The C-terminal region of LIC1 is intrinsically disordered and contains two short segments with α-helical propensity

To gain mechanistic insight into how LIC-C interacts with dynein adaptors that are diverse in structure and function (Fig 1A and 1B, S1 Fig), we first characterized LIC1-C (residues 388–523) by NMR spectroscopy. The ¹⁵N-¹H heteronuclear single quantum coherence (HSQC) spectrum of LIC1-C at 25°C showed a narrow 7.8 to 8.6 parts per million (ppm) amide-proton chemical shift range, indicating a predominance of structural disorder (Fig 1C). Near-complete backbone resonance assignment was achieved (104 out of the 117 nonproline residues; Fig 1C, S2 Fig). An overlay of ¹⁵N-¹H HSQC spectra of two smaller constructs, consisting of residues 388–471 and 472–523, reproduced the spectrum obtained from the entire LIC1-C (S2A Fig). This finding indicates that there are no long-range interactions between N- and C-terminal segments of LIC1-C.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. LIC1(388–523) is intrinsically disordered and contains two short segments with α-helical propensity.

(A) Subunit composition of dynein. The N-terminal RAS-like domain of the LIC binds to the N-terminal tail of the HC, and the flexible LIC C-terminal region interacts with cargo adaptor proteins, as illustrated in (B). (B) Cargo adaptor proteins examined in this study that bind the C-terminal region of dynein LIC. (C) ¹⁵N-¹H HSQC spectrum of human LIC1(388–523) with backbone NH assignments. The spectra were acquired at 25°C and 900 MHz. Labels marked with an asterisk indicate peaks with heights below the cutoff contour level. (D) SSP scores for LIC1(388–523) calculated using ¹H^N, ¹⁵N, CO, CA, and CB chemical shifts for each assigned residue. “+1” indicates a fully formed α helix, “-1” indicates a fully formed β sheet, and “0” indicates disorder. GOR4 helix prediction scores are also shown. Values from 0 to 1,000 reflect absence of a fully formed α-helical structure. Scores obtained for residues 442–453 and 493–502 indicate transient α helix formation. (E) ¹⁵N relaxation measurements for LIC1(388–523). R₁ and R₂ relaxation rate constants at 700 MHz are shown for each assigned residue. (F) ¹⁵N-¹H heteronuclear NOEs of LIC1(388–523) probing amide bond vector motion. Relaxation rate constants (E) and NOEs (F) are indicative of intrinsic structural disorder, and larger values indicate more restricted sub-nanosecond motion, which are observed for the helical regions. Underlying data for Fig 1 can be found in S1 Data. BICD2, bicaudal D homolog 2; dynein, cytoplasmic dynein 1; HC, heavy chain; HOOK3, Hook homolog 3; HSQC, heteronuclear single quantum coherence; LIC, light intermediate chain; MIRO, mitochondrial Rho; NIN, ninein; NOE, nuclear Overhauser enhancement; ppm, parts per million; RAB11FIP3, RAB11 family-interacting protein 3; RILP, RAB-interacting lysosomal protein; RZZ, ROD-ZW10-ZWILCH; SPDL1, Spindly; SSP, secondary structure propensity; TRAK1, trafficking kinesin-binding protein 1.

https://doi.org/10.1371/journal.pbio.3000100.g001

To assess residual secondary structure in LIC1-C, we computed the secondary structure propensity (SSP) score developed by Forman-Kay and colleagues [24]. The method combines different backbone chemical shifts into a single residue–specific SSP score. The results closely matched those obtained from the GOR4 secondary structure prediction program [25] (Fig 1D, S1 Data). An SSP score of +1 and −1 indicates a fully formed α helix and β sheet, respectively. The SSP scores for LIC1-C residues with predicted α-helical propensity (residues 444–454 and 497–504) were positive values up to 0.35. However, assuming a linear relationship between the score and the population, helical structure was only transiently sampled at about 20%. These two segments of LIC1-C, which are highly conserved (S1A Fig), will be referred to as helix 1 and helix 2, respectively. The remaining segments scored mostly slightly negative SSP values and GOR4 values around zero, indicating complete structural disorder (note that β sheet and random coil have both extended secondary structure with similar scores in practice).

To characterize the dynamic properties of LIC1-C, we performed ¹⁵N NMR relaxation measurements (Fig 1E, S3 Fig). The relaxation rates are sensitive probes of the amplitudes and timescales of residue-specific structural fluctuations. The longitudinal relaxation rates R₁, which report on motion faster than nanoseconds, were 1.2 s⁻¹ on average. Although such a value is not compatible with a globular protein, it is indicative of an intrinsically disordered protein that undergoes extensive local reorientation. Residues 441–456 and 492–508 had larger R₁ values, which are suggestive of increased structural order of the residues with α-helical propensity. The residues between helices 1 and 2 had consistently lower-than-average R₁ values, indicating a highly flexible interhelical linker.

The fast dynamics inferred from R₁ measurements reduce the transverse relaxation rates R₂ typically observed for folded proteins, but in addition, potential motion slower than nanoseconds might increase them. Whereas the measured R₂ average of 4.7 s⁻¹ is again only reconcilable with high structural disorder, the helical regions showed larger values in line with partial helix formation (Fig 1E). Since the R₁ values suggest less flexibility in the helices on the fast timescale, it is likely that the increased R₂ values originate also exclusively from reduced fast-motion amplitudes. In support of this, relaxation dispersion experiments did not indicate any slow micro-millisecond dynamics. The segment with the smallest values was again the interhelical linker. Since slow motion is absent throughout the protein, the ratio R₂/R₁ allowed extraction of an effective residue-specific reorientational tumbling time that is independent of the motional amplitude [26]. The average tumbling time was 4.5 ns, which is much smaller than a typical value expected for a folded protein consisting of approximately 140 residues.

The ¹H-¹⁵N nuclear Overhauser enhancement (NOE) is another probe of the amplitudes of motions taking place on the sub-nanosecond timescale (Fig 1F). The values of nearly all residues fall below 0.5, which is again indicative of structural disorder with large amplitudes. In support of the findings described above, the largest values were typically found for the helix 1 and 2 segments, which suggests higher structural order than the other regions.

In conclusion, the relaxation data show that LIC1-C is intrinsically disordered with large amplitudes of motions faster than nanoseconds. In helix 1 and 2, the amplitudes are smaller and the local reorientation slower but still on the fast timescale. These results support the highly transient character of the helices derived from the SSP score.

For independent confirmation of the residual helical propensity in LIC1-C, we collected circular dichroism (CD) spectra of LIC1-C (S4 Fig). Whereas fully formed α-helical proteins show prominent negative bands at 222 and 208 nm and a positive band at 193 nm, disordered proteins have very low ellipticity above 210 nm and negative bands near 195 nm [27]. The spectrum of LIC1-C corresponded largely to structural disorder. However, there were negative values at 208 nm and a negative shoulder at 222 nm. These features are indicative of residual helical structure, in agreement with our findings from NMR spectroscopy. The helical content was maintained over a large temperature range (25, 37, 50, 60, and 70°C), similar to previous reports on other intrinsically disordered proteins [28].

The C-terminal helix 1 in LIC1 is the main binding site for SPDL1, BICD2, and HOOK3

To identify adaptor binding sites in LIC1-C, we recorded ¹⁵N-¹H HSQC spectra of LIC1-C bound to SPDL1(2–359), BICD2(2–422), and HOOK3(2–239) using molar ratios of 1:0, 1:0.5, and 1:1 (Fig 2A, 2B, 2C and 2D). The spectra obtained from each titration series revealed similar modes of interaction for the three adaptors. Rather than chemical shifts being perturbed, many peaks were significantly attenuated because of binding. This indicates decrease in the local motion of the bound residues; increase of the hydrodynamic radius of the entire complex, resulting in increase in the effective tumbling time; and/or contributions from conformational and chemical exchange [29]. Inspection of the peak height ratios between bound and free LIC1-C showed that the residues affected were mainly located in helix 1 and to a lesser extent in helix 2 (Fig 2B, 2C and 2D). In addition, we observed intensity reduction for the peaks of residues 418–421, which also tend to be conserved (S1A Fig). The quenching of segment 418–421 and helix 2 was more pronounced for SPDL1(2–359) than for BICD2(2–422) or HOOK3(2–239) (Fig 2B, 2C and 2D).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. The LIC1(388–523) interaction with the N-terminal regions of SPDL1, BICD2, and HOOK3 at single-residue resolution.

(A) Overlay of NMR ¹⁵N-¹H HSQC spectra of free ¹⁵N-labeled LIC1(388–523) (orange) and in a 1:1 mixture with unlabeled SPDL1(2–359) (blue). (B–D) Relative peak intensity quenching in the ¹⁵N-¹H HSQC spectra of ¹⁵N-labeled LIC1(388–523) upon 1:1 titration of unlabeled SPDL1(2–359) (B), BICD2(2–422) (C), and HOOK3(2–239) (D). Gray bars indicate residues for which no ratio could be obtained because of peak overlap or because they were proline residues. (E–G) SPR sensograms from injection of dilution series of SPDL1(2–359) (E), BICD2(2–422) (F), and HOOK3(2–239) (G) over immobilized LIC1(388–523). Fitted K_D values are indicated as mean ± SEM. Underlying data for Fig 2 can be found in S1 Data. BICD2, bicaudal D homolog 2; HOOK3, Hook homolog 3; HSQC, heteronuclear single quantum coherence; LIC1, light intermediate chain 1; NMR, nuclear magnetic resonance; ppm, parts per million; SPDL1, Spindly; SPR, surface plasmon resonance.

https://doi.org/10.1371/journal.pbio.3000100.g002

To assess binding affinities between LIC1-C and adaptors, we conducted surface plasmon resonance (SPR) experiments in which LIC1-C::6xHis was immobilized on a nitrilotriacetic acid (NTA) chip. We obtained a K_D of 5.7 ± 1.9 μM for the interaction between LIC1(388–523) and SPDL1(2–359) (Fig 2E). LIC1(388–471), which contains helix 1 but not helix 2, bound to SPDL1(2–359) with the same affinity as LIC1(388–523) (5.40 ± 0.25 μM). Similarly, LIC1(388–523) and LIC1(388–471) bound to BICD2(2–422) with comparable affinity (K_D of 1.07 ± 0.05 μM and 0.7 ± 0.01 μM, respectively) (Fig 2F). We also measured binding to fluorescently labeled LIC1(388–523) in microscale thermophoresis (MST) experiments. We obtained K_D values of 13.1 ± 0.4 μM and 6.0 ± 1.3 μM for the interaction with SPDL1(2–359) and BICD2(2–422), respectively, which is in reasonable agreement with the SPR analysis (S5A and S5B Fig). From SPR and MST experiments with LIC1(388–523) and HOOK3(2–239), we obtained K_D values of 1.9 ± 0.4 μM and 4.0 ± 0.8 μM, respectively (Fig 2G, S5C Fig), which are similar to those obtained with BICD2(2–422). We conclude that SPDL1, BICD2, and HOOK3 bind to LIC1-C with an affinity in the single-digit micromolar range and that an LIC1-C fragment comprising residues 388–471 is sufficient for binding. These results are in agreement with the NMR titration experiments, which showed the most pronounced intensity quenching for helix 1 residues.

The C-terminal helix 1 in LIC1 and LIC2 is essential for binding to structurally diverse adaptors

NMR spectroscopy and SPR analysis strongly suggested that helix 1 in LIC1-C is the major binding site for BICD2, SPDL1, and HOOK3. To directly test whether helix 1 is important for adaptor binding, we performed in vitro pull-down experiments using purified glutathione S-transferase (GST)-tagged versions of LIC1-C (Fig 3A and 3B) and LIC2-C (Fig 4A and 4B) as bait. In addition to BICD2(2–422), SPDL1(2–359), and HOOK3(2–552), we purified full-length versions of RILP (residues 1–401) and FIP3 (residues 2–756), as well as an N-terminal fragment of NIN (residues 1–693) (Fig 3C, S1B Fig). Using Coomassie Blue staining and immunoblotting for the Strep-tag II at the adaptor C terminus, all adaptors were readily detected in GST pull-downs with LIC1(388–523) and LIC1(388–471) (Fig 3D). By contrast, no significant binding was observed with LIC1(472–523). The same result was obtained with corresponding constructs of LIC2 (residues 375–492, 375–450, and 451–492) (Fig 4C). Deleting helix 1 in LIC1-C (Δ440–455), or mutating either the two phenylalanines or the two leucines in helix 1 to alanine (F447A/F448A and L451A/L452A) abrogated binding between LIC1-C and all six adaptors (Fig 3D). Mutating the two leucines in helix 1 of LIC2-C to alanine (L436A/L437A) had the same effect (Fig 4C). We conclude that helix 1 in LIC1-C and LIC2-C is essential for binding to cargo adaptors that are diverse in structure and function.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. LIC1 C-terminal helix 1, but not helix 2, is essential for binding to structurally diverse adaptors.

(A) LIC1-C fragments used for binding assays. (B, C) Coomassie Blue–stained protein gels of purified GST::LIC1-C fragments (B) and purified N-terminal or full-length versions of adaptors (C) (also see S1B Fig). All adaptors contain a C-terminal Strep-tag II. (D, E) GST pull-downs using the proteins shown in (B) and (C). Protein fractions bound to beads were visualized on gels by Coomassie Blue staining and by immunoblot against the Strep-tag II. Molecular mass is indicated in kDa on the left. Each pull-down was performed at least three times. BICD2, bicaudal D homolog 2; FIP3, RAB11 family-interacting protein 3; GST, glutathione S-transferase; HOOK3, Hook homolog 3; LIC1-C, C-terminal light intermediate chain 1; NIN, ninein; ppm, parts per million; RILP, RAB-interacting lysosomal protein; SPDL1, Spindly.

https://doi.org/10.1371/journal.pbio.3000100.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. LIC2 also uses its C-terminal helix 1 for binding to adaptors.

(A) LIC2-C fragments used for binding assays. (B) Coomassie Blue–stained protein gel of purified GST::LIC2-C fragments. (C) GST pull-downs using LIC2-C fragments shown in (B) and the Strep II–tagged adaptors shown in Fig 3C. Protein fractions bound to beads were visualized on gels by Coomassie Blue staining and by immunoblot against the Strep-tag II. Molecular mass is indicated in kDa on the left. Each pull-down was performed at least three times. BICD2, bicaudal D homolog 2; FIP3, RAB11 family-interacting protein 3; GST, glutathione S-transferase; HOOK3, Hook homolog 3; LIC2-C, C-terminal light intermediate chain 2; NIN, ninein; ppm, parts per million; RILP, RAB-interacting lysosomal protein; SPDL1, Spindly.

https://doi.org/10.1371/journal.pbio.3000100.g004

Helix 1 is sufficient for binding to RILP, but other adaptors require additional contacts with LIC1-C

Next, we asked whether helix 1 of LIC1-C (residues 440–455) was sufficient to bind adaptors. In GST pull-down experiments, RILP bound robustly to LIC1(388–523) and LIC1(440–455) (Fig 3A, 3B and 3E). SPR experiments in which LIC1-C was immobilized through its N-terminal GST tag confirmed that RILP binds LIC1(440–455) with similar affinity as LIC1(388–523) (K_D of 1.67 ± 0.21 and 1.18 ± 0.14 μM, respectively; S6 Fig). We conclude that helix 1 of LIC1-C is sufficient for RILP binding. By contrast, none of the other five adaptors were detected in GST pull-downs with LIC1-C helix 1, suggesting that efficient binding requires additional segments in LIC1-C (Fig 3E).

NMR spectroscopy suggested that, in addition to helix 1, residues N-terminal to helix 1, as well as residues in helix 2, also make contact with adaptors. To investigate the contribution of these sites to adaptor binding, we first truncated LIC1(388–471) from its N terminus to generate progressively smaller fragments (residues 402–471, 414–471, 424–471, and 440–471), all of which retained helix 1 (Fig 3A). SPDL1(2–359), BICD2(2–422), HOOK3(2–552), FIP3(2–756), and NIN(1–693) showed reduced binding to LIC1-C with decreasing fragment size, whereas RILP bound equally well to all LIC1-C fragments (Fig 3E). Thus, for most adaptors, efficient binding to LIC1-C requires residues located N-terminally of helix 1. To assess the contribution of helix 2, we generated a fragment from the beginning of helix 1 until the end of LIC1-C (residues 440–523). Although SPDL1(2–359), BICD2(2–422), and HOOK3(2–552) were not detected in pull-downs using LIC1(440–455), they bound weakly to LIC1(440–523), potentially reflecting a modest contribution of helix 2 to the interaction (Fig 3E). Overall, our binding experiments with purified components demonstrate that helix 1 of LIC-C is essential for the interaction with adaptors but suggest that most adaptors also need to make additional contacts with the flexible LIC-C scaffold for efficient binding. Specifically, residues preceding helix 1 appear to make a significant contribution. The exception is RILP, for which LIC-C helix 1 on its own is sufficient for binding.

Point mutations in the C-terminal helix 1 of C. elegans LIC cause severe developmental defects, impair locomotion, and shorten life span

To determine whether helix 1 of LIC-C contributes to dynein function in vivo, we turned to the animal model C. elegans, which expresses a single dynein LIC, dli-1 [30]. Just as in other LIC orthologs, the C-terminal helix 1 is highly conserved in DLI-1 (Fig 5A). We used clustered regularly interspaced short palindromic repeat/CRISPR-associated 9 (CRISPR/Cas9)–mediated genome editing to mutate either the two phenylalanines or the two leucines in helix 1 to alanine (F392A/F393A and L396A/L397A), in analogy to the human LIC-C mutations we characterized in vitro (Fig 5B). For comparison, we also deleted the entire DLI-1 C-terminal region (Δ369–443). For all three dli-1 mutants, homozygous offspring from heterozygous mothers developed to adulthood but were sterile, suggesting that the C-terminal region of DLI-1, and helix 1 in particular, is essential for dynein function in vivo (Fig 5B). Because RNA interference (RNAi)-mediated depletion of DLI-1 showed that DLI-1 is required for the stability of dynein HC 1 (DHC-1; S7 Fig), we addressed the possibility that the phenotype of our dli-1 mutants was a consequence of reduced DHC-1 levels. Immunoblotting of homozygous adults with an antibody against DHC-1 demonstrated that this was not the case. On the contrary, DHC-1 levels appeared slightly increased in all three dli-1 mutants (Fig 5C and S7B Fig). This indicates that the C-terminal DLI-1 mutants do not interfere with the binding of the N-terminal GTPase-like domain to DHC-1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Point mutations in the C-terminal helix 1 of C. elegans DLI-1 cause severe developmental defects, impair locomotion, and shorten life span.

(A) C-terminal sequence alignment of the two human LIC paralogs and the sole LIC ortholog in C. elegans, DLI-1, showing high conservation of helix 1. Residues mutated to alanine are marked by asterisks. (B) (left) Cartoon showing DLI-1 WT and mutant versions (Δ369–443, F392A/F393A, or L396A/L397A) expressed in animals after CRISPR/Cas9-mediated genome editing. (right) Genetics of dli-1 mutant alleles. Mothers heterozygous for the mutant dli-1 allele contain a balancer chromosome with a WT copy, dli-1(wt). Progeny homozygous for the mutant dli-1 allele develop to adulthood but are sterile. All three dli-1 mutants behave the same. (C) Immunoblot showing levels of DHC-1 in adults of dli-1 mutants. α-Tubulin is used as a loading control. Also see S7 Fig. (D–F) Differential interference contrast images of animals at the day 1 adult stage, showing extensive morphological defects in dli-1 mutants. Scale bars: (D) 100 μm; (E) and (F) 20 μm. (G) Locomotion of day 1 adult animals, assessed by determining body bending frequency in liquid medium. Graph shows mean ± SEM for n number of animals from two independent experiments. Statistical significance (mutant versus WT dli-1) was determined by one-way ANOVA on ranks (Kruskal-Wallis nonparametric test) followed by Dunn’s multiple comparison test. ****P < 0.0001. (H) Life span curves. Animals were collected as L4 larvae (day 0) and observed every 1–3 d until they died. Mean life span (± SEM) is indicated for n number of animals. Underlying data for Fig 5 can be found in S1 Data. CRISPR/Cas9, clustered regularly interspaced short palindromic repeat/CRISPR-associated 9; DHC-1, dynein heavy chain 1; DLI-1, dynein LIC 1; LIC, light intermediate chain; WT, wild type.

https://doi.org/10.1371/journal.pbio.3000100.g005

Differential interference contrast imaging revealed severe morphological defects in day 1 adults homozygous for the dli-1 mutations (Fig 5D, 5E and 5F). All three dli-1 mutants had a slightly dumpy body with a protruding vulva and a highly disorganized and underdeveloped gonad (Fig 5D and 5E), which is indicative of problems with postembryonic cell divisions. Furthermore, the prevalence of vacuoles suggested widespread necrotic-like cell death (Fig 5F). Consistent with this, dli-1 mutants exhibited severe defects in locomotion (Fig 5G), which was assessed by determining the body bending frequency in liquid medium (wild type, 1.50 ± 0.02 Hz; Δ369–443, 0.50 ± 0.03 Hz; F392A/F393A, 0.46 ± 0.04 Hz; L396A/L397A, 0.50 ± 0.03 Hz). Dli-1 mutants had a significantly shorter life span compared to wild-type animals (wild type, 18.3 ± 0.7 d; Δ369–443, 13.7 ± 0.4 d; F392A/F393A, 12.3 ± 0.5 d; L396A/L397A, 13.0 ± 0.5 d) (Fig 5H). None of the dli-1 mutant animals lived beyond 24 d, whereas 27% of wild-type animals did. We conclude that deletion of the DLI-1 C-terminal disordered region and point mutations in the conserved helix 1 cause similar defects in development, locomotion, and life span. These phenotypes are reminiscent of those reported for other dynein mutants, including null mutants of dli-1 [30,31].

Point mutations in the C-terminal helix 1 of DLI-1 cause accumulation of membrane cargo at nerve endings of mechanosensory neurons and perturb retrograde axonal transport

Next, we sought to assess the impact of the C-terminal dli-1 mutations on dynein-dependent cargo transport. We chose to examine the distribution and transport kinetics of membrane cargo in axons of touch receptor neurons, which are mechanosensory neurons whose processes extend along the length of the animal just underneath its cuticle (Fig 6A). Axonal microtubules in touch receptor neurons are polarized with their plus ends oriented toward the axonal tip, and dynein is therefore responsible for retrograde transport of cargo toward the cell body [31,32]. Using Mos transposase-mediated single-copy insertion [33], we constructed transgenic animals expressing mKate2::RAB-5 from the mec-7 promoter and crossed them with animals expressing soluble green fluorescent protein (GFP) from the mec-4 promoter (allele zdIs5). This allowed simultaneous visualization of early endosomes (mKate2), a cargo of dynein, and neuronal morphology (GFP) (Fig 6A). In contrast to control day 1 adults, all three dli-1 mutants exhibited a pronounced misaccumulation of mKate2::RAB-5 at the axonal tips and nerve ring synapses of anterior lateral mechanosensory (ALM) and anterior ventral mechanosensory (AVM) neurons (Fig 6B, 6C, 6D and 6E). This defect resembled the previously described misaccumulation of synaptic vesicle proteins in touch receptor neurons of dynein mutants [31].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. C-terminal dli-1 mutations disrupt retrograde axonal transport of early endosomes.

(A) Cartoon showing the architecture of anterior touch receptor neurons. Animals coexpress mKate2::RAB-5, a marker for early endosomes, and soluble GFP in touch receptor neurons. ALM and AVM are neurons that extend processes into the nose and the nerve ring. Results shown in (B–I) derive from imaging specific regions in these neurons, as indicated. (B, C) Fluorescence images of axonal tips (B) and nerve rings (C) in day 1 adults, showing misaccumulation of early endosomes in dli-1 mutants. Scale bars, 10 μm. (D, E) Quantification of early endosome misaccumulation in axonal tips (D) and nerve rings (E), using fluorescence intensity measurements of mKate2::RAB-5 in images as shown in (B) and (C), respectively. Graphs represent the mean ± SEM signal in A.U. for n number of neurons imaged in two independent experiments. Statistical significance (mutant versus WT dli-1) was determined by one-way ANOVA on ranks (Kruskal-Wallis nonparametric test) followed by Dunn’s multiple comparison test. ****P < 0.0001. Also see S1 Movie. (F) Kymographs of mKate2::RAB-5-labeled early endosome motility in axons of ALM neurons, imaged at the L4 larval stage. Time-lapse sequences of confocal images were recorded every 200 ms in an axonal region approximately 50 μm away from the cell body, which is positioned to the right. Scale bar, 10 μm. (G–I) Quantification of early endosome motility, based on the analysis of kymographs as shown in (F). Graphs represent the mean ± SEM. For (G), n is the total number of axons. For (H) and (I), n is the total number of segments within tracks of moving particles framed by a pause or a reversal. Results are derived from 2–5 independent imaging sessions. Statistical significance was determined as described for (D, E). ****P < 0.0001; ***P < 0.001; ns indicates P > 0.05. Also see S8A Fig. Underlying data for Fig 6 can be found in S1 Data. ALM, anterior lateral mechanosensory; A.U., arbitrary units; AVM, anterior ventral mechanosensory; GFP, green fluorescent protein; ns, not significant; WT, wild type.

https://doi.org/10.1371/journal.pbio.3000100.g006

To determine how dli-1 mutants affect the kinetics of axonal transport, we imaged mKate2::RAB-5-labeled early endosomes at high temporal resolution in an axonal segment close to the cell body of ALM neurons at the larval L4 stage (Fig 6A and 6F and S1 Movie). Quantitative analysis of kymographs revealed that the three dli-1 mutants had near-identical defects in axonal transport. The fraction of particles moving in the retrograde direction was reduced significantly, with a corresponding increase in the fraction of stationary particles (Fig 6G). Analysis of individual runs showed that particles spent less time in retrograde motion and paused more frequently and for longer periods of time (S8A Fig). Particles also had a decreased run length and moved at reduced speed, especially in the retrograde direction (Fig 6H and 6I). Overall, this analysis revealed that all three dli-1 mutants strongly impaired retrograde axonal transport of early endosomes, consistent with compromised dynein function.

We also examined the axonal distribution of two additional types of dynein cargo, each labeled with a marker expressed from a single-copy transgene: synaptic vesicles (SNB-1 [synaptobrevin 1]::mKate2) and mitochondria (TOMM-20 [translocase of outer mitochondrial membrane 20][1–54]::mKate2). Because the three dli-1 mutants had identical defects in the distribution and transport kinetics of mKate2::RAB-5, we restricted our analysis to the dli-1(L396A/L397A) mutant. We found that dli-1(L396A/L397A) animals misaccumulated SNB-1::mKate2 at axonal tips (S8B, S8C and S8D Fig), similar to the results of a prior study that examined the distribution of SNB-1::GFP in the null mutant dli-1(js351) [31].

When examining the distribution of TOMM-20(1–54)::mKate2 particles (Fig 7A), we found that axons of ALM neurons contained an average of 21 ± 1 mitochondria in control day 1 adults, corresponding to a density of 5.6 mitochondria per 100 μm, and that mitochondria were evenly distributed along axons (Fig 7B, 7C, 7D and 7E). These findings are consistent with prior reports [34,35]. Mitochondria remained evenly distributed in ALM axons in dli-1(L396A/L397A) animals (Fig 7E). However, the average number of axonal mitochondria increased to 33 ± 1, corresponding to a density of 9 mitochondria per 100 μm (Fig 7C and 7D). Close inspection of mitochondrial morphology suggested that the increase was primarily due to an excess of small TOMM-20(1–54)::mKate2 puncta, which may represent fragmented mitochondria (Fig 7B). Thus, the dli-1(L396A/L397A) mutation does not result in distribution bias of mitochondria along the axon but appears to promote mitochondrial fission. Similar results were obtained for the dli-1(Δ369–443) mutation (Fig 7C, 7D and 7E). Analysis of mitochondrial transport kinetics in dli-1(L396A/L397A) animals revealed a reduction in the speed of transport, with a more pronounced effect in the retrograde direction (Fig 7F and 7G, and S2 Movie). However, none of the other transport parameters were significantly altered (S8E and S8F Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. C-terminal dli-1 mutants have increased numbers of mitochondria that move with reduced retrograde velocity.

(A) Cartoon highlighting the axons of ALM neurons examined in (B–G). Animals coexpress TOMM-20(1–54)::mKate2, a marker for mitochondria, and soluble GFP in these neurons. (B) Fluorescence images of ALM axons in animals at the day 1 adult stage. Blow-ups highlight the prevalence of small (<1 μm) TOMM-20(1–54)::mKate2 puncta in the dli-1(L396A/L397A) mutant. Scale bars, 10 μm; blow-ups, 2.5 μm. (C, D) Total number of mitochondria per ALM axon (C) and axonal length (D) in day 1 adults. Graphs represent the mean ± SEM. Statistical significance (mutant versus WT dli-1) was determined by one-way ANOVA on ranks (Kruskal-Wallis nonparametric test) followed by Dunn’s multiple comparison test. ****P < 0.0001; ns indicates P > 0.05. (E) Relative distribution of mitochondria along the ALM axon from the proximal (0%) to the distal (100%) end in day 1 adults. Graph corresponds to the mean, and the number n of axons imaged for each condition is indicated. (F) Kymographs of mitochondrial motility in axons of ALM neurons, imaged at the L4 larval stage. Time-lapse sequences of confocal images were recorded every 5 s in an axonal region approximately 50 μm away from the cell body, which is positioned to the right. Scale bar, 10 μm. Also see S2 Movie. (G) Quantification of mitochondrial transport velocity, based on the analysis of kymographs as shown in (F). Graphs represent the mean ± SEM. n is the total number of segments within tracks of moving particles framed by a pause or a reversal. Results are derived from 5–8 independent imaging sessions. Statistical significance was determined as described for (C, D). ****P < 0.0001; ***P < 0.001. Also see S8E and S8F Fig. (H) Coomassie Blue–stained protein gel of purified GST-tagged human LIC1-C and purified MBP::TRAK1 fragments containing a C-terminal Strep-tag II (also see S1B Fig). Two different versions of the first coiled-coil segment of TRAK1 were fused to a fragment of the yeast transcriptional activator GCN4 for artificial dimerization. (I) GST pull-downs using the proteins shown in (H). Protein fractions bound to beads were visualized on gels by Coomassie Blue staining and by immunoblot against the Strep-tag II. Molecular mass is indicated in kDa on the left. Each pull-down was performed at least three times. Underlying data for Fig 7 can be found in S1 Data. ALM, anterior lateral mechanosensory; GCN4, general control nondepressible 4; GFP, green fluorescent protein; GST, glutathione S-transferase; LIC1-C, C-terminal light intermediate chain 1; MBP, maltose-binding protein; ns, not significant; TRAK1, trafficking kinesin-binding protein 1; WT, wild type.

https://doi.org/10.1371/journal.pbio.3000100.g007

Given the modest impact of the dli-1(L396A/L397A) mutation on mitochondrial transport, we asked whether the mitochondrial adaptor TRAK1, which contains a CC1 box similar to SPDL1 and BICD2, binds LIC1. GST pull-downs with purified human proteins showed that maltose-binding protein (MBP)::TRAK1(103–187), which corresponds to the first coiled-coil segment, was able to bind GST::LIC1(388–523), albeit weakly and only when artificially dimerized using a general control nondepressible 4 (GCN4) sequence (Fig 7H and 7I). The interaction was abolished when the two leucines in LIC1-C helix 1 were mutated, suggesting that LIC1 binds TRAK1 through a similar mechanism as the other adaptors.

The C-terminal helix 2 of DLI-1 is largely dispensable for dynein function in vivo

Just like helix 1, helix 2 in LIC-C shows high sequence conservation, including in C. elegans (residues 420–428; Fig 8A and S1A Fig). To assess the role of helix 2 in vivo, we used genome editing to generate animals that express DLI-1 without its C-terminal 30 residues (Δ414–443). This truncated version of DLI-1 retains helix 1 (residues 388–400) (Fig 8A). In contrast to the helix 1 point mutants, homozygous dli-1(Δ414–443) animals were fully viable and fertile (embryonic viability 98.4% ± 0.6% and 97.3% ± 0.7% for wild type [>300 progeny, 12 mothers] and mutant [>300 progeny, 14 mothers], respectively). Furthermore, we observed no increase in mKate2::RAB-5 signal at axonal tips of touch receptor neurons, indicating that dynein-dependent transport of early endosomes was not significantly affected in this mutant (Fig 8B, 8C and 8D). Because homozygous dli-1(Δ414–443) animals produced progeny, we were able to examine the first mitotic division of embryos coexpressing GFP::β-tubulin and mCherry::histone H2B, which mark microtubules and chromosomes, respectively (Fig 8E). During mitosis, dynein-dependent pulling on centrosome-nucleated microtubules is essential for the separation and positioning of centrosomes, which form the spindle poles after nuclear envelope breakdown (NEBD). At the time of NEBD in control animals, centrosomes were fully separated and the centrosome–centrosome axis was oriented approximately parallel to the anterior–posterior axis (Fig 8E and 8F). In dli-1(Δ414–443) embryos, centrosomes separated normally, but the centrosome–centrosome axis was frequently severely tilted relative to the anterior–posterior axis (angle > 45° in 7 out of 21 mutant embryos versus 0 out of 13 controls), resulting in misoriented spindles. Despite initial misorientation, spindles ultimately correctly aligned with the anterior–posterior axis, and chromosome segregation completed without defects. The delay in spindle orientation indicates a subtle impairment of dynein-dependent pulling forces [36,37]. To further probe the functional significance of helix 2, we combined the dli-1(Δ414–443) mutation with a null allele of the dynein cofactor nud-2, nud-2(ok949), which represents a sensitized dynein background that also exhibits a spindle orientation delay and that we had previously exploited for an enhancer screen [36,38]. We observed neither an increase in lethality nor enhanced defects in spindle orientation in the double mutant relative to the single mutants (embryonic viability 85.9% ± 1.6% and 89.8% ± 1.4% for ok949 [>300 progeny, 15 mothers] and ok949; dli-1[Δ414–443] [>300 progeny, 19 mothers], respectively), consistent with the idea that dynein function in dli-1(Δ414–443) animals is relatively unperturbed.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Deletion of the C-terminal helix 2 in DLI-1 affects mitotic spindle positioning in the one-cell embryo.

(A) Sequence alignment of human and C. elegans LIC-C showing the deleted region in DLI-1 that includes the conserved helix 2. (B) Cartoon highlighting the axons of ALM neurons examined in (C) and (D). Animals coexpress mKate2::RAB-5 and soluble GFP in these neurons. (C) Fluorescence images of ALM axonal tips in day 1 adults. Scale bar, 10 μm. (D) Quantification of mKate2::RAB-5 levels at axonal tips, using fluorescence intensity measurements in images as shown in (C). Graph represents the mean ± SEM signal in A.U. for n number of neurons imaged in two independent experiments. Statistical significance (mutant versus WT dli-1) was determined with the Mann-Whitney test. ns indicates P > 0.05. (E) (top) Cartoon illustrating that animals homozygous for dli-1(Δ414–443) are viable and fertile. (bottom) Stills from time-lapse sequences in the one-cell embryo showing the delay in mitotic spindle orientation in dli-1(Δ414–443). Scale bar, 10 μm. (F) (top) Cartoon showing the angle (α) between the C-C axis and the A-P axis in one-cell embryos at NEBD. (bottom) Rose diagrams of angle α measured at NEBD in images as shown in (E). A significant fraction of dli-1(Δ414–443) embryos have a severely tilted C-C axis relative to controls (α > 45°; Fisher’s exact test, P = 0.029). Underlying data for Fig 8 can be found in S1 Data. ALM, anterior lateral mechanosensory; A-P, anterior–posterior; A.U., arbitrary units; C-C, centrosome–centrosome; DLI-1, dynein LIC 1; GFP, green fluorescent protein; LIC, light intermediate chain; LIC-C, C-terminal LIC; NEBD, nuclear envelope breakdown; ns, not significant; WT, wild type.

https://doi.org/10.1371/journal.pbio.3000100.g008

We conclude that helix 2 of the DLI-1 C-terminal region makes a modest contribution to dynein-dependent pulling forces during mitosis but overall plays a minor role in vivo compared to helix 1, which is critical for dynein function. These results are consistent with our biochemical characterization of human LIC-C in vitro, which shows that helix 1, but not helix 2, is essential for the interaction between LIC-C and cargo adaptors.

DNA constructs for protein expression

The cDNAs for expression of human DYNC1LI1 (UniProt ID: Q9Y6G9; residues 388–523, 388–471, 402–471, 414–471, 424–471, 440–471, 440–455, 440–523, and 472–523) and human DYNC1LI2 (UniProt ID: O43237; residues 375–492, 375–450, and 451–492) were cloned into vector pGEX-6P-1 with a single N-terminal tryptophan and a C-terminal linker (GSGSG) followed by 6xHis. The cDNAs for human BICD2 (UniProt ID: Q8TD16; residues 2–422), RAB11FIP3 (UniProt ID: O75154; residues 2–756), HOOK3 (UniProt ID: Q86VS8; residues 2–239 and 2–552), NIN (UniProt ID: Q8N4C6; residues 1–693), SPDL1 (UniProt ID: Q96EA4; residues 2–359), and TRAK1 (UniProt ID: Q9UPV9; residues 103–167 and 103–187, with and without a C-terminal fusion to the GCN4 dimeric coiled-coil sequence VKQLEDKVEELLSKNAHLENEVARLKKLV [GCN4^CC]) were cloned into a 2CT-derived vector containing an N-terminal 6xHis::MBP fusion followed by a linker with a Tobacco Etch Virus nuclear-inclusion-a endopeptidase (TEV protease) cleavage site and containing a C-terminal linker (GSGSGR) followed by the Strep-tag II. The cDNA of RILP (UniProt ID: Q96NA2; residues 1–401) was cloned into the pACEBac1 vector with a C-terminal linker (GSGSGR) followed by the Strep-tag II.

Protein expression and purification from bacteria

All bacterial expression vectors were transformed into the Escherichia coli strain BL21, except for the NIN and HOOK3 constructs, which were transformed into the E. coli strain Rosetta. Expression was induced with 0.1 mM IPTG at 18°C overnight at an OD₆₀₀ of 0.9, and cells were harvested by centrifugation for 20 min at 4,000g.

For GST::LIC::6xHis constructs used in pull-down experiments, bacterial pellets were resuspended in lysis buffer A (50 mM HEPES, 250 mM NaCl, 0.1% [v/v] Tween 20, 10 mM EDTA, 10 mM EGTA, 5 mM DTT, 1 mM phenylmethanesulfonyl fluoride [PMSF], 2 mM benzamidine-HCl, 1 mg/mL lysozyme [pH 8.0]), disrupted by sonication, and cleared by centrifugation at 34,000g for 45 min. GST::LIC::6xHis was purified by tandem affinity chromatography using glutathione agarose resin (Thermo Fisher Scientific) followed by HisPur Ni-NTA resin (Thermo Fisher Scientific). Glutathione agarose resin was incubated in batch with the cleared lysate and then washed with wash buffer A (25 mM HEPES, 250 mM NaCl, 0.1% Tween 20, 1 mM DTT, 2 mM benzamidine-HCl [pH 8.0]), and proteins were eluted on a gravity column with elution buffer A (50 mM HEPES, 150 mM NaCl, 10 mM reduced L-glutathione, 1 mM DTT, 2 mM benzamidine-HCl [pH 8.0]). Fractions containing the recombinant proteins were pooled, incubated in batch with Ni-NTA resin, and washed with wash buffer B (25 mM HEPES, 250 mM NaCl, 25 mM imidazole, 0.1% Tween 20, 1 mM DTT, 2 mM benzamidine-HCl [pH 8.0]). Proteins were eluted on a gravity column with elution buffer B (50 mM HEPES, 150 mM NaCl, 250 mM imidazole, 1 mM DTT, 2 mM benzamidine-HCl [pH 8.0]). Fractions containing the proteins were pooled and dialyzed against storage buffer (25 mM HEPES, 150 mM NaCl [pH 7.5]) or further purified by size-exclusion chromatography using a Superose 6 10/300 column (GE Healthcare) equilibrated with storage buffer. Glycerol and DTT were added to final concentrations of 10% (v/v) and 1 mM, respectively, and aliquots were flash-frozen in liquid nitrogen and stored at −80°C.

Purification of LIC1::6xHis (residues 388–523, 388–471, and 472–523) for NMR spectroscopy, SPR, and MST experiments was carried out as described above with the following modifications: GST::LIC1::6xHis was captured using a GSTrap FF column (GE Healthcare) and eluted with elution buffer A. The GST moiety was cleaved off in solution with PreScission Protease, glutathione was removed by dialysis (50 mM HEPES, 150 mM NaCl [pH 8.0]), and the sample was applied again to a GSTrap FF column to remove GST and GST-tagged Prescission Protease. The flow-through containing LIC1::6His was subjected to size-exclusion chromatography using a Superdex 75 increase 10/300 GL column (GE Healthcare) in NMR buffer (50 mM sodium phosphate, 150 mM NaCl [pH 6.5]).

For purification of cargo adaptors, bacterial pellets were resuspended in lysis buffer B (50 mM HEPES, 250 mM NaCl, 10 mM imidazole, 0.1% Tween 20, 1 mM DTT, 1 mM PMSF, 2 mM benzamidine-HCl, 1 mg/mL lysozyme [pH 8.0]), disrupted by sonication, and cleared by centrifugation at 34,000g for 45 min. The 6xHis::MBP::adaptor::Strep-tag II proteins were purified by tandem affinity chromatography using HisPur Ni-NTA resin followed by Strep-Tactin Sepharose resin (IBA). HisPur Ni-NTA resin was incubated in batch with the cleared lysate and then washed with wash buffer B, and proteins were eluted on a gravity column with elution buffer B. Fractions containing the recombinant proteins were pooled, incubated overnight with TEV protease to cleave off the 6xHis::MBP moiety (except for TRAK1 fragments, which were not cleaved), incubated in batch with Strep-Tactin Sepharose resin, and washed with wash buffer A. Proteins were eluted on a gravity column with elution buffer E (100 mM Tris-HCl, 150 mM NaCl, 1 mM EDTA, 2.5 mM desthiobiotin [IBA] [pH 8.0]). Fractions containing the proteins were pooled and dialyzed against storage buffer or further purified by size-exclusion chromatography using a Superose 6 10/300 column equilibrated with storage buffer. Glycerol and DTT were added to final concentrations of 10% and 1 mM, respectively, and aliquots were flash-frozen in liquid nitrogen and stored at −80°C.

Protein expression and purification from insect cells

Bacmid recombination and virus production were performed as described previously [54]. A 500-mL culture (SFM4 medium; Hyclone) of Sf21 cells (0.8 × 10⁶ cells/mL) was infected with RILP::Strep-tag II-encoding virus. Cells were harvested by centrifugation at 800g for 5 min. Pellets were resuspended in lysis buffer C (50 mM HEPES, 250 mM NaCl, 1 mM DTT [pH 8.0]) supplemented with EDTA-free cOmplete Protease Inhibitor Cocktail (Roche), sonicated, and cleared by centrifugation at 34,000g for 45 min. RILP::Strep-tag II was purified by batch affinity chromatography using Strep-Tactin Sepharose. The resin was washed with wash buffer C (25 mM HEPES, 250 mM NaCl, 0.1% [v/v] Tween 20, 1 mM DTT [pH 8.0]), and the protein was eluted on a gravity column with elution buffer E. Fractions containing RILP::Strep-tag II were pooled and dialyzed against storage buffer. Glycerol and DTT were added to final concentrations of 10% and 1 mM, respectively, and aliquots were flash-frozen in liquid nitrogen and stored at −80°C.

GST pull-down assays

Purified GST::LIC::6xHis constructs (50 pmol) were incubated with 250 pmol SPDL1(2–359)::Strep-tag II, 50 pmol BICD2(2–422)::Strep-tag II, 250 pmol NIN(1–693)::Strep-tag II, 50 pmol HOOK3(2–552)::Strep-tag II, 50 pmol RAB11FIP3(2–756)::Strep-tag II, 50 pmol RILP(1–401)::Strep-tag II, 250 pmol 6xHis::MBP::TRAK1(103–187)::Strep-tag II, 250 pmol 6xHis::MBP::TRAK1(103–167)::GCN4^CC::Strep-tag II, or 250 pmol 6xHis::MBP::TRAK1(103–187)::GCN4^CC::Strep-tag II for 1 h at 4°C in 150 μL pull-down buffer (50 mM HEPES, 100 mM NaCl, 5 mM DTT [pH 7.5]) containing 0.1% Tween 20 and supplemented with 15 μL of glutathione agarose resin. After washing the resin with 3 × 500 μL of the same buffer, proteins were eluted with pull-down buffer containing 15 mM reduced L-glutathione.

Immunoblotting

For immunoblots of purified proteins, samples were resolved by 10% SDS-PAGE and transferred to 0.2-μm nitrocellulose membranes (GE Healthcare). Membranes were blocked in PBS (4 mM KH₂PO₄, 16 mM Na₂HPO₄, 115 mM NaCl [pH 7.4]) containing 3% (w/v) BSA and 0.5% (v/v) Tween 20 and probed at 4°C overnight with mouse StrepMAB-Classic antibody (IBA) at 1 μg/mL in PBS containing 0.2% BSA and 0.1% Tween 20. Membranes were washed three times with PBS/0.1% Tween 20 (PBST), incubated with goat anti-mouse antibody coupled to HRP (Jackson ImmunoResearch, 1:10,000) for 1 h at room temperature, and washed again three times with PBST. Proteins were visualized by chemiluminescence using Pierce ECL Western Blotting Substrate (Thermo Fisher Scientific) and X-ray film (Amersham, GE Healthcare).

For immunoblots of C. elegans, 100 adult hermaphrodites were collected into M9 buffer and processed for immunoblotting as described [37]. Samples were resolved on a gradient gel (4%–20%) and transferred to 0.2-μm nitrocellulose membranes. Membranes were blocked with 5% (w/v) nonfat dry milk in TBST (20 mM Tris, 140 mM NaCl, 0.1% Tween [pH 7.6]) and probed at 4°C overnight with rabbit anti-DHC-1 antibody GC4 (1:1,400, made in-house), mouse anti-FLAG M2 antibody (1:1,000, Sigma-Aldrich), or mouse anti-α-tubulin B512 antibody (1:5,000, Sigma-Aldrich). Membranes were sequentially rinsed 3× with TBST, 1× with 5% nonfat dry milk in TBST, and 3× with TBST. Membranes were incubated with goat secondary antibody coupled to HRP (Jackson ImmunoResearch, 1:10,000) for 1 h at room temperature and washed again 3× with TBST, 1× with 5% nonfat dry milk in TBST, and 3× with TBST. Proteins were visualized as described above.

NMR spectroscopy

For backbone resonance assignment of LIC1(388–523)::6xHis, ¹⁵N-¹H HSQC, HNCACB, CBCA(co)NH, HNCO, HN(ca)CO, and HNN spectra [55] were recorded on a triple-resonance Varian 900 NMR cryoprobe spectrometer at 25°C using a ¹³C/¹⁵N-labeled sample in 50 mM sodium phosphate and 150 mM NaCl at pH 6.5 and 375 μM protein concentration in a standard 5-mm Shigemi tube. The 3D spectra were acquired with a nonuniform sampling (NUS) scheme generated by the NUS@HMS scheme generator software [56] with 1,024 complex data points in the direct dimension and 30% sampling of the original 96 and 80 points in the indirect ¹³C and ¹⁵N dimension, respectively. The spectral widths were 14,045 Hz (¹H), 3,200 Hz (¹⁵N), 3,770 Hz (C = O), and 15,835 Hz (Cα/Cβ); the interscan delay was 1.7 s; and the number of scans was 16 for all experiments. The NUS-acquired data were reconstructed using the hmsIST software [56]. Zero-filling was achieved by addition of 256 points in both indirect dimensions. A solvent subtraction function was applied in the direct dimension. Further data processing and visualization were performed using NMRPipe/NMRDraw [57] and NMRFAM Sparky [58]. Resonance assignment was performed using CCPNmr Analysis software [59].

Because of high sequence redundancy and extensive peak overlap, we used a “divide-and-conquer” approach for chemical shift assignment. We measured and overlaid ¹⁵N-¹H HSQC spectra for two smaller LIC1 constructs, 388–471 and 472–523, with 388–523 to facilitate the verification of assignment. The ¹⁵N-¹H HSQC spectra of the small constructs were measured with 128 scans and 128 complex points in the indirect dimension.

We assessed residual secondary structure using the SSP score program developed by Forman-Kay and colleagues with the re-referencing algorithm for ¹³CA and ¹³CB shifts [24]. The method combines different chemical shifts into a single residue–specific SSP score. Our input shifts were those of ¹H^N, ¹⁵N, ¹³CO, ¹³CA, and ¹³CB.

The ¹⁵N R₁ and R_1ρ relaxation rate constants and ¹⁵N-¹H heteronuclear NOEs of LIC1(388–523)::6xHis were measured on a Bruker 700 MHz spectrometer equipped with a triple-resonance cryoprobe at 25°C using a ¹⁵N-labeled sample at 400 μM protein concentration in 50 mM sodium phosphate buffer, 50 mM NaCl, 0.05% (w/v) NaN₃, and 5 mM DTT at pH 6.5. For the R₁ and R_1ρ experiments, the sampling time points were 40, 88, 136, 192, 288, 392, 592, 688, 792, and 992 ms and 30, 60, 120, 150, 180, and 210 ms, respectively. During the R_1ρ relaxation time, a ¹⁵N spin-lock field of 1,433-Hz strength was applied. R₂ was calculated from R₁ and R_1ρ using the following equation: R₂ = R_1ρ + (R_1ρ–R₁)*tg²(θ), where θ = tan^-1(2πΔν/γ_NB₁), Δν is the resonance offset, |γ_NB₁|/2π is the strength of the spin-lock field B₁, and γ_N the gyromagnetic ratio of the ¹⁵N spin. The ¹⁵N-¹H heteronuclear NOEs were determined from two spectra recorded in presence and in the absence of ¹H saturation in an interleaved manner.

For NMR titration analysis, ¹⁵N-¹H HSQC spectra with 128 complex points in the indirect dimension and 128 scans were recorded on a triple-resonance Varian 900 NMR cold-probe spectrometer at 25°C of 40 μM samples of ¹⁵N-labeled LIC1(388–523)::6xHis alone and in the presence of 0.5 and 1 equivalent of unlabeled SPDL1(2–359)::Strep-tag II, BICD2(2–422)::Strep-tag II, or HOOK3(2–239)::Strep-tag II in 50 mM sodium phosphate and 150 mM NaCl at pH 6.5. We used freshly prepared samples for each titration step to limit confusion with potential degradation peaks. Data processing and visualization were performed using NMRPipe/NMRDraw [57] and NMRFAM Sparky [58].

The ¹H, ¹³C, and ¹⁵N chemical shift assignments have been deposited in the BioMagResBank database (http://www.bmrb.wisc.edu) under the accession number 27401.

SPR

SPR analysis with SPDL1, BICD2, and HOOK3 fragments was conducted with a Biacore 3000 system. His-tagged LIC1(388–523) and LIC1(388–471) constructs (ligand) were immobilized on three different flow cells (FC1–3) on an NTA sensor chip at densities of 200–300 RU, whereas the fourth flow cell (FC4) was spared for blank sensogram measurement. Concentrated stocks of Strep-tagged SPDL1(2–359), BICD2(2–422), and HOOK3(2–239) (analyte) were dialyzed exhaustively against HBS-P flow buffer (10 mM HEPES, 150 mM NaCl, 0.005% [v/v] surfactant P₂₀ [pH 7.4]). The background-blank sensogram was subtracted from sensograms measured with immobilized ligands. Injections for each analyte concentration were performed in triplicate. Data processing was done on Biacore evaluation software.

SPR analysis with RILP::Strep-tag II was performed using a Biacore X100 system equipped with a CM5 sensor chip (GE Healthcare). Anti-GST antibody was immobilized using amine-coupling chemistry using the Amine Coupling Kit (GE Healthcare) and the GST Capture Kit (GE Healthcare) according to manufacturer's instructions. The surfaces of flow cells 1 and 2 were activated for 7 min with a 1:1 mixture of 0.1 M N-hydroxysuccinimide and 0.4 M 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide at a flow rate of 5 μL/min. Anti-GST antibody at a concentration of 30 μg/mL in 10 mM sodium acetate (pH 5.0) was immobilized at a density of 7,500 RU on flow cells 1 and 2. Surfaces were blocked with a 7-min injection of 1 M ethanolamine (pH 8.0). Anti-GST antibody high-affinity sites were blocked with 3 cycles of a 3-min injection of recombinant GST at 5 μg/mL in running buffer (10 mM HEPES, 150 mM NaCl, 3 mM EDTA, 0.05% surfactant P₂₀ [pH 7.4]) followed by a 2-min injection of regeneration solution (10 mM glycine-HCl [pH 2.1]). To collect kinetic binding data, at the beginning of each cycle, GST::LIC1(388–523)::6xHis or GST::LIC1(440–455)::6His in running buffer (ligand) was injected over flow cell 2 at a density of 720–530 and 550–460 RU, respectively. Flow cell 1 was injected with GST::6xHis at a density of 575–430 RU to serve as a reference surface. RILP::Strep-tag II (analyte) was injected in running buffer over the two flow cells at concentrations of 50, 11, 4.5, 1.8, and 0.3 μM at a flow rate of 30 μL/min and a temperature of 25°C. The complex was allowed to associate and dissociate for 120 and 600 s, respectively. Surfaces were regenerated with two 2-min injections of regeneration solution. Three independent runs were performed for each condition. The data were fit to a 1:1 interaction model using the evaluation module of the Biacore X100 software, version 2.0.1 (GE Healthcare).

MST

Measurements were carried out on a NanoTemper Monolith NT.115 pico instrument (NanoTemper Technologies) at 25°C using medium power and 20% excitation power (auto-detect-pico-red). LIC1(388–523)::6xHis was fluorescently labeled by interacting 100 μL protein solution (200 nM concentration) with 100 μL Red-Tris NTA dye (100 nM concentration). The reaction mixture was incubated at room temperature for 30 min followed by centrifugation for 10 min at 4°C with 15,000g. Then, 10 nM of labeled LIC1(388–523)::6xHis and 16 two-fold dilution series of the adaptor were loaded into 16 standard capillaries (NanoTemper Technologies) (SPDL1[2–359]::Strep-tag II highest concentration 42.5 μM; BICD2[2–422]::Strep-tag II highest concentration 36.5 μM; HOOK3[2–239]::Strep-tag II highest concentration 70 μM). We observed sigmoidal behavior of the fluorescence level over time, which allowed characterization of the interactions. Raw data were analyzed using the NanoTemper software (MO affinity analysis v 2.2.7). The signal-to-noise ratios for SPDL1(2–359)::Strep-tag II, BICD2(2–422)::Strep-tag II, and HOOK3(2–239)::Strep-tag II were 18.8, 22.1, and 7.1, respectively.

CD

CD spectra of LIC1(388–523)::6xHis were collected on a J-815 CD spectrometer (Jasco) with a wavelength range from 190 to 250 nm, data pitch 1 nm, standard sensitivity, 1 nm bandwidth and 20 nm/min scanning speed, at temperatures of 25, 37, 50, 60, and 70°C. The baseline of the spectrum was obtained from measurement of the buffer (50 mM sodium phosphate, 150 mM NaCl [pH 6.5]) and subtracted from the spectra of the samples to remove artificial CD signals that might originate from the optical system. Data measurement and analysis were performed using Spectra Manager version 2 (Jasco). Molar ellipticity was calculated as m°M/(10LC), where m° is the degree value (mdeg), M is the molecular weight in g/mole, L is the path length in cm, and C is the concentration in g/L.

C. elegans strains

Worm strains (S1 Table) were maintained at 20°C on standard nematode growth media (NGM) plates seeded with OP50 bacteria. A Mos1 transposon-based strategy (MosSCI) was used to generate strains stably expressing mKate2::RAB-5, SNB-1::mKate2, and TOMM-20(1–54)::mKate2 in touch receptor neurons [33]. Transgenes were cloned into pCFJ151 for insertion on Chromosome 2 (ttTi5605 locus), and transgene integration was confirmed by PCR. C-terminal mutants of dli-1 (Δ369–443, F392A/F393A, L396A/L397A, and Δ414–443) and 3xflag::dli-1 were generated by CRISPR/Cas9-mediated genome editing, as described previously [60,61]. Genomic sequences targeted by sgRNAs are listed in S2 Table. Modifications in genomic DNA sequence were confirmed by sequencing, and strains were outcrossed 6 times against the wild-type N2 strain to remove potential background mutations. Other fluorescent markers were subsequently introduced by mating. The dli-1 mutants Δ369–443, F392A/F393A, and L396A/L397A were maintained as heterozygotes using the GFP-marked genetic balancer nT1 [qIs51]. Homozygous F1 progeny from balanced heterozygous mothers were identified by the absence of GFP fluorescence.

Life span assay

Animals were collected at the L4 stage (day 0) and transferred every 2 d to a new NGM plate with bacteria. Animals were scored as alive or dead every 1–3 d. Animals were considered dead if they did not respond when touched with a platinum wire and if there was no evidence of pharyngeal pumping. Animals that were found dead on the edge of the plate, escaped, or died because of internal hatching of progeny were excluded from the assay.

Imaging

Image analysis

Image analysis was performed using Fiji software (Image J version 1.52d).

Statistical analysis

Statistical analysis was performed with GraphPad Prism 7.0 software. Values in figures and text are reported as mean ± SEM. The type of statistical analysis performed is described in each figure legend. Differences were considered significant at P < 0.05.