Correlates between host and viral transcriptional program associated with different oncolytic vaccinia virus isolates

Abstract

Vaccinia virus (VACV) has emerged as an attractive tool in oncolytic virotherapy. VACV replication efficiency plays a crucial role in the therapeutic outcome. However, little is known about the influence of host factors on viral replication efficiency and permissiveness of a host cell line to infection and oncolysis. In this study, replication of the attenuated VACV GLV-1h68 strain and three wild-type VACV isolates was determined in two autologous human melanoma cell lines (888-MEL and 1936-MEL). Host gene expression and viral gene expression in infected cells were evaluated via respective expression array platforms. Microarray analyses followed by sequential statistical approaches characterized human genes that change specifically due to virus infection. Viral gene transcription correlated with viral replication in a time-dependent manner. A set of human genes revealed strong correlations with the respective viral gene expression. Finally we identified a set of human genes with possible predictive value for viral replication in an independent dataset. The results demonstrate a probable correlation between viral replication, early gene expression, and the respective host response, and thus a possible involvement of human host factors in viral early replication. The characterization of human target genes that influence viral replication could help answer the question of host cell permissiveness to oncolytic virotherapy and provide important information for the development of novel recombinant vaccinia viruses with improved features to enhance replication rate and hence trigger therapeutic outcome.

FIG. 1.

Flow chart, describing the experiment for analysis of VACV-dependent changes in host cell transcription. 888-MEL and 1936-MEL cells were infected with four different VACV isolates at an MOI of 0.01 (GLV-1h68, LIVP 1.1.1, LIVP 5.1.1, and LIVP 6.1.1) in separate experiments or were treated with virus-free infection medium as negative control. Total RNA was amplified, purified, and labeled before generating transcriptional profiles via 36k human or custom-made VACV array platforms. (A) Expression profile of 888- and 1936-MEL cells after virus and mock infection at 2, 10, and 48 hr (filter 95%, 2-fold). (B) Statistical approach to obtain a list of human genes affected by VACV infection specifically. (C) Human transcriptional profile after applying SD exclusion/inclusion method. (D) Correlation between VACV transcription and virus-affected host genes (Pearson correlation, r=0.6). SD, standard deviation; ctr, control samples; hpi, hours post infection; VACV, Vaccinia virus; MOI, multiplicity of infection.

FIG. 2.

VACV-isolate-specific changes in gene transcription over time and correlates with viral replication. 888-MEL and 1936-MEL cells were infected with four different VACV isolates (GLV-1h68, LIVP 1.1.1, LIVP 5.1.1, and LIVP 6.1.1) in separate experiments (MOI 0.01) or were treated with virus-free infection medium as negative control. Viral titers were determined by plaque assay and total RNA was isolated, amplified, purified, and labeled before generating transcriptional profiles via custom-made VACV array platforms. (A) Transcription profile including subcluster classification of four VACV isolates and negative control after 2 and 10 hr. (B) Principal components analysis (PCA, Partek Genomics Suite) of the VACV microarray data. Infection period and infectional status appeared to be a major source of variation, as indicated by red ellipsoids. The different colors specify the VACV isolate utilized. Green=GLV-1h68; purple=LIVP 1.1.1; yellow=LIVP 5.1.1, light blue=LIVP6.1.1; and pink=untreated controls. (C) VACV growth curves from 0 to 10 hpi. (D) Transcriptional ranking of VACV isolates. Transcription values were averaged considering the different time points and individual subclusters. Subsequently, viruses were ranked from lowest to highest mean gene transcription. (E) Hierarchical clustering of individual genes representing the transcriptional ranking, low to high: GLV-1h68, LIVP 5.1.1, LIVP 1.1.1, LIVP 6.1.1 or GLV-1h68, LIVP 1.1.1, LIVP 5.1.1, LIVP 6.1.1 in C3a (top) and C4 (bottom). C1, 2, 3a, 3b, 4, cluster 1, 2, 3a, 3b, 4; hpi, hours post infection; hrs, hours; VACV, Vaccinia virus; MOI, multiplicity of infection; RUC-GFP, Renilla luciferase–green fluorescent protein (Aequorea) fusion; gus A, β-glucoronidase.

FIG. 3.

Pathway analyses of correlates between VACV and host cell transcription. Canonical pathways and top networks were identified by ingenuity pathway analysis (IPA). (A) Top canonical pathways of 114 early human correlates with averaged VACV early gene transcription. (B) Top canonical pathways of 84 early human correlates with averaged VACV intermediate/late gene transcription. (C, D) Network: post-translational modification, free radical scavenging, and gene expression revealed an overlap of 14 genes out of an intercept of 66 genes between early (C) and intermediate/late (D) correlates. Red circles, positive correlation; green circles, negative correlation; hpi, hours post infection; h, human; VACV, Vaccinia virus.

FIG. 4.

Prediction of viral replication in an independent dataset. GI101A and HT29 cells were infected with six different VACV isolates (GLV-1h68, GLV-1h70, GLV-1h71, GLV-1h72, GLV-1h73, and GLV-1h74) in separate experiments (MOI 0.01). Total RNA was isolated, amplified, purified, and labeled before generating transcriptional profiles via custom-made VACV array platforms or human whole genome (36k) array platforms. Predictor genes were determined by the following criteria: r in melanoma cell lines ≥0.5 (absolute value); r in all cell lines has to have same direction (positive or negative); and r in GI101A and HT29 has to be either at least moderate in both cases or, if one correlation is weak in one cell line, it must be strong in the other one. (A) Ten predictors of viral replication were determined by Pearson Correlation. (B) Heat map of the six predictive positive human correlates at 2 hpi in 888-MEL, 1936-MEL, GI101A, and HT29; averaged viral gene expression of seven VRIs at 2 hpi from cluster C3a (bar graphs). VRI, viral replication indicators.