Delineation of functional subdomains of Huntingtin protein and their interaction with HAP40

Summary

The huntingtin (HTT) protein plays critical roles in numerous cellular pathways by functioning as a scaffold for its many interaction partners and HTT knock out is embryonic lethal. Interrogation of HTT function is complicated by the large size of this protein so we studied a suite of structure-rationalised subdomains to investigate the structure-function relationships within the HTT-HAP40 complex. Protein samples derived from the subdomain constructs were validated using biophysical methods and cryo-electron microscopy, revealing they are natively folded and can complex with validated binding partner, HAP40. Derivatized versions of these constructs enable protein-protein interaction assays in vitro, with biotin tags, and in cells, with luciferase two-hybrid assay-based tags, which we use in proof-of-principle analyses to further interrogate the HTT-HAP40 interaction. These open-source biochemical tools enable studies of fundamental HTT biochemistry and biology, will aid discovery of macromolecular or small-molecule binding partners and help map interaction sites across this large protein.

eTOC Blurb:

Alteen et al. designed structure-rationalised HTT subdomain constructs for structure-function interrogation of this huge protein. HTT subdomain proteins were validated using biophysical approaches and cryo-electron microscopy. Derivatized versions of these constructs were used for proof-of-principle protein interaction analyses in vitro and in cells to probe the HTT-HAP40 interaction.

Introduction

Huntingtin (HTT) is a 348 kDa protein hypothesised to act as a scaffolding protein, governing numerous cellular functions including axonal transport ^1,2, transcriptional regulation ³ and proteostasis regulation by mediating protein-protein interactions ^4,5. HTT also plays a critical role in development ⁶ and knock out is embryonic lethal ⁷. Repeat expansion mutation of the HTT gene in its CAG repeat-tract above a critical threshold of approximately ~35 repeats results in the autosomal dominant neurodegenerative condition Huntington’s Disease (HD). HD patients suffer debilitating cognitive, motor and psychiatric symptoms and have a life expectancy of ~18 years after symptom onset ^8,9. CAG expansion results in an expanded polyglutamine (polyQ) tract near the N-terminus of the encoded protein huntingtin (HTT), resulting in an aberrantly functioning form of the protein ^9,10. HTT. Despite identification of the HTT gene almost thirty years ago ¹¹, there are currently no disease-modifying therapies available, in part due to limited understanding of HTT protein biochemical function, or how these functions might be modulated by polyglutamine expansion. There is therefore a pressing need for the development of chemical and biochemical tools that can be used to illuminate the molecular details of the biological functions of HTT and enable therapeutic strategies to combat the pathology of expanded HTT.

The structure of HTT has been solved by cryo-electron microscopy ^12–14, in complex with 40 kDa huntingtin-associated protein (HAP40) ¹⁵, forming a ~389 kDa multidomain complex. HTT is primarily composed of HEAT repeats (huntingtin, elongation factor 3, protein phosphatase 2a, and yeast kinase TOR1) ¹⁶ that form a large solenoid-like structure at the N-terminal region of the protein as well as a more compact C-terminal region. The N-HEAT and C-HEAT domains are connected by a bridge domain, composed of repetitive alpha-helical structures more closely resembling armadillo-like repeats, with HAP40 sandwiched between these regions (Figure 1A). HTT is reported to interact with over 3000 proteins ^4,17–19. The similarity of the HTT-HEAT motifs to related armadillo repeats, which are known to mediate PPIs, may provide a structural basis for HTT interaction network. Additionally, HTT contains many large, disordered regions, including most of exon 1 which contains the polyQ region. Importantly, these disordered regions have been shown to contain post-translational modifications (PTMs) and may mediate interactions with protein binding partners ²⁰. HAP40 remains the only biochemically and structurally validated interaction partner ^12,13 and thus HAP40 interaction is the only molecularly defined function of HTT in vitro. Interestingly, while HTT has been recombinantly purified as an apo protein, efforts to recombinantly purify HAP40 on its own have not been successful, suggesting that HTT may aid expression and stabilization of HAP40 structure in these systems ^12,13,21. This observation has been supported by genetic knockdowns of HTT that show a corresponding decrease in HAP40 expression ^12,22 and studies which show that the two proteins coevolved ²³.

Figure 1: NTD and CTD subdomain constructs of HTT can be recombinantly expressed and produced to high purity in milligram quantities, enabling structural, functional and biophysical studies.

A, Schematic representation of the domain architecture of HTT and regions within the domain boundaries that are unresolved in cryo-EM models (grey). Constructs analysed within this study are shown on the HTT-HAP40 structural model in surface representation with HTT in grey and HAP40 in yellow. ¹² The constructs corresponding to aa. 97–406, 97–1721, 97–2069 and 2095–3138 are shown on the model in orange, blue, teal and purple respectively. B, Representative SDS-PAGE and size-exclusion chromatogram of purified NTD. C, Representative SDS-PAGE and size-exclusion chromatogram of purified CTD.

We previously reported a toolkit of resources for the scalable production of high-purity HTT and HTT-HAP40 protein samples from various eukaryotic expression systems ²¹. Milligram quantities of samples can be generated with various Q-lengths as well as an exon 1 deleted (Δexon 1) form of the protein that are suitable for numerous in vitro assays. This expression platform has enabled high-resolution structural analysis of the HTT-HAP40 complex using cryo-EM, and associated biophysical studies such as small-angle X-ray scattering, allowing detailed examination of the structural organisation and potential functional roles of this large multidomain complex ¹².

Nevertheless, the large size of the protein and its associated HAP40 complex presents challenges for many downstream applications. For example, the analysis of protein-ligand interactions using surface plasmon resonance (SPR), which relies on optical methods to detect mass changes on an immobilised surface, suffer from decreasing signal-to-noise with increasing mass ratio between ligand and analyte molecules ²⁴. Smaller subdomain fragments of HTT could serve as useful biochemical tools for the study of HTT function and its interaction with other proteins, permitting a wider array of in vitro biochemical and biophysical assays. Additionally, well-defined subdomain fragments could help clarify the relative contributions of each domain in stabilizing HTT-HAP40 interaction, a potential therapeutic target for HD. HTT fragment constructs have been reported in the literature prior to the determination of the cryoEM structure, such as a commonly used fragment encompassing the exon1 region and a portion of the N-HEAT domain (HTT aa. 1–586) ^25,26. However, these constructs frequently have start and stop sites in the middle of HEAT domains or have boundaries in the middle of extensive disordered regions, which likely result in a non-native protein product. To date, no robust biophysical or structural validation of any such construct has been published.

In this study, we report the cloning, expression, purification and validation of highly pure recombinant HTT subdomain proteins and the open-source tools required to generate these samples. We also report the generation of full-length HTT and HTT subdomain constructs with biotinylated tags at both the N-terminus and C-terminus. We show that constructs encompassing the N-HEAT and bridge domains (NTD, aa. 97–2069) as well as the C-HEAT domain (CTD, aa. 2095–3138), can be stably expressed in eukaryotic cell culture and behave well in biophysical assays. We demonstrate that these subdomains can be combined in vitro or co-expressed to reconstitute the HTT-HAP40 complex, as shown by our cryoEM analysis, indicating that they are folded properly and can serve as suitable surrogates for structural or biophysical analyses. Additionally, we show that HAP40 can form a stable complex with the C-HEAT domain in the absence of the N-HEAT and bridge domains, and that this CTD-HAP40 complex can be resolved by cryoEM. Using these subdomain constructs, we provide the first kinetic analysis of the HTT-HAP40 interaction, revealing high affinity binding of the two proteins. Finally, we validate the interaction between the CTD and HAP40 in cells using a luciferase two-hybrid assay and use subdomain constructs to further probe the HAP40-stabilisation function of HTT in cells. We expect these subdomains will serve as useful biochemical reagents for further structural investigation of the HTT-HAP40 complex, as well as for screening, characterisation, and validation of other members of the HTT protein-protein interactome.

Results

Design and cloning of HTT and HTT subdomain constructs with FLAG and Avi tags

Cloning of biotinylated full-length HTT constructs, as well as subdomain constructs, with or without biotin tags, was performed as previously reported ²¹. Ligase-independent cloning (LIC) was used to clone gene sequences into a pBMDEL baculovirus transfer vector which can be used for expression in either Sf9 insect cells or mammalian cells. Structure-rationalised clones for subdomain constructs were designed by examining the domain boundaries observed in the cryo-EM model of full-length HTT-HAP40 (Figure 1A) and span aa. 97–2069 for the N-HEAT and bridge domain containing construct, and aa. 2095–3138 for the C-HEAT domain construct. All clones were designed to incorporate C-terminal FLAG tags for affinity purification. Biotinylated constructs were cloned to include a 15 amino acid AviTag peptide sequence at either the N- or C-terminus of the protein, enabling biotinylation by co-expression with BirA biotin ligase and addition of exogenous biotin ²⁷.

Scalable production of high-purity HTT and HTT subdomain samples

After cloning and verifying the gene sequences of the proposed subdomains, as well as the biotinylated full-length and subdomain constructs, we next aimed to express and purify these constructs using baculovirus-mediated expression in Sf9 insect cell culture. Samples of HTT obtained from either mammalian or insect cell expression have resulted in comparable cryoEM structures ^12,13 and our constructs may be used for production in either system. SDS-PAGE analysis of anti-FLAG purified samples from small-scale test cultures showed the presence of bands corresponding to the expected molecular weights of each subdomain (Figure S1). To further characterise each sample, expression volumes were scaled up to 4 L and the recombinant subdomains were subsequently isolated from lysed cells via FLAG affinity purification and then further purified by size-exclusion chromatography (SEC). We found that a subdomain construct consisting of the folded N-terminal region up to the intrinsically disordered domain (IDR, aa. 97–406) was capable of being expressed and purified by SEC but was found to be poorly stable in biophysical assays (Figures S2). Similarly, a construct consisting of the complete N-terminal solenoid region (aa. 97–1721) showed a non-optimal elution profile on SEC, suggesting that it was not a suitable protein for in vitro assays (Figure S3). As such, these two constructs were deprioritised for further investigation in vitro. However, we discovered that the inclusion of the bridge domain in this construct (aa. 97–2069) caused it to elute as a monodisperse sample, permitting its isolation at >85% purity as determined by densitometry (Figure 1B). The yield of the N-terminal HEAT and bridge domain (NTD) construct after this two-step purification sequence was comparable to that of full-length HTT Δexon 1 at ~1.4 mg/L. Additionally, a construct consisting of the C-terminal HEAT domain (CTD) could also be purified as a monodisperse sample in high purity (Figure 1C), albeit at lower yields of ~0.3 mg/L, but nonetheless permissible for downstream structural and biophysical assays using this construct. Biotin tagged variants of HTT, HTT-HAP40 and the NTD and CTD constructs bearing AviTag sequences at either the N- or C-terminus were co-expressed with BirA and purified in the same manner as constructs bearing only FLAG tags. Size-exclusion chromatography profiles and corresponding yields of biotinylated proteins were nearly identical to non-biotinylated constructs (Figure S4). The installation of the biotin moiety was confirmed by streptavidin gelshift ²⁷ (Figure S5).

Subdomain constructs show good properties in multiple orthogonal biophysical assays

With purified samples of each subdomain in hand, we next sought to assess their quality and characteristics through a series of biophysical analyses. We performed differential scanning fluorimetry (DSF) ²⁸ to determine the thermal stability of each construct under identical buffer, salt and pH conditions as full-length HTT samples (Figure 2A–2C). We found that the NTD, with a calculated melting temperature (${\text{T}}_{\text{m}}$) of 48.3 ± 0.2°C, has significantly greater thermal stability relative to full-length apo HTT (${\text{T}}_{\text{m}}$ ~43.2 ± 0.1°C). Conversely, we observed a ${\text{T}}_{\text{m}}$ of 42.4 ± 0.1°C for the CTD, suggesting this subdomain is less thermostable than full-length HTT. Nevertheless, both subdomains showed good properties in these biophysical assays, including a sharp transition upon heating and high dynamic range of the fluorescence signal between folded and unfolded states (Figure S6). We also analysed the protein constructs using differential static light scattering (DSLS) to assess the aggregation behaviour of each sample upon heating (Figure 2D–2F). Mirroring the trend observed by DSF, the NTD showed a significantly higher aggregation temperature (${\text{T}}_{\text{agg}}$) than the CTD. Given the large, multidomain structure of full-length HTT, it is possible that its ${\text{T}}_{\text{m}}$ and ${\text{T}}_{\text{agg}}$ values represent some combination of the values for the NTD and CTD. Notably, the unfolding of full-length HTT as monitored by DSF occurs over a broader temperature range, suggesting that the domains may unfold sequentially. Taken together, these data suggest that the two purified subdomains each comprise a soluble, folded protein that is suitable for use in biophysical and screening assays, as well as structural analyses.

Figure 2: NTD and CTD subdomains of HTT have good biophysical properties.

A-C, DSF profiles and calculated melting temperature (${\text{T}}_{\text{m}}$) values for full-length HTT (orange) vs NTD (teal) and CTD (magenta). Melting temperatures were determined from the inflection point of curves obtained by fitting the data to the Boltzmann sigmoidal function. D-F, DSLS profiles of full-length HTT (orange) vs. NTD (teal) and CTD (magenta) showing calculated aggregation temperature (${\text{T}}_{\text{agg}}$).

CryoEM reveals the HTT subdomains can stabilise HAP40, forming a functional HTT-HAP40 complex

Given the results of our biophysical assessment of the NTD and CTD samples, which suggested the proteins were properly folded, we next examined if they could be used to re-constitute the HTT-HAP40 complex. To test this, we co-expressed NTD, CTD, and HAP40 in Sf9 cells and attempted to purify the protein as before. Strikingly, after affinity purification from lysed cells with anti-FLAG resin, we observed protein bands corresponding with the molecular weight of these three fragments by SDS-PAGE (Figure S7). Further purification of the crude samples by size-exclusion chromatography as before revealed a profile nearly identical to that obtained with full-length HTT-HAP40, with a major peak eluting at the same volume as expected for the intact construct (Figure 3A). After collecting this peak and performing further analysis by DSF, we observed a thermal transition profile and ${\text{T}}_{\text{m}}$ value nearly identical to that of full-length HTT (Figure 3B, Figure S8). As further confirmation of the protein bands observed through SDS-PAGE analysis of the purified triplet complex (Figure 3C), we carried out western blotting using antibodies for the N-terminal region of HTT as well as HAP40 (Figure 3D, Figure S9). This analysis showed that the 218 kDa band corresponds to the NTD and that the band at ~40 kDa is HAP40. We also noted the presence of an additional band at ~−160 kDa, which persisted despite several rounds of purification. This impurity also possessed a recognised epitope of the D7F7 $\text{α}$-HTT antibody, suggesting that it may be a partial fragment of the N-HEAT region that either does not fully express in Sf9 cells or has been proteolysed by an unknown mechanism. HTT is susceptible to proteolysis within the IDR (30), and the molecular weight of this fragment suggests this band corresponds to a cleavage product within this region.

Figure 3: HTT subdomains bind HAP40 and can be used to reconstitute the HTT-HAP40 complex.

A, Size exclusion chromatography profiles of co-expressed NTD, CTD and HAP40 (black trace) after FLAG purification from Sf9 cells. An elution volume nearly identical to full-length HTT-HAP40 (orange trace) indicates formation of the trimeric complex. B, Reconstituted and full-length HTT-HAP40 complexes have similar melt profiles as determined by DSF. C, SDS-PAGE of full-length HTT (lane 1), NTD (lane 2), CTD (lane 3) and co-expressed subdomains (lane 4). D, Western blots of full-length HTT-HAP40, NTD, CTD, and reconstituted HTT-HAP40. E, Overlay, of cryo-EM map obtained from reconstituted HTT-HAP40 superimposed on the model generated for full-length HTT-HAP40 6X9O.

To further verify the structural organisation of the reconstituted complex, we performed cryo-EM of this sample using similar conditions as performed for the full-length HTT-HAP40 complex (Table 1, Figure S10). The 3.3 Å resolution map obtained from this sample is highly similar to the previously solved atomic resolution model (Figure 3E) ¹². A model was not built into this map due to a high degree of anisotropy. However, the striking structural resemblance of this reconstituted complex map to the full-length HTT-HAP40 complex indicates the subdomains are capable of properly folding and associating to a degree that enables the concomitant expression of HAP40 and validates these constructs for the expression of functional and folded HTT protein subdomains.

Table 1:

Cryo-EM data collection, refinement and validation statistics.

	Full-length HTT-HAP40 (EMDB-28767)	CTD-HAP40 (EMDB-28766) (PDB 8SAH)
Data Collection and Processing
Magnification	100,000	100,000
Voltage (kV)	200	200
Electron exposure (e-/Å2)	51.3	51.3
Defocus range (μm)	−2.0 to −0.4	−2.0 to −0.4
Pixel size (Å)	0.81	0.81
Symmetry imposed	C1	C1
Initial particle images (no.)	112,717	986,229
Final particle images (no.)	64,597	134,849
Map resolution (Å)	3.3	3.2
FSC threshold	0.143	0.143
Refinement
Initial model used (PDB code)		6X9O
Model resolution (Å)		3.2
FSC threshold		0.143
Model resolution range (Å)		2.85–3.42
Map sharpening B factor (Å2)		−163
Model composition
Non-hydrogen atoms		9175
Protein residues		1180
Ligands		0
B factors (Å2)
Protein		44.09
Ligand		N/A
R.m.s. deviations
Bond lengths (Å)		0.009
Bond angles (°)		0.952
Validation
MolProbity score		2.80
Clashscore		19.66
Poor rotamers (%)		1.18
Ramachandran plot
Favored (%)		89.14
Allowed (%)		9.66
Disallowed (%)		1.21

The HTT CTD is sufficient to stabilise HAP40, forming a complex which can be resolved by cryoEM

Having confirmed the robust structural and biophysical integrity of the NTD and CTD constructs, we next asked whether either the NTD or CTD alone is capable of stabilizing HAP40 when co-expressed. Co-expression experiments with the NTD and analysis by SDS-PAGE did not show evidence of HAP40 expression (Figure S11). However, after co-expressing HAP40 with the CTD, a band corresponding to HAP40 was present on SDS-PAGE at the expected molecular weight (Figure 4C, Figure S12). Analysis of this sample using analytical size-exclusion chromatography revealed a significant shift in elution volume, corresponding to an increase in size relative to the C-HEAT construct (Figure 4A). We performed DSF to determine the thermostability of this complex and found that it possessed a marginally higher melting temperature of approximately 1.6 degrees and a sharper melting transition curve than the CTD (Figure 4B, Figure S13). This shift in thermal stability is below the typical significance cut off ²⁹, which suggests that complex formation of the CTD with the HAP40 binding partner has limited effect on the overall stability of the CTD.

Figure 4: HAP40 can be purified as a complex with CTD alone.

A, Overlay of size-exclusion chromatograms of purified co-expressed CTD and HAP40 (black) vs apo CTD (magenta). B, The CTD-HAP40 complex displays higher thermostability and a sharper transition upon thermal unfolding as determined by DSF. C, SDS-PAGE analysis of purified co-expressed C-HEAT and HAP40, with bands at ~120 kDa and ~40 kDa indicated. D, Overlay, of the map of purified co-expressed CTD-HAP40 obtained by cryo-EM (grey mesh) with the carton structure of the domains determined from the full-length cryo-EM model 6X9O. E, Structure overlay of CTD (purple) and HAP40 (grey) complex (PDB ID: 8SAH) with CTD-HAP40 of full-length model (grey) (PDB ID: 6X9O).

We again studied the structure of this complex by cryo-EM to determine its overall 3D structure relative to full-length HTT-HAP40 (Figure 4D, Table 1, Figure S10). We found that the 3.2 Å resolution map obtained from the CTD-HAP40 complex fit well into the structural model derived from the full-length complex for these regions of the protein, with only slight deviations at the termini of the proteins. A model of CTD-HAP40 (PDB ID: 8SAH) was built into the map (Figure 4E). The model of CTD-HAP40 shows no significant structural differences compared to this portion of the structure which we solved previously (6X9O) with an RMSD of 0.436 over 1180 residues. Overall, these data indicate that the CTD alone is sufficient to stabilize and permit expression of HAP40 and that the CTD HTT fragment is structurally similar to the CTD of the full-length HTT protein.

HTT subdomain constructs allow interrogation of protein-protein interactions by biophysical methods

With the CTD-HAP40 complex in hand, we proceeded to measure its binding affinity to the NTD in vitro. Using analytical size-exclusion chromatography, we showed that pre-mixing the purified protein subdomains results in a decreased retention volume compared to the elution of CTD-HAP40 alone, suggesting that the majority of the CTD-HAP40 associates with the NTD (Figure 5A). To quantify the interaction, we purified CTD-HAP40 bearing a C-terminal biotin tag to measure binding kinetics using biolayer interferometry (BLI). Biotinylated CTD-HAP40 was immobilised onto streptavidin biosensors at a concentration of 1 μg/mL and, after equilibration, the sensors were incubated with various concentrations of NTD (Figure 5B, Figure S14). The resulting association and dissociation curves revealed a concentration-dependent rate of binding and dissociation of the two protein subunits. By fitting the curves to a global 1:1 binding model, a dissociation constant (${\text{K}}_{\text{d}}$) of 10 ± 0.3 nM was obtained (${\text{k}}_{\text{a}}$ (1/Ms) = 2.3 × 10⁵ ± 0.03 × 10⁵ M, ${\text{k}}_{\text{d}}$ (1/s) = 2.3 × 10⁻⁴ ± 0.21 × 10⁻⁴ M, R² = 0.93 ± 0.07). These data represent the first quantitative in vitro assessment of the strength of HAP40 binding to HTT, albeit measuring a fraction of the interface mediated through CTD-HAP40 and the NTD, and support earlier observations that the HTT-HAP40 interaction is remarkably stable under a wide array of conditions ¹².

Figure 5: Purified co-expressed CTD-HAP40 binds the NTD with high affinity.

A, Overlayed size-exclusion chromatography profiles of purified CTD-HAP40 (magenta), NTD (teal) and equimolar amounts of pre-mixed NTD and CTD-HAP40 (black). B, Characterisation of NTD binding affinity to immobilised CTD-HAP using biolayer interferometry, revealing a ${\text{K}}_{\text{d}}$ of 10 ± 0.3 nM. Kinetic parameters were determined by fitting the data to a 1:1 binding association and dissociation model (black lines) using GraphPad Prism 9, R² = 0.927. Three independent experiments were performed.

Validation of HTT subdomain protein-protein interactions in cells

To further verify that HAP40 can be bound by HTT CTD to form a stable complex under physiological conditions, we performed a luciferase two-hybrid (LuTHy) assay ³⁰ in HEK293 cells (Figure 6A). In this assay, mCitrine-Protein A (mCit-PA)-conjugated HTT constructs corresponding to the full-length and CTD sequences were co-expressed with HAP40-NanoLuc luciferase (NL) fusion protein. If a protein-protein interaction takes place, and if the donor (NL) and acceptor (mCit) proteins come in proximity of ~10 nm or less ³¹, the bioluminescence resonance energy transfer (BRET) can be assessed. Screening HTT and HAP40 constructs showed significantly increased BRET ratios for the pairing of HTT CTD and HAP40 compared to tag-only controls (Figure 6B). This confirms that the CTD of HTT is sufficient to bind HAP40 in a cellular context.

Figure 6: LuTHy assay shows interaction of HTT full-length (FL) and its C-terminal domain (CTD) with HAP40 in live cells.

A, Graphical illustration of the LuTHy-BRET assay performed. HTT FL or HTT CTD were expressed as mCitrine-Protein A (mCit-PA)- and HAP40 as NanoLuc luciferase (NL)-tagged fusion proteins in HEK293 cells. After expression for 48 h and addition of luciferase substrate, BRET was quantified from live cells. B, BRET ratios between HAP40-NL and HTT FL-mCit-PA and HTT CTD-mCit-PA. As a control, NL only was co-transfected with the HTT acceptor constructs, respectively, as well as PA-mCit only with the HAP40-NL donor construct. HAP40-NL co-expressed with HTT FL-mCit-PA and HTT CTD-mCit-PA showed significantly increased BRET ratios compared to controls, respectively. Bars represent means ± SEM from two independent experiments performed in triplicate. One-way ANOVA with Tukey’s multiple comparisons test, **p < 0.002, *** p < 0.001.

Investigating HTT cellular functions with subdomain constructs

Next, we sought to scrutinise a defined biomolecular function of HTT using our subdomain tools. HAP40 expression cannot be detected in HTT knockout HEK293 cells (Figure 7A), in line with previously published data which show the levels of HAP40 are dependent on HTT protein levels and that HTT functions to stabilise HAP40 ^12,22. Thus, by monitoring the protein levels of HAP40 in HTT KO cells in the presence of exogenous full-length or subdomain constructs of HTT we could observe a rescue of the HAP40 null phenotype in these cells. Overexpression of N-HEAT, CTD or full-length HTT in these cells showed that only the full-length HTT can restore some degree of HAP40 expression in HTT knock out HEK293 cells. Together with our LuTHy assay data, this indicates that although HTT CTD can bind HAP40 in a cellular context, it is not sufficient to restore the HAP40 stabilisation function of HTT.

Figure 7: Overexpression of full-length HTT can rescue HAP40 protein levels in HTT-null cells.

A-B, Representative western blot images of total cell lysates from HEK8293 wildtype and HEK293 HTT-null lines after transfection with of various FLAG-tagged constructs for HTT full-length (FL) and truncated species of HTT. A, Nitrocellulose membrane was incubated with rabbit anti-F8A1 antibody, followed by mouse anti-tubulin to detect endogenous HAP40 and tubulin. B, Nitrocellulose membrane was incubated with mouse anti-FLAG antibody, followed by mouse anti-GAPDH to detect overexpressed FLAG-tagged constructs and endogenous GAPDH. Calculated molecular weight for HTT 81–1643 is 175 kDa, for HTT 2088–3144 121 kDa, and for FL HTTQ23 352 kDa. Two independent experiments were performed.

Discussion

The expansion of the HTT polyQ tract has been well-defined as the cause of HD pathobiology, but the structure-function paradigm of this mutation remains to be precisely defined, and the biochemical functions of wildtype HTT, or the changes to HTT function as consequences of expansion, are incompletely understood. Given a lack of biochemical tools to interrogate HTT function in cells, genetic studies have formed the basis of most functional studies, either through knockdown of HTT expression or the generation of expanded polyQ allelic series in various animal models or cell lines ^32–34. Additionally, affinity-based proteomic methods coupled with mass spectrometry have been applied to identify putative protein binders that interact with HTT ^17,18. Efforts to interrogate HTT function by many biochemical and biophysical assays are thwarted by the large size of HTT, which renders these assays non-tractable. Smaller HTT subdomains provide a solution to this problem, but many of the constructs published to date are not designed with knowledge of the HTT structure in mind. Often the boundaries of published HTT fragments are predicted to interrupt structural elements which likely result in non-native protein products. To date, no robust biophysical or structural validation of any HTT construct has been published. The toolkit of subdomain constructs described in this manuscript provides a solution to this problem, through the rational structure-based design of subdomain constructs and their validation by numerous assays and structural analysis by cryoEM.

Our functional HTT subdomains have enabled fundamental insight into the key structural interactions governing the HTT-HAP40 complex. Our observation that HAP40 can be co-expressed with the CTD alone, and form a stable, soluble protein complex, shows the importance of the HTT interaction interface for HAP40 stability. Analysis of the CTD-HAP40 interaction surface shows that this subdomain accounts for approximately 64% of the total interaction area between HAP40 and full-length HTT, highlighting the significance of the C-HEAT on binding to HAP40. The generation of this complex permitted quantitative characterisation of its binding to the NTD, providing additional evidence of the robust nature of HTT-HAP40 affinity. Such analysis has not been possible with full-length HTT to date as HAP40 can only be isolated in the presence of HTT, or, as we demonstrate in this study, the CTD of HTT. Given that no other putative protein binding partners of HTT have been characterized through a direct binding assay in this manner, the use of the CTD-HAP40 and NTD association provides an important positive control for prospective PPI analyses.

Because of the suspected role of expanded HTT in triggering HD progression, HTT lowering therapies have been hypothesised as a potential therapeutic strategy for HD. One proposed pharmacological approach to HTT lowering is through the use of proteolysis-targeting chimeras (PROTACs) – heterobifunctional ligands that combine a recognition scaffold for a protein of interest combined with an E3 ligase motif, resulting in ubiquitin-mediated proteasomal degradation of target proteins ^35,36. To date, however, all small-molecule ligands being developed as PROTACs bind mutant aggregated HTT fragments ^37,38. As a result, these compounds may suffer from off-target effects because they do not bind a homogenous or specific binding site on the protein structure nor soluble HTT that is present within cells. There is therefore a need for the discovery and validation of novel small-molecule scaffolds that bind soluble HTT in a reversible and specific manner.

To support efforts to validate the complex network of HTT protein-protein interactions, as well as to support pre-clinical characterisation and optimisation of small-molecule ligands, high quality biochemical tools and assay methods are needed. The suite of 19 full-length and subdomain structure-rationalised constructs we have developed for recombinant protein production with FLAG or biotin tags, as well as for cell-based Luthy assays, have a range of potential applications to support fundamental biology and drug discovery efforts. We expect that the combined use of NTD and CTD constructs will permit approaches to counter-screen novel ligands for selectivity of binding, as well as to characterise the degree and site of interaction between HTT and potential PPIs. Furthermore, the smaller size and robust nature of the biotin tags on these constructs enable a variety of biophysical assays to accurately quantify the affinity of potential HTT binders. Additionally, the smaller molecular weight of the subdomains and, in some cases, their higher expression yield relative to full-length HTT, improves their suitability for high-throughput applications.

Using our defined and validated subdomain construct boundaries, we were also able to characterise the CTD-HAP40 interaction in live cells. This revealed that the CTD can bind HAP40 when measured using the BRET-based LuTHy readout. Interestingly, in a HTT knockout background where HAP40 protein levels are ablated due to lack of stabilisation by HTT ^12,22,39, we showed that the full-length, but not the CTD alone, is able to restore HAP40 expression. This suggests that the interdependent relationship between HTT and HAP40 expression is perhaps not just one of structural stabilisation, as the CTD is sufficient to form a stable HAP40 complex, but perhaps linked in a more complex co-translational dependent manner.

In summary, we have generated and validated a suite of high-quality biochemical tools to support the structural and functional characterisation of HTT. We report the design and purification of HTT subdomains encompassing the N-terminal and C-terminal regions of the protein, as well as biotinylated variants that permit convenient immobilisation on streptavidin medium. These protein constructs, which are stably expressed and folded, permit a wide array of biophysical assays to probe the diverse network of HTT protein-protein interactions and facilitate future screening efforts. Additionally, these subdomains have provided important information on the nature of the HTT-HAP40 interaction, demonstrating that the C-HEAT domain forms key interactions with HAP40 and is sufficient to stably interact and retain HAP40 in vitro in the absence of the NTD. Finally, we used our defined subdomains to probe the HAP40-stabilisation function of HTT, revealing expression of the full-length HTT protein molecule is required to rescue this function in a knockout background. We expect that these biochemical tools will find widespread use within the HD community and will support ongoing efforts to elucidate the biological roles of HTT.

STAR Methods

Resource Availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Rachel J. Harding (rachel.harding@utoronto.ca)

Materials availability

All unique/stable reagents generated in this study are available via Addgene (see Data S1) or from the lead contact upon request.

Data and code availability

Cryo-EM maps can be downloaded at EMDB-28767 and EMDB-28766 and PDB accession 8SAH. This paper does not report any original code. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Experimental Model and Study Participant Details

Mach1 T1R competent cells were unauthenticated and purchased from ThermoFisher, catalog number C862003. DH10Bac competent cells were unauthenticated and purchased from Gibco, catalog number 10361012. Sf9 cells were unauthenticated and purchased from ThermoFisher Scientific, catalog number, 12659017. HEK293T cells were unauthenticated and purchased from the American Type Culture Collection (ATCC), catalog number CRL-11268. HEK293T HTT null cells were unauthenticated and a kind gift from the laboratory of Marcy MacDonald ⁴⁰.

Method Details

Cloning of expression constructs

Expression constructs for the NTD (amino acids 97 to 2069) and the CTD (amino acids 2095 to 3138) were constructed with three different affinity tag arrangements. The first version had a C-terminal Flag tag, the 2nd version had a N-terminal biotinylation tag / C-terminal Flag tag, and the 3rd version had the biotinylation and Flag tag on the C-terminal. The biotinylation tag contains the sequence GLNDIFEAQKIEWHE which facilitates the in vivo conjugation of a single biotin to the lysine residue when the HTT protein is co-expressed with biotin ligase in SF9 cells. To assemble each plasmid, the DNA encoding for the HTT gene fragments was PCR amplified from cDNA (Kazusa clone FHC15881). PCR primers were designed to add coding sequence for N-terminal or C-terminal biotinylation tags and Flag tags as required. The HTT expression constructs were assembled using the In-Fusion PCR cloning kit (Takara) in the mammalian/insect cell vector pBMDEL (Addgene plasmid #111751), an unencumbered vector created for open distribution of these reagents.

Two expression constructs were made for the expression of full length (FL) HTT (amino acids 1 to 3144). One had an N-terminal biotinylation tag and C-terminal flag tag. The other had both the biotinylation tag and Flag tag on the C-terminal. These plasmids were created through modification of Addgene plasmid #111726, whose construction we previously described ²¹. Briefly, these modifications were done in two steps. First, a N-terminal or C-terminal fragment the HTT gene with the required tags was assembled into an intermediate plasmid. Second, these tagged fragments were transferred into Addgene plasmid #111726 by PCR amplification followed by insertion into the expression construct using the BD In-Fusion PCR cloning kit. The cDNA of E. coli BirA, corresponding to the full-length 1–321 aa protein, was subcloned into pFBOH-LIC notag, also using the BD In-Fusion PCR cloning kit.

Expression vectors for C-terminal FLAG-tagged constructs were generated by Gateway cloning. To this end, 150 ng of entry vector in a pDONR221 vector for full-length HTTQ23, HTT 81–1643, and HTT 2088–3144 was mixed with 600 ng of the destination vector pDEST_gateway-2xFLAG (Addgene plasmid #118372), 1 μl of LR clonase, and filled up to 10 μl with TE buffer pH 8.0. Reaction was incubated overnight at room temperature. Afterwards 1 μl of Proteinase kinase K was added and incubated at 37°C for 10 minutes. Reaction was then transformed into chemically competent Mach1 bacterial cells, plated onto ampicillin agar plates. Clones were picked, cultured in liquid LB medium, and DNA isolated using QiAprep Spin Miniprep Kit (Qiagen). Final constructs were sequenced to validate correct insertion using the following primers; GAGGTCTATATAAGCAGAGC and AACCATTATAAGCTGCAATAAAC.

LuTHy plasmids were generated as described previously ³⁰. In brief, open reading frames (ORFs) of full-length HTT(Q23) or HTT’s CTD were amplified from previously generated expression plasmid (Addgene plasmid #111723) and resulting attB1 and attB2-flanked PCR products were subcloned into pDONR221 entry vectors using BP Clonase (Gateway Cloning System, Invitrogen). For HAP40 (F8A1), a cDNA entry clone was obtained from Source BioSciences (OCAAo5051D1091D). To generate LuTHy expression plasmids, HAP40 was cloned into a LuTHy donor vector (Addgene plasmid #113447) and the HTT constructs into a LuTHy acceptor vector (Addgene plasmid #113449) by LR Clonase reactions (Gateway Cloning System, Invitrogen). All expression plasmids were finally validated by restriction enzyme digest, agarose gel electrophoresis, and Sanger sequencing.

The DNA sequences and expressed protein sequences of plasmids used in this paper are summarized in Data S1.

Protein expression and purification

Expression of HTT subdomains was performed in Sf9 insect cells as previously described ^12,21,48 with minor modifications. Plasmids were transformed into DH10Bac E. coli competent cells, plated onto LB-agar plates and then successfully transformed white colonies were selected and grown in 3 mL LB media supplemented with 50 μg/mL kanamycin, 7 μg/mL gentamicin, and 10 μg/mL tetracycline. Bacmids were purified using a Miniprep kit (Qiagen) but following cell lysate neutralisation, the supernatants were mixed with 0.8 mL sterile isopropanol in a fresh tube and incubated on ice for 10 mins to precipitate the bacmid DNA. The DNA was pelleted by centrifugation, the supernatant removed and the pellets left to airdry prior to resuspension in 50 μL of elution buffer from the kit.

Exponentially growing Sf9 cells were diluted to 4 × 10⁵ cells/mL in serum free insect media and 0.5 mL cell suspension/well used to seed 24-well plate to form a monolayer of cells. Plates were incubated at 27 °C for 1 h. Per well seeded in the 24 well plate, 2 μL of the X-tremeGENE 9 transfection reagent was mixed with 100 μL unsupplemented insect cell growth medium and 10 μL of a 0.2 μg/μL solution of recombinant bacmid DNA. The transfection mixture was incubated for 15–20 min to enable complex formation before addition of this mixture to a seeded well in the 24 well plate. Transfection plates were incubate for 4–5 h at 27 °C before addition of 1.5 mL of insect serum-free medium supplemented with 10% (v/v) of heat inactivated fetal bovine serum and 1% (v/v) antibiotic-antimycotic (100 units/mL of penicillin, 100 μg/mL of streptomycin, and 0.25 μg/mL of amphotericin B). Cells were incubated at 27 °C incubator for 72–96 h prior to harvesting of the P1 virus stocks and the recombinant virus titer was sequentially amplified by this method to generate P3 virus stocks.

Cells at ~4.5 million cells/ml were infected with 8 mL P3 recombinant baculovirus per 4 L of culture in 5 L reagent bottle and grown at 37°C until cell viability reached 80–85%, normally ~72 hours post-infection. The volume of culture grown can be scaled according to the required yield of protein. For full-length HTT-HAP40, a ratio of 1:1 HTT:HAP40 P3 baculovirus was used for infection. For CTD-HAP40, a ratio of 1:3 CTD:HAP40 P3 baculovirus was used. Biotinylation of AviTag constructs was achieved by co-expression with recombinant BirA in the presence of 10 μg/mL D-biotin. For purification, cells were harvested by centrifugation (JLA8.1000 (Beckman), 2500 rpm, 10 °C, 10 min) and resuspended in ~40 mL lysis buffer (20 mM HEPES pH 7.4, 300 mM NaCl, 2.5% (v/v) glycerol) per L of production. Cell suspensions were lysed with two freeze-thaw cycles. Lysates were then diluted 3-fold in lysis buffer and clarified by centrifugation (JLA16.250 (Beckman), 14,000 rpm, 10 °C, 1 h). The supernatant was incubated with ~1 mL anti-FLAG resin (Sigma) per L of production processed and rocked for 2 h at 4 °C. The resin was washed with ~100 CV lysis buffer and then proteins were eluted with 5 CV lysis buffer supplemented with 250 μg/mL 3x FLAG peptide. Crude protein samples from FLAG-eluted fractions were then pooled, concentrated using Amicon spin filters to ~0.1 CV (MWCO 100,000) and purified by size-exclusion chromatography using a Superose 6 Increase 10/300 GL column (Cytiva Life Sciences) in buffer containing 20 mM HEPES pH 7.4, 300 mM NaCl, 2.5% (v/v) glycerol, 1 mM TCEP. To avoid overloading the gel filtration column, only 0.5–1 mL sample was applied per run. Peaks corresponding to monodisperse protein were pooled, concentrated and flash frozen at 5–10 mg/mL and stored at −80°C until use. Sample purity was assessed by SDS-PAGE.

Differential scanning fluorimetry

Determination of protein thermostability by DSF was performed using a Roche Applied Science Light Cycler 480 II. Samples were prepared in LightCycler 480 white 384-well plates (Roche) at a volume of 20 μL per well using a final concentration of 0.1 mg/mL protein and 5X Sypro Orange (Invitrogen). All samples were diluted in buffer containing 20 mM HEPES pH 7.4, 300 mM NaCl, 2.5% (v/v) glycerol, and 1 mM TCEP. Thermal shift assays were carried out over a temperature range of 20 to 95°C at a ramp rate of 0.02°C/sec. Fluorescence measurements were taken using a 465 excitation / 580 emission filter set. Reactions were performed in triplicate in three independent experiments and data were analyzed using GraphPad Prism 9. Melt curves were fitted to a Boltzmann sigmoidal curve and apparent ${\text{T}}_{\text{m}}$ values were determined by calculating the inflection point of the fitted curves.

Differential static light scattering

${\text{T}}_{\text{agg}}$ values for purified proteins were determined by DSLS using a Stargazer instrument (Harbinger Biotech). Protein samples were diluted to 0.4 mg/mL in buffer containing 20 mM HEPES pH 7.4, 300 mM NaCl, 2.5% (v/v) glycerol and 1 mM TCEP using a volume of 50 μL per well. Samples were heated from 20°C to 85°C at a rate of 1°C/min and protein aggregation was monitored by measuring the intensity of scattered light every 30 s with a CCD camera. Scattering intensity was then plotted and fitted to the Boltzmann equation, and ${\text{T}}_{\text{agg}}$ values were determined by measuring the inflection point of each curve.

Cryo‐EM sample preparation and data acquisition

Full-length HTT-HAP40 or CTD-HAP40 samples were diluted to 0.2 mg/mL in 25 mM HEPES pH 7.4, 300 mM NaCl, 0.025% w/v CHAPS and 1 mM DTT and adsorbed onto gently glow-discharged suspended monolayer graphene grids (Graphenea) for 60 s. Grids were then blotted with filter paper for 1 s at 100% humidity, 4 °C and frozen in liquid ethane using a Vitrobot Mark IV (Thermo Fisher Scientific).

Data were collected in super-resolution counting mode on a Talos Arctica (Thermo Fisher Scientific) operating at 200 kV with a BioQuantum imaging filter (Gatan) and K3 direct detection camera (Gatan) at 100,000x magnification, corresponding to a real pixel size of 0.81 Å/pixel. Movies were collected at a dose rate of 22.8 e⁻/Ų/s, exposure time of 2.25 s, resulting in a total dose of 51.3 e⁻/Ų fractionated across 53 frames.

Cryo-EM data processing, model building and refinement

Movies were processed in real time using the SIMPLE 3.0 pipeline ⁴², using SIMPLE-unblur for patched motion correction, SIMPLE-CTFFIND for patched CTF estimation and SIMPLE-picker for particle picking. Particles were extracted in 288 × 288 pixel boxes, sampling of 0.81 Å/pixel.

For full-length HTT-HAP40, 112,717 particles were generated from 166 movies and subjected to 2D classification (50 classes, 160 Å mask) within cryoSPARC ⁴³. Particles (64,597) belonging to the most defined, highest resolution classes were used to generate an ab initio map. This map was lowpass filtered to 30 Å and used as reference for non-uniform refinement within cryoSPARC, yielding a 3.3 Å volume as assessed by Gold standard Fourier Shell Correlations (FSC) using the 0.143 criterion.

For CTD-HAP40, 986,229 particles were generated from 1,461 movies and subjected to initial 2D classification in SIMPLE 3.0 (200 classes, 140 Å mask). “Good” particles (819,568) were subjected to an additional round of 2D classification within cryoSPARC (200 classes, 140 Å mask). Retained particles (297,261) underwent multiclass (4 classes) ab initio volume generation within cryoSPARC, producing one sensible class composed of 134,849 particles. The volume corresponding to this class was lowpass filtered to 30 Å and used as reference for non-uniform refinement within cryoSPARC, generating a 3.2 Å volume as assessed by Gold standard Fourier Shell Correlations (FSC) using the 0.143 criterion.

The model for CTD-HAP40 was generated from our previous HTT-HAP40 model (PDBID: 6X9O) and rigid body fitted into our globally sharpened, local resolution filtered map. Multiple rounds of manual real-space refinement using Coot v. 0.9.6 ⁴⁴ and automated real-space refinement in PHENIX v. 1.19.2–4158 ⁴⁵. HTT-HAP40 model was validated using MolProbity ⁴⁶ within PHENIX. Figures were prepared using UCSF ChimeraX v.1.2 ⁴⁷ and PyMOL v.2.4.0 (The PyMOL Molecular Graphics System, v.2.0; Schrödinger).

Western blotting of recombinant protein samples

General protocols for western blot analyses were performed as previously described ¹². Primary antibodies used include anti-HTT D7F7 (Cell Signalling Technology), anti-HAP40 54731 (Novus Biologicals) and anti-FLAG F1804 (Sigma). Secondary antibodies used are goat-anti-rabbit IgG-IR800 (LI-COR) and donkey anti-mouse IgG-IR680 (LI-COR). Membranes were visualised using an Odyssey CLx imaging system (LI-COR).

Biolayer interferometry

The affinity of the CTD-HAP40 binding interaction to the NTD was determined using an OctetRED96 BLI system (ForteBio). Experiments were performed at 25°C in 96-well black microplates (Greiner, 655209) with shaking at 1000 rpm. Purified CTD-HAP40 with a C-terminal biotin tag was diluted to 1 μg/mL in buffer containing 100 mM sodium phosphate pH 7.4, 500 mM NaCl, 0.05% Tween-20, and 0.1% BSA and were loaded onto streptavidin biosensors (ForteBio, 18–5019) for 300s. Sensors were then washed in buffer for 120 s to establish a baseline reading. A serial dilution of NTD in identical buffer was prepared in the same plate, and then loaded sensors were transferred into NTD wells for a 60 s association period to allow association of the complex. Sensors were then transferred to wells containing only buffer for 300 s to allow complex dissociation. Binding kinetics were determined using GraphPad Prism 9 by fitting the data to a 1:1 association then dissociation model. Association and dissociation curves from all concentrations of NTD were fitted globally to a shared ${\text{k}}_{\text{on}}$ and ${\text{k}}_{\text{off}}$ value ⁴⁹. Background signal from reference sensors without immobilised CTD-HAP40 was subtracted from experimental sensor wells.

Cell culture and transfection

Human embryonic kidney line 293 (HEK293) wildtype and HTT-null cells were grown in Dulbecco’s modified Eagle’s medium (Thermofisher #41965) supplemented with 10% heat-inactivated fetal bovine serum (ThermoFisher #10500064), and 1% penicillin/streptocillin (ThermoFisher #15140122) at 37 °C, and 5% CO2. Cells were subcultured every three to four days. For overexpression studies, one million cells in a final volume of 2 ml complete DMEM medium were reversed transfected in 6-well plates with a transfection mix composed of two μg of DNA in a volume of 200 μl Opti-MEM and 5 μl Fugene transfection reagent (2:5:1 ratio). After 48 hours cells were trypzined, washed with ice-cold PBS, pelleted and stored at −80°C.

Western blotting of cell lysates

HEK293 cell pellets collected from a 6-well plate were lyzed in 30–50 μl HEPES lysis buffer (50 mM HEPES pH 7.0, 150 mM sodium chloride, 10 % glycerol, 1 % NP-40, 20 mM NaF, 1.5 mM MgCl2, 1 mM EDTA, 1 mM PMSF, 0.5% sodium deoxycholate, 1x Benzonase, 1x, Roche Complete EDTA-free protease inhibitor cocktail (Merck, 5056489001) for 30 min on ice. Lysates were centrifuged at 14,000 rpm for 10 min at 4°C and supernatants collected. Protein concentrations were determined using the PierceTM BCA assay (Thermo Scientific) and 20 μg total protein was combined with 50 mM DTT and 1x NuPAGE LDS sample buffer, followed by 5 min at 95°C. Proteins were separated by SDS-PAGE using a NuPAGE 4–12% Bis-Tris gel and transferred onto nitrocellulose membranes (Cytiva, 10600002). Membranes were blocked for 1 hour in 3% milk in PBS with 0.05% Tween. The following primary antibodies were applied overnight at 4°C: mouse anti-FLAG (1:1000; Sigma #F3165), rabbit anti-HAP40 (1:1000; Atlas Antibodies #HPA046960), mouse anti-tubulin (1:80,000; Sigma #T6074), and mouse anti-GAPDH (1:1000; Santa Curz #sc-47724). The following secondary antibodies diluted to 1:6000 in 3% milk and applied for two hours at room temperature: goat anti-Rabbit IgG peroxidase (Sigma #A0545) and goat anti-mouse IgG peroxidase (Sigma #A0168). Each membrane was incubated with WesternBrightTM Quantum (Advansta, K-12042-D20) solution for two minutes, followed by acquisition of a chemiluminescence image using an iBright imaging system (ThermoFisher).

In-cell interaction validation with LuTHy

LuTHy-BRET assays were performed as described previously ³⁰. In brief, cells were reverse transfected using linear polyethyleneimine (25 kDa, Polysciences 23966) with LuTHy constructs and cells were subsequently incubated for 48 h. Previously generated LuTHy control vectors expressing only NL (Addgene #113442) or PA-mCit (Addgene #113443) were used as background controls. Live cell BRET measurements were carried out in flat-bottom white 96-well plates (Greiner, 655983) with 24 PPIs per plate (each PPI in triplicate). Infinite^® microplate readers M1000 or M1000Pro (Tecan) were used for the readout with the following settings: fluorescence of mCitrine recorded at Ex 500 nm/Em 530 nm, luminescence measured using blue (370–480 nm) and green (520–570 nm) band pass filters with 1,000 ms integration time. BRET ratios were calculated by dividing the background corrected luminescence intensity at 520–570 nm by the intensity obtained at 370–480 nm and subsequent donor bleed-through subtraction from NL only expressing wells.

Quantification and Statistical Analysis

Data presented for thermal shift assays, both DSF and DSLS, are shown as representative replicates from three biological replicates. DSLS data were not normalised, but DSF data were normalized to the fraction of maximal fluorescence and both were fitted to a Boltzmann sigmoidal curve using GraphPad Prism to determine Tagg and Tm respectively. Tagg and Tm values are mean ± S.D across replicates. BLI binding kinetics were determined using GraphPad Prism 9 by fitting the data to a 1:1 association then dissociation model. Association and dissociation curves from all concentrations of NTD were fitted globally to a shared kon and koff value. Kinetic parameters reported are mean ± S.D across replicates. For LuTHy analyses, the chart shows bars which represent means ± SEM from two independent experiments performed in triplicate. Statistical significance of BRET signal between different protein pairs was calculated using one-way ANOVA with Tukey’s multiple comparisons test, **p < 0.002, *** p < 0.001.

Supplementary Material

1

Data S1: Summary of study construct details including sequences, primers and Addgene accessions, related to STAR Methods subsection, Cloning of expression constructs.

Key Resources Table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit monoclonal anti-HTT	Cell Signalling Technologies	Cat # 5656S
Rabbit polyclonal anti-HAP40	Novus Biologicals	Cat # NBP2-54731
Mouse monoclonal anti-FLAG	Sigma	Cat # F1804
Mouse monoclonal anti-FLAG	Sigma	Cat # F3165
Rabbit polyclonal anti-HAP40	Atlas Antibodies	Cat # HPA046960
Mouse monoclonal anti-tubulin	Sigma	Cat # T6074
Mouse monoclonal anti-GAPDH	Santa Cruz	Cat # sc-47724
Goat anti-Rabbit IgG-IR800	LI-COR	Cat # 926-32211
Donkey anti-Mouse IgG-IR800	LI-COR	Cat # 926-32212
Goat anti-Rabbit IgG peroxidase	Sigma	Cat # A0545
Goat anti-Mouse IgG peroxidase	Sigma	Cat # A0168
Bacterial and virus strains
E. coli: Mach1 T1R competent cells	ThermoFisher Scientific	Cat # C862003
E. coli: DH10Bac competent cells	Gibco	Cat # 10361012
Chemicals, peptides, and recombinant proteins
BP Clonase	Invitrogen	Cat # 11789013
LR Clonase	Invitrogen	Cat # 11791019
Sfx-Insect Cell Culture Media	Cytiva	Cat # SH3027802
Dulbecco’s modified Eagle’s medium	ThermoFisher	Cat # 41965
Heat-inactivated fetal bovine serum	ThermoFisher	Cat # 10500064
Opti-Mem	Gibco	Cat # 31985062
X-tremeGENE 9 Transfection Reagent	Roche	Cat #6365787001
Fugene transfection reagent	Promega	Cat # E2311
Linear polyethyleneimine, 25 kDa	Polysciences	Cat # 23966
Penicillin-Streptomycin (10,000 U/ mL)	Gibco	Cat # LS15140122
Amphotericin B	Gibco	Cat # 15290018
5X Sypro Orange	Invitrogen	Cat # S6650
Roche cOmplete EDTA-free protease inhibitor cocktail	Merck	Cat # 5056489001
1X NuPAGE LDS Sample Buffer	Invitrogen	Cat # NP9007
WesternBright Quantum Western Blotting HRP Substrate	VWR	Cat # K12042-D20
Critical commercial assays
In-Fusion HD EcoDry Cloning Kit	Takara	Cat # 638915
QiAprep pin Miniprep Kit	Qiagen	Cat # 27104
Pierce BCA Assay	Thermo Scientific	Cat # 23227
Deposited data
HTT-HAP40 model	Harding et al. 12	PDB-6X9O
Full-length HTT-HAP40 cryo-EM map	This paper	EMDB-28767
CTD-HAP40 cryo-EM map and model	This paper	EMDB-28766, PDB-8SAH
Experimental models: Cell lines
S. frugiperda: Sf9 cells	ThermoFisher Scientific	Cat # 12659017
Human: HEK293T	ATCC	Cat # CRL-11268
Human: HEK293T HTT null	Jung et al. 40	N/A
Oligonucleotides
Details in Data S1	This paper	N/A
Recombinant DNA
cDNA HTT	Kasuza	Clone # FHC15881
cDNA HAP40	Source BioSciences	Clone # OCAAo5051D1091D
pBMDEL entry vector	Harding et al. 21	Addgene Plasmid # 111751
pBacMam2-DiEx-LIC-C-flag_huntingtin_full-length_Q23	Harding et al. 21	Addgene Plasmid # 111726
pBacMam2-DiEx-LIC-N-flag_huntingtin_full-length Q23	Harding et al. 21	Addgene Plasmid # 111723
pFBOH-LIC	Arrowsmith Lab Plasmid Collection	Addgene Plasmid # 26099
pDONR221	Invitrogen	Cat # 12536017
pDEST_gateway-2xF LAG	Viita et al. 41	Addgene Plasmid # 118372
pcDNA3.1 NL	Trepte et al. 30	Addgene Plasmid # 113442
pcDNA3.1 PA-mCit	Trepte et al. 30	Addgene Plasmid # 113443
pcDNA3.1 GW-NL-myc	Trepte et al. 30	Addgene Plasmid # 113447
pcDNA3.1 GW-mCit-PA	Trepte et al. 30	Addgene Plasmid # 113449
Software and algorithms
Prism 9	GraphPad Software	https://www.graphpad.com/
Simple 3.0 Pipeline	Caesar et al. 42	https://github.com/hael/SIMPLE/releases/tag/v3.0.0
cryoSPARC	Punjani et al. 43	https://cryosparc.com/
Coot v. 0.9.6	Casanal et al. 44	https://www2.mrc-lmb.cam.ac.uk/personal/pemsley/coot/
PHENIX v. 1.19.2-4158	Afonine et al. 45	https://phenix-online.org
Molprobity	Prisant et al. 46	http://molprobity.biochem.duke.edu
ChimeraX v.1.2	Petterson et al. 47	https://www.rbvi.ucsf.edu/chimerax/
Pymol v.2.4.0	The PyMOL Molecular Graphics System, v.2.0; Schrödinger	https://pymol.org
Other
Anti-FLAG M2 Affinity Gel	Millipore	Cat # A2220
Amicon® Ultra-15 Centrifugal Filter Unit	Millipore	Cat # UFC9100
Superose 6 Increase 10/300 GL column	Cytiva	Cat # 29-0915-96
Suspended monolayer graphene grids	Graphenea	https://www.graphenea.com/products/suspended-monolayer-graphene-on-tem-grids
Streptavidin biosensors	ForteBio	Cat # 18-5019
Nu-PAGE 4-12% Bis Tris	Invitrogen	Cat # NP0322BOX
Nitrocellulose membranes	Cytiva	Cat # 10600002

Highlights:

• Expression and purification of human huntingtin (HTT) subdomain protein samples

• CryoEM analysis of HTT subdomain-HAP40 protein complexes

• Analysis of HTT subdomain-HAP40 interactions with in vitro and in cell assays