KSHV MicroRNA Sequence Analysis and KS Risk in an European AIDS-KS Case Control Study

Abstract

Background

We recently identified polymorphisms in KSHV encoded microRNA (miRNA) sequences from clinical subjects. Here, we examine whether any of these may contribute to KS risk in a European AIDS-KS case control study.

Methods

KSHV viral load in peripheral blood was determined by real-time quantitative PCR. Samples that had detectable viral loads were used to amplify the 2.8 kbp microRNA encoding region plus a 646 bp fragment of the K12/T0.7 gene. Additionally, we characterized an 840 bp fragment of the K1 gene to determine KSHV subtypes.

Results

KSHV viral DNA was detected in PBMC of 49.6% cases and 6.8% controls and viral loads tended to be higher in cases. Sequences from the miRNA encoding regions were conserved overall but distinct polymorphisms were detected some of which occurred in pri-miRNAs, pre-miRNAs or mature miRNAs.

Conclusions

Patients with Kaposi’s sarcoma were more likely to have detectable viral loads than controls without disease. Despite high conservation in KSHV miRNA encoded sequences, polymorphisms were observed including some that have been previously reported. Some polymorphisms could affect mature miRNA processing and appear to be associated with KS risk.

Introduction

Kaposi’s sarcoma associated herpesvirus (KSHV), also known as human herpesvirus 8 (HHV-8), is the cause of Kaposi’s sarcoma (KS) [1]. KS has been categorized into four epidemiological types including the classical form affecting older men of Mediterranean descent, an aggressive endemic KS identified in sub-Saharan Africa, iatrogenic KS resulting from immunosuppression regiments following organ transplantation, and epidemic or AIDS-associated KS. AIDS-KS represents the vast majority of KS cases worldwide and much remains to be understood regarding the risk factors for KS in KSHV and HIV co-infected subjects. The introduction of HAART therapy in developed countries has lead to a dramatic decrease in KS incidence among HIV infected subjects [2, 3]. However, KS still occurs at significantly increased risk in HIV infected subjects and rates of KS have been stable in HIV infected populations during the HAART era. Recent reports of KS occurring in patients with high CD4 count and low or undetectable HIV viral loads emphasize the need for further insights into the correlates of KS in this population [4, 5].

KSHV has an atypical worldwide prevalence and genotypic distribution. Most studies have failed to show any association between viral genotype and disease risk [6–8]. A recent report in Italy has suggested that KSHV genotype A predisposes patients with classical KS to a more rapid clinical disease pattern correlated with higher viral loads but the study was limited in size [9].

Recently, the discoveries of microRNAs (miRNA) in large double-stranded DNA viruses, including herpesviruses, have presented another layer of complexity in understanding KSHV pathogenesis [10–12]. MiRNAs are small non-coding RNAs, typically 19–22 nt in length, that post-transcriptionally regulate gene expression by recognizing and binding to specific targets in the 3’UTR of mRNAs resulting in either downregulation or degradation of the target mRNA. KSHV encodes 12 distinct pre-miRNAs resulting in 17 mature miRNAs, all located within the latency associated region that is highly expressed in all KSHV-associated malignancies. A cluster of 10 viral pre-miRNAs are encoded by a 2.8 KB intragenic region located between the viral FLIP and kaposin genes. The remaining two pre-miRNAs are found within the K12 open reading frame [11].

Cellular miRNAs have been linked to many processes including proliferation, differentiation, and apoptosis. Few targets of virally encoded miRNAs are known but it is likely they play key roles in latency, immune evasion, and pathogenesis. We recently showed that the KSHV encoded miRNA region is highly conserved in KSHV cell lines and clinical samples from patients with KS and MCD [13]. However, we observed a distinct cluster of sequences that differed phylogenetically in the miRNA region. Moreover, single nucleotide polymorphisms were observed in some pre-miRNA sequences, one of which has been shown to affect processing of the miRNAs by drosha and consequently affect mature transcript expression levels.

We recently reported a study of risk factors for KS in HIV infected subjects [14]. In this study, we have analyzed KSHV viral load detection in PBMC and examined miRNA sequences from KSHV infected cases and controls in order to determine if any specific polymorphisms correlate with KS risk.

PATIENTS, MATERIALS, AND METHODS

Patient samples

The samples were collected in 1993 through a collaboration of researchers in European countries and funded by the European Union (EURO-SHAKS: European Study on HIV Associated Kaposi’s Sarcoma). The study design included HIV positive, KS positive cases matched with two HIV positive patients who either had low CD4 cell counts (less than 200) or another AIDS-defining illness. Informed consents and detailed questionnaire data was collected from each study participant. Participants residing in Spain were followed bi-yearly until 2004. Analysis of serological data from this study has been previously reported[14]. Peripheral blood samples were collected for each patient and DNA was extracted [15].

Real-time polymerase chain reaction (PCR) KSHV load quantification

A standard input of approximately 200 ng per PCR reaction was used when possible. Samples with lower concentrations were run at the maximum volume input of 10 µl. KSHV viral load was determined using a quantitative real-time PCR assay which specifically targets the K6 gene region [16, 17]. A second real time assay for human endogenous retrovirus 3 (ERV-3), present at two copies per genomic cell, was used to quantify cellular DNA [18]. For each real-time PCR assay, each sample was assayed in triplicate and an average of the three individual reactions was used to determine target copy. KSHV viral load was then calculated as copies per million cells.

Samples were designated qualitative positive if the estimated viral copy average was less than 1 or if the sample failed to amplify in any of the triplicate reactions. Quantitative estimates for these samples were not determined as target amplification occurred at the lower limit of assay sensitivity. Samples that fell into this category were retested to confirm the findings.

Amplification of KSHV miRNA and K1 gene regions

Amplification of each gene region was performed using a nested PCR strategy with Jumpstart Readymix (Sigma) as previously described [19].

Three overlapping PCR amplifications were performed for the miRNA cluster region. The overlapping sequences included a 1,065 bp fragment containing miRNAs 1–5, a 1,014 bp fragment encoding miRNAs 6, 11, and 7; and a 1,266 bp fragment encoding miRNAs 7–9. The nested PCR primers used to amplify the individual fragments are listed in Table 1 and the positions are shown in Figure 1.

Table 1.

Nested PCR primers used to amplify miRNA cluster region

Primer	Primer type	Sequence	Region Amplified
TLP OR	Outer nested	GAATGCGTGCTTCTGTTTGA	MiRNAs 1–5
MAS027F	Outer nested	GACTTGTAGGCGAGGGGAG	MiRNAs 1–5
TLP IR	Inner nested	GGGGAGGAGGAAAAAGTACG	MiRNAs 1–5
MAS036F	Inner nested	CAGAAGTTGACACCAGCCC	MiRNAs 1–5
MAS034F	Outer nested	TAGCAGGGCCATCCACAC	MiRNAs 6, 11, 7
MAS037R	Outer/Inner	CCGCTCATAAGACCATAAC	MiRNAs 6, 11, 7
MAS035F	Inner nested	TTTTGCGCCCCTCTTTGG	MiRNAs 6, 11, 7
TLP OF	Outer nested	CTAGCTCCCCTCCCATCGA	MiRNAs 7–9
MAS036R	Outer nested	GGGCTGGTGTCAACTTCTG	MiRNAs 7–9
TLP IF	Inner nested	TTCCGGAAATACCACCTGAG	MiRNAs 7–9
MAS027R	Inner nested	CTCCCCTCGCCTACAAGTC	MiRNAs 7–9

All primer sequences are based upon GenBank accession no. U75698. OR, outer reverse; F, forward; IR, inner reverse; R, reverse.

Figure 1.

Diagram showing the overlapping nested PCR amplification strategy used to generate the complete sequence of the KSHV microRNA encoding region and the fragment of K12 encoding mirK12-10 and mirK12-12. The primer names, sequence, and locations are consistent with our previous publication [13] using the GenBank reference U75698. The arrows indicate the direction of extension and are shaded in relationship to the fragment amplified. Asterisks indicate outer nested primer sets. All positions are approximate and the features are not to scale. IF, inner forward; IR, inner reverse; OF, outer forward; OR, outer reverse; F, forward; R, reverse. The primer sequences are listed in Table 1.

DNA from KS cases and controls were grouped separately for nested PCR testing. In addition, samples with estimated viral copy per million cell equivalents greater than 1000 were segregated from lower copy samples to reduce the chance of cross contamination. Universal standards were employed to prevent non-specific or cross contamination of samples including using separate rooms for reagent preparation, first and second round assay plate setup, and gel analysis. All equipment, including pipettes and hoods, were cleaned with Clorox and/or UV irradiated before and after each assay. A checkerboard plate setup was used so that no sample was directly beside another either horizontally or vertically. Molecular grade water was used as no template controls and they were included in every row of the 96 well plates used for amplification. Samples were repeated if any control wells showed evidence of non-specific amplification.

Cloning and sequencing of KSHV miRNA and K1 gene regions

The PCR products from the second round nested reactions were electrophoresed on 1.5% agarose gels and bands of the expected size were excised. DNA was extracted using QiaQuick (Qiagen) and 5 µl was used in a confirmatory gel to verify size and as an estimate of DNA concentration. Products with low concentrations were cloned using the pGEM T-Easy vector system (Promega). Otherwise, all products were sequenced directly, using the amplification primers as sequencing primers, by an ABI Prism 3130xl sequence detector system (Applied Biosystems). The sequence information was assembled using data from both forward and reverse overlapping sequences for the K1, T0.7, and three individual miRNA cluster region fragments. The entire miRNA cluster region was assembled using three overlapping sequences as shown in Figure 1 [13].

Phylogenetic analysis

DNA sequences obtained in this study were aligned using CLUSTAL X (version 1.81) with relevant reference sequences, available from the GenBank public database, representing all known subtypes for the genes analyzed. Alignments were made using the nucleotide sequences with the exception of the K1 gene, which was translated into amino acid sequences prior to alignment. The CLUSTAL X files were converted into Mega formats (version 2.1) for neighbor-joining analysis and phylogenetic tree generation using the Kimura 2-parameter evolutionary model.

Statistical Analysis

Viral load DNA distributions were compared using ANOVA. Odds ratios of detecting a SNP were calculated independently for each SNP using Cornfields’ method for computing confidence intervals. Correction for data sparsity and incomplete tables was attempted by adding a small constant to all cells[20]. All analyses were performed using STATA v 10.1 (StataCorp, College Station, TX).

RESULTS

KSHV Viral Load Detection in PBMC

DNA was available from 398 subjects for screening with real-time quantitative PCR to determine viral load. Of these, 157 samples failed to meet the optimum DNA concentration requirement of 200–250 ng per 10 µl. These samples were still included in the study since the primary purpose of the viral load testing was to identify samples for genotyping. The rate of detection of KSHV DNA did not differ significantly between the samples with optimal and suboptimal DNA however, 21.6% vs. 19.9% respectively, and therefore all samples were included in subsequent analyses.

KS cases were more likely to be positive for KSHV DNA (50.7%) than controls (6.8%) consistent with previous reports (OR = 14.08, 95% CI: 7.6–26.8) [21]. KSHV viral load estimates, in subjects who tested positive, also tended to be higher in KS cases, (23–130,159/ 10⁶ cells, geometric mean 321.4/10⁶ cells, n= 64) compared to controls (15–1,600/10⁶ cells, geometric mean 131.3/ 10⁶ cells, n= 16) as shown in Figure 2. Five KS case samples were qualitative positive by PCR. Additionally, 5 samples from KSHV antibody negative subjects (6.2%) were weakly positive by PCR (1–5 raw copies: viral load 42–235/10⁶ cells).

Figure 2.

Box and whisper plot representing log-transformed viral load in KS patients and KSHV seropositive controls. In each box, the upper and lower margin represent the upper and lower quartile, while the midsecting line represents the median. The top and bottom whiskers indicate maximum and minimum valuea, excluding outliers, which are represented by dots. The two distributions were compared by ANOVA.

Phylogenetic Analyses and sub-type determination

All KSHV viral strains analyzed from study subjects were subtype A or C based upon K1 amino acid analysis. K1 subtypes were determined for 19 of 21 cases, of which 13 were designated subtype C (68.4%). Ten of the 13 (70%) control subject K1 sequences were also subtype C. Similarly, for T0.7, most subjects were subtype C (20 of 27, 74%) and sequences were very similar. More variation was observed in the 7 A subtype sequences, 6 of which were from KS cases. Phylogenetic analysis of the microRNAs cluster region resulted in a tree with similar topology to that we previously reported[13]. KSHV sub-type information is shown in table 2. Phylogentic trees are available upon request.

Table 2.

Summary of patient information and subtype analysis.

Sample	Status	Birthplace	Residency	KSHV copies/ 1×106 cells	K1	T0.7 K12	MiRNA region encoding 1–5	MiRNA region encoding 6, 11,7	MiRNA region encoding 7–9
Spn0023	Case	Spain	Spain	1,590	C3	C	C	N/D	C
Bar0029	Case	Barcelona	Barcelona	18,605	A4	A	C’	C’	C
Bar0030	Case	Córdoba, Spain	Barcelona	98,000	N/D	C	N/D	N/D	N/D
Bar0035	Case	Jaen, Spain	Barcelona	15,349	C2	A	A	N/D	N/D
Bar0040	Case	Spain	Spain	545	C2	C	C	C	A
Bar0044	Case	Pelopones, Greece	Barcelona	12,941	A1	C	C	C	C
Bar0045	Case	Badajoz, Spain	Barcelona	23,000	C1	C	A’	C	C
Bar0046	Case	Not reported	Not reported	3,714	C2	A	A	C	A
Bar0049	Case	Ceuta, Spain	Barcelona	1,667	C2	A	A”	A	A’
Spn0057	Case	Spain	Spain	7,875	C1	A	A”	A	B/Q
Bar0080	Case	Barcelona	Barcelona	5,647	C3	C	C	C	C
Spn0084	Case	Spain	Spain	729	C2	C	C	C	A
Bar0102	Case	Barcelona	Barcelona	8,824	C2	C	C	C	A
Spn0103	Case	Spain	Spain	651	N/D	N/D	C	C	C
Bar0110	Case	Lugo, Spain	Barcelona	130,159	A4	C	C	C	C
Bar0112	Case	Argentina	Barcelona	20,588	A4	C	C’	C’	N/D
Spn0115	Case	Spain	Spain	176	C2	C	C	C	C
Spn0116	Case	Spain	Spain	2,435	A1	N/D	C	C	C
Spn0118	Case	Spain	Spain	1,231	C3	C	C	C	C
Spn0451	Case	Belgium	Belgium	941	A1	A	C	N/D	A
Brs0489	Case	Brussels, Belgium	Brussels	8,485	C3	C	N/D	N/D	N/D
Bar0171	Control	Barcelona	Barcelona	1,333	C2	N/D	C	C	A
Bar0193	Control	Huelva, Spain	Barcelona	1,500	C2	N/D	N/D	N/D	N/D
Spn0216	Control	Spain	Spain	89	N/D	C	C	C	A
Mad0222	Control	Madrid	Madrid	1,458	A1	A	A	C	A
Bar0321	Control	Barcelona	Barcelona	462	A1	C	C	C	C
Bar0332	Control	Granada, Spain	Barcelona	1,468	A3	C	C	C	C
Bar0336	Control	Pontevedra, Spain	Barcelona	133	C1	C	C	N/D	C
Spn0340	Control	Spain	Spain	15	N/D	N/D	N/D	C	N/D
Bar0355	Control	Barcelona	Barcelona	1,733	C3	C	C	C	C
Spn0370	Control	Spain	Spain	31	N/D	C	N/D	C	N/D
Bar0376	Control	Salamanca, Spain	Barcelona	1,600	C3	N/D	N/D	N/D	N/D
Spn0379	Control	Spain	Spain	18	N/D	N/D	C	C	N/D
Bar0386	Control	France	Barcelona	129	C2	C	C	C	C
Bar0452	Control	Flemish-Barabant, Belgium	Brussels, Belgium	160	C3	N/D	C	N/D	N/D

Note: N/D, not determined.

Conservation of KSHV-encoded MiRNAs in Cases and Controls

Within the 1,065 bp fragment encoding KSHV-mir-K12-1 to 5, the conservation was 93.7%. The 1,014 bp fragment encoding KSHV mir-K12-6, 11, and 7 was the most conserved (94.3%) while the least conserved (90.8%) was the 1,229 bp fragment encoding KSHV mir-K12-7 to 9. The lower conservation for the mirK12-7 through 9 portion of the cluster region is a consequence of multiple polymorphisms located in the sequence between mir-K12-8 and 9 including the very polymorphic mir-K12-9.

A detailed analysis of single nucleotide polymorphisms detected in the KSHV miRNA cluster region and the K12/T0.7 gene are shown in figures 3a and 3b.

Figure 3.

a Detailed map of SNPs observed in the 2,713 bp region encoding 10 miRNAs. The reference sequence used to determine the nucleotide position is found at genbank accession number U75698. The analysis includes all 29 current study subjects as well as 21 previously reported full length sequences. All SNPs observed in 2 or more subjects are represented including insertions and deletions. The number of sequences in which the SNPs were observed is shown in the right hand columns. The position of the premiRNAs are indicated in green. Only nucleotide positions at which SNPs occur are included in the figure.

b Analysis if SNPs observed in the 646 bp region sequenced from the K12/T0.7gene that encodes mir-K10 and 12. The figure shows data from 27 subjects from the current study as well as 21 previously reported full length sequences.

Detection of Pre-MiRNA Polymorphisms

Consistent with our previous findings[13], polymorphisms within pre-miRNA regions were rare overall with six miRNAs, namely KSHV-mir-K12-1, 6, 8, 10, 11, and 12 showing complete conservation in the current study. Polymorphisms within the seed region, nucleotides 2–8 in the mature miRNA, were even rarer with complete conservation observed in 27 of 29 study subjects. However, one study subject, Bar0049, had three SNPs in the seed region of KSHV mir-K12-9, previously reported in the BCP-1 cell line. An insertion was found in the seed region sequence of KSHV mir-K12-3 from Bar0080, a KS case.

We disregarded polymorphisms present in only a single sequence as they may be attributable to PCR artifacts. Such SNPs were found in KSHV mir-K12-2, 3, 4, and 9. However, it should be noted that many SNPs discovered in single sequences in our initial report have been verified in the current study due to expanding the number of sequences available for analysis.

We have previously reported polymorphisms in KSHV mir-K12-4, 6, 7, 8, and 9 [13]. Many of these polymorphisms were observed in at least one of the sequences in the current study. An additional polymorphism was identified in the mature transcript sequence of KSHV-mir-K12-2 from one KS case and one control. Detailed representation of polymorphisms occurring in the KS case and control samples are shown in figures 3a and 3b. The mapping of single nucleotide polymorphisms, insertions, and deletions in the current study suggests that the changes are more common in KS cases than controls.

Polymorphisms outside pre-MicroRNA Sequences

Recent studies have shown that polymorphisms in pri-miRNA are important determinants of Dicer and Drosha processing affecting mature miRNA generation [22, 23]. We therefore examined the sequence outside the pre-miRNA coding region for additional polymorphisms. We found 52 polymorphisms outside the pre-miRNAs in the cluster region and 15 in the T0.7 amplicon analyzed. Polymorphisms occurring within a few base-pairs of the start of pre-miRNA sequences were noted for mir-K12-3, 5, and 6. The region between KSHV-mir-K12-8 and 9 is highly variable to a degree unexpected for a herpesvirus. Polymorphisms preceding KSHV-mir-K12-1 and 10 and proceeding 9 and 12 are not included in our analysis as they are external to the region sequenced. The distribution of polymorphisms in the miRNA cluster region is illustrated in Figure 3a and Figure 3b illustrates the distribution of polymorphisms in the T0.7 region.

Disease association of polymorphisms

Figure 4 shows the odds ratio of detecting a SNP in KS cases vs controls for both the T0.7 and miRNA cluster region. Since amplification and cloning of the regions was possible in 29 or fewer subjects the standard error for any odds ratio estimate is expected to be large. Thus, the study has a limited power to detect even a large effect size (70% for the largest observed), if type I error is to be set at 5%. In other words, all observations are bound to be not statistically significant at the conventional 0.05 level. Nevertheless, it is possible to observe that polymorphisms in miRNAs 4, 5, 7 and 9 were strikingly more common in cases than in controls. Several others outside of pre-miRNAs, especially in the T0.7 amplicon, were similarly more common in cases, while most were seen only slightly more frequently in cases than controls or at similar rates in the two groups.

Figure 4.

Odds ratio of detecting a SNP in KS cases vs controls. Only SNP present in more than 10% of the study population are shown. Point estimates are shown as red dots and confidence intervals as blue bars. The T0.7 region is indicated in blue and the miRNA cluster is in pink.

DISCUSSION

We have previously reported sequence analysis of the KSHV miRNA encoding regions from PEL cells lines and clinical subjects that demonstrated two major phylogenetic groups consisting of a large number of very conserved A and C sequences and a second smaller group of more diverse sequences designated B/Q [13]. We also previously reported SNPs within the pre-miRNA sequences of several of the KSHV encoded miRNAs, at least one of which has been shown elsewhere to affect miRNA processing and hence expression levels [13, 24]. Studies to determine the effect of other SNPs on miRNA processing and expression are warranted.

Here we report the analysis of sequences of the KSHV miRNA encoding regions in the context of an AIDS-KS case-control study in order to examine the possibility that polymorphisms in miRNA encoding regions may be associated with KS risk. We have shown that polymorphisms in miRNAs 4 and 5 are more common in KS cases than controls, although this could not reach statistical significance due to the limited sample size of our study, discussed below. These findings are highly suggestive of an important role for KSHV encoded miRNAs in KS pathogenesis and that some viruses may have evolved to be less pathogenic by modulating the expression of the microRNAs they encode. In addition we have, for the first time, identified polymorphisms in the KSHV encoded pri-miRNA sequences. Since polymorphisms in pri-miRNA sequences have been shown to affect miRNA processing and maturation [22, 23], we suspect that at least some of these may have effects on the expression levels of KSHV encoded miRNAs in infected cells, which is in concordance with the fact that some of them also occur more frequently in cases than in controls.

Several polymorphisms observed in the T0.7 region were more common in KS cases than in controls. However, the K12 gene has also been reported to be a target for cellular immune responses [25]. Two of the non synonymous polymorphisms we observed (nt 119955 and nt 118037) were within previously reported CTL epitopes [26, 27]. This raises the intriguing possibility that CTL escape may drive changes in the miRNA encoding sequences affecting miRNA expression that may result in changes in pathogenesis. The increasingly significant preponderance of polymorphisms within KS cases suggests that the changes observed may actually serve to promote KSHV pathogenesis and the development of cancer. It was recently reported that mir-K12-7 is involved in the induction IL-6 and IL-10 secretion by macrophages and monocytes, modulating immune responses to KSHV tumor progression [28]. Also, recent findings indicate mi-K12-9 targets ORF50, which functions as the switch between latency and lytic viral replication [29]. We had initially hypothesized that microRNAs that were highly conserved would be more likely to play critical roles in the KSHV life cycle, however, these recent reports are more suggestive of an important role for the microRNAs in which SNPs occur, increasing the significance of our findings.

Previous studies have attempted to find an association with KSHV sequence variation and KS risk but failed to do so, finding instead that sequence variation, especially in the K1 gene is associated with ethnicity and geographical variations. [6–8]. Consequently, the SNPs we report here may represent the first “virulence factor” to be discovered for KSHV.

Our study has some limitations. The study design was cross sectional and subjects were recruited during the pre-HAART era. Consequently, we cannot be sure that some subjects included as controls may not have developed KS at a later date or would have done so if they had not succumbed to another AIDS associated illness. DNA was not available from all subjects or was limited which affected our ability to amplify the gene regions of interest for sequencing. In addition, the viral load tended to be higher in cases than in controls, therefore amplification of the large amplicons we used in this study was more successful in cases. Obviously, we were only able to examine sequences from subjects with detectable viral DNA, which limited the sample size and consequently the power of the study. Nevertheless, we feel our study has added substantially to the available body of evidence on the possible pathogenic role of SNP related to KSHV miRNA. In particular, our study points to individual SNPs for which in vitro studies may be warranted to elucidate possible effects on miRNA activity.

In conclusion, we have extended our previous observations of variation in the miRNA encoding regions of KSHV in clinical subjects by examining sequences from subjects of an AIDS KS case control study. We have reported for the first time SNPs in the pri-miRNA sequences of KSHV that may have additional effects on miRNA processing and expression. Sequence differences observed in this and our previous studies may have consequences for disease risk and pathogenesis and we suggest this is an area which merits further study.