Depleting endogenous DISC1 results in loss of primary cilia

To investigate this question, we first asked if depleting endogenous DISC1 protein affected cilia number in NIH3T3 cells. To accomplish this, we identified two independent siRNA duplexes producing reliable knockdown of endogenous DISC1 (Fig. 2A; scanning densitometry estimated >80% reduction in immunoreactive DISC1 protein). As a positive control, we verified two RNA duplexes depleting the essential ciliary protein IFT88 (Fig. 2B).

Open in a separate window

Figure 2

Depletion of endogenous DISC1 and IFT88 by RNA interference.

NIH3T3 cells were transfected with the indicated siRNA duplexes (sequences are listed in Materials and Methods). Total cell extracts were prepared 72 hours later, and levels of DISC1 (A) or IFT88 (B) protein were assessed by immunoblot using antibodies recognizing the endogenous proteins. Representative immunoblots are shown. Densitometric scanning across multiple experiments verified >80% depletion of both proteins by the respective siRNA duplexes.

Using the nuclear stain DAPI (blue) to identify individual cells, and acetylated tubulin (green) to mark cilia, we observed primary cilia on the majority of cells transfected with control (non-silencing) siRNA duplex (Fig. 3A, top row of panels; a particular example is indicated by arrow). In cells transfected with DISC1 siRNA, in contrast, we rarely observed a detectable cilium (Fig. 3A, middle row). This effect of DISC1 depletion was observed using both of the siRNA duplexes silencing DISC1. This effect was qualitatively similar to that of knocking down IFT88 (Fig. 3A, bottom row), whose depletion is already known to prevent cilia formation [10], [11]. We verified these results quantitatively by counting cilia-bearing cells across multiple experiments (Fig. 3B). To test specificity, we attempted to rescue this phenotype using a lentiviral vector encoding a human-derived DISC1-GFP construct that is resistant to knockdown by the rodent-specific duplexes (see Materials and Methods ). Primary cilia were observed in the majority of knockdown cells following re-expression of DISC1-GFP (Fig. 4A), and quantification across multiple experiments indicated essentially complete rescue relative to the scrambled siRNA control (Fig. 4B). Together, these results suggest that DISC1 plays an important role in the normal formation or maintenance of primary cilia in this model cell system.

Open in a separate window

Figure 3

DISC1 knockdown reduces cilia number.

(A) NIH3T3 cells were transfected with the indicated RNA duplexes and then stained with anti-acetylated tubulin (to mark primary cilia, left panels) and DAPI (to mark nuclei, middle panels). Merged images (right panels) show acetylated tubulin and DAPI staining in green and blue, respectively. Cilia were prominently observed in the majority of cells transfected with control (non-silencing) RNA duplexes (top row of images, an example is indicated by arrow). In cells transfected with either siRNA targeting DISC1, the vast majority of cells did not project a detectable primary cilium. Depleting IFT88 (bottom row of images) produced a similar effect. (B) Quantification of cilia loss by counting the number of cells extending a primary cilium, as marked by acetylated tubulin immunostaining, observed in blinded analysis of the indicated populations of siRNA-transfected cells. Bars represent the mean fraction of cells with a visible cilium, averaged across 5 experiments counting ≥250 cells/condition in each. Error bars represent the s.e.m. calculated across the experiments (***, p<0.0001).

Open in a separate window

Figure 4

Rescue of the DISC1 knockdown phenotype by human-derived DISC1-GFP.

NIH3T3 cells were transfected with siRNA targeting endogenous murine DISC1 and transduced with lentiviral particles encoding human-derived DISC1-GFP lacking the target sequence. (A) Example images from the rescue experiment. Top panels show the negative control condition, in which cells were transfected with scrambled (non-silencing) siRNA and infected with a lentivirus encoding GFP. Middle panels show the knockdown condition, in which cells were transfected with DISC1 siRNA and infected with lentivirus encoding GFP. Bottom paneles show the rescue condition, in which cells were transfected with DISC1 siRNA and infected with lentivirus encoding the siRNA-resistant DISC1-GFP construct. Left panels show acetylated tubulin immunoreactivity, middle panels show GFP fluorescence, and right panels show the merge image with inset as indicated by box. (B) Quantification of the rescue effect by cilia count (n=4 experiments).

Knockdown and recombinant protein expression

For knockdown of endogenous DISC1 in NIH3T3 cells, the following siRNA duplexes were obtained from Qiagen:

siRNA#1 (mM_DISC1_4): r(CACGGAGACCAGGCUACAUGA)

siRNA#2 (mM_DISC1_2): r(CAGCUGGAGGUCACUUCCUUA)

For knockdown of IFT88 in NIH3T3 cells we used the following siRNA duplexes, also obtained from Qiagen:

siRNA#1 (mM_IFT88_1), r(AAGGCAUUAGAUACUUAUAAA)dTdT

siRNA#2 (mM_IFT88_4)r(UUGGAGCUUAUUACAUUGAUA)dTdT.

The scrambled control sequence used was r(AAU UCU CCG AAC GUG UCA CG)dT

Duplexes were transfected using Lipofectamine RNAi-max (Invitrogen) using the optimized protocol provided by the manufacturer for NIH3T3 cells. In all experiments reagent amounts were scaled according to surface area of the specific culture dishes used, based on the optimized protocol listed for 24-well plates. Experiments were conducted 3 days after siRNA transfection without starvation.

For knockdown of endogenous DISC1 in neurons, short hairpin sequences were cloned as DNA oligos into pLKO.1 with AgeI/EcoRI, then shuttled into the dual short hairpin GFP expression vector pHUGW-GFP [42], [43].

shRNA DISC1 #1: GGCTACATGAGAAGCACAG

shRNA DISC1 #2: CAGCTGGAGGTCACTTCCTT

shRNA DISC1#1 is a previously validated sequence which has been shown to target mouse DISC1 [44]. The identical target sequence is present in rat DISC1.

For knockdown of endogenous IFT88 in neurons,

shRNA IFT88 #1: GCCCTCAGATAGAAAGACCAA

shRNA IFT88 #2: GCAGGAAGACTGAAAGTGAAT

The scrambled control sequence used was TCCTAAGGTTAAGTCGCCCTCT

For recombinant protein re-expression in both NIH3T3 cells and cultured neurons, DISC1-GFP was mutated by site directed mutagenesis to C GGGTATATGC in order to assure resistance to knockdown by rodent-targeted siRNA#1 and sh#1duplex (the duplexes used in rescue experiments; underlined residues show synonymous mutations introduced). This construct was then cloned by PCR into the pHUGW vector downstream of the Ub promoter using the restriction enzyme sites XbaI/AgeI.

Lentiviruses were generated from the constructs described above in fresh HEK293FT cells (Invitrogen) cultured in 10 cm dishes containing DMEM-H21, 10% FBS, 4 mM L-Glutamine, 1 mM MEM sodium Pyruvate, 0.1 mM MEM Non-Essential Amino Acids, 1% penicillin-streptomycin, and 500 ug/ml G418. Cells were transfected with pLKO.0 or pHUGW and the packaging vectors psPAX2, pVSV-G, at a ratio of 2/2/1 respectively, utilizing Lipofectamine 2000 (Invitrogen) as tranfection reagent. Medium was changed the next day and new media collected the following day. We centrifuged the media at 3000×g for 15 min to pellet cell debris. We took the supernatant and filtered through a Millex-HV 0.45 µM PVDF filter (Millipore) and transferred it to a PEG-it Virus Precipitation solution (System Biosciences) and refrigerated overnight. Twelve hours later we centrifuged the lentivector-containing particles at 1500×g for 5 minutes and removed all traces of fluid, taking care to not disturb the pellet. We resuspended the pellet in 100 µl of OPTI-MEM and snap-froze individual aliquots at −70 C.

In order to achieve efficient knockdown of endogenous DISC1 without cytotoxic effects (which we frequently observed in both NIH3T3 cells and striatal neurons expressing the tagged construct at high levels as estimated by GFP fluorescence intensity), we co-transduced with different viruses, one encoding the silencing shRNA alone and another including the shRNA-resistant replacement construct. This achieved efficient knockdown of endogenous DISC1 while allowing independent titration of recombinant DISC1-GFP expression. With our preparations of viral particles we achieved optimal rescue at a ratio of 5/1, with 5 representing lentivirus containing shRNA and 1 being the lentivirus encoding DISC1-GFP (or GFP as a negative control). This strategy resulted in >80% knockdown of endogenous DISC1 (estimated from densitometry of immunoblots, see Results ) and uniform, low levels of expression of DISC1-GFP in 90–95% of neurons (assessed by visualization of GFP fluorescence in the transduced cell population).

Fluorescence microscopy

Colocalization of DISC1-GFP with acetylated tubulin or adenylyl cyclase III (ACIII) was visualized with cells plated on nitric acid etched coverslips and then treated with poly-D-lysine (Sigma) overnight. Cells were fixed with 3.7% formaldehyde and permeabilized with 0.1% Triton X- 100, and 3% milk in PBS. We incubated cells with mouse anti acetylated tubulin (Sigma, 1 µg/ml for 60 min) and rabbit anti adenylyl cyclase III-C20 (Santa Cruz Biotech, 0.8 µg/ml for 60 min), and then probed with goat anti rabbit Alexa594 (Invitrogen) and goat anti mouse Alexa488 (Invitrogen) respectively for 20 minutes. DISC1-GFP was localized relative to PCM1 using a rabbit antibody (H262, Santa Cruz Biotech, sc-67204, 1200 for 60 min) that has been previously validated [45]. In triple localization with mouse anti acetylated tubulin, detection of each protein was accomplished using goat anti rabbit Alexa594 and goat anti mouse Alexa647 (both from Invitrogen).

For cilia counts in NIH3T3 cells and primary striatal neurons, we stained for acetylated tubulin in NIH3T3 cells and ACIII in neurons as indicated above. Approximately 60% of NIH3T3 cells formed acetylated tubulin positive cilia in control conditions, and 40% formed ACIII positive cilia in primary dissociated striatal cultures. We arrived at the above numbers by counting the number of acetylated tubulin or ACIII positive cilia, respectively, and dividing it by the number of DAPI-positive cells. Each siRNA and short hairpin RNA condition was measured in the same manner. For each siRNA and short hairpin, experiments were conducted at least 5 times on separate days. In each experiment, ≥250 cells were examined for each condition.

Surface receptor immunoreactivity was assayed by incubating intact, non-permeabilized NIH3T3 cells with rabbit anti FLAG antibody (Sigma, 1 µg/ml for 15 min), washed, and fixed with 3.7% formaldehyde. Surface receptor immunoreactivity for non-permeabilized striatal primary dissociated neurons was assayed by incubating with mouse M1 anti-FLAG antibody (Sigma, 1 µg/ml for 15 min). Cells were washed and permeabilized with 0.1% Triton X- 100 in PBS, 3% milk, then incubated mouse anti acetylated tubulin (Sigma, 1 µg/ml for 60 min) or Rabbit anti AC III (Santa Cruz Biotech, 0.8 µg/ml for 60 min) followed by goat anti-rabbit Alexa594 and goat anti-mouse Alexa488 conjugate (Invitrogen) respectively.

All specimens were imaged by epifluorescence microscopy, using standard dichroic filter sets (Chroma) and a 60×, numerical aperture 1.4 objective (Nikon). Images were captured using a cooled CCD camera (Princeton Instruments) and exposures adjusted to avoid saturation. Acquired images were rendered with Adobe Photoshop software using linear lookup tables.

Biochemical methods

Cell monolayers were washed three times in ice-cold phosphate-buffered saline (PBS) and lysed in extraction buffer (0.1% Triton X- 100, 150 mM NaCl, 25 mM KCl, 25 mM Tris, pH 7.4) supplemented with a standard protease inhibitor mixture (Roche Applied Science). Extracts were clarified by centrifugation (20,000×g for 15 min) and then mixed with lithium dodecylsulfate (LDS) sample buffer for denaturation and 1% β-mercaptoethanol for reduction, and incubated for 5 minutes at room temperature. Total protein levels for each well were normalized to each other by averaging 3 measurements of Coomassie Plus in a 96 well plate reader. Proteins present in the extracts were resolved by LDS-PAGE using 4–12% BisTris gels (NuPAGE; Invitrogen), transferred to nitrocellulose membranes, and probed for tagged protein by immunoblotting using the indicated primary antibody. Horseradish peroxidase-conjugated donkey anti-rabbit IgG (Amersham Biosciences) was used as secondary antibody, as appropriate, followed by detection of immunoreactivity using SuperSignal detection reagent (Pierce). Apparent molecular mass was estimated using commercial protein standards (Novex Sharp prestained protein standard, Invitrogen). Band intensities of unsaturated immunoblots were analyzed and quantified by densitometry using FluorChem 2.0 software (AlphaInnotech Corp.). Antibodies used were Rabbit-DISC1 (midi and C-term; Invitrogen catalog number 40–6900 and 40–6800, respectively; used at 1.5 µg/ml overnight), Rabbit anti IFT88 (Proteintech; used at 1/1000). The midi antibody was raised to a proprietary peptide immunogen corresponding to a middle portion (between residues 400 and 450) of NP_777279.1. The C-term antibody was raised to a proprietary peptide between residues 520 and 570 of NP_777279.1. Both antibodies have been previously characterized [46], [47].

We are indebted to Vu Dang for expert advice on working with cultured neurons, and for providing neuronal cultures used in the present study. We also thank Seth Shipman, Jon Levy and Eva LaDow for advice on the preparation and handling of lentiviral constructs. We thank Nick Brandon, Steve Hamilton, James Hislop, Sarah Kotowski, Robert Malenka, Jeremy Reiter, Nicole Santos and Hubert Van Tol for valuable discussion and generously sharing reagents.