The stress kit: a new method based on competitive reverse transcriptase–polymerase chain reaction to quantify the expression of human αB-crystallin, Hsp27, and Hsp60

Abstract

Abstract We describe a reverse transcriptase–polymerase chain reaction method for the semiquantitative detection of mRNAs encoding the human heat shock proteins αB-crystallin, Hsp27, and Hsp60. The method involves the coamplification of cellular mRNA-derived cDNA with a dilution series of a competitor fragment (internal standard), using 1 primer pair common to both templates. Internal standards were based on cellular-derived cDNA engineered to be slightly smaller to differentiate between the target and the standard on electrophoretic separation. Initial cDNA quantitations can be corrected for possible variations during cDNA synthesis by standardizing to the levels of β-actin–encoding cDNA. We show that the coamplified templates accumulate in a parallel manner with the cellular-derived cDNA throughout both the exponential and the nonexponential phase of amplification. Furthermore, we illustrate the utility of this technique by quantifying increased expression of αB-crystallin, Hsp27, and Hsp60 mRNA in astroglioma cells on heat shock.

INTRODUCTION

Several lines of evidence indicate that the cellular mechanisms by which Hsp expression is regulated in various types of cells, for example, in the human central nervous system, are different (Freedman et al 1992; Satoh and Kim 1995; Bajramović et al 1997). In addition, there is also evidence for differential expression patterns of different Hsp in the same cell type in response to the same stimulus (Wiegant et al 1994; Lee et al 1995; Head et al 1996). Primary rat astrocytes, for example, show an increase in mRNA expression levels of αB-crystallin but not of Hsp27 in response to tumor necrosis factor–α stimulation (Head et al 1994). Such observations indicate that Hsp expression, in response to stressors other than heat shock, is regulated by differential control mechanisms rather than by uniform mechanisms. In many cases the promoter sequences that are involved are unknown or still await characterization. Consequently, prediction of possible Hsp induction by specific stressors is prevented. To allow studies on the regulation of Hsp, reliable methods to quantitate Hsp-encoding mRNAs are an important prerequisite.

In this paper we describe a sensitive competitive reverse transcriptase–polymerase chain reaction (RT-PCR) method to quantify the mRNA levels of 3 human Hsps: αB-crystallin, Hsp27, and Hsp60. These 3 Hsps were selected on the basis of their presumed involvement in several diseases. αB-crystallin has been described as a putative autoantigen in multiple sclerosis (van Noort et al 1995); Hsp27 is associated with enhanced tumorgenicity of a variety of cell types (Garrido et al 1998; Lemieux et al 1997); and Hsp60 is thought to be involved in the pathogenesis of atherosclerosis (Roma and Catapano 1996), type I diabetes (Abulafia et al 1999); multiple sclerosis (Selmaj et al 1992); and rheumatoid arthritis (De Graeff-Meeder et al 1991).

Quantitation by PCR is difficult because the efficiency of PCR amplification may vary among samples, mainly because of differences in the quality of mRNA and cDNA preparations and variations occurring during the PCR reaction. Therefore, we chose to design a competitive PCR in which a competitor control fragment (internal standard) is coamplified with sample cDNA using 1 set of primers within a single PCR mixture (Wang et al 1989; Siebert and Carrick 1992). The internal standard was engineered to be slightly smaller than the target to allow separation of the respective amplicons using agarose gel electrophoresis. PCR reaction tubes containing the target samples were spiked with a dilution series of internal standard. When the molar ratio of the PCR products generated from target and internal standard is equal to 1, the amount of target cDNA is equal to that of the internal standard. As the amount of internal standard is known, the amount of target can thus be determined. Initial cDNA quantitations were corrected for possible variations during cDNA synthesis by standardizing to cDNA levels of the endogenous housekeeping gene β-actin. In this report we describe the optimization and validation of this assay.

MATERIALS AND METHODS

In vitro culture of the U373 MG astroglioma cell line

U373 MG (Number HTB-17; American Type Culture Collection, Manassas, VA, USA) cells were cultured in 1:1 v/v Dulbecco modified Eagle medium (high glucose; Life Technologies, Breda, The Netherlands)/HAMF10 (with l-glutamine; Life Technologies) plus 10% v/v fetal calf serum (Gibco; Life Technologies, Breda, The Netherlands) and antibiotic supplement (penicillin 100 U/mL and streptomycin 0.1 mg/mL) at 37°C in a humidified atmosphere containing 5% carbon dioxide. Cells (2.5 × 10⁵) were seeded into poly-l-lysine (15 μg/mL; Sigma, St Louis, MO, USA)-coated 25-cm² flasks. To passage cells, cells were rinsed with phosphate-buffered saline, incubated for 5–10 minutes with 0.25% w/v porcine trypsin (Sigma) at 37°C, washed once with culture medium containing 10% v/v fetal calf serum, and replated in poly-l-lysine–coated flasks. Confluent cell cultures were exposed to heat shock at 43°C by the addition of an equal volume of prewarmed medium at 49°C. Cells were left for 30 minutes at 43°C before they were allowed to recover at 37°C.

Competitive RT-PCR

Total RNA was isolated directly from cells in the culture flasks using RNAzolB according to the manufacturer's protocol (Campro Scientific, Veenendaal, The Netherlands) and was precipitated with isopropanol. Using 2.5 μg of RNA as a template, 50 μL of cDNA was produced using the reverse transcription system (Promega, Madison, WI, USA). For amplification, sample cDNA (1 μL), internal standard cDNA (1 μL), and the appropriate primer pair (1 μL for each primer of a 20 μM solution) were added to 1 μL 10 mM diethylnitrophenyl thiophosphate mix, 5 μL 10× PCR buffer (0.1 M Tris/HCl [pH 8.4], 0.5 M KCl, and 0.6 mg bovine serum albumin/mL), 1 μL Taq mix (PCR buffer containing 1 U Taq polymerase [Life Technologies, Amsterdam, The Netherlands]) and 39 μL Depc-H₂O. For β-actin and Hsp60, 22.5 mM MgCl₂; for αB-crystallin, 30.0 mM MgCl₂; and for Hsp27, 15.0 mM MgCl₂ was added to the 10× PCR buffer. Competitive RT-PCR was performed at 30 cycles for β-actin, αB-crystallin, and Hsp60 and at 27 cycles for Hsp27. One cycle consisted of 30-second melting at 94°C, 30 seconds' annealing at 57°C, and 30-second elongation at 72°C. Primers used and amplified fragment sizes are given in Table 1. Sample cDNA was amplified together with increasing amounts of the appropriate internal standard. The 50% equivalence point was calculated following agarose gel (2%) electrophoresis and densitometrical analysis (Bioprofil V6.0 software, Vilber Lourmat, France) of the ethidium bromide–stained amplicons. Linear regression was applied on five 2-fold dilution analyses centered around the 50% equivalence point. Initial cDNA quantitations were corrected for possible variations that might have occurred during cDNA synthesis by standardizing to the levels of β-actin–encoding cDNA and thus converted into relative content values. When absolute cDNA values for β-actin were below 0.1 ng/mL, cDNA synthesis was repeated using more mRNA.

Table 1.

 Primers used to amplify internal standards and sample cDNA and the amplified fragment sizes

Development of the internal standards

Internal standards were developed using RNA of unstimulated U373 MG cells. The mRNA was transcribed into cDNA and used for RT-PCR as described above. After agarose gel electrophoresis, the PCR product was isolated from gel and purified using the Geneclean kit (Bio 101 Inc, Vista, CA, USA). Purified PCR products were then cloned into a pGEM-T vector (Promega) and competent Escherichia coli cells (strain JM109) were transformed with these vectors. Positive (white) clones were selected using a conventional blue-white screening method. Isolated plasmids (Qiagen, Leusden, The Netherlands) were subjected to restriction analysis to verify the length of the insert. Correct plasmids were subjected to digestion with specificially chosen restriction enzymes resulting in amplified fragments that were approximately 50 bp shorter than the original fragments. For β-actin, BalI and Bsu36I (Promega) were used; for αB-crystallin, BamHI (Promega) and DraII (Roche, Almere, The Netherlands) were used; for Hsp27, Bsu36I and NheI (Promega) were used; and for Hsp60, HpaI (Promega) and DraII (Roche) were used. Nonhomologous sticky ends generated by the different restriction enzymes were blunt ended by treatment with 1 U/μL mung bean nuclease (Promega) for 30 minutes at 30°C. After subsequent ligation and transformation of the plasmids to E coli, clones were analyzed by restriction analysis. Plasmids displaying the expected restriction pattern were selected and sequenced using the T7 sequencing kit (Pharmacia, Roosendaal, The Netherlands) to confirm the correct sequence of the complete internal standards.

Parallelism assays

Parallelism assays were performed using approximately the same amounts of predetermined sample cDNA and internal standard. RT-PCR was performed as described above. From cycle 20 until cycle 40, PCR reactions of 2 samples were terminated every 2 cycles. These samples were then subjected to gel electrophoresis to compare amplification efficiencies of the sample cDNA and the internal standard by densitometrical analysis.

RESULTS AND DISCUSSION

A competitive RT-PCR assay was developed to allow quantitation of mRNA levels of 3 human heat shock proteins. Table 1 shows the sequences of the primers selected and the amplicon lengths that were generated. All amplified fragments were analyzed using sequence analysis (data not shown). Internal standards were synthesized and used to quantify mRNA-derived cDNA levels of Hsp and β-actin. cDNA levels of the housekeeping gene β-actin were used to correct for possible variations during cDNA synthesis from different samples.

To assess whether amplification efficiencies of sample cDNA and of the smaller internal standards were comparable, parallelism assays were performed (Becker-André and Hahlbrock 1989; Gilliland et al 1990). Results show that the amplification of the internal standards and the samples are indeed parallel up to the plateau phase (Fig 1). As the amplification efficiencies were identical for targets and internal standards throughout the amplification process, the amount of Hsp-cDNA in the total cDNA can theoretically be calculated at any cycle number in the range examined (Bouaboula et al, 1992). However, it should be noted that during the competitive PCR for Hsp27 at higher cycle numbers (≥cycle 32), heteroduplex formation, composed of single-stranded internal standard cDNA and single-stranded sample cDNA, was observed (Fig 1c). This heteroduplex formation, visible on the gel as a faint third band that migrated more slowly than the sample cDNA, can interfere with precise quantification. This phenomenon, also reported by others (Henley et al 1996; Boer and Ramamoorthy 1997), was observed only when a 1:1 ratio of sample and internal standard was used. In our system, heteroduplex formation can be prevented either by using fewer PCR cycles for quantitation of Hsp27 cDNAs or by performing agarose gel electrophoresis under denaturing conditions. As fewer amplification cycles (25–30) still generate a sufficient amount of amplicons for quantitation, this solution for prevention of heteroduplexes is preferable.

Fig 1.

 Parallelism assays for β-actin, αB-crystallin, Hsp27, and Hsp60. Parallelism assays were performed using approximately the same amounts of internal standard and sample cDNA. Starting at cycle 20 and ending at cycle 40, amplicons were sampled in duplicate every second cycle. Amplicon samples were subjected to gel electrophoresis (as shown in the lower figures) and densitometrical analysis (as shown in the upper graphs; each point is the mean of 2 analyses). A low molecular weight marker consisting of 50-bp–spaced markers from 100–500 bp and 100-bp–spaced markers from 500–1000 bp (Biozym, Landgraaf, The Netherlands) was used as a reference at the right-hand side of the samples. (A) β-Actin parallelism assay with 1.25 ng internal standard/mL. (B) αB-crystallin parallelism assay with 0.5 ng internal standard/mL. (C) Hsp27 parallelism assay with 0.4 ng internal standard/mL. The extra band appearing as a result of heteroduplex formation (visible ≥ cycle 32) is indicated by an open arrowhead at cycle 34. (D) Hsp60 parallelism assay with 2 ng internal standard/mL.

The applicability of the technique was tested by subjecting U373 MG astroglioma cells to heat shock and analyzing Hsp mRNA levels at 3 subsequent times. cDNA values for all proteins, as well as both their standardization to β-actin cDNA levels and final standardization to the levels of unstimulated cells, are shown in Table 2. Quantitation of mRNA-derived cDNA was highly reproducible, even when separately synthesized cDNA batches of the same mRNAs were used (data not shown). Figure 2 shows that heat shock indeed resulted in an upregulation of the mRNAs of all 3 Hsps tested. This was confirmed at the protein level by immunohistochemistry (data not shown). Interestingly, the kinetics of the 3 Hsps studied differed. Where both Hsp27 and Hsp60 mRNA levels were upregulated within 2 hours, αB-crystallin showed markedly slower kinetics. The sensitivity of the quantitative PCR assay was within the 10–100 pg Hsp cDNA/mL range. mRNA isolated from 5 000–10 000 U373 cells still allowed the production of sufficient amounts of cDNA to perform reproducible quantitative PCR analyses. However, sensitivity should be assessed for each cell type separately, as constitutive Hsp mRNA levels vary considerably between different types of cells.

Table 2.

 β-Actin, αB-crystallin, hsp27, and hsp60 mRNA levels in U373 cells at different times following heat shock as determined by competitive rt-PCR^a

Fig 2.

 Upregulation of αB-crystallin, Hsp27, and Hsp60 following heat shock of U373 cells. Relative values of αB-crystallin, Hsp27, and Hsp60 mRNA levels are shown as calculated in Table 2.

The differential expression patterns of Hsp in the same cell type in response to the same stimulus (Wiegant et al 1994; Lee et al 1995; Head et al 1996; Bajramović et al personal communication) indicate that Hsp expression in response to stressors other than heat shock is regulated by differential control mechanisms rather than by uniform mechanisms. The assay described here provides a powerful tool to analyze such control mechanisms for αB-crystallin, Hsp27, and Hsp60 in response to various stimuli.