Stereospecific NANOG PEST Stabilization by Pin1

Abstract

NANOG protein levels correlate with stem cell pluripotency. NANOG concentrations fluctuate constantly with low NANOG levels leading to spontaneous cell differentiation. Previous literature implicated Pin1, a phosphorylation-dependent prolyl isomerase, as a key player in NANOG stabilization. Here, using NMR spectroscopy, we investigate the molecular interactions of Pin1 with the NANOG unstructured N-terminal domain that contains a PEST sequence with two phosphorylation sites. Phosphorylation of NANOG PEST peptides increases affinity to Pin1. By systematically increasing the amount of cis PEST conformers, we show that the peptides bind tighter to the prolyl isomerase domain (PPIase) of Pin1. Phosphorylation and cis Pro enhancement at both PEST sites lead to a 5–10-fold increase in NANOG binding to the Pin1 WW domain and PPIase domain, respectively. The cis-populated NANOG PEST peptides can be potential inhibitors for disrupting Pin1-dependent NANOG stabilization in cancer stem cells.

NANOG is a critical marker for stem cell pluripotency.^1,2 Regulated by master transcription factors such as OCT4, SOX2, and KLF4,^3,4 NANOG levels are in constant flux, allowing cells transient windows of opportunity for differentiation.^1,2 We have previously shown that NANOG oligomerization via its prion-like C-terminal domain plays a critical role in NANOG stabilization and in establishing stem cell pluripotency.⁵ Phosphorylation can be another mechanism of regulating NANOG stability.^6-8 NANOG has a PEST motif⁹ in the N-terminal domain (NTD, Figure 1) that targets NANOG for ubiquitination and proteasomal degradation.⁸ Within this PEST sequence are phosphorylation sites for proline-directed kinases such as ERK1/2 and CDKs.^6,7,10 Pin1 is a phosphorylation-dependent cis–trans prolyl isomerase that acts on substrates involved in cell signaling, cell cycle control, and proliferation.^11-27 In embryonic stem cells (ESCs), Pin1 interacts with NANOG to suppress its ubiquitination and degradation.⁶ Disruption of Pin1-NANOG interaction inhibits ESC self-renewal potential.⁶

Figure 1.

NANOG NTD PEST phosphorylation. (a) NANOG disorder profile. Rapid Intrinsic Disorder Analysis Online (RIDAO) plots a disorder score per residue with scores >0.5 predicted to be disordered while residues with scores <0.5 are predicted to be ordered. RIDAO²⁸ incorporates disorder predictors PONDRs VLXT,⁴⁹ VL3,⁵⁰ and VSL2,⁵¹ as well as IUPred-Long,⁵² IUPred-Short,⁵² and PONDR-FIT.⁵³ MDP (gray area) represents the mean disorder profile. (b) NANOG structure prediction by Alphafold.^29,30 (c) Domain organization of NANOG (NTD, N-terminal domain; DBD, DNA-binding domain; WR, tryptophan repeat domain). Protein sequence of NANOG NTD residues 49–74 showing the PEST sequence (highlighted in green) and the phosphorylation sites by p38d kinase (red). (d) Superimposition of the 2D NMR ¹⁵N-HSQC spectra of unphosphorylated (green) and phosphorylated (red) ¹⁵N NANOG NTD(49–74). Arrows indicate the chemical shift changes upon phosphorylation. The minor peaks are due to minor cis populations associated with proline isomerization (e.g., pS52′ is a minor cis population).

Results and Discussion.

Our previous work has shown that human NANOG (Uniprot ID: Q9H9S0) is a predominantly intrinsically disordered protein with a structured DNA-binding domain and unstructured N-terminal and C-terminal domains.⁵ The experimental data are consistent with disorder and structural predictions by RIDAO²⁸ and Alphafold^29,30 (Figure 1a,b). To gain insights into the NANOG NTD PEST structure, we collected triple resonance NMR data using the NANOG NTD(49–74) construct (Figure 1c,d), and we determined the HN, N, Cα, and Cβ chemical shifts. The assigned NMR chemical shifts approximated random coil values³¹ (Table 1), suggesting that NANOG NTD(49–74) is unstructured. To investigate the effect of phosphorylation, we prepared in vitro phosphorylated NANOG NTD(49–74) using p38δ kinase. Phosphorylation of NANOG NTD resulted in large chemical shift changes (~1–2 ppm of Cβ, ~0.3–0.5 ppm of HN) for the phosphorylated serine residues, pS52, pS65, and pS71 (Table 1, Figure 1d). Aside from the serine residues, most of the residues display chemical shifts close to those of unphosphorylated peptide, suggesting that phosphorylation did not alter the disordered nature of the NANOG PEST sequence. Our data confirm previous mass spectrometry data⁷ that the NANOG PEST sequence can be phosphorylated by proline-directed kinases.

Table 1.

NMR Chemical Shifts of Unphosphorylated and Phosphorylated NANOG NTD (49–74; BMRB IDs 52388 and 52389, Respectively)^a

		Nan NTD (49-74)				Phos Nan NTD(49-74)
Res. No.	Res	CA	CB	N	HN	CA	CB	N	HN
49	E	56.49	30.45			56.75	30.29
50	T	62.08	69.87	116.9	8.388	61.71	69.97	115.9	8.288
51	V	62.18	32.85	123.6	8.279	61.91	33.02	122.8	8.245
52	S	56.32	63.39	121.7	8.458	55.84	65.39	122.4	8.702
53	P	62.06	32.89			62.92	32.15
54	L	53.1	41.67	123.7	8.284	53.18	41.61	123.9	8.352
55	P	63.23	32.05			63.23	32.05
56	S	58.42	64.01	115.9	8.418	58.42	64.01	115.9	8.415
57	S	58.69	63.73	117.5	8.328	58.62	63.82	117.6	8.333
58	M	55.92	32.66	121.7	8.302	55.75	32.83	122	8.275
59	D	54.79	40.95	120.9	8.164	54.59	40.95	121.4	8.248
60	L	55.32	42.18	121.7	7.979	55.29	42.36	121.8	7.998
61	L	55.32	42.18	122.6	8.1	55.29	42.12	122.8	8.117
62	I	61.14	38.43	122.2	8.02	61.09	38.45	122.4	8.043
63	Q	55.76	29.8	124.6	8.393	55.65	29.83	124.9	8.411
64	D	54.43	41.26	122.1	8.364	54.45	41.32	122.3	8.376
65	S	55.55	64.56	118.5	7.998	55.4	65.85	117.1	8.541
66	P	63.62	31.99			63.51	32.11
67	D	54.47	41.19	119.7	8.326	54.46	41.13	120.1	8.35
68	S	58.75	63.77	116.7	8.249	58.59	63.85	116.6	8.228
69	S	59.02	63.74	117.8	8.414	58.64	64.02	117.9	8.4
70	T	61.79	69.67	114.8	8.075	61.75	69.79	115.3	8.193
71	S	56.61	63.25	119.5	8.155	55.71	65.27	119.8	8.49
72	P	63.42	32.11			63.6	32.06
73	K	56.21	33.2	122.1	8.435	55.72	33.22	122.5	8.489
74	G	46.12		116.3	7.952	46.12		116.2	7.997

^aPhosphorylated serine residues are highlighted in red. Proline residues are shaded in gray.

To systematically characterize the role of phosphorylation in Pin1 (Uniprot ID: Q13526) interaction with NANOG PEST, we synthesized several peptides encompassing the PEST region (Figure 2). The use of synthetic peptides ensures completely phosphorylated and pure states for S65 and S71. Previous studies have shown that the Pin1 PPIase domain has preferred bias toward the cis Pro conformer.^32,33 Thus, we tested whether an increase in the % cis (Pro) conformer in NANOG PEST peptides correlated with an increase in binding affinities to Pin1. Various peptides were synthesized by replacing Pro residues (adjacent to the phosphorylated Ser) with pipecolic acid (pSPip and pSPippSPip), which have been shown to increase the cis population.^{34 31}P NMR spectra (phosphorylated Ser residues) confirm the increase in the cis peaks by ~10–20% (Figure 2).

Figure 2.

NANOG PEST peptides. Residue compositions and the % cis vs % trans population of the phosphorylated Ser residues preceding Pro as determined from ³¹P NMR spectroscopy.

To characterize the interaction of Pin1 with NANOG PEST peptides, NMR chemical shift perturbation (CSP) studies were performed by titrating ¹⁵N-labeled Pin1 with saturating concentrations of peptides (up to 3–4 mM, Figure 3). Consistent with previous studies of Pin1 and its substrates, we observed minimal shifts with unphosphorylated (SP) or with the pSA, indicating that the specificity of the interaction is due to phosphorylation and proline dependence of Pin1 interaction²⁵ (Figure 3a,b). Serine phosphorylation (pSP or pSPpSP peptides) results in significant backbone chemical shift changes, with most residues affected in the Pin1 WW domain (Figure 3a). Increasing the cis Pro conformers (pSPip and pSPippSPip peptides) dramatically increases the chemical shift changes for residues in the PPIase domain while maintaining similar shift changes in the WW domain. For example, representative ¹H–¹⁵N NMR HSQC spectral slices (Figure 3c) show hardly any shifts for PPIase residues M130, D153, and S115 and for pSP and pSPpSP peptides but large shifts for the cis-enhanced peptides (pSPip and pSPippSPip). This agrees with previous studies showing Pin1’s PPIase domain’s preference for cis-locked peptides.^32,33 More interestingly, our results show that increasing the cis Pro conformers in both sites (pS65 and pS71) results in even larger chemical shifts compared to modification at only one site, suggesting synergistic effects in Pin1 interaction. We determined the specific binding constants from each Pin1 domain (Figure 3d, Table 2). The singly phosphorylated NANOG peptides pSP and pSPip bound with K_d values of 320 ± 100 and 390 ± 120 uM, respectively, to the Pin1 WW domain. We were unable to obtain reliable affinities for SP, pSA (both WW and PPIase domain), pSP, and pSPip (PPIase domain only) peptides because either the chemical shifts were too small (δΔ_av < 0.05 ppm) or the binding affinities were weak (i.e., no detectable saturation). Doubly phosphorylated (pSPpSP or pSPippSPip) peptides result in 5–6-fold tighter affinity to Pin1 WW domain residues compared to singly pS65 phosphorylated (pSP or pSPip) peptides (K_d ~ 60–70 vs 300–400 μM; Table 2). This suggests that both sites can interact with Pin1 in a bivalent manner, similar to IRAK1 phosphorylated substrates.³⁵ Remarkably, cis Pro isomer enhancement at both sites (even with only a ~10% increase per site) significantly increases the affinity by ~10-fold (Figure 3d,e). Our data suggest that further increasing the cis Pro content, either by substitution to 5,5-dimethylproline (shown to increase by as much as 96% cis)³⁴ or by designing PEST-based cis-locked peptides,³³ could potentially result in stronger Pin1 binders that can be integrated in current Pin1-based drug discovery efforts.^36-38

Figure 3.

Enhanced cis NANOG PEST peptides increase Pin1 interaction. (a) Chemical shift perturbations (CSPs) of the various NANOG PEST peptides. Dashed line at δΔ_av = 0.05 ppm represents cutoff for significant chemical shift changes. (b) Mapping of the chemical shift changes (δΔ_av > 0.05 ppm) on the surface of Pin1 structure. (c) ¹⁵N-HSQC ¹⁵N Pin1 spectra (selected regions) upon titration with different concentrations of NANOG PEST peptides (pSP, pSPpSP, pSPip, pSPippSPip; top to bottom panels, respectively). The selected peaks are representative residue markers for the WW domain (W34HE1, left) and PPIase domain (M130, D153; middle panels, and S115; right panels). (d) Representative binding curves (WW domain, top; PPIase domain, bottom) for the interaction of Pin1 with different NANOG peptides. (e) Box plots (median line and 10–90 percentile range are shown) for the dissociation constants obtained from representative residues (see Table 2 for K_d and the residues; ns = not significant; p values are presented in the graph).

Table 2.

Binding Affinities of Pin1 with Different NANOG PEST Peptides (Reported Values Are in μM)

peptide	WW	PPIaseb
pSPpSP	70 ± 80a	1200 ± 1000c
pSPippSPip	60 ± 40a	16 ± 30c
pSP	390 ± 120b	ndd
pSPip	320 ± 100b	nd
SP	nd	nd
pSA	nd	nd

^aave and stdev from eight peaks: S16N, Y23N, F25H, S32N, Q33H, W34N, W34HE1, E35N.

^bave and stdev from nine peaks: S16N, G20N, R21H, Y23N, F25H, S32N, W34N, W34HE1, E35N.

^cave and stdev from nine residues: S67N, S72N, S115H, M130H, K132H, D153H, S154N (M130N and D153N for pSPpSP).

^dnd, not determined.

NANOG expression is implicated in the generation of cancer stem cells (CSC) and progression of cancers.^39-47 NANOG dosage correlates with increased metastasis and invasion of ovarian and breast cancer cells^46,48 and with castration-resistant tumor growth and regeneration in prostate cancer.⁴³ A previous study showed that disruption of Pin1-NANOG interaction inhibits cell stemness.⁶ Our study demonstrates that specific NANOG PEST cis Pro-based peptides are potential inhibitors for preventing Pin1-dependent NANOG stabilization and mitigating CSC formation and cancer progression and recurrence.