB 3O43). In the α/β-peptide, yellow indicates α residues, blue indicates β residues that are homologous to (same side chain) natural a residues, and orange indicates β residues with a cyclic constraint.

Figure 2 shows the crystal structure of a 38-residue α/β-peptide (yellow, orange and blue) bound to an engineered partner protein (gray) that represents a critical portion of the HIV protein gp41. Viral infection depends on gp41, which orchestrates fusion of the viral envelope with the target cell membrane. Our α/β-peptide, like the conventional peptide on which it is based, blocks HIV infection in a cellular assay (Horne et al. PNAS 106:14751 (2009)). The crystal structure shows that the α/β-peptide forms a 10-turn helix and binds into a long, complementary groove presented by gp41. The α/β-peptide is an excellent structural mimic of the α-helix that naturally binds to the target site on gp41, despite the fact that every turn of the α/β-peptide helix contains one more carbon atom than does one turn of an authentic α-helix. We have deposited many crystal structures of type in the PDB. We are currently extending this approach to respiratory viruses, including parainfluenza viruses, respiratory syncytial virus and coronaviruses (e.g., Outlaw et al. J. Am Chem. Soc. 141:12648 (2019); J. Am Chem. Soc. 142:2140 (2020)).

Cartoon showing a polypeptide hormone (blue) such as GLP-1 or PTH bound to a cognate B-family GPCR.

Figure 3. Cartoon showing a polypeptide hormone (blue) such as GLP-1 or PTH bound to a cognate B-family GPCR.
Three of the intracellular proteins (a G-protein and two β–arresins) that sense the receptor’s binding to the hormone and
initiate signaling cascades are indicated.

The anti-viral example illustrates how polypeptides with modified backbones can bind to specific recognition sites on target proteins and thereby block pathogenic protein-protein interactions while resisting proteolytic destruction. More recently, we have discovered that backbone modification of polypeptide hormones can generate agonists with unique receptor-activation profiles. The B-family GPCRs that are activated by hormones such as GLP-1 or PTH exert powerful physiological effects (Figure 3) and are targets of drug development. Receptor conformational changes induced GPCR engagement with an agonist are detected by intracellular proteins, such G proteins and β-arrestins, which then initiate distinct signaling processes. Backbone modifications can alter the signaling profile (Cheloha et al. Nature Biotech. 32:653 (2014); Hager et al. J. Am. Chem. Soc. 138:14970 (2016); Liu et al. PNAS 115:12383 (2018); Liu et al. J. Am. Chem. Soc. 141:14486 (2019)). We are currently working to elucidate and harness these remarkable effects.