We have known for decades that proteins consist of tens to thousands of amino acids strung together, but it has proven difficult to predict the function of a given protein based on its amino acid sequence. This difficulty rests on the overarching principle that the 3D structure of a protein determines its functionality by specifying with which other proteins and molecules it can interact. Conventional structural biology approaches, like X-ray crystallography, have enabled us to figure out the shapes and interactions of many rigidly folded proteins or their constituent domains. However, sometimes invisible to crystallographers are so-called low-complexity and intrinsically disordered regions of proteins, which often do not form stable and predictable 3D structures in physiological conditions or in the absence of their binding partners. Despite this conformational heterogeneity and flexibility, intrinsically disordered proteins (IDPs) and unstructured stretches of amino acids participate in many essential processes and provide new paradigms for protein-protein interactions. IDPs are generally thought to either conditionally establish secondary structure elements upon binding to their target or remain unfolded and participate in low-affinity, high-avidity dynamic interactions with other proteins. The former case can be exemplified by the formation of an α-helical region in the disordered domain of the transcription factor CREB upon binding to its transcriptional coactivator, CBP/p300 (Sugase et al., 2007Sugase K. Dyson H.J. Wright P.E. Mechanism of coupled folding and binding of an intrinsically disordered protein.Nature. 2007; 447: 1021-1025Crossref PubMed Scopus (852) Google Scholar). Transient and multivalent interactions are perhaps best illustrated by the core selectivity filter of nuclear pore complexes, where proteins with low-complexity sequence regions dynamically bind and unbind at many sites on the surface of nuclear import or export receptors, which is thought to drive fast directional transport. Indeed, intrinsically disordered regions of proteins mediate many of the multivalent interactions that form the basis of phase-separated liquid droplets, which are now widely appreciated in myriad cellular processes like RNA processing (Schmidt and Gorlich, 2016Schmidt H.B. Gorlich D. Transport Selectivity of Nuclear Pores, Phase Separation, and Membraneless Organelles.Trends Biochem. Sci. 2016; 41: 46-61Abstract Full Text Full Text PDF PubMed Scopus (257) Google Scholar). Technological innovation in protein structural biology, biophysics, molecular dynamics simulations, and single-molecule studies have enabled many advances in our understanding of how proteins or protein segments without defined structure uniquely contribute to important cellular processes. Several recent studies are challenging the long-held paradigms of how unstructured protein regions and peptides interact with their cognate partners to achieve specificity, affinity, and mechanostability—and to do what canonical interactions between globular domains simply cannot do. A recent study published in Nature uncovered an extremely high-affinity interaction between two IDPs, the linker histone H1 and its nuclear chaperone prothymosin-α. Surprisingly, the authors did not detect the formation of any transient secondary structure elements upon binding of these two unstructured proteins, despite the formation of a complex with incredibly strong binding and picomolar affinity. A variety of structural, biochemical, and modeling techniques led to the conclusion that the two proteins form an extremely dynamic complex wherein long-range electrostatic interactions lead to rapid interconversion between countless different combinations of oppositely charged residues and patches (Borgia et al., 2018Borgia A. Borgia M.B. Bugge K. Kissling V.M. Heidarsson P.O. Fernandes C.B. Sottini A. Soranno A. Bulhozer K.J. Nettels D. et al.Extreme disorder in an ultrahigh-affinity protein complex.Nature. 2018; 555: 61-66Crossref PubMed Scopus (369) Google Scholar). This striking new example expands the known universe of protein-protein interaction determinants and may defy conventional thinking about how protein structure determines function. While low-complexity protein domains have been implicated in liquid-liquid phase separation and the formation of membraneless organelles, like stress granules and P-bodies in eukaryotic cells, the precise molecular determinants governing these condensates have, in most cases, remained obscure. In a study published in Science, scientists determined the atomic structures of segments from several low-complexity domains implicated in phase separation and hydrogel formation and found that the short peptides consistently formed kinked beta sheet structures. While highly stable beta strands are core features of amyloid-forming proteins, the kinked nature of these LARKS (low-complexity, aromatic-rich, kinked segments) prevent cross-beta strand interactions among amino acid side chains that give amyloid fibrils their characteristic strength and irreversible nature. These molecular determinants lead to reversible protein meshworks among LARKS-containing proteins, forming the basis for phase-separated droplets. Strikingly, the authors found that hundreds of human proteins have the sequence propensity to form similar kinked beta strands and therefore may take part in multivalent interaction networks and phase separation (Hughes et al., 2018Hughes M.P. Sawaya M.R. Boyer D.R. Goldschmidt L. Rodriguez J.A. Cascio D. Chong L. Gonen T. Eisenberg D.S. Atomic structures of low-complexity protein segments reveal kinked B sheets that assemble networks.Science. 2018; 359: 698-701Crossref PubMed Scopus (242) Google Scholar). With more and more cellular processes relying on phase separation continually being uncovered, this list of proteins may lead to a treasure trove of new biology. In another extreme example of protein acrobatics, a paper recently published in Science showed that an adhesin protein from Staphylococcus species defies canonical interaction paradigms to establish the strongest known non-covalent protein association with its target—a staggering 10-fold higher mechanostability than biotin and streptavidin. The strength of this interaction with a segment of human fibrinogen is important for pathogenic staphylococci to form recalcitrant biofilms. The authors found that this remarkable stability is largely independent of amino acid side chains on the fibrinogen peptide, since mutating many residues and even replacing all the amino acids with glycine in a simulation did not impact the stability of the complex once it was bound. By bending the fibrinogen peptide into an elongated, corkscrew-like coiled shape that is buried within the adhesin protein and locked into place, the adhesin can achieve multivalent hydrogen bonding with the peptide backbone in a direction perpendicular to the shear force, thereby dissipating the mechanical stress over many individually weak bonds (Milles et al., 2018Milles L.F. Schulten K. Gaub H.E. Bernardi R.C. Molecular mechanism of extreme mechanostability in a pathogen adhesion.Science. 2018; 359: 1527-1533Crossref PubMed Scopus (124) Google Scholar). The evolutionary pressures facing pathogenic bacteria could have favored the development of this unconventional mechanism for extreme mechanostability. In addition to these new examples of proteins breaking the traditional rules, scientists are also finding IDPs and low-complexity domains popping up in all kinds of interesting biological processes. From keeping tardigrades alive in extreme dessication (Boothby et al., 2017Boothby T.C. Tapia H. Brozena A.H. Piszkiewicz S. Smith A.E. Giovannini I. Rebecchi L. Pielak G.J. Koshland D. Goldstein B. Tardigrades Use Intrinsically Disordered Proteins to Survive Desiccation.Mol. Cell. 2017; 65: 975-984Abstract Full Text Full Text PDF PubMed Scopus (183) Google Scholar) to the formation of tooth enamel (Wald et al., 2017Wald T. Spoutil F. Osickova A. Prochazkova M. Benada O. Kasparek P. Bumba L. Klein O.D. Sedlacek R. Sebo P. et al.Intrinsically disordered proteins drive enamel formation via an evolutionarily conserved self-assembly motif.Proc. Natl. Acad. Sci. USA. 2017; 114: E1641-E1650Crossref PubMed Scopus (38) Google Scholar), it is clear that biology is not constrained by neatly folded protein domains. At the same time, protein engineers are closer than ever to being able to design IDPs to assemble in programmable ways (Simon et al., 2017Simon J.R. Carroll N.J. Rubinstein M. Chilkoti A. López G.P. Programming molecular self-assembly of intrinsically disordered proteins containing sequences of low complexity.Nat. Chem. 2017; 9: 509-515Crossref PubMed Scopus (172) Google Scholar). As many labs around the world take a deep dive into the biophysics of protein-protein interactions with new tools at their disposal, who knows what they might find (or create) next? After all, rules are made to be broken.