DNA extracted from cultures of a cytopathic virus isolated from a patient with chronic fatigue syndrome was cloned into pBluescript plasmid. The nucleotide sequences of the plasmid inserts were analyzed using the BlastN and BlastX programs of the National Center for Biotechnology Information. In confirmation of earlier studies, many of the sequences show partial homology to various regions within the genome of human cytomegalovirus (HCMV). The matching regions were unevenly distributed throughout the HCMV genome. No matches were seen with either the UL55 or UL83 genes, which provide the major antigenic targets for anti-HCMV cytotoxic T cell mediated immunity. This finding is consistent with the notion that certain viruses can avoid immune elimination by deleting genes required for effective antigenic recognition by the cellular immune system. The term "stealth" has been applied to such viruses. Comparisons were also made between the sequences of the stealth virus and the limited sequence data available on cytomegaloviruses from rhesus and from African green monkeys. These comparisons unequivocally establish that the virus was derived from an African green monkey simian cytomegalovirus (SCMV).

A major function of the cellular immune system is to recognize and respond to virus infections (1). A successful immune response can eradicate foreign viruses by destroying infected cells prior to the release of progeny viruses. Symptoms of an acute virus infection may occur during the time period required to generate a primary cellular immune response, and can also be a byproduct of cellular immune damage inflicted on viral infected cells. Unusually severe infections can occur if the immune system is impaired, for example, as a result of immaturity, chemotherapy, coincident infection with human immunodeficiency virus (HIV), or specific genetic deficits in immune competence (2). Certain viruses may also interfere with immunological defenses through such mechanisms as downregulation of the expression of histocompatibility antigens, induction of immunosuppressive cytokines and related virokines, and by remaining inactive, as occurs during periods of virus latency (1-10).

Stealth adaptation was proposed as a distinct process whereby a virus could remain actively cytopathic in the absence of an accompanying inflammatory cellular immune response (11). Based on a series of histological findings, and on repeated isolations of atypically structured cytopathic viruses, it was suggested that these viruses lacked certain critical viral genes that coded for the antigens that would have provided effective targets for the cellular immune system (11-20). This hypothesis is still unacceptable to many virologists who intuitively believe, either that all viruses possess ample targets for immune recognition and/or that viruses without such critical elements would be unable to replicate, cause cell damage or pass between individuals.

Sequence studies on a prototype stealth-adapted virus has provided data consistent with this hypothesis. The results presented in this paper also confirm the derivation of this cyptopathic virus from an African green monkey simian cytomegalovirus (SCMV).

Patient. A 43 year old female developed an acute illness in 1990. It began with a sore throat and was followed by intense headaches, panic-like reactions, altered level of consciousness, generalized muscle pain and diarrhea. Although the acute neurological symptoms subsided during a week of hospitalization, the patient has never regained her normal health. Her life has changed from that of an energetic, socially adept, joyful and artistically creative individual to that of an overly fatigued, emotionally labile, generally depressed and cognitively impaired person struggling to survive financially on social disability payments.

Virus Culture. A blood sample obtained several months after the onset of the patient’s illness induced a foamy vacuolating cytopathic effect (CPE) in human foreskin fibroblasts (11). Viral particles were seen by electron microscopy and the cultures gave positive polymerase chain reactions similar to those observed using the patient’s blood. The CPE was readily transmissible to a wide range of cell types of several species, including Sf9 insect cells of Spodoptera frugiperda (16).

DNA Extraction and Cloning. Infected MRC-5 cells were used to extract DNA for cloning. In the first series of experiments, DNA was extracted from the material present in filtered culture supernatants that was pelleted by ultracentrifugation. The DNA was cut using EcoRI enzyme and the digestion products ligated into pBluescript plasmid (Stratagene, La Jolla). Bacteria were transfected and 200 clones containing inserts were randomly selected and assigned a 3B number (11). A second series of pBluescript clones, designated C16, was obtained using SacI digestion of agarose banded DNA. The DNA had been extracted from the material released by lysis of viral infected MRC-5 and pelleted by ultracentrifugation (16).

DNA Sequencing. The T3 and T7 promoter sites of the pBluescript plasmid were used to obtain partial sequence data from the ends of each of the inserts. Extended sequences of the inserts in selected clones were obtained using primers based on the T3 and T7 readouts. The sequencing services of the City of Hope Cancer Center, Durate CA; Midland Certified Reagent Company, Midland, Texas; and Lark Technology, Houston, Texas, were used with excellent correlation of duplicate and triplicate testing of the same clones. The individual sequences were analyzed against GenBank entries using the gapped BlastN and unfiltered BlastX programs of the National Center for Biotechnology Information (NCBI) of the National Library of Medicine (21). A "p" value of <e -2.0 was used as a cutoff for a significant nucleotide or amino acid sequence homology. Known sequences of human, rhesus and simian cytomegaloviruses were obtained using the Entrez Program of NCBI.

At least partial sequencing has also been obtained on 300 clones from the virus infected cultures. For many of the clones, the sequencing merely comprises a relatively short readout from the T3 and T7 promoter sites of the pBluescript plasmid. For selected clones, complete (C) sequencing has been obtained. Several of the clones contained sequences that could be statistically aligned by BlastN analysis to at least some portion of the protein coding regions of the HCMV genome. The BlastX program, which is based on the deduced amino acid sequences coded by a nucleotide sequence, identified many additional clones with significant partial sequence homology to various HCMV proteins. The matching was often incomplete with apparent gaps and occasional insertions in the sequences of the corresponding stealth virus clones. An overall indication of the HCMV proteins for which at least some portions could be aligned with a sequence contained in one or more of the stealth virus clones is shown in Table 1.

As shown in the Table, the nucleotide and protein matching regions of the stealth virus were widely, yet unevenly, distributed throughout much of the HCMV genome. Of the 300 clones, 10 or more clones matched to the UL36, UL52, UL86 and US28 genes of HCMV, while certain other genes were not represented by any of the clones from which sequence data have been obtained. The homologues of the UL55 and UL83 HCMV genes were not represented in any of the sequenced stealth virus clones. The terminal end of six clones were identified as matching to the UL84 gene, but all of the clones extended in a forward direction. Similarly, 10 clones were found that matched to the UL52 gene but all extended in a backward direction.

Limited nucleotide and amino acid sequence data are available for rhesus cytomegalovirus (RhCMV). Several of these sequences corresponded to regions of the HCMV genome that matched to portions of the prototype stealth virus sequence. For these regions, it was possible to compare the relative relatedness of the stealth virus sequence with that of RhCMV and HCMV. A comparison of the p values seen when matching the nucleotide sequences using BlastN program and deduced amino acid sequences using BlastX program of the stealth virus clones against homologous regions of HCMV and RhCMV sequences are listed in Table 2. This analysis showed a significantly higher homology of the stealth virus sequences to RhCMV compared to HCMV.

Only a small amount of genetic information is available for cytomegaloviruses of other primates. Five sequences are recorded in GenBank for the Colburn strain of African green monkey simian cytomegalovirus (SCMV). Sequences for the assembly protein (GenBank accession numbers M24205), and for a cell virus homology region (GenBank accession M24205) of SCMV, were not matched by any of the stealth virus clones. Portions of the other three known regions of SCMV were, however, closely matched by one or more of the stealth virus clones. Specifically, 96% identity occurred over the 641 nucleotide length of the T3 sequence of stealth virus clone C16278 with the sequence of SCMV single strand DNA binding protein (GenBank accession number D00750). Stealth virus clone C16136 showed a 95% identity to a region within the origin of lytic replication of SCMV (GenBank accession number M57681). Extensive homology was also seen between the fully sequenced stealth virus clone 3B546 and the longest of the known contiguous sequences of SCMV. This section covers the immediate-early gene transcriptional unit IE94 (GenBank accession numbers U38308, M16019, L06819 and U18245) of SCMV, and potentially encodes 18 proteins. As detailed in Table 3, the alignment between SCMV and clone 3B546 occurred over 8 discrete regions, with small gaps and occasional overlapping of sequences which differed in size between SCMV and the stealth virus clone.

The genome of the fully sequenced, laboratory-adapted, AD169 strain of HCMV comprises 235,000 nucleotides base pairs (22). The virus has two linear segments, designated unique long (UL) and unique short (US). Each segment is flanked by relatively small regions of repetitive sequences. The potential proteins coded by the UL and US regions are designated numerically, and extend from UL1-UL132 and US1-US36. Additional potential open reading frames are present within the repeat regions that flank both the UL and US segments. Fresh clinical isolates of HCMV contain additional protein coding sequences, designated UL133-151, that are not present in the laboratory adapted AD169 strain (23). Sequence comparisons of animal and human herpesviruses indicate a greater conservatism of the central core sequences from UL30 to UL110, compared to the sequences prior to UL30 and beyond UL110.

The matching sequences contained within the stealth virus clones corresponded to numerous genes of HCMV. The matching sequences were not uniformly distributed throughout the HCMV genome with some regions being matched by multiple clones, and yet other regions not being represented by any of the clones so far sequenced.

The UL83 lower matrix protein is the dominant HCMV antigen for immune recognition of infected cells by cytotoxic T lymphocytes (24). The homologue of this gene was not represented in any of the sequenced stealth virus clones. Similarly, the gB glycoprotein coding UL55 gene of HCMV, which also provides a major target for anti-HCMV cytotoxic and helper T cells (25,26), was not identified by any of the stealth virus clones. It could be argued that since sequence data on the UL83 genes of SCMV and RhCMV are not known, an SCMV UL83 gene sequence could have been missed. This problem does not apply to the UL55 gene since the HCMV and RhCMV UL55 genes show relatively minor differences (27), and indeed there is significant sequence homology to the gB glycoprotein of murine cytomegaovirus.

The apparent lack of the UL83 and UL55 genes among the stealth virus clones, is consistent with the underlying assumption concerning gene deletion as a means for stealth adapted viruses to bypass cellular immune defense mechanisms (11). Significant deletions were also seen in portions of many of the proteins that were identified using the BlastX program. Moreover, some of the deduced amino acid sequences were not included in analyses based on projected open reading frames (data not shown) and may not be expressed in infected cells.

Where the comparison could be made, the stealth virus sequences were more closely related to RhCMV that to HCMV. This finding helps exclude an origin from HCMV and points to a primate origin. The extremely high degree of homology between the stealth virus and regions of the SCMV genome (28), with p values often recorded as zero, unequivocally identifies the virus as a derivative of SCMV.

Although not specifically determined in this study, the aggregate lengths of the cloned stealth viral DNA sequences that match to non-overlapping regions of the HCMV genome, exceeds 100 kb. Since the stealth virus DNA migrates in agarose gels with a size of only approximately 20 kilobase (11), it is assumed that the genome consists of multiple fragments, rather than as an entire full length cytomegaloviral genome. As shown in the accompanying papers, the situation is even more complex since many of the clones contain sequences that cannot be aligned to a conventional cytomegalovirus(29,30).