Tubulin tyrosine ligase like 12 links to prostate cancer through tubulin posttranslational modification and chromosome ploidy



Prostate cancer is a common cause of death, and an important goal is to establish the pathways and functions of causative genes. We isolated RNAs that are differentially expressed in macrodissected prostate cancer samples. This study focused on 1 identified gene, TTLL12, which was predicted to modify tubulins, an established target for tumor therapy. TTLL12 is the most poorly characterized member of a recently discovered 14-member family of proteins that catalyze posttranslational modification of tubulins. We show that human TTLL12 is expressed in the proliferating layer of benign prostate. Expression increases during cancer progression to metastasis. It is highly expressed in many metastatic prostate cancer cell lines. It partially colocalizes with vimentin intermediate filaments and cellular structures containing tubulin, including midbodies, centrosomes, intercellular bridges and the mitotic spindle. Downregulation of TTLL12 affects several posttranslational modifications of tubulin (detyrosination and subsequent deglutamylation and polyglutamylation). Overexpression alters chromosomal ploidy. These results raise the possibility that TTLL12 could contribute to tumorigenesis through effects on the cytoskeleton, tubulin modification and chromosome number stability. This study contributes a step toward developing more selective agents targeting microtubules, an already successful target for tumor therapy.

Prostate cancer is a common cause of death in Western males, and there is no effective cure for men with advanced disease.1, 2 Prostate cancer lesions are heterogeneous with juxtaposition of benign glands, preneoplastic (PIN) foci and neoplastic foci of varying severity. Tumors are ranked by the Gleason grade: the sum of the 2 most prevalent grades of neoplastic foci.3 Prostate cancer metastasizes preferentially to bone. An important goal is to establish molecular pathways of prostate cancer initiation and progression and the functional roles of candidate genes and regulatory pathways. There is limited availability of homogeneous material from human prostate cancer specimens. We used a sensitive method to screen human specimens for differences in gene expression based on our previous differential display study of head and neck squamous cell carcinoma.4 We identified a number of differentially expressed RNAs with macrodissected prostate cancer samples. One was selected for this study because it was predicted to code for an uncharacterized enzyme that modifies tubulin, which is a successful target for tumor treatment. Furthermore, the connections between tubulins and oncogenic pathways are still poorly understood.5

The mRNA encodes a member of the TTLL superfamily.6 Humans have 14 predicted TTLL family members [review7]. Tubulin tyrosine ligase (TTL), the founding member of the family, readds tyrosine to α-tubulin that has been terminally detyrosinated,8, 9 producing a “TTL cycle.” The TTL cycle is important for neuronal organization,10, 11 trafficking of intermediate filament proteins12–14 and cell morphology and spindle positioning.15 TTL expression is suppressed during tumor progression,16–18 and the resulting increase in detyrosinated tubulin is associated with increased tumor aggressiveness.17–19 TTL was identified in 1993,20 whereas many other tubulin modifiers have only recently been identified, including a candidate tubulin carboxy peptidase for detyrosination21 and the TTLLs [recent review:7]. TTLL1, 4, 5, 6, 7, 9, 11 and 13 have been shown to polyglutamylate α-tubulin and/or β-tubulin22, 23 and other substrates.24 TTLL2 is predicted to have polyglutamylase activity. TTLL3 and TTLL10 are tubulin glycine ligases.25–27 TTLL12 is the only family member without an assigned enzymatic function.

Tubulins are an important target for tumor therapy, and many compounds are still being developed. They are classified according to their effects on tubulin dynamics (polymerization and depolymerization) and their binding sites. However, their precise mechanisms of action are still poorly understood, and more precise tubulin-targeted molecular therapeutics could be useful [review:28]. Posttranslational modifications are a good target for the development of new therapeutic agents, as has been shown for kinases [review:29]. Posttranslational modifications of tubulins are complex and have been compared to the histone code.30 Surprisingly, little is known about the roles of these modifications in cancer. The very recent discovery and characterization of the TTLL family is a promising field for new discoveries. We present here the characterization of TTLL12. We suggest that TTLL12 deregulation in tumors could contribute to tumorigenesis through its effects on microtubule dynamics and ploidy.

Material and Methods

Reverse Northerns, semiquantitative PCR and prostate RNA samples

The techniques were adapted from our previous studies.4, 31 Tumor and histological samples were obtained with informed consent from patients undergoing diagnosis and treatment at the University Hospital of Innsbruck and the Regional Hospital of Bolzano, respectively. Tumor specimens were characterized by uropathologists (HR, GS) according to the Gleason grading system.3, 32 Benign and cancer RNA sample pairs were isolated from macrodissected frozen prostate specimens. For immunohistochemical expression analysis, sections of paraffin-embedded radical prostatectomy specimens and tissue arrays of local recurrent tumors, lymph node and distant metastases were used.

Expression vectors

The TTLL12 cDNAs (2 Kb) from the 22Rv1 were cloned into pSG5-Puro Fnt.

Tumor specimens, in situ hybridization and immunohistochemistry

Primary tumor specimens were obtained from radical prostatectomy of untreated patients that were diagnosed in a PSA early tumor detection program. In situ hybridization (ISH) and immunohistochemistry (IHC) were performed by standard techniques. Immunoreactivity was scored using a 4-point scaling system: 1, no staining; 2, weak staining; 3, intermediate staining; 4, strong staining.

Cell culture and transfections

Cell lines were obtained from ATCC, and HCT116 wt and p53−/− from B Vogelstein.33 siTTLL12_1: 5′-GAGUUCA UCCCCGAGUUUG-3′; siTTLL12_2: 5′-GGAACGAGCUGU GCUACAA-3′; siTTLL12_3: 5′-AAGGCCAUCUUCUCUUA AA-3′; siTTLL12_4: 5′-ACGCCGACAUCCUCUUCAA-3′; siTTLL12_5: 5′-GUAGCGGUGUCUCCUCUUU-3′.


TTLL12 antibodies were raised against peptides (amino acids 78–96, PAb2089) and mouse (230–247, MAb1B5).


Standard techniques were used, with: anti-TBP (clone TBP 3G3, IGBMC; full name TATA box-binding protein, HGNC:11588), anti-PolyGlu (B3 Sigma), anti-Δ2Tub (CHEMICON and AbCys) and anti-GluT (AbCys).

Cell cycle analysis

A FACScalibur and Cellquest software (Becton Dickinson) were used for acquisition, and Modfit for analysis.

Karyotyping and immunofluorescence

We performed standard metaphase spreads with Giemsa staining. Immunofluorescence techniques are described in detail in the Supporting Information Material and Methods.

Cell proliferation assays

Growth was followed with MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) (Chemicon Int.).34

Statistical tests

The data were analyzed using the Student test (Excel).


Identification of TTLL12 as a differentially expressed transcript in prostate cancer

We identified differentially expressed RNA in macrodissected prostate tumors. Prostate cancer specimens were validated by a pathologist, staged according to Gleason score and macrodissected to isolate tumor and benign areas. Probes prepared from purified RNA were hybridized with CVDD (comprehensive validated differential display) membranes that were spotted with differential display cloned cDNAs.4, 31 We compared the signals generated with tumor and benign RNAs; control clones were used for normalization (Figs. 1a and 1b). Fifty-eight clones gave differential signals (45 overexpressed and 13 underexpressed). Nineteen were selected and analyzed by semiquantitative PCR, using RPLP0 and PCA335 as normalization and positive controls, respectively. One clone (5–17) was selected for further study based on: the differential signal on the CVDD membranes (Fig. 1b, sample 3), its increased expression in 7/10 matched tumor/benign samples analyzed by semiquantitative PCR (Fig. 1c) and its potential enzymatic activity. In silico analysis predicted a gene with 14 exons (KIAA0153, now called TTLL12; Fig. 1d), and a protein with a C-terminal TTLL domain and a weaker homology to SET domains has also been noted in orthologs from worms (WP:CE01535; www.wormbase.org, release WS198) and mosquitoes (Culex quinquefasciatus, NCBI Reference Sequence: XM_001848762.1, 8/2008).

Figure 1.

General scheme for the selection of KIAA0153/TTLL12. (a) Differential gene expression between prostate cancer and benign samples was investigated with CVDD membranes (comprehensive validated differential display clones spotted on membranes) and reverse Northerns.4, 31 The 58 differentially expressed clones were analyzed in silico, and 19 were selected for further analysis by reverse transcription (RT) and semiquantitative PCR (RT-SQPCR). One clone was selected for further evaluation using tumor samples and functional assays. (b) Identification of 5–17 (clone 3, KIAA0153/TTLL12) by reverse Northerns on CVDD membranes. Top panel: membrane with test clones 1–10 and control clones 1–4 was hybridized with cDNA from a pool of 10 prostate tumors. Bottom panel: a duplicate membrane was hybridized with a probe from the benign part of the prostate from the same patients. The arrows point to 5–17 (clone 3, TTLL12) that is highly differentially expressed in tumors versus benign tissues. (c) Semiquantitative RT-PCR (RT-SQPCR). RNAs from prostate tumors (T) and corresponding benign tissues (B) were amplified (20–40 cycles) with primers for TTLL12 and RPLP0 (upper and lower blots) and run on agarose gels (optimum cycles are shown: 30 for TTLL12, 25 for RPLP0). A total of 7/10 patients expressed TTLL12 at a higher level in the tumors compared to the corresponding benign tissues (Patients 2, 3, 5, 6, 8, 9 and 10). (d, e) Schematic representations of (d) the TTLL12 gene that is composed of 14 introns and (e) the TTLL12 protein that harbors a potential SET-like domain in the N-terminal part and a TTL-like (TTLL) domain that contains predicted ATP-binding sites (light bars) in the C-terminal part.

TTLL12 expression in prostate benign glands, cancers and metastases

We cloned the cDNA and developed peptide antibodies (Supporting Information Material and Methods and Supporting Information Fig. 1). The rabbit polyclonal PAb2089 (amino acids 79–96) and the mouse monoclonal MAb1B5 (amino acids 230–249) detected, on Western blots, 1 major expected-size band in protein extracts from various cell lines (arrowhead, Supporting Information Figs. 1bd, and data not shown). The intensity of the band was increased by overexpression (Supporting Information Fig. 1b) and decreased by knockdown with siRNAs (Supporting Information Figs. 1c and 1d). These results show that the predicted protein is expressed, and that the antibodies efficiently and specifically detect endogenous TTLL12.

We studied the expression of TTLL12 in human prostate cancers and metastases. TTLL12 was detected in the basal layer of benign prostate glands, using ISH (Fig. 2a, panels 1 and 2) and IHC (Fig. 2b, panel 1 and higher magnifications not shown; left: H&E-stained sections, right: adjacent IHC sections). There was only low immunoreactivity with some focally increased staining in the secretory epithelial cells. TTL12 RNA and protein were more highly expressed in tumor cells (see ISH and IHC, Fig. 2a, panels 3 and 4, and Fig. 2b; compare adjacent benign and prostate intraepithelial neoplasia (PIN) in b1; representative ISH and IHC pictures are shown). Immunoreactivity was located predominantly in the cytoplasm, with focal nuclear localization in some cases.

Figure 2.

Expression of TTLL12 in prostate specimens. (a) ISH detection of TTLL12 RNA in prostate sections (1–4) with an antisense TTLL12-specific probe. TTLL12 RNA is expressed in the basal layer of prostate glands (1–2) and in tumor cells (3–4). Original magnifications: ×400 or ×200. (b) IHC detection of TTLL12 in local tumors and recurrent tumor metastases. Left panels = haematoxylin–eosin. Right panels = IHC. The TTLL12 antibody (PAb2089) specifically stains the basal layer in benign prostate and PIN (1), hyperplasia and atrophy (2), different stages of carcinoma (3–6), local recurrent hormone refractory cancer (7) and metastases to the lymph node (8) and bone (9). CA, cancer; GS, Gleason score. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

We further analyzed 61 cases of primary tumors obtained from radical prostatectomies from a PSA early tumor screening program. Staining was scored on a 4-point scale (Supporting Information Material and Methods). Staining was similar in benign tissues and hyperplastic and atrophic glands (0.93, 0.91 and 1.0, respectively; Table 1). Staining was higher in PIN glands (average 2.08) and in tumors, without a clear association with the Gleason pattern (1.81 for 21 cases with a Gleason score <7, 1.85 for 40 with a score ≥7). Staining was consistently and significantly heterogenous in tumors, with absent and low staining as well as areas with various degrees of intense staining.

Table 1. Levels of TTLL12 determined by IHC of human prostate tissues, tumors and metastasis
  1. Number of cases and mean TTLL12 staining scores stratified according to histopathology. Immunoreactivity was scored by a pathologist and stratified according to the Gleason pattern (GP) using a 4-point scaling system. 1: no staining; 2: weak staining; 3: intermediate staining; 4: strong staining.

inline image

Metastatic lesions and local recurrent, hormone ablation-refractory tumors were assessed with tissue microarrays. Immunoreactivity was high in tumors that progressed locally after failure of androgen-ablation therapy (mean score 2.3 for 27 tumors) and in lymph node and distant metastasis (2.6 and 2.3, respectively). These results show that TTLL12 expression increases in prostate cancer and metastasis, raising the possibility that TTLL12 has a role in tumor progression.

TTLL12 expression in normal cells and tumor cell lines

We analyzed TTLL12's cellular localization and expression levels in cell lines, using immunocytochemistry (ICC) and Western blotting, respectively. TTLL12 was overexpressed in DU145 prostate cancer cells, detected by ICC with PAb2089 and examined by fluorescence microscopy. TTLL12 was detected mainly in the cytoplasm of TTLL12-transfected cells (Fig. 3a, panel 1), but not detectably at this exposure in surrounding nontransfected cells and in the control transfected cells using the same short exposure (panel 2). Endogenous TTLL12 was detected in the cytoplasm of nontransfected cells upon longer exposure (Fig. 3b and data not shown). TTLL12 was also detected in the nucleus, with the highest intensities relative to the cytoplasm in GM33480 and Hs677.Tg (normal fibroblasts from skin and tongue, respectively). The endogenous signals were decreased by siRNA against TTLL12 (data not shown), confirming the specificity of detection. Western blots with PAb2089 gave a predominant band of the expected size (Fig. 3c and data not shown). TTLL12 expression was quantified by scanning suitably exposed autoradiograms and corrected for loading using TBP (TATA box-binding protein, HGNC:11588) and GAPDH (glyceraldehyde-3-phosphate dehydrogenase, EC, data not shown). The highest levels were found in many of the metastasis-derived cell lines (LNCaP, DuCaP, VCaP, MDA-PCa and DU145), although there was a low level in PC3 (Fig. 3d, graph, dark blue bars). The lowest levels were found in normal cells (HUVEC, endothelial; GM33480, Hs677.Tg, fibroblasts; GM01953C, lymphocytes; HeKa, epithelial keratinocytes, red bars). Intermediate levels were found in prostate tumor-derived lines, establish by passage in mice as xenographs (22Rv1, light bleu bar) or by expression of oncogenes (PRW-1E and CA-HPV10, orange bars), as well as in normal cells established by oncogene expression (PZHPV7 and RWPE-1, orange bars). These subcellular localization and the general trend of expression levels are similar to those in prostate tumors and metastases (Fig. 2).

Figure 3.

Expression analysis of TTLL12. (a) Immunocytochemistry staining (ICS) of overexpressed TTLL12 in prostate cancer cells (DU145) probed with PAb2089. Cells were transfected with a TTLL12 expression vector [1, pSG5 puro-flag-Nter TTLL12 (22Rv1, 94-A)] or the empty vector (2, pSG5 puro-flag-Nter). TTLL12 is stained in red. Nuclei are stained with DAPI (blue). (b) ICS localization of endogenous TTLL12 probed with PAb2089 in prostate cancer cell lines (1–3), PWR-1E (4, immortalized human prostatic epithelial cells, GM 3348 (5, normal skin fibroblasts) and Hs677.Tg (6, normal tongue fibroblasts). (c) WB analysis of endogenous TTLL12 expression in cell lines from prostate cancer metastases (1–4, 6 and 7), from prostate cancer (lane 5) and those established with oncogenes introduced into cells from prostate tumors (lanes 8 and 9) and benign prostate tissue (lanes 10 and11). The Western blots of whole cell extracts were probed for TTLL12 (PAb2089, upper panel), detyrosinated tubulin (detyr-T, GluT antibody from AbCam, middle panel) and TBP (MAb3G3, lower panel). The bands were quantified by scanning suitably exposed films. The graph represents the relative expression levels of TTLL12 and detyrosinated tubulin, adjusted for the internal control TBP, using the illustrated and other Western blots. Values are relative to LNCaP, set to 100 for TTLL12 and 50 for detyr-T. Tum., derived from prostate cancer without introducing oncogenes; Tum*, established with oncogenes from prostate cancer; Normal*, established with oncogenes from benign prostate tissue. The solid and dotted lines indicate the average levels of expression of TTLL12 and detyr-T, respectively, for the 5 metastatic (blue), normal (red) and oncogene-induced (Tum* and Normal, yellow) cell lines. The error bars represent the standard deviation for 2 independent experiments. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

TTLL12 has a TTL-like domain, suggesting that high levels of TTLL12 expression could affect the levels of modification of the C-terminus of α-tubulin. We measured the levels of detyrosinated tubulin using Western blotting (Figs. 3c and 3d). We found that there was an overall trend in the levels of detyrosinated tubulin, with lower levels in many of the metastasis cell lines compared to the normal cells (about half the amount; Fig. 3d, compare the blue and red dotted lines). The oncogene-derived cell lines (see cell lines in yellow), PC3 and 22Rv1 had intermediate levels (see yellow dotted line). These trends suggest that TTLL12 is somehow linked to detyrosination of tubulin.

We chose to focus further studies on the DU145 cell line, because it expresses relatively high levels of TTLL12 compared to the normal lines, and is amenable to study in cell culture. This choice does not seem to have introduced a bias, as inferred from the experiments where we compared DU145 with other cell lines (see below and data not shown).

Cellular localization of TTLL12

TTLL12 was localization by dual-label indirect immunofluorescence of DU145 cells using 2 antibodies (MAb1B5 and PAb2089; red labeling in Fig. 4 and Supporting Information Figs. 2 and 3) and either standard light (Fig. 4) or confocal (Supporting Information Figs. 2 and 3) microscopy. The cytoskeleton was detected with antibodies from other species (green labeling). DNA was labeled with DAPI (blue labeling in the bottom row). The merged images are in the third row. Using optical light microscopy, we detected TTLL12 in intercellular bridges linking daughter cells and in midbodies (Fig. 4, yellow arrows) and apparently also in centrosomes (orange arrows) as well as along mitotic spindles (blue arrows in the panels on the right). Similar localizations were also observed by confocal microscopy, using as markers various forms of tubulin (acetylated tubulin, α tubulin and tyrosinated tubulin, Supporting Information Fig. 2). TTLL12 localized to midbodies (a, d and i) and centrosomes (g; compare with the centrosome specific marker γ-tubulin) and mitotic spindles (e, f).

Figure 4.

Immunofluorescent localization of endogenous TTLL12 in DU145 cells. Dual labeling was used to compare the localization of TTLL12 with β-tubulin (a) and cytoskeleton components (b). The cells were fixed with Cytoskelfix™, revealed with antibodies against: TTLL12 (a, MAb1B5 and PAb2089; b MAb1B5), β-tubulin (a, rabbit polyclonal sc-9104, left panel; mouse monoclonal TUB 2.1, right 3 panels), α-tubulin (b, rat monoclonal Abcam YOL1/34), detyrosinated tubulin (b, rabbit polyclonal, AbCys GluT) and vimentin (b, goat whole antiserum, Sigma V4630). TTLL12 was fluorescently labeled with Molecular Probe Alexa Fluor® 594 Secondary Antibodies (red; anti-mouse or -rabbit) and the other proteins with Fluor® 488 (green; anti-rabbit, -mouse, -goat or -rat). (a) Top row, TTLL12 (red); second row, β-tubulin (green); third row, merge of the top 2 and bottom panels; bottom row, DAPI (blue). The 2 columns on the right are blow-ups of the second column. (b) Left column, α-tubulin; middle column, detyrosinated tubulin; right column, vimentin; top row, cytoskeleton components; second row, TTLL12; third row, merge of the top 2 and bottom panels; bottom row, DAPI. Arrows point, in a, to: centrosomes (red), midbodies (yellow) and mitotic spindle fibers (blue) and, in b, to areas of colocalization. DAPI (blue) labels DNA. Cells were observed under a fluorescent microscope. Representative images are shown. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

TTLL12 was detected in filamentous structures using MAb1B5 (Fig. 4b, second row). It partially colocalized with vimentin filaments (right columns) and appeared to run along some detyrosinated tubulin fibers (middle column) but did not precisely overlap (see yellow arrows). It did not appear to overlap with α-tubulin (left column), except for several small regions. These results were confirmed by confocal microscopy (Supporting Information Figs. 3ac for α-tubulin, detyrosinated tubulin and vimentin, respectively). There was very little overlap between TTLL12 and α-tubulin and detyrosinated tubulin (see yellow arrows), yet there appeared to be partial overlap with vimentin. Thick detyrosinated tubulin-rich fibers are known to partially colocalize with vimentin filaments. The partial colocalization of TTLL12 filaments with vimentin and possibly detyrosinated tubulin is a potential link with tubulin posttranslational modification.

TTLL12 levels affect tubulin modification

We investigated the posttranslational modifications of tubulins using siRNAs to knockdown endogenous TTLL12 levels and modification-specific antibodies for Western blotting of cell extracts (Fig. 5a). TTLL12 levels were decreased with 5 siRNAs that target different parts of the mRNA, 3 in the coding region and 2 in the 3′ UTR. The controls were siControl and siLuc (Dharmacon) and the transfection mix without siRNAs. Titration experiments showed that the minimum amount of siTTLL12s required to decrease TTLL12 levels by more than 80% was 12 nM with HiPerFect. The Western blots shown are 1 representative of 3 complete experiments (several pilot experiments gave similar consistent results). The Western blots were quantitated by scanning suitably exposed autoradiograms and corrected for loading with TBP (Figs. 5be). All 5 siRNAs efficiently decreased TTLL12 levels in DU145 cells (Fig. 5a, compare lanes 3–7 with 1, 2, 8 and 9; data not shown for ICC). In a time course, they decreased TTLL12 levels efficiently (≥75%) for at least 4 days (Fig. 5b, a scan of 1 representative Western blot is shown). Decreasing TTLL12 levels reproducibly increased the amounts of detyrosinated tubulin (Fig. 5c), Δ2-tubulin (Fig. 5d) and polyglutamylated tubulin (Fig. 5e) after 3–4 days (the quantifications in graphs ce are the averages of 3 experiments normalized in each case to the control siRNAs). These results show that there is a link between TTLL12 protein levels and tubulin modifications.

Figure 5.

Effect of downregulation of TTLL12 with siRNAs on tubulin modifications. DU145 was transfected without siRNAs (Con), with control siRNAs (siC1, siLuc; siC2, control, both from Dharmacon) or siTTLL12s (si1-5), and total cell extracts were prepared after 1–4 days. Protein extracts (20 μg) were analyzed by Western blotting, using antibodies against TTLL12 (PAb2089), detyrosinated tubulin (detyrT), Δ2-tubulin (Δ2Tub), polyglutamylated tubulin (PolyGlu Tub) and TBP. (a) The membranes were cut using the molecular markers as a guide and probed with antibodies against TTLL12, detyrT and TBP (top 3 panels), Δ2Tub and TBP (middle 2 panels) and TTLL12, PolyGlu tubulin and TBP (bottom 3 panels). (be) Histograms of the levels, relative to the controls, of TTLL12 after 1–4 days (b), and detyrT (c), Δ2Tub (d) and PolyGlu Tub (e) after 3 and 4 days, determined by scanning suitably exposed films and correction for the TBP internal control. TxT, Taxotere (500 μM for 2 hr). The quantifications (ce) are the averages of 3 experiments normalized in each case to control siRNAs. The error bars for the controls indicate the variations between the 3 experiments for the controls, the others the variations relative to the normalization control set to 1.

Involvement of TTLL12 in cell growth, cell cycle distribution and chromosome number stability

We tested whether changes in TTLL12 levels affect cell growth (Fig. 6), cell cycle distribution (Fig. 7) and DNA/chromosome content (Fig. 8). TTLL12 was downregulated with siTTLL12 (si1, si2 and si1+2) in 5 prostate cancer cell lines, DU145 (our working model) and LNCaP, MDA-PCa, 22Rv1 and PC3, which express a spectrum of TTLL12 levels (see Fig. 3 above). TTLL12 was efficiently downregulated in the 4 cell lines that express detectable levels (LNCaP, MDA-PCa, DU145 and 22Rv1, data not shown). TTLL12 is barely detectable by Western blotting in PC3. To ensure that the transfection of PC3 cells was efficient, we showed that GAPDH is downregulated by about 80% at the protein level by specific siRNA (data not shown). Cell growth was followed with the MTT assay for 5 days (zero time is 5 hr after beginning the transfection, when the cells had attached). The experiment was repeated 3 times, 1 representative experiments is shown. The 3 experiments gave similar results (data not shown). Downregulation of TTLL12 decreased the growth of all of the cell lines that express TTLL12 (Figs. 6a–6d, the combined data for the control siRNAs and for the siTTLL12s are compared. Time points indicated with stars were significantly different with p values below 0.05, calculated by the Student test). Similar results were obtained in 2 other independent experiments. These results show that TTLL12 is required for efficient cell growth in these cells. There was very little effect in PC3, as would be expected from the low level of expression. These results indicate that the effects of the siRNAs do not result from nonspecific mechanisms. Overall, the results suggest that changes in TTLL12 can affect the efficiency of cell division, but that TTLL12 is not essential for proliferation.

Figure 6.

Downregulation of TTLL12 with siRNAs in prostate cancer cell lines inhibits cell growth. Cell lines (a, DU145; b, LNCaP; c, MDA-PCa; d, 22Rv1; e, PC3) were transfected with HiPerFect (Qiagen) and 12 nM control siRNAs (siCon; average of separate experiments with siGFP and siLuc combined with siCtrl from Dharmacon) or siL12 (siTTLL12s; average of experiments with siRNAs 1, 2 and 1+2 mixed) in 96-well plates and grown for 5 days in complete medium. Each siRNA was transfected in sextuplicate and proliferation followed daily using the MTT assay. The experiment was done 3 times, and 1 representative experiment is shown. The error bars represent 1 standard deviation, and the stars p values below 0.05.

Figure 7.

Downregulation of TTLL12 increases the proportion of cells in G2/M. DU145 (a), LNCaP (b) and PC3 (c) cells were transfected with 12 nM of siCon (controls, siLuc and siControl) or siTTLL12s (si1, si2 and si1+2), and after 48 hr the cells were harvested and treated for FACS analysis. DNA was labeled with propidium iodide (PI). The values shown are the averages and the standard deviations for 3 independent experiments for DU145, LNCaP and PC3.

Figure 8.

TTLL12 overexpression affects cell growth and alters the DNA profiles and ploidy of cells. (a) HCT116 p53−/− cells that stably overexpress TTLL12 (L12-A pool), contain the empty vector clone (control, Con-A) or were used to derive the lines (parental HCT116 p53−/−) were seeded, and the MTT assay was used to follow cell growth (96-well plates, 6 wells per cell line) for 5 days. TTLL12 overexpression reproducibly inhibited growth compared to the control clone and the parental cell line (p value < 0.001). The results for the TTLL12 and the control clone pools are typical. (b) FACS profiles of HCT116 p53−/− parental cells and clones (L12-A and control Con-A). c, The haploid number of chromatids (2c, G1 phase DNA content). The arrow points to the additional 8c peak in TTLL12-overexpressing cells. (c) TTLL12 overexpression affects HCT116 p53−/− cell chromosome numbers. Karyotype analysis was performed on control (Con-A) and TTLL12-overexpressing (L12-A) pools of clones. The modal number of mitotic chromosomes (N) is 45 ± 4 in the control cells. In the TTLL12-overexpressing cells, 20% of the cells has 92 ± 7 mitotic chromosomes and 5% has 131 ± 10. (d) FACS profiles of HCT116 p53 wild-type parental cells and clones (sense = L12-pA, vector = Con-pA). 2c, 4c and 8c, diploid, tetraploid and octaploid DNA contents, respectively. The arrow points to the 8c peak in the TTLL12-overexpressing cells. (e) TTLL12 overexpression affects the karyotype of HCT116 p53+/+ cells. Karyotype analysis was performed on cells derived from the control (pA) and TTLL12-overexpressing (L12-pA) clones. The modal number of mitotic chromosome (N) is 46 ± 4 in the control cells. In the TTLL12-overexpressing cells, 5% has 90 ± 7 mitotic chromosomes. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

DU145 and LNCaP cells were analyzed for cell cycle distribution. Cells were transfected and 2 days later fixed, permeabilized, RNAase treated, propidium iodide stained and analyzed with a FACS Calibur instrument (Fig. 7). Downregulation of TTLL12 in DU145 and LNCaP (Figs. 7a and 7b) reproducibly increased the proportion of cells with 2 DNA complements, corresponding to the G2/M phase of the cell cycle. Similar changes were observed with MDA-PCa and 22Rv1 (data not shown). In contrast, with PC3 that expresses barely detectable levels of TTLL12, there was no reproducible difference. These results show that TTLL12 can affect cell cycle distribution but is not essential for passage through the cell cycle.

We investigated the effects of TTLL12 overexpression on changes in chromosome numbers with the diploid cell lines, HCT116 and HCT116 p53−/−. The p53−/− background sensitizes the cells to alterations in chromosome number.36 We selected pools of cells that stably overexpress TTLL12 and control clones with the equivalent empty vector and studied cell growth, cellular DNA content and chromosome numbers (see representative pools in Fig. 8; similar results were observed with other pools and in Hep2, data not shown). Growth assays showed that overexpression of TTLL12 inhibited cell growth in the L12-A HCT116 p53−/− pool relative to control (Con-A) and the parental line (Fig. 8a). FACS analysis showed that stable TTLL12 overexpression in HCT116 p53−/− generated cells with a tetraploid DNA content that was not present in the control and parental cells (Fig. 8b, see arrow). Counts of the number of chromosomes in mitotic spreads (Fig. 8c) showed that essentially all of the cells in the controls had the diploid complement of chromosomes (45 ± 4, Fig. 8c), whereas 20% was tetraploid and 5% hexaploid amongst the TTLL12-overexpressing cells. The changes were studied in the p53+/+ background. Using FACS analysis, we found that overexpression of TTLL12 increased the number of cells with an 8c complement of DNA, compared to the control and parental cells (Fig. 8d), but to a lesser extent than in the p53−/− background (about 50% less). In karyotype analysis, about 5% of the TTLL12-overexpressing cells was tetraploid (Fig. 8e, 90 ± 7), whereas no tetraploid cells were observed in the control cells. These results show that stable TTLL12 overexpression inhibits cell growth and increases the DNA content and the number of chromosomes. The increase in chromosome copy number is a potential mechanism by which TTLL12 overexpression could drive cancer progression to metastasis.


We describe the initial characterization of TTLL12, isolated using an approach4, 31 that: (i) is not restricted to predefined transcripts, (ii) is sensitive and (iii) can be applied to small amounts of RNA. We initially selected CSRP2, RABL2, LPRP and TTLL12. Many of these proteins have since been shown to have interesting features.37–43 TTLL12 and 13 other proteins with TTL domains have been identified in mouse23, 44 and humans. TTLL12, including its TTL- and SET-like domains, is phylogenetically conserved (data not shown), suggesting that TTLL12 has a basic cellular function. TTL is often suppressed in human cancers,16 and tubulin detyrosination is frequent in breast cancer of poor prognosis,18 suggesting that other members of the TTL family could be linked to cancer.

Compared to benign and hyperplastic prostate glands, TTLL12 immunoreactivity increases in metastatic lesions, local recurrent tumors and prostatic intraepithelial neoplasia, which is considered to be a precursor lesion of prostate cancer. Heterogenous expression in primary tumors could be functionally significant, if both increasing and decreasing expression deregulate a finely tuned pathway. Heterogeneity confounds comparisons of IHC with RNA expression levels, because IHC scoring stresses the regions with the highest immunoreactivity, whereas isolation of RNA from microdissected samples yields an average result. The tumor cells with high TTLL12 expression could be crucial for tumor progression, because immunoreactivity was higher and much more uniform in recurrent tumors and metastatic lesions. Deregulated TTLL12 expression may confer a selective advantage during cancer progression. TTLL12 expression is high in many metastasis-derived cell lines, intermediate in cell lines derived from prostate tumors and established with oncogenes and low in normal cells. The overall expression pattern is consistent with the IHC studies in tumors.

TTLL12 is mainly cytoplasmic, as shown by both immunolocalization and fractionation (see above and data not shown). TTLL12 was reproducibly detected in both punctate and fiber-like structures, including intercellular bridges, midbodies, centrosomes and the mitotic spindle. It partially colocalized with vimentin and to a lesser degree with microtubules. Similar colocalizations have not, to our knowledge, been described for other TTL family members. Many of these structures contain tubulins. Microtubules within the constricted portion of intercellular bridges are close together, perhaps accounting for the concentration of TTLL12.45 Intercellular bridges in DU145 contain tubulins.46 These colocalizations link TTLL12 to tubulin and its posttranslational modifications.

TTLL12 knockdown significantly increases the amounts of detyrosinated and polyglutamylated tubulin and Δ2-tubulin. α-tubulin undergoes tyrosination, detyrosination47 and deglutamylation to Δ2-tubulin.48, 49 Detyrosinated and Δ2 tubulins are present in stable, long-lived MTs50, 51 but do not directly cause stabilization [reviews:7, 30, 52]. Polyglutamylation is detected on both tyrosinated and detyrosinated tubulin and is catalyzed by distinct enzymes6, 23, 53 [reviews:7, 30, 52]. The multiple effects of TTLL12 suggest that its effects are indirect, as opposed to having a direct enzymatic activity. TTLL12's TTL domain lacks amino acids predicted to be essential for TTL activity7, 23 and does not appear to have polyglutamylation activity.23 The TTL-like domain of TTLL12 is highly conserved in evolution, suggesting that TTLL12 has an important function. It could have a regulatory role, similar to many other phylogenetically conserved nonfunctional proteins that retain significant homology to enzyme families.54

TTLL12 is apparently not essential for cell division but can affect its efficiency. TTLL12 downregulation decreases proliferation and increases G2/M phase distribution, except in PC3 that expresses barely detectable levels of the protein. TTLL12 could affect microtubule functions in centrosomes, the mitotic spindle and midbodies, as they all contribute to efficient progression through mitosis. TTLL12 overexpression increased the DNA content and number of chromosomes in karyotypically normal HCT116. Increased karyotype instability could account for both TTLL12's effects on cell proliferation and its apparent link to cancer. The importance of aneuploidy in cancer causation is not entirely clear because cellular context can greatly influence its effect on cellular physiology [review:55]. The levels of detyrosinated tubulin are decreased in many cancer cell lines used in our study. This is not necessarily a contradiction of previous work showing decreased TTL suppression in cancer,16–19 as human prostate cancer samples have not been analyzed, and other factors besides TTL are involved in detyrosination of tubulin.

Our results fit into a tentative working model that links TTLL12 to cancer through the regulation of tubulin modifications, the cytoskeleton and chromosome number stability. We suggest that TTLL12 is a previously unrecognized part of the molecular machinery that is involved in tubulin modification. It could be involved in cancer progression and could be a molecular target for cancer therapy.


The authors thank Dr. D. Dembele for help with the statistical analysis, all the members of Procure (FP5) and Prima (FP6) networks, Dr. P. Ramain, Dr. C. Birck, Dr. A. McEwen, Dr. J.P. Samama, Dr. R. Drillien, Dr. J.M. Zahm and the Wasylyk laboratory for useful discussions, support and encouragement; the IGBMC core facilities and Miss Aline Lux and Mrs. Heidi Huebl for perfect technical assistance.