Gene Expression Analysis in Platelets from a Single Donor: Evaluation of a PCR-Based Amplification Technique

BACKGROUND
Genetic analysis of platelet mRNA may facilitate the diagnosis of disorders affecting the megakaryocytic-platelet lineage. Its use, however, is limited by the exceptionally small yield of platelet mRNA and the risk of leukocyte contamination during platelet preparation.


METHODS
We depleted platelet suspensions of leukocytes by filtration and used a PCR-based RNA amplification step [switching mechanism at the 5' end of RNA templates (SMART)]. We tested the reliability and precision of the RNA amplification procedure by use of real-time PCR to measure quantities of specific transcripts: von Willebrand factor (vWF), A-subunit of coagulation factor XIII (F13A), and glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Microarray analysis was performed on platelet RNA with and without amplification.


RESULTS
Microgram quantities of platelet-specific cDNAs were produced from as little as 50 ng of total platelet RNA or 40 mL of whole blood. At cycle numbers <16, amplification of all transcripts tested was exponential with slightly more efficient amplification of low-abundance transcripts. Expression profiling of 9850 genes gave identical results for 9815 genes (1576 positive/8239 negative). Eight transcripts failed to be amplified by the SMART procedure. Expression of vWF, F13A, and GAPDH transcripts showed only minor day-to-day variations in three healthy individuals.


CONCLUSION
The proposed protocol makes extremely small amounts of platelet RNA available for gene expression analysis in single patients.

mechanism at the 5Ј end of RNA templates (SMART) 4technique (12 ).The main concern about amplification procedures, especially with PCR-based methods, is that they provide genetic information with high yield but an altered gene expression profile (10,17,20,21 ).
We carefully evaluated this aspect by comparative microarray analysis and by quantitative PCR.In the microarray experiments, original and amplified platelet RNA was profiled across 9850 genes.To determine the bias between transcript ratios, selected transcripts were quantified by real-time PCR during the amplification process.von Willebrand factor (vWF) was selected for two reasons: (a), with a length of 8.3 kb, vWF mRNA is one of the longest human messages; and (b), it is expressed exclusively in platelets and endothelial cells.We selected RNA coding for subunit A of coagulation factor XIII (F13A) as a second platelet typical RNA.This 3.8-kb spanning RNA represents the group of medium-sized RNAs and is highly expressed in platelets (2,3,5,6 ).Recently, expression of fibrinogen ␣-chain (FGA; 2.2 kb) has been found in platelets by array analysis.It has been ranked in position 632 of the most abundant platelet messages, lying between F13A (rank 199) and vWF (rank 2465) (5 ).On the basis of these data, we decided to use FGA as an additional candidate RNA representing platelet RNAs that are expressed in low to moderate concentrations.Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was chosen because this 1.3-kb gene is highly expressed in nearly all cells, including platelets.
Because the concentration of mRNA in platelets is much lower than in leukocytes, a small number of leukocytes can distort platelet gene expression profiles (22 ).To address this problem, we included a leukocyte depletion step in the platelet preparation protocol and monitored its efficiency by analyzing leukocyte-specific gene products.
Finally, we tested the reproducibility and sensitivity of our approach and the day-to-day variations by measuring the amounts of vWF, F13A, and GAPDH RNA over time in three healthy individuals.

Materials and Methods isolation and preparation of platelets
For large-scale RNA preparation, platelets were collected from healthy blood donors by platelet apheresis on an Amicus Crescendo cell separator (Baxter).Each leukocytereduced apheresate contained Ͼ2 ϫ 10 11 platelets in 200 -300 mL of autologous plasma.Contaminating leukocytes (Ͻ1 ϫ 10 6 per unit of apheresate) were completely removed by filtration (PXL2; Pall Biomedizin) as judged by flow cytometric analysis (LeucoCount TM ; Becton Dick-inson).The platelets were washed with Tyrode's buffer and collected by centrifugation at 1000g for 15 min, after which the platelet pellets were frozen in liquid nitrogen and stored at Ϫ80 °C.To obtain RNA from a single donor blood, we collected 20 -80 mL of citrate-anticoagulated blood from healthy volunteers.Platelets were separated from erythrocytes by centrifugation at 150g for 20 min at room temperature.The upper two thirds of the plateletrich plasma (PRP) was resuspended in twice a volume of Tyrode's buffer, filtered through a Purecell PL leukocyte removal filter (Pall Biomedizin), and processed as described above.

rna extraction
For microarray analysis, total RNA was isolated from six single donor platelet apheresis concentrates after leukocyte depletion with the use of TRIzol reagent (Invitrogen) as described previously (3 ).From each platelet concentrate with total platelet numbers in the range of 2.4 -2.8 ϫ 10 11 (mean, 2.7 ϫ 10 11 ) we could obtain 18.5-28.2g (mean, 22.5 g) of total RNA.When we worked with whole blood, we directly extracted mRNA from isolated platelets with oligo(dT)-coupled paramagnetic beads [Dynabeads Oligo(dT) 25 ; Dynal] according to the procedure described by Jakobsen et al. (23 ).Bead-coupled platelet mRNA was eluted in 10 L of RNase-free water.

cDNA generation and amplification
Platelet RNA was transcribed to cDNA and subsequently amplified by SMART technology (BD Biosciences Clontech) according to the manufacturer's instructions.Samples were amplified in a PTC-200 thermal cycler (MJ Research, Biozym Diagnostika, Hess) by use of the following program: 95 °C for 1 min, then 2-18 cycles at 95 °C for 15 s, 58 °C for 30 s, and 68 °C for 6 min.After every two cycles, 5 L of the reaction mixture was transferred to a fresh tube and kept at 4 °C; the remaining mixture was subjected to additional cycles.

quantitative real-time pcr
One-step quantitative real-time reverse transcription-PCR (RT-PCR) was used for measuring the concentrations of RNA transcripts coding for vWF, FGA, F13A, and GAPDH.Oligonucleotide primers and probes for quantification of vWF and FGA transcripts were designed by use of Primer Express software, Ver.1.5 (Perkin-Elmer).Primer and probe sequences for amplification of GAPDH mRNA were taken from the TaqMan Gold RT-PCR Kit protocol (Perkin-Elmer).Oligonucleotides for quantification of vWF, FGA, and GAPDH transcripts were purchased from Eurogentec.Detailed sequence information is provided in Table 1.For quantification of F13A transcripts, we used a predeveloped primer/probe set from Applied Biosystems (Assay-on-Demand TM product, containing a 5Ј-6-carboxyfluorescein-labeled MGB ® probe with a black hole quencher at the 3Ј end).Corresponding sequence information was not available.All sequences were chosen to prevent amplification of genomic DNA by overlapping or spanning exon/intron boundaries.
Quantitative RT-PCR reactions were performed with the QuantiTect Probe RT PCR Kit (Qiagen).Reactions were performed in a final volume of 20 L containing 1ϫ master mixture (including PCR buffer, deoxynucleotide triphosphates, 4 mM MgCl 2 , and ROX reference dye), FGA or vWF forward and reverse primers (150 nM each), GAPDH forward and reverse primers (100 nM each), FGA or vWF probe (200 nM), GAPDH probe (100 nM), 0.2 L of QT Probe RT Mix, and 1 L of each dilution of calibrators and samples.For quantification of F13A transcripts, 1 L of the primer/probe mixture was added to each reaction.Samples were amplified in a 96-well spectrofluorometric thermal cycler (ABI Prism SDS 7700; Applied Biosystems) using the following program: 50 °C for 30 min, 95 °C for 15 min, 40 cycles at 94 °C for 15 s, and 60 °C for 1 min.
To measure amplified cDNA, we performed real-time PCR using probes and primers identical to those used for the quantitative RT-PCR.Multiplex reactions were performed in a final volume of 25 L containing 1ϫ PCR buffer [20 mM Tris-HCl (pH 8.5), 50 mM KCl], 4 mM MgCl 2, 200 mM each deoxynucleotide triphosphate, 0.5 L of ROX reference dye, 1.25 U of PlatinumTaq DNA polymerase, and 1 L of each calibrator or sample preparation.Oligonucleotide concentrations were the same as those described for RT-PCR reactions.PlatinumTaq DNA polymerase and ROX reference dye were purchased from Invitrogen.Thermal cycling using the Prism SDS 7700 was performed with the following profile: 95 °C for 5 min followed by 40 cycles consisting of 20 s of denaturation at 95 °C and 60 s of annealing and extension at 60 °C.
All calibrators and samples were run in duplicate.

rna synthesis and calibrator preparation
RNA calibrator for absolute quantification of vWF and GAPDH transcripts were prepared by in vitro transcription.Starting points were PCR-amplified cDNAs coding for vWF and GAPDH that were cloned in the SrfI site of pPCR-Script Amp SK (ϩ) (Stratagene).These constructs were used to generate DNA templates containing a T7 RNA polymerase promoter sequence at their 5Ј end with the following sequences of vWF and GAPDH downstream: for vWF, bp 5633-5919 (GenBank accession no.NM_000552); and for GAPDH, bp 8 -525 (GenBank accession no.M33197).In vitro transcription was performed with these modified DNA templates using T7 RNA polymerase (Invitrogen) according to the manufacturer's instructions.After digestion of DNA templates with DNase I (Roche), RNAs were purified by use of the RNeasy Mini Kit (Qiagen).RNA was quantified by photometric measurement (A 260 reading of 1 ϭ 44 mg/L).All RNA stock solutions were stored at Ϫ70 °C.Total placental RNA (BD Biosciences Clontech) was used for relative quantification of F13A transcripts.One unit was defined as the number of F13A transcripts present in 10 pg of the placental RNA.
Calibrators for quantification of SMART-generated vWF and GAPDH cDNA products were prepared from cDNA plasmids containing the sequences described above.The plasmids were isolated from transformed XL10-Gold (Stratagene) cultures and quantified by photometric measurement (A 260 reading of 1 ϭ 50 mg/L).The DNA concentration is expressed in molecules/mL.All DNA stock solutions were stored at Ϫ70 °C.For relative quantification of SMART-generated F13A sequences, total placental RNA was transcribed to cDNA by use of Super-Script RNase H Ϫ reverse transcriptase (Invitrogen).

comparative microarray analysis
To compare gene expression profiles from platelet RNA with and without amplification, we used total RNA isolated from single donor platelet apheresis concentrates in six individual experiments.Unamplified samples were fluorescently labeled by use of 10 g of platelet RNA, Cy3-or Cy5-dCTP (Amersham Biosciences), and the La-belStar reagents (Qiagen).Amplified samples were generated each from 0.5 g of total platelet RNA by use of the SMART fluorescent probe amplification reagents (BD Before the hybridization on microarray glass slides representing 9850 human genes (MWG-Biotech AG), the Cy3-and Cy5-labeled samples were combined as follows: slide 1, unamplified sample 1 (Cy3) ϩ unamplified sample 4 (Cy5); slide 2, unamplified sample 2 (Cy3) ϩ unamplified sample 5 (Cy5); slide 3, unamplified sample 3 (Cy3) ϩ unamplified sample 6 (Cy5); slide 4, amplified sample 1 (Cy5) ϩ amplified sample 4 (Cy3); slide 5, amplified sample 2 (Cy5) ϩ amplified sample 5 (Cy3); slide 6, amplified sample 3 (Cy5) ϩ amplified sample 6 (Cy3).Arrays were hybridized and washed according to the manufacturer's protocol and scanned by a laser scanner (GMS 417Ј MWG-Biotech).Computer-assisted data evaluation was performed with use of the ArrayVision software (Imaging Research, Inc.) as described previously (3 ).In brief, negative hybridization signals revealed intensity values Ͻ3000, the gray area range was 3000 -5000, and positive hybridization signals showed intensity values Ͼ5000.Mean values were calculated from the six unamplified and the six amplified samples for each gene spot on the microarray.The mean values from the unamplified and the amplified samples were compared in a scatter plot.

platelet preparation and precision of rna isolation
Platelets were prepared from 40 mL of citrate-anticoagulated blood from 16 healthy volunteers.Platelet count ranged from 105 000 to 259 000/L.Leukocyte counts ranged from 3600 to 6100/L.After preparation of PRP and leukofiltration, no leukocytes were detectable by flow cytometric analysis.Mean (SD) platelet loss attributable to preparation of PRP and leukofiltration was 72 (3)%.
To determine the precision of mRNA isolation with oligo(dT)-coupled magnetic beads, we aliquoted platelets obtained from one donor by platelet apheresis after filtration (1 ϫ 10 10 platelets each), and mRNA was extracted in triplicate on 3 different days.vWF and GAPDH mRNA was analyzed by quantitative RT-PCR.The mean (SD) the threshold cycle (Ct) values were 30.1 (0.65) and 19.1 (0.27) for vWF and GAPDH mRNA, respectively; the CVs indicate the low intraassay variation.For the run-to run-variation, the mean (SD) Ct values were 30.21 (0.65) and 19.52 (0.45) for vWF-and GAPDH mRNA, respectively.

efficiency of amplification
One of the most critical steps in PCR-based RNA amplification is the maintenance of the original message profile.To obtain optimum results, PCR should be in the exponential phase of the reaction, and irrespective of length and abundance, all mRNAs should be amplified with equal efficiency.To establish optimum PCR conditions, purified total platelet RNA at concentrations of 50 ng, 100 ng, and 1 g was subjected to SMART-PCR amplification.During the process of PCR amplification, aliquots were repeatedly taken from each sample, and the concentrations of vWF, F13A, and GAPDH cDNA were measured.
Copy numbers exponentially increased from cycles 10 to 18 for all RNA species tested when 50 or 100 ng of initial RNA was used (Fig. 1).At a starting RNA concentration of 1 g, the reactions became nonexponential after 16 amplification cycles for F13A and GAPDH, whereas vWF amplification remained in the exponential phase.Interestingly, we failed to detect message for FGA in our samples.We confirmed this result by testing 10 additional unamplified and SMART-amplified platelet RNA samples from 10 healthy individuals who tested positive for vWF transcripts.
The amplification efficiencies were calculated for each initial RNA concentration according to the formula: E ϭ e s Ϫ 1, where E is the efficiency of the PCR and s is the slope of the corresponding interpolation curve (Fig. 1).Efficiencies for vWF RNA amplification were 0.81, 0.84, and 0.80 when we used 50 ng, 100 ng, and 1 g of total RNA, respectively, as the initial starting concentration.In contrast to the nearly identical amplification efficiencies observed for vWF, the amplification efficiencies of F13A (0.66, 0.65, and 0.36) and GAPDH (0.72, 0.76, and 0.65) transcripts decreased with increasing concentrations of starting RNA (50 ng, 100 ng, and 1 g).Overall, the amplification efficiency was higher for vWF amplification than for GAPDH or F13A amplification.

precision of amplification results
The assay-to-assay variation, which defines the precision of the amplification process, was calculated by analysis of cDNAs generated from identical RNA samples (0.5 g) using 14, 18, and 22 amplification cycles three times on 3 different days.The mean Ct values of the multiplex vWF/GAPDH PCR decreased with increasing SMART amplification cycles.The corresponding SD increased with increasing SMART amplification cycles, but they did not exceed 0.73 for both transcripts tested.Considering that these values include the intraassay variability of the PCR, they are remarkably low.We determined the intraassay variability of the multiplex PCR by processing each cDNA calibration dilution six times in one experiment.The maximum SD values calculated on the basis of Ct values were 0.19 and 0.23 for vWF and GAPDH, respectively.

gene expression profiles of unamplified and smart-amplified platelet rna
The platelet gene expression profile was determined in six microarray hybridization experiments with either unamplified or SMART-amplified platelet RNA obtained from six single donor platelet concentrates.Mean hybridization signals were calculated for each gene and compared between the unamplified samples (directly labeled with Cy3 or Cy5 by reverse transcription) and the amplified samples (SMART-amplified and labeled with Cy3 or Cy5).The scatter plot of mean signal intensity values of the unamplified vs the amplified samples revealed a strong linear relationship (R 2 ϭ 0.915; Fig. 2).Positive hybridization signals were detected for 1576 genes (16%) in all samples.The most prominent genes corresponded to those described previously (3 ), such as platelet factor 4, RANTES, glycoprotein Ib␤, and others.For 8239 genes (83.6%), we found signal intensities in the gray area or negative range in all samples analyzed.The CVs were in the range of 0.001-1.6[mean (SD), 0.34 (0.17)] for the unamplified samples and 0.001-1.4[0.31 (0.16)] for the amplified samples.The hypothetical protein KIA0433 revealed that the highest CV values were 1.6 in the unamplified and 1.4 in the amplified samples.This was attributable to the highly positive signal intensity values in unamplified and amplified RNA sample 5, whereas the signals were negative or in the gray area range in all other RNA samples.Similar results could be seen for the ribosomal protein L29 (CV unamplified ϭ 1.22; CV amplified ϭ 1.39), the ubiquitin-specific protease 1 (1.26 and 1.12, respectively), the acid fibroblast growth factor-like protein GLIO703 (1.22 and 1.20, respectively), and others.In summary, 10 genes had CV values Ն1 in both sample types.Furthermore, 28 genes had CV values Ն1 in at least one of the two sample types (25 genes only in the unamplified samples; 3 genes only in the amplified sam-  ples).In all cases, the high CV values resulted from one or two positive signals among the six analyzed RNA samples, indicating an interindividual difference in gene expression.
Only 35 of the 9850 genes (0.4%) revealed discrepant results when we compared signal intensities for the unamplified and the SMART-amplified samples.A full list of these genes is available as a table in the Data Supplement that accompanies the online version of this article at http://www.clinchem.org/content/vol50/issue12/.Among these genes we identified 27 with negative signals in the unamplified samples and unambiguously positive signals in the amplified samples, such as the matrix metalloproteinase 14, troponin C2, and others.This may reflect the higher sensitivity of the SMART amplification and may not be regarded as a "true" discrepant result.Positive signals in the unamplified samples but negative signals in the SMART-amplified samples were found for eight genes, such as serine racemase, actin-binding LIM protein 1, and others.Microarray analysis without amplification revealed positive hybridization signals for these genes in all RNA samples, but the signals were negative when we used the amplification technique.These should be regarded as true discrepant results because the SMART technique in combination with microarray hybridization analysis failed to detect the gene transcripts.

minimum sample size
The minimum sample size was defined as the volume of whole blood needed as starting material to produce more than 5 g of platelet cDNA with detectable vWF transcripts by SMART PCR amplification using a maximum of 18 cycles.The minimum platelet number needed for RNA extraction and cDNA generation was first determined by testing different numbers of platelets obtained by platelet apheresis.When we started with 2 ϫ 10 6 platelets, GAPDH or vWF transcripts were not detectable by quantitative PCR even after 22 amplification cycles.When we increased the platelet number to 2 ϫ 10 7 , we obtained measurable GAPDH transcripts but vWF remained undetectable.At starting concentrations of 2 ϫ 10 8 platelets or more, both marker transcripts became detectable in a platelet concentration-dependent manner.Because 2 ϫ 10 8 platelets are equivalent to 2-4 mL of whole blood, we started the evaluation of the minimum volume of whole blood with a minimum of 5 mL.However, because of losses of thrombocytes during sample preparation and leukofiltration, 40 mL of whole blood was required to achieve reliable and reproducible results with a mean yield of 25-30 g of cDNA.Typical results obtained with blood samples from healthy volunteers (platelet counts, 143 000 -398 000/L) are listed in Table 2.

inter-and intraindividual variability of vWF, F13A, and gapdh transcripts in platelets
Blood samples (40 mL of whole blood) from three healthy volunteers were obtained weekly over a period of 3 weeks.Platelet counts in citrate-anticoagulated blood ranged from 138 000 to 146 000/L, 159 000 to 169 000/ L, and 217 000 to 265 000/L in individuals 1, 2, and 3, respectively.After mRNA isolation, vWF, GAPDH, and F13A were quantified by RT-PCR.The results (Fig. 3) demonstrated interindividual variations for all three transcripts with the lowest concentrations seen in individual 1.These differences remained stable over time, indicating a low intraindividual variability.

Discussion
Our aim was to establish a protocol that makes analysis of the platelet transcriptome applicable to single patients with suspected platelet disorders.
Contaminating leukocytes are a potential problem in platelet RNA analysis because the content of mRNA in leukocytes is Ͼ10 000-fold higher than in platelets (22 ).To overcome this problem, several techniques, including laser-assisted microdissection and manipulation, serial filtration, and antibody-mediated depletion of leukocytes  by magnetic beads, have been used (2,3,22 ).Filtration is the only technique that is used on a routine basis (24,25 ), and filters adapted for the processing of small volumes of PRP are commercially available.Taking these advantages into account, we studied the efficiency of leukocyte filtration by flow cytometry analysis and determination of CD45 and CD14 transcripts.After one filtration, leukocytes were not detectable by flow cytometry analysis, and negative array results were obtained for CD45 and CD14 transcripts.These results support previously published data obtained on large volume preparations of platelets (3,26 ).When we used a more sensitive quantitative RT-PCR approach, however, low concentrations of CD45 transcripts were detectable, indicating that the filtered PRP was not completely devoid of leukocytes.Analyzing the efficiency of leukocyte filtration, we found that one filtration step reduced the amount of CD45 transcripts by three orders of magnitude, corresponding to a reduction in the leukocyte/platelet ratio from 1:2000 to 1:2 000 000. Adding more filtration steps did not significantly reduce the concentration of CD45 mRNA but led to additional platelet loss.On the basis of these results, we included one filtration step in our experimental protocol.The quality of the mRNA preparation is of critical importance when working with a tissue source that contains only minute amounts of mRNA.We used oligo(dT)-coupled magnetic beads, which directly bind platelet mRNA representing 1-3% of total RNA.To control our mRNA isolation procedure, we isolated mRNA at various time points from one subsampled platelet preparation.Subsequently, the numbers of vWF and GAPDH transcripts were measured.The SD for the Ct values never exceeded 0.73, demonstrating the high reproducibility and precision of this procedure.
The use of an amplification procedure before gene expression analysis requires maintenance of the original message profile.In our microarray investigation, we compared mRNA profiles from 10 g of unamplified and 0.5 g of SMART-amplified platelet RNA and found that 91.5% of the gene expression information was maintained.A total of 508 genes revealed negative signals in the unamplified samples but were positive (27 genes) or in the gray area range (481 genes) in the SMART-amplified samples.These results may be regarded as indicative of the higher sensitivity of the SMART technique, but not as true discrepancies.However, we observed a false-negative microarray result in the SMART-amplified samples for eight transcripts.Obviously, amplification of these transcripts failed for unknown reasons.Their lengths varied between 1.0 and 6.8 kb; therefore, length can probably be ruled out as a main reason for the amplification failure.One can speculate that failure of template switching at the 5Јend might occur because of secondary structures in these transcripts.Such failures must be considered for future studies on individual mRNA profiling of platelet RNA.
SMART amplification revealed consistent results when we compared microarray data from six individual experiments.The CV for the majority of genes (8913 of 9850; 90.5%) was Ͻ0.Quantification of vWF, F13A, and GAPDH as marker transcripts to determine how accurately PCR-generated cDNA reflects the original platelet mRNA population revealed exponential amplification between cycles 10 and 18 for all transcripts when 50 or 100 ng of total platelet RNA was used as starting material.However, amplification efficiency was slightly higher for vWF than for F13A and GAPDH.Most likely, this is attributable to the more than 200-fold lower concentration of vWF transcripts compared with GAPDH transcripts in the starting mRNA, which led to the C 0 t effect (28,29 ).The bias toward more efficient amplification of low-abundance transcripts during SMART amplification has already been described in other tissues (15,18 ).However, when we compared platelet RNA with SMART-amplified RNA by microarray analysis, the correlation was excellent.Presumably, the semiquantitative microarray method is not sensitive enough to detect such differences, which can be seen with quantitative real-time PCR (29 ).
Using quantitative PCR, we were not able to detect FGA transcripts in any of the amplified or unamplified samples tested.This is in contrast to recently published array data (5 ), but agrees well with data obtained by in situ hybridization and PCR testing (30 ).Both sensitive techniques failed to demonstrate FGA mRNA in platelets or megakaryocytes.The reason for this discrepancy remains unknown, but it emphasizes that array results need to be validated by independent methods such as RT-PCR.
The platelet count is regulated in a narrow range with minimal day-to-day individual variation (31 ).At present, however, no data are available on intra-and interindividual variability in platelet gene expression.In prelimi-Clinical Chemistry 50, No. 12,2004 nary experiments, we measured changes in vWF, F13A, and GAPDH mRNA concentrations over time in three healthy individuals.The interindividual differences in the expression of all three transcripts remained stable over 3 weeks, indicating that there is only minor intraindividual variation.
In conclusion, the assessed protocol allows production of micrograms of platelet-specific cDNA from blood volumes as low as 40 mL.SMART amplification does not cause distortion of the gene expression profile as estimated by microarray analysis.Preliminary results on samples from three healthy volunteers indicated interindividual differences of platelet gene expression profiles that seemed to be consistent over time.Therefore, this protocol makes platelet RNA isolated from single patients available for gene expression studies that may detect target genes involved in the development of inherited megakaryocytic/platelet disorders.

Fig. 1 .
Fig. 1.Gene-specific monitoring of SMART amplification by quantitative real-time PCR.Total platelet RNA at concentrations of 0.05 g (OE), 0.1 g (f), and 1 g (ࡗ) was amplified by the SMART method.At the indicated amplification cycle numbers (x axis), vWF (A), F13A (B), and GAPDH cDNA (C) concentrations were measured by quantitative real-time PCR.Results are shown as mean (SD; error bars) of three independent experiments.

Fig. 2 .
Fig. 2. Scatter plot of mean signal intensity values from comparative microarray analysis.Glass slide microarrays representing 9850 human genes were used to characterize the mRNA profile in human platelets.Mean values of hybridization signals were calculated from six individual experiments for each individual gene and compared between unamplified (x axis) and SMART-amplified samples (y axis), indicated by gray circles in the scatter plot.The line indicates the linear relationship between the microarray hybridization results from unamplified and amplified samples (R 2 ϭ 0.915) calculated by the equation: y ϭ 0.8226x ϩ 203.5.

Table 1 .
Oligonucleotide primers and probes used for quantitative PCR.FGA and vWF probes were labeled with the fluorescent reporter dye 6-carboxyfluorescein (FAM) at the 5Ј end and the quencher 6-carboxytetramethylrhodamine (TAMRA) at the 3Ј end.To enable multiplex reactions, a different reporter dye (JOE) was labeled at the 5Јend of the GAPDH probe, but TAMRA was still used as quencher at the 3Ј end.bNucleotide positions based on the GenBank accession nos.J00127 (FGA), NM_000552 (vWF), and NM-002046 (GAPDH). a

Table 2 .
RNA amplification results for three representative blood samples.
(27 )d only 13 genes had CV values Ն1.Mean CV values did not differ between amplified and unamplified samples.This confirmed the maintenance of gene expression profiles after SMART amplification.The observed differences may not result from the amplification technique but may represent individually different gene expression patterns.Comparability of array results depends on the array format used(27 ).The array results presented here are consistent with our previously published data obtained with an identical platform (MWG Biotech) (3 ).When we compared our results with the data published by Gnatenko et al. (2 ) and McRedmond et al. (6 ),we had to consider that these results were generated with a different microarray platform (Affymetrix).Among the 50 most abundant platelet transcripts identified by Gnatenko et al. (2 ) and McRedmond et al. (6 ), 25 are arrayed on both platforms.For 22 of these genes, we confirmed strongly positive hybridization signals.Only three genes gave discrepant results: one gave signal intensities in the gray area (cofilin 1, nonmuscle) and two in the negative range (progesterone receptor component and nonmuscle myosin light chain).