Neuron,

Vol. 7, 381-390,

September,

1991, Copyright

0 1991 by Cell Press

The Octamer-Binding Protein Ott-2 Represses HSV Immediate-Early Genes in Cell Lines Derived from latently Infectable Sensory Neurons K. A. Lillycrop,* C. 1. Dent,* S. C. Wheatley,* M. N. Beech,+ N. N. Ninkina,+* J. N. Wood,+ and D. S. Latchman* *Medical Molecular Biology Unit Department of Biochemistry University College and Middlesex School of Medicine London, WIP 6DB England +Department of Neuroimmunology Sandoz institute for Medical Research London, WCIE 6BN England

Summary Transcription of herpes simplex virus (HSV) immediateearly (IE) genes does not occur in sensory neurons latently infected with the virus or following infection of neuronal cell lines. In neuronal cell lines this inability results from the weak activity of the viral IE promoters, which is caused by a neuron-specific repressor factor that binds specifically to the TAATGARAT motif in these promoters and to related octamer elements. Cells expressing this repressor contain an additional octamerbinding protein that is absent from permissive cells. We identify this factor as the lymphocyteand neuronspecific octamer-binding protein Ott-2 and show that Ott-2 mRNA is also present in dorsal root ganglion neurons, the natural site of HSV latency in vivo. Moreover, artificially elevated expression of Ott-2 can repress the IE promoter. The potential role of Oct.2 in the initiation and maintenance of in vivo latent infection with HSV is discussed. Introduction Herpes simplex virus (HSV) establishes life-long, asymp tomatic latent infections of sensory neurons, which provide a reservoir for repeated cycles of productive lytic infection elsewhere in the body (for reviews see Roizman and Sears, 1987; Latchman, 1990). These latently infected cells do not contain the viral immediate-early (IE) proteins and the mRNAs that encode them (Stevens et al., 1987; Croen et al., 1987), and hence the viral lytic cycle is aborted at an early stage. It is likely, therefore, that differences in the ability of different cell types to support viral IE gene expression play a critical role in determining the outcome of an initial infection with HSV. In lytic infection, IE gene expression is dependent upon the interaction of cellular transcription factors present in the infected cell with binding sites in the viral promoters. Thus, the * Permanent address: Institute of Molecular Biology, 32 Vavilov Street, Moscow 117-984, Union of Soviet Socialist Republics.

viral IE promoters contain binding sitesforthecellular factor Spl as well as multiple copies of the octamer-related sequence TAATGARAT, which binds the octamer-binding transcription factor Ott-1 following formation of a complex with the viral protein Vmw65 (Jones and Tjian, 1985; O’Hare and Coding, 1988). It is likely, therefore, that differences in the nature of these cellular transcription factors in latently infected neuronal cells are responsible for failure of IE gene expression and may reflect differences in the regulation of cellular gene expression mediated by these factors. To studythe interaction of HSV with neuronal cells, we have previously used the Cl300 mouse CNS neuroblastoma cell line and a series of cell lines (ND cells) prepared by fusing these cells with primary sensory neurons (Wood et al., 1990), the major site of HSV latency in vivo. Both the Cl300 cells and the ND lines are nonpermissive for HSV infection (Vahlne and Lycke 1978; Ash, 1986; Wheatley et al., 1990), and this nonpermissivity is due to a failure of IE gene transcription following infection (Kemp and Latchman, 1989; Wheatley et al., 1990). The lack of IE gene transcription following infection is in turn dependent upon the weak activity of the IE promoters in these cells; constructs in which these promoters drive expression of the readily assayable chloramphenicol acetyltransferase (CAT) gene (Gorman, 1985) are expressed very weakly compared with permissive cell types (Kemp et al., 1990). Most interestingly, however, the activity of IE-CAT constructs in Cl300 cells (Kemp et al., 1990) and ND cells (Wheatley et al., 1991) can be dramatically increased by cotransfecting plasmids containing isolated TAATGARATelements of the type found in the IE promoters or the related octamer motif found in cellular gene promoters. Moreover, the permissivityof these neuronal cells for HSV infection can also be increased by transfecting them with cloned octamer or TAATGARAT elements (Kemp et al., 1990; Wheatley et al., 1991). These results suggest that the weak activity of the IE promoters and the consequent failure of IE gene transcription following infection arecaused byacellular octamer-binding factor present in neuronal cells that binds to the TAATGARAT motif in the IE promoters and inhibits gene activity. Hence the removal of this factor by cotransfection of its binding site increases IE promoter activity and viral permissivity. In the work presented here we show directly that these neuronal cells contain an additional octamer/ TAATGARAT-binding protein that is absent from permissive cells. The relative binding of this protein to different octamer/TAATGARAT sequences in neuronal cells determines the degree of repression of promoters containing these sequences. We identify this factor as the lymphocyteand neuron-specific octamer-binding protein Ott-2 and show directly that arti-

abed

Figure 1. DNA from ND7 and

Mobility Shift 3T3 Cells

a

Assay

Using

Whole-Cell

Extracts

Lanes a and b, ND7 cells; lanes c and d, 3T3 cells. In lanes a and c the probe was the OctamerRAATCARAT sequence ATCCTAATCATAT; in lanes b and d the probe was the mutant octamer ATAATAATAA.

ficially

elevated

expression

of Ott-2

can

repress

the

IE promoter. In addition, we demonstrate that Ott-2 mRNA is expressed by dorsal root ganglion (DRG) neurons, the natural site of HSV latency in vivo.

Results Neuronal Cells Contain an Additional Octamer-Binding Protein The cell lines used in these studies (ND cells) were prepared by fusing primary neuronal cells isolated from neonatal rat DRGs (the predominant site of HSV latency in vivo) with N18 TG2 cells-a HAT-sensitive neuroblastoma cell line derived from Cl300 cells. Ganglion cultures had been previously treated with cytosine arabinoside for 2 days to kill dividing cells, such as fibroblasts and Schwann cells, and produce an essentially pure population of sensory neurons. The resulting immortalized cell lines (Wood et al., 1990) provide a large-scale source of material while retaining many of the properties of the parental DRG cells, including the synthesis of characteristic glycolipid surface markers (Dodd and Jessell, 1985), opioid receptors, and sensory neuropeptides such as calcitonin gene-related peptide. These cells also exhibit electrophysiological responses to a number of sensory neuron activators, such as bradykinin and capsaicin, which are not exhibited by Cl300 cells. Although both ND and Cl300 cells are nonpermissive for HSV infection (Vahlne and Lycke, 1978; Wheatley et al.,

Figure

2. DNA

Mobility

bcde

Shift

Assay

Using

ND

Cell Extract

The labeled octamer sequence ATGCTAATCATAT was used in the absence of unlabeled competitor (lane a) or in the presence of a IO- or 100.fold excess of octamer (lanes b and c) or Spl (lanes d and e) oligonucleotide competitor.

1990), only ND cells (Wheatley Cl300 cells (Kemp and Latchman, HSV latency-associated transcript

et al., 1990) and not 1989) express the following infection.

Since the latency-associated transcript is the only HSV transcript present in latently infected sensory neurons in vivo (Croen et al., 1987; Stevens et al., 1987), ND cells offer a relevant model system for studying the interaction of HSV with neuronal cells. In view of our previous findings (Kemp et al., 1990; Wheatley et al., 1991) that the IE promoters are inhibited in neuronal cells by a cellular octamer/TAATGARAT repressor factor, we attempted to obtain physical evidence for this factor. To do this, we carried out DNA gel retardation assays (Fried and Crothers, 1981) using an OctameriTAATGARAT oligonucleotide from the IEI promoter (Perry et al., 1986). Two major DNAprotein complexes formed on this motif when extracts from ND7 cells were used (Figure 1, lane a). These failed to form on a mutated octamer (Figure 1, lane b) that does not bind previously characterized octamerbindingproteins(Lenardoetal.,1987),confirmingthat they were due to the presence of two sequencespecific octamer-binding proteins in the ND cells. In contrast, when extracts from 3T3 cells, which synthesize high quantitites of IE protein following infection (Kemp and Latchman, 1989; Wheatley et al., 1990), were used, only a single low mobility complex was observed and the second complex was absent (Figure 1, lanes c and d). As expected, both extracts showed similar patterns of binding with other oligonucleotides containing binding sites for other transcription factors, such as Spl and ATF (data not shown).

Neuronal 383

Oct.2

Represses

Octametconsensus HSV A

IE consensus

HSV IE Gene

A

T

G

C

A

A

A

T

N

Figure 3. Sequence of the OctamermAATCARAT Oligdnucleotides and Comparison with the Octamer Consensus and the HSV TAATCARAT Consensus

A

RYGNTAATGARAT ATGCTAATGAGAT

Oigonucleotide A is a composite overlappi& octamer/TAATGARAT ‘motif not identified in any known promoter; the seATGCAAATAA quence in oliRonucleotide C, which differs f;om this perfect consensus by 1 base, is located immediately upstream of the HSV-1 IEI promoter (Perry et al., 1986). Oligonucleotide B is a TAATCARAT motif found upstream of the HSV-1 IE4/5 promoter (Whitton and Clements, 1984); oligonucleotide D is a perfect octamer motif. The octamer consensus is from Faulkner et al. (1986); the HSV IE consensus is from Whitton and Clements (1984).

B

GCGGTAATGAGAT

C D

ATGCTAATGAfAT

In competition experiments the two octamer-dependent bands could be removed only by octamerorTAATGARAT-containingoligonucleotides(Figure2, lanes b and c) and not by competition with the octamer mutant oligonucleotide or an oligonucleotide containing the Spl-binding site (Dynan and Tjian, 1983; Figure2, lanes d and e). It is clear, therefore, that in addition tothe ubiquitousoctamer-binding protein Ott-I (which is responsible for the largest complex), ND cells contain another octamer-binding protein that is absent from 3T3 cells and is likely to be the repressor factor identified in our previous competition experiments. A similar additional octamer-binding protein is also present in Cl300 cells (Kemp et al., 1990) and in another ND cell line (ND3; data not shown).

Binding Specificity of the Octamer-Binding Protein Correlates with Repression of Gene Expression Having identified this factor, we investigated its binding specificity using a panel of different octamer or TAATGARAT oligonucleotides (see Figure 3 for sequences of these oligonucleotides). In these experiments (Figure 4) strong binding was observed as before to an oligonucleotide (C) containing the overlapping octamer/TAATGARAT motif (ATGCTAATGATAT) found upstream of the IEI promoter (Perry et al., 1986). In contrast, a single base change to produce a perfect consensus TAATGARAT portion (oligonucleotide A: ATGCTAATGAGAT), which is not present in any IE promoter, virtually abolished binding of the neuronal factor. It was also noteworthy that binding of the lower band predominated over that of Ott-I on both the oligonucleotides (6 and C) that contain sequences present in HSV IE promoters, although the absolute level of binding to these two sequences differed greatly. Hence the repressor factor exhibits strong sequence specificity (binding being dramatically affected by a single base change), while it binds more strongly than Ott-I to two distinct sequences present in HSV IE promoters. To test whether these differences in binding have any functional consequences, we cloned each of these oligonucleotides upstream of the thymidine kinase promoter in the vector pBL2 CAT (Luckow and Schutz, 1987). These constructs were then transfected into permissive BHK and ND7 cells. In these experi-

ments (Figure 5) each of the oligonucleotides directed a similar level of CAT activity in the BHK cells, which contain only Ott-1. This activity was considerably higher than that observed with the pBL2 CAT vector, indicating that all the oligonucleotides activated the thymidine kinase promoter. In contrast, in ND cells, the two constructs containing oligonucleotides that bound predominantlythe repressor protein (oligonucleotides B and C) showed low levels of CAT activity, similar to that of the vector. This indicates that they did not activate promoter activity, although the two constructs bearing oligonucleotides that bound predominantly Ott-1 (oligonucleotides A and D) increased promoter activity above that in the parental vector. Hence a single base change (oligonucleotide A compared with C) that dramatically affects binding of the lower band also dramatically affects the level of gene expression directed in ND cells. Moreover, a direct correlation exists between the gene activity directed bytheoligonucleotidesandtheirabilitytobind

abc

d