A novel magnetic resonance-based method to measure gene expression in living cells.

In unicellular and multicellular eukaryotes, elaborate gene regulatory mechanisms facilitate a broad range of biological processes from cell division to morphological differentiation. In order to fully understand the gene regulatory networks involved in these biological processes, the spatial and temporal patterns of expression of many thousands of genes will need to be determined in real time in living organisms. Currently available techniques are not sufficient to achieve this goal; however, novel methods based on magnetic resonance (MR) imaging may be particularly useful for sensitive detection of gene expression in opaque tissues. This report describes a novel reporter gene system that monitors gene expression dynamically and quantitatively, in yeast cells, by measuring the accumulation of inorganic polyphosphate (polyP) using MR spectroscopy (MRS) or MR spectroscopic imaging (MRI). Because this system is completely non-invasive and does not require exogenous substrates, it is a powerful tool for studying gene expression in multicellular organisms, as well.

INTRODUCTION

In situ hybridization is the most precise, currently available method for obtaining spatial information on gene expression at the subcellular level. However, a major drawback of this method is that it cannot be performed in living cells or in real time: the cells or tissues to be analyzed must be immobilized and exposed to harsh chemicals that are incompatible with living cells. Thus, spatial and temporal information on gene expression cannot be measured simultaneously using in situ hybridization technology. In recent years, several non-invasive methods for detecting gene expression have been developed, which do not have this limitation. These methods use reporter genes, such as luciferase or green fluorescent protein, whose expression stimulates an optically detectable signal (1,2). Although these methods can be used in living cells, the optical signals on which the methods depend only penetrate 1–2 cm of tissue; thus, internal tissues of animal or human subjects cannot be analyzed, because the gene expression signal is obscured by light absorption and/or light scattering before it can reach the signal detector (3). Radionuclide imaging methods such as positron emission tomography (PET) and single-photon emission tomography (SPECT) offer the advantages of a non-invasive method and can measure gene expression in internal tissues (3). These techniques are well-suited for clinical use, but the spatial resolution, which is 1–2 mm, is not sufficient for research applications that demand resolution at or below the single cell level. A second disadvantage of these methods is that they require exposure of the subject to radioactive isotopes, which could be undesirable or disallowed for some applications.

MR spectroscopic imaging (MRI) is a non-invasive method with a spatial resolution at least 10 times greater than radionuclide imaging that can be used to visualize deep tissues (3). Weissleder (4) and Meade (5) were among the first to use MRI-based detection with reporter genes encoding the transferrin receptor and β-galactosidase, respectively, to analyze gene expression in cells treated with specially designed substrates for these gene products. More recently, an MRI-based reporter system was developed using either the heavy subunit of ferritin or the light and heavy subunits of ferritin. This system does not require use of a specially designed substrate, but depends on the fact that ferritin is a metalloprotein that becomes superparamagnetic when it binds endogenous iron (6,7). This method, which was developed and applied in mice, may be more versatile than other non-invasive methods which require exogenous substrate. However, MRI contrast obtained by the ferritin system is limited, since the spin–lattice (7) or spin–spin (6,7) NMR relaxation rate of water proton magnetizations was reported to be typically enhanced only about a factor of two by the presence of the iron–ferritin complex, and thus the complex serves as a negative contrast agent in T2- and T2*-weighted images. The acquired MRI images may not be straight-forwardly interpreted, as the relaxation rates can be affected by other factors and may not be suitable for fully quantitative measure of gene expression.

Polyphosphate (polyP), a linear polymer of orthophosphate residues linked by high-energy phosphoanhydride bonds, is found in all organisms from bacteria to mammals (8). In yeast, most of the cellular polyP (∼99%) is in the vacuole where it presumably serves as both phosphate storage and a cation chelator (8). The amount of polyP in the cell depends on the function of the vacuolar transporter chaperone (Vtc) complex (9) and the vacuolar H+-ATPase (V-ATPase) (10). Strains defective in either complex produce almost no detectable polyP, but still grow normally on acidic YPD media (data not shown) (11). These observations indicate that polyP is dispensable for growth under normal conditions and suggest that the polyP system could be manipulated without deleterious effects in living cells.

This study describes a novel MRS/MRI-based reporter system that quantifies gene expression in living cells by measuring the accumulation of inorganic polyP. This method does not require exogenous substrate and is suitable for quantitative studies of gene expression in internal tissues. In order to characterize this method, initial studies were carried out in Saccharomyces cerevisiae, because yeast accumulate polyP at a higher concentration (∼120 mM) than other eukaryotes (8) and because yeast cells are amenable to genetic manipulation.

MATERIALS AND METHODS

Yeast strains

Standard techniques were used for yeast growth and transformation (12). The yeast strains used in this study are listed in Supplementary Table 1. Oligonucleotide sequences used for the strain construction are listed in Supplementary Table 2. More detailed information is also available on request.

The YTK6337 strain was generated from BY4742 (EUROSCARF) by a fusion PCR-mediated gene replacement method (13) using two pairs of primers, TK3950/4012 and TK4011/3977. The strain was developed as a control strain for studies examining the function of the TATA box binding protein (TBP) associated factor 1 (TAF1). In the present study, the endogenous promoter spanning up to 100 bp upstream of the translational initiation site (A of ATG as +1) of VTC1 of the YTK6337 strain was replaced with one of six different promoters to generate the YTK6352, 6353, 6356, 6359, 6362 and 6365 strains by a fusion PCR-based method. The PCR fragments used for each promoter replacement were amplified by two or three pairs of primers as follows (the amplified marker gene or promoter region is shown in parentheses): TK3755/3756 (His3MX6) + TK3757/3758 (RPS5, −593 to +139 bp, transcriptional initiation site is +1), TK3755/3756+TK3759/3758 (RPS5, −87 to +139 bp), TK3755/3756+TK3760/3762 (RPS5, −450 to −361 bp) + TK3761/3758 (RPS5, −87 to +139 bp), TK3755/3756 + TK3763/3765 (CLN2, −100 to +175 bp, transcriptional initiation site is +1), TK3755/3756+TK3760/3767 (RPS5, −450 to −361 bp)+TK3766/3765 (CLN2, −100 to +175 bp), and TK3755/3756+TK3768/3765 (CLN2, −543 to +175 bp), respectively. Similarly, the YTK6395, 6415 and 6489 strains were generated from BY4742 by a PCR-based method using primer pairs of TK3755/3756+TK4390/4391 (GAL1, −453 to +1 bp, translational initiation site is +1), TK4985/4986 [His3MX6-GAL1 promoter (−541 to +1 bp)-3 × HA epitope tag] and TK4981/3756 (His3MX6) + TK4390/4982 (GAL1, −453 to +1 bp), respectively. The YTK6397 strain was generated from YTK6395 by a PCR-based method using the primer pair TK4650/4651 (3 × HA epitope tag-ADH1 terminator-kanMX6).

The endogenous promoter of VMA2 in BY4741 (EUROSCARF) was replaced with the six promoters described above to generate YTK6526, 6530, 6537, 6533, 6534 and 6527 using the following primer pairs: TK4981/3756 + TK3757/5275 (RPS5, −593 to +139 bp), TK4981/3756 + TK3759/5275 (RPS5, −87 to +139 bp), TK4981/3756 + TK3760/3762 (RPS5, −450 to −361 bp) + TK3761/5275 (RPS5, −87 to +139 bp), TK4981/3756 + TK3763/5276 (CLN2, −100 to +175 bp), TK4981/3756 + TK3760/3767 (RPS5, −450 to −361 bp) + TK3766/5276 (CLN2, −100 to +175 bp) and TK4981/3756 + TK3768/5276 (CLN2, −543 to +175 bp), respectively. The PCR templates used were pFA6a-kanMX6 (14) for TK4011/3977, pFA6a-His3MX6 (14) for TK3755/3756 and TK4981/3756, pFA6a-3HA-kanMX6 (14) for TK4650/4651, pFA6a-His3MX6-PGAL1-3HA (14) for TK4985/4986 and yeast genomic DNA for the other primers.

All gene replacements were confirmed by PCR, Southern blot and genomic DNA sequencing.

Northern blot analyses

Northern blot analyses were performed as described previously (15). To prepare the probes, DNA fragments surrounding the initiating methionine (VMA2 and ADH1) or corresponding to the entire open reading frame (VTC1) were amplified by PCR from yeast genomic DNA, purified and 32P-labeled by random priming.

Immunoblot analyses

Immunoblot analyses were conducted as described previously (16). Polyclonal antibodies directed against TBP were prepared as described (16). Antibodies against the HA epitope tag (F7, mouse monoclonal) were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA).

In vivo MR spectroscopy and MRI

31P-NMR spectra were recorded on a DRX-500 spectrometer (Bruker, Germany) equipped with a standard bore (54 mm) 11.7 T magnet. 1H-MRI and 31P-MRI were recorded on the same spectrometer equipped with a Micro5 imaging probehead.

For MR spectroscopy, samples were transferred into a 5 mm NMR tube and filled with a mixture of 450 μL YPD media and 50 μL 2H2O (CIL, USA) containing 100 μg methylene diphosphonic acid (MDP) (Fulka, Germany), as both a concentration and chemical shift standard. The assignments of the peaks in the 31P-NMR spectrum were given by reference (17). The data were acquired by applying π/2 pulses at 1 s intervals with a sweep width of 16 181 Hz. A total of 16 384 complex points were acquired. For each measurement 512 transients were accumulated. Data acquisition and analysis were performed using XWinNMR software (Bruker). The quantities of inorganic polyP were estimated by the ratio of integrated peak areas of polyP and MDP.

For MRI, samples were transferred into a 0.8 mm outer diameter capillary glass tube (Hilgenberg, Germany). The bottom of the tube was sealed with a sealing compound (Fisher, USA). The tube was then centrifuged at 1000 r.p.m. for 3 min to pellet the yeast cells. The tubes containing test samples and a positional reference (water) were put into an 8 mm NMR tube, as depicted in Figure 3A. A 3D 1H image was acquired to define the positions of samples. Subsequently, the 2D image of polyP in which the polyP signal was integrated longitudinally along the axis of each capillary tube was acquired by the chemical shift selective imaging method (CHESS) (18). For the CHESS, a π/2 pulse to excite ±1500 Hz region centered at the polyP signal with a Gaussian shaped and a non-selective rectangular π pulse were used. The sweep width was 5000 Hz, and TR and TE were 750 and 1.4 ms, respectively. The matrix size was 32 × 16 (zero-filled to 64 × 32 for analysis) with the field of view (FOV) of 1 cm × 1 cm. For each measurement, 256 transients were accumulated. No slice selective gradient was used. The obtained images were analyzed using ParaVision software (Bruker).

RESULTS AND DISCUSSION

Initial studies were carried out to determine whether VTC1 and VMA2, which encode subunits of the Vtc complex and the V-ATPase, respectively, are suitable as quantitative reporter genes with MRS detection in intact cells. This was done by substituting six heterologous promoters of variable strength (19) for the endogenous promoter region of chromosomal VTC1 or VMA2 (Figure 1A). PolyP was then quantified in yeast strains carrying the chromosomal reporter genes with heterologous or endogenous promoters. The results were evaluated to determine if the relationship between the expression of VTC1 and VMA2 and the amount of polyP was quantitative.

Figure 1

Transcriptional activity VTC1 or VMA2 under control of six different promoters. Gene activity was detected by northern blot and in vivo MR spectroscopy (MRS). (A) Schematic diagram of the reporter constructs used to express VTC1 or VMA2. The sequences derived from the RPS5 and CLN2 promoters are shown as hatched and dotted rectangles, respectively. The regions included in each construct (#1–6) are indicated at the right. (B) Total RNA was extracted from cells growing logarithmically at 30°C in YPD and analyzed by northern blot using a probe for VTC1 mRNA (upper panel). Transcription of VTC1 was driven by the endogenous promoter (WT, lane 1) or heterologous promoters #1–6 (lanes 2–7). The Δvtc1 strain was used as a negative control (lane 8) to show that the gene expression signals are specific to VTC1. The blot was probed for ADH1 mRNA as a control for equal loading (lower panel). (C) Transcription of VMA2 and ADH1 in the indicated strains were measured by northern blot as described for panel (B). (D) 31P-MRS analysis of VTC1 gene expression was performed using the same strains as in panel (B). Strain number is indicated to the left of each MRS scan. Peaks corresponding to MDP, inorganic phosphate (Pi) and inorganic polyP are marked at the top. MDP was added at a known concentration as a chemical shift standard. (E) 31P-MRS analysis of VMA2 expression was performed as described in (D) using the same strains as in panel (C). Strain number is indicated to the left of each MRS scan.

Yeast cells were cultured to mid-log phase in liquid YPD media and divided into two aliquots. One sample was analyzed by northern blot (Figure 1B and C) and the other was analyzed by 31P-MRS (Figure 1D and E). The northern blot shows very similar expression profiles for VTC1, VMA2 (Figure 1B and C) and the mini-CLN2 reporter gene characterized previously (19), suggesting that no post-transcriptional regulatory mechanisms influence the steady-state level and/or stability of VTC1 or VMA2 mRNA. Furthermore, the amount of mRNA correlates well with the amount of polyP in the vacuole in all strains with heterologous promoters (Figure 3C #1–6 and D). In contrast, the wild-type (WT) strains accumulate more polyP than expected based on the level VTC1/VMA2 mRNA (Figure 3C). Although the mechanism for this enhanced accumulation is unknown, it is possible that the endogenous VTC1/VMA2 promoters are regulated by an uncharacterized feedback mechanism that leads to increased production of polyP.

In order to provide an accurate measure of gene expression in vivo, a reporter system must be temporally responsive; i.e. it must be able to respond quickly to rapid changes in gene expression. This aspect of the polyP system was tested using the inducible GAL1 promoter to drive gene expression. When the endogenous VTC1 or VMA2 promoters are replaced with the GAL1 promoter, gene expression can be rapidly turned on or off by changing the carbon source in the medium from raffinose to galactose or from galactose to glucose, respectively (20). Kinetic analysis (on-rate kinetics) showed that polyP begins to accumulate (Figure 2A) by 1–2 h after cells are exposed to galactose and induction of VTC1 mRNA begins (Figure 2C, left). In contrast, the level of polyP does not decrease for at least 24 h after mRNA synthesis is suppressed by changing carbon source (Figure 2B, right); the concentration of polyP remains high, even though VTC1 mRNA returns to a low basal level within 1 h (Figure 2C, right) and VTC1 protein is also suppressed within 2 h (Figure 2D). One explanation for this result is that may be VTC1 plays a role in initiating vacuolar fusion (21) but does not play a direct role in synthesis of polyP; if this is true, then induction of VTC1 would stimulate accumulation of polyP in the vacuole, but subsequent suppression of VTC1 might not cause a decrease in the level of polyP.

Figure 2

Kinetic analysis of expression of VTC1 and VMA2 reporter genes under control of the GAL1 promoter. (A) Log phase yeast strains expressing VTC1 from the GAL1 promoter were cultured at 30°C in raffinose-containing media and shifted to galactose-containing media at t = 0. Samples were withdrawn at the indicated times and analyzed by 31P-MRS. (B) Samples were prepared as described in panel (A) using cells shifted from raffinose to galactose (open bar, left) or from galactose to glucose (hatched bar, right). PolyP was measured by 31P-MRS as shown in panel (A). The integral of the polyP peak was normalized to the MDP peak and the percentage of the maximum polyP signal (t = 24 h after the shift from raffinose to galactose) was calculated. (C) Samples were prepared as described in panel (B) and analyzed by northern blot with a probe for VTC1. The VTC1 transcript was normalized to the ADH1 transcript and the percentage of the maximum value for each data was calculated. (D) Samples were prepared as described in panel (B) and used for immunoblot analysis with antibody to HA-tagged VTC1. VTC1 protein was quantified and normalized to TATA box binding protein (TBP). Values in bar graph are the percentage of VTC1 at t = 0 (lane 1). Note that the strains expressing HA-tagged or untagged VTC1 protein produced almost the same amounts of polyP irrespective of endogenous or heterologous promoters. (E) Samples were prepared as described in panel (B) using yeast cells expressing HA-tagged VMA2 under the control of the GAL1 promoter. Northern blot analysis was carried out using probes for VMA2 (upper) or ADH1 (lower). VMA2 transcript was normalized to ADH1 and the percentage of maximum VMA2 expression (t = 6 h after the shift from raffinose to galactose) was calculated. (F) PolyP was quantified by 31P-MRS in yeast cells expressing HA-tagged VMA2 under the control of the GAL1 promoter. Samples were prepared at the indicated time after media switching as described in (B). The integral of the polyP peak was normalized to the MDP peak and the percentage of the maximum polyP signal (t = 12 h after the shift from raffinose to galactose) was calculated.

Similar results were observed with VMA2, in that the amount of polyP correlated the amount of VMA2 transcript during induction but not during repression of VMA2 mRNA synthesis (data not shown). However, the correlation between gene expression and polyP improved when the reporter gene was modified with an N-terminal HA epitope tag. Thus, when an HA-tag was fused to the amino-terminus of VMA2, a decline in VMA2 mRNA correlated with a decline in polyP, with a lag of a few hours (Figure 2E and F). This more rapid response might indicate that HA-tagged VMA2 has a shorter half-life than VMA2 (data not shown).

The VTC1 and/or VMA2 reporter system described above was then tested with MRI detection. The strains shown in Figure 1 were cultured in YPD to mid-log phase and transferred into 0.8 mm glass capillary tubes (Figure 3A) and yeast cultures and a reference sample (i.e. water) were analyzed by MRI (Figure 3B). Signals were integrated longitudinally in each capillary tube and are presented as a bottom view (Figure 3B, right). The average intensity of the polyP image was calculated from 31P-MRI data and the area corresponding to each sample was defined spatially using 1H-MRI data (Figure 3B, left).

Figure 3

Comparison of MRI, MRS and northern blot to quantify VTC1 and VMA2 reporter gene expression in yeast. (A) Schematic diagram of MRI apparatus. A total of 16 experimental and one reference samples, loaded into 0.8 mm capillary glass tubes, were placed in an 8 mm NMR tube which was then inserted into an NMR spectrometer with an imaging probe. (B) The proton density and the polyP-selective images were generated by 1H-MRI (left) and 31P-MRI (right), respectively. The sample number or letter and its position in the grid are shown below the MRI data. ‘R’ denotes reference and Arabic numerals indicate the promoter construct as described in Figure 1A. ‘W’ and ‘Δ’ denote wild type and Δvma2 (or Δvtc1) strains, respectively. (C) Gene expression was quantified by northern blot (open bar), MRS (grey bar) and MRI (dark grey bar). Results are shown as a percentage of the maximum value in each dataset (asterisk). The values are the average of two (MRI) or three (northern blot, MRS) independent experiments and shown along with the SEM. (D) The correlation coefficients between the northern blot and MRS or MRI data were calculated. Data points for gene expression driven by the endogenous VTC1 or VMA2 promoters were omitted from the calculations, because they deviated significantly from the exogenous promoters (see text).

Experimental values were measured by northern blot, MRS and MRI, and values were expressed as a percentage of the maximum value for each dataset. The results are summarized in Figure 3C and D. Figure 3D shows the correlation coefficients between northern blot and MRS or MRI methods. These results strongly suggest that gene expression can be quantified accurately by MRS and MRI using VTC1 or VMA2 as a reporter gene in yeast cells. To our knowledge, this is the first report describing a fully quantitative reporter system with MRI detection that can be used to monitor spatial and temporal variation in gene expression.

In higher eukaryotes, the endogenous level of polyP is much lower than in yeast (8). In addition, VTC1 orthologs have not been identified in mammals whereas mutations of VMA2 orthologs cause genetic disease in humans (22,23). Therefore, in order to apply a similar reporter system in mammalian cells, an exogenous MRI reporter gene may be required. Preliminary results suggest that the Escherichia coli PPK1 gene, which encodes polyP kinase, is a candidate reporter gene for use in mammalian cells. PPK1 converts the terminal phosphate of ATP to polyP (24), and expression of PPK1 from a mammalian promoter induced polyP at a level readily detected by MRS or MRI in mammalian cells (unpublished results). Unfortunately, it has not been possible to characterize a PPK1-based reporter system in yeast, because the endogenous background of polyP interferes with measurement of induced PPK1 activity (data not shown). Although organism-specific reporter genes may be needed for future studies in mammalian cells, the system described in this report should be widely applicable.

In summary, this study presents a novel MR-based method for quantifying temporal and spatial variation in gene expression. This method can potentially be used in any tissue of any organism. Furthermore, this method may have important applications in the context of high-throughput cell-based assays, which are used in drug discovery screening and in many other areas of basic and clinical research.

SUPPLEMENTARY DATA

Supplementary data are available at NAR online.