Titin UN2A Acts as a Stable, Non‐Polymorphic Scaffold in its Binding to CARP

The N2A segment of titin functions as a pivotal hub for signal transduction and interacts with various proteins involved in structural support, chaperone activities, and transcriptional regulation. Notably, the “unique N2A” (UN2A) subdomain has been shown to interact with the stress‐regulated cardiac ankyrin repeat protein (CARP), which contributes to the regulation of sarcomeric stiffness. Previously, the UN2A domain's three‐dimensional structure was modelled based on its secondary structure content identified by NMR spectroscopy, considering the domain in isolation. In this study, we report experimental long‐range distance distributions by electron paramagnetic resonance (EPR) spectroscopy between the three helixes within the UN2A domain linked to the immunoglobulin domain I81 in the presence and absence of CARP. The data confirm the central three‐helix bundle fold of UN2A and show that this adopts a compact and stable conformation in absence of CARP. After binding to CARP, no significant conformational change was observed, suggesting that the UN2A domain retains its structure upon binding to CARP thereby, mediating the interaction approximately as a rigid‐body.


Introduction
Cells are under the constant influence of stressors derived from metabolism, immunogenic processes, aging, chronic disease, chemical toxicity, and mechanical overload, among others.However, cells are able to trigger adaptive gene programs that help to resist these insults.A prominent example of such an adaptive molecular mechanism is that mediated by the cardiac ankyrin repeat protein (CARP, also termed Ankrd1/MARP1), which is a universal stress response factor of striated muscle. [1]ARP is rapidly and robustly up-regulated in both heart and skeletal muscle upon mechanical, metabolic, or toxic stress. [2]pon up-regulation, CARP is recruited to the elastic I-band of the sarcomere, where it binds to the N2A component of the elastic protein titin.Titin is a gigantic (4200 kDa), intrasarcomeric filamentous protein that spans half the sarcomere, from the Z-disc to the M-line (Figure 1A). [3]It provides elasticity to the sarcomere through the molecular springs of its I-band region and it exists in many differentially-spliced isoforms with different spring compositions and diverse levels of compliance. [4]The physiological outcome of N2A targeting by CARP had remained enigmatic until very recently, when it was shown that CARP induces the cross-linking of titin and actin myofilaments in the sarcomere at the N2A locus. [5,6]Such cross link acts as a "clamp", effectually transforming pliant titin isoforms into shorter and less elastic variants.This confers mechanical resilience to the myofibril. [5,6]By this mechanism, CARP acts as a rapid regulator of sarcomere mechanics that supports muscle performance upon mechanical overload in the early stages of tissue insult.The CARP-mediated mechanical adaptation of muscle becomes effective within minutes to hours, well before other mechanisms of response (such as titin isoform replacement or posttranslational modification) can take place.Particularly in the heart, this mechanism can be expected to preserve ejection force and protect against cardiac insufficiency in the short term.
The CARP-receptor site in titin, the N2A element, is a multidomain chain segment composed of four immunoglobulin (Ig) domains (I80, I81, I82, I83) and a unique sequence insert (UN2A), ca. 100 aa in length, between domains I80 and I81 (Figure 1B).In addition to binding CARP, the N2A element acts as signalling hub in the sarcomere, recruiting cellular proteins that include MARP, the lysine methyltransferase SMYD2, the chaperone HSP90, the protease calpain3/p94, and myopalladin. [7]The structural features of titin N2A have been recently revealed.The UN2A insertion was modelled using a secondary structure composition drawn from NMR spectroscopy data, [5] while the structure of the downstream I81-I83 tandem was elucidated using X-ray crystallography (Figure 1C). [8]The data revealed that UN2A contains a small, central three-helical fold (ca.45 aa long) that is joined to its N-and C-terminal flanking domains, I80 and I81, by long flexible linkers with helical content. [5]However, the model was calculated based only on secondary structure data, with no long-range NMR distance information being available that could inform on the spatial arrangement of the three helices in the bundle.CARP has been shown to strongly bind the dual domain fraction UN2A-I81, where UN2A constitutes the primary binding site [1] and I81 provides a secondary site that confers high affinity to the interaction. [9]CARP binds titin through its C-terminal ankyrin repeat domain (CARP ANK ), which is necessary and sufficient for a high affinity interaction. [1,9] fundamental open question relates to the mechanism by which CARP enables the titin/actin association.Titin and actin co-exist in the confined environment of the sarcomere as laterally aligned filaments, being in permanent close proximity.Yet, their direct association, if present, is only weak.[10,11] A robust association occurs only in presence of CARP.Given that neither titin nor CARP individually bind actin, it is still unknown what makes the CARP/titin complex have such high affinity to actin.In this regard, changes in hydrogen/deuterium exchange rates monitored by mass spectrometry (HDX-MS) between UN2A in isolation and complexed to CARP ANK suggested that the UN2A domain might play a key role in regulating the affinity of the titin/CARP complex to actin by undergoing a modest conformational change.[5] Additionally, the comparison with a close homologue, the m-domain of the myosin binding protein C (MyBP-C) (71 % sequence similarity) [9] might be interesting as it adopts a very similar three-helical bundle.MyBP-C seemingly acts as a polymorphic scaffold that undergoes conformational changes upon binding to calmodulin, when its tri-helical fold adopts an extended arrangement.[12] To gain insight into this question, we investigated the conformation of UN2A in the presence and absence of CARP binding using electron paramagnetic resonance (EPR) spectroscopy.
Due to the significant size of the titin UN2A-I81/CARP ANK complex, only a limited number of methods are suitable to study this question.[15] A pair of EPR-active spin labels can be introduced site-specifically into the UN2A domain by various bioengineering methods. [16]The pulsed EPR method double electron electron resonance (DEER), also known as PELDOR, can then be used to extract distance information between the two spin labels in the nanometre range. [17,18]As titin itself, like many biomolecules, is natively EPRsilent, i. e. diamagnetic, the application of DEER is basically unlimited regarding the size of the investigated molecule or the complexity of the environment.This makes it possible to follow conformational changes within the UN2A domain of titin alone or in complex with other proteins like CARP. [19]However, DEER gives not only a mean distance, but a probability weighted distance distribution of the whole conformational ensemble.2]

Results and Discussion
For the investigation of CARP binding by titin, the construct UN2A-I81 was chosen, as it spans the full dual CARP binding site in titin (Figure 1C). [5,9]Three different constructs were designed, with labelling sites within the UN2A domain, to investigate the three-dimensional orientation of the three-helix bundle (Figure 1D and E).The pair of labelling sites E4/R19 Orange indicates the CARP binding interface identified using NMR and HDX-MS. [5]The construct experimentally studied in this work spanning the dual domain UN2A-I81 (residues 9505-9671) is indicated; D. Representation of the tri-helical bundle in the UN2A domain, with the labelling sites being marked by yellow spheres and the sampled distances shown as blue, red and green lines.;E. Amino acid sequence of the tri-helical bundle, with the labelling sites marked in yellow; F. The spin labelled side chain upon labelling with 3-Maleimido-PROXYL.
monitored the distance and flexibility between Helix 1 and Helix 2, the pair E4/R27 between Helix 1 and Helix 3, and the pair R19/K37 between Helix 2 and Helix 3 (Figure 1D).To introduce the spin labels site-specifically, three variants were designed with cysteine residues at the chosen labelling sites, E4C/R19C, E4C/R27C, and R19C/K37C.It was confirmed that these cysteine mutants are still functional in binding to CARP ANK using analytical gel filtration (see Figure S1).The cysteine pairs were labelled quantitatively with 3-Maleimido-PROXYL (MProx, Figure 1F).The two native cysteine residues are buried in the protein and not accessible for labelling, which was confirmed experimentally (see Figure S2).
Subsequent DEER distance measurements between the two spin labels within each UN2A-I81 mutant resulted in the distance distributions shown in Figure 2.
Distance distributions between labels were simulated, based on the available model of the UN2A tri-helical bundle (PDB: 7NIP).For that, the whole model ensemble, consisting of the top ten calculated models, was included.The ten structures show high conformity (see Figure S3).Possible spin label rotamers, arising from the linker flexibility of the label, were calculated in silico at the labelling sites (see Figure S3) and the distance between the respective pairs of spin labels was predicted (see Figure 2, dashed lines).The results are distance distributions that include the flexibility of the spin label linker.They can now be used to validate the modelled structure with the experimental long-range distance data from the DEER experiments.Comparing the width of the distance distributions, all three mutants show high similarity between modelled and experimental data (Table S2).Since the modelled distance distributions are based on the assumption of a rigid protein structure, we observe no significant flexibility of the three-helix bundle.The maxima of the distance distribution extracted from mutant E4C/R19C and R19C/K37C show a high similarity to the modelled data.However, the mutant E4C/R27C shows an experimental distance distribution with a maximum, that is reduced by about 1 nm compared to the model (Table S2).This might mean, that the model is correct in the position of Helix 1 and Helix 2 towards each other.However, one could speculate that the N-terminus of Helix 3 is slightly more tilted towards the other two helices than the model has suggested.These results indicate that the proposed model had in general a good coverage of the existing conformer ensemble in solution.The three-helix bundle is well defined and shows no large flexibility of one of the helices towards the others.However, the data indicates that Helix 3 might be slightly more tilted inwards, than displayed by the model.
To identify conformational changes within the UN2A domain upon binding of CARP, the three spin labelled mutants were incubated with two molar equivalents of CARP ANK .The respective EPR spectra exhibit a small but significant appearance of a broad spectral feature at the low field peak for all three mutants (Figure S2).This indicates that the rotational mobility of the spin label is restricted, suggesting an interaction with CARP ANK in the region of the spin labelling sites.A pulldown binding assay, using magnetic beads, additionally confirmed that the binding of UN2A-I81 to CARP ANK is not significantly affected by the spin-labelled cysteine mutations (Figure 3A).The mutant E4C/R27C-MProx showed a slightly reduced binding, but as DEER is an ensemble method and the majority of protein is bound, an effect on the DEER data is still expected to be visible.DEER measurements of the doubly labelled UN2A-I81 mutants in presence of CARP ANK were conducted.The resulting distance distributions are shown in Figure 3 BÀ D. There are no significant differences between the distance distributions with and without CARP ANK for all three mutants.This result suggests that there are no global conformational changes of the three-helix bundle of the UN2A domain when UN2A-I81 binds to CARP, neither in the mean distance, nor in the conformational flexibility between the three helices.

Conclusions
The N2A region of titin is an essential signalling hub, interacting with different proteins with structural functions, [24] chaperones , [25] and transcription modulators. [26]In particular, the UN2A domain was shown to interact with the stress-induced CARP, which regulates sarcomere stiffness. [5,7]Previously, the three- The distance distributions are shown as solid line with a 95 % confidence interval after bootstrap analysis. [23]The simulated distance distributions based on PDB 7NIP and rotamer calculations are shown as a dashed line.For details about data analysis see the Experimental Section and Figures S5, S7,  and S9.dimensional structure of the UN2A domain was modelled based on secondary structure content experimentally identified using NMR spectroscopy and where the UN2A domain had been calculated in isolation.Here, we present experimental longrange distance distributions between the three helices in the UN2A domain joined to I81 in the presence and absence of CARP ANK .The experimental data supports the proposed model of UN2A, regarding the arrangement of the three-helices in the bundle.Furthermore, the three-helix bundle is found to be well defined in solution, showing no disordered behaviour in the absence of CARP ANK .The slight differences in the mean distance of the experimental data and the modelled data might indicate that the C-terminus of the Helix 3 is slightly tilted towards Helix 1 and Helix 2. Upon binding of CARP ANK , no significant structural rearrangement was observed within the resolution of DEER spectroscopy.Thus, if a rearrangement of Helix 3 occurs upon CARP binding, as suggested by the H/DX-MS data, this must be small. [5]This means, that the CARP binding site in the UN2A domain exists in a folded state before binding of CARP and that it retains its overall fold upon association.So, despite the highly similar structure, the binding mechanism seems different from the homologue m-domain of MyBPC.However, it will be interesting to investigate the conformation of the UN2A domain during the CARP induced cross-linking to actin in future studies.

Experimental Section General Information
If not specified otherwise, chemicals and reagents were obtained from Sigma-Aldrich, Carl Roth, and Thermo Fisher Scientific.

Sample preparation
Protein site-directed mutagenesis: A segment of human titin UN2A-I81 spanning the CARP binding site (i.e. the three-helix bundle in the unique sequence, the C-terminal linker and the Ig domain I81) [5] (UniProtKB Q8WZ42; residues 9503-9671) was cloned into the pET-Trx1a vector that fuses a His 6 -tag, a thioredoxin domain, and a tobacco etch virus (TEV) protease cleavage site to the N-terminus of the inserted gene.The vector was kindly provided by Dr. Julius Bogomolovas. [27]Next, this clone was subjected to site directed mutagenesis to produce cysteine containing mutants that allow for the attachment of 3-Maleimidoproxyl spin labels.For ease of description, mutants are described in relation to their construct numbering where residue 1 corresponds to residue 9503 in titin Q8WZ42.Site-specific mutagenesis was performed on the clone described using Q5 Site-Directed Mutagenesis (New England Biolabs) following manufacturer's instructions.Specific non-mutagenic forward and reverse primers were designed using NebBaseChanger (New England Biolabs) (Table S1).The following constructs were created: UN2A-I81 E4C/R19C , UN2A-I81 E4C/R27C , UN2A-I81 R19C/K37C .Plasmid DNA was then purified using NucleoSpin Plasmid (Macherey-Nagel, Düren, Germany).All UN2A-I81 mutated variants generated were confirmed by sequencing.
The expression vector for human CARP ANK (residues 106-319; UniprotKBQ15327) has been previously reported. [9]otein production: Protein samples were produced recombinantly in E. coli Rosetta (DE3) cells (Merck Millipore) using protocols based on those previously reported.[9] Briefly, bacterial cultures were grown at 37 °C in Luria-Bertani medium supplemented with 25μg/ ml kanamycin and 34μg/ml chloramphenicol up to OD 600 = 0.6.Protein expression was then induced with 0.5 mM isopropyl β-D-1thiogalactopyranoside (IPTG) and cultures further grown overnight at 18 °C.Cells were harvested by centrifugation (4000 g, 4 °C) and sonicated in lysis buffer (25 mM HEPES pH 7.5, 300 mM NaCl) in the presence of an EDTA-free protease inhibitor cocktail (Roche).The lysate was clarified by centrifugation (40,000 g, 4 °C, 45 min) and proteins isolated from the supernatant by Ni 2 + -affinity chromatography on a Histrap ™ HP 5 mL column (Cytiva) in lysis buffer.Tag removal was by cleavage with TEV protease in dialysis overnight. Cleved samples were further purified by subtractive metal affinity chromatography and size exclusion chromatography (SEC), the latter used a HiLoad 26/60 Superdex 75PG column (Cytiva) equilibrated in 25 mM HEPES pH 7.5, 100 mM NaCl, 1 mM DTT.The resulting protein sample was concentrated to 500-800 μM by ultracentrifugal filtration at 4000 g and 4 °C in SEC buffer.Purified samples were confirmed to be > 98 % pure via Coommassie blue shown as solid lines with a 95 % confidence interval after bootstrap analysis.[23] For raw data and data analysis details see Figure S5-S10 and the Experimental Section.Superimpositions of the raw data and the background corrected data with and without CARP ANK are shown in Figure S11.
stained SDS-PAGE.Protein samples were then aliquoted, flash frozen in liquid nitrogen and stored at À 80 °C until further use.

Analysis of complexes using analytical Size Exclusion Chromatography:
Complexes of CARP ANK and titin UN2A-I81 mutated variants were formed by mixing purified samples at 1 : 1 molar ratio in 25 mM HEPES pH 7.5, 100 mM NaCl.Subsequently, analytical size exclusion chromatography was performed on the sample mixture using a Superdex 200 Increase 10/300 GL column (Cytiva).The total amount of protein injected in the column was 1 mg (at an approximate concentration of 200 μg/μL as quantitated using A 280 ).
Labelling of Titin UN2A-I81 constructs: Before labelling, DTT was removed from the protein solution using Zeba™ Spin Desalting Colum, equilibrated with 25 mM HEPES pH 7.5, 100 mM NaCl.A 10 mM solution of 3-Malemido-PROXYL in DMSO was prepared.A 4x molar excess of 3-Malemido-PROXYL was added to a 50 μM protein solution and incubated for 3 hours on ice.Afterwards, excess spin label was removed from the solution by filtration using Amicon Ultra 0.5 mL centrifugal filters, 3 K MWCO at 14 000 g and 4 °C.Flowthroughs were collected and tested for the presence of free spin label by X-band CW-EPR spectroscopy with an EMXnano (Bruker Biospin) at room temperature.The degree of doublelabelling was also determined by X-band CW EPR spectroscopy and was between 70 and 100 % for each sample.Labelling of the two native cysteine residues was excluded by treating the wild type UN2A-I81 with the same labelling protocol as the cysteine mutants.The respective CW EPR spectra showed no EPR signal for the wild type protein (see Figure S1).This suggests that the native cysteine residues are buried in the protein and therefore not accessible for labelling.Therefore, the native cysteine residues did not interfere with the site-specific labelling of UN2A-I81.

Titin-CARP binding assay:
The binding of 3-Maleimido-proxyl labelled UN2A-I81 mutants (UN2A-I81 E4C-R19C , UN2A-I81 E4CÀ R27C, and UN2A-I81 R19C-K37C ) to CARP ANK was assessed using a pull-down assay as follows.Magnetic Ni 2 + beads (His Mag Sepharose TM excel; Cytivia) were equilibrated into 500 μL of assay buffer (25 mM HEPES pH 7.5, 100 mM NaCl, 1 mM TCEP).Then, 100 μg of His 6 -Trx-CARP ANK was bound to the beads via overnight incubation.Next, 100 μg of either wild-type or labelled mutant UN2A-I81 was applied and incubated at 4 °C for 40 min with gentle agitation.After incubation, the beads were washed to remove unbound sample (3 × 500 μL assay buffer) and resuspended in 150 μL of assay buffer.15 μL of the bead suspension was analysed using 12 % SDS-PAGE.As a control for non-specific binding of the UN2A-I81 samples, the experiments were repeated using untreated beads with no His 6 -Trx-CARP ANK bound.Binding was assessed using densitometry (GelAnalyzer 19.1; www.gelanalyzer.com).To compare the bands of the His 6 -Trx-CARP ANK and the UN2A-I81 samples, the lane and bands were manually selected and the background detected using the rolling ball method (as defined by the manufacturer: http://www.gelanalyzer.com/gelanalyzer-19.1-user-manual.pdf).The calculated band volume of each UN2A-I81 sample was then divided by the respective volume of the His 6 -Trx-CARP ANK band.

Preparation of DEER samples:
For DEER experiments 60 μL samples of 25 μM labelled UN2A-I81 mutants in deuterated 25 mM HEPES, 300 mM NaCl, pH 7.5 buffer containing 20 % of deuterated glycerol were prepared.For the samples of UN2A-I81 bound to CARP, a twofold molar excess of CARP in 25 mM HEPES, 100 mM NaCl, pH 7.5 was added and incubated for 40 min at 4 °C.Samples were transferred into a quartz EPR tube (12 cm length, 3 mm outer diameter), shock frozen with liquid nitrogen and stored at À 80 °C.

Spectroscopy
Continuous Wave (CW) EPR: CW measurements were performed using the BRUKER EMXnano X-band continuous wave (CW)-EPR spectrometer.A sample volume of 20 μL was filled into HIRSCH-MANN ringcaps® and sealed using Cha-seal tube sealing compound.All samples were measured at 20 °C, a modulation amplitude of 1 G, a microwave power of 3.162 mW and a sweep time of 60 seconds.30 scans were accumulated for each sample.To compare the spectral shape, they are represented normalized to the maximum.The unlabelled control is normalized to the same noise as the signal containing samples.DEER spectroscopy: All DEER measurements were performed on a Bruker ELEXYS E580 spectrometer (Bruker Biospin) equipped with an arbitrary waveform generator (AWG), a Flexline probe head, an ER5106QT-II resonator (Bruker) and a 150 W TWT-amplifier (model 187 Ka, Applied Systems Engineering, Fort Worth, TX, USA).The measurements were performed at Q-band (34 GHz) and a temperature of 50 K using a CF935O helium gas-flow cryostat (Oxford Instruments, Abingdon, UK) and a MercuryITC temperature controller (Oxford Instruments).
For the DEER measurements, the standard four-pulse sequence was used: π/2(ν A )À τ 1 À π(ν A )À (τ 1 + t)À π(ν B )À (τ 2 -t)À π(ν A )À τ 2 À echo. [28]The magnetic field was set to the maximum of the nitroxide spectrum for a pump frequency of ν A = 34 GHz, which is at the maximum of the resonator profile.The observer frequency was set to ν B = 33.92GHz, resulting in a frequency offset of 80 MHz.The shot repetition time was set to 4080 μs and the interpulse delay τ1 between the first and second observer pulse was 400 ns.The dipolar evolution time τ2 was chosen between 5 and 7 μs.The time step rate was set to 8 ns.As observer pulses, rectangular shaped pulses were used and length was optimized using a nutation experiment (π-π/2-π-echo) and was about 32 ns for all experiments.The pump pulse was a symmetric hyperbolic sectant pulse HS{1,1} with a length of t P = 100 ns, a β parameter of 6/t P , a frequency sweep width of 90 MHz and an offset of 80 MHz.For nuclear modulation averaging, τ 1 was incremented eight times by 16 ns and the individual traces were summed up.An 8-step phase cycle was applied on the observer pulse according to Tait & Stoll, 2016. [29]he integration gate was positioned symmetrically around the refocused echo with a width of the observer π pulse.Each scan was saved separately in a two-dimensional experiment and the accumulation time was around 16 h for each sample.

Data analysis
Continuous Wave (CW) EPR: The cw spectra of the UN2A-I81 mutants without CARP and the wild type spectrum were normalized to the same noise intensity, as the same number of scans was accumulated for all samples.The spectra of the mutants in presence of CARP were normalized to the same maximum as the respective spectra without CARP, to compare the spectral shape.

DEER spectroscopy:
The DEER data was analyzed with the python package DeerLab (version 1.0.0). [23]The distance distribution was calculated in a one-step procedure, by fitting the background and the modulated contribution in one step, instead of successively.A three dimensional homogenous background function was used and the regularization parameter was chosen according to the Akaike information criterion.The validation was performed with bootstrapping by generating 1000 samples with artificial noise that were subsequently analyzed.The error was shown as the 95 % confidence interval.

Figure 1 .
Figure 1.The N2A element of titin.A. Schematic representation of a half sarcomere; B. Domain composition of the titin N2A element; C. Reconstruction of the 3D-structure of UN2A-I83.Experimental structures (PDB: 7NIP, 7AHS) are shown in surface representation, parts modelled based on NMR secondary structure calculation are flexible.Orange indicates the CARP binding interface identified using NMR and HDX-MS.[5]The construct experimentally studied in this work spanning the dual domain UN2A-I81 (residues 9505-9671) is indicated; D. Representation of the tri-helical bundle in the UN2A domain, with the labelling sites being marked by yellow spheres and the sampled distances shown as blue, red and green lines.;E. Amino acid sequence of the tri-helical bundle, with the labelling sites marked in yellow; F. The spin labelled side chain upon labelling with 3-Maleimido-PROXYL.

Figure 2 .
Figure 2. Distance distributions obtained by DEER of the three doublecysteine mutants of UN2A-I81 in the absence of CARP ANK .In the first line the mutant E4C/R19C is shown in red, in the second line the mutant E4C/R27C is shown in blue, and in the third line the mutant R19C/K37C is shown in green.The protein structures of UN2A [34-73] on the left side are based on PDB 7NIP and the respective labelling sites are shown in the respective colour.The distance distributions are shown as solid line with a 95 % confidence interval after bootstrap analysis.[23]The simulated distance distributions based on PDB 7NIP and rotamer calculations are shown as a dashed line.For details about data analysis see the Experimental Section and FiguresS5, S7, and S9.

Figure 3 .
Figure 3. Binding of UN2A-I81 to CARP ANK .A. The binding assay reports the binding affinity of the double cysteine mutants, labelled with 3-Maleimido-PROXYL (MProx).The values are normalized to the wildtype data.For raw data see Figure S4; The distance distributions of the UN2A-I81 mutants E4C/ R19C (B), E4C/R27C (C), and R19C/K37C (D) are displayed in absence and presence of 2 molar equivalents of CARP ANK .The distance distributions areshown as solid lines with a 95 % confidence interval after bootstrap analysis.[23]For raw data and data analysis details see FigureS5-S10 and the Experimental Section.Superimpositions of the raw data and the background corrected data with and without CARP ANK are shown in FigureS11.