The ¹H, ¹⁵N and ¹³C resonance assignments of the low-complexity domain from the oncogenic fusion protein EWS-FLI1

Abstract

The RNA-binding protein EWS is a multifunctional protein with roles in the regulation of transcription and RNA splicing. It is one of the FET (FUS, EWS and TAF15) family of RNA binding proteins that contain an intrinsically disordered, low-complexity N-terminal domain. The FET family proteins are prone to chromosomal translocations, often fusing their low-complexity domain with a transcription factor derived DNA-binding domain, that are oncogenic drivers in several leukemias and sarcomas. The fusion protein disrupts the normal function of cells through non-canonical DNA binding and alteration of normal transcriptional programs. However, the exact mechanism for how the intrinsically disordered domain contributes to aberrant DNA binding and abnormal transcription is unknown. The purification and ¹H, ¹³C, and ¹⁵N backbone resonance assignments of the amino terminal domain comprising 264 residues of EWS is described. This segment is common to all known EWS-fusions that are the hallmark of the pediatric cancer Ewing sarcoma. This domain is intrinsically disordered and features significant sequence degeneracy resulting in spectra with poor chemical shift dispersion. To alleviate this problem, the domain was divided into three overlapping fragments, reducing the complexity of the spectra and enabling almost complete backbone resonance assignment of the full domain. These solution NMR chemical shift assignments represent the first steps towards understanding, at atomic resolution, how the low-complexity domain of EWS contributes to the aberrant functions of its oncogenic fusion proteins.

Biological Context

The FET family proteins (FUS, EWS, TAF15) are RNA binding proteins that share similar sequence features (low-complexity, SYGQ repeats, RRM and RG/RGG motifs) and have been shown to form membraneless organelles through the process of liquid-liquid phase separation (Schwartz et al. 2015). FET proteins have been implicated in the progression of ALS and other neurodegenerative diseases and as contributors to oncogenic fusion proteins central to malignancies like prostate cancer, leukemias and small round blue cell sarcomas. Natively, RNA-binding protein EWS (EWS) is vital for regulating transcription and splicing, may serve as a transcriptional repressor and has been implicated in the initiation of the DNA damage repair response (Gorthi et al. 2018; Boulay et al. 2017). Human EWS is a 656 residue protein containing an N-terminal low-complexity domain rich in SYGQ repeats, (1–280), a glycine rich region (283–348), an RNA recognition motif (360–446), two RGG regions (360–446, 548–640) that flank a zinc finger motif (517–548) and a C-terminal PY motif that functions as a nuclear localization signal (Leemann-Zakaryan et al. 2011). The structure of the RRM domain has been solved by solution NMR (PDB: 2CPE).

In Ewing sarcoma, an aggressive soft tissue and bone cancer that is primarily diagnosed in children and young adults, the N-terminal third of EWS (1–264) is fused, through a balanced chromosomal translocation, to the DNA-binding domain (DBD) of an ETS family transcription factor, most commonly FLI1 (Grunewald et al. 2018). The presence of the fusion gene remains the diagnostic mark for Ewing sarcoma since the tumor histology is indistinguishable from many other small round blue cell sarcomas (Machado et al. 2011). The oncogenic activity of the EWS-FLI1 fusion is thought to partially reside with aberrant binding of the FLI1 DBD to upstream GGAA-repeat microsatellites where it acts like a pioneer factor, enabling dysregulated gene expression (Gangwal et al. 2008). Alteration of the transcriptional landscape is evident from upregulation of genes implicated in the development and progression of Ewing sarcoma. Additionally, EWS-FLI1 may dysregulate normal ETS-controlled genes as well as act as a dominant negative, interfering with the native functions of EWS (Jully et al. 2012).

Previous studies have determined that the low-complexity domain (LCD) of EWS is centrally important to the oncogenic potential of the EWS-FLI1 fusion, yet the mechanistic details of its role in EWS-FLI1 pioneering-like and transcription factor-like functions remain to be determined (Boulay et al. 2017). Divergent theories concerning the role of the EWS LCD in EWS-FLI1 have been proposed that implicate liquid-liquid phase separation mediated by the prion-like SYGQ-rich segments of the EWS LCD. The SYGQ-repeats are thought to be central to the aberrant binding of EWS-FLI1 to DNA GGAA microsatellites and subsequent recruitment of chromatin remodelers and transcriptional machinery although other studies have failed to identify significant condensates at these sites (Boulay et al. 2017; Riggi et al. 2014; Chong et al. 2018).

EWS-FLI1 type I and type II fusions that incorporate EWS residues 1–264 occur with a higher frequency than the rarer type III fusion (EWS 1–280) in Ewing sarcoma tumors (Zucman et al. 1993). Therefore, an expression construct representing EWS 1–264 was designed for NMR studies and resonance assignment. Due to the severe sequence degeneracy and low-complexity amino acid composition including the SYGQ repeat regions of EWS 1–264, three smaller overlapping constructs were designed to facilitate resonance assignments. These constructs roughly recapitulate the two SYGQ repeat regions (1–120 and 171–264) with a third fragment (91–199) that is devoid of SYGQ repeats. Here the backbone resonance assignment of the N-terminal 264 low-complexity residues common to both EWS and EWS-FLI1 are reported. The assignments presented here are the first step toward acquiring atomic level mechanistic data that will enable determination of the function of the EWS-LCD in the context of EWS-FLI1 function.

Methods and Experiments

Expression and Purification of N-terminal EWS Constructs

Sequences of the N-terminal 264 residues of EWS (identical to the N-terminal 264 residues of EWS-FLI1) and three truncation mutants (residues 1–120, 91–199 and, 171–264) were codon optimized for E. coli expression and synthesized (Genscript). The synthetic genes were cloned (KasI and BamHI restriction sites) into a panel of modified pET expression vectors that place either an 8x His-tag or 8x His-tag and a solubility enhancement tag (SET) followed by a tobacco etch virus (TEV) protease cleavage site N-terminal to the expressed protein. Correct insertion was confirmed by DNA sequencing and the plasmid was transformed into BL21 Star^™ (DE3) E. coli cells (Invitrogen) for expression. Generally, one colony was used to inoculate 3 mL of Luria Broth (LB) starter culture supplemented with 100 μg/mL ampicillin, grown for 6 hours at 37 °C, the cells were harvested (3000 × g, 5 min.) and resuspended in 3 mL of fresh M9 media. The suspension was added to 100 mL of M9 medium supplemented with 0.1 g ¹⁵NH₄ Cl, 0.3 g ¹³C₆-glucose and 100 μg/mL ampicillin and grown overnight at 30 °C. The 100 ml overnight culture was diluted in 900 mL M9 containing 0.9 g ¹⁵NH₄ Cl and 2.7 g ¹³C₆-glucose and supplemented with 100 μg/mL ampicillin. When OD₆₀₀ ~ 0.6 recombinant protein expression was induced with 0.5 mM isopropyl β-D-1-thigalactopyranoside (IPTG) and grown for 3.5 hours post induction. Cells were harvested by centrifugation at (3,000 × g, 30 min.) and stored at −80 °C.

The fragments were purified as follows with Sumo protease substituted for TEV to remove the purification tag of EWS 91–199. E. coli cells expressing the appropriate construct were resuspended in 20 mM NaP_i pH 7.3, 300 mM NaCl, 10 mM imidazole lysis buffer, EDTA-free protease inhibitor (Roche), and lysed by three passes through a high-pressure homogenizer (Avestin). The lysate was sonicated for three cycles (10 s on, 10 s off) to shear genomic DNA and cleared by centrifugation (40,000 × g, 30 min.) and passed through a 0.45 μm Millex filter (MilliporeSigma). The cleared lysate was applied to a 5 mL HisTrap FF IMAC column (Cytiva), washed with 10 column volumes of lysis buffer and eluted with 20 mM NaP_i pH 7.3, 300 mM NaCl, 300 mM imidazole. Eluted fractions were pooled, buffer exchanged with an Amicon® Ultra-15 10 kDa cut-off centrifugal filter (Merck) to remove excess imidazole, TEV was added (1:20 TEV:EWS OD₂₈₀ ratio) (Kapust et al. 2001) and incubated with gentle rocking at room temperature overnight. Cleaved protein was passed over a 5 mL HisTrap FF column equilibrated in lysis buffer to remove the solubility tag and TEV protease, the flow-through was collected, concentrated to less than 1 mL, diluted to 10 mL in 20 mM MES pH 5.5 and applied to a Superdex 75 16/60 HiLoad size exclusion column equilibrated with 20 mM MES pH 5.5 (Cytiva). The fractions containing protein were pooled, concentrated, aliquoted, flash frozen and stored at −80°C.

The EWS 1–264 expression construct contained an N-terminal 8x-His tag followed by a TEV cleavage site. E. coli cells expressing the EWS 1–264 construct were resuspended in 20 mM CAPS pH 11, 8 M urea, lysed and cleared as described for the other constructs and applied to a 5 mL HisTrap FF IMAC column (Cytiva) equilibrated in 20mM CAPS pH 11, 8 M urea, washed with 10 column volumes of 20 mM CAPS pH 11, 8 M urea and eluted with 20 mM CAPS pH 11, 8 M urea, 500 mM imidazole. Eluted fractions were pooled, concentrated to less than 1 mL with an Amicon® Ultra-15 10 kDa cut-off centrifugal filter (Merck) and rapidly diluted in 20 ml 50 mM Tris, pH 9, 1% Hexanediol. TEV was added (1:20 TEV:EWS OD₂₈₀ ratio) (Kapust et al. 2001) and incubated with gentle rocking at room temperature overnight. Cleaved target protein was cleared (3,000 × g, 10 min.), concentrated to less than 1 mL and rapidly diluted in 20 mM CAPS pH 11, passed over a 5 mL HisTrap FF IMAC column (Cytiva). The flowthrough was collected, concentrated, and applied to a Superdex 75 16/60 HiLoad size exclusion column (Cytiva) equilibrated in 20 mM CAPS pH 11. Purified EWS 1–264 was concentrated to approx. 1 mM in 20 mM CAPS pH 11, aliquoted, flash frozen and stored at −80°C. The TEV cleavage leaves a two residue (Gly-Ala) N-terminal cloning scar that is not included in the numbering scheme. Protein purity for all constructs was assessed by SDS-PAGE and mass spectrometry.

NMR Experiments

NMR experiments were performed at 298 K using a Bruker NEO spectrometer operating at a proton Larmor frequency of 700.13 MHz and equipped with a 5 mm TCI z-axis gradient cryogenically cooled probe. EWS samples were diluted from storage stocks to 100 μM in 20mM MES, pH 5.5, 5% (v/v) D₂O, 0.2 mM EDTA, 0.5 mM PMSF, 0.1 mM EDTA, and 0.05% (w/v) NaN₃ in a Shigemi tube. The data sets used to assign each EWS construct consisted of a 2D ¹H-¹⁵N HSQC to obtain ¹H_N and ¹⁵N chemical shifts, and a series of complementary 3D experiments (HNCO, HNCACO CBCACONH, HNCACB) to obtain the ¹³Cα, ¹³Cβ and ¹³C’ chemical shifts. All triple resonance experiments were recorded in non-uniform sampling (NUS) mode (Delaglio et al. 2017). Acquisition parameters for the experiments recorded on each of the four EWS constructs are shown in Table 1. Offsets (F3 (¹³C), F2 (¹⁵N)) were 172.5 ppm and 119 ppm for the HNCO, HNCACO pair and 40 ppm and 119 ppm for the CBCACONH, HNCACB pair. For all samples, water suppression was achieved by a WATERGATE scheme (Piotto et al. 1992). The SMILE algorithm implemented in NMRPipe was used to reconstruct NUS data (Ying et al. 2017; Delaglio et al. 1995). Generally, the spectra were apodized by applying a shifted sine bell function and zero filled to twice the number of acquisition points before being Fourier transformed. Spectra were processed using tools in NMRPipe (Delaglio et al. 1995) and assignment of the backbone resonances was completed with CCPNMR analysis version 2.5 (Vranken et al. 2005).

Table 1:

NUS sampling density, data matrix and acquisition times used to acquire the triple-resonance experiments used to assign EWS LCD constructs.

Construct		HNCO	HNCACO	CBCACONH	HNCACB
1–120	NUS density (%)1	7.5	7.5	7.5	7.5
	Complex points2	32 × 64 ×1024	64 × 64 × 1024	64 × 64 × 1024	128 × 64 × 1024
	Acquisition time (ms)2	22.7 × 41 × 112.6	45.4 × 41 × 112.6	5.5 × 41 × 112.6	11.1 × 41 × 112.6
91–199	NUS density (%)1	7.5	7.5	7.5	7.5
	Complex points2	64 × 64 × 1024	64 × 64 × 1024	128 × 64 × 1024	128 × 64 × 1024
	Acquisition time (ms)2	45.4 × 41 × 112.6	45.4 × 41 × 112.6	11.1 × 41 × 112.6	11.1 × 41 × 112.6
171–264	NUS density (%)1	7.5	7.5	7.5	7.5
	Complex points2	64 × 64 × 1024	64 × 64 × 1024	64 × 64 × 1024	128 × 64 × 1024
	Acquisition time (ms)2	45.4 × 41 × 112.6	45.4 × 41 × 112.6	5.5 × 41 × 112.6	11.1 × 41 × 112.6
1–264	NUS density (%)1	15	15	7.5	7.5
	Complex points2	32 × 64 × 1024	32 × 64 × 1024	128 × 64 × 1024	128 × 64 × 1024
	Acquisition time (ms)2	16.5 × 41 × 112.6	16.5 × 41 × 112.6	11 × 41 × 112.6	11 × 41 × 112.6

¹non-uniform sampling density, % of data matrix recorded, non-weighted sampling grid

²indirect (F1, ¹³C) × indirect (F2, ¹⁵N) × direct (F3, ¹H)

Assignment and Data Deposition

The function of the EWS portion in the context of the EWS-FLI1 fusion is not well understood mostly owing to its intrinsically disordered nature, low-sequence complexity, and propensity to self-associate (Schwartz et al. 2015). These features make studying the structural biology of EWS-FLI1 particularly challenging. While solution NMR is well suited to the study of intrinsically disordered proteins, several challenges needed to be solved to successfully prepare a suitable EWS 1–264 sample for resonance assignments. The predominant role of the EWS LCD is to mediate self-association therefore suitable solution conditions amenable to both NMR spectroscopy and EWS 1–264 stability were identified (generally, low protein concentration and no salt). A library of solubility enhancement tags (SETs) was screened and the SET-construct combination that resulted in the highest expression of soluble protein post-purification was chosen as the expression construct. Expression and solubility of each EWS fragment was optimal with different solubility tags: EWS 1–120, N-terminal 8x-His tag, GFP solubility tag, and TEV cleavage site; EWS 91–199, N-terminal 8x His-tag, Sumo solubility tag; EWS 171–264 N-terminal 8x His-tag, GB1 solubility tag and TEV cleavage site, EWS 1–264, N-terminal 8x His tag, TEV cleavage site. EWS 1–264 was purified with a denaturing protocol since the ¹H,¹⁵N HSQC spectra did not yield detectable chemical shift differences between the proteins purified from native or denaturing conditions. SETs were cleaved from all protein constructs prior to recording NMR spectra.

The resonance overlap arising from low sequence complexity (EWS 1–264 is comprised of 18% Glu, 15% Ser, 15% Thr, 14% Tyr, 11% Pro, 10% Gly, and 10% Ala) was addressed using truncation mutants that roughly divide EWS 1–264 into overlapping thirds namely EWS 1–120, EWS 91–199 and EWS 171–264 (Fig. 1). The N- (1–120) and C-terminal (171–264) fragments each contain an SYGQ repeat element while the middle fragment (91–199) is practically devoid of SYGQ repeats. This approach resulted in decreased spectral overlap for each of the smaller fragments compared to the full length NTD easing the difficulty of determining chemical shift assignments for the smaller fragments. The resonances in the spectra of each of the fragments were well resolved, enabling almost complete backbone resonance assignments with 94 % (1–120), 94 % (91–199), and 85 % (171–264) of backbone residues assigned (Table 2). While the use of EWS fragments proved useful to disentangle the resonances arising from repetitive dipeptides like QQ and SS, this strategy was less effective when the pairs occurred within the same fragment. This was particularly acute for the SSYGQQ tri-repeat in 171–264. Ambiguously assigned residues are marked with asterisks in the spectra of the fragments and discussed in the context of EWS 1–264 below (Fig. 1). Fortuitously, the chemical shifts of the fragments were virtually identical to those in the full length NTD, except the terminal residues, thus the resonance assignments from the fragments provided a useful reference point and an aid to resolve overlap when assigning EWS 1–264 (Fig. 1). The use of HA-start carbon-detected experiments (e.g. HACON) did not result in improved discrimination of overlapped chemical shifts since the gains from using the improved dispersion of the C’ nuclei, were offset by the relatively low-sensitivity of these experiments on conventional TCI-cryoprobes (Cook et al. 2018). This was further exacerbated by the low protein concentrations in the NMR samples of the EWS constructs necessary to prevent self-association and aggregation.

Figure 1:

Annotated ¹H, ¹⁵N-HSQC spectrum of EWS LCD fragments 1–120 (red), 91–199 (green), and 171–264 (blue) acquired at 298 K on a 100 μM ¹³C, ¹⁵N-labeled sample in 20 mM MES pH 5.5, 0.5 mM PMSF, 0.2 mM EDTA supplemented with 5 % D₂O. Asterisks denote ambiguously assigned residues.

Table 2:

Percentage of assigned backbone resonances for EWS LCD constructs.

Construct	1H+15N	13Cα	13Cβ	13C’
EWS 1–120	95	94	94	94
EWS 91–199	96	94	93	94
EWS 171–264	84	86	84	84
EWS 1–264	91	89	91	90

With the resonance assignments of the EWS truncated constructs, 214 of the 235 expected resonances of EWS 1–264 were unambiguously assigned (Fig. 2). Backbone carbons for 25 of the 30 proline residues in EWS 1–264 were assigned; 10 prolines occur as PP pairs and thus resonances from five prolines could not be identified using conventional H_N-detected experiments. An additional 13 resonances from the recurrent dipeptides QQ (223, 229), GQ (222, 228), YG (221, 227), SY (232, 226) and SS (219, 225, 231) or tripeptide TTT (102, 108) were identified in the spectra but determined to be ambiguous due to practically identical chemical shifts for all measured resonances. This ambiguity arises primarily from residues in the C-terminal portion of EWS 1–264, particularly from the degenerate tri-repeat SSYGQQ sequence, and are denoted with asterisks in the EWS 1–264 spectrum (Fig. 2). Due to the intrinsically disordered nature and the extreme sequence degeneracy of the EWS LCD, several severely overlapped residues are observed but could be disentangled and assigned using the spectra of the fragments. The overlapped residues in the EWS 1–264 spectra are: T4/T198, A13/Y190, Q38/Q222,Q228, G62/G221,G227, S69/S196, Y70/Y214, A82/A135, Y99/Y195/Y203, S153/S230, Q167/Q234, and Y220/Y226,Y232, (Fig. 2). Some of these peaks are overlapped with ambiguously assigned pairs further complicating the assignment but were identified in the EWS 1–264 data set using the superior dispersion of the C’ resonances and guided by the resonance positions identified in the spectra of the fragments (Fig. 1) Despite the challenges arising from spectral overlap and sequence degeneracy inherent in EWS 1–264, 90 % of the backbone resonances were unambiguously assigned (Table 2).

Figure 2:

Annotated ¹H, ¹⁵N-HSQC spectrum of EWS 1–264 acquired at 298 K on a 100 μM ¹³C, ¹⁵N-labeled sample in 20 mM MES pH 5.5, 0.5 mM PMSF, 0.2 mM EDTA supplemented with 5 % D₂O. The central region (dashed box) is detailed in the inset for clarity. Asterisks denote ambiguously assigned residues.

The EWS LCD, in the context of either wildtype EWSR1 or the EWS-FLI1 fusion, is predicted to be devoid of stable secondary structure by the PSIPRED algorithm (Buchan and Jones 2019). The chemical shift data from EWS 1–264 was used to probe secondary structure propensity at the residue level. The δ2D algorithm uses chemical shift data to calculate the probability that a protein adopts certain secondary structure elements (Camilloni et al. 2012). EWS 1–264 chemical shift data was used to predict inherent secondary structure propensity for helix, sheet, coil and poly-proline structural elements (Fig. 3). These calculations indicate that EWS 1–264 is primarily disordered in solution, with most residues adopting predominantly (> 90%) polyproline II and random coil conformations, consistent with PSIPRED predictions. The assignments for the backbone resonances have been deposited in the Biological Magnetic Resonance Bank under the accession numbers 51111 (1–120), 51112 (91–199), 51113 (171–264), and 51114 (1–264).

Figure 3:

Secondary structure predictions for EWS 1–264 calculated from experimental chemical shifts using the ∂2D algorithm, plotted as a function of the amino acid sequence. Propensity for each secondary structure element is fractionally represented. White spaces correspond to proline pairs or triplets.

Abbreviations:

EWS

RNA-binding protein EWS

FUS

RNA-binding protein FUS

TAF15

TATA-binding protein associated factor 2N

ETS

E-26 transformation-specific

TEV

tobacco etch virus

IPTG

isopropyl β-d-1-thiogalactopyranoside

DBD

DNA-binding domain

NUS

non-uniform sampling

SET

solubility enhancement tag

Data Availability:

NMRPipe processing scripts are available upon reasonable request, expression plasmids containing the EWSR1-FLI1 constructs were deposited with Addgene (180464, 180465, 180466 and 180467), the backbone resonance assignments were deposited in the BMRB (51111, 51112, 51113 and 51114)