S2 cells deficient in Rheb or TOR are reduced in size

To test the aforementioned hypotheses,Drosophila Schneider line 2 (S2) cells were treated with double-stranded RNA (dsRNA, RNAi) to deplete Rheb, TSC2, or other relevant proteins that function in the insulin and TOR signal-transduction pathways. To confirm dsRNA knockdown of the targeted mRNAs, quantitative real time reverse transcription (RT) PCR or Northern analyses were performed (Additional file 1). Using a Beckman Coulter Counter to measure cell volumes, we noted that S2 cells subjected to 72 hours of dsRheb- or dsTOR-treatment were smaller than control cells (Figure 1A). This is consistent with previously published results from S2 cells and other systems, and is also seen in vivo in Drosophila rheb, TOR, or s6k (a TOR target) gene mutants [1, 2, 30, 31].

General amino acid import is not impaired in cells deficient in Rheb or TOR

To test the hypothesis that Rheb directly controls the cellular import of amino acids, we developed a simple amino-acid import assay. At 72 hours after treatment with dsRNA, cells were transferred to an isosmotic salt solution containing a mix of 15 tritiated amino acids at 22°C. At time-points following this, the cells were washed and lysed, and the amount of ³H in the lysates was determined using a scintillation counter. Time courses of cellular uptake of amino acids revealed that treatment of S2 cells with dsRNA directed against Rheb or TOR consistently caused a slight increase in amino-acid import relative to control cells, which were treated with dsRNA against GFP, when the data were normalized per cell volume (Fig 1B) or per total cellular protein (not shown). Similar assays were performed on cells treated for 30 min with rapamycin, an inhibitor of TOR, or LY294002, an inhibitor of PI3K as well as TOR. These short treatments did not change cell size, and detected no significant deficits in amino-acid import. Longer exposures to rapamycin (5–50 hrs) also had no detectable effect on amino-acid uptake. As a control for the biological relevance of the assay, we performed several import assays on ice at 0°C. This blocked ³H-amino acid uptake nearly completely, and the effect was rapidly reversed by restoring the cells to 22°C (data not shown). Overall, these results indicate that the activity of Rheb and TOR is not an important regulator of bulk amino-acid import. Hence, we suggest that the promotion of general amino-acid import is not likely to be a critical function of either Rheb or TOR in regulating S2 cell growth.

Arginine import is increased, not reduced, in cells deficient in Rheb or TOR

Two recent reports suggest that TOR might control cell growth by specifically regulating the import of basic amino acids via a cationic amino-acid permease encoded by the slimfast (slif) gene [32, 33]. These studies show that slif is required for normal rates of organ growth in flies, that its localization to the plasma membrane of adipose cells in the fat body is controlled by TOR activity, and that it is a limiting regulator of arginine import in S2 cells. The role of TOR in regulating arginine import, however, was not tested directly. To investigate the possibility that TOR regulates cell growth by specifically controlling the import of basic amino acids, we used RNAi to deplete Rheb or TOR and then measured the uptake of ³H-arginine in S2 cells. Experiments performed in triplicate indicated that rates of ³H-Arg uptake were not reduced, but instead showed a small but reproducible increase, in cells depleted of Rheb or TOR (Figure 1C). These data are inconsistent with models in which the growth effects of TOR are mediated via Slif and the import of basic amino acids, but they concur with the observation that arginine import is increased in yeast mutant for Rheb [34]. The basis of this effect is not clear; Rheb/TOR signaling might negatively regulate arginine importers, or the effect may be an indirect, compensatory response to reductions in another TOR-regulated process, such as protein synthesis (see below).

Glucose import is not impaired in cells deficient in Rheb or TOR

Genetic epistasis tests and biochemical evidence place Drosophila Rheb downstream of the insulin receptor (InR), phosphoinositide-3-kinase (PI3K) and the oncogenic protein kinase, Akt (also known as PKB), in the insulin-signaling cascade [1–3, 12, 18, 30]. In mammals, insulin signaling functions to enhance glucose import in skeletal muscle and adipose tissue [24, 35]. Moreover, the small GTPases Rab11 and Rab4, as well as a Rab-GAP, have been implicated in mediating translocation of the glucose importer GLUT4 to the plasma membrane upon insulin stimulation [21, 22]. The influence of Rheb on glucose uptake was therefore tested. S2 cells were treated with dsRNA directed against Rheb for 72 hours, after which time rates of cellular import of tritiated 2-deoxyglucose (2DG) were measured. 2DG is a glucose analog that enters the cell and is phosphorylated to 2-DG-6-phosphate. It is not metabolized further and hence accumulates intracellularly [36]. dsRheb-treated S2 cells had a consistent, significant reduction in cellular 2DG import compared with dsGFP-treated control cells (Figure 1D). Inhibition of TOR with either dsRNA or 16-hour treatment with rapamycin yielded similar results. However, when these 2DG import data were corrected for differences in cell size or protein content, no effects of either loss of TOR or Rheb were apparent (Figures 1E, F). Moreover, a time-course of inhibition of TOR by rapamycin demonstrated that a deficiency in 2DG import was not observed until there was also an observable change in cell size (data not shown). Therefore, the decreased 2DG uptake can be attributed to the reduction in cell size, provided that we assume that the number of glucose importers per cell drops proportionally to the loss of surface area. These findings weigh against models in which Rheb activates TOR by stimulating glucose import. Moreover, they imply that the decreased size of cells lacking Rheb or TOR cannot be attributed to decreased glucose or amino-acid import.

Insulin signaling has only minor effects on glucose import

Insulin signaling is an important regulator of glucose import in some mammalian cell types [24, 35], and in vivo studies suggest that this function is conserved in Drosophila [37, 38]. It has been proposed that insulin signaling might promote TOR activity by increasing glucose import and thereby supporting ATP pools [5, 39]. To assess this possibility in Drosophila, we measured rates of glucose import in S2 cells stimulated with 10 μg/mL bovine insulin after various periods of serum starvation. As shown in Figure 1G, insulin markedly stimulated glucose import only in cells that had been starved of serum (and the insulin within it) for short periods of time (2–6 hours). Insulin stimulation of S2 cells that had not been serum-starved, or that had been serum-starved for 24 hours or more, had no significant effect on rates of glucose import. This indicates that S2 cells are capable of importing glucose independently of their insulin-signaling activity. We also assessed, in serum-stimulated cells, the effects of inhibiting insulin-signaling components with inhibitors or RNAi. Short-term inhibition of insulin signaling was achieved with the PI3K inhibitor wortmannin (see [40, 41]) or an AKT inhibitor (1L-6-Hydroxymethyl-chiro-inositol 2-(R)-2-O-methyl-3-O-octadecylcarbonate; Calbiochem), and long-term inhibition of insulin signaling was achieved using dsRNA directed against PI3K or Akt. None of these treatments had major effects on rates of glucose import (Figure 1G). We also used RNAi to assess the potential role of Glut1, the Drosophila glucose transporter most homologous to mammalian Glut4, which mediates insulin-dependent glucose import in fat and muscle [24, 35]. Although the Glut1 gene is essential for viability in vivo in Drosophila, RNAi-mediated depletion of Glut1 in S2 cells had no detectable effect on glucose import (Figure 1G). All these results indicate that insulin signaling via PI3K and Akt plays only a minor, conditional role in controlling glucose import into S2 cells.

These results are in marked contrast to the strong dose-dependent inhibition of 2DG import seen following a short (30 minute) pre-treatment with the kinase inhibitor LY294002. This drug inhibits PI3K relatively specifically at low concentrations, but inhibits TOR and other kinases at higher concentrations. LY294002 blocked glucose import nearly completely at either 20 μM or 100 μM (Figure 1G). This indicates that glucose import is in fact regulated in S2 cells, but by an LY294002 target other than PI3K or TOR.

To further assess the role of glucose in controlling cell size, we used dsRNA to deplete adenosine monophosphate activated protein kinase (AMPK), a kinase that is activated by AMP when cellular energy levels are low, and which suppresses TOR activity by activating TSC2 [14, 42, 43]. Depletion of AMPK by RNAi caused an increase in cell size (mean volume for AMPKi = 681fL; GFPi = 636fL, p = 0.0011, n = 4; Figure 2A), providing corroborating evidence for a link between cellular energy levels and cell size control that could be mediated by TOR.

Rheb and TOR are required for normal rates of protein synthesis

Having determined that Rheb and TOR do not drive S2 cell growth by promoting nutrient import, we considered the established role for TOR and its downstream targets in the control of protein synthesis. A translation assay was developed to test whether decreased translation could account for the small size of dsRheb- and dsTOR-treated cells. S2 cells were treated with dsRNA for 72 hours, after which the incorporation of radiolabeled amino acids into protein, relative to total protein, was measured over a 3-hour period. To validate this assay, cells were pretreated for 2 hours with the general protein synthesis inhibitor cycloheximide (100 μg/mL). This caused a reduction in the incorporation of amino acids to approximately 7% of control values. Cells treated with dsRNA directed against Rheb or TOR had a significantly reduced proportion of protein labeled compared to dsGFP-treated controls (approximately 60% of controls) (Figure 2A).

In mammalian and Drosophila cells, TOR has been found in two separate complexes; a Rictor-TOR complex, and a Raptor-TOR complex [44–46]. The Raptor-TOR complex appears to be the primary functional complex for serum-dependent or rapamycin-dependent and cell-growth functions of TOR [44]. Accordingly, Raptor was tested for roles in protein synthesis and cell size control. dsRaptor treatment phenocopied dsRheb and dsTOR treatments, in that both cell size and protein synthesis were significantly reduced (Figure 2A). We also found that protein synthesis was markedly reduced, but cell size was unchanged, after a short (20 minute) pre-treatment with 10 ng/mL rapamycin (Figure 2B). Together, these data suggest that a primary effect of Rheb, TOR, and Raptor is to control translation rates.

Overexpressed Rheb drives protein synthesis in vivo

To ascertain the effects of increasing Rheb activity on protein synthesis, we turned to an in vivo assay in Drosophila larvae. The hsFLP/Gal4 system [47] was used to induce high levels of Rheb expression in most cells of third instar, pre-wandering larvae. The larvae were inverted 20 hours after gene induction, and then incubated for 1 hour in Ringer's saline solution containing ³H-amino acids. Protein samples from larval lysates were then prepared and processed as in the S2 cell assay used above. Fed larvae overexpressing Rheb showed an increase in protein synthesis of 140–180% relative to controls (Figure 3A). Previous studies have shown that larvae starved of dietary protein have reduced levels of insulin and TOR signaling activity [33, 48, 49], and that overexpressed Rheb can drive cell growth in such larvae, which are normally growth-arrested [2]. Thus, we tested whether Rheb was also capable of stimulating protein synthesis in protein-starved larvae. Larvae starved of dietary protein for 32 hours prior to heat-shock induction of Rheb had more than twice the translation rate of controls starved in parallel (Figure 3B). These data indicate that Rheb is a potent regulator of protein synthesis in Drosophila, and are consistent with the view that Rheb and TOR are used to modulate rates of protein synthesis in response to changes in the levels of dietary protein.

Insulin signaling also regulates translation in larvae

Given the abundance of data demonstrating a relationship between TOR activity and insulin signaling, we were surprised to find that suppression of PI3K signaling in S2 cells had no significant effect on cell size or translation (Figure 2A). To further investigate the relationship of insulin signaling to TOR, insulin-signaling components were overexpressed in whole larvae using the hsFLP/Gal4 system. Larvae overexpressing Dp110, the catalytic subunit of PI3K, had a significantly enhanced level of translation relative to controls, though the effect was less profound than that of overexpressed Rheb. Consistently, larvae overexpressing PTEN, a negative regulator of PI3K, had lower levels of protein synthesis than controls (Figure 3C). This result corroborates previous findings that suggest a link between insulin signaling and activation of Rheb-TOR signaling in vivo.

S6K and 4EBP cannot account for the translational effects of TOR

The ribosomal protein S6 kinase (S6K) and eukaryotic initiation factor 4E binding protein (4EBP) are both targets of TOR in mammals and flies [50] and are well-known regulators of protein synthesis. Flies and mice mutant for s6k have reduced body size due to a reduction in cell size, although they have a normal number of total body cells [30, 51]. Treatment of S2 cells with dsRNA directed against S6K also reduced cell size and protein synthetic rate, but not as drastically as did dsRheb or dsTOR treatment (Figure 2A). This is consistent with our previous finding that overexpressed Rheb can drive cell growth in s6k mutant animals [2], and indicates that S6K is not the sole mediator of cell size and protein synthesis downstream of TOR.

4EBP is a well-characterized TOR target that functions by sequestering the mRNA cap-binding protein, eukaryotic initiation factor 4E (eIF4E). Phosphorylation of 4EBP by TOR releases eIF4E and allows it to bind the eIF4G scaffold protein, which, in a complex with other components, binds the 5' cap region of mRNAs to initiate translation. Ectopic expression of an active form of Drosophila 4EBP (known as Thor), a negative regulator of translation, results in reduced wing size due to reductions in cell size and cell number [52]. Given these observations, we attempted to rescue the cell size and translation defects observed in dsTOR-treated S2 cells by codepleting 4EBP with dsRNA. Treatment with ds4EBP alone had no detectable effect on cell size or protein synthesis rates. Moreover, in co-depletion experiments, ds4EBP did not lessen the reductions in cell size and translation caused by dsTOR or dsRheb treatment (data not shown). Therefore, negative regulation of 4EBP by Rheb and TOR appears not to be an important mode of growth control in S2 cells. We did find, however, that dsRNA directed against Drosophila eIF4E or eIF4G resulted in small cells with greatly reduced translation (Figure 2A). Considering the similarity to the dsTOR and dsRheb phenotypes, this is consistent with the proposal that TOR is required for proper eIF4G function and translation initiation [53].

Small cell size is a TOR pathway-specific phenotype

Because inhibition of any component of the protein synthesis machinery could be reasoned to have these same phenotypic consequences, two ribosomal proteins were also tested. Treatment of cells with dsRNA targeting ribosomal proteins S3 or S13 did not cause a demonstrable alteration in cell size, despite severely reducing rates of translation (Figure 2A) and cell division (data not shown). Although this failure to affect cell size could be due to differences in the kinetics of target depletion by different types of RNAi, we believe it is meaningful because, whereas the TOR pathway suppression invariably reduces cell size, reducing protein synthesis by other means rarely does so. For instance, heterozygosity for ribosomal gene mutations (Minutes) slows growth dramatically in vivo, yet does not reduce cell size, and sometimes even increases it (our unpublished observations). Moreover, whole-genome screening using RNAi in S2 cells has failed to detect ribosomal protein genes as regulators of cell size [54], whereas TOR pathway components and targets are readily detected [31, 54]. Hence, we surmise that smaller cell size is not a necessary consequence of reduced protein synthesis, and that other targets in addition to S6K and 4EBP mediate the effects of TOR on cell size. A recent report duplicating many of these results corroborates these conclusions [31].

Autophagy

Recent studies have demonstrated that TOR activity suppresses autophagy, a catabolic process whereby starved cells engulf and break down their cytoplasm and organelles, which occurs in flies, yeast, and mammalian cells [9, 55, 56]. We therefore considered the possibility that depletion of TOR in S2 cells may induce autophagy, and that this might account for the reduced cell size seen in dsTOR-treated or dsRheb-treated cells. To test this hypothesis, two proteins required for autophagy, ATG5 and ATG7 [9], were targeted for depletion using RNAi, either alone or in combination with dsTOR or dsRheb treatment. In some cases, we pre-depleted the cells of ATG5 or ATG7 by applying the RNAi against these targets 3days prior to RNAi directed against Rheb or TOR. In all cases, we found that depletion of these ATG mRNAs had no significant effect on cell size, and did not rescue the reduction in cell size caused by depletion of either Rheb or TOR (Figure 2C). These results support the hypothesis that the reduction in cell size is primarily due to the reduction in protein synthesis, and not to an increase in protein breakdown via autophagy.

TOR and Rheb promote ribosome biogenesis

Given that S6K, 4EBP, and autophagy appeared to be insufficient to account for the effects of Rheb and TOR on protein synthesis and cell size, we sought other explanations. Because TOR signaling regulates ribosome biogenesis in both yeast and mammalian cells [7, 8], the effect of loss of either Rheb or TOR on ribosome synthesis in S2 cells was examined. S2 cells treated with dsRheb and dsTOR showed reduced levels of rRNA synthesis as measured by Northern blotting using an internal transcribed spacer probe that hybridizes to the 45S pre-rRNA (Figure 4C). Similarly, rapamycin-treated S2 cells also had reduced levels of pre-rRNA (data not shown). Levels of total (mature) rRNA were also reduced in both dsRheb-treated and dsTOR-treated cells, indicating that these cells have fewer ribosomes (Figure 4C). Consistent with this result, S2 cells treated with rapamycin had smaller nucleoli than controls (Figures 4A, B). In contrast, depletion of either PI3K or Akt had no detectable effect on either pre-rRNA or total rRNA levels in S2 cells (data not shown). This is consistent with our finding that neither dsPI3K nor dsAkt treatment significantly affected either translation nor cell size in our S2 cell assays. Thus, Rheb and TOR regulate ribosome biogenesis in Drosophila S2 cells, and do so independently of PI3K and Akt.

dsRNA synthesis

cDNAs encoding the proteins of interest were obtained and amplified by PCR using composite primers (details available on request), of which the 3' 20 base pairs were gene-specific and the 5' ends had a T7 RNA polymerase binding site. The double-stranded PCR products, which contained a portion of the coding sequence for the protein to be depleted flanked by T7 polymerase binding sites, were purified using a commercial kit (QIAquick Gel Extraction kit; Qiagen, Valencia CA, USA) and then used as template for RNA synthesis using a T7 RNA polymerase (Megascript T7 kit; Ambion, Austin TX, USA). RNA was precipitated by standard procedures using sodium acetate and ethanol for precipitation. To anneal complementary transcribed products, the RNA was heated to 65°C for 30 minutes and slowly cooled to room temperature before being quantified by UV spectrophotometry and diluted to a working concentration of 1 μg/μL. Template DNA for T7-cDNA synthesis was obtained from the DGC clone collection or by reverse transcription (SuperScript II; Invitrogen, Carlsbad, CA, USA) and PCR amplification of an S2 RNA product using sequence-specific primers. In the latter cases, the T7-cDNA was sequenced to verify that the expected product was obtained. The Drosophila genome encodes seven highly related eIF4E genes [82], and the dsRNA we used is predicted to target the four most abundantly expressed isoforms.

dsRNA interference

Interference was performed as described previously [83]. Briefly, S2 cells from an actively growing cell culture were collected by centrifugation and resuspended in serum-free media (SFM, Hy-Q, Hy-Clone, Logan UT, USA) at a density of 1 × 10⁶ cells/mL in one well of a six-well plate, then 20 μg dsRNA was added to the cells. After 1 hour of incubation in the SFM, 2 mL Schneider's Drosophila medium containing 10% fetal bovine serum (Invitrogen/Gibco, Carlsbad, CA, USA) (SDM-FBS) was added to each well. After 72 hours of incubation at 25°C to allow for the turnover of the protein of interest, cells were collected for use in the assays.

Tritiated-nutrient import assays

Cells were pelleted and resuspended to a density of about 1.5 million cells/mL in uptake assay buffer (UAB; 5.4 mM KCl, 140 mM NaCl, 1.8 mM CaCl₂, 0.8 mM MgSO₄, 25 mM Hepes, 25 mM Tris), pH 7.5 [5] containing 10 μg/mL bovine insulin (Sigma), and if used, inhibitors such as LY294002 (Calbiochem, San Diego, CA, USA), AKT inhibitor (1L-6-Hydroxymethyl-chiro-inositol 2-(R)-2-O-methyl-3-O-octadecylcarbonate; Calbiochem) or rapamycin (LC Laboratories, Wobern MA, USA). After 30 minutes equilibration time, 10 μCi/mL tritiated nutrient (15 ³H-amino acid mix; 16–92 Ci/mmol; TRK-440), ³H-arginine (60 Ci/mmol; TRK698), or 2DG (Amersham Biosciences, Piscataway NJ, USA) was added to each cell suspension. The suspension was mixed by inversion. Assays were carried out at room temperature over a period of 45–90 minutes. Duplicate aliquots of each cell suspension were removed at 15–20 minute intervals for the duration of each assay. Aliquots were collected and washed twice with UAB to remove any extracellular tritiated nutrient. The cells were then lysed with cell lysis buffer (100 mM Tris-Cl, pH 8.0, 100 mM NaCl, 0.5% Triton X-100), vortexed, and incubated in ice for at least 10 minutes to ensure complete lysis. Lysates were transferred to scintillation vials containing 3 mL scintillation buffer (EcoScint Original; National Diagnostics, Atlanta GA, USA). The vials were capped and mixed by inversion before the activity count was obtained using a scintillation counter (Beckman LS6500, Beckman-Coulter, Fullerton CA, USA). The activity (counts per minute; cpm) for each time point analyzed was graphically illustrated using MS Excel for visual comparison of nutrient uptake kinetics of the control versus experimental cell populations. If two separate populations were used for control and experimental groups, as was the case with dsRNA-treated cells, the activity was normalized to the number of cells/mL, cell volume/cell/mL, or protein content. Experiments were carried out a minimum of three times.

Determination of cell size

The mean cell size for each culture was determined using a counter (Beckman Coulter Z2) set to count cells of size 5–20 μm. dsRNA-treated cultures were compared to paired control dsGFP-treated cultures.

Determination of protein content

Protein content of a cell culture was determined using the DC microplate protein assay (Biorad). Cell lysis was carried out using the cell lysis buffer as described above.

Translation assay, S2 cells

In total, 2 mL of S2 cells in cultured medium was transferred to one well of a six-well dish, then 1 mL of fresh SDM-FBS containing 15 μCi/mL tritiated amino acid mix (Amersham Biosciences) was added to each well for a final concentration of 5 μCi AA per mL culture. Following 3 hours of amino-acid incorporation at room temperature, 900 μL of suspended cells were split into three aliquots, pelleted, and the media removed. The cells were washed once with 200 μL UAB to remove excess non-imported amino acids. The cells were lysed in 150 μL cell lysis buffer as above then 12 μL resin (Strataclean; Stratagene) was added to each vial of cell lysate. The vials were vortexed to disperse the resin and then left for at least 3 minutes to allow the resin to bind the protein. The resin was pelleted by centrifugation and the remaining lysate was decanted. The resin was washed once with 100 μL UAB before being transferred to scintillation vials containing 3 mL scintillation liquid (EcoScint Original; National Diagnostics). Using a scintillation counter (Beckman LS6500) 10-minute scintillation counts were performed. To validate the assay conditions, a resin-binding test was performed. Radioactivity of resin-bound material (protein), as measured in cpm, was linear over the range of 5–20 mL resin volume. The resin did not bind native amino acids. Therefore, using a constant resin volume normalizes for protein content in this assay. Incorporation of ³H-amino acids into resin-binding material (protein) was also linear over a range of 25–350 minutes incorporation time.

Translation assay, whole larvae

The S2 translation protocol above was modified for use with whole larvae. Experimental larval genotypes were: hs-Flp; Act>CD2>Gal4 UAS-GFP/UAS-Rheb (Figures 3A, B), and hs-Flp; Act>CD2>Gal4 UAS-GFP/UAS-Dp110 or hs-Flp; Act>CD2>Gal4 UAS-GFP/UAS-PTEN (Figure 3C). Control genotypes were: hs-Flp; Act>CD2>Gal4 UAS-GFP in each case. Triplicate sets of 10 larvae were inverted in Ringer's solution and transferred to an eppendorf tube along with 100 μL Ringer's solution for the assay; 750 μL Ringer's solution containing 15 μCi/mL tritiated amino acid mix was added to each eppendorf tube. The larval carcasses were rocked at room temperature for 1 hour, washed in 500 μL cold Ringer's solution, and lysed using mechanical disruption of the larval carcasses in a total of 350 μL cell-lysis buffer. Larval debris was pelleted, and 250 μL lysate was transferred to a new vial. A quantity (15 μL) of resin (Strataclean) resin was added to the lysate. Protein-bound resin was then pelleted, washed with 150 μL Ringer's solution and transferred to 3 mL scintillation buffer, and 1-minute counts were obtained (Beckman LS6500).

Northern blotting

A quantity (1 mL) of dsRNA-treated S2 cell culture (~2 million cells) was used for RNA extraction. Total RNA was extracted using 1 mL of reagent (TRIzol; Invitrogen) using standard procedures, followed by DNase I treatment for 15 minutes at 37°C and purification of the RNA (RNEasy kit; Qiagen). RNA was stored at -70°C until it could be used for Northern blot analysis. Equal volumes of total RNA were loaded into a MOPS/formaldehyde gel for electrophoretic separation. The RNA was blotted onto a positively charged nylon membrane (Roche) according to standard procedures, and the transcript of interest was detected using DIG-labeled antisense probes (DIG labeling and detection, Roche).