|Home | About | Journals | Submit | Contact Us | Français|
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Pseudorabies virus (PRV), a neurotropic herpesvirus of pigs, serves as an excellent model system with which to investigate the herpesvirus life cycle both in cultured cells and in vivo. Real-time RT-PCR is a very sensitive, accurate and reproducible technique that can be used to detect very small amounts of RNA molecules, and it can therefore be applied for analysis of the expression of herpesvirus genes from the very early period of infection.
In this study, we have developed and applied a quantitative reverse transcriptase-based real-time PCR technique in order to profile transcription from the whole genome of PRV after lytic infection in porcine kidney cells. We calculated the relative expression ratios in a novel way, which allowed us to compare different PRV genes with respect to their expression dynamics, and to divide the PRV genes into distinct kinetic classes. This is the first publication on the whole-genome analysis of the gene expression of an alpha-herpesvirus by qRT2-PCR. We additionally established the kinetic properties of uncharacterized PRV genes and revised or confirmed data on PRV genes earlier examined by traditional methods such as Northern blot analysis. Our investigations revealed that genes with the same expression properties form clusters on the PRV genome: nested overlapping genes belong in the same kinetic class, while most convergent genes belong in different kinetic classes. Further, we detected inverse relationships as concerns the expressions of EP0 and IE180 mRNAs and their antisense partners.
Most (if not all) PRV genes begin to be expressed from the onset of viral expression. No sharp boundary was found between the groups of early and late genes classified on the basis of their requirement for viral DNA synthesis. The expressions of the PRV genes were analyzed, categorized and compared by qRT2-PCR assay, with the average of the minimum cycle threshold used as a control for the calculation of a particular R value. In principle, this new calculation technique is applicable for the analysis of gene expression in all temporally changing genetic systems.
The pseudorabies virus (PRV; also called Aujeszky's disease virus or suid herpesvirus type 1) is a member of the alpha-herpesviruses and an important pathogen of pigs, causing Aujeszky's disease . The PRV is a valuable model organism in herpesvirus research  and also a powerful tool in neuroscience, employed to map neural circuits [[3-5], and ] and gene delivery . The PRV life cycle is primarily controlled at the level of transcription. The genes of α-herpesviruses are divided into three major temporal classes (immediate-early, IE or α; early, E or β, and late, L or γ), which are regulated in a coordinated, cascade-like fashion [ and ]. First, the IE genes are expressed, independently of de novo protein synthesis from the virus. The products of these genes are transcription factors and other regulatory proteins. PRV IE180 protein [homologous to herpes simplex virus (HSV) ICP4 protein] is the major regulator of the transcription of the E genes, which in turn are mainly involved in replication of the viral DNA. Blockers of DNA synthesis inhibit the E gene expression to a much lesser extent than the L gene expression. On a finer scale, the E genes can be subdivided into β1 (E) and β2 (E/L) genes. Finally, the L genes are transcribed, encoding mainly structural proteins involved in virion assembly. The expression of the L genes is partially (γ1, leaky L genes) or completely (γ2, true L genes) dependent on the viral DNA replication. HSV encodes 5 IE genes: icp0, icp4, icp22, icp27 and icp47. By contrast, PRV has been reported to express ep0 (homologous to icp0 of HSV;  and ul54 (the homolog of icp27 of HSV; ) proteins in the E kinetics, and it lacks the icp47 gene. The ie180 gene of PRV has been shown to be a true IE gene in various experimental systems [ and ]; however, there is no consensus as to whether the US1 protein (Rsp40; homologous to HSV ICP22) is expressed in the IE  or the E kinetics . Herpesviruses are capable of two types of infections: lytic (or productive) and latent . In the lytic pathway, the entire transcription machinery of the herpesvirus is initiated, and the progress of infection eventually leads to the production of new virions and the lysis of infected cells. In contrast, in latency, only a limited segment (LAT region) of the herpesvirus genome is transcriptionally active, no new virus particles are produced and the cells survive the infection. A characteristic feature of the organization of the herpesvirus genome is the presence of nested genes producing 3'-coterminal transcripts. The read-through of overlapping genes is regulated by the ICP27 protein, which helps in the recognition of the internal polyA signals, resulting in differential transcript lengths . The gene expression of α-herpesviruses, such as HSV types 1 and 2, Varicella-zoster virus, bovine herpesvirus type 1 and PRV has been investigated earlier by traditional methods including Northern blot analysis, ribonuclease protection assay, and end-point RT-PCR analysis. However, each of these techniques is associated with a number of disadvantages. For example, Northern blot analysis is labor-intensive, allows only semiquantitative determination of the mRNA level and is unsuitable for multiple mRNA analysis. Moreover, hybridization-based membrane arrays profile changes in a nonlinear fashion, tending to overemphasize large alterations, and they are insensitive to smaller variations. Ribonuclease protection assay-based methods require the use of polyacrylamide gel electrophoresis and typically utilize radioactively labeled probes. The limitations of endpoint PCR technique are the time-consuming procedure, the poor precision and the variable endpoints between samples. Over the past few years, microarray techniques have revolutionized practically all disciplines of molecular biology, including herpesvirus research. As compared with traditional methods, microarray analysis is superior in that it is applicable for the simultaneous analysis of a large number of genes, and even whole genomes. The disadvantage of DNA chip technology is associated with the uncertain quality control: it is impossible to assess the identity of DNA immobilized on any microarray. Further, there are many artifacts associated with image and data analysis. Real-time RT-PCR is an alternative to microarray techniques for the analysis of transcription from multiple genes. The main advantage of real-time PCR is that it is more sensitive to low-fold changes than other high-throughput assays. In addition, the real-time PCR technique provides a reproducible quantitation of DNA copies, and has a large dynamic range, and various controls can be included to ensure accuracy, such as a loading control to verify equal cDNA loading, a no-primer control to prove a measure of non-amplification-related background, a no-template control to screen for contamination of reagents or false amplification, and a no-RT control with confirm the absence of DNA contamination. In contrast to microarray techniques, in real-time PCR the parameters for each gene can be optimized individually. Moreover, the identity of PCR products can be confirmed through melting curve analysis, restriction endonuclease analysis, or DNA sequencing. In fact, real-time PCR is often used to verify gene expression data obtained by microarrays. Nevertheless, considerable pitfalls may be associated with this technique. The major limitations of real-time RT-PCR relative to microarray techniques are the higher cost and labor intensiveness for a large number of samples. Another disadvantage of real-time PCR as compared with blotting techniques is that only the accumulation, but not the size of the transcripts can be monitored. Microarray techniques have recently been applied to investigate herpesvirus gene expression [[17-19] and ]; to analyze the effects of the deletion of particular viral genes or of the specific experimental conditions on whole-genome viral gene expression [[21,22], and ]; and to analyze the impact of virus infection on the expression of cellular genes [[24-26] and ]. The expression of PRV genes has been studied by traditional methods, but many its genes have not yet been characterized at all. Flori and co-workers  investigated the dialog between PRV and epithelial cells, but obtained poor resolution for viral transcripts that did not provide conclusive data on the temporal expression of the PRV genes. To date, Real-time RT-PCR has not been frequently utilized in herpesvirus research for global gene expression analysis. With this technique Oster and Höllsberg  carried out a kinetic analysis of 35 genes of human herpesvirus 6B, a β-herpesvirus, and Dittmer and colleagues  performed a whole-genome profiling of the rhesus monkey rhadinovirus, a γ-herpesvirus. As far as we are aware, no genome-wide expression data obtained by qRT2-PCR have been published so far on α-herpesviruses. In the present study, we describe the development and utilization of a real-time RT-PCR assay for the global analysis of PRV gene expression. We applied a novel method that allows the kinetic characterization of individual PRV genes, and also the comparison of the expression dynamics of different viral genes. Expression profiles were constructed on the basis of relative expression ratios (Rs) calculated as described in the Materials section. This method is applicable for the analysis of gene expression in any genetic system that progressively changes in time.
For each gene, a minimum of 3 independent replicates were carried out for statistical confidence, and the median of these values along with the standard error was calculated. In these experiments, porcine kidney (PK)-15 epithelial cells were infected with PRV with a low multiplicity of infection [MOI; 0.1 plaque-forming units (pfu)/cell]. The reason for this was that our preliminary experiments indicated that infecting cells with a low dose of the virus produced a much better resolution of the cycle threshold (Ct) values for PRV genes at different time points than infecting with a higher MOI (data not shown). However, the use of a low MOI led to a larger proportion of the cells remaining uninfected; accordingly, in order to avoid the initiation of a new infection cycle we chose a relatively short maximal incubation period. Initially, we attempted an 8-h investigation period, but observed a second infection wave of the virus (data not shown), and therefore we restricted our analysis to 6 h post-infection (pi). The transcription of PRV genes was monitored at 5 time points: 0, 1, 2, 4 and 6 h. Prior to PRV infection, cells were either untreated or treated with cycloheximide (CHX), a protein synthesis blocker, or phosphonoacetic acid (PAA), an inhibitor of DNA synthesis. We used strand-specific primers for the reverse transcription, one of the reasons for this being that we intended to exclude the distorting effects of potential overlapping antisense RNAs transcribed from the antiparallel DNA strands, which cannot be distinguished by oligo-dT- or random priming-based RT (this issue will be described elsewhere). Another reason was that it produced a much higher amount of specific cDNAs than oligo-dT priming (data not shown), which was especially important under our experimental conditions with a low MOI. We utilized the mathematical model for relative quantification described by Soong et al. ; however, we calculated the R values in a different way: we used the average of the maximal ECt-sample values for each gene as a control, which was normalized with the average of the corresponding 28S values (ECt-reference). Thus, in our calculation, a particular R value indicates the ratio of the mRNA level of a gene at a given time point to the maximal RNA level of this gene (maximal gene expression occurred at 6 h pi for all genes except us3 and llt2, which peaked at 4 h and at 2 h pi, respectively). The relative amounts of the transcripts of different genes cannot be compared because the primer efficiencies may vary both in RT and in PCR. Even so, use of the maximal expression as the control value allows a comparison of the expression dynamics of viral genes both at individual time points and across the entire infection period. The specificity of the PCR products was confirmed by quality control experiments, including Tm analysis and PAGE (formation of a single product with the appropriate molecular weight was accepted); if some doubts remained, the amplification products were subjected to DNA sequencing or restriction endonuclease analysis using amplicons containing recognition sites for a particular restriction endonuclease [see Additional file 1]. The accuracy of sampling and qRT2-PCR analysis were ensured by using loading controls, as well as no-RT, no-primer and no-template controls (Figure (Figure11).
A major concern regarding the choice of internal reference gene control in viral gene expression studies is the constancy of the reference gene products throughout the entire examination period. We tested whether the 28S ribosomal RNA of the pig fulfilled this requirement, and found that this gene was suitable as reference gene for the study of PRV infection, at least in our 6-h examination period. The 28S rRNA values were highly reproducible: the mean value ± SD of all measurements the experiments was 8.03 ± 0.61 cycles throughout (Figure (Figure22).
Traditionally, lytic herpesvirus genes are classified into distinct kinetic groups on the basis of the effects of protein and DNA synthesis inhibition on the gene expression. In general, inhibitory drugs are applied to infected cells for a prolonged incubation period (24, 48 or 72 h; [[31,32] and ]. Instead, similarly to Stingley et al , we used shorter incubation periods (2, 4 and 6 h for CHX, and 4 and 6 h for PAA) and compared the inhibitory effects of these drugs by calculating the R value ratios for the treated and untreated samples.
To test the requirement of de novo protein synthesis for PRV mRNA production, PK-15 cells were untreated or treated with 100 μg/ml CHX prior to the infection of cells with the virus. At the indicated time points, RNA was isolated and converted to cDNA, which was subsequently analyzed by qRT2-PCR. The degree of inhibition [1-Ri-CHX) × 100] was found to range between 97.3 and 100% for all but one protein encoding PRV gene (the ie180 gene) and two antisense transcripts [the long-latency transcript-1 (llt1), antiparallel to ep0, and the long-latency transcript-2 (llt2), antiparallel to ie180). IE180 mRNA displayed a significantly increased level of expression in the CHX-treated samples at the analyzed 3 time points: 2.27-fold at 2 h; 5.55-fold at 4 h; 1.4-fold at 6 h pi (Figure (Figure3).3). We explain this phenomenon in that the IE180 proteins exert an inhibitory effect on their own synthesis (upon the binding of their own promoters; ), which is resolved by CHX blocking protein synthesis from the IE180 mRNAs. The other exception for the negative CHX effect is llt2: a 1.72-fold increase at 2 h and a 3.983-fold increase at 4 h. Interestingly, at 6 h pi llt2 is significantly repressed by CHX. The repression of llt1 expression at 2 h pi is relatively low (Ri-CHX-2h = 0.281), but this antisense transcript is significantly blocked by CHX at 4 and 6 h pi. Overall, our CHX analysis indicated that the only true IE protein-encoding gene of PRV is ie180. While the icp27 and icp0 genes in the HSV are IE genes, their PRV homologs ul54  and ep0  genes were earlier described as E genes, which was confirmed by our CHX analysis. The icp22 gene of HSV is expressed with IE kinetics; our analysis revealed that its homologous counterpart, the us1 gene of PRV, is significantly blocked by CHX, and hence it is not an IE gene. As a result of CHX treatment, LLT2 exhibits elevated levels at 2 and 4 h pi, which indicates that IE180 transcription factor exerts a negative effect on its expression. Thus, LLT2 appears to be an IE transcript. However, llt2 displays fairly low relative expression at 1 h pi [see Additional file 2], and therefore it cannot be regarded as an IE gene beyond doubt.
For examination of the dependence of the PRV genes on DNA replication, PK-15 cells were infected with the virus in the absence or presence of 400 μg/ml PAA, an inhibitor of DNA polymerase. It was expected that PAA would exert a more drastic effect on the L genes because the expressions of these genes are highly dependent on DNA replication. However, PAA must affect the E gene expression, too. The reason for this is that the gene expression is dependent not only on the promoter activity, but also on the copy numbers of the genes, which are higher after DNA replication than in the initial phase of infection when the PRV DNA is represented in a single copy in a cell (at least in our system). The inhibitory effect of PAA on gene expression was calculated by using the ratio of the R values for the PAA-treated/untreated samples at 6 h pi (Ri-PAA = R6h-PAA/R6h-UT): a low ratio indicates a strong inhibitory effect and vice versa. The PRV genes were ranked from the least inhibited (highest Ri-PAA) to the most inhibited (lowest Ri-PAA). We expected that such a ranking of the PRV genes on the basis of the inhibitory effect would provide natural clusters of E and L genes. Indeed, the results obtained conformed fairly well to the data published earlier on HSV and PRV genes, though there is a certain degree of disagreement between the data published by the different authors on the HSV genes, and only incomplete expression data are available for the PRV genes. Moreover, in many cases a PRV gene is characterized without comparative analysis, e.g. "it appears 3 h pi". For comparison of the HSV and PRV genes, we used five main sources for the HSV data: microarray data published by Wagner [35,36]; various expression data compiled by Roizman [37,38]; and the data collected by Mettenleiter . For the PRV genes, we used the data compiled by Mettenleiter . We classified the analyzed but uncategorized PRV genes via the following principles: if the mRNA was detected between 1 and 2 h pi,, between 3 and 4 h pi, or 5 h pi, the particular gene was classified as E, E/L or L, respectively. In the upper part of the ranking list (Table (Table1,1, Figure Figure4),4), [see Additional file 3], we mostly find typical E genes, with typical L genes in the bottom part of the list. However, we could not draw a clear-cut line between the E and L genes; instead, we put the 15 genes with intermediate values in the E/L group. The precise locations of borderlines between kinetic classes were drawn arbitrarily, because we found smooth transitions between the Ri-PAA values. We chose a typical E gene (ul52) of the HSV as the last member of the E group, and a typical L gene (ul32) of the HSV as the first member of the L group. The 3 protein-encoding genes with the highest Ri-PAA values were ul30 (0.851), ul23 (0.752) and us3 (0.739). These high Ri-PAA values indicate that PAA does not exert a significant cytotoxic effect on the gene expression apart from blocking DNA replication. The 2 LLT transcripts display a surprising response to PAA treatment: the level of LLT1 increases to 2.94-fold at 4 h, and drops to 0.007-fold at 6 h pi relative to the untreated sample; while the level of LLT2 increases markedly close to 40-fold at 4 h, and 3-fold at 6 h pi (this phenomenon will be discussed later). The genes with the lowest Ri-PAA values are all but one (us1: 0.042) L genes in the HSV: ul16 (0.000), ul1 (0.041) and ul38 (0.042). The strong inhibitory effect of PAA on the us1 expression is surprising because it is an IE gene in the HSV. The ie180 gene expression is also significantly inhibited by PAA (0.211); however, we found that ie180 and to a certain extent us1 were expressed in an "irregular" manner in other analyses, too (see later). ul11 and ul21, which are L genes in HSV, do not appear to belong in the E kinetic class. Both genes were characterized as L genes in the PRV by De Wind et al.  using Northern blot analysis. Furthermore, ul13 is an L gene in the HSV, but its transcripts were detected at 2 h pi in the PRV. The ul2 and ul5 genes are E genes in the HSV, but were characterized as E/l and L genes, respectively, in the PRV by Dean and Cheung [55,60], respectively, which is in agreement with our results. We classified the kinetically still uncharacterized PRV genes as follows; E genes: us3, ul29 and ul30, L genes: ul17, ul18, ul27, ul32, ul33, ul35, ul37 and ul41. Peaking of the us3 gene expression at 4 h can be explained by the important role of US3 protein kinase in blocking the apoptotic pathway of the compromised cell in the E phase of infection . Our kinetic classifications accorded more strongly with those published on the HSV than with those on the PRV. The reason may be that the expression profiling of PRV genes was performed by others with low resolution techniques. Our data allow the following conclusions: (1) there is no sharp boundary between the E and L genes; (2) classification of the PRV genes on the basis of the Ri-PAA values yielded similar results those for the homologous HSV genes. A noteworthy exception is the us1 gene, which according to our PAA analysis is expressed in L kinetics; (3) the ie180 and llt genes cannot be classified on the basis of the Ri-PAA data alone, because they display unique expression kinetics; (4) LLT1 and LLT2 respond differently to CHX and PAA treatment, indicating that these antisense transcripts are, at least partially, under different regulation. We assume that the putative antisense promoter (ASP; ) controls the expression of LLT2.
Real-time RT-PCR is a high-resolution technique, which allows a detailed analysis of gene expression dynamics and a reliable temporal classification of viral mRNAs without drug treatment. The data from these experiments are presented in detail in the text [see Additional file 2 and 4]. The us3 PRV gene peaks at 4 h, llt2 at 2 h, while the other genes reach their maximum at 6 h within the 0-6 h examination period.
The accurate and reproducible data generated by real-time PCR allow the analysis of gene expression throughout the entire time course of infection. In order to classify genes on the basis of their expression kinetics, we performed Pearson's correlation analysis for each gene pair of the PRV, using the net increase (RΔ) in each time interval for the comparison [see Additional file 4b]. Genes were clustered into the same group if their pairwise correlation coefficients were high, ranging from r ≥ 0.9 to 0.99 depending on the group. Surprisingly, genes belonging in a particular group very rarely give similarly high values as any of the genes belonging in other groups. Albeit Pearson's coefficient expresses the kinetic properties of a gene in single numbers, the expression curves of genes belonging in the same group proved to be very similar (Figures (Figures5,5, ,6,6, ,7,7, ,8,8, ,9,9, ,10,10, ,11,11, ,12,12, ,13,13, ,14,14, ,15).15). We distinguished 10 gene sets. Group L1 of PRV genes (each pair shared r ≥ 0.989 values) contains 24 genes, of which 22 were identified as L and two as E/L genes by our PAA analysis. The only exception was ul13 which although an L gene in the HSV, behaved as an E gene in all of our analyses. The profile curves of the genes clustered on the basis of Pearson's correlation of pairwise RΔ values, gave very similar patterns of expression within the groups. In each period of time, the expression curves can also be characterized by the formula: RΔ(t+1) - RΔt (Table (Table2).2). For example, group L1 was characterized by hardly any change in the first 2 h, a moderate rise in mRNA level between 2 and 4 h, and a sudden increase in mRNA copy number from 4 to 6 h (0/+/++). Group L2 (r ≥ 0.99) contains one E/L and 6 L genes. The expression profiles of these genes are characterized by a continuous intensive increase from 2 to 6 h (0/+/+). Our PAA analysis revealed that all of the 6 members of group L3 (r ≥ 0.99) are L genes, except for us1, which is an IE gene in the HSV. In our system, therefore, most of the data (except the relatively high R2h for us1), suggest that it is an L gene. This group is characterized by the expression profile 0+/0-/+++. Group E1 (r ≥ 0.91; with 5 members) and group E2 (r ≥ 0.901, with 5 members) comprise E genes, except for ul53, which is an L gene in the HSV. However, this gene is characterized as an E gene in the PRV , which was confirmed by our PAA analysis. These genes are characterized by the expression profile: 0+/+/-. Group E3 (r = 0,964, 2 members: ul11 and ul9), expression profile: +/+/0+; Group E4 (r = 0.996, 2 members: ul12 and us4), profile: 0/+/0. Group M1 (r ≥ 0.933) is composed of 8 members, shown by our PAA analysis to exhibit either E or L kinetics. The expression profiles of these genes are very similar (0/+/++), which draws attention to the fact that PAA analysis alone is not sufficient for the grouping of genes with a similar expression profile. The 3 members of group M2 exhibit a very high correlation (r ≥ 0.998) and similar expression profiles (0/++/+). Group LLT (r > 0.946) contains 2 llt genes that have an unusual expression profile (++/--/+++). The following 4 genes cannot be clustered on the basis of Pearson's coefficients: ie180 (--/++/++), ul30 (--/--/++), ul34 (+/+/+) and us3 (+/++/---), which are likewise unusual Indeed, these genes have peculiar expression profiles (Figure (Figure16).16). The expression profile can also be visualized by heatmap presentation, using the expression data generated with the RΔtvalues (Figure (Figure1717).
We utilized our data to analyze whether the expression properties of PRV genes localized in adjacent genomic loci display similarities.
Figure Figure1818 shows an intriguing relationship between genes categorized on the basis of the data of the PAA experiments. Nested 3' co-terminal genes were found to belong in the same kinetic classes, indicating the existence of shared regulatory mechanisms of gene expression. An exception is the late ul27 gene, which forms a cluster overlapping in parallel with the ul28 and ul29 genes expressed with E kinetics. Of the 11 divergent pairs, 6 genes were found to belong in the same and 5 in different kinetic classes. A further special feature of the PRV genome is that the convergent genes mostly belong in different kinetic classes (Figure (Figure18).18). The potential role of such organization of the herpesvirus genome is discussed later. It may be noted that most US genes belong in the E/L kinetic class accounting for 37.5% of this group.
Analysis of the gene clusters generated by using Pearson's correlation coefficients revealed that genes within the same clusters are often located in neighboring genomic positions. Group L1 is composed of 24 members. The adjacent ul1, ul2, ul3, ul5 and ul6 genes are members of this group. The high pairwise correlation coefficients and the similar profile curves of the RΔ values of these genes cannot be explained by the nested co-localization alone, because ul5 and ul6 are divergent genes and do not overlap with ul1, ul2 and ul3. The ul44 gene is convergent with ul24 and ul26; and ul22 is divergent with ul20. Moreover, ul27 is convergent with ul47 and ul48. It should e borne in mind that ul26 and ul27, an adjacent gene pair in the HSV, are relocalized into a remote position in the PRV, due to a large inversion, but retain their expression properties. This is also true for ul44, ul47, and ul48 (there is no ul45 gene in the PRV). Several genes (us2, us6, us8 and us9) in the US region of the PRV genome belong in this group, too. The ul17 and ul19 genes, which are oriented parallelly, but non-overlapping, likewise belong in group L1. The ul34 and ul35 genes are in a divergent position with ul37. Group L2 is composed of 7 members. The ul32 and ul33 genes are divergently oriented, while ul4 and ul3.5 are convergent genes. Group L3 is a small group with 6 members. The ul36 and ul38 genes are arranged in a divergent orientation. Group E2: Two of the 4 members of this group, ul28 and ul29, are overlapping genes. Group E3: Two (ul11 and ul12) of the 4 members of this group are divergently oriented with the third member, ul9. Group M1: ul7 and ul8 are a convergent gene pair. The ul49.5 and ul51 are L genes oriented in parallel and separated by the divergent early ul50 gene. Furthermore, ul15 and ul17 are a convergent adjacent gene pair. Genes with similar expression profiles appear to be under a common regulation, which cannot be explained solely by the nested localization (these genes also have common promoters) of genes. Group LLT: the llt1 and llt2 expression patterns show a high degree of similarity. However, there is noteworthy inverse relationship in the expression profiles of LLT1 and LLT2 with those of their complementary partners, EP0 and IE180 mRNAs, respectively (Figure (Figure1919).
Gene expressions of α-herpesviruses have already been analyzed by various techniques including Northern-blot and microarray analysis. Different authors have often assigned the same genes to different kinetic groups, which may be explained by the poor resolution of the applied techniques. Only partial data are available on the expression of PRV genes. Flori et al.  carried out genome-wide gene expression analysis via a microarray technique, but this approach did not produce evaluable data on the kinetic properties of the PRV genes. We employed a novel qRT2-PCR technique using (1) strand-specific primers for the RT to obtain a higher yield and to eliminate the effects of the potential antisense transcripts; and (2) the average maximal ECt values as the controls for the calculation of relative expression ratios (Rs). Other approaches, such as the models of Pfaffl's  and Souazé et al., , and the ΔCt and ΔΔCt models  are simpler than the model we used; however, they all neglect the variation in the amplification efficiencies between cDNA samples, which was present in our system. Traditionally, herpesvirus genes are grouped into kinetic classes on the basis of gene expression inhibitory effects of protein and DNA synthesis-blocking reagents. The IE genes should not be significantly inhibited by any of the above substances, the E genes are inhibited only by DNA synthesis blockers, and the expression of the L genes is substantially inhibited by both protein and DNA synthesis blockade. We grouped the PRV genes into IE, E, E/L and L categories on the basis of the results of CHX and PAA treatment (Table (Table1),1), [see Additional file 3]. Our findings conformed well with those published for the HSV genes. The expression of the ie180 gene is enhanced by CHX treatment. The explanation of this phenomenon lies in the fact that the ie180 protein normally inhibits its own expression a short time after the onset of virus infection, a feature which is absent in CHX-treated cells due to novel protein synthesis being blocked by the drug. The other PRV genes were inhibited to a significant extent by CHX at every examined time point. Consequently, ie180 is the only IE gene of the PRV. The ep0 and ul54 genes (IE genes in the HSV) were shown to be E genes by others [ and ], and this was confirmed by our analyses. Interestingly, us1, which is an IE gene in the HSV, appeared to be an L gene in most of our analyses: its expression was significantly inhibited by PAA; it afforded a low amount of mRNAs at 1 h pi, and a high rate of increase of transcript level at 4-6 h pi, all of these being L characteristics. Furthermore, together with a set of L genes, us1 belongs in group L2 according to the clustering of the genes on the basis of pairwise Pearson correlation coefficients. The moderately high relative amount of us1 transcripts at 2 h pi is the only E-like characteristic of this gene. However, us1 displays a high Ct value, suggesting a low absolute amount of transcripts at 2 h pi. We may mention here that the Ct value of a transcript is dependent on the primer efficiency; these data therefore cannot be reliably used to the estimate the absolute transcript amount. If the us1 gene has an important function early in the lytic cycle of the virus, it might fulfill this without de novo synthesis by being released upon infection from the tegument layer of the virus where it might be incorporated. Our data do not demonstrate a clear-cut demarcation line between the groups of E and L genes, suggesting that the categorization of herpesvirus genes on the basis of the requirements for DNA replication might be arbitrary. The kinetic types of genes identified by our PAA analysis resemble homologous HSV genes more than the published PRV genes, which can be explained by the semiquantitative methods used for the kinetic analysis of the PRV genes. The present study revealed that temporal classification of herpesvirus genes is also possible through study of the gene expression of untreated cells in consequence of the high sensitivity of the real-time RT-PCR technique. As a further interesting point, we detected mRNAs by qRT2-PCR in all but 3 PRV genes as early as 1 h pi [see Additional file 4]. This observation indicates that PRV genes are either leaky or subject to regulatory mechanisms which have not yet been elucidated. We found that E genes are generally expressed in a higher proportion than L genes at 1, 2 and 4 h, as indicated by their higher R values at these time points [see Additional file 4a]. The E gene products proved to display a higher net increase (RΔ) than the L genes within the 0-1 h, 1-2 h and 2-4 h pi periods. However, by the 4-6 h pi interval, this trend had definitely reversed, i.e. the L genes exhibited high R(6h-4h) values (Additional file 4b). Unexpectedly, the ratio of increase (Ra) between 1 and 2 h was higher for a number of L genes than any of the E genes [see Additional file 4c]. Since the R2h values of these L genes were lower than those of the E genes, the high R(2h/1h) value is a result of the very low R1h values of late genes. Most of the L genes also have higher R(4h/2h) and R(6h/4h) values than those of E genes, which is explained by the fact that the E genes reach a relatively high expression level by 1 h pi. The classification of the PRV genes through the CHX and PAA analyses led to results very similar those generated by the analysis of untreated cells via the following calculation: R(6h-4h)/R(1h-0h) [see Additional file 5]. The differences between the results of the two approaches can be explained in that PAA analysis alone gives only a rough picture of the gene expression; a detailed kinetic analysis of the viral gene expression furnishes a more sophisticated picture. Some genes classified as E or L by PAA analysis display "irregular" kinetics under untreated conditions. The ie180 and perhaps us1 genes cannot be classified by PAA analysis alone. Analysis of the transcription kinetics of overlapping transcript sets is an important issue. It is not known whether downstream genes on polycistronic mRNAs are translated, and if so, to what extent. This makes interpretation of the mRNA expression data difficult because the mRNA levels cannot be correlated with the amounts of the corresponding proteins, which are the workhorse molecules in regulating cellular physiology. In principle, a downstream gene is translated if it is transcribed from its own promoter. Theoretically, downstream genes could also be translated from a polycistronic RNA if alternative splicing removed upstream gene(s) from the pre-mRNA, or if potential IRES-like sequences helped recruit ribosomes, thereby initiating a cap-independent translation from the downstream gene (we have found no data in the literature concerning these possibilities). Thus, analysis of the downstream genes without discriminating between transcription from their own promoters or by read-through from upstream genes might result in a false categorization. The PRV genes were also classified by analysis of the gene expression throughout the entire examination period by using Pearson's correlation analysis. Genes with similar expression profiles (high pairwise r values) were placed in the same group. Genes in the same kinetic class (identified by PAA analysis) are generally situated in the same group generated by the use of Person's correlation. Moreover, members of a particular group (with high r values) rarely display high correlation coefficients with genes belonging in different groups. The PRV genes conformed well to 10 patterns of expression. Only 4 viral genes (ie180, ul30, ul34and us3) did not fit into any of the 10 groups. We found an inverse expression profile of the sense/antisense partners in the ep0-llt1 and ie180-llt2 genes (Figure (Figure19).19). Combination of the untreated data with the results of the CHX and PAA analyses revealed an interesting relationship for coordinated regulation of sense/antisense partners. The CHX analysis suggested that IE180 protein facilitates LAP (latency-associated transcript promoter) activity, and inhibits ASP (putative antisense promoter activity). The PAA treatment resulted in a significantly elevated antisense transcript level at 3 of the 4 time points in LLTs, which indicates the existence of another regulatory level besides IE180 protein action. We assume that transcription from one DNA strand negatively influences the expression of transcripts from the complementary DNA strand. The interaction can occur at the level of transcription (RNA polymerase moving in one direction along one of the DNA strands inhibits RNA polymerase moving in another direction) and/or translation by forming double-stranded RNAs by the sense and antisense transcripts. As an example, PAA has a negative effect on the transcription of ie180 (the level of IE180 mRNA is reduced to a quarter) at 4 h pi, which results in a lower rate of transcription from ie180 (thereby facilitating the expression of LLT2); and a lower amount of inhibitory IE180 proteins, which also facilitates LLT2 expression. Overall, the LLT2 level increases 39.4-fold relative to the untreated conditions. Furthermore, genes with the same kinetic properties exhibit a distinctive distribution pattern along the PRV genome. Nested genes appear to belong in the same kinetic group. Additionally, convergent genes and gene clusters in most (3 out of 11) cases belong to different kinetic classes. It should be remembered that the PAA-based approach and the time intervals applied for kinetic analysis produce only a coarse resolution of gene expression. It is possible that convergent genes allocated to the same kinetic classes might display different expression profiles on a finer scale. The above genome organization principles may point to the existence of as yet unknown regulatory mechanisms. It may be speculated whether a read-through of transcription across convergent genes is the basis of this regulation. Moreover, groups formed in terms of high Pearson's correlation coefficients contain many genes localized at adjacent loci on the PRV genome. Most genes with high correlation coefficients are not nested genes. This means that the similarity of their expression profiles cannot be explained by the control of a common promoter. Further, several genes in the same group are separated by genes that display different kinetic profiles. These results suggest the existence of a genetic mechanism that synchronizes gene expression on a higher-order (chromatin?) scale.
Although the relative amounts of the transcripts between genes cannot be compared because primer efficiencies may vary both in RT and in PCR, our method allows a comparison of genes on the basis of their expression dynamics. We detected the expression of almost all genes from the earliest time point (1 h), Further, we have found that a sharp boundary cannot be drawn a between the groups of E and L genes in terms of expression profiles. We have also found that the PRV has only a single IE gene (the ie180 gene) and that the us1 gene is expressed in L kinetics. Our kinetic studies revealed that PAA analysis alone is not sufficient for the classification of genes on the basis of their expression profiles. The analysis of gene expression throughout the 6 h examination period demonstrated that these genes can be subdivided into further clusters, and that genes showing significant differences in response to PAA treatment can have very similar expression characteristics in untreated conditions. In our analysis, nested genes displayed the same expression profiles, whereas convergent genes mostly exhibited different kinetic properties. It should be noted that a finer resolution of the gene expression might expose slight differences in the expressions of nested genes. Approximately half of the divergent genes belong in the same class, and the other half of them in different kinetic classes. The new calculation technique that we have developed is also applicable to evaluate the loss-of-function phenotypes of mutant herpesviruses viruses and, in principle, for the analysis of gene expression in every temporally changing genetic system.
Monolayer cultures of immortalized porcine kidney 15 (PK-15) cells were maintained at 37°C in an atmosphere of 95% air, 5% CO2 in Dulbecco's modified Eagle medium (DMEM) supplemented with 5% fetal bovine serum (Gibco) and 80 μg gentamicin per ml. Strain Kaplan of pseudorabies virus (PRV-Ka) was used for gene expression analysis. The virus stock was prepared by infecting PK-15 cells with 10 plaque forming units (pfu)/cell PRV-Ka followed by incubation of the cells until a complete cytopathic effect was observed. For the analysis of the PRV transcriptome, rapidly-growing semi-confluent PK-15 cells were infected with a low multiplicity of infection (MOI; 0.1 pfu/cell) of the virus, and incubated for 1 h, after which the virus suspension was removed and the cells were washed with phosphate-buffered saline (PBS). Subsequently, new culture medium was added to the cells, which were further cultivated for 0, 1, 2, 4 or 6 h. Cells were incubated in the presence or absence of 100 μg/ml cycloheximide (CHX), a translation inhibitor, or 400 μg/ml phosphonoacetic acid (PAA), an inhibitor of DNA synthesis (both purchased from Sigma-Aldrich) 1 h prior to virus infection. Mock-infected cells, treated in the same way as infected cells, were used as controls.
Viral DNA was used to test the efficiency and specificity of the primers applied in real-time PCR. The viral DNA was purified as follows. Monolayers of PK-15 cells were infected with the PRV at an MOI of 10, and cultivated at 37°C until a complete cytopathic effect was observed. Subsequently, culture medium was collected without disrupting the cells and clarified by centrifugation at 4,000 rpm for 10 min using a Sorvall GS-3 rotor. Next, the virus in the supernatant fluids was sedimented on a 30% sucrose cushion by ultracentrifugation at 24,000 rpm for 1 h using a Sorvall AH-628 rotor. The sedimented virus was resuspended in sodium Tris-EDTA buffer. After the addition of proteinase-K (l00 μg/ml final concentration) and sodium dodecyl sulfate (SDS; 0.5% final concentration), the lysate was incubated at 37°C for 1 h, which was followed by phenol-chloroform extraction and dialysis.
Primers were designed through use of the Primer Express (Applied Biosystems) and FastPCR Professional (Primer Digital Ltd.) oligonucleotide design software according to the given guidelines. Primer pairs were designed from the 3'-end regions of ORFs for each gene (Table (Table3).3). The ul26 and ul26.5, and orf1 and orf1.2 genes contain overlapping ORFs; and we therefore, did not employ distinct primers for them. Primer specificity was verified by BLAST searches of the GenBank database (National Center for Biotechnology Information (NCBI) website http://www.ncbi.nlm.nih.gov/BLAST/. All primers were purchased from Bio Basic Inc. (Mississauga, Ontario, Canada).
A conventional PCR technique was used to check on the potential DNA contamination and the quality and specificity of the primers. The PCR reaction was carried out by standard methods on a Veriti™ 96-Well Thermal Cycler (Applied Biosystems) with the following parameters: 1 cycle of 94°C for 4 min; 30 cycles of 94°C for 1 min, 60°C for 1 min and 72°C for 2 min; and 1 cycle of 72°C for 7 min. The PCR reaction was carried out by using the GC Rich PCR System (Roche). The complete DNA sequence of the PRV is a composite of 6 different strains , but the variation between them is very small (around 1%). Most of the sequence data (86.7%) have been derived from strain Kaplan (PRV-Ka), which we used in our experiments. To circumvent the problem of genetic background differences, we tested 2 or more primers for RT and PCR reactions for non-Ka sequences and selected those that performed best; alternatively, we sequenced the particular DNA region and designed new primers on the basis of the sequence data. For each gene, we designed and tested various primers, and selected those, which did not produce primer dimers or other nonspecific products. For those genes where we could not eliminate the problem of primer dimer formation, in every cycle we applied an extra extension step with an elevated temperature (below the Tm of the specific product and well above the Tm of the primer dimers) for the detection.
PK-15 cells (5 × 106 cells per flask) were washed in PBS and harvested for RNA isolation at 0, 1, 2, 4 and 6 h pi. For quantitative RT2-PCR, total RNA was extracted from the cells with the NucleoSpin RNA II Kit (Macherey-Nagel GmbH and Co. KG) according to the manufacturer's instructions. Briefly, cells were collected by low-speed centrifugation, lysed in a buffer containing the chaotropic ions needed for the inactivation of RNases and providing the conditions for binding of nucleic acids to a silica membrane. Next, contaminating DNA was removed with RNase-free rDNase solution (included in the NucleoSpin RNA II Kit). Subsequently, the RNA solution was treated with Turbo DNase (Ambion Inc.) to remove potential residual DNA contamination. Finally, RNA was eluted in RNase-Free Water (supplied with the kit) in a total volume of 60 μl. RNA concentration was determined in triplicate by spectrophotometric analysis of the absorbance at 260 nm with a BioPhotometer Plus (Eppendorf). The RNA solution was stored at -80°C until use.
A two-step quantitative real-time RT-PCR was carried out for the transcriptional analysis. For each gene, a minimum of 3 independent replicates (separate infections) were performed.
Reverse transcription (RT) Total RNA isolated from infected cell cultures subjected to drug treatment (CHX or PAA) was reverse-transcribed into cDNA for PCR analysis. RTs were performed in 5 μl of solution containing 0.07 μg of total RNA, 2 pmol of the gene-specific primer, 0.25 μl of dNTP mix (10 μM final concentration), 1 μl of 5× First-Strand Buffer, 0.25 μl (50 units/μl) of SuperScript III Reverse Transcriptase (Invitrogen) and 1 U of RNAsin (Applied Biosystems Inc.) and the mixture was incubated at 55°C for 60 min. The reaction was stopped by raising the temperature to 70°C for 15 min. No-RT control reactions (RT reactions without Superscript III enzyme) were run to test the potential viral DNA contamination by conventional PCR. RNA samples with no detectable DNA contamination were used for quantitative RT-PCR reactions. First-strand cDNAs were diluted 10-fold with DEPC-treated water (Ambion Inc.), then subjected to real-time PCR analysis.
Real-time quantitative PCR experiments were performed with a Rotor-Gene 6000 cycler (Corbett Life Science). All reactions were carried out in 20-μl reaction mixtures containing 7 μl of cDNAs, 10 μl of ABsolute QPCR SYBR Green Mix (Thermo Fisher Scientific), 1.5 μl of forward and 1.5 μl of reverse primers (10 μM each). The running conditions were as follows: (1) 15 min at 95°C, followed by 30 cycles of 94°C for 25 sec (denaturation), 60°C for 25 s (annealing), and 72°C for 6 s (extension). The absence of nonspecific products or primer dimers was indicated by observation of a single melting peak in melting curve analysis. An additional extension and detection step was applied for those primers that produced primer dimmers: for 2 s at a temperature just below the Tm of the specific product and substantially above the Tm of the primer dimers. With this technique we could eliminate nonspecific fluorescent signals produced by primer dimers. Following the PCR reaction, melting curve analysis was performed to control amplification specificity (specificity was defined as the production of a single peak at the predicted temperature and the absence of primer dimers) by measuring the fluorescence intensity across the temperature interval from 55°C to 95°C. The 28S ribosomal (r)RNA used as the loading control (reference gene) was amplified in each run. H2O was included as a no-template control, and cDNA derived from the reverse-transcribed RNAs of non-infected cells was used as a negative mock-infected control. We applied SYBR Green-based real-time PCR because of the lower costs and simpler protocol than for TaqMan probe-based methods for instance. It has recently been demonstrated that the SYBR-based method of detection is as sensitive and specific, and has a similar dynamic range to that of the TaqMan-based technique .
We calculated the R value by using the following equation:
where R is the relative expression (quantification) ratio; E is the efficiency of amplification; Ct is the cycle threshold value; sample refers to any particular gene at a given time point; and ref is the 28S rRNA, which was used as a reference gene and was amplified in each run. Average Ct values with their standard error (SE) values and amplification efficiencies with SE are shown in Additional file 6. This equation is similar to that used by Soong et al . However, instead of individual values, we used the average maximal value of ECt for each gene as the control. The relative copy numbers of mRNAs were calculated by normalizing cDNAs to 28S rRNA using the Comparative Quantitation module of the Rotor-Gene 6000 software (Version 1.7.28, Corbett Research), which automatically calculates the real-time PCR efficiency sample-by-sample. Thresholds were set automatically by the software.
Data were analyzed by the Microsoft Excel program, using the average and the standard deviance functions. The inhibitory effect of CHX or PAA was calculated via the ratio of the drug-treated and untreated R values at 2, 4 and 6 h pi for CHX: Ri-CHX = RCHX/RUT, or 4 and 6 h pi for PAA: Ri-PAA = RPAA/RUT. Thus, a low value indicates a high inhibitory effect and vice versa. The net increase in a product was calculated by subtracting the R value at time point t+1 from that at t (RΔ = R(t+1)-Rt), where t = 0, 1, 2, 4 or 6 h and (t+1) = 1, 2, 4 or 6 h. The ratios of adjacent R values (rate of change; Ra) were calculated with the following equation: Ra = R(t+1)/Rt; t = 1, 2, 4 or 6 h; (t+1) = 2, 4 or 6 h. Pearson's correlation analysis was used to evaluate qRT2-PCR data, as an alternative method for the grouping of PRV genes into kinetic classes. Pearson's correlation coefficient (r) was calculated as follows.
A correlation is a number between -1 and +1 that measures the degree of association between two variables [labeled here as X and Y, which are the RΔ values of two different genes in the same time interval (i)]. and are the average values, n is the sample number, and SX and SY are the standard deviances (errors) for X and Y, respectively. A positive value for the correlation implies a positive association and a negative value implies a negative or inverse association. Genes were clustered by using a complete linkage hierarchical clustering method with a centered correlation similarity metric with Cluster 3.0 software (Stanford University). To view the clustering results generated by Cluster 3.0, we used Alok Saldanha's Java TreeView.
Larger DNA fragments generated by conventional PCR were run on 1% agarose/TBE gels containing ethidium bromide and visualized under UV illumination, using Marker 16 (Lambda DNA/Eco130I; Fermentas) to size DNA fragments. Smaller DNA fragments generated by qRT2-PCR were run in a 12% polyacrylamide gel to ensure that the amplified products had the correct size. DNA fragments were visualized with ethidium bromide staining under UV illumination. A GeneRuler™ Low Range DNA Ladder (Fermentas) was included in each run.
If there were doubts concerning the identity of the amplified product, restriction endonuclease analysis was performed to confirm the specificity of the qRT2-PCR products.
Plasmids containing PRV DNA fragments were constructed for DNA sequencing analysis, which was performed if primers designed on the basis of sequence data relating to non-Ka strains performed badly; or if the identity of amplicons generated by real-time PCR was uncertain. PRV DNAs were subcloned by using two methods: PCR amplification of the particular DNA segment; or subcloning of the desired DNA region by standard molecular cloning protocols. Amplified products were subcloned to the pGEM (Promega) vector in accordance with the manufacturer's instructions. PRV BamHI fragments to be sequenced were subcloned to the pRL525 vector .
Subcloned DNA fragments were subjected to DNA sequencing with the ABI Prism™ 3730xl DNA sequencer (AME Bioscience Ltd.). DNA sequences were analyzed by using the Chromas Lite 2.01 software (Technelysium Pty Ltd).
The authors declare that they have no competing interests.
DT carried out the standard and real-time PCR, the agarose and polyacrylamide gel electrophoresis, and the DNA sequencing, and participated in the evaluation of the primary data. JT took part by performing the reverse transcription reactions, purified PRV RNA, and propagated PK-15 cells. PP participated in performing the reverse transcription reactions. ZB coordinated the study, propagated viruses and isolated viral DNAs. All authors have read and approved the final manuscript.
Validation of qRT2-PCR products by restriction endonuclease analysis. In a few cases, the specificity of PCR products was confirmed by restriction endonuclease assay. Polyacrylamide gel electrophoresis of qRT2-PCR products, using primers ul54 (lanes 1 and 2) and us4 (lanes 3 and 4). Amplicon specificity was confirmed by restriction endonuclease analysis, using SalI (lane 2) and HpaII (lane 4) enzymes. The GeneRuler™ Low Range DNA Ladder is shown on the left.
Grouping PRV genes according to their expression in untreated cells. Data of R values of untreated samples.
PRV genes ranked on the basis of the effect of PAA on gene expression. PRV genes were classified on the basis of the inhibitory effect of PAA (Ri-PAA) on gene expression. Early genes are separated by E/L genes from late genes. The ie180 gene was classified as IE gene because its expression was not inhibited by CHX treatment. a R values of PAA-treated samples at 4 h post-infection. b Relative expression ratios of PRV genes after PAA-treatment at 6 h post-infection. c Ri-PAA values at 4 h post-infection. d PRV genes ranked on their Ri-PAA (6 h pi) values. e Ri-PAA values of PRV genes after 6 hours post-infection.
PRV genes ranked on the basis of their expression profiles. a. R values Genes were ranked on the basis of their R values at different time points. a PRV genes ranked on the basis of R values at 1 h pi. b Relative expression ratios of PRV genes at 1 h pi. c Order of R values of PRV genes at 2 h pi. d R values after 2 hours post infection. e PRV genes ranked on the basis of R values at 4 h pi. f R values at 4 h pi. b. RΔ values Genes were ranked on the basis of their RΔ values at different time intervals. a Order of RΔ values of PRV genes in the interval 1 h-2 h. b RΔ values in the interval between 1 and 2 h. c PRV genes ranked on the basis of their RΔ values in the infection period from 2 to 4 hours. d RΔ values in the interval 2 h-4 h. e PRV genes ranked on their basis of their RΔ values in the interval 4 h-6 h. f RΔ values in the infection period: 4 h-6 h. c. Ra values Genes were ranked on the basis of their Ra values at different time intervals. a Order of Ra values of PRV genes in the interval 1 h-2 h. b Ra values in the interval between 1 and 2 hours. c PRV genes ranked on the basis of their Ra values in the infection period from 2 to 4 hours. d Ra values in the infection period from 2 to 4 hours. e PRV genes ranked on their basis of their Ra values in the interval 4 h-6 h. f Ra values in the interval 4 h-6 h.
PRV genes ranked on the basis of their R(6h-4h)/R1h ratios. R(6h-4h)/R1h values.
Average cycle threshold (a) and amplification values (b) with standard errors.a Average of cycle threshold (Ct) values based on almost three separated reactions. b Standard error of the mean.
This study was supported by Hungarian National Fund for Human Frontiers Science Program Young Investigator grant (No. RGY0073/2006) to Z.B.