Multivalent Angiomotin‐like 1 and Yes‐associated protein form a dynamic complex

Abstract

Multivalent complexes formed between the cancer‐promoting transcriptional co‐activator, Yes‐associated protein (YAP), and proteins containing short linear motifs of type PPxY modulate cell proliferation and are attractive therapeutic targets. However, challenges producing PPxY polypeptides containing the full binding domain has limited understanding of the assembly process. Here, we successfully produced a polypeptide containing the complete set of three PPxY binding sites of Angiomotin‐like 1 (AMOTL1), a scaffolding protein that regulates the nucleo‐cytoplasmic shuttling of YAP via WW‐PPxY interactions. Using an array of biophysical techniques including isothermal titration calorimetry, size‐exclusion chromatography coupled to multi‐angle light scattering, and solution nuclear magnetic resonance spectroscopy, we show that the AMOTL1 polypeptide is partially disordered, and binds the YAP WW domains to form an ensemble of complexes of varying stabilities. The binding process is initiated by the binding of one YAP WW domain to one AMOTL1 PPxY motif and is completed by transient interactions of the second YAP WW domain with a second AMOTL1 PPxY motif to form an equilibrating mixture composed of various species having two YAP sites bound to two conjugate AMOTL1 sites. We rationalize that the transient interactions fine‐tune the stability of the complex for rapid assembly and disassembly in response to changes in the local cellular environment.

Abbreviations

CD

circular dichroism spectroscopy

HSQC

heteronuclear single quantum coherence

ITC

isothermal titration calorimetry

NMR

nuclear magnetic resonance spectroscopy

WT

wild type

1. INTRODUCTION

Yes‐associated protein (YAP) is a transcriptional co‐activator that promotes cell survival by activating cell proliferation and anti‐apoptosis. ¹ YAP‐mediated processes are linked to various cancers in which cellular levels and nuclear localization of YAP are increased, and subsequent interactions of DNA‐binding transcription factors with YAP are important first steps. ² Given their central role in cell survival, we have undertaken foundational structure–function studies to elucidate YAP regulatory mechanisms. Results will elucidate fundamental scientific questions as well as inform the design of novel small molecules for therapeutic intervention.

Multiple proteins are implicated in the regulation of YAP cellular concentrations and localization. Among them are the Motin family of proteins Angiomotin (AMOT), Angiomotin‐like 1 (AMOTL1), and Angiomotin‐like 2 (AMOTL2) which function in cell differentiation, proliferation, and migration. ³ , ⁴ , ⁵ , ⁶ , ⁷ The three proteins bind directly to YAP to regulate YAP‐promoted cell proliferation. ⁵ , ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² Key to Motin interactions with YAP are 2–3 short linear motifs (SliMs) of type L/PPxY (hereafter referred to as PPxY) which recognize tandem YAP WW domains, autonomous folding units which adopt a triple‐stranded beta‐sheet fold and are characterized by the presence of two conserved tryptophans. ¹³ Here we focus on molecular interactions between YAP and AMOTL1.

AMOTL1 is a 106 kDa protein with three PPxY motifs (P1, P2, and P3) which precede three structured domains: a Bin/Amphiphysin/Rvs (BAR) domain, a coiled‐coil (CC) domain, and a postsynaptic density protein (PDZ)‐binding domain ³ , ¹⁴ , ¹⁵ (Figure 1). High structural disorder is predicted for the PPxY segment and challenges producing large primarily disordered fragments may explain why biophysical studies of fragments containing all three PPxY sites have been limiting. An attractive workaround to produce shorter polypeptides with one or two PPxY sites has led to conflicting reports; some studies suggest that P2 is the most relevant motif ⁵ , ¹⁰ , ¹¹ ; and others demonstrate that both P1 and P2 sites function in complex formation. ⁸

FIGURE 1.

Domain architecture and solution properties of apo and bound AMOTL1 and YAP polypeptides. (a) Full‐length AMOTL1 (956‐residues) contains three PPxY motifs which precede, a Bin/Amphiphysin/Rvs (BAR), coiled‐coil (CC) and a postsynaptic density protein (PDZ)‐binding domains. The AMOTL1 A₁₂₃ construct (residues 178–384) includes all three PPxY sites. (b) Far UV CD spectrum of A₁₂₃ is consistent with a partially disordered polypeptide. (c) YAP, a multidomain protein, contains two WW domains, WW1 and WW2. The YAP construct, YWW_TD (residues 157–277) was designed to include both WW domains. (d) Far UV CD spectrum of YWW_TD shows the characteristic spectrum of a WW domain. (e) SEC‐MALS elution profiles of A₁₂₃ (gray line), YWW_TD (black line), and a 1:1 mixture of both proteins (dashed black line). (f) A plot of the SEC‐MALS‐detected molar mass of varying ratios of A₁₂₃:YWW_TD complexes. The average molar mass values range between 34.9–37.8 kDa. The theoretical molar masses of A₁₂₃, YWW_TD, and the 1:1 A₁₂₃‐YWW_TD complex are indicated by dashed lines

Our experiments resolve the question of which PPxY motifs are required for in vitro assembly of the YAP‐AMOTL1 complex. A battery of formidable and mutually reinforcing molecular biophysical methodologies are employed, and the experimental results are collectively interpreted in the context of the function of the YAP‐AMOTL1 complex.

2. RESULTS

2.1. SEC‐MALS and CD experiments of apo and bound proteins

In these experiments, we use A₁₂₃, a multivalent 207‐residue AMOTL1 polypeptide (residues 178–384) with three PPxY motifs—LPTY, PPEY, and PPEY (Figure 1a). The far ultra‐violet (UV) circular dichroism (CD) spectrum of A₁₂₃ (Figure 1b) shows a strong signal at 204 nm, suggestive of random coil‐like structure, and a relatively weak signal at 222 nm indicative of nascent helical structure. The experimental CD spectrum was analyzed with the webserver, DichroWeb, ¹⁶ , ¹⁷ to provide calculated secondary structure contents of 24.7% helix, 0.6% strands, 15.9% turns, and 43.7% disordered segments.

The YAP polypeptide, YWW_TD, (residues 157–277, Figure 1c) contains both WW domains (WW1 and WW2). The far UV CD spectrum of YWW_TD (Figure 1d) shows a positive peak at 230 nm and a negative peak at 220 nm which are signature WW domain peaks attributed to the packing of the two tryptophans and the β‐strands, respectively. Taken together, the CD analysis is consistent with a partially folded A₁₂₃ polypeptide, and a largely folded YWW_TD polypeptide.

Size‐exclusion chromatography (SEC) combined with multi‐angle light scattering (MALS) can be used to determine the molar mass of a protein without the need for a column calibration curve. ¹⁸ Completely or partially disordered proteins have relatively large hydrodynamic radii, which precludes the use of SEC alone to infer molecular mass, whereas SEC‐MALS is particularly attractive for computing their molar mass. The average molecular mass of A₁₂₃ probed by SEC‐MALS is 26.1 ± 1.2 kDa, a value close to the theoretical monomer molecular mass of 25.9 kDa. The SEC‐MALS calculated molar mass of YWW_TD is 14.1 ± 1.1 kDa (expected monomer molar mass is 14.2 kDa) (Figure 1e; Table 1). An equimolar mixture of A₁₂₃ and YWW_TD elutes earlier than unbound A₁₂₃ or YWW_TD with an SEC‐MALS‐measured molecular mass of 35 kDa, slightly less than the theoretical molar mass of 40.1 kDa expected for a 1:1 complex. When resolved on SDS‐PAGE, the peak for the equimolar mixture migrates as two bands with migration rates similar to unbound A₁₂₃ and YWW_TD (data not shown). A₁₂₃ and YWW_TD complexes formed by mixing 0.25–6‐fold molar excess of YWW_TD with A₁₂₃ give SEC‐MALS measured molecular masses in the range 34.9–37.8 kDa (Figure 1f), values slightly less than the theoretical molar mass of 40.1 kDa for a 1:1 complex (Table 1). Taken together, the SEC‐MALS data indicate that under the experimental conditions, the A₁₂₃ and YWW_TD polypeptides are monomeric in solution, and the YWW_TD‐A₁₂₃ complex is formed between one YWW_TD molecule and one A₁₂₃ molecule.

TABLE 1.

Experimental and theoretical molar masses of apo and bound AMOTL1 A₁₂₃ and YAP YWW_TD

Protein	Stoichiometry	Molar mass (kDa)
		Experimental	Theoretical
A123 (AMOTL1)	–	26.1 ± 1.2 a	25.9
YWWTD (YAP)	–	14.1 ± 1.1 a	14.2
YWWTD‐A123 complex	0.25:1	34.9	–
	0.5:1	37.3	–
	0.75:1	37.2	–
	1:1	35 ± 3 a	40.1
	2:1	36.4 ± 0.7 a	54.3
	3:1	36.8	68.5
	4:1	37.0	82.7
	5:1	36.8	96.9
	6:1	37.8	111.1

^a Reported value is the average of triplicate experiments.

2.2. Solution NMR experiments of apo and YAP‐bound AMOTL1 A₁₂₃

NMR experiments were used to determine for A₁₂₃ the local secondary structure propensity, backbone dynamics, and binding interface residues. The ¹H‐¹⁵N heteronuclear single quantum coherence (HSQC) spectrum of A₁₂₃ shows poor dispersion in the NH‐region, as expected for a protein with random coil‐like and/or helical structure (Figure 2a). Despite significant peak overlap, we assigned 79% (145 of 183 non‐proline residues) of backbone nuclei, which enabled residue‐specific analyses. Resonance assignments for the 26 additional residues that are part of the expression vector are not included in the analysis.

FIGURE 2.

NMR backbone assignments and solution dynamics of AMOTL1 A₁₂₃ polypeptide. (a) ¹H‐¹⁵ N‐HSQC of A₁₂₃ showing assignments for 145 of 183 non‐proline residues. (b) A plot of the deviation of the chemical shifts from random coil values (Δ13Cα − Δ13Cβ). (c) CLEANEX‐HSQC analysis of A₁₂₃. The gray bars represent the rapidly exchanging amide protons, and the black dots are the relatively slow exchanging amide protons. (d) A plot of the steady‐state heteronuclear NOE (Het‐NOE) values. (e) The ratio of the transverse and longitudinal relaxation (R ₂/R ₁) as a function of the residue number. Average values are shown as dashed lines in plots b, d, and e. The predicted secondary structure of A₁₂₃ is shown above the plots as a line diagram where disordered segments are represented by a solid line. Three predicted helices, H1 (residues 191–202) and H2 (residues 247–254), and H3 (residues 261–272) are located between the first (P1) and second (P2) PPxY sites

PSIPRED, ¹⁹ a sequence‐based secondary structure prediction algorithm, shows that the A₁₂₃ polypeptide is largely disordered except for three short helical segments designated H1, H2, and H3 (Figure 2). The experimental secondary structure propensities of A₁₂₃ were determined from the relative deviations of assigned CA and CB chemical shift values from standard random coil values (∆CA‐∆CB). A plot of the ∆CA‐∆CB as a function of the residue number (Figure 2b), shows sequential and substantial positive values for residues 193–197, 261–263, 265–267, and 269–270, within the predicted helical segments H1 and H3. An additional stretch of residues (245–257) within the predicted H2 α‐helical segment has positive deviations slightly below the computed average value of 0.4 (Figure 2b, dashed line), suggestive of a relatively weak or nascent helix. Using the backbone chemical shifts and the δ2D analysis software, ²⁰ we determined the probability distribution of secondary structural elements as 69.1% coil, 15.8% polypropylene II helix, 8.0% regular helix, and 7.1% strands consistent with the conclusion from CD analysis that A₁₂₃ has limited folded structure.

To identify solvent‐exposed residues in the A₁₂₃ polypeptide, a clean chemical exchange (CLEANEX)‐HSQC experiment which measures fast backbone amide proton exchange with water was recorded. The plot in Figure 2c shows residues with rapidly exchanging amide protons (gray bars) and solvent‐protected residues (black circles). The latter residues are not detected in the CLEANEX‐HSQC spectrum (Figure S1). As expected for a predominantly unstructured polypeptide, most residues in A₁₂₃ are accessible to solvent, with the most rapidly exchanging amide protons (residues with the highest intensities) within the linker segments connecting the P1 and P2 sites. Consistent with a predicted helical segment, the amide protons for residues in the H2 segment are protected from solvent.

To determine backbone dynamics at multiple timescales, we measured a suite of NMR experiments, including steady‐state ¹H‐¹⁵N heteronuclear NOE (HetNOE), ¹⁵N longitudinal (R ₁) and transverse (R ₂) relaxation. HetNOE measurements are sensitive to the strength of the magnetic field and report backbone dynamics in the pico‐ to nanosecond (ps‐ns) timescale. Generally, positive and negative HetNOE values are indicative of restricted and mobile residues respectively, but at high magnetic field strengths, mobile residues may display low positive values. ²¹ HetNOE values for the A₁₂₃ polypeptide, at a magnetic field strength of 800 MHz, are mostly positive, with an average of 0.4 (Figure 2d). At this relatively low positive average value, most segments of residues are mobile. The most mobile residues, 179 and 384 (negative NOE value), are at the N and C‐termini. Intriguingly, the first residue of the native protein sequence (178) has an NOE value greater than the average value of 0.4. A possible explanation for this anomaly is that the non‐native poly‐histidine sequence adjacent to 178 restricts the motion of this residue.

The ratio of the transverse and longitudinal relaxation (R ₂/R ₁) reflects motions in the nanosecond timescale and identifies segments with slower tumbling. The plot in Figure 2e shows relatively high values for residues 190–198 and 259–271 which implies relatively slower tumbling for the two segments.

A₁₂₃ residues involved in binding YWW_TD were mapped by NMR titration experiments in which unlabeled YWW_TD and isotopically labeled A₁₂₃ were mixed at molar ratios in the range of 0.25:1–2:1 (YWW_TD:A₁₂₃). As a reporter of the binding interactions, we monitored changes in the intensities of peaks corresponding to the tyrosine residues, Y191, Y313, and Y370, at the 3 PPxY sites. The plot in Figure 3a shows a decrease in the intensity of each tyrosine peak with increasing concentrations of added YWW_TD. Note that peaks for all three tyrosine residues are not detected in the partially bound 0.75:1 (YWW_TD:A₁₂₃) complex. Loss in peak intensity is attributed to binding‐induced exchange broadening, binding‐induced conformational changes and/or slower tumbling of the complex. ²² Line broadening occurs, and peaks disappear when the rate of exchange between the bound and unbound conformations is intermediate on the NMR timescale.

FIGURE 3.

NMR mapping of the A₁₂₃‐YWW_TD binding interface. (a) Changes in peak intensities of the three tyrosine residues, Y191 (P1), Y313 (P2), Y370 (P3), of each motif as a function of YWW_TD‐A₁₂₃ molar ratio. (b) Normalized intensities of assigned A₁₂₃ residues plotted as a function of the residue number for apo A₁₂₃ (black dashed lines), and upon addition of 0.75 (top panel, green), 1 (middle panel, blue), and 2 (bottom panel, magenta) molar equivalents of YWW_TD. The predicted secondary structure is shown above the plots. At the final molar equivalents of 2:1 (bottom panel), only residues primarily located in the P1‐P2 linker segment are observed in the spectrum. (c) A bar plot of the intensities of peaks in the CLEANEX‐HSQC spectrum of unbound A₁₂₃. Peaks that disappear upon addition of 0.25 molar equivalents of YWW_TD are shown in cyan and include residues at the three PPxY sites (shaded in pink)

Complex formation also induces resonance intensity changes in other residues and leads to many missing peaks in spectra of the YAP‐bound protein (Figure S2). These missing peaks imply that the residues are close to binding interfaces and/or involved in binding‐induced allosteric changes. ²³ At the 0.75:1 molar ratio (Figure 3b, upper panel), missing cross‐peaks map to residues in the sequence vicinity of P1 (186–204), H2 (261, 265, 268 and 273), P2 (305–317), P3 (365–371), and the linker segment between P2 and P3 (322, 335, 340, 342). Additional cross‐peaks corresponding to residues 245, 262–264, 269–274 and C‐terminal residues 302–384 (except 328, 349, 358) are missing in the spectrum of the 1:1 YWW_TD‐A₁₂₃ complex (Figure 3b, middle panel). At the final molar ratio of 2:1 (YWW_TD: A₁₂₃), the only remaining spectral peaks correspond to residues primarily located in the P1‐P2 linker segment (Figure 3b, lower panel).

To identify protected segments in the partially bound A₁₂₃ polypeptide, the CLEANEX‐HSQC experiment recorded for the unbound A₁₂₃ polypeptide (Figure 2c) was compared to a CLEANEX‐HSQC experiment recorded on a partially bound protein at a YWW_TD:A₁₂₃ molar ratio of 0.25:1. Peaks that disappear upon addition of 0.25 molar equivalents of YWW_TD are mapped onto the CLEANEX‐HSQC data of unbound A₁₂₃ and are shown as cyan bars in Figure 3c. Amide protons which are most protected from the solvent map to residues in the sequence vicinity of P1, P2, P3, and the predicted H2 helical segment.

To summarize the NMR titration and CLEANEX‐HSQC data, the partially bound A₁₂₃ polypeptide formed by addition of a sub‐stoichiometric concentration of YWW_TD to A₁₂₃ results in disappearance of peaks in the sequence vicinity of all three PPxY motifs, which indicates that YWW_TD binds all three sites concurrently. Most of the missing peaks in the stoichiometric complex are in the sequence vicinity of the P1 site, which suggests more favorable interactions at the P1 site. Further, peaks corresponding to most residues in the predicted helix (H3), which is not a putative binding site, also disappear, possibly due to binding‐induced conformational changes. Finally, spectral peaks remaining in the fully bound complex map to the linker segment between P1 and P2; an indication that residues in this linker region of A₁₂₃ remain largely disordered in the YAP‐bound protein.

2.3. ITC experiments of WT and site‐directed mutants

Reaction thermodynamics, stoichiometry, and effective binding affinity of the A₁₂₃‐YWW_TD interaction were determined by ITC. Binding of the two proteins is enthalpically driven, occurs with a binding stoichiometry (N) close to 1:1 and has an effective dissociation constant (K _d) of 0.26 ± 0.01 μM (Figure 4a; Table 2). Because there are two putative binding sites on the YWW_TD polypeptide (2 WW domains) and three putative binding sites on the A₁₂₃ polypeptide (3 PPxY motifs), binding of a single WW‐PPxY site or of both WW sites to two PPxY sites will result in a binding stoichiometry of 1:1.

FIGURE 4.

ITC binding isotherms of the A₁₂₃‐YWW_TD interaction. Representative binding isotherms for interactions of A₁₂₃ with (a) YWW_TD (b) P202A_TD and (c) P261A_TD or YWW_TD with A₁₂₃ mutants (d) A_Δ1, (e) A_Δ2 (f) A_Δ3 in which sites P1, P2 and P3, respectively are inactivated. Active domains or motifs are labeled in the schematic of the constructs above each binding isotherm. ITC data were collected at 25°C in pH 7.5 buffer composed of 50 mM sodium phosphate, 50 mM NaCl, 5 mM β‐mercaptoethanol, and 0.5 mM NaN₃

TABLE 2.

Thermodynamics parameters for the YAP‐AMOTL1 interactions

Titrant	Cell	N	K d (μM)	ΔH (kcal/Mol)	TΔS (kcal/Mol)	ΔG (kcal/Mol)
YWWTD	A123	1.1	0.26 ± 0.01	−39.6 ± 0.9	−30.3 ± 0.9	−8.98 ± 0.03
	AΔ1 a	1.0	1.8	−33.8	−26.0	−7.8
	AΔ2	1.1	0.6 ± 0.1	−35.5 ± 0.3	−27.1 ± 0.3	−8.47 ± 0.01
	AΔ3 a	1.2	0.11	−41.1	−31.8	−9.5
P202ATD	A123	1.0	7.6 ± 0.2	−37 ± 4	−30 ± 4	−6.98 ± 0.01
P261ATD	A123	1.2	0.70 ± 0.02	−30 ± 2	−22 ± 2	−8.40 ± 0.02

^a Reported value is the average of duplicate experiments. All other values are the average of triplicate experiments.

To determine whether a single YAP WW domain is sufficient for binding, we introduced mutations that inactivate either the first (P202A_TD) or second (P261A_TD) YAP WW domain. Binding of A₁₂₃ to the P202A_TD mutant occurs with a K _d of 7.6 μM (Figure 4b), a value ~10‐fold higher than 0.7 μM computed for the A₁₂₃‐P261A_TD interaction (Figure 4c). The binding stoichiometry of 1:1 or 1:1.2 implies that in each case the active WW domain binds at least one A₁₂₃ PPxY. However, the weaker but significantly different binding affinities of the mutants relative to the WT YWW_TD polypeptide suggest that both YAP WW domains contribute to the stability of the WT YWW_TD:A₁₂₃ complex.

To determine if inactivating specific PPxY sites destabilizes the A₁₂₃‐YWW_TD complex, we used A₁₂₃ “knockout” variants, A_Δ1, A_Δ2, and A_Δ3, each with one inactivated motif (indicated by subscript). Figures 4d–f show that the reactions are enthalpically driven and have a binding stoichiometry in the range of 1:1 to 1:1.2 for each “knockout” site (Table 2). The effective K _d varies from 1.8 μM in A_Δ1 (Figure 4d) to 0.6 μM in A_Δ2 (Figure 4e) and 0.1 μM in A_Δ3 (Figure 4f; Table 2). In summary, the binding of YWW_TD to A₁₂₃ is more stable (lower K _d) when both the YAP WW1 domain and AMOTL1 P1 sites are active. Inactivating the P2 site leads to a modest decrease (higher K _d) in the stability of the complex, while inactivating the P3 site has the opposite effect, that is, a modest increase (lower K _d) in the stability of the complex. Finally, the non‐integer binding stoichiometries are indicative of inhomogeneity in the complexes formed, possibly due to variations in the number of occupied binding sites.

3. DISCUSSION

Direct binding of the AMOTL1 scaffold protein to YAP regulates cell proliferation. While it is known that assembly of the complex occurs via multivalent WW‐PPxY interactions, and studies with shorter polypeptides containing one or two PPxY sites have provided some insight into the binding interactions, it remains unknown how polypeptides containing the full binding domain interact, specifically which pairs of the three PPxY sites bind the two YAP WW domains. We have clarified the uncertainty as to which AMOTL1 PPxY motifs are required for the assembly of its complex with YAP. Our first advancement was to produce a construct of AMOTL1 (A₁₂₃), which contains the entire multivalent PPxY segment rather than only one or two PPxY sites, and to use site‐directed mutagenesis of A₁₂₃ to inactivate specific PPxY sites. Then we characterized interactions of the tandem WW domains of YAP (YWW_TD) with A₁₂₃ and its PPxY variants. Our conclusions are drawn from collective analyses of data measured by an extensive array of methodologies applied both to apo proteins and to the complex formed by A₁₂₃ and YWW_TD.

We conclude that the apo AMOTL1 PPxY domain, A₁₂₃, has a limited folded structure, localized in two short helical segments, residues 193–197 and 245–257 as shown by CD and NMR. This first experimental demonstration that the full AMOTL1 PPxY segment is partially disordered is consistent with the structures of other multivalent PPxY proteins. ²⁴ , ²⁵ , ²⁶ Second, the AMOTL1‐YAP PPxY‐WW complex is formed by one molecule of AMOTL1 bound to one molecule of YAP, as shown by the hydrodynamic method of SEC‐MALS and ITC. Third, binding of the YAP tandem WW domains to A₁₂₃ perturbs residues at all three PPxY sites as shown by solution NMR spectroscopy. More residues are perturbed in the sequence vicinity of the P1 site, a clear indication of more intermolecular interactions at P1 relative to P2 or P3, and evidence that P1 is critical for the assembly of the AMOTL1‐YAP complex. The P1 site in the AMOTL1 paralog, AMOT, is also indispensable for binding specific WW domain targets ⁸ and plays a critical role in the function of the protein. ²⁷ Likewise, the P1 site in AMOTL1 may provide some functional advantages to the protein. Fourth, inactivating the P3 site results in a ~2‐fold binding enhancement which implies that P3 contributes negative entropy to the overall interaction. We speculate that the negative entropy may be eliminated when the P3 site is occupied by another WW domain protein such as the kidney and brain expressed protein KIBRA, which is reported to preferentially bind the P3 site in AMOT. ⁷ Inactivating the YAP WW1 domain also de‐stabilized the complex to a much greater degree than the WW2 domain, findings in agreement with the different binding preferences of the YAP WW domains. ²⁸

Considered together, the novel information from these complementary methodologies indicates that the AMOTL1‐YAP PPxY‐WW complex is structurally dynamic, and the YAP WW tandems interact with all three PPxY sites in AMOTL1 as illustrated by the model in Figure 5. The dynamic structure is composed of different complexes of varying stabilities as previously noted for other WW‐PPxY complexes. ²⁴ , ²⁶ Complexes in which the YAP WW domains bind AMOTL1 sites P1‐P2 or P1‐P3 are the most stable; and the least stable complex is formed by relatively weak or transient binding of the YAP WW domains to AMOTL1 sites P2‐P3. Simultaneous binding of both YAP WW domains to varying pairs of AMOTL1 PPxY sites leads to the formation of an equilibrating mixture of interconverting species transiently formed by two YAP sites bound to two conjugate AMOTL1 sites.

FIGURE 5.

Model of the species formed in the AMOTL1‐YAP complex. The YAP tandem WW domains and the predominantly unstructured AMOTL1 PPxY domain‐containing three PPxY sites, designated P1, P2, and P3, form a dynamic ensemble of interconverting complexes. Complexes in which AMOTL1 sites P1‐P2 or P1‐P3 are bound are relatively more stable than the complex in which sites P2‐P3 are bound

3.1. Functional implications of a dynamic YAP‐AMOTL1 complex

A hallmark of regulatory complexes in signaling pathways is their rapid and spontaneous response to cellular signals. The modest micromolar binding affinity of the dynamic complex can facilitate rapid assembly and disassembly of the complex in response to changes in the local environment. Further, recent findings that link YAP, ²⁹ multivalency, intrinsic disorder, and low complexity regions, to phase separation and the formation of membraneless organelles in human cells ³⁰ , ³¹ , ³² , ³³ could organize the ensemble in space. Phase separation may be triggered by changes in the local environment such as a change in the local concentration of YAP. The presence of another WW domain‐containing protein could also shift the dynamic equilibria in favor of specific complexes. For instance, AMOTL1 forms a ternary via WW‐PPxY interactions with both YAP and the kidney and brain expressed protein KIBRA ³⁴ ; and simultaneous binding to both WW domain partners may shift the dynamic equilibria in favor of specific complexes in the ensemble. We propose that forming dynamic complexes underlie AMOTL1 function.

4. MATERIALS AND METHODS

4.1. Cloning of constructs

The genes encoding AMOTL1 (Uniprot ID Q8IY63, residues 178–384, designated A₁₂₃) and YAP, (Uniprot ID P46937‐2, residues 157–277, designated YWW_TD) were inserted into a modified pET24 vector (Millipore‐Sigma, MA, USA) with an N‐terminal His₆ tag and a tobacco‐etch virus (TEV) protease cleavage site. A₁₂₃ mutants with a single Tyr to Ala substitution of the first (Y191A, designated A_Δ1), second (Y313A, designated A_Δ2), or third (Y370A, designated A_Δ3) L/PPxY motif; YWW_TD mutants with Pro to Ala substitutions in the first (P202A_TD) or second (P261A_TD) WW domains were generated using the Q5® site‐directed mutagenesis kit (New England Biolabs, MA). Tyr to Ala substitutions in L/PPxY motifs or Pro to Ala substitutions in WW domains inactivate the mutant L/PPxY motif or WW domain. ²⁵ , ³⁵ The YAP cDNA was a gift from Kunliang Guan (Addgene plasmid # 24637).

4.2. Recombinant protein production

Escherichia coli BL21 (DE3) cells with the gene of interest were cultured at 37°C in terrific broth (A₁₂₃ proteins) or lysogeny broth (YWW_TD proteins). For isotopically labeled proteins, cells were cultured in MJ9 supplemented with ¹⁵NH₄Cl and ¹²C or ¹³C glucose. Proteins were overexpressed at 20°C for 5 (A₁₂₃ variants) or 16 (YWW_TD proteins) hours after induction with 0.1 mM IPTG. His₆‐tagged recombinant proteins were purified as previously reported. ²⁵ The A₁₂₃ polypeptides were prone to proteolytic degradation, therefore, to minimize degradation the purification tag which is part of the expression vector was not removed from the N‐terminal end of the polypeptide. This introduced 26 non‐native residues to the N‐terminal end of the A₁₂₃ polypeptides. A₁₂₃ proteins expressed in MJ9 accumulated in inclusion bodies and were purified following protocols described elsewhere. ³⁶ Protein concentrations were determined from the absorbance at 280 nm and extinction coefficient values computed from the protein sequence (http://web.expasy.org/protparam/).

4.3. Circular dichroism

Far UV circular dichroism (CD) measurements were recorded at 25°C on a JASCO 720 spectrophotometer using a path length of 1 mm, and a bandwidth of 1.0 nm. Prior to data collection, the protein was dialyzed against 10 mM sodium phosphate, pH 7.5 with or without 100 mM NaF. The final concentration of the protein for CD analysis was 3 or 5.4 μM and reported CD data are the average of experimental repeats.

4.4. Size‐exclusion chromatography‐multi‐angle light scattering (SEC‐MALS)

The average molar masses and association states of the proteins were determined from SEC (AKTA FPLC; GE Healthcare) connected to multi‐angle light scattering (DAWN; Wyatt Technology) and refractive index (Optilab; Wyatt Technology) detectors. 100 μL of proteins in the concentration range of 25–150 μM were injected at a flow rate of 0.7 or 1.0 mL/min onto a Superdex²⁰⁰ 10/300 (Cytiva life sciences) chromatographic column pre‐equilibrated with a pH 7.5 buffer composed of 50 mM sodium phosphate, 0.4 M NaCl, 1 mM NaN₃, 5 mM β‐mercaptoethanol. Average molar masses were computed with the ASTRA software package, version 8 (Wyatt Technologies).

4.5. NMR data collection and analysis

NMR experiments were performed at 10°C on a Bruker Avance III, 800 MHz spectrometer (Bruker BioSpin) equipped with a triple resonance cryogenic probe. Data were collected on isotopically labeled A₁₂₃ at concentrations of 75 or 200 μM and in a pH 6.8 buffer composed of 50 mM sodium phosphate, 100 mM NaCl, 50 mM arginine, 50 mM glutamate, 1 mM NaN₃, and 2 mM tris(2‐carboxyethyl) phosphine. All samples contained 10% D₂O as the lock nucleus and 0.5% DSS for internal referencing.

Backbone resonances were assigned from BEST‐TROSY ¹H‐¹⁵N HSQC experiments ³⁷ and triple resonance (3D) experiments HNCACB, HN(CO)CACB, HNCO, HN(CA)CO, and ¹H‐¹⁵N HSQC‐TOCSY. The TROSY‐based pulse sequences were used to reduce line broadening due to rapid signal decay. All 3D experiments used non‐uniform sampling (NUS) to reduce data collection times. The NUS data were reconstructed using the iterative shrinkage thresholding approach in NMRPipe. ³⁸ NMR spectra were processed in NMRPipe ³⁸ and visualized with Sparky ³⁹ or NMRView. ⁴⁰

Per‐residue secondary structure propensities were calculated from the deviations of experimental CA and CB chemical shifts from the random coil values of Poulsen et al (https://spin.niddk.nih.gov/bax/nmrserver/Poulsen_rc_CS/). ⁴¹ , ⁴² , ⁴³ Substantial positive deviations greater than a pre‐determined mean value, for four or more sequential residues indicate helical propensity while negative deviations for three or more sequential residues suggest an extended structure.

Heteronuclear NOE (HetNOE), longitudinal (R ₁), and transverse (R ₂) spin relaxation rates were measured using TROSY‐based interleaved pulse sequences. ⁴⁴ HetNOE experiments were collected with and without proton saturation using a relaxation delay of 8 s. NOE errors were calculated as previously reported. ²⁵ R ₁ spin relaxation rates were measured with relaxation delays of 0.02, 0.06 (×3), 0.1, 0.2, 0.4, 0.6, 0.8, and 1.2 s. R ₂ spin relaxation rates were obtained with delays of 17, 34 (×3), 51, 68, 85, 140, 170, and 240 ms with a recycle delay of 1.5 s. R ₁ and R ₂ data were fit to a single exponential decay function, I(t) = I ₀e^−Rt , where t is the variable relaxation delay and R is the relaxation rate. Rates were computed using the rate analysis tool in NMRView. ⁴⁰ Experimental errors were estimated from standard deviations of triplicate (×3) experiments. Results are reported only for unambiguously assigned resonances with reliably quantified peak intensities.

CLEANEX NMR experiments were collected with a mixing time of 100 ms using a recycle delay of 1.5 s. The A₁₂₃‐YWW_TD complex was formed by adding 0.25 molar equivalents of YWW_TD to A₁₂₃.

For NMR titration experiments, unlabeled YWW_TD was added to ¹⁵N‐labeled A₁₂₃ to final molar ratios (unlabeled: ¹⁵N‐labeled:) of 0.25:1, 0.5:1, 0.75:1, 1:1, and 2:1. To correct for minor variations in sample concentrations at each molar ratio, peak intensities (measured as peak height) were normalized as the ratio of the intensity of the peak in the bound spectrum to the intensity of the peak in the unbound spectrum.

4.6. Isothermal titration calorimetry (ITC)

A VP‐ITC instrument (Malvern Instruments Inc., MA) set to 25°C, was used to record ITC data. Prior to the titrations, all samples were extensively dialyzed against a pH 7.5 buffer composed of 50 mM sodium phosphate, 50 mM NaCl, 0.5 mM NaN₃, and 5 mM β‐mercaptoethanol. Twenty seven or 28 injections of 96–148 μM WW proteins were titrated into 6–16 μM A₁₂₃ or its variants. Data were collected in duplicates, triplicates, or quadruplicates, using proteins from two different preparations. Reported data are the average of experimental repeats. Isotherms were analyzed by single‐site fits of the thermograms using the Origin 7.1 software. The free energy of binding (ΔG) was calculated from the equation ΔG = −RTln (K _a), where R is the universal gas constant, T is temperature in Kelvin, and K _a is the association constant.

4.7. delta2D method

The residue‐level disorder probability of A₁₂₃ was calculated with CA, CB, CO, N, and HN chemical shifts using the delta2D web server. ²⁰

4.8. DichroWeb method

The percent helicity of A₁₂₃ was calculated using DichroWeb ¹⁷ with data collected at 25°C on a 5.4 μM sample in buffer composed of 10 mM sodium phosphate and 100 mM NaF at pH 7.5. Reported helicity was obtained the Contin‐LL method and the reference set SP175. ⁴⁵

CONFLICT OF INTEREST

The authors declare that they have no conflicts of interest with the contents of this article.

AUTHOR CONTRIBUTIONS

Amber Vogel: Conceptualization (equal); formal analysis (equal); investigation (lead); methodology (equal); validation (equal); visualization (equal); writing – original draft (equal); writing – review and editing (equal). Alexandra Crawford: Investigation (supporting). Afua Nyarko: Conceptualization (lead); formal analysis (equal); funding acquisition (lead); methodology (equal); project administration (lead); resources (lead); supervision (lead); validation (equal); visualization (equal); writing – original draft (equal); writing – review and editing (lead).

DATA DEPOSITION

NMR chemical shifts for AMOTL1 (residues 178–384) have been deposited in the Biological Magnetic Resonance Data Bank (BMRB) under accession code 51029.

Supporting information

Appendix S1.

Vogel A, Crawford A, Nyarko A. Multivalent Angiomotin‐like 1 and Yes‐associated protein form a dynamic complex. Protein Science. 2022;31(5):e4295. 10.1002/pro.4295