Stocks, cultures and chemicals

Cells (stock 51s P. tetraurelia, sensitive to killer) were grown in wheatgrass medium inoculated with Klebsiella pneumoniae or Aerobacteraerogenes (adapted from [28]). Allchemicals were obtained from Sigma-Aldrich (St Louis, MO, USA) unless otherwisenoted.

Sequence analysis and construct design

BLAST searches in the Paramecium annotated genome were completed using thehuman sequence for TMEM67 (Q5HYA8) for MKS3 and the humanIFT88 (NP_783195) and mouse Tg737 (NP_033402) sequences forIFT88 orthologs. Searches identified GSPATG00015939001 as a potentialortholog for MKS3, which was used to create the RNAi construct. Fivepotential orthologs (GSPATG00038505001, GSPATG00021390001, GSPATG00011771001,GSPATG00022644001 and GSPATG00039556001) were identified for IFT88. Theconstruct to target IFT88 mRNA was designed from GSPATG00038505001. Homologyof these genes to those in other organisms is shown in Additional file 1: Tables S1 and S2.

All constructs were created from genomic DNA, which was collected by organicextraction. Briefly, 100 μl of cells were mixed 1:1 with denaturing buffer(Promega, Madison, WI, USA), mixed 1:1 with phenol:chloroform:isoamyl alcohol(25:24:1) and centrifuged for 5 minutes at 12,000 × g(Centrifuge 5424; Eppendorf, Hauppauge, NY, USA). The aqueous phase was removed,mixed 1:1 with chloroform:isoamyl alcohol (24:1) and spun again. The DNA wasprecipitated 2:1 with cold isopropanol for 20 minutes at -20°C and spun for 10minutes at 4°C (Centrifuge 5424). Pellets were rinsed twice with 75% ethanol,dried and resuspended in water.

FLAG-tag of MKS3

To localize MKS3p, we added the coding sequence for a threefold repeated FLAGsequence (DYKDDDDK) to the 5′ end of the genomic DNA sequence forGSPATG00015939001 in the pPXV plasmid using the restriction enzymes ApaI and SacI(USB/Affymetrix, Cleveland, OH, USA). These cut sites were created using largeprimers to add them to either end of the sequence: forward(5′-gcggggcccatgctaatttatatcg-3′) and reverse(5′-cgcgagctctcatattagaaaccttttgtc-3′). PlatinumPfx Polymerase(Invitrogen/Life Technologies, Grand Island, NY, USA) was used per the vendor’sinstructions to amplify the sequence. A total of 75 ng of genomic DNA was usedin each PCR: 94°C for 5 minutes; five cycles of 94°C for 1 minute,40°C for 1 minute and 68°C for 3 minutes; five cycles of 94°C for 1minute, 48°C for 1 minute and 68°C for 3 minutes; ten cycles of 94°Cfor 1 minute, 58°C for 1 minute and 68°C for 3 minutes; seventeen cycles of94°C for 1 minute, 65°C for 1 minute and 68°C for 3 minutes; and onecycle of 68°C for 15 minutes (Techne TC-4000 Thermal Cycler; KrackelerScientific, Albany, NY, USA). The products were cleaned using the PrepEasy GelExtraction Kit (Affymetrix). The resulting DNA was treated with restriction enzymes,cleaned again using the PrepEasy Gel Extraction Kit and ligated into thepPXV-5′-3xFLAG plasmid using the Ligate-IT Kit (Affymetrix). The mixture wasthen transformed into OneShot competent cells (Invitrogen/Life Technologies), and theresulting colonies were screened for positives. Positive clones were sequenced at theVermont Cancer Center DNA Analysis Facility (University of Vermont, Burlington, VT,USA).

Plasmid injection

Approximately 200 μg of pPXV-3xFLAG-MKS3 was linearized with NotI(Affymetrix) overnight at 37°C and then cleansed using an organic extractionmethod modified from that described earlier. This procedure required two washes inphenol:chloroform:isoamyl alcohol (25:24:1) followed by two washes ofchloroform:isoamyl alcohol (24:1). The final pellet was resuspended in50 μl of MilliQ water (EMD Millipore, Billerica, MA, USA), and theconcentration was checked using a spectrophotometer (Agilent Technologies, SantaClara, CA, USA). The sample was spun at 16,000 × g(Eppendorf Centrifuge 5424) for 10 minutes to pellet debris. The top 45 μlwas carefully removed and placed in a fresh RNase/DNase-free 1.5-ml Eppendorf tubeand again dried in a speed vac. The final pellet was resuspended in MilliQ water toobtain a concentration between 3 and 9 μg/μl and stored at 4°Cuntil injection.

Approximately 20 cells which had recently undergone autogamy were placed underhigh-temperature silicon oil to immobilize them. Approximately 5 to 50 pg of theplasmid was injected into the macronucleus of each cell using a pulled capillary anda Narishige micromanipulator (Narishige International USA, East Meadow, NY, USA).Individual injected cells were transferred to 750 μl of inoculated culturefluid in depression slides and incubated in a humidifying chamber at RT for 2 days,allowing the cells to recover and divide. Cells were then transferred to test tubeswith inoculated culture fluid and maintained at 15°C as individual clones.Genomic DNA was extracted from the clone cultures as described previously (seeSequence analysis and construct design text section) and tested by PCR usingplasmid-specific primers: the forward primer for the plasmid pPXV(5′-taagatgaatggaatataatg-3′) and a reverse primer(5′-gaaaacccaagccaatcaatac-3′), which was sequence-specific forMKS3. DNA (1 μl, approximately 400 ng) was used in eachPCR: one cycle at 95°C for 5 minutes followed by 30 cycles at 95°C for 1minute, 40°C for 1 minute and 72°C for 3 minutes followed by one15-minute cycle at 72°C.

Localization, visualization and analysis of FLAG-MKS3p

We tested small cultures of individual clones to ascertain whether the cellsexpressed the protein and where it was localized. A 10-ml culture of injected cellswas added to 50 ml of inoculated culture fluid and grown at 22°C forapproximately 48 to 72 hours. The cells were immunostained and imaged as describedbelow. Images were analyzed for colocalization using softWoRx Pro software (AppliedPrecision, Issaquah, WA, USA). Experiments were repeated five times.

To isolate pellicle membrane and whole cilia membrane, wild-type (stock 51s P.tetraurelia) cells expressing FLAG (control) or FLAG-MKS3 (Test) weremaintained in large cultures (3 to 6 L of culture fluid) at 22°C until adensity of 8,000 to 12,000 cells/ml was achieved. For pellicular membrane, cells wereharvested as described previously [29]. Inseparate experiments, cilia were separated from cell bodies and collected aspreviously described [30] up to the point ofseparation of the ciliary membrane from the axoneme. Protein concentrations weredetermined using a bicinchoninic acid protein assay (Thermo Scientific, Pittsburgh,PA, USA) and equalized between the test and control. Samples were separated on a 12%SDS-PAGE gel after adding 1 μl of β-mercaptoethanol and boiling for 5minutes. One hundred micrograms of pure pellicular membrane and 400 μg ofwhole cilia were loaded, along with 10 μl of a Pierce Biotechnologythree-color prestained protein molecular weight marker (Thermo Scientific). Proteinswere transferred onto nitrocellulose membrane (Pall Gelman Versapor; KrackelerScientific, Albany, NY, USA) and blocked for 1 hour using 5% nonfat dry milk, 2%Telost gelatin from fish, 3% normal goat serum (Vector Laboratories, Burlingame, CA,USA), in Tris-buffered saline Tween 20 (TBS-T) (15 mM Tris-Cl, 140 mM NaCl,0.1% v/v Tween 20, pH 7.5). Blots were probed with a 1:2,500 dilution ofrabbit Anti-FLAG M2 clone or 1:10,000 mouse anti-tubulin in the blocking buffer.Blots were incubated overnight while rocking at 4°C. Buffers were removed, theblots were rinsed three times in TBS-T and then incubated for 1 h in 1:10,000goat anti-rabbit or anti-mouse alkaline phosphatase (AP)-conjugated secondaryantibody. Blots were rinsed again four times in TBS-T for 15 minutes for each washand developed using nitroblue tetrazolium/5-bromo-4-chloro-3′-indolyl phosphateAP (Moss, Inc, Pasadena, MD, USA).

RNAi by feeding construct

Constructs for RNAi were created from genomic DNA using the following primers:MKS3 forward, 5′-gaaaacccaagccaatcaatac-3′ and reverse,5′-ggtcgacaatctgaaggataag-3′; and IFT88 forward,5′-caattaaggaaaaccacctg-3′ and reverse,5′-aaaactaacaggattgtcatct-3′. All PCR conditions began with an initialstep at 95°C for 5 minutes and ended with a final stage at 72°C for 20minutes. The MKS3 RNAi construct was amplified by 30 cycles at 95°C for1 minute, 52°C for 1 minute and 72°C for 2 minutes. The IFT88construct was amplified by five cycles at 95°C for 1 minute, 47°C for 1minute and 72°C for 2.25 minutes; followed by twenty-five cycles at95°C for 1 minute, 50°C for 1 minute and 72°C for 2.25 minutes(Techne Thermal Cycler; Bibby Scientific, Burlington, NJ, USA). The final PCRproducts were analyzed on 0.75% or 1.0% agarose gel (Invitrogen/Life Technologies)and visualized with ethidium bromide. Resulting PCR products were cloned directlyinto pCR2.1-TOPO vector (Invitrogen/Life Technologies), transformed into OneShotcells (Invitrogen/Life Technologies), and sequenced. Correct sequences were cut fromthe pCR2.1-TOPO vector and ligated into the double-T7 promoter vector L4440 (AddGene,Cambridge, MA, USA) using the Ligate-IT Kit (USB/Affymetrix) as per the kitinstructions. Escherichia coli strain Ht115 (DE3), which lacks RNaseIII, wastransformed with 50 ng of plasmid DNA for either MKS3 orIFT88. As a control, Ht115 cells were transformed with L4440 with no insert.Bacterial cultures were maintained with tetracycline (12.5 μg/ml) andampicillin (AMP) (100 μg/ml).

RNAi by feeding

Overnight cultures of Ht115(DE3) transformed with RNAi or control plasmids were usedto inoculate 50 ml of LB-AMP (100 μg/mL) and grown until the 595-nmoptical density reached 0.3 to 0.4, at which point isopropylβ-D-1-thiogalactopyranoside (IPTG) (RPI, Mount Prospect, IL, USA) was added to afinal concentration of 0.125 mg/ml. Cultures were incubated with shaking for 3hours at 37°C to induce the production of double-stranded RNA. Paramecia thathad recently undergone autogamy were collected by centrifugation and resuspended in10 ml of Dryl’s solution (1 mM Na₂HPO₄,1 mM NaH₂PO₄, 1.5 mM CaCl₂, 2 mMNa-citrate, pH 6.8) to purge bacteria from their surfaces and food vacuoles.

The induced bacteria were collected by centrifugation at4,000 × g (Beckman J2-21 centrifuge, JA-14 rotor;Beckman Coulter, Brea, CA, USA) at 4°C and resuspended in 100 mL of wheatculture medium containing an additional 8 μg/mL stigmasterol,0.125 mg/mL IPTG (RPI), and 100 μg/ml AMP. Approximately 50 to 100 ofthe purged paramecia were added to the control culture. In the case of theMKS3 and IFT88 RNAi cultures, 4,000 and 8,000 cells were addedto 100 ml, respectively. Cultures were maintained at 28°C, and after 24hours, an additional 0.125 mg/ml of IPTG (RPI) and 800 μg ofstigmasterol were added. Growth rates of cultures were determined by counting cellsat 24, 48 and 72 hours of growth. All experiments were repeated a minimum of threetime and all cultures were harvested or observed after approximately 72 hours ofgrowth unless noted otherwise.

Immunofluorescence

Cultured cells (100 ml) were collected by centrifugation (Damon IEC DivisionClinical Centrifuge, Needham Heights, MA, USA) and rinsed twice in 100 ml ofDryl’s solution. The cell volume was reduced to approximately 100 μlin a 1.5-ml Eppendorf tube before 1 ml of PHEM and 0.1% or 0.5% Triton X-100)was added. Cells were undisturbed for 1 to 4 minutes, then spun at250 × g (Damon IEC Division Clinical Centrifuge) andthen the supernatant was removed and the pellet (cells) was mixed with 1 ml offixation buffer (2% or 4% paraformaldehyde (Electron Microscopy Sciences, Hatfield,PA, USA), 2 mM NaH₂PO₄•H₂O, 8 mMNa₂HPO₄, 150 mM NaCl, pH 7.5). Samples wereundisturbed for 10 minutes or rocked for 1 hour at room temperature (RT) and washedthree times in 1 ml of blocking buffer (2 mMNaH₂PO₄•H₂O, 8 mMNa₂HPO₄, 150 mM NaCl, 10 mM EGTA, 2 mMMgCl₂, 0.1% Tween 20, 1% or 3% bovine serum albumin (BSA),pH 7.5).

Primary antibodies for the immunostaining for localization were as follows:FLAG-MKS3: mouse anti-FLAG, M2 clone at a 1:300 dilution (Sigma-Aldrich) andanti-centrin at a 1:1,000 dilution (anti-Tetrahymena centrin, gift from MarkWiney, University of Colorado, Boulder, CO, USA). For ciliary measurements, we usedmouse anti-α-tubulin at a dilution of 1:200 (Sigma-Aldrich). For visualizationof basal bodies, we used anti-centrin at a dilution of 1:1,000. For cortical unitvisualization, we used anti-2F12 at a dilution of 1:200 (gift from Jean Cohen,Gif-sur-Yvette, France). For the visualization of the kinetodesmal fibers (KDFs), weused anti-KDF at a 1:400 dilution (gift from Janine Beisson, Centre deGénétique Moléculaire, Gif-sur-Yvette, France) andanti-Glu-α-tubulin at a 1:500 dilution (Synaptic Systems, Göttingen,Germany). Primary antibodies in 100 μl of blocking buffer were mixed withthe cells and rocked at RT for 1 hour. Cells were washed three times in blockingbuffer or wash buffer (2 mM NaH₂PO₄•H₂O,8 mM Na₂HPO₄, 150 mM NaCl, 0.1% Tween 20, 1% BSA,pH 7.5). The cells were mixed with 100 μl of blocking buffer with a1:200 dilution of secondary antibodies. Secondary antibodies (MolecularProbes/Invitrogen, Grand Island, NY, USA) included Alexa Fluor 488 or 555 goatanti-mouse and Alexa Fluor 488 or 568 goat anti-rabbit. After 30 minutes to 1 hour ofincubation while rocking, cells were washed three to five times with blocking or washbuffer and, to the final 20 μl of cells, one drop (approximately15 μl) of VECTASHIELD mounting medium (Vector Laboratories, Burlingame, CA,USA) was added. Tubes were wrapped in aluminum foil and stored at 4°C untiluse.

Imaging of the immunostained cells was done using a DeltaVision RestorationMicroscopy System (Applied Precision), consisting of an inverted Olympus IX70microscope (Olympus America, Center Valley, PA, USA) and a Kodak CH350E camera(Rochester, NY, USA). Prepared cells (7 μl) were placed under a glasscoverslip and imaged at 20°C to 22°C using either a PlanApo 60× or100×/1.40 oil-immersion lens objective and deconvolved and analyzed usingsoftWoRx Pro software.

Colocalization of FLAG-MKS3 and centrin (basal bodies) was analyzed using softWoRxPro software or ImageJ software [31]. Elevencells were analyzed for the colocalization of these two proteins. To examine thestaining patterns and calculate the number of basal bodies with FLAG-MKS3 staining,15 μm × 15 μm grids were chosen from both theventral and dorsal surfaces of each of three cells. Basal bodies within that gridwere counted, and we noted whether they had FLAG-MKS3 staining. A total of 463 basalbodies were analyzed on these three cells.

Scanning electron microscopy

RNAi cultured cells (200 ml) were collected by brief centrifugation at800 × g (Damon IEC Division Clinical Centrifuge), washedtwice in Dryl’s solution and fixed as described by Lieberman et al.[32]. After critical point drying,coverslips were glued onto an aluminum chuck using colloidal graphite cement andallowed to dry in a desiccator overnight. The samples were sputter-coated and storedin a desiccator until imaged using a JSM-6060 scanning electron microscope (JEOL USA,Peabody, MA, USA).

Transmission electron microscopy

RNAi cultured cells (100 ml) were collected by brief centrifugation at800 × g (Damon IEC Division Clinical Centrifuge) andwashed twice in 100 ml of Dryl’s solution, then approximately100 μL of the cell pellet was removed and placed in 1.5-ml Eppendorf tubes.One milliliter of Fixation Solution A (1% gluteraldehyde (Electron MicroscopySciences), 0.05 M sodium cacodylate, pH 7.2) was added, rocked for 30minutes on ice and washed three times for 10 minutes under the same conditions. Cellswere resuspended in postfix Solution B (1% gluteraldehyde (Electron MicroscopySciences), 0.05 M sodium cacodylate buffer, 1% osmium tetroxide, pH 7.2)and again washed and rinsed as described above. Cells were preembedded in 2% agarosegel (Invitrogen/Life Technologies) in 0.05 M sodium cacodylate buffer andallowed to set, then sliced into 1 mm × 1 mm blocks.Blocks were placed in glass vials with 50% ethanol and rocked on a specimen rotatorfor 30 minutes during each of the following washes: ethanol at concentrations of 50%,70% and 90% and two times at 100%, with both of the latter in propylene oxide. Cellswere left overnight on a specimen rotator in 1:1 propylene oxide and Spurr’ssolution (Electron Microscopy Sciences). The next day, samples were placed in freshSpurr’s solution for 6 hours and placed in flat embedding molds with freshSpurr’s solution at 60°C for 48 hours. Sections were cut to 90-nmthickness, placed on copper 200-mesh grids and contrasted on droplets of 2% uranylacetate in 50% ethanol for 6 minutes followed by lead citrate (120 mM sodiumcitrate, 2.66% lead nitrate and 0.65% sodium hydroxide in water) for 4 minutes.Sections were imaged using a JEM-1210 electron microscope (JEOL USA). These studieswere repeated three times.

Glutathione S-transferase pull-down and mass spectrometry analysis

The coiled-coil domain of MKS3 was expressed with a glutathioneS-transferase (GST) tag for use in a GST pull-down assay. The construct wascreated by amplifying positions +2,183 to +2,273 of GSPATG00015939001 using thefollowing forward and reverse primers, respectively:5′-gcgggatccatgaattttgtcgatctc-3′ and5′-gcggaattctgatggattttctccatg-3′. The PCR product was treated with BamHIand EcoRI restriction enzymes (New England Biolabs, Ipswich, MA, USA) and cleanedusing gel purification and the PrepEase Gel Extraction Kit (Affymetrix, Santa Clara,CA, USA), then ligated into a pGEX-2TK plasmid vector (GE Healthcare Life Sciences,Pittsburgh, PA, USA) using the Ligate-IT Rapid Ligation Kit (Affymetrix). ThepGEX-2TK plasmid vector had already been opened using the same restriction enzymes,treated with 1 U of calf intestinal alkaline phosphatase at 37°C for 5 minutesto remove the phosphate groups, followed by heat inactivation with 5 mMNa₂-ethylenediaminetetraacetic acid at 72°C for 20 minutes. TheGST-MKS3 coiled-coil domain and GST were expressed in BL-21 cells and bound toglutathione sepharose beads (GE Healthcare Life Sciences) as described previously[33]. After beads were collected frombacterial cell lysates, they were washed in a 1 M MgCl₂ buffer toremove bacterial proteins from the GST and GST-MKS3 proteins. Protein-bound beadswere stored at 4°C in phosphate-buffered saline for up to 2 weeks.

Stock 51s P. tetraurelia cells were cultured and harvested as describedpreviously [34] for whole-cell extract (WCE).Glutathione sepharose beads (GE Healthcare Life Sciences) were prepared by washingthree times in LAP200 buffer (50 mM HEPES, 200 mM KCl, 1 mM EGTA,1 mM MgCl₂, pH 7.4) buffer with 1% Triton X-100. Washed beads(200 μl) were added to 20 ml of WCE. This precleared WCE was thensplit in half and incubated with 200 μl of glutathione sepharose beadsattached to either GST or GST-MKS3. Beads in the supernatant were allowed to rock onice at 4°C for 1 hour. Control and test beads were recovered and washed threetimes in LAP200 buffer with 1% Triton X-100. Samples were run on a 7% to 14% gradientacrylamide gel and silver-stained, then gel slices were trypsin-digested as describedpreviously [34].

Samples were analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) ona linear ion trap LTQ XL Linear Ion Trap Mass Spectrometer (Thermo Fisher Scientific,Asheville, NC, USA). Half the material was loaded onto a100-μm × 120 mm capillary column packed with MAGIC C18(5-μm particle size, 20-nm pore size; Michrom Bioresources, Auburn, CA, USA) ata flow rate of 500 nl/min. Peptides were separated by a gradient of 5% to 35%CH₃CN/0.1% formic acid for 30 minutes, 40% to 100%CH₃CN/0.1% formic acid for 1 minute and 100% CH₃CN for 10minutes.

Product ion spectra were searched using the SEQUEST search engine on ProteomeDiscoverer 1.4 (Thermo Fisher Scientific) against a curated P. tetraureliadatabase with sequences in forward and reverse orientations. The 13 raw files fromcontrol samples and the 13 raw files from test samples were searched as onecontiguous input file, and a single result file was generated for each. The databasewas indexed to allow for full trypsin enzymatic activity, two missed cleavages andpeptides between the molecular weights of 350 to 5,000 Da. Search parameters setthe mass tolerance at 2 Da for precursor ions and 0.8 Da for fragment ions.The result files were then searched against Scaffold version 4.0.5 software (ProteomeSoftware, Portland, OR, USA). Cross-correlation (xcorr) significance filters wereapplied to limit the false-positive rates to less than 1% in both data sets. Thexcorr values were as follows: (+1): 1.8, (+2): 2.7, (+3): 3.3 and (+4): 3.5. Otherfilters applied were a minimum peptide cutoff of 2 as well as DeltaCN >0.1.Ultimately, the confidence parameters resulted in 0% false discovery rate at theprotein and peptide level for both the control and test results.

Sequence analysis

The sequence for MKS3 in Paramecium (GSPATG00015939001,4e^-57, 23% identity) was found using the human sequence forTMEM67 (Q5HYA8) and the annotated Paramecium genome(ParameciumDB, http://paramecium.cgm.cnrs-gif.fr/). ParameciumMKS3 (TMEM67) codes for 2,906 nucleic acids and 951 amino acids. TheRNAi construct design for Paramecium MKS3 comprises bases from positions+1,101 to +2,019.

Five potential homologues for IFT88 were found using the humanIFT88 (NP_783195) and mouse Tg737 (NP_033402) sequences. Thebest match was GSPATP00038505001 (e^-146, 38% identity), which codes for2,341 nucleic acids and 743 amino acids. The RNAi construct for ParameciumIFT88 spans positions +48 to +2,121. Using a feature of ParameciumDB toidentify potential off-target effects [35],we found that the MKS3 RNAi plasmid will target only GSPATG00015939001,whereas the RNAi plasmid for IFT88 will target all five homologues but noother gene sequences outside this gene family. Included in the Supplemental Materialare tables comparing Paramecium IFT88 and MKS3 (Additional file1: Table S1 and Table S2) with sequences from otherorganisms. To further document the conservation of these proteins inParamecium, amino acid alignments of full length and conserved regions ineach protein are included in Additional file 2: Figure S1and Additional file 3: Figure S2.

FLAG-MKS3p immunostaining and localization

We used a 5′-3xFLAG-tagged (FLAG-MKS3) expression vector to produceFLAG-MKS3p to localize the MKS3 protein. Control paramecia were derived from cellsthat were injected with the empty FLAG vector to confirm that cells were unaffectedby the expressed FLAG peptide. Cells were permeabilized, stained with anti-centrinand anti-FLAG and imaged. In Figure 1, images are stacksto ensure that basal bodies and FLAG-MKS3p staining are visible. Cells expressingFLAG or FLAG-MKS3 showed similar centrin staining patterns across the cellsurface (Figure 1A). The control cells show almost nostaining by anti-FLAG, but the FLAG-MKS3-expressing cells show very clearFLAG staining near the basal bodies and faint staining in the cilia(Figure 1A and arrows in Figure 1C; see also Western blots of the tagged protein from cell membrane andcilia in Additional file 4: Figure S4). Additional file5: Movie S1 demonstrates this pattern of FLAG-MKS3pstaining above the staining of centrin. When scanning through the same cell, startingfrom the surface, the green FLAG staining can be seen prior to the red staining ofthe centrin. Figure 1B also demonstrates the FLAG-MKS3pstaining at the distal side of the centrin staining, that is, above the staining ofthe basal body. The anti-centrin antibody recognizes Tetrahymena centrin 1,which is homologous to Paramecium centrin 2, which is found in the basalbody along the shaft [24]. The transitionzone of the Paramecium cilium has been defined as stretching from the basalbody, near the proximal surface of the epiplasm, to the base of the cilium, where thetriplets of microtubules become doublets and the central pair of microtubule doubletsbegins [25, 26]. Thelocalization of the MKS3 protein is therefore consistent with that of the transitionzone for these cells.

Additional file 5: Movie S1: FLAG-MKS3-injected cell shown in Figure 1in the main text begins outside the cell, with its anterior side on the right.As the movie plays, each frame is one z-section, showing the FLAG stain (green)in the frame before the centrin stain (red). Notice that the FLAG stain appearsbefore (above) the basal bodies, which are positioned slightly below themembrane, and is positioned to the side and posteriorly. (WMV 4 MB)

To quantify the basal body and FLAG-MKS3p staining,15 μm × 15 μm squares on the dorsal and ventralsurfaces of three different cells expressing FLAG-MKS3 were randomly chosen.Of the 463 basal bodies observed, 95.2% ± 2.2% of them hadFLAG-MKS3 staining. These data suggest that where a basal body ispresent, we expect to find MKS3 protein. To quantify the extent of colocalization ofthe centrin and FLAG-MKS3p staining, the images were analyzed using softWoRx softwareto obtain a Pearson’s coefficient (r). Eleven FLAG-MKS3p cells showedan average colocalization score of 0.46 ± 0.11(r ± SD), indicating partial colocalization. FLAG-MKS3pstaining is clearly seen in the oral groove (Figure 1A,yellow arrows), but we were unable to differentiate individual basal bodies in thisregion because of their close packing and the spatial limitations of fluorescencemicroscopy. Therefore, these oral groove basal bodies were not included in ouranalysis.

Ultrastructure

We utilize RNAi by feeding of paramecia because of its ease of use and because thecreation of knockouts by homologous recombination is not possible. RNAi allows us toobserve the cells in a depleted state of a targeted protein quickly and effectivelyin a wild-type background. RNAi is estimated to be 80% effective [36]. It allowed us to leave variable amounts of thetargeted protein in the cells and thereby protect them from lethal effects ofcomplete loss of MKS3. We found that very aggressive RNAi treatment quickly led tocell death. In this way, RNAi had an advantage over gene knockout.

Scanning electron micrographs (SEMs) show that the control cells were covered incilia and display a highly organized cell surface with one cilium protruding fromeach cortical unit (Figures 2A and 2B). The cilia on the control cells appear normal, as shown in therepresentative cell in Figure 2A. TheMKS3-depleted cells displayed very short and sparse cilia(Figures 2C to 2F) and lookdramatically different from the controls. The cilia that are present do not resemblethe control cilia. They have wrinkled surfaces and bulges at the tips(Figures 2E and 2F). Of the 23MKS3-depleted cells observed, 56.5% displayed the “blebby”cilia. This was not observed on any of the control cells. The control cell inFigure 2A is fixed with the metachronal wave intact,but the MKS3-depleted cell shown in Figure 2B isnot, which is far more common. This difference between the two cells is not aconsequence of the reduction of MKS3 mRNA. The disturbance to the cell andcilia surface by reduction of MKS3 mRNA is more evident at highermagnification (Figures 2D and 2F).We used RNAi to reduce mRNA for IFT88 that is known to cause loss of ciliaby failure of intraflagellar transport (IFT), a mechanism that is specific to ciliarydevelopment and maintenance. The IFT88-depleted cells display a normallypatterned cell surface (Figures 2G and 2H), with very few and short cilia present on the surface of the majorityof the cells (Figure 2G, yellow arrows).

MKS3 RNAi resulted in cells that appear, by SEM examination(Figure 2), to have missing or shortened ciliaeverywhere on the cell, except in the oral groove. For more details regarding shortcilia, refer to Additional file 6. The reduction ofMKS3 using RNAi also caused severe distortions in the cell surface, incontrast to the surfaces of cells depleted of IFT88, which show missingcilia but no other disruptions.

Transmission electron microscopy (TEM) was employed to examine the ultrastructure ofbasal bodies to determine whether they were structurally equivalent in control andMKS3-depleted cells. The loss of cilia that we observed was not due totheir inability to properly form basal bodies. Cross-sections of basal bodiesobserved using TEM were measured for both height and width using ImageJ software[31]. No differences were observedbetween control basal bodies, which were 379.6 ± 42.4 nm inlength and 202.9 ± 22.8 nm in width(n = 13), and MKS3-depleted basal bodies, which were367.7 ± 35.5 nm long and 191.8 ± 21.9 nmwide (n = 14). In addition, no obvious differences in basal bodydocking were observed.

Immunofluorescence: basal bodies and cortical units

Immunostaining of the MKS3-depleted cells with anti-centrin revealed a basalbody pattern that differed from that of the control and IFT88-depletedcells. The images shown in Figure 3 are stacks ofZ-sections approximately 10 μm thick, which we used to ensure all basalbodies could be visualized. The representative views shown are of the anterior dorsalsurfaces of the cells. The control cell shows rows of basal bodies that run fromanterior to posterior (Figure 3). The basal body rows atthe midline of the typical MKS3-depleted cell show disorganization andtwisting (Figure 3, white arrows). Distortions of rows canbe seen elsewhere on the dorsal sides of the cells, but are most commonly observed atthe dorsal midline. The control and IFT88-depleted cells maintain straight,organized rows.

In Figure 4, the ridges of the cortical units arehighlighted using anti-2F12 at 60× magnification. Light staining of a basal bodycan be seen at the center of each unit. The dorsal surface of a control cell(Figure 4A) demonstrates the high level of organizationof the cortical units. An area of the control cell has been enlarged (yellow box) tobetter highlight this surface (Figure 4a). The lowerimages in Figures 4a to 4c have beentraced for clarity and are to the right of each image. The contractile vacuole poresare indicated by gray arrows. The two MKS3-depleted cells(Figures 4B and 4C) show twomajor types of differences from the control: an insertion of cortical units thatincorporates a short row into another row of units (a kinety) (Figure 4B, yellow arrows) and clustering of basal bodies that should beorganized in a row (Figure 4C, yellow arrows). TheMKS3-depleted cell with the insertion of an abbreviated kinety has abasal body for almost every cortical unit (Figure 4B),whereas the complete surface disruption, the clustering, shows chaotic organizationof the cortical units, some of which are missing a basal body (Figure 4C). Of the MKS3-depleted cells, 70% show kinetydisruptions. Of those 70%, 90% had clusters of basal bodies, as shown inFigure 4C, and 10% had an insertion of a partial kinetyrow, as shown in Figure 4B. These changes in ridgepatterns were always observed on the dorsal surfaces of the cells, often near themidline and never at the extreme poles of the cell, in over 30 control and 70MKS3-depleted cells.

Each cortical unit has one or two basal bodies with corresponding microtubulerootlets. The transverse microtubule (TM) and the postciliary microtubule (PCMs) areoriented in 5 o’clock and 7 o’clock directions, with the anterior end ofthe cell pointing to 12 o’clock. We examined the orientation of TMs and PCMsusing an anti-α-tubulin antibody and the basal bodies using theTetrahymena anti-centrin antibody. The MKS3 -depletedcells lost most of their cilia, which facilitated the imaging of the basal bodies andthe cortical microtubule cytoskeleton. However, the cilia on control cells obscuredthe image of the cortical microtubules that were visualized with theanti-α-tubulin antibody. Therefore, we used IFT88-depleted cells as acontrol because they lose their cilia but do not lose alignment of basal bodies inorderly rows of cortical units (Figure 3).

Figure 5 shows representatives of bothIFT88-depleted cells (Figure 5A) andMKS3-depleted cells (Figure 5B). Basal bodiesof the control IFT88-depleted cells showed organized rows and microtubulerootlets that maintained their polarity and orientation. In contrast, therepresentative MKS3-depleted cell showed twisting of a basal body row andwith it a new alignment of the TM and PCM rootlets. The organized pattern of theIFT88-depleted cell is enlarged in Figure 5A(yellow box) and traced to show the basal bodies (red) and their microtubule rootlets(black). The same pattern is shown for the MKS3-depleted cell inFigure 5B (yellow box), where the orientation of themicrotubule rootlets, as well as the basal bodies, can clearly be seen. The anglebetween the TM and PCM ribbons that emanate from the basal body was maintained in theMKS3-depleted cells (Figure 5C), but theorientation of the rootlets relative to the anteroposterior axis was changed. Both ofthe images of representative cells shown in Figure 5 areof the dorsal surfaces, and the enlarged areas are from near the dorsal midline. Notethat the microtubule rootlet misalignments coincide with basal body misalignments,but not vice versa. The third rootlet, the striated rootlet (SR), alsocalled the kinetodesmal fiber, was visualized using anti-KDF [37] and the basal bodies withanti-Glu-α-tubulin.

Figure 6A shows a control RNAi-fed cell with basal bodies(red) forming clear, organized rows, or kineties, along the cell surface. Emanatingfrom the left side of each basal body is a striated rootlet (green). These fibersextend toward the anterior pole of the cell and span two or more cortical units[19]. The control cell clearlydemonstrates the anterior orientation of the SRs. In the case of two basal bodyunits, this fiber projects only from the posterior of the basal body pair(Figure 6A, yellow arrows). The large red structures inFigures 6A and 6B are thecontractile vacuoles and are not the subject of this study. In theMKS3-depleted cell within the areas of basal body misalignment, the SRs donot always project anteriorly and often veer in oblique directions (Figure 6B(b) and 6B(b′)). The basal bodiesare no longer maintained in their kinety rows, and, much like the twisted orientationof the PCMs and TMs shown in Figure 5, the SRs are chaoticin their orientations. These data, in conjunction with the TM and PCM data(Figure 5), suggest that these rootlets normallydevelop from the basal body, but the basal body has lost its orientation and does notmaintain its position along the anteroposterior axis of the cell (see also Additionalfile 7: Figure S3).

Mass spectrometry and potential interacting partners

Reduced MKS3leads to abnormal and missing cilia

We expressed FLAG-tagged MKS3 protein to localize it within the Parameciumcell and used feeding RNAi to explore for new functions of this protein.IFT88 served as a control for our approach because reduction ofIFT88 mRNA would inhibit ciliary transport and help to determine whethershort and missing cilia are sufficient to explain the RNAi phenotype for MKS3. BothIFT88- and MKS3-depleted cells showed shortening and loss ofcilia over the entire cell, except in the oral groove. These results forMKS3 depletion are in agreement with those of Dawe et al.[27], who reported that smallinterfering RNA (siRNA) duplexes against MKS3 caused short or missing ciliain inner medullary collecting duct cells (IMCD3). In the same study, siRNA duplexeswere used against IFT88 in IMCD3 cells, leaving more than 90% of the cellswithout a primary cilium. The authors concluded that loss of MKS3 disrupts polarityof centrioles and their migration to the cell surface for ciliary formation[27]. Other studies have found avariety of changes in cilia numbers and length, possibly because of differencesbetween cell types and methods of interfering with MKS3 expression through reductionin amount or mutation [27, 38–41].

We found that aspects other than short and missing cilia differed between theParamecium MKS3 RNAi phenotypes and IFT88 RNAi phenotypes. Forexample, the cilia that remained on the IFT88-depleted cells appeared shortbut not misshapen, whereas those on MKS3-depleted cells were short withbulging membranes, giving them a blebby appearance, especially at the tips.

In other systems, MKS3 (TMEM67) functions as part of the filter or as a gatekeeper inthe transition zone, which is the region between the basal body and the ciliarynecklace [13, 42–44]. Failure of transition zonefunction to control ciliary structure and membrane composition can lead to short andbulbous cilia [45], which is similar to ourobserved blebby cilia on cells depleted of MKS3 by RNAi. Ourimmunofluorescence data of paramecia expressing FLAG-MKS3 suggest thatFLAG-MKS3p is in the transition zone, which in Paramecium has been definedas an area that spans from the epiplasm to the base of the cilium below the ciliarynecklace [25, 26](Figure 1 and Additional file 5: Movie S1). The antibody we used to label basal bodies recognizesTetrahymena centrin 1, which is most homologous to Parameciumcentrin 2 and stains the full length of the Paramecium basal body below thecell surface [24]. The transverse sectionthrough the surface of the cell shows anti-FLAG labeling for FLAG-MKS3 nearthe cell surface and at or above the distal end of the basal body. This location ofMKS3p in P. tetraurelia is also consistent with the observations reported byDawe et al. [27], who were the firstto show the localization of MKS3p at the base of the primary cilium at the transitionzone in IMCD3 and HEK293 cells transfected with N-terminal tagged proteins. Othergroups have reported similar findings [13, 46].

We propose that the depletion of MKS3p from the transition zone accounts for the lossof cilia and blebbing of the membrane of the short remaining cilia by causing afailure of the transition zone to regulate ciliary structure and membranecomposition. Our data also suggest the presence of MKS3p in the distal portion of thecilium (observed by Western blotting) (Additional file 2:Figure S1). The cilia for the Western blot preparations are severed from the cellbody above the ciliary necklace, which means that if MKS3p is in the cilia, theproteins on the blot come from above the transition zone [47].

New phenotypes of MKS3mRNA depletion suggest interaction with basal bodystriated rootlets

The repetitive stereotypical rows of cortical units of P. tetraureliaallowed us to identify subtle deviations due to reduction of MKS3p. RNAi forMKS3 led to basal bodies out of kinety rows on the dorsal surface, mostlyat the midline. The disorganized basal bodies were in patches, mostly in clusters or,less often, in small extra rows and with misshapen morphology of cortical ridges.These phenotypes were not seen in the IFT88-depleted cells, indicating thatthe shortening and loss of cilia are not sufficient to explain these changes inMKS3-depleted paramecia.

Cortical units across the surface of Paramecium contain either one or twobasal bodies (mono- or dikinetids, respectively). In preparation for cell division,basal bodies duplicate and the cell must enlarge and elongate. This first stage ofdivision involves the conversion of all dikinetids to monokinetids [21], with the exception of those in the invariantzones at the extreme anterior and posterior ends of the cells [19, 21]. This conversion of di-to monokinetids is the earliest stage in preparation for cell division, and, oncecomplete, the cell will begin basal body duplication and formation of the fissionfurrow at the midline of the cell [21]. Forboth the conversion of di- to monokinetids and basal body duplication, where a newbasal body is produced anterior to the parental basal body, the anterior basal bodiesor the new basal bodies move away from the parental basal bodies using the SR as aguide, thus maintaining orderly rows [19, 21]. Although basal bodies in all areas outside theinvariant zones must duplicate for cell division, we did not observe a distortion ofthe kinety rows of basal bodies except on the dorsal side and primarily at themidline. Therefore, we propose that the MKS3 RNAi phenotype ofdisorientation of basal bodies and rootlet orientations that we observed is primarilydue to failure of the anterior basal body of two basal body units to migrateappropriately along the SR and maintain a straight kinety row. The conversion fromdi- to monokinetids occurs prior to basal body duplication first on thedorsal surface of the cell, where a large number of randomly distributed dikinetidsexist [20], and at the midline because theanterior movement occurs in advance of basal body duplication.

The posterior basal body in the dikinetid has a cilium and full complement ofrootlets (TM, PCM and SR) projecting in stereotypical orientations. The anteriorbasal body of the pair has no cilium and only the transverse microtubules associatedwith it. In addition to the rootlets, a set of three cytoskeletal nodes link thebasal bodies of a dikinetid to each other and to the SR [20, 21]. While moving, the anterior basalbody remains linked to the SR, which extends for two or more cortical units towardthe anterior and appears to act as a guide that keeps the migrating basal bodiesaligned with the cortical row. Once the anterior basal body has separated from theposterior basal body, it develops a PCM and SR in addition to its preexisting TM[21]. A schematic of this process isshown in Figure 7.

We propose that misguidance in the early movement of the anterior basal body ofdikinetids can account for the observed RNAi phenotype of misalignments, primarilynear the dorsal midline. We also propose that all newly forming basal bodies thatalso use the SR as a guide could require MKS3p for their attachment to the SR. Weexpect that only the errors in anterior basal body movements in the dikinetics arenoticed in MKS3 RNAi-depleted cells because they occur early, before basalbody duplication, and because cells stop growing and do not proceed further withbasal body duplication and cell division. Indeed, these RNAi-treated cells stopgrowing after 24 hours of RNAi feeding (data not shown). This hypothesis ofinteraction of MKS3p with the SR is strongly supported by the results of a GSTpull-down assay. Using the MKS3p coiled-coil domain as bait, we have identified akinetodesmal fiber protein (KdB2:GSPATG00008129001).

MKS3p bait also pulls down two Paramecium centrin binding protein 1 s(PtCenBP1: GSPATG00034434001 and GSPATG00034433001), which have been shown to be themain components of the infraciliary lattice (ICL). The ICL is a contractile corticalcytoskeletal network that is nucleated from the basal body region [48, 49]. An interaction of MKS3pand PtCenBP1 could help to stabilize the basal body within the cortical unit,allowing a cilium to be properly established.

GST pull-down results included Ran-GTPase-activating protein 1(Ran-GAP:GSPATG00009639001). An interaction of RanGAP1 with MKS3p is interesting ifthe ciliary pore functions in a fashion similar to the nuclear pore complex, whichhas been suggested previously [50, 51]. A Ran-GTPase/Ran-guanosine diphosphatase gradient betweenthe ciliary and cellular compartments has been suggested to be involved in theentrance of select proteins such as Kif17 into the cilia in mammalian cells[52]. The interaction of MKS3p withRanGAP1 may prove to be a reflection of MKS3p function in the transition zone andpart of the explanation for loss and deformation of cilia in MKS3 depletion.

It might be suggested that the misalignment phenotypes in MKS3-depletedcells is a result of inappropriate development of the SRs around the basal body. Wedo not favor this explanation, because in areas of disruption the PCM and TM formwith a normal angle between them and KDFs develop. These results suggest that theentire basal body unit with its rootlets appears to be misdirected and not alignedwith the anteroposterior axis of a kinety, as opposed to a dysfunction in SRdevelopment.

We did not identify a second location for MKS3 in our immunofluorescence studies oftagged MKS3p outside the transition zone. Although physical interactions of basalbodies and the SR were identified in structural studies [21], the transient nature of the attachments made it difficultto identify the interacting components. Our findings open up a new opportunity todissect these transient but critical interactions.