Reverse transcriptase inhibitors promote the remodelling of nuclear architecture and induce autophagy in prostate cancer cells

Highlights

• RT inhibitors inhibit cell proliferation and induce autophagy in prostate cancer cells, but not in non-cancer cells.

• RT inhibitory treatment induces DNA damage, increased intensity of H3K9me2 and fragmentation of the nuclear lamina.

• Concomitant with this nuclear reorganisation, cancer cells activate the autophagic pathway.

• These results confirm that RT inhibitors are anticancer agents and provide novel insight into their molecular mechanisms.

Abstract

Emerging data indicate that the reverse transcriptase (RT) protein encoded by LINE-1 transposable elements is a promising cancer target. Nonnucleoside RT inhibitors, e.g. efavirenz (EFV) and SPV122.2, reduce proliferation and promote differentiation of cancer cells, concomitant with a global reprogramming of the transcription profile. Both inhibitors have therapeutic anticancer efficacy in animal models. Here we have sought to clarify the mechanisms of RT inhibitors in cancer cells. We report that exposure of PC3 metastatic prostate carcinoma cells to both RT inhibitors results in decreased proliferation, and concomitantly induces genome damage. This is associated with rearrangements of the nuclear architecture, particularly at peripheral chromatin, disruption of the nuclear lamina, and budding of micronuclei. These changes are reversible upon discontinuation of the RT-inhibitory treatment, with reconsititution of the lamina and resumption of the cancer cell original features. The use of pharmacological autophagy inhibitors proves that autophagy is largely responsible for the antiproliferative effect of RT inhibitors. These alterations are not induced in non-cancer cell lines exposed to RT inhibitors. These data provide novel insight in the molecular pathways targeted by RT inhibitors in cancer cells.

Introduction

Human LINE-1 elements constitute the largest family of autonomous retrotransposons, accounting for about 17% of the human genome [1]. LINE-1 retrotransposons harbour two open reading frames (ORFs): ORF1, encoding an RNA-binding protein, and ORF2, encoding reverse transcriptase (RT) and endonuclease (EN) activities essential for the mobility of LINE-1 elements and of other non-autonomous retroelements, e.g. Alu and SVA elements [2,3].

LINE-1s are typically expressed in cells with low differentiation levels and high proliferation rates, such as cancer cells, and are virtually silent in terminally differentiated cells - with the remarkable exception of neuronal cells, which offer a permissive environment for retrotransposition [4]. Consistently, a massive increase of retrotransposal insertions have been identified and mapped in the genomes of many human cancers, e.g. lung [5], colorectal [6,7], prostate, multiple myeloma and glioblastoma [6], hepatoma [8], esophagus [9], pancreas [10], gastric [11],
 ovary [6,12] [reviewed in ref. [13]]. These patterns suggest a circumstantial link between the retroelement machinery and tumorigenesis. It is as yet unclear, however, whether retrotransposal insertions are “driver” mutations with causative roles in tumorigenesis, or whether they are oncologically irrelevant “passengers” [14].

Consistent with the increased level of retrotransposition occurring in cancer cells, a body of data identify expression of LINE-1-encoded ORF1p as a common feature in a variety of tumors [[15], [16], [17]]. Evidence regarding ORF2p is more controversial: high expression was reported in gastric [18] and breast carcinomas [19], and nuclear localization of both ORF1p and ORF2p was proposed as a marker of poor prognosis in some breast cancer types [19,20], whereas ORF2p was not detected in colon cancer tissues or cell lines [21]. These data suggest that ORF1 and ORF2 proteins, though being encoded by bicistronic transcripts, are either uncoupled in their synthesis, or have markedly different individual stability in cancer cells. Despite of the variability in ORF2p detection, the massive waves of retrotransposal insertions documented in human cancers indicate that high RT enzymatic activity is widespread in neoplastic tissues. That was confirmed by measuring reverse transcription in in vitro assays using protein extracts from cancer cells and purified RNA template [22]. Together, these data raise the question of the role played by LINE-1-encoded genes in tumorigenesis.

Previous evidence from our [[22], [23], [24], [25], [26]] and other groups [[27], [28], [29], [30], [31]] showed that nonnucleoside RT inhibitors (NNRTIs) strongly inhibit proliferation and promote cell death in a variety of carcinoma and sarcoma cell lines. Among them, the commercially available efavirenz (EFV; commercial name, Sustiva) has attracted attention because it has long been used and is still currently in use in AIDS treatment as an inhibitor of the HIV-1-encoded RT [32]. Remarkably, downregulation of LINE-1 expression via RNA interference (RNAi) also reduced proliferation and restored differentiation in cancer cell lines [23,24,33]: these findings implicate LINE-1 expression in tumorigenesis and make it unlikely that the anticancer activity of NNRTIs reflects a random off-target effect. The anticancer potential of EFV was further assessed in assays with animal models [23,33] and, more recently, in a phase two-trial with oncological patients [34,35]. Certain nucleoside RT inhibitors (NRTIs), characterised as effective inhibitors of LINE-1 RT [36,37], also exert anticancer effects [38,39] similar to those observed with NNRTIs and LINE-1-specfic RNAi.

In the last few years, another class of potent pyrimidinone-based NNRTIs, called F2-DABOs [40,41], as well as structurally related triazine compounds, ADATs, were investigated for their effects on tumor cells [42,43]. Based on the combination of the obtained structure–activity relationship (SAR), a series of novel pyrimidinones derivatives were designed and screened for their antiproliferative effects in cancer cell lines. Among tested derivatives, compound SPV122, and particularly the stereoisomer SPV122.2, was identified as the most potent [44]. Indeed, SPV122.2 inhibited A375 melanoma cell proliferation with comparable suppressive effects, and in a similar time scale - i.e., within a few days - compared to those observed after RNAi-mediated silencing of LINE-1 elements. Moreover, SPV122.2 inhibited the growth of tumor xenografts in nude mice, even more effectively than reported for other NRTIs or NNRTIs, including EFV [44].

In earlier studies, we analysed the transcription profiles of melanoma cells with or without EFV treatment. A comparative transcriptome analysis indicated that the biogenesis of a sub-population of microRNAs was defective in untreated vs. EFV-treated cells, due to the conversion of their RNA precursors in RNA:DNA hybrid structures [24,45]. Such hybrids are unsuitable substrates for Dicer-dependent maturation, resulting in impaired miRNA biogenesis in untreated melanoma cells. In turn, this induces altered transcription profiles which involve protein-coding mRNAs, miRNAs and long non-coding RNAs [24]. This transcriptome profile was associated with dedifferentiation and highly proliferative capacity, typical of the cancer state. RT inhibitors abolished the formation of RNA:DNA hybrids and re-established the biogenesis of miRNAs, entailing a reduced proliferation rate and antagonizing tumor progression. The observation that the effects of RT inhibitors are reversible upon treatment discontinuation would hardly be compatible with a mechanism solely based on retrotranspositional insertions, which are permanent events. The reversible effects of NNRTIs rather suggest that LINE-1 exert epigenetic regulatory functions on the cell transcriptome, compatible with the impairment of miRNA biogenesis via reverse transcription of RNA precursors and not necessarily associated with retrotransposition proper [46].

These observations raise the question of whether the cancer cell transcriptome reprogramming is related to the reshaping of the nuclear landscape. To address that question, here we have chosen to concentrate on PC3 prostate metastatic carcinoma cells, which proved sensitive to inhibition by NNRTIs when xenografted in mice [23]. In addition, NNRTIs also yielded promising results when administered to metastatic prostate cancer patients, albeit in a small size trial [34,35]. We now show that PC3 cell exposure to both EFV and SPV122.2 generates a cascade of events that ultimately alter nuclear functions, trigger the autophagic pathway and drive the cells towards death. Pivotal to this process is a global rearrangement of the nuclear architecture induced by genome damage and disruption of the nuclear lamina. We suggest that these combined actions of NNRTIs induce an extensive remodelling of the chromatin architecture that has the potential to redirect the fate of cancer cells.