GCSH antisense regulation determines breast cancer cells’ viability

Since it is known that cancer cells exhibit a preference for increased glycine consumption, the respective glycine metabolizing enzymes are in focus of many research projects. However, no cancer associated studies are available for the Glycine Cleavage System Protein H (GCSH) to date. Our initial analysis revealed a GCSH-overexpression of the protein-coding transcript variant 1 (Tv1) in breast cancer cells and tissue. Furthermore, a shorter (391 bp) transcript variant (Tv*) was amplified with an increased expression in healthy breast cells and a decreased expression in breast cancer samples. The Tv1/Tv* transcript ratio is 1.0 in healthy cells on average, and between 5–10 in breast cancer cells. Thus, a GCSH-equilibrium at the transcript level is likely conceivable for optimal glycine degradation. A possible regulative role of Tv* was proven by Tv1-Tv*-RNA-binding and overexpression studies which consequently led to serious physiological alterations: decreased metabolic activity, release of the lactate dehydrogenase, increased extracellular acidification, and finally necrosis as a result of impaired plasma membranes. In contrast, Tv1-overexpression led to an additional increase in cellular vitality of the tumor cells, primarily due to the acceleration of the mitochondrial glycine decarboxylation activity. Ultimately, we provide the first evidence of a sensitive GCSH-antisense regulation which determines cancerous cell viability.

391 bp (Fig. 3A). Nucleotide identity between Tv* and Tv1 was calculated to 97%. Obviously, Tv* lacks 30 bp of the N-terminal mitochondrial transit peptide region which makes mitochondrial localization unlikely. However, based on the altered transcriptional expression between non-tumorigenic and cancerous cells, a regulative function for Tv* would be possible. Speculating on an antisense regulating mechanism, we analyzed whether Tv1 and Tv* RNA sequences can bind each other. Using northern blotting analysis, a stable Tv1-Tv*-RNA-binding was observed, and, moreover, revealed that Tv* is present in antisense orientation in the breast cancer cell line MCF-7 (Fig. 3B). Transient Tv* elevation reduces Tv1 transcript and protein content. The assumed antisense regulation mechanism was analyzed by nanoparticle driven Tv*-overexpression in all three cancer cell lines and in both non-tumorigenic control cell lines (see SFig. 1). All further experiments were conducted with the breast cancer cell line MCF-7, presenting the most common breast cancer subtype (positive for estrogen and progesterone receptor expression). For transient Tv*-transfection purposes, silica and magnetic, red fluorescent nanoparticles were chosen. These nanoparticles mediated mitochondrial localization after cellular internalization, visualized by yellowish co-localization with green mitochondrial tracker (Fig. 4A). Stable binding of Tv*-DNA with nanoparticles was verified by DNA-agarose electrophoresis (Fig. 4B), and therefore its suitability for transfection is guaranteed. Hence, breast cancer cells were next transfected with GCSH-Tv*-loaded nanoparticles. A considerable increase of Tv* PCR products was detected by RT-PCR, with a simultan decrease in the amount of Tv1 PCR products (Fig. 4C). This result represents the first indication for an antisense-regulative mechanism. In parallel, GCSH protein expression of Tv*-transfected cells was analyzed by confocal laser-scanning microscopy (Fig. 4D). Mock transfected cells revealed a mitochondria-associated localization of red fluorescent nanoparticles. Also, GCSH protein distribution in mock transfected cells was not altered in comparison to the non-transfected MCF-7 cells. In contrast, Tv*-transfected MCF-7 cells showed an accumulation of red nanoparticles around the nucleus and a lowered GCSH protein content in general.
Stable transfection with Tv* decreases cell viability. Nanoparticle mediated Tv*-overexpression was used as a transient, primary test system, while physiological alterations can only be monitored with stably transfected cells. Thus, the GCSH transcript variant Tv1 as well as Tv* were cloned in frame with the GFP-ORF of the vector pCMV6-AC-GFP (SFig. 2) and transfected into the cells (Fig. 5A). As positive control, the GFP-vector was used, known to cause cytoplasmic green fluorescence in MCF-7 cells. The Tv*-GFP-construct revealed a minor cytoplasmic fluorescence, which indicates the absence of the full-length mitochondrial transit peptide. However, Tv1-GFP overexpressing cells showed the expected mitochondrial distribution (Fig. 5A, right). Notably, Figure 2. GCSH transcript and protein analysis. (A) qRT-PCR analysis using GCSH Tv1 specific primers showed an increase in GCSH Tv1 expression in all three breast cancer cell lines (red) as well as in the nontumorigenic control cell line MCF-12A compared to the normal breast cells of MCF-10A (green). MCF-7 and BT-20 revealed the highest GCSH Tv1 transcript levels. Mean ± SD, n = 5, ***P < 0.001, **P < 0.01, *P < 0.05, significantly different compared to MCF-10A, unpaired t-test. (B) RT-PCR analysis using GCSH Tv1 full length primers confirmed the qRT-PCR results. A 2 nd transcript variant (Tv*) with 391 bp in size was amplified and significantly lower expressed in all cancer cell lines. Tv1/Tv* ratios ranged from 5-10 for the cancer cells. (C) GCSH protein content in all 5 cell lines was visualized by western blotting. Native GCSH protein (19 kDa) content was increased for MCF-7 and BT-20, as analogous to the transcript analysis. However, further signals could be detected at ~58 kDa and ~65 kDa. The densitometrically calculated sum factor revealed an approximately 10-fold higher GCSH protein content in all breast cancer cell lines. (D) Stainfree imaging to guarantee identical loaded protein contents, used for the protein content normalization additionally. (E) Control Western blots of PCNA (proliferating nuclear antigen), ß-actin (intended as housekeeping protein) and AMT (aminomethyltransferase, direct GCSH interacting enzyme). Protein contents were calculated densitometrically and set in relation to MCF-10A protein levels.
mitochondria showed a round, swollen shape in comparison with non-treated MCF-7 cells (Fig. 5A, left). Tv*-overexpression modulated the Tv1/Tv* ratio to a similar extent as transient particle-driven Tv*-expression, that reduced Tv1 and increased Tv* transcript (Fig. 5B, compare with Fig. 4C). In contrast, overexpression of Tv1-GFP did not significantly increase the Tv1 transcript content but induced formation of a third transcript variant (~300 bp). Simultaneous visualization of the endogenous GCSH protein (red) and the GCSH-GFP-constructs (green) confirmed low cytosolic expression of Tv* and high mitochondrial expression of Tv1 (Fig. 5C), too. The physiological impact of Tv1 and Tv*-overexpression was examined in GFP-sorted MCF-7 cells. First, the impact on the general cell viability was checked (Fig. 6A). Whilst overexpression of Tv1 induced a significant increase (+30%), overexpression of Tv* mediated a decrease (−40%) in the metabolic viability in comparison with corresponding control treatments. This observation was verified by lactate dehydrogenase (LDH) release measurements, accounting for a first indication of plasma membrane permeabilization, a measure of necrosis or apoptosis induction (Fig. 6B). Tv1-transfected cells harbored a significantly reduced extracellular LDH activity while Tv* showed an upregulation of up to 400%. Metabolic real-time monitoring of viable MCF-7 breast cancer cells confirmed the rise of the extracellular acidification over time (Fig. 6C). 24-30 h after Tv*-transfection, tumor cells are no longer viable. In general, membrane impairment leads to leakage of intracellular molecules and enzymes like the abundant LDH. MCF-7 cells first lost their normal shape and subsequently died by damaged plasma membrane (Fig. 6D). By immunoblotting it was shown that Tv* reduces the multimerized GCSH bands and increases the monomeric GCSH band at 19 kDa. This result clearly proves that Tv* influences more the GCSH expression mainly by impairing multimerization events. Next, a conclusion can be drawn between the GCSH expression ratio of the multimerized upper bands and the monomeric 19 kDa bands. The lower ratio between those bands is the higher is the impact on apoptosis induction (cleaved caspase 7) and proliferation inhibition (PCNA-proliferating nuclear antigen). Further, immunoblotting showed that Tv1 overexpression, here fused with GFP-protein, also enhanced AMT and lowered the glycine N-methyltransferase protein expression (Fig. 6E). Expression of the mitochondrial serine hydroxymethyltransferase (mSHMT) was not altered at any cells. The cytosolic SHMT (cSHMT) was not expressed although co-expression was predicted for GLDC, GNMT and both SHMT variants (Fig. 6F). In contrast, Tv* overexpression induced caspase-7 cleavage and a slight reduction of PCNA (Fig. 6F). Loading control was achieved by stain free imaging prior and after blotting as well as  detection of ß-actin. Finally, the impact on GCS activity was examined by determination of the glycine-serine ratio (Table 1). Notably, reduction of the GCSH content, achieved either by the transfection of Tv* or siRNAs, resulted in an approximately 20% lowered glycine levels. On the contrary, overexpression of GCSH variant Tv1 elevated the glycine levels up to 25%. The glycine-to-serine ratio revealed a shift in glycine-to-serine equilibrium after Tv* transfection, since the ratio between both amino acids decreased from 2.4 to 1.7 on average.

Discussion
Given the fact that cancer cells are characterized by a high demand for C1 bodies to ensure continuous growth, enzymes of glycine metabolism, such as GLDC, are overexpressed in tumorigenic tissue 6,19 . In addition, a combination of low-GLDC/negative-HIF-1α expression has emerged as early prognostic factor for long-term survival in non-small lung cell cancer 20 . However, GLDC is only one out of four protein components of the GCS, which is ubiquitously essential for one-carbon metabolism 9 . Another protein of this system is the GCS H-protein (GCSH) which forms the core of the entire GCS, since it functionally connects all GCS enzymes via its lipoyl arm 9 . Under in vitro conditions, GCS activity is stimulated by external GCSH supply. Consequently, a ~50-fold stimulation of the glycine-bicarbonate exchange rate, relative to the rate measured in the absence of exogenous H-protein, could be reached 7 . In vivo this phenomenon was also verified in the model plant Arabidopsis. Transgenic GCSH overexpressors showed accelerated glycine turnover, yielding higher biomass production 8 . This fundamental finding raises the question, if higher GCSH protein amounts have an impact on the proliferation of breast cancer cells?
Our first result confirmed that the GCSH protein is overexpressed in breast cancer tissue and breast cancer cells. Unexpectedly this overexpression does not depend on the histological subtype. Both, luminal, hormone receptor positive, and, basal, triple negative breast cancer cells exhibited a 10-fold increased GCSH content (Fig. 2C), which is congruent with the reported higher GCS in various cancer types 6 . However, we here demonstrate for the first time that additional GCSH, which very likely results in higher overall GCS activity, strengthens the viability of the breast cancer cells and therefore forces tumorigenesis (Fig. 6). This finding is further supported by the fact that higher GSCH contents correlate with a poorer long-term and relapse free survival rate (SFig. 3).
Secondly, due to the identification of a second transcript variant (Tv*), which is downregulated in all tested breast cancer cell lines, we hypothesized that Tv* could be important for the regulation of the cellular GCSH content, perhaps via antisense repression (Fig. 2). This hypothesis was verified by RNA-binding studies and Tv*-overexpression. Antisense binding of Tv* to Tv1 was shown by northern blotting (Fig. 3). Physiologically, transfection with Tv* mediated a weak cytosolic expression that causes reduced cell viability and cell membrane impairment of MCF-7 cells. Excess of Tv* reduces the expression of Tv1 and induces necrosis of the breast cancer cells within 24-30 h, which was verified by caspase-7 cleavage, an executioner protein of apoptosis and necrosis (Fig. 6). This result implies that the Tv1/Tv* ratio decisively determines GCSH expression and subsequently the capacity for providing C1-units to several biosynthetic pathways. Furthermore, measurement of the cellular glycine and serine levels implicate that the GCSH content significantly affects the cellular glycine-serine equilibrium ( Table 1). To our surprise Tv* overexpression did not significantly change glycine levels, and the small change is in the opposite direction as expected -one expects increased glycine levels due to decreased glycine degradation. On explanation is that the amount and/or activity of the glycine producing serine hydroxymethyltransferase (SHMT) is increased. The cytoplasmic and mitochondrial SHMTs catalyze the conversion of serine to glycine with the transfer of β-carbon from serine to tetrahydrofolate (THF) to form 5, 10-methylene-THF. SHMTs are directly interacting with proteins of the glycine decarboxylating system and thus with GCSH, too (Fig. 4G). By western blotting we showed that the amount of the mitochondrial SHMT (mSHMT) was not altered and the cytosolic SHMT protein was not expressed. Therefore, we checked several glycine degrading and converting enzymes. On protein level, we found that the glycine N-methyltransferase (GNMT) expression is repressed in Tv1 overexpressors. Thus, one could speculate that glycine is not further converted into N-methylglycine in Tv1 overexpressors and therefore accumulates intracellularly. Tv* cells overexpress GNMT protein, as a result, more glycine is metabolized and thus measured in reduced concentrations by LC-MS. However, the effects of such a shift in the Gly-Ser balance on cell vitality and its consequences must be determined in the future studies.
Nevertheless, Jain et al. 5 proved that glycine metabolism plays an essential role for high cell proliferation rates. In addition, metabolic studies showed that fast-growing cells take up more glycine. On the other hand, non-proliferative active cells even released glycine into the medium. Radiolabeling of the ingested glycine has been shown to be used directly for purine synthesis. And here we are, glycine degradation by the glycine decarboxylation system is the primary C1 unit donor for purine synthesis.
Collectively, our results suggest that the cellular GCSH content is a key factor that determines the metabolic state and viability of cells, including tumorigenesis, and thus is a potent tumor marker for highly proliferative breast cancers. The cellular GCSH content itself is sensitively regulated by antisense binding of a 391 bp GCSH transcript variant Tv* to the coding transcript variant Tv1. We hypothesize that overexpressed Tv* directly binds to the coding transcript variant Tv1 whereby GCSH translation is restrained. No GCSH protein could be imported into the mitochondria whereby overall mitochondrial GCS activity is lowered over time leading to  Table 1. Cellular glycine (Gly) and glycine/serine (Gly/Ser) ratios of MCF-7 cells 24 h after transfection with Tv1, Tv* plasmids or GCSH-siRNAs measured by LC-MS. Gly control was set to 1. Mean ± SD, n = 7-10, * P < 0.5, significantly different compared to control (C), unpaired t-test. plasma membrane impairment and necrosis. Thus, it is likely to argue that the GCSH content could be a key player in tumorigenesis, and, therefore, a potent metabolic tumor marker.

Material and Methods
Cell lines and culture conditions. Experiments were performed as reported in our previous work 21 .
Nanoparticle mediated transfection. Sicastar ® -redF and BNF-Starch-redF (micromod Partikeltechnologie GmbH, Rostock, Germany) are amorphous silica nanoparticles and magnetic bionized nanoferrite nanoparticles, respectively, both 100 nm in size, precoated with amino groups (NH 2 ), and labeled with a red fluorophore (Sicastar: ex. λ = 569 nm, BNF: ex. λ = 552 nm). The cell-internalization of magnetic BNF nanoparticles was stimulated by a conventional permanent magnet. To prevent nanoparticle aggregation, all nanoparticle associated experiments were performed in serum-free medium Dulbecco's modified Eagle's medium with 1% gentamicin 23,24 . 50 µg/ml nanoparticle working solutions were prepared at room temperature, protected from light, and coated with 4.5 µg/ml of a poly-D-lysine solution (Sigma-Aldrich, St. Louis, USA) to ensure DNA binding. Nanoparticles were incubated with 50 ng/ml GCSH (Tv1 or Tv*) cDNA for 1 h. Cells were transfected with labeled nanoparticles for 4 h, washed twice with serum-free medium and cultivated up to 24 h in cell culture medium. Following controls were used: 1. Non-transfection control (w/o DNA, w/o nanoparticles); 2. mock control (nanoparticles, w/o cDNA), and 3. GAPDH-control (GAPDH-cDNA, nanoparticles). Transfection efficiency, nanoparticle internalization, and transient expression of GCSH variants were determined by confocal laser-scanning microscopy and flow cytometry.
Cell viability and cytotoxicity. Cell  Statistical analysis. All experiments were replicated at least three times with individually passaged cells, and data sets were expressed as means ± standard deviations (SD). Statistical significance was determined by the unpaired student's t-test or one-way ANOVA (***P < 0.001; **P < 0.01; *P < 0.05) using the software Graphpad Prism Version 5 (http://www.graphpad.com/scientific-software/prism/).