Early intervention the hiv

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.


Early intervention before the acute phase of HIV infection is recommended to reduce viral dissemination and its harmful effects on immune function in infected individuals. Preservation of HIV-specific CD4+ and broad CD8+ virus specific immune responses appear to play pivotal roles in controlling viral replication. Individuals treated with HAART before HIV-1 seroconversion show stronger HIV-1 specific T helper cell responses and reduced HIV-1 diversification than individuals treated with HAART after seroconversion [1]-[2]. Therefore, development of a diagnostic tool to detect HIV infection before seroconversion may help improve clinical outcomes of patients by allowing initiation of effective therapy before seroconversion.

Rapid ongoing advances in gene chip technology not only offers a powerful approach to provide biomarkers for diagnosis and prognosis of diseases as evidenced in treatment of cancers [3]-[11], it can also offer insights into the molecular mechanisms that may operate in the progression of an infectious disease through using an integrated approach of data mining, text mining and sophisticated multivariate statistical and clustering tools to identify the key molecular players in HIV infection [12]-[13].

Here, we show that gene expression changes that occur in CD4+ T-cells within 48 hrs of exposure to HIV-1 may be useful as novel diagnostic biomarkers of HIV infection.



The laboratory HIV-1 strains HTLVIIIB (wildtype RT, NRTI-sensitive, NNRTI-sensitive) HIV RTMDR-1/MT-2 (catalog no 252; NRTI-resistant and nonnucleoside analog RT inhibitor-resistant laboratory strain of HIV), A17 and A17V (NNRTI-resistant as well as a primary clinical HIV-1 isolates, 92BR019 (catalog no 1778; envelope subtype B), were obtained from the AIDS Research and Reference Reagent Program, NIAID.


MT2 cells were grown in RPMI 1640 containing 15% fetal bovine serum. 50 X 106 MT2 cells were infected at a multiplicity of infection of 0.01 with one of five different HIV viral strains (HIV-1RT-MDR, HIV-1HTLVIIIB, HIV-192BR019, HIV-1A17, HIV-1A17V). Also a negative control (lacking virus) was included for each time point. The cells were incubated with virus or media (negative control) for 1 hour at 37o C. After incubation the cells were diluted in 50 ml of 15% media and transferred to a T75 tissue culture flask. The flasks were incubated in a CO2 incubator for 24 hrs and 48 hrs. After the appropriate incubation the cells were lysed in TriPure (Boehringer Mannheim, Indianapolis, IN).


Total RNA was prepared from cells infected with virus using TriPure isolation reagent (Boehringer Mannheim), DNA was removed from this RNA preparation by DNase treatment (RQ1 RNase-Free DNase, Promega) and phenol chloroform extraction. Further purification of the RNA was achieved by binding to an RNeasy column prior to in-vitro transcription reaction. Total RNA was quantified using standard spectrophotometric methods. First strand cDNA was synthesized with a T7-(dT)24 primer starting with 8 mg of total RNA . Second strand cDNA was synthesized with E coli DNA polymerase I and ligase. The cDNA product was in-vitro transcribed and labeled with the Enzo BioArray High Yield RNA transcript labeling kit. The in vitro transcription products were purified on an RNeasy spin column prior to fragmentation in buffer containing 40mM Tris-Acetate (pH 8.1), 100mM KOAc and 30 mM MgOAC for 35 minutes at 94oC. Biotinylated fragmented cDNA was hybridized to an Affymetrix U95Av2 array for 15h prior to wash and staining, as per manufacturer's recommendations. Arrays were scanned with a confocal scanner manufactured for Affymetrix by Agilent.


We measured levels of RNA transcripts using an U95Av2 GeneChip microarrays from Affymetrix which interrogates the expression level of 12,625 genes. The chips are widely used for a variety of experimental applications and spotted on each chip are a robust series of controls to minimize chip to chip variation. On this type of an array each human gene is represented by at least one probe set composed of multiple probe pairs (16-20 pairs). Each probe pair consists of two sets of 25mer sequences. One set is a perfect match (PM) and the other set has a 13th base mismatch (MM) to serve as an internal control for the signal produced by the perfect match probe. A quantitative Signal metric was used to measure the level of each transcript on the chip (developed by Affymetrix). The algorithm calculates the signal using a one-step Tukey's Biweight Estimate which determines a weighted mean that is relatively insensitive to outliers. The estimated real signal is calculated by subtracting the log of the Perfect Match intensity from the stray signal estimate. The mismatch signal is used to estimate the stray signal where appropriate. The probe pair is weighted more strongly to calculate the Signal if the signal is closer to the median value.

To extract useful biological significance of gene expression changes, sources of biological and experimental variation have to be adequately accounted for and considered in the statistical design. Analysis of variance techniques provide an efficient and powerful way to analyze designs in which there are multiple experimental treatments and complex factorial structures. We used a one-way ANOVA technique to identify differentially expressed genes using a virus factor (Genesight 3.0) to screen for the most significantly affected genes.

ANOVA groups for the virus effect (performed on signal values):

  1. No virus control (mean of 8 treatments (4X24 and 4X48 hr timepoints))
  2. HIV-1RT-MDR (mean of 4 (2X24 and 2X48 hr time points))
  3. HIV-1A17 (mean of 4 (2X24 and 2X48 hr time points)).
  4. HIV-1A17V (mean of 4 (2X24 and 2X48 hr time points))
  5. HIV-1HTLVIIIB (mean of 2 (1X24 and 1X48 hr time points))
  6. HIV-192BR019 (mean of 2 (1X24 and 1X48 hr time points))

We used a two-way agglomerative hierarchical clustering technique to organize expression patterns using the average distance linkage method such that genes (columns) having similar expression across HIV-1 subtypes (rows), and similar expression of HIV-1 sub-types across each gene were grouped together. Dendrograms were drawn to illustrate similar gene expression profiles from joining pairs of closely related gene expression profiles, whereby genes joined by short branch lengths showed most similarity in expression values. For each row (HIV-subtype), we compared the mean and standard error of the HIV-subtype with that of control to calculate the T-value. Heat maps depicted the gene expression such that positive and negative t-values represented an increase and decrease in expression relative to control with HIV-1 infection for each gene.

Typically informative genes are chosen for classification by ranking genes according to a test statistic and then choosing a number of the highest ranking genes. The problem with this approach is that many of the genes will be correlated with each other hence redundant information is included in the classifier. We used Principal Component Analysis (PCA) to transform a number of correlated gene variables (76 genes) into a smaller number of uncorrelated variables called principal components composed of a linear combination of the 76 gene variables. The first principal component accounted for 54% of the variation performed on the correlation matrix between the genes (10 principal components accounted for 88% of the variation, 16 components accounted for 96% and 23 components accounted for 100% of the variation).

Several principal component variables were included in the multivariate model to discriminate between treatment groups (5 viral, 1 no-virus) groups. A matrix of total variances and covariances was compared using multivariate F tests in order to determined whether or not there are any significant differences (with regard to all variables) between groups. Canonical Analysis was performed to determine the combination of the original variables (PCA components and Treatment groups) that explains the largest possible variation. The first two canonical variables were plotted to visualize the separation between the groups. The Mahalanobis distance (metric that takes into account both the multivariate mean and variance/co-variances) of each sample from the centroid of each group was calculated to determine group membership of each sample expressed as a probability value).

A discriminant function (canonical variables to define groups) can easily be built using many variables to define class membership. Leave-one-out-cross validation (LOOCV) procedure was used to determine the generalized error rate for classification and to test for overfitting of the data. In this method one sample was left out as a test sample and the other 23 samples (training set) were used to build discriminant function to assign group memberships. The misclassification error was calculated both for the training set and the test sample. The procedure was repeated for all the samples to obtain the overall prediction error rate. Classification prediction error was calculated using 8 to 16 principal component variables in the model. Too few (not enough variation explained) or too many variables (loss of degrees of freedom and inclusion of redundant information) in the model results in high misclassification rates.

The LOOCV procedure resulted in the minimum number of principal components required for the lowest misclassification rate. To better understand the gene loading on the principal components, a varimax method was employed (JMP Software) to make sharper distinctions between the factors; this transformation of principal components approximated a simple structure using an orthogonal rotation criterion by maximizing the variance of the squared elements in the columns of a factor matrix. The correlations between gene variables and each factor were examined to identify co-regulated genes that were also loaded on to common factors. In the common factor model, gene variables are linear combinations of common factors and unique factors, in which we determined factors that explain the most variance (factors that distinguish between the viral sub types) and the genes that are highly loaded on these factors. Factor loadings were expressed as correlation coefficient of the gene versus the factor. All calculations for the cluster and factor analyses were performed using JMP software (SAS, Cary, NC).

The U.C.S.C. Genome Bioinformatics web site (http://genome.ucsc.edu/) was used to obtain 1000 bp of upstream genomic sequence for the gene list in Table 8. Promoter sequences were isolated for each group of genes with more than one member and analyzed on TRES [Transcription Regulatory Element Search (http://bioportal.bic.nus.edu.sg/tres/)] to identify common transcription factor binding sites.

1 Ejarr-an50


The goal of the present study was to determine whether changes in gene expression profiles of MT-2 cells that occur following HIV-1 exposure could be used to detect cellular infection with drug-resistant HIV-1 strains. We expect that this will be a very useful approach to predict infection of host cells that have actually experienced viral entry rather than present methods that utilize PCR detection or presence of HIV RNA/DNA that assay for biomarkers in the blood. Furthermore, characterization of gene expression changes may aid in the discovery of novel host cell biochemical machinery usurped by the virus for its own replication [12]-[13]. Here, we chart out a method to develop gene expression signatures for sub-type specific viral entry into the host cell. We interrogated over 12600 gene transcripts and used a combination of statistical filters and multivariate statistical methods to identify genes differentially expressed in MT2 cells infected with different HIV-1 viral sub types. After using statistical filters to identify genes that have a high signal to noise ratio, we organized the data utilizing a clustering algorithm for both treatment and gene categories revealing two sets of related gene expression changes among 397 genes: HIV-1RT-MDR, HIV-1HTLVIIIB, and HIV-192BR019 formed one cluster (greenred); HIV-1A17, HIV-1A17V formed the other cluster group (Figure 1; redgreen cluster) when changes were compared to the no-virus controls using t-value calculations. More genes were up regulated in the red cluster (256; HIV-1A17 and HIV-1A17V) than the green cluster (141 genes; HIV-1RT-MDR, HIV-192BR019 and HIV-1HTLVIIIB ) suggesting that HIV-1 viral subtypes potentially utilize distinct molecular mechanisms for infection and replication. To gain insights into some of these differences we examined gene expression changes for each of the 5 viral sub types and we sought to characterize a sub type specific molecular fingerprint from gene expression changes using a smaller set of genes. Correlation analysis using a multivariate technique identified the key molecular drivers for the differences observed in the gene expression signatures for each viral sub type.

We examined gene expression changes with exposure of MT2 cells to each subtype of HIV-1 virus to identify prominent gene expression changes. The significantly affected genes were ranked according to the most significantly affected genes across all treatment groups for each virus sub-type using ANOVA from calculated p-values calculated from ANOVA (Table 1). Treatment with HIV-1A17V (Table 1A) resulted in greatest fold change for interferon-induced protein with tetratricopeptide repeats 4 (IFIT4; Fold-difference = 2.22) and single-minded Drosophila homolog 2 (SIM2; Fold Difference=1.97). Four other gene products up regulated in HIV-1A17V (IFI44 [14]; PTGDS [15]; ABCC4 [16]-[17], BCL2 [18]) have been previously shown to be associated with HIV infections suggesting strong precedence for these set of genes for their role in disease progression. Five gene products: Mevalonate (diphospho) decarboxylase (MVD); polycystic kidney disease 1 (PKD1); zyxin (ZYX); phosphate cytidylyltransferase 2, ethanolamine (PCYT2); and interferon, alpha 6 (IFNA6) showed more than two fold decrease relative to no-virus controls (Table 1A). Three of the down regulated genes have been implicated in HIV infections from previous studies (PTK2B [19], PDE1B [20] and MVD [21]). The expression of the most significantly HIV-1A17V affected genes were markedly different when treated with other viral subtypes: a smaller effect was observed with HIV-1A17 (Table1B; Fold Difference=1.48); larger effects on IFIT4 were observed in the presence of HIV-192BR019 (Table 1C; Fold Difference = 2.66) and HIV-1HTLVIIIB (Table 1D; Fold difference = 3.08). In the case of SIM2 that was up-regulated with HIV-1A17V treatment (Table 1A; Fold Difference = 1.97), expression values in the presence of HIV-192BR019 resulted in down regulation of gene expression (Table 1C; Fold Difference = 0.32). Notably, structural maintenance of chromosomes 2-like 1 (SMC2L1) mRNA resulted in the most significant increase in expression when treated with HIV-1A17V (Table 1A; Fold Difference = 1.72) and was the most significant down-regulated gene in HIV-1HTLVIIIB (Table 1D; Fold Difference = 0.39). Whereas ZYX showed a decrease with HIV-1A17V (Table 1A; Fold Difference = 0.41), addition of HIV-1HTLVIIIB resulted in increase in expression for this gene (Table 1D; Fold difference = 1.85). Taken together, these results suggest that expression of SMC2L1, SIM2 and ZYX can distinguish HIV-1A17V from HIV-192BR019 and HIV-1HTLVIIIB sub-types and 7 of the gene products have been previously implicated in HIV infection and progression of the disease.

Treatment with HIV-1A17 (Table 1B) showed the largest increase for two of the probe sets for the chemokine (C-C motif) ligand 5 (CCL5) gene suggesting that the A17 variant of the virus strongly activates the production of this gene product compared to the other virus types (Fold Difference = 1.76 and 2.05 for two transcripts). CCL5 is one of the natural ligands for the chemokine receptor CCR5 that suppresses in vitro replication of the R5 strains of HIV-1, and polymorphisms in this gene may play an important part in disease progression [22]-[24]. Other genes that were stimulated with HIV-1A17 included tumor necrosis factor (TNFSF8), Kruppel-like factor 5 (KLF5), IFIT4 and apoptosis-related cysteine protease (CASP2). The role of IFIT4 in HIV-1 infected cells has previously been linked to involvement in chemokine induction to facilitate the expansion of the virus to other cells [14]. Two other genes (TNFSF8 [25] and SF1 [26]) were also found to be associated with HIV infections. Interestingly, MVD and WFS1 showed decreases in expression relative to control to similar levels for both HIV-1A17V and HIV-1A17 variants of virus (Table 1A and Table 1B). The largest fold decrease was observed for peroxisome biogenesis factor 10 (PEX10; Fold Difference = 0.32) and all other changes were less than two fold difference. Six of the down regulated genes were found to be implicated with HIV infections from previous studies (TNFRSF6 [27], ABCC4 [16], GH1 [28], MAN1A1 [29] GNAQ [30] and PRPF4B [31]).

Examining gene expression changes in HIV-192BR019 (Table 1C) showed 7 genes with greater than two fold increases. Three genes of these genes: thyroid hormone receptor-associated protein (TRAP100), CD83 and WFS1, were up regulated for three viral subtypes HIV-192BR019 (Table 1C), HIV-1HTLVIIIB (Table 1D) and HIV-1RT-MDR (Table 1E). TRAP100/MED24 showed the greatest fold difference for the three virus sub-types (3.34 for HIV-192BR019, 3.71 for HIV-1HTLVIIIB, and 1.96 for HIV-1RT-MDR). This protein is a component of the mediator complex, a co-activator involved in the regulated transcription of nearly all RNA polymerase II-dependent genes, and has been reported to bind to Transcription Activation Domain of RelA (NF-KB transcription factor) in JURKAT T-cells [32]. CD83 is an adhesion receptor belonging to the SIGLEC family CD83 and its expression represents a regulatory component for CD4-positive T-cell development in the thymus [33]. Up regulation of CD83 has been shown to be induced by HIV-1 infection in activated dendritic cells to initiate dendritic cell migration to lymphoid tissue [34]. Two genes: chemokine (C-C motif) ligand 4 (CCL4) and valosin-containing protein (VCP), were up regulated for HIV-192BR019 (Table 1C; Fold Difference = 2.85 for CCL4 and 1.75 for VCP) and HIV-1HTLVIIIB (Table 1D; Fold Difference = 2.73 for CCL4 and 1.9 for VCP). Studies have shown that polymorphisms in the CCL4 gene have influenced the rate of HIV-1 transmission and disease progression [35]. It appears that the HIV-1 can induce CCL4 with the addition of HIV-192BR019 or HIV-1HTLVIIIB while CCL5 is activated by HIV-1A17. Examination of the down regulated genes revealed that 4 genes: SIM2, STCH, GNAQ, GNA13 were greater than 3 fold reduced in expression compared to control levels of expression. GNAQ is involved in the Galpha(q) signaling pathway which has been shown to be required for the HIV envelope fusion to the cell membrane [30]. Interferon regulatory factor 4 (IRF4) was strongly down regulated in the presence of HIV-192BR019 (Table 1C; Fold Difference = 0.35) and HIV-1RT-MDR (Table 1E; Fold Difference = 0.6). Expression of two guanine nucleotide binding proteins: GNA13 and GNAQ, and a non-catalytic subunit of RAB3 GTPase-activating protein (RAB3-GAP150), were decreased with HIV-192BR019 (Table 1C; Fold Difference = 0.26 for GNA13 and 0.3 for GNAQ) suggesting a differential inhibition of signal transduction mechanisms can occur in a viral sub-type specific manner.

Tumor necrosis factor (TNFSF8), POU domain transcription factor (POU6F1) and CASP2 were down regulated when treated with HIV-1HTLVIIIB (Table 1D). Amiloride-sensitive cation channel (ACNN1) was markedly reduced with HIV-1HTLVIIIB (Table 1D; Fold Difference = 0.34) and HIV-1RT-MDR (Table 1E; Fold Difference = 0.58). Other down regulated genes with HIV-1RT-MDR treatment included (interferon regulatory factor 4) IRF4 and tumor necrosis factor receptor 6 (TNFRSF6; CD95; Apoptosis antigen 1). IRF4 has been suggested to be immune-regulatory and involved in resistance to HIV infection, whereas TNFRSF6 in HIV infection has been suggested to have a role in the cell death of natural killer cells [36]-[37]. IFIT4 was found to be up regulated for four viral subtypes (HIV-1A17V (Table1B), HIV-192BR019 (Table 1C), HIV-1HTLVIIIB (Table1D) and HIV-1RT-MDR (Table 1E)) and may be a general response to viral infection. In summary, TRAP100, CD83, WFS1, IRF4, GNA13, ACCN1 distinguish between the two clusters of host transcriptomes (Figure 1, HIV-1A17V and HIV-1A17 versus HIV-192BR019, HIV-1HTLVIIIB and HIV-1RT-MDR).

We identified 76 genes whose expression changed most significantly when compared across the virus groups, and therefore provided significant information about differences in the gene expression changes caused by these viruses (Table 2). These genes were used in the Principal Component Analysis to derive uncorrelated variables to be used for the Discriminant Analysis. The first 16 principal components explained 97% of the variance and showed significant separation between the treatment groups. The predictive power of the linear discriminator functions were tested using LOOCV procedure and repeated with fewer principal components. Group membership of the test sample was determined using discriminant functions calculated from the 23 training samples. Comparison of the actual and predicted group membership showed that using 9 (4 misclassifications), 10 (3 misclassifications), 11 (5 misclassifications), 16 principal components (16 misclassifications) resulted in the first 10 principal components explaining 88% of the variance and the fewest number of miscalculations (MANOVA means contrast, Wilks' Lambda, F(45, 47.8) = 4.35, p<0.0001; Pillai's Trace, F(45, 70) = 3.67, p<0.0001). None of the no-virus controls were misclassified (false positive rate).

We next examined the correlation structure of the 76 genes that showed differential expression across the 5 HIV-1 viral sub types. In this multi-dimensional analysis, we defined new axes that were uncorrelated with each other and assessed the correlation of each gene variable onto these axes, thereby identifying genes that were co-regulated and distinguished from other sets of co-regulated genes. In this analysis the 10 principal components that gave the fewest misclassifications were rotated using the varimax method to determine the correlation structure of the 76 gene variables. Table 3 shows the highest correlations for gene variables with each of the 4 defined axes that explained the greatest variation in gene expression changes across HIV-1 viral subtypes. Two genes strongly correlated with axis 1 were up-regulated in HIV-1A17 infected cells; CCL5 and SWAP70. CCL5 expression was most correlated gene with axis 1 (r=0.91) and four other genes showed high correlation with axis 1 (a part of the TREX (transcription/export) complex (THOC), SWAP70, hypoxanthine phospho- ribosyltransferase 1 (HPRT1) and disintegrin/metalloproteinase domain 19 (ADAM19). Gene expression for switch associated protein 70 (SWAP70) was increased with HIV-1A17 (Table 1B), decreased in the presence of HIV-1HTLVIIIB and strongly correlated to axis 1 (r = 0.74). This protein exists both in the cytoplasm and the nucleus and is thought to be a unique signaling protein that specifically binds to PtdIns(3,4,5)P3 to mediate B-cell activation [38]. Furthermore, previous studies report activated CD3-positive T cells having a higher amount of SWAP70 expressed at the surface, suggesting high surface expression of SWAP70 as a marker for HIV-1 infection [39]. Correlations with axis 2 resulted in high correlations of two genes affected by HIV-1A17V infection: IFI44 and IFNA6. Notably, these are signal transduction enzymes and two interferon related gene products (interferon alpha 6 (IFNA6); interferon-induced protein 44 (IFI44)). IFNA6 showed strong negative correlation (-0.784), and IFI44 strong positive correlation (0.829) with axis 2. Infection of chimpanzees with hepatitis C virus showed induction of protein 44 and may be one of the mediators involved in the antiviral action of interferon [40]-[41]. Sixteen genes showed strong correlations with axis 3, which explained the greatest amount of variation in the data (22.42%). TRAP100 was negatively correlated with factor 3 and was found to be up regulated in HIV-192BR019 (Table 1C), HIV-1HTLVIIIB (Table 1D) and HIV-1RT-MDR (Table 1E) and therefore reflected the gene expression signature using the 397 gene set (Figure 1). Secretory carrier membrane protein (SCAMP1; r = 0.85) and cysteine conjugate-beta lyase (CCBL1; r= 0.78) were most strongly correlated with axis 3, and CTBP1 was most negatively correlated with axis 3. One gene TNF receptor-associated factor 3 (TRAF3) showed a strong negative correlation with axis 4, while KRTHA3B and SIM2 showed positive correlations with axis 4. Comparing expression values relative to control showed SIM2 to be up regulated with HIV-1A17V (Table 1A) and down regulated with HIV-192BR019 infection. (Table 1C). These results suggest that the discriminating power of the 397 genes for the 5 HIV-1 viral subtypes may be driven by a smaller set of highly co-regulated genes that can be grouped into 4 basic gene expression profiles, and that one of these genes, SWAP70, has shown to be a marker for HIV infection [39].

In order to determine if there may be a common regulatory motif governing the expression of each type of gene expression profiles, we used the TRES: Transcription Regulatory Element Search to look for common regulatory motifs in the promoter regions of the genes shown in table 3. For axis 1 correlated genes this analysis showed common regulatory motifs of S8, dorsal, bicoid, myoblast determining factor and maternal gene product in the gene's promoter regions. Axis 2 correlated genes shared binding motifs for Ikaros 1, dorsal, activating protein 4, and homeo domain factor Nkx-2.5/Csx, tinman homolog. At least 13 out of 16 genes correlated to axis 3 contained binding sites for ikaros 1. S8, AP 4. SP 1, thing1/E47GC box elements and complex of Lmo bound to Tal1. Further studies are required to determine if any of these common transcription factors actually drive the co- regulated expression of these gene groups.


Analysis of the gene expression profiles of T-lymphoblastoid cells following HIV infection yielded a subtype selective transcription signature of genes on the Affymetrix genechip. PCA analysis of 76 genes followed by a LOOCV procedure using linear discriminant functions results in 3 misclassifications (1 false negative, 2 false viral) and no false positives (0/8) out of the 24 samples. All HIV-1A17 and HIV-1A17V samples were correctly identified in the LOOCV procedure. These findings prompt the hypothesis that a diagnostic microchip assay may be developed to allow for the diagnosis of drug resistant forms of infectious agents within 2 days of infection.


We are grateful for technical assistance in performing the experiments by Sharon Pendergrass and manuscript revisions by Patricia Goodman.

Cluster figure showing 397 differentially expressed genes (columns) in MT2 cells treated with five subtypes of HIV-1 virus (rows). The heat map represents the color-coded expression value reported as T-values comparing expression of the genes in control sample versus the HIV-1 virus treated samples (blue color indicated genes were down regulated and red color indicates genes were up regulated relative to control). The two-way dendrogram shows genes that showed similar expression across samples and samples whose transcriptome profiles were similar to each other across all genes. Each virus subtype elicited a transcriptome signature that was compared to each other and showed two classes of responses (red and green clusters) whereby HIV-1A17 and HIV-1A17V showed up regulation of 256 genes, and HIV-1RT-MDR, HIV-192BR019 and HIV-1HTLVIIIB up regulated 141 genes.

Table 1. Identification of HIV-1 subtype specific genes. Analysis of variance for each gene was used to test for the effect of treatment (P-value for 5 viral, 1 no-virus control comparison are shown). Expression values for each HIV-1 virus subtype was expressed as fold difference relative to control levels for HIV-1A17V (A), HIV-1A17 (B), HIV-192BR019 (C), HIV-1HTLVIIIB (D) and HIV-1RT-MDR (E) virus subtypes.

Table 2. Host cell genes used for classification of viral subtypes. The data were pre-filtered using ANOVA tests between the 6 treatment groups (1 control, 5 virus treated) after removal of each sample, therefore 24 p-values were calculated for each gene. The final set of 76 genes were the most differentially expressed genes across treatments (p<0.01 for full set of samples, p<0.05 for all sets of ANOVAs in which 1 sample was removed, and mean signal for no-virus control>30).

Table 3. Identification of co-regulated genes using factor rotation of gene expression values. The LOOCV procedure determined that using 10 principal components in the MANOVA model resulted in the fewest number of misclassifications. Factor rotation of these 10 components using the varimax method revealed the groups of co-regulated genes that were most correlated to each axis. Correlation coefficient of each gene to each axis and the amount of variance explained by correlation to each axis are shown in the table. The table shows the 4 axes that most explain the variation in the gene expression values and the genes that were most correlated to each axis.


1. G. Alter, G. Hatzakis, C. M. Tsoukas, K. Pelley, D. Rouleau, R. LeBlanc, J. G. Baril, H. Dion, E. Lefebvre, R. Thomas, P. Cote, N. Lapointe, J. P. Routy, R. P. Sekaly, B. Conway and N. F. Bernard, “Longitudinal assessment of changes in HIV-specific effector activity in HIV-infected patients starting highly active antiretroviral therapy in primary infection”, J Immunol., Vol. 171, 2003, pp.477-88.

2. M. Altfeld, E. S. Rosenberg, R. Shankarappa, J. S. Mukherjee, F. M. Hecht, R. L. Eldridge, M. M. Addo, S. H. Poon, M. N. Phillips, G. K. Robbins, P. E. Sax, S. Boswell, J. O. Kahn, C. Brander, P. J. Goulder, J. A. Levy, J. I. Mullins and B. D. Walker, “Cellular immune responses and viral diversity in individuals treated during acute and early HIV-1 infection”, J Exp Med. Vol. 193, 2001, pp. 169-80.

3. U. Alon, N. Barkai, D. A. Notterman, K. Gish, S. Ybarra, D. Mack and A. J. Levine, “ Broad patterns of gene expression revealed by clustering analysis of tumor and normal colon tissues probed by oligonucleotide arrays”, Proc Natl Acad Sci U S A., Vol. 96, 1999, pp. 6745-50.

4. S. A. Armstrong, J. E. Staunton, L. B. Silverman, R. Pieters, M. L. den Boer, M. D. Minden, S. E. Sallan, E. S. Lander, T. R. Golub and S. J. Korsmeyer, “MLL translocations specify a distinct gene expression profile that distinguishes a unique leukemia”, Nat Genet., Vol. 30, 2002, pp. 41-7.

5. M. Bittner, P. Meltzer, Y. Chen, Y. Jiang, E. Seftor, M. Hendrix, M. Radmacher, R. Simon, Z. Yakhini, A. Ben-Dor, N. Sampas, E. Dougherty, E. Wang, F. Marincola, C. Gooden, J. Lueders, A. Glatfelter, P. Pollock, J. Carpten, E. Gillanders, D. Leja, K. Dietrich, C. Beaudry, M. Berens, D. Alberts and V. Sondak, “Molecular classification of cutaneous malignant melanoma by gene expression profiling”, Nature, Vol. 406, 2000, pp. 536-40.

6. I. Hedenfalk, D. Duggan, Y. Chen, M. Radmacher, M. Bittner, R. Simon, P. Meltzer, B. Gusterson, M. Esteller, O. P. Kallioniemi, B. Wilfond, A. Borg and J. Trent, “Gene-expression profiles in hereditary breast cancer”, N Engl J Med., Vol. 344, 2001, pp. 539-48.

7. D. V. Nguyen and D. M. Rocke, “Tumor classification by partial least squares using microarray gene expression data”, Bioinformatics, Vol. 18, 2002, pp. 39-50.

8. A. A. Ferrando, D. S. Neuberg, J. Staunton, M. L. Loh, C. Huard, S. C. Raimondi, F. G. Behm, C. H. Pui, J. R. Downing, D. G. Gilliland, E. S. Lander, T. R. Golub, and A. T. Look, “Gene expression signatures define novel oncogenic pathways in T cell acute lymphoblastic leukemia”, Cancer Cell, Vol. 1, 2002, pp. 75-87.

9. L. J. van 't Veer, H. Dai, M. J. van de Vijver, Y. D. He, A. A. Hart, M. Mao, H. L. Peterse, K. van der Kooy, M. J. Marton, A. T. Witteveen, G. J. Schreiber, R. M. Kerkhoven, C. Roberts, P. S. Linsley, R. Bernards and S. H. Friend, “Gene expression profiling predicts clinical outcome of breast cancer”, Nature, Vol. 415, 2002, pp. 30-6.

10. E. J. Yeoh, M. E. Ross, S. A. Shurtleff, W. K. Williams, D. Patel, R. Mahfouz, F. G. Behm, S. C. Raimondi, M. V. Relling, A. Patel, C. Cheng, D. Campana, D. Wilkins, X. Zhou, J. Li, H. Liu, C. H. Pui, W. E. Evans, C. Naeve, L. Wong and J. R. Downing, “Classification, subtype discovery, and prediction of outcome in pediatric acute lymphoblastic leukemia by gene expression profiling”, Cancer Cell, Vol. 1, 2002, pp. 133-43.

11. S. Dan, T. Tsunoda, O. Kitahara, R. Yanagawa, H. Zembutsu, T. Katagiri, K. Yamazaki, Y. Nakamura, and T. Yamori, “An integrated database of chemosensitivity to 55 anticancer drugs and gene expression profiles of 39 human cancer cell lines”, Cancer Res., Vol. 62, 2002, pp. 1139-47.

12. J. Corbeil, D. Sheeter, D. Genini, S. Rought, L. Leoni, P. Du, M. Ferguson, D. R. Masys, J. B. Welsh, J. L. Fink, R. Sasik, D. Huang, J. Drenkow, D. D. Richman and T. Gingeras, “Temporal gene regulation during HIV-1 infection of human CD4+ T cells”, Genome Res., Vol. 11, 2001, pp. 1198-204.

13. G. K. Geiss, R. E. Bumgarner, M. C. An, M. B. Agy, A. B. van 't Wout, E. Hammersmark, V. S. Carter, D. Upchurch, J. I. Mullins and M. G. Katze, “Large-scale monitoring of host cell gene expression during HIV-1 infection using cDNA microarrays”, Virology, Vol. 266, 2000, pp. 8-16.

14. E. Izmailova, F.M. Bertley, Q. Huang, N. Makori, C. J. Miller, R. A. Young and A. Aldovini, “HIV-1 Tat reprograms immature dendritic cells to express chemoattractants for activated T cells and macrophages”, Nat Med., 2003, Vol. 9, pp. 191-7.

15. W. Li, T. M. Malpica-Llanos, R. Gundry, R. J. Cotter, N. Sacktor, J. McArthur and A. Nath, “Nitrosative stress with HIV dementia causes decreased L-prostaglandin D synthase activity”, Neurology, 2008, Vol. 70, pp. 1753-62.

16. S. Jorajuria, N. Dereuddre-Bosquet, K. Naissant-Storck, D. Dormont and P. Clayette, “Differential expression levels of MRP1, MRP4, and MRP5 in response to human immunodeficiency virus infection in human macrophages”, Antimicrob Agents Chemother., 2004, Vol. 48, pp. 1889-91.

17. A.W. Ravna and G.Sager, “Molecular modeling studies of ABC transporters involved in multidrug resistance”, Mini Rev Med Chem., 2009, Vol. 9, pp. 186-93.

18. E. Priceputu, I. Rodrigue, P. Chrobak, J. Poudrier, T.W. Mak, Z. Hanna, C. Hu, D. G. Kay and P. Jolicoeur, “The Nef-mediated AIDS-like disease of CD4C/human immunodeficiency virus transgenic mice is associated with increased Fas/FasL expression on T cells and T-cell death but is not prevented in Fas-, FasL-, tumor necrosis factor receptor 1-, or interleukin-1beta-converting enzyme-deficient or Bcl2-expressing transgenic mice”, J Virol., 2005, Vol. 79, pp. 6377-91.

19. D. Eggert, P. K. Dash, N. Serradji, C. Z. Dong, P. Clayette, F. Heymans, H. Dou, S. Gorantla, H. A. Gelbard, L. Poluektova and H. E. Gendelman, “Development of a platelet-activating factor antagonist for HIV-1 associated neurocognitive disorders”, J Neuroimmunol., 2009, Vol.213, pp. 47-59.

20. G. Zauli, D. Milani, P. Mirandola, M. Mazzoni, P. Secchiero, S. Miscia and S. Capitani, “HIV-1 Tat protein down-regulates CREB transcription factor expression in PC12 neuronal cells through a phosphatidylinositol 3-kinase/AKT/cyclic nucleoside phosphodiesterase pathway”, FASEB J., 2001, Vol. 15, pp. 483-91.

21. E. Patsouris, O. Katsarou, P. Korkolopoulou, P. Kotsi, A. Kouramba, A. Androulaki and A. Karafoulidou, “Increased microvascular network in bone marrow of HIV-positive haemophilic patients”, HIV Med., 2004, Vol. 5, pp. 18-25.

22. H. Liu, D. Chao, E. E. Nakayama, H. Taguchi, M. Goto, X. Xin, J. K. Takamatsu, H. Saito, Y. Ishikawa, T. Akaza, T. Juji, Y. Takebe, T. Ohishi, K. Fukutake, Y. Maruyama, S. Yashiki, S. Sonoda, T. Nakamura, Y. Nagai, A. Iwamoto and T. Shioda, “Polymorphism in RANTES chemokine promoter affects HIV-1 disease progression”, Proc Natl Acad Sci U S A., Vol. 96, 1999, pp. 4581-5.

23. D. H. Jang, B. S. Choi and S.S. Kim, “The effects of RANTES/CCR5 promoter polymorphisms on HIV disease progression in HIV-infected Koreans”, Int J Immunogenet., 2008, Vol. 35, pp. 101-5.

24. A. Rathore, A. Chatterjee, P. Sivarama, N. Yamamoto, P. K. Singhal and T. N. Dhole, “Association of RANTES -403 G/A, -28 C/G and In1.1 T/C polymorphism with HIV-1 transmission and progression among North Indians”, J Med Virol., 2008, Vol. 80, pp. 1133-41.

25. R. S. Gomez, P.E. de Souza, J.E. da Costa and N.S..Araújo, “CD30+ lymphocytes in chronic gingivitis from HIV-positive patients: a pilot study”, J Periodontol., 1997, Vol. 68, pp. 881-3.

26. D. Missé, J. Gajardo, C. Oblet, A. Religa, N. Riquet, D. Mathieu, H. Yssel, F. Veas, “Soluble HIV-1 gp120 enhances HIV-1 replication in non-dividing CD4+ T cells, mediated via cell signaling and Tat cofactor overexpression”, AIDS., 2005, Vol.19, pp. 897-905.

27. H. Arokium, M. Kamata, and I. Chen, “Virion-associated Vpr of human immunodeficiency virus type 1 triggers activation of apoptotic events and enhances fas-induced apoptosis in human T cells”, J Virol., 2009, Vol. 83, pp. 11283-97.

28. J. Teichmann, U. Lange, T. Discher, J. Lohmeyer, H. Stracke and R. G. Bretzel, “Growth hormone and bone mineral densitiy in HIV-1-infected male subjects”, Eur J Med Res., 2008, Vol. 13, pp. 173-8.

29. M. Knaś, M. Choromańska, K. Karaszewska, D. Dudzik, D. Waszkiel, M. Borzym-Kluczyk, A. Zaniewska and K. Zwierz, “Activity of lysosomal exoglycosidases in saliva of patients with HIV infection”, Adv Med Sci., 2007, Vol. 52, pp. 186-90.

30. B. Harmon and L. Ratner, “Induction of the Galpha(q) signaling cascade by the human immunodeficiency virus envelope is required for virus entry”, J Virol., 2008, Vol. 82, pp. 9191-205.

31. E. M. Bennett, A. M. Lever and J. F. Allen, “Human immunodeficiency virus type 2 Gag interacts specifically with PRP4, a serine-threonine kinase, and inhibits phosphorylation of splicing factor SF2”, J Virol., 2004, Vol. 78, pp. 11303-12.

32. H. R. Owen, M. Quadroni, W. Bienvenut, C. Buerki and M. O. Hottiger, “Identification of novel and cell type enriched cofactors of the transcription activation domain of RelA (p65 NF-kappaB)”, J Proteome Res.,2005, Vol. 4, pp. 1381-90.

33. Y. Fujimoto, L. Tu, A. S. Miller, C. Bock, M. Fujimoto, C. Doyle, D. A. Steeber and T. F. Tedder, “CD83 expression influences CD4+ T cell development in the thymus”, Cell, Vol. 108, 2002, pp. 755-67.

34. D. Wilflingseder, B. Müllauer, H. Schramek, Z. Banki, M. Pruenster, M.P. Dierich and H. Stoiber, “HIV-1-induced migration of monocyte-derived dendritic cells is associated with differential activation of MAPK pathways”, J Immunol., 2004, Vol. 173, pp. 7497-505.

35. W. S. Modi, J. Lautenberger, P. An, K. Scott, J. J. Goedert, G. D. Kirk, S. Buchbinder, J. Phair, S. Donfield, S. J. O'Brien and C. Winkler, “Genetic variation in the CCL18-CCL3-CCL4 chemokine gene cluster influences HIV Type 1 transmission and AIDS disease progression”, Am J Hum Genet., 2006, Vol. 79, pp. 120-8.

36. T. B. Ball, H. Ji, J. Kimani, P. McLaren, C. Marlin, A.V. Hill and F. A. Plummer, “Polymorphisms in IRF-1 associated with resistance to HIV-1 infection in highly exposed uninfected Kenyan sex workers”, AIDS., 2007, Vol. 21, pp. 1091-101.

37. A. Iannello, S. Samarani, O. Debbeche, R. Ahmad, M. R. Boulassel, C. Tremblay, E. Toma, J.P. Routy and A. Ahmad, “Potential role of interleukin-18 in the immunopathogenesis of AIDS: involvement in fratricidal killing of NK cells”, J Virol., 2009, Vol. 83, pp. 5999-6010.

38. M. Shinohara, Y. Terada, A. Iwamatsu, A. Shinohara, N. Mochizuki, M. Higuchi, Y. Gotoh, S. Ihara, S. Nagata, H. Itoh, Y. Fukui and R. Jessberger, “SWAP-70 is a guanine-nucleotide-exchange factor that mediates signalling of membrane ruffling”, Nature, Vol. 416, 2002, pp. 759-63.

39. N. Kimbara, N. Dohi, M. Miyamoto, S. Asai, H. Okada and N. Okada, “Diagnostic surface expression of SWAP-70 on HIV-1 infected T cells”, Microbiol Immunol. ,2006, Vol. 50, pp. 235-42.

40. A. Kitamura, K. Takahashi, A. Okajima and N. Kitamura, “Induction of the human gene for p44, a hepatitis-C-associated microtubular aggregate protein, by interferon-alpha/beta”, Eur J Biochem., Vol. 224, 1994, pp. 877-83.