Cell cultures

All chemicals used were purchased from Sigma (St. Louis, MO, USA) unless otherwise noted. P. tetraurelia, 51-s (sensitive to killer) were maintained as described in Sasner and Van Houten [29].

RNAi plasmids

In all construct designs, homologues in other organisms were used to find sequences in the Paramecium annotated genome using the dedicated database ParameciumDB (http://paramecium.cgm.cnrs-gif.fr/) [30]. Genes with the highest homology were then used to design constructs for RNAi using genomic DNA (Additional file 1: Table S1). All inserts were ligated into the double T7-promoter vector, L4440 (Addgene, Cambridge, MA, USA). Off-target sequences were searched for using the ParameciumDB database (http://paramecium.cgm.cnrs-gif.fr/cgi/alignment/off- target.cgi). We found that both sequences of each paralog pair would be affected by the same RNAi sequence; BBS2 showed one off-target match of 23 nucleotides from a hypothetical protein; otherwise no other gene sequences in the genome would be targeted by our BBS RNAi plasmids.

Reverse transcription-PCR (RT-PCR)

This method was used as a check on the efficacy of the RNAi feeding, according to our procedures in Yano et al., [31]. We consider the data from RT-PCR to be semi-quantitative and certainly not a suitable way to quantify the degree of mRNA reduction by RNAi. These experiments were repeated a minimum of three times. See Additional file 2: Figure S1 for a representative example.

Calmodulin primers were also used in RT-PCR as a check on the methodology. Concentrations of the cDNA used as template were undiluted, and diluted 10 fold, and 100 fold. Calmodulin primers used were 5^′-CTGAAGCTGAACTTCAAG-3^′ (forward) and 5^′-CAGAATGATGGTTTCTAAATGA-3^′ (reverse).

RNAi feeding method

We followed the methods for the BBS RNAi as previously described (http://paramecium.cgm.cnrs-gif.fr/RNAi/index.php) [30]. Cells were fed HT115 bacteria transformed with the control (L4440) or with L4440 containing the RNAi insert of interest. After 2 h of incubation while shaking at 37°C, HT115 bacteria transformed with the L4440 plasmid or plasmid with BBS insert were induced to produce double stranded RNA by adding isopropylthio-β-galactoside (IPTG) (RPI Corp., Mt. Prospect, IL, USA) to a final concentration of 0.125 mg/ml and incubated for an additional 4 h. The induced culture was centrifuged at 3,439 × g for 10 minutes at 4°C (Beckman J2-21 centrifuge, Beckman Coulter, Brea, CA, USA) and the pellet was re-suspended in 100 ml of wheat grass medium. The 51-s Paramecium cells were washed in Dryl’s solution (1 mM Na₂HPO₄, 1 mM NaH₂PO₄, 1.5 mM CaCl₂, 2 mM Na-citrate, pH 6.8) and approximately 50 to 100 paramecia were added to the induced culture. Additional stigmasterol, ampicillin, and IPTG were also added to the final concentrations of 8 μg/mL, 0.1 mg/mL, and 0.125 mg/mL, respectively. Cultures were maintained at 28°C. When required, additional induced bacteria, stigmasterol, ampicillin, and/or IPTG were added at 24 and 48 h after feeding. All experiments were carried out at 72 h of RNAi feeding.

The RNAi treatment of cells expressing FLAG-VGCC1c was somewhat different because large numbers of cells were required to harvest the cilia. Paramecia were fed bacteria for BBS8 RNAi or control RNAi as above. The expression of double stranded RNA was induced in 500 mL LB medium with the same concentration of IPTG for 4 h. The final pellets of bacteria were re-suspended in 1.5 L of wheat grass medium containing the same concentrations of stigmasterol (8 μg/mL), ampicillin (0.1 mg/mL), and IPTG (0.125 mg/mL). About 10,000 cells expressing FLAG-VGCC1c were added to the bacterial cultures with the BBS8 RNAi or control plasmids. For three consecutive days, induced bacteria and additional IPTG of 0.125 mg/mL were added to keep cells in log phase. Cells were harvested at 96 h of RNAi feeding for the ciliary membrane immunoprecipitation (IP).

Fluorescence imaging and ciliary measurements

Cells were imaged using the DeltaVision microscope system and SoftWoRx® Pro software (Applied Precision/GE Healthcare, Issaquah, WA, USA). Images were taken using 20×, 60× or 100× oil emersion objectives on an inverted Olympus IX70 microscope with a Photometrics Coolsnap HQ camera (Photometrics, Tucson, AZ, USA). Lenses used were the UPlanApo 20×/0.80 oil; PlanApo 60×/1.40 oil; PlanApo 100×/1.40 oil. Optical z-sections were 0.5 μm thick. For cilia length measurements, the entire course of a curved cilium in different z-sections was traced using the deconvolved images and softWoRx3.3.6 software in multiple segment mode. Care was taken to match up the segments of the cilia that crossed optical sections. Mann–Whitney U-tests of both the raw and normalized data were used to determine significant differences, with no differences in the outcomes. These experiments were repeated three times.

Scanning electron microscopy

We used scanning electron microscopy to examine 200 mL of cells grown for 72 h in RNAi bacteria. Cells were washed twice in Dryl’s solution using a table top centrifuge to remove debris (Damon/IEC Clinical centrifuge, Needham Hts, MA, USA). Pelleted cells were then treated with 1% osmium tetroxide in 10 mM sodium cacodylate for one minute. Cells were again collected by brief centrifugation and immersed immediately in fresh 2% gluteraldehyde in 10 mM sodium cacodylate buffer. After 10 minutes, cells were centrifuged and rinsed in the same buffer for one hour at room temperature (RT). Cells were then collected by brief centrifugation, placed on 13-mm glass cover slips which had been coated with 0.1% poly-L-lysine (high molecular weight) and rinsed in PBS (137 mM NaCl, 2.7 mM KCl, 10.4 mM sodium phosphate dibasic, 1.7 mM potassium phosphate monobasic, pH 7.4). Cells were allowed to settle for 15 minutes and were then rinsed, stacked, and dried at critical point. Cover slips were glued to an aluminum chuck using graphite cement and allowed to dry. The chuck was then sputter coated and imaged using a JEOL 6060 scanning electron microscope (JEOL USA, Inc., Peabody, MA, USA). These experiments were repeated twice.

Assays of behavior in response to ionic stimuli

All solutions used to test behavior in ionic stimuli contained a base buffer of 1 mM citric acid, 1 mM Ca(OH)₂, and 1 mM Tris base. Salts were added from 100 mM stock solutions prepared to desired concentrations (see below) and pH was adjusted to 7.0 using 100 mM Tris Base. After 72 h of growth in RNAi bacteria, approximately 200 cells were removed from their culture and allowed to acclimate in resting buffer (4 mM KCl in the base buffer above) for 30 minutes. Individual cells were transferred to testing solutions in glass depression slides and timed for length of backward swimming; 10 to 20 cells were tested per solution. The experiments were repeated 3 to 10 times. The following solutions were used with the base buffer above: 30 mM KCl; 8 mM BaCl₂; 25 mM TEA with 10 mM NaCl; and 25 mM TEA with 5 mM MgCl₂. In some cases, backward swimming durations were normalized to the control backward swimming in order to combine data from many BBS-depleted lines. Mann–Whitney U-tests performed on the raw data and normalized data showed no difference in significance outcomes. These experiments were repeated a minimum of three times; we often used the swimming in TEA solutions with Na⁺ or Mg²⁺ as indicators of whether the RNAi for BBS7, 8, or 9 was working.

Deciliation and recovery of motility

Cells were deciliated using trituration in an ethanol solution and observed for recovery of motility. After culturing in 100 mL of RNAi bacteria for 72 h, the cells were collected by centrifugation (Damon/IEC Clinical Centrifuge), washed and re-suspended in Dryl’s solution. Cells were centrifuged again and re-suspended in 4 mM potassium chloride (KCl) buffer as for ionic stimulation. Cells that swam upward in the tube were collected after 5 minutes using a Rainin Pipetman (Mettler Toledo, Columbus, OH, USA) and placed in the same KCl buffer. A sample was removed to confirm that all cells were motile. We rapidly added 100% ethanol to the cells for a final concentration of 5%, sampled cells again to determine motility, and began triturating with a Pasteur pipet to sheer off the cilia. After each course of trituration, the cells were examined to determine how many were no longer motile. These experiments were repeated three times.

Preparing FLAG pPXV plasmid construct for microinjection

We prepared an N-terminal FLAG pPXV plasmid for expression of three FLAG sequences at the N terminus of the BBS proteins, SK1a (GSPATP00031195001) and VGCC1c (GSPATP00017333001) and a C-terminal FLAG pPXV plasmid with three FLAG sequences for PKD2 (GSPATP00005599001). The pPXV plasmid (courtesy of Dr. W. John Haynes, University of Wisconsin, Madison, WI, USA) with the 3× FLAG or 3× FLAG with insert was extracted using the Wizard^TM Plus Mini-Prep (Promega, Madison, WI, USA) and linearized with Not I restriction enzyme (New England BioLabs, Inc., Ipswich, MA, USA). The linearized plasmid was purified, re-suspended at a concentration of 5 to 10 μg/μl in sterile H₂0, and 5 to 9 pl was injected into the macronucleus of approximately 20 wild type cells. Individual cells were placed into depressions containing 500 μL of inoculated culture fluid and allowed to recover and divide at RT for 24 to 48 h in a humidification chamber. From each depression, 5 to 7 cells were removed and placed in 10 mL of inoculated culture fluid. Each depression was maintained as a separate cell line at 15°C and the cells were re-fed by transferring 5 to 7 cells to fresh culture fluid every 4 days. Cell lines were tested for the presence of the plasmid using PCR with extracted genomic DNA as a template.

Immunostaining and deconvolution microscopy image analysis

Collection of 100 mL of cultured cells by centrifugation and washing in Dryl’s solution was followed by permeabilization in PHEM solution (60 mM piperazine ethanesulfonic acid (PIPES), 25 mM hydroxyethyl piperazineethanesulfonic acid (HEPES), 10 mM ethylene glycol tetraacetic acid (EGTA), 2 mM MgCl₂ and 0.1% Triton X-100, pH 6.9) and fixation for 60 minutes in freshly made 4% paraformaldehyde in PHEM. The fixed cells were washed three times with blocking buffer (2 mM sodium phosphate monobasic, 8 mM sodium phosphate dibasic, 150 mM sodium chloride, 1% Tween20, 1% BSA, 10 mM EGTA and 2 mM MgCl₂; pH 7.4) by centrifugation and incubated for 1 h at RT with primary antibodies: monoclonal anti-FLAG M2 (Sigma) and Tetrahymena rabbit anti-centrin-1 (gift from Dr Mark Winey, University of Colorado Boulder, Boulder, CO, USA), or rabbit anti-folate binding protein (FBP) [32]; FBP gene GSPATP00025147001[GENBANK: AAS57871]. The cells were collected and washed three times by light centrifugation with 1 mL PBS-T (2 mM sodium phosphate monobasic, 8 mM sodium phosphate dibasic, 150 mM sodium chloride, 1% Tween20; pH 7.4) per wash and incubated for 1 h at RT with 100 μL PBS containing 1:10,000 dilution of secondary antibodies: Alexa fluor® 568-labeled goat anti-mouse and Alexa fluor® 488 goat anti-rabbit (Molecular Probes/Invitrogen, Carlsbad, CA, USA). Cells were washed five times with PBS-T solution and suspended in Vectashield® mounting medium (Vector Labs, Burlingame, CA, USA) for imaging using the DeltaVision® restoration microscopy system (Applied Precision/GE Healthcare, Issaquah, WA, USA) (see fluorescence microscopy and cilia lengths). These experiments were repeated at least three times.

Whole cell extract (WCE) preparation for immunoprecipitation

The WCE protocol was adapted from previous publications [14, 33]. Cells expressing FLAG-tagged BBS8 or BBS9 or control cells with the pPXV vector were grown in four 1.5 L wheat grass cultures at 22°C. The cells from the cultures were collected once densities were between 8,000 and 12,000 cells per mL. Cells were washed twice in 200 mL HM Buffer (20 mM Maleic Acid, 20 mM Trizma Base, 1 mM EDTA, pH 7.8), once in 200 mL LAP200 Buffer (50 mM HEPES, 200 mM KCl, 1 mM EGTA, 1 mM MgCl₂, pH 7.8) and then in 100 mL LAP200 with protease inhibitors: 1 mM phenylmethylsulfonyl fluoride, 1 μg/mL leupeptin (RPI Corp., Mt. Prospect, IL, USA) and 1 μg/mL pepstatin A (RPI Corp.) in addition to 100 μL protease inhibitor cocktail. Cells were then homogenized and the protein concentration was determined using a Pierce protein assay (Thermo Scientific/Pierce, Rockford, IL, USA). Equal concentrations of test and control protein were solubilized by adding Triton X-100 to a final concentration of 1%. Cell lysates were rocked on ice at 4°C for one hour and insoluble proteins were removed by centrifugation at 31,000 × g (Beckman J2-21, Beckman Coulter, Brea, CA, USA) for 20 minutes and then 100,000 × g (Beckman L8-80 M Ultracentrifuge, Beckman Coulter, Brea, CA, USA) for 1 h, both at 4°C.

Immunoprecipitation with anti-FLAG M2 agarose beads

The protocol for IP of WCE was followed as described previously [14] with some modification: 5 to 6 mL of both control and test WCE were clarified by incubating each lysate with 30 μl of Protein A beads (Amersham/GE Healthcare, Pittsburgh, PA, USA). Anti-FLAG M2 agarose beads (Sigma-Aldrich, St. Louis, MO, USA) were prepared by washing eight times in cold LAP200 buffer containing 1% BSA and 1% TritonX-100 [14]. These prepared beads were added to the clarified sample, incubated on ice while rocking for 2 h, and collected by centrifugation (Damon/IEC Clinical Centrifuge). Beads were washed five times in cold LAP200 buffer with 1% TritonX-100 followed by a final wash in cold LAP200 buffer without TritonX-100. An equal volume (30 to 60 μl) of 2× SDS sample buffer (62.5 mM Tris–HCl, 10% w/v glycerol, 2% SDS, 0.01 mg/mL bromophenol blue, pH 6.8) with 3% β-mercaptoethanol (BME) was added and the sample was boiled for five minutes, and centrifuged at 16,000 × g (Eppendorf centrifuge 5424, Hauppauge, NY, USA) for one minute. The supernatant was then loaded and separated by SDS-PAGE on a 7 to 18% gradient SDS gel. BenchMark^TM prestained protein ladder (Invitrogen/Life Technologies, Carlsbad, CA, USA) was loaded to ascertain the approximate molecular mass of the resolved protein samples. Experiments were repeated twice.

Whole cilia were isolated following Adoutte et al.[34] and a total of 5.3 mg from control or test RNAi-treated cells was used for IP. The whole cilia were re-suspended in membrane buffer (10 mM Tris buffer, 50 mM KCl, 5 mM MgCl₂, 1 mM EGTA, pH 7.4) with 1% Triton X-114, and then agitated for 1 h at 4°C. After centrifugation at 16,000 × g (Eppendorf centrifuge 5424) for 10 minutes at 4°C, the supernatant was clarified as previously described using protein A beads (Amersham/GE Healthcare). The clarified lysate was centrifuged at 16,000 × g for 10 min at 4°C and the supernatant was incubated with 20 to 30 μl of prepared anti-FLAG M2 beads (Sigma-Aldrich) for 1 h at 4°C. Beads were prepared by washing four times in membrane buffer with 1% Triton X114. Beads were collected by brief centrifugation and washed in membrane buffer with 1% Triton X-114 and 0.1% BSA three times and then in membrane buffer three times. Samples were prepared as in the WCE before separation by SDS-PAGE.

Western blots

The proteins separated by SDS-PAGE were transferred to BioTrace^TM nitrocellulose blotting membrane (PALL Life Sciences, Ann Arbor, MI, USA). Blots were incubated in blocking buffer comprising 0.5 g skim milk powder, 200 μl of Telost fish gelatin, and 300 μl of normal goat serum (Vector Labs) dissolved in 10 ml of TBS-T (15 mM Tris, 140 mM NaCl, 0.1% Tween, pH 7.5)] at RT for 1 h with rocking. Blots were probed using the following primary antibodies: 1:2500 rabbit anti-FLAG M2 or 1:2000 mouse anti-tubulin. Secondary antibody was either alkaline phosphatase (AP)- or horseradish peroxidase (HRP)-conjugated goat-anti-mouse or anti-rabbit at 1:10,000 dilution and developed accordingly (all antibodies from Sigma-Aldrich, St. Louis, MO, USA).

Silverstained gels and mass spectrometry analysis

After electrophoresis, the SDS-PAGE gel was stained using directions of the FASTSilver^TM kit (G-Biosciences, St. Louis, MO, USA). From the BBS8 and 9 IP, eight regions of each of the silver stained gels were removed (see Additional file 3: Figure S2). The same regions were removed from the control lanes and sliced into small pieces. The molecular mass ranges covered the masses of the BBS proteins (88 kd to 38 kd). Before the trypsin digest, gel slices were destained with 500 μL of destain solution (30 mM K₃Fe(CN)₆ and 100 mM sodium sulfate, pentahydrate (EMD Chemicals, Billerica, MA, USA)) at RT for 15 minutes with occasional vortexing. Gel slices were each rinsed twice in 1 mL HPLC water for 5 minutes, then were covered in 100 μL 100 mM NH₄CO₃. This was removed after incubating at RT for 10 minutes, and samples were incubated at 37°C for 10 minutes in 500 μL 50% MeCN in 50 mM NH₄CO₃. This was removed, 500 μL was again added and the samples were incubated at 37°C for 30 minutes. Samples were dehydrated using 100 μL 100% MeCN and incubated at RT for 10 minutes. Samples were spun briefly and the solution removed. Gel pellets were allowed to dry completely and were then incubated in 30 to 60 μL of 12.5 ng/μL trypsin (Promega, Madison, WI, USA) in 50 mM NH₄CO₃ at 37°C. After 10 to 15 minutes, samples were given an additional 30 μL of 50 mM NH₄CO₃ if needed and then incubated at 37°C overnight. The following morning, the solution was removed and saved. The gel slices were vortexed in 150 μL of 50% MeCN and 2.5% formic acid and spun for 10 minutes at 16,000 × g (Eppendorf centrifuge 5424). The supernatant was removed and added to the trypsin sample removed earlier for each sample. The gel slices were lastly dehydrated in 100 μL of 100% MeCN for 5 minutes at RT, then centrifuged at 16,000 × g for 5 minutes. Each supernatant was again removed and added to the other collected supernatants for each sample. These collected digested peptides were then dried completely in a SpeedVac, re-suspended in 8 μL of 2.5% MeCN and 2.5% formic acid, and 4 μL was transferred to a 100 μL glass deactivated tube with polymer feet (Agilent Technologies, Santa Clara, CA, USA). Each tube was then capped in a glass autosampler vial.

Tryptic peptides in an autosampler vial were loaded using a Micro-autosampler (ThermoElectron, Waltham, MA, USA) onto a microcapillary column packed with 12 cm of reversed-phase MagicC18 material (5 μm, 200 Å; Michrom Bioresources, Inc., Auburn, CA, USA). After a 15-minute isocratic loading at 2.5% MeCN and 0.5% formic acid, peptides were eluted with a 5 to 35% MeCN (0.1% FA) gradient over 60 minutes. Ten mass spectrometry (MS/MS) scans followed each survey scan for the entire run (75 minutes). Mass spectra were acquired in a LTQ-XL linear ion trap mass spectrometer (Thermo Electron).

The raw MS/MS data were searched against the Paramecium tetraurelia forward (target) and reverse (decoy) proteome databases (http://aiaia.cgm.cnrs-gif.fr/download/fasta/) [35], using the Sequest algorithm with a precursor mass tolerance of 2 Da. A static increase in 71.0 Da of cysteine residues for acrylamide adduction was required and differential modification of 16.0 Da on methionine residues was permitted. The top matches were filtered using a unique delta correlation score (dCn2) of 0.16 and Xcorr (cross-correlation) values of 1.8, 2.4 and 2.8 for single-, double- and triple-charged ions, respectively. At the protein level, only proteins for which two unique peptides were assigned to a given Genoscope annotation identification were retained. The number of protein entries having an identical peptide match is listed for each peptide in the column ‘redu’ for redundancy. The estimated false-discovery rates were calculated based on the number of reverse database assignments after the above filtering was applied. Specifically the false discovery rate equals the number of peptides assigned to a reverse database entry times two (to account for unknown false positives) divided by the number of peptides assigned to the forward database. The estimated false discovery rate of proteins identified by more than two unique peptides for the BBS8 WCE IP sample was 0.36% and for the BBS9 WCE IP sample was < 0.00%. The tryptic peptides were analyzed twice for each BBS IP.

Paramecium BBSgenes

A search of the P. tetraurelia genome identified eight BBS homologs (BBS1, 2, 3, 4, 5, 7, 8 and 9) (Additional file 1: Table S1). BBS3 and BBS5 are each encoded by two paralogs, originating from genome-wide duplications [35]. The BBS3a BBS3b and BBS5a BBS5b pairs share 85% and 86% nucleotide identity respectively, a sequence conservation that allows co-inactivation of the paralog when one gene is silenced by RNAi (see Materials and Methods).

BBS protein interactions in Paramecium

In order to determine the BBS protein interactions in P. tetraurelia, we expressed FLAG-tagged BBS8 and BBS9 genes separately and analyzed the proteins immunoprecipitated from WCEs after separation by SDS gel electrophoresis. Control cells expressed the empty FLAG vector. The presence of FLAG-BBS8 or -BBS9 was confirmed using western blotting (Figure 1) compared to the control cells. We do not know the identities of the bands below 37 kD in the test lanes. The positive control lanes in Figure 1A and B (labeled P) show the reactivity of a small FLAG-fusion protein.

IPs from WCEs of control cells and those transformed with FLAG-BBS8 or -BBS9 expression vectors were separated by SDS gel electrophoresis and were silver stained (Additional file 3: Figure S2). The gel lanes were divided into eight sections that spanned molecular masses from 38 to 88 kD, treated with trypsin, and analyzed by tandem MS/MS (for details, see Materials and Methods). We considered peptides found only in the lanes from the FLAG-BBS IPs and not in the control lanes. Proteins were identified by comparison with the Paramecium annotated genome. We required each protein positively identified to have two or more peptides and for these peptides to be absent from the control (Table 1). Using these methods, in IPs from cells expressing FLAG-BBS9, we found BBS1, 2, 4, 5, 7, and 8 (Table 1a). In IPs from cell expressing FLAG-BBS8, we found BBS1, 2, 5, 7, and 9 (Table 1b). Although mammalian BBS4 has been shown to precipitate with BBS8 and BBS9 [14], we failed to identify BBS4 in our FLAG-BBS8 IP experiment, possibly because the FLAG tag interfered with that specific protein interaction, or the interactions were not maintained under our IP conditions. There is no Paramecium ortholog of BBSome protein BBIP10 [13] and we did not find a candidate for BBIP10 in our IPs. BBS3 is not part of the mammalian BBSome [14], and did not precipitate with BBS8 or BBS9 in our experiments. Peptides for the specific BBS proteins were found in the molecular mass range expected, indicating no anomalous distribution of peptides due to degradation of the proteins in the gels (Additional file 3: Figure S2). The non-BBS proteins that precipitated with BBS9 or 8 are listed in Additional file 4: Table S2 and Additional file 5: Tables S3.

Gene name	Annotated gene accession number	MW (kD)	Peptides identified in FLAG-BBS9 IP (number)		Peptides identified in control IP (number)
			Unique	Total	Total
BBS9	GSPATP00027545001	83.33	11	30	0
BBS7	GSPATP00026091001	77.44	9	16	0
BBS2	GSPATP00000964001	75.48	10	18	0
BBS1	GSPATP00033252001	64.85	11	18	0
BBS8	GSPATP00028481001	58.61	8	26	0
BBS4	GSPATP00005292001	50.07	7	11	0
BBS5	GSPATP00036912001	38.19	5	12	0
Gene name	Annotated gene accession number	MW (kD)	Peptides identified in the FLAG-BBS8 IP (number)		Peptides identified in control IP (number)
			Unique	Total	Total
BBS8	GSPATP00028481001	58.61	4	8	0
BBS9	GSPATP00027545001	83.33	13	32	0
BBS7	GSPATP00026091001	77.44	7	9	0
BBS2	GSPATP00000964001	75.48	7	8	0
BBS1	GSPATP00033252001	64.85	7	19	0
BBS5	GSPATP00036912001	38.19	2	2	0

Ciliary phenotypes of BBS depleted paramecia

P. tetraurelia provides the interesting advantage of monitoring channel activity by observation of swimming behavior. For example, transient backward swimming caused by the reversal of the cilia power stroke correlates with a calcium action potential. Depolarization above threshold initiates a graded Ca²⁺ action potential by opening the voltage-gated Ca²⁺ channels (Ca_v) that are exclusively in the cilia [24, 25, 36]. The resulting increase in intra-ciliary Ca²⁺ changes the power stroke of the cilia, sending the cell backward. A rapidly activated voltage-gated K⁺ conductance (I_Kv) and slower calcium activated K⁺ conductance (I_KCa) return the membrane potential to resting levels. The K⁺ channels responsible for this conductance are also thought to be in the cilia and the Ca²⁺ that activates the calcium-dependent potassium channel (K_Ca) is local (not cellular) Ca²⁺, entering cilia through the ciliary Ca_v channels [26, 37]. The duration of backward swimming is a function of the number and activity of Ca_v channels as well as the repolarizing K⁺ conductance. In addition, other Ca²⁺-dependent channels that conduct Na⁺ or Mg²⁺ can prolong the plateau of the action potential, and increase the duration of backward swimming. Since paramecia survive in a range of buffers, we are able to assess the separate contributions of K⁺, Na⁺ and Mg²⁺ conductances to the backward swimming.

Throughout, we refer to cells fed bacteria transformed with the L4440 vector as the control cells and to cells fed bacteria transformed with a vector containing an RNAi insert as the depleted cells. While we have confirmed by RT-PCR that mRNA has been reduced by RNAi (Additional file 2: Figure S1), we assume that the BBS proteins are likewise depleted but not eliminated. (See Additional file 6: Figure S3 for evidence of loss of the FBP with RNAi).

Paramecium cells were depleted for each of the BBS proteins individually. Overall examination with a dissecting microscope revealed no anomalies in cell shape, growth rate and swimming speed upon depletion of BBS1-5 and BBS8. A slow swimming speed was observed for BBS7- and BBS9-depleted cells, which eventually died after 72 h of RNAi.

To examine the distribution of cilia, we made scanning electron microscope images of the BBS-depleted cell lines. Figure 2 shows a representative selection of cells all of which fall within the normal range of size and shape for cells in an asynchronized culture. The BBS2- depleted cell in Figure 2B is representative of BBS3-, BBS4-, BBS5-, and BBS8-depleted cells, all of which show a distribution of cilia comparable to that of the control cells (Figure 2A). In contrast, the BBS7- and BBS9-depleted cells (Figures 2C,D) differ dramatically from the control and BBS2-depleted cells by showing patches of surface with no cilia and even totally bald cells. Additional file 7: Figure S4 shows the distributions of bald and partially ciliated cells for BBS7-, BBS8-, and BBS9-depleted cells. Figure 3A shows more detail of control normal cilia and surface and the normal surfaces with stubs of cilia, short cilia and bald patches of BBS7- and BBS9-depleted cells (Figure 3B,C).

We examined ciliary length using anti-tubulin antibodies for immunofluorescence imaging (Figure 4). The RNAi feeding had little effect on ciliary length with the exception of BBS7 and BBS9 depletion. The length of cilia in other BBS-depleted cell lines range from 88% to 108% of controls, while the few cilia that remained on the BBS7- and BBS9-depleted cells were 62% and 72% of the control length, respectively.

To be sure that bald patches on BBS7- and BBS9-depleted cells were not just due to more fragile cilia than in the control cells and therefore more easily lost during fixation, we assayed their deciliation capacity by trituration in 5% ethanol (see Materials and Methods). Deciliation required as many trituations of the BBS7- and BBS9-depleted cells as control cells to render the cells non-motile: 10 triturations immobilized 52% of both control and BBS7-depleted cells and 63% of both control and BBS9-depleted cells.

Behavioral phenotypes of BBS-depleted paramecia

In order to assess the sensory function of the cilia on the RNAi-treated cells, we analyzed swimming behavior by measuring the duration of backward swimming in a battery of depolarizing solutions (see Materials and Methods). With the exception of BBS2-depleted cells, which show normal behavior in all the testing solutions, BBS-depleted cells show prolonged backward swimming both in TEA solutions with Na⁺ (Figure 5A) and TEA solutions with Mg²⁺ (Figure 5B) (BBS1, 3, 4, 7–9). These behaviors are consistent with a reduced I_K(Ca) of the cilia.

TEA inhibits the voltage-dependent potassium channels (K_v), allowing us to focus on the remaining K_Ca above. To examine the function of K_v, we employed 30 mM KCl. Only BBS3-and BBS5-depleted lines showed slightly but significantly longer backward swimming in 30 mM KCl (Figure 5C). Given that all the BBS-depleted cells including BBS3 and BBS5 show normal behavior in 8 mM BaCl₂ (Additional file 8: Table S4), we presume that their Ca_v channels are operating normally, and that a reduced I_Kv is causing a prolonged backward motion of BBS3- and BBS5-depleted cells in 30 mM KCl.

It should be noted that while BBS7 and BBS9 depleted cells swim more slowly due to the loss of cilia, their backward swimming durations in solutions other than those with TEA were normal. That is, the short or lost cilia do not create a general long backward swimming phenotype in all depolarizing solutions, but rather the cells show a phenotype in TEA with Mg²⁺ and TEA with Na⁺ that is consistent with a loss of specific K_Ca channels.

Effect of BBS depletion on ciliary membrane proteins

In the behavioral experiments, all the BBS-depleted cells except BBS2 show phenotypes consistent with reduced function of K_Ca, K_v, or both K_Ca and K_v. One possible reason for the failure of ciliary K⁺ channels to function in BBS-depleted cells is that they are no longer located in the cilia. In order to test this possibility, we expressed a FLAG-tagged small conductance potassium channel (SK1a) that is a member of the K_Ca channels. We use this SK1a as a proxy for the two K_Ca channels of the cilia. We also examined cilia for the presence of PKD2 and voltage-dependent calcium channel isoform 1c (VGCC1c), a Ca_v that is exclusively in the cilia. We chose these channels because we know from proteomics studies and epitope-tag expression studies that they are expressed in P. tetraurelia cilia [38]. Figure 6 (and Additional file 9: Figure S5, Additional file 10: Figure S6) show the results of these experiments using cells transformed with vectors for expression of FLAG alone or FLAG-tagged channels combined with separate RNAi treatments for each of three BBS orthologs, BBS7, 8, or 9. We chose BBS8 because its depletion leads to the same apparent loss of K_Ca channel function as BBS7 and BBS9 depletions while having very little effect on cilia number or length. We chose BBS7 and BBS9 because they caused dramatic phenotypes in cilia number and length when depleted. We used immunofluorescence to examine SK1a and PKD2 only on cells that retained cilia as shown in the differential interference contrast (DIC) images (Figure 6). For VGCC1c, we examined western blots of IPs of ciliary membrane because of the low expression of this channel in the cilia (Figure 7).

Before immunostaining of the cells (Figure 6 and Additional file 9: Figure S5 and Additional file 10: Figure S6), we confirmed that the BBS7, 8 or 9 depletion was effective by assaying the cell behavior in solutions of TEA with Mg²⁺ and TEA with Na⁺ (see Additional file 11: Figure S7). The data from three replicate experiments show that the RNAi treatments result in significantly longer backward swimming in these solutions, while the control cells show normal durations of backward swimming.

There are two control conditions for each experiment shown in Figure 6 (and Additional file 9: Figure S5 and Additional file 10: Figure S6) with FLAG-SK1a and PKD2-FLAG: paramecia transformed with the expression vector with FLAG, and fed bacteria with the control RNAi vector (FLAG Control) and paramecia expressing the FLAG-tagged SK1a or PKD2 channels and fed bacteria with the control RNAi vector (Empty RNAi vector Control). The test cells expressing FLAG-tagged SK1a or PKD2 were fed RNAi bacteria to deplete BBS7, 8 or 9 (Figure 6, BBS7, BBS8 and BBS9). In the first column of images, we used monoclonal anti-FLAG antibody to detect FLAG-SK1a (red) and as a contrast (green) an antibody against a folate chemoreceptor that is known to be in the cilia (anti-FBP). The corresponding DIC images are shown to assure that the cells were ciliated. In the second column of fluorescence images, we used the monoclonal anti-FLAG antibody to detect PKD2-FLAG (red) and a polyclonal anti-Centrin1 (green) that is specific for basal bodies so that the surface structure would be apparent. In Figure 6 we show only the merged fluorescence and DIC images; the separate images can be found in Additional file 9: Figure S5 and Additional file 10: Figure S6.

In both sets of experiments for FLAG-SK1a and PKD2-FLAG in Figure 6, the tagged channels are clearly elaborated in the cilia (Empty RNAi Vector Control) and are no longer found in the cilia when BBS7, BBS8 or BBS9 is depleted. Notably, the folate chemoreceptor remains on the cilia, especially at the tips, even though the FLAG-SK1a channel is no longer evident in the cilia with BBS depletion. The DIC images demonstrate that we have chosen cells that retain cilia and matched them for region of the cell and dorsal vs. ventral side. In this way, we demonstrate that the loss of FLAG-SK1a or PKD2-FLAG staining is not due to lack of cilia.

As a measure of the effectiveness of our RNAi technique in reducing protein levels, we repeated the experiment with FLAG-SK1a expression in BBS7-, BBS8- or BBS9-depleted cells, and concurrently used RNAi feeding to reduce the folate chemoreceptor as well. As shown in Additional file 6: Figure S3, we see the same results for loss of SK1a in the cilia of BBS7-, BBS8- or BBS9-depleted cells as in Figure 6 and also greatly diminished folate chemoreceptor detected by the anti-FBP antibody.

The behavioral tests predicted that the Ca_vs, which are exclusively in the cilia, are present and functional in BBS7 , BBS8- and BBS9-depleted cells. In order to confirm this, we expressed the full length FLAG-tagged sequence for one of the Ca_vs (VGCC1c) that we have found in the cilia by MS/MS [38]. This channel is in very low abundance, which made it necessary to use IP of ciliary membrane and immunoblotting rather than immunofluorescence. Figure 7 shows a representative blot of anti-FLAG reactive proteins from ciliary membrane of control cells and those depleted for BBS8. Since we must harvest large amounts of cilia for this experiment, it could not be repeated on BBS7- or BBS9-depleted cells. The FLAG-VGCC1c band of about 272 kD is present in both control and depleted cells. As for the experiments in Figure 6, the efficacy of the RNAi was monitored by the duration of backward swimming in TEA-Na solutions (Additional file 11: Figure S7).