|Home | About | Journals | Submit | Contact Us | Français|
In this review, we compare four assays that are currently used to measure HIV integration and discuss their strengths and weaknesses. We then outline advances that have been made toward development of a more robust, more sensitive, quantitative HIV integration assay suitable for clinical use. The assay that we have developed uses repetitive-sampling Alu-gag PCR. The detailed protocol describes our assay step-by-step, the creation of an integration standard cell line and accompanying standard curve, as well as the quantitation of integration and calculation of associated error estimates. Finally, we speculate on fundamental, unresolved issues in HIV latency that can be addressed by measuring HIV integration.
Integration is a central step in the HIV life cycle and is defined as the insertion of the HIV reverse transcript into the host cellular DNA. Integration is required for efficient spreading infection (1–9) and so is an important step to measure. Measurements of integration in vitro continue to enhance our understanding of basic retrovirology and how cells restrict HIV integration (10–13). Measuring integration in vivo has also enhanced our understanding of latent HIV infection as it has been demonstrated that resting CD4+ T cells contain HIV, but fail to produce infectious virus unless stimulated (14). Thus, such latently infected cells are resistant to antiretroviral therapy (15, 16).
With the advent of more sensitive assays for HIV integration, it may be useful to measure the level of integrated HIV DNA in various cellular subsets within HIV infected individuals especially in combination with other viral intermediates (17, 18). Because of the limitations of prior HIV integration assays, there is very little data on the level of HIV integration in various CD4+ T subsets (19) and less information regarding the level of integration in CD4+ non-T cell subsets. Monitoring integrated DNA within subsets over time could demonstrate, for example, if HIV is cleared from short lived CD4+ non-T cells (20) in the presence of antiretroviral therapy. By combining measurements of total DNA and integrated DNA (18), it may even be possible to indirectly determine the relative level of replication within different subsets. One recent study provides evidence that the half-life of integrated HIV DNA is greater than the half-life of unintegrated HIV DNA (18). This would suggest that cells with higher levels of total HIV DNA relative to integrated HIV DNA experienced more rounds of replication. Thus, this approach might reveal if HIV replication persists within specific cellular subsets for patients on HAART.
There are three main hurdles to measuring integration well. One is distinguishing integrated from unintegrated DNA since not all reverse transcripts integrate (14). Thus, one cannot measure total DNA as a surrogate marker for integrated DNA. Measuring integration in vivo presents two additional hurdles. These challenges include enhancing sensitivity since the level of integration is low in vivo (14) and detecting all variants of the integrated population since it is known that HIV has a relatively high mutation rate (21–24).
Apart from the method that we use to measure integration, Alu-gag PCR (25–29), three other methods have been used to assay integration: gel separation, inverse PCR, and linker ligation PCR. In the first method, integration in HIV infected individuals is measured by first using gel separation methods to segregate genomic DNA from episomal DNA (18, 30) and then measuring the amount of HIV DNA within the genomic DNA by routine quantitative real-time (kinetic) PCR. However, the separation method is too laborious for large numbers of clinical samples and it is unclear how effectively this method separates episomal DNA from genomic DNA. The two other methods, inverse PCR (14, 31, 32) and linker ligation PCR (33), are conceptually similar to Alu-gag PCR, the method we prefer. They distinguish integrated from unintegrated DNA by designing primers that only allow exponential amplification of HIV DNA when it is integrated. These methods tend to use two PCR steps and approach quantitation via endpoint dilution analysis. In all three methods, inverse, linker ligation, and Alu-gag PCR, one primer is HIV-specific and binds to both integrated and unintegrated DNA. By design, the second primer only binds to the same target DNA template in the correct orientation if HIV has integrated. In this way, only integrated HIV is exponentially amplified. In all three methods, primer extension (or linear amplification) of unintegrated HIV DNA occurs. In all three techniques, only amplicons that contain HIV DNA sequences are detected.
Exponential amplification of human chromosomal DNA is a limitation in both linker ligation and Alu-gag PCR as it results in limiting substrate. Thus, it limits the sensitivity of the assays. However, the amplified human chromosomal DNA that lacks HIV DNA is not detected. While inverse PCR avoids amplification of chromosomal DNA, it is significantly more labor intensive than the other PCR approaches, making it less attractive for widespread clinical use.
Detecting all HIV variants is one of the biggest challenges when measuring integration in vivo. Inverse and linker ligation PCR are most affected by this problem. In both assays, a restriction enzyme digestion step is included. Given the relatively high mutation rate, the restriction enzymes used in these two methods will at some frequency fail to digest some of the HIV sequences in the DNA sample at the predicted site(s). These mutated sequences would often be undetected (or under certain circumstances detected at a much lower efficiency). For example, we found that approximately 5% of the sequences in the Los Alamos database (34) were mutated in a commonly used restriction enzyme site (unpublished observation).
Applications of inverse PCR (14, 31, 35) and Alu-gag PCR (25) to the measurement of integration in HIV-infected individuals led to groundbreaking discoveries and provided the first demonstration that integrated DNA existed in resting CD4+ T cells in infected individuals (14, 25, 31). However, these methods were laborious and had weaknesses that prevented their widespread application to clinical samples. For example, end point PCR is a semiquantitative technique and thus is not amenable to robust quantitation. Both methods lacked a rigorous integration standard. In addition, the Alu method failed to account for events lying too far away from an Alu site to be detected. Neither method provided rigorous background controls to account for unintegrated DNA.
Given the above limitations, several groups tried to improve the quantitation of integration by adapting the endpoint Alu-gag based PCR assay approach as first described by Chun et al (14, 25, 31, 36). These modified assays (26, 27, 29, 37) added a kinetic PCR step, and provided an integration standard as well as a background gag only control to enhance quantitation and control for false positives. Nonetheless, these attempts to improve the original assays resulted in inferior sensitivity.
We recently developed an Alu-gag PCR based assay for HIV integration that has the sensitivity of the endpoint PCR assay, but also provides robust quantitation. We chose to develop the Alu-gag PCR method because it is the least labor intensive, least susceptible to lack of detection (i.e., does not require digestion with a restriction endonuclease), and thus the most amenable to adaptation for clinical use. We overcame the sensitivity limit by incorporating a repetitive sampling step (38) and by additional modifications that enhance sensitivity by forgoing a large dynamic range (39). Repetitive sampling also provides a means to calculate confidence intervals (39). Thus, our new assay should provide a measurement that will allow hypothesis testing to determine, for example, if a treatment regimen reduces the level of integrated DNA within a cellular subset.
Using this approach, we recently showed that patients on HAART have lower levels of integration than patients off HAART (39). This suggests that monitoring patients over time might show that integration is decreased with HAART. In addition, we have demonstrated routine detection of as few as ~0.5 provirus in 10,000 cellular genomes.
Assays to measure HIV integration have improved in a stepwise fashion over several years. Below we present the assay we currently perform. We also mention future directions that could lead to further improvements in assay sensitivity.
Our two-step, Alu-gag PCR assay for detection and quantitation of integrated HIV DNA is depicted schematically in Figure 1. After isolating total DNA from HIV-infected cells, the first PCR is performed with one primer that anneals to Alu, and the other that anneals to gag. Alu is a repeat element in the human genome that occurs approximately every 5,000 base pairs (40, 41). Gag is an HIV gene that encodes for the structural components of the virion particle (42). The second PCR detects HIV-specific products by using primers to the R and U5 regions within the HIV long terminal repeat (LTR).
In the case of integrated HIV DNA (Fig. 1A), the Alu primer serves as an anchor in the human genome and the gag primer serves as an anchor in the HIV genome. Binding sites for these two primers will be present in the DNA target template only when HIV has integrated into the human genome. Furthermore, when the primer binding sites are close enough and aligned correctly, the region between them can be amplified exponentially. The first PCR is followed by a second, HIV-specific, real-time or kinetic PCR step to quantify the level of Alu-gag amplicons produced in the first round of PCR. Quantification is achieved by comparing the resultant signals to those obtained with an integration standard (IS).
In contrast to integrated DNA, which is selectively and exponentially amplified in this PCR strategy, unintegrated HIV DNA, whether linear or circular, is only linearly amplified (Fig. 1B). This is because, although the Alu and gag primers bind to their cognate sites, the primers are not bound to the same DNA template. In this way, integrated HIV DNA (Fig. 1A) is always preferentially amplified in comparison to unintegrated HIV DNA (panel B). This, in turn, results in a stronger signal from Alu-gag amplicons than gag-only amplicons in the second PCR step. Not explicitly shown in the schematic is a control PCR performed with only the gag primer added. This serves as a background measurement and approximates the signal expected from unintegrated HIV DNA.
Finally, human genomic DNA is not detected by the second PCR step (Fig. 1C). Although the Alu primers bind to the same DNA target template resulting in exponential amplification, these Alu-Alu amplicons are not detected by the second PCR step because it is HIV-specific.
HIV integration occurs at numerous positions throughout the host cell genome. It is therefore essential to prepare the assay standard that best mimics the natural polyclonality of integration during infection. Using our assay, when integration occurs nearby an Alu site, amplification occurs more efficiently than when HIV integrates further away from an Alu site. With a polyclonal standard we are able to compensate for this heterogeneity in our proviral estimates. To prepare the IS, we used a retroviral construct (kind gift of Bob Siliciano) that is engineered to only allow a single round of infection (11). In this engineered virus, HIV env is deleted and replaced with a cassette that encodes for green fluorescent protein (GFP) followed by an IRES and hygromycin resistance gene. Since this construct only allows a single round of infection, over time we expect that unintegrated DNA will be diluted to undetectable levels while integrated HIV DNA will remain stably associated with the cell as it doubles.
We performed several steps to validate our IS sample. In agreement with our expectations, we demonstrated that, by one week post-infection, only integrated DNA could be detected by probing a southern blot of the IS with HIV specific sequences, which showed that HIV DNA co-localized with chromosomal DNA after gel electrophoresis of undigested genomic DNA (not shown, but described in (37)). Before preparing the IS DNA, we purified GFP positive cells by fluorescence-activated cell sorting and detected approximately ~1 copy of HIV DNA/cell using single round HIV-specific kinetic PCR. We also determined that the integration insertion sites in the prepared IS sample were, as predicted, polyclonal by performing a southern blot on the Alu-gag amplicons with an HIV-specific probe. This demonstrated a range of amplicon lengths, reflecting the range of integration site distances to Alu elements within the infected cell population. These last three steps are not necessary, but we performed them to validate our assay conceptually.
Below is a list of the primers and probes used in this protocol. We designed HIV-specific primers using the Los Alamos database (34). We chose to use a standard Alu primer that was described in the literature.(45–47). We scanned the entire gag gene, found the most conserved region, and designed our gag primer to this region. We took the same approach to determine the primers for the R and U5 regions by looking at the entire LTR to find the most conserved area, then designing our nested PCR primers and probe to bind these regions. We included two degenerate probes to account for the two most common mutations in the probe site, which were unlinked. The number of sequences analyzed and the percent identity are described in (39). Also listed below are the primers that we use in the β-globin assay for determining the concentration of human genomes in the sample DNA.
To date, the sequences in the Los Alamos database (34) for the LTR region of the HIV genome are somewhat limited. However, with the advent of pyrosequencing, we expect that more sequences will soon be available. As more sequences are added to the database, it may be necessary to further refine the primers and probes for this step.
We prepare a large volume of PCR master mixture once a year to minimize systematic variation. The master mixtures contain all the reagents required for PCR except the Platinum Taq enzyme. The master mixture should be aliquoted and stored at −20°C for up to 12 months. At the time of PCR, the master mixture is mixed with the DNA sample and the Platinum Taq enzyme and the PCR protocol is run with this mixture. Before using the new master mixture, we always compare it to an old master mixture at several dilutions of our IS, to check for consistent amplification efficiency between master mixtures. We add the Platinum Taq enzyme immediately before performing PCR in order to optimally preserve enzyme activity and store it separately at −20°C until use. Specifically, we add 1 part Taq Polymerase (5 U/μL) for the β-globin assay, and 2 parts Taq Polymerase for the others (Alu-gag, gag only, and RU5).
The following recipes make 2x master mixtures. When we set up the reaction, we add equal amounts of sample and master mixture to the wells, creating a 1x master mixture during PCR. Specifically, for RU5 we add 10 μL sample and 10 μL master mixture to each well (total reaction volume = 20 μL). For the other reactions (β-globin, Alu-gag, and gag only) we add 25 μL sample and 25 μL master mixture to each well (total reaction volume = 50 μL).
Add together the following:
Add together the following:
Add together the following:
Add together the following:
We use Platinum Taq DNA Polymerase (Invitrogen Life Technologies). Since different enzyme lots can affect the efficiency of PCR, we prescreen small samples of several lots. We determine the lot with the best activity and then buy sufficient quantity for one year of reactions.
We use the QIAamp DNA Micro Kit as described in the QIAamp DNA Micro Handbook (“Isolation of Genomic DNA from Small Volumes of Blood”), except that we dilute our final sample in 10 mM Tris-HCl pH 8.0 rather than Buffer AE.
To estimate the number of proviruses per cell, first determine the concentration of human genomes in the isolated sample DNA. For this, our lab uses a β-globin-based PCR assay, described below. Alternatively, OD260/280 can be used.
Make a standard curve by running the nested PCR protocol on several dilutions of the IS DNA sample. We use 40 replicates of both Alu-gag and gag-only per dilution, for 8 different dilutions ranging from 1.25 provirus/well to 160 provirus/well (dilute in uninfected PBMC DNA). To make a standard curve, proceed with the following steps:
The level of HIV integration in the different samples is calculated using the standard curve. First, perform the nested PCR protocol described above and obtain Ct values. Then, follow these steps:
The assay described here was specifically designed for use with samples derived from HIV-infected patients. However, it can easily be adapted for experiments performed in basic virology research laboratories, for example, to determine the level of HIV integration in tissue culture cells infected ex vivo. The biggest change is that the PCR primers should be chosen based on the particular virus isolate, rather than by consulting the Los Alamos Database (34) for conserved regions of HIV DNA. With new primers, a new standard curve will need to be generated. However, the rest of the procedure and analysis is the same.
The dynamic range of this assay can be expanded by reducing the number of cycles in the first PCR round from 40 to 20 (38). This will result in less sensitivity, but allows measurement of integration at higher copy numbers. We demonstrated using previously published primers specific for the NL4-3 strain of HIV-1 (38) that the dynamic range of our assay is ~1 to 10,000 proviruses when 20 cycles of Alu-gag PCR are performed in the first reaction.
As the assay is currently described, with 40 Alu-gag repeats, the sensitivity limit with patient samples is ~0.5 proviruses among 10,000 genomes (39). The sensitivity limit is determined by the point when the average Alu-gag signal is no longer statistically different from the gag-only signal. The enhancement in sensitivity is proportional to the number of experimental replicates analyzed in each experiment.
It should be possible to enhance the sensitivity of our assay further by simply performing more repeats. In other words, the sensitivity of the assay is only limited by the number of repeats we are willing to perform given that we detect integration approximately 10% of the time (38, 39). However, to detect integration below 0.5 proviruses in 10,000 genomes, it is no longer feasible to correlate the average Alu-gag Ct with proviral copy number because the Alu-gag and gag only signals on average no longer statistically differ. Nonetheless, individual Alu-gag signals are statistically different than the gag only signal.
We are in the process of testing a new mathematical correlation and statistical tools to allow us to detect integration at levels below 0.5 in 10,000 genomes. In this method, we record the percent positive samples and use the binomial distribution to estimate the errors. Positive samples are determined by showing that they differ by greater than ~2 standard deviations from the gag only signal. Using this approach, we have detected as few as 1 provirus in 100,000 cells (unpublished observations). The downside to this approach is the number of samples required. For example, to detect one provirus in 100,000 genomes, we performed 200 Alu-gag PCRs to obtain 5 positive signals.
We present a method for measuring HIV integration that is more accurate, more sensitive, and more suitable for clinical use than previous assays. HIV integration levels in patients’ PBMCs have only been monitored in a few studies longitudinally (18, 48, 49), and no study has monitored the level of integrated DNA in different CD4+ cell subsets over time. Furthermore, the longitudinal studies that were performed monitoring integration in PBMCs have provided conflicting results regarding the effect of HAART on integrated DNA levels. This may be related to potentially wide and uncertain errors inherent in the assays utilized. With improved methods to measure integration that provide precise confidence intervals, it will be possible to monitor more precisely the level of integration in patients to conclusively determine if it is affected by therapy. The use of the repetitive sampling technique improves the Alu-gag PCR assay by both enhancing sensitivity and permitting calculation of confidence intervals. Thus, this method should provide a means to robustly measure integration and perform rigorous hypothesis testing for the first time within HIV-infected individuals on different therapeutic regimens. Furthermore, these methods may provide new information about the level and the half-life of integrated DNA in CD4+ T and non T cell subsets in the presence of therapy. By combining these measurements with measures of total DNA, important information regarding the potential for ongoing HIV replication within different subsets may be obtained. This in turn should provide a deeper understanding of the differential effect of antiretroviral therapy on different cell subsets.
We would like to thank Troy Brady, Bruce Levine, Liz Colston, Wei Yang and Luis Agosto for their assistance in preparation of this manuscript. This study was supported in part by NIH grant R01 AI058862-01 (U.O.).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.