DNA microarrays of the complex human cytomegalovirus genome: profiling kinetic class with drug sensitivity of viral gene expression

Abstract

We describe, for the first time, the generation of a viral DNA chip for simultaneous expression measurements of nearly all known open reading frames (ORFs) in the largest member of the herpesvirus family, human cytomegalovirus (HCMV). In this study, an HCMV chip was fabricated and used to characterize the temporal class of viral gene expression. The viral chip is composed of microarrays of viral DNA prepared by robotic deposition of oligonucleotides on glass for ORFs in the HCMV genome. Viral gene expression was monitored by hybridization to the oligonucleotide microarrays with fluorescently labelled cDNAs prepared from mock-infected or infected human foreskin fibroblast cells. By using cycloheximide and ganciclovir to block de novo viral protein synthesis and viral DNA replication, respectively, the kinetic classes of array elements were classified. The expression profiles of known ORFs and many previously uncharacterized ORFs provided a temporal map of immediate-early (alpha), early (beta), early-late (gamma1), and late (gamma2) genes in the entire genome of HCMV. Sequence compositional analysis of the 5' noncoding DNA sequences of the temporal classes, performed by using algorithms that automatically search for defined and recurring motifs in unaligned sequences, indicated the presence of potential regulatory motifs for beta, gamma1, and gamma2 genes. In summary, these fabricated microarrays of viral DNA allow rapid and parallel analysis of gene expression at the whole viral genome level. The viral chip approach coupled with global biochemical and genetic strategies should greatly speed the functional analysis of established as well as newly discovered large viral genomes.

FIG. 1

Oligonucleotides used in the microarray deposition.

FIG. 2

Microarray analysis of HCMV gene expression. (Top panel) Scatterplot of the normalized square root (sqrt) of the median level of expression of mock-infected cells (triplicates, n = 6) on abscissa in contrast to normalized square root of the median level of expression of infected cells (triplicates, n = 6) under IE conditions. (Middle and lower panels) The same comparison shown in the top panel is shown for E expression (n = 3) (in the presence of ganciclovir) and L expression (n = 2) (in the absence of drug at 72 h postinfection). The points above the line of equivalence exhibit the ORFs expressed. Representative ORFs are indicated (see text for details). The point adjacent to IRL-4 represents expression from a triplicate set of TRL-4 (n = 3).

FIG. 3

Sensitivity of HCMV gene expression to ganciclovir treatment. The fold decrease in levels of gene expression in cells 72 h postinfection treated with ganciclovir or left untreated are represented by concentric circles. Each ORF was arbitrarily assigned a radial spacing of approximately 1°. The inner, shaded circle shows values that exhibit threefold-or-less change in normalized median levels of expression (n = 3). The outer boxed point is off scale, and the name of the ORF (UL130) and fold decrease (919-fold) are marked. Representative ORFs are indicated.

FIG. 4

Northern blot analysis of selected HCMV ORF transcripts. Whole-cell RNA was harvested either from uninfected cells (lane 1) or HCMV-infected cells at 13 h postinfection in the presence of cycloheximide (lane 2) or 72 h postinfection in the absence (lane 4) or presence (lane 3) of ganciclovir. RNAs were separated on formaldehyde-agarose gels, transferred to a nylon membrane, and hybridized with radiolabelled probes specific for selected viral transcripts (details in Materials and Methods). The probes used are indicated at the top of the lanes. The position and approximate size of the major viral transcript(s) detected in each case are indicated on the right of each panel. The 1.4-kb GAPDH band detectable in all lanes is indicated by an arrow to the left of the lower panels.