2.3 Backbone Dynamics of L-hIGFBP2 from 15N relaxation
Insight into the functional regulation of IGF-1 by L-h IGFBP2 was achieved by studying the dynamics of L-h IGFBP2 in the free and IGF-1-bound forms using 15N relaxation. Reduced spectral density mapping was used to examine dynamics in the microsecond-millisecond (µs-ms) and picosecond-nanosecond (ps-ns) regimes. The 15N relaxation rates (R­1, R2, R) and15N-1H heteronuclear nuclear Overhauser effects (HetNOE) were measured at a 1H resonance frequency of 800 MHz (Fig. 4 ). Based on15N R1 and Rrelaxation values, an average overall rotational correlation time of ~3 ns for the disordered linker domain was obtained for unbound L-h IGFBP2. The average overall correlation time for the linker domain in full-length h IGFBP2 (32 kDa) determined by a similar method was ~4 ns at 293 K, implying that the disordered linker domain retains a high degree of flexibility in the full-length form, largely unaffected by the presence of the N- and C-domains.
The 15N R1, R2, and15N-1H het-NOE values and a plot of the calculated spectral density functions- J (0),JN), and J (0.87*ωH) for L-h IGFBP2 in free and IGF-1-bound forms are shown in Fig. 4 and Fig. 5 , respectively. Several important observations can be made. First, L-h IGFBP2 exhibits a high degree of flexibility in both the free and IGF-1-bound forms, as reflected by theJ (0.87*ωH) and J (0) values, with the latter being significantly less than 2/5τc for most residues (where τc is the rotational correlation time of a rigid isotropically tumbling protein of equivalent size). Second, the15N relaxation rates for residues K150-E161, Q165, and M166 of the linker domain are significantly perturbed by IGF-1 in complex with L-h IGFBP2, as the large complex causes a great increase in correlation time (τc ) which in turn causes the fast T2 relaxation (Fig. 4) . In the full-length h IGFBP2 complex the intensities of cross-peaks corresponding to these residues in the 2D [15N,1H] HSQC spectrum are reduced owing to the formation of a large complex which causes an increase in correlation time (τc ), resulting in the fast T2relaxation and increased NMR line-width (Fig. S6 ). Third, largeJ (0) values indicative of dynamics in the µs-ms regime are significantly enhanced for L-domain residues involved in binding IGF-1 (K150-E161), as well as those distant from the binding site (V110, N113, H117, H172, Q165, M166, L174, and L182). This increase in J (0) values can be attributed to the larger size of the complex and slow conformational exchange in the µs-ms regime, and is quantified by the exchange rate, Rex, which was estimated by measuring the15N transverse relaxation rate in the rotating frame (R) at 800 MHz for both the unbound and bound forms of L-h IGFBP2. Values of Rex calculated using the difference in J (0) values obtained with R2 and R are plotted in Fig. 5 . Notably, an overall increase in Rex is observed in the IGF-bound complex for residues of L-h IGFBP2 both close to and distant from the binding site. This implies that in the IGF-1-bound complex L-h IGFBP2 populates an ensemble of alternate conformations that interconvert on the µs-ms timescale. The region bound to IGF-1 exhibits a ‘reduced’ level of conformational dynamics, such that some of the ns-ps timescales have now entered the µs-ms timescale (and therefore now entered the exchange regime in these experiments). Interestingly, it is known that proteolytic cleavage of h IGFBP2 by the pregnancy-associated plasma protein-A (PAPP-A) is enhanced in the IGF-1-bound state. As shown in Fig. 7 , the region K150-E161 from the disordered linker domain of h IGFBP2 has a helical propensity, as predicted by AlphaFold; this binds to IGF-1 and shows a change in exchange in µs-ms timescale due to the formation of the larger complex upon binding. The IGF-1 bound complex has a higher correlation time (τc ) which in turn causes the fast T2 relaxation, resulting in a higher relaxation rate, as the relaxation times and relaxation rates are simple inverses of each other. The majority of the L-h IGFBP2 residues recognized by proteases (Fig. 3 ) show increases in Rex upon IGF binding (Fig. 5 ). This explains the IGF-dependent dynamic modulation of a protease cleavage site region in the intrinsically disordered linker domain of h IGFBP2.
To estimate the conformational entropy associated with binding IGF-1, approximate backbone NH order parameters (S 2) using J (0) and JN) were calculated for L-h IGFBP2. J (0) and JN) calculated from R were used to avoid the effect of conformational exchange when estimating S 2values. Order parameters were calculated for both free and bound forms of L-h IGFBP2 and the change in conformational entropy (ΔS) was estimated using the calculated S 2 values (for residues with S 2 < 1) (Eq. 2; Fig. 6 ). The overall ΔS value (summed over all residues) of ~100 J/mole (0.024 kcal/mol) implies an increase in entropy for the system upon IGF-1 binding. The contribution of the conformational entropy to the free energy of binding is given by -TΔS, which yields a contribution of -7 kcal/mol.
Discussion
In recent years, the concept of “fuzzy complexes” in IDPs has been described , which proposes that functionally important regions of IDPs in protein complexes can retain their structural disorder. In fuzzy complexes, dynamic regulation ensues when the ensemble average population of conformers of the IDP and/or their flexibility are affected upon ligand binding. The current study involving the intrinsically disordered linker domain of human IGFBP2 exemplifies such a case.
In the IGF system, proteolysis plays a crucial role in regulating the bioavailability of IGFs. IGFBP levels are regulated by proteolysis following their secretion from the cell and the resulting proteolytic fragments have reduced affinity for IGF ligands. The net effect is an increase in IGFs availability for interaction with the IGF-1R. Thus, efforts to reduce protease action could have a beneficial effect on reducing IGF-1R activity in cancer. Proteolysis of IGFBPs has been observed to be both IGF-dependent and IGF-independent; IGF-dependent proteolysis has been observed for IGFBP2 and in IGFBP4, whereas IGFBP3 and IGFBP5 undergo IGF-independent proteolysis. K150-E161 residues in IGFBP2 are more ordered (less disorder disposition, Fig. S4 ), and our experimental results show an increase in dynamics for those residues in IGFBP2 after binding with IGF-1. As these K150-E161 residues in the linker region become more flexible after binding, the conformational ensemble populated by the linker domain of IGFBP2 shifts so that it is more readily recognized by the protease and/or is more amenable to proteolysis. This brings an interesting link for why binding to IGF-1 can be so important for proteolysis for h IGFBP2, which is not the case for IGFBP3 and IGFBP5. An example of an IGF-dependent protease action on h IGFBP2 is PAPP- A, which cleavesh IGFBP2 in an IGF-dependent manner at a single site between Gln165 and Met166, to yield two proteolytic fragments having weak IGF binding affinity.
The different susceptibilities of the different IGFBPs to proteolysis have been attributed to ligand-induced conformational changes. Our studies demonstrate that K150-E161, Q165, and M166 residues of L-h IGFBP2, which are involved in binding IGF-1, exhibit enhanced conformational exchange upon IGF binding (Fig. 4 ). Moreover, this enhanced conformational exchange is not confined to the binding site but extends to some distant residues (V110, N113, H117, H172, Q165, M166, L174, and L182) (Fig. 4 ) and is accompanied by an increase in conformational entropy. This implies distant dynamic regulation, where changes in protein dynamics induced by ligand binding extend to residues distant from the ligand-binding site, even in the absence of a well-defined conformational change. The increase in µs-ms motions of residues in the vicinity of protease cleavage sites of L-h IGFBP2 provides an interesting link to proteolytic cleavage upon binding IGF-1 as it is well known that changes in conformational dynamics upon ligand binding are important for regulation of proteolysis. In the presence of IGF-1, the conformational ensemble populated by the linker domain of IGFBP2 shifts so that it is more readily recognized by the protease and/or is more amenable to proteolysis. Considering the current findings, we therefore propose that dynamic regulation in the linker domain of IGFBP2 plays an important role in its susceptibility to PAPP-A proteolytic cleavage.
These results have significant implications for the development of IGFBPs (mutants and/or chimeras) as antagonists of IGF-1R activation that can block IGF-1R mediated tumor progression. Most current cancer therapeutics target the IGF-signaling pathway and focus on blocking the IGF-1R directly (kinase inhibitors) and/or its downstream effectors. However, a drawback of this approach is the resulting high serum IGF-1 levels in response to targeted inhibition of IGF-1R and adverse side-effects and/or toxicities arising from potential interference with the insulin pathway. It has been suggested recently that therapeutics targeting the interaction of IGFs with IGFBPs may overcome these serious drawbacks. For example, IGFBPs engineered to be protease–resistant by mutating or deleting the protease cleavage sites in the linker domain should act as IGF antagonists. Recently, in separate studies, engineered protease-resistant h IGFBP2 and h IGFBP4 were found to inhibit tumor growth in breast cancer. Interestingly, the engineered protease-resistant form of h IGFBP2 lacking residues 114-170 (des(114-170)) retains high-affinity binding to IGF-1 and IGF-2, with only a 1.6-2-fold reduction in affinity compared to full-lengthh IGFBP2. The present study may now explain the loss in binding affinity of des(114-170) towards both IGFs compared to the full-length protein by the fact that residues K150-E161 of the linker domain, which facilitate IGF binding (Fig. S7 ), were deleted from the construct. This suggests that, in addition to alteration of the protease cleavage sites, more potent IGFBP-based antagonists could be designed by considering the binding affinity of the linker domain for the IGFs and taking into account the resulting change in dynamics upon binding. These studies will facilitate the development of future IGFBP-based antagonists.
In summary, our studies of the intrinsically disordered linker domain of human IGFBP2 provide new insights into the regulatory mechanisms in the IGF system. Contrary to currently held models, the intrinsically disordered linker domain of IGFBP2 is involved in binding IGF-1. L-h IGFBP2 does not undergo a well-defined conformational change upon binding its ligand, but binding is accompanied by a significant change in dynamics on both the millisecond-microsecond and picosecond-nanosecond time scales. This is an example of functional regulation in an intrinsically disordered protein complex by dynamic regulation, which is being recognized increasingly in recent years.
Materials and Methods