**3. Type I interferon and endometriosis**

We will now focus on the putative implication of the important immune modulating cytokine family type I interferons in endometriosis, as a line of circumstantial evidence

Virus Infection and Type I Interferon in Endometriosis 255

natural killer cells and thus reduce the growth of endometriosis tissue (Acién et al., 2002). Rats with surgically induced endometriosis showed persistent significant reductions in the implant sizes upon intraperitoneal or subcutaneous administration of human IFNα-2b. However, a clinical study showed that intraperitoneal administration of IFNα-2b following surgical treatment for endometriosis increased the cyst reoccurrence rate significantly after 21 months. Hence, the normal anti-proliferative effects of type I IFNs appear to be

The endometrial level of IFNα mRNA is increased in the mid-secretory phase of the human menstrual cycle (Li et al., 2001), which is the putative window of implantation (figure 2). It has been found that the transcription of IFNα and the interferon α receptor 2 (IFNAR2) is highly up-regulated in the endometrium of women suffering from endometriosis compared with healthy women (Kao et al., 2003). Moreover, *JAK1*, which has an important function in type I IFN signaling, is up-regulated in endometriosis stromal cells compared with the endometrial cells from endometriosis patients (Matsuzaki et al., 2006). The sum of these findings suggests a role of type I IFN in the human endometrium and a possible

The possible involvement of type I IFNs in endometriosis has recently been investigated directly (Vestergaard et al., 2011). A type I interferon-specific PCR Array indicated significantly down-regulated transcription of the genes *HOXB2* and *ISG20* in endometriosis lesions compared with endometrium from endometriosis patients and healthy controls, but no difference in the expression of any other interferon stimulated genes were observed. These results were independent of the menstrual phase. As only two out of 84 genes of the type I IFN response was significantly dysregulated in endometriosis, the type I IFNs do not appear to be generally involved in the pathogenesis of endometriosis. Yet, specific gene regulation involving type I IFNs could still play distinct roles in endometriosis. The putative involvement of *ISG20, HOXB2* and *HOXA10*

The transcriptional expression of *ISG20* (interferon-stimulated gene product of 20 kDa, also called *HEM45*) was shown by validated qRT-PCR to be highly down-regulated in endometriosis lesions (Vestergaard et al., 2011). *ISG20* transcription is induced synergistically by type I and II IFNs, induced by estrogen, and regulated in a progesteronedependent manner in mice, especially in the mouse endometrium. This gene encodes a 3′ to 5′ exonuclease with specificity for single-stranded RNA (and to a lesser extent for DNA), which is localized in the nucleus and is associated with promyelocytic leukemia (PML) protein nuclear bodies. The exonuclease has been proposed to down-regulate the estrogendependent transcriptional response by degrading estrogen-induced mRNAs within PML oncogenic domains (PODs). The subcellular localization of ISG20 in the nucleus argues for

ISG20 has also been implicated in the anti-angiogenic properties of interferon (Taylor et al., 2008). In an *in vitro* angiogenic assay system, ISG20 was found to be up-regulated in endothelial cells treated with interferon, however, overexpression of ISG20 did not lead to

its involvement in the maturation rather than in the degradation of mRNAs.

challenged in endometriosis.

dysregulation in endometriosis.

in the pathogenesis will now be discussed.

**3.1 The ISG20 protein** 

indicates that these cytokines could be involved. A key constituent of the immune system that has been only slightly investigated in relation to endometriosis is the family of interferons (IFNs). IFNs are cytokines secreted from cells in response to viral challenge and various other stimuli. The human interferons are by structural homology classified into type I, which consists of IFNα (counting 13 subtypes), IFNβ, IFNε, IFNκ, and IFNω, type II, which is only IFNγ, and type III, the more recently discovered IFNλ (comprising 3 subtypes) (Platanias, 2005; Hall & Rosen, 2010). Secreted IFNs bind to specific plasma membrane receptors, which initiate intracellular pathways leading to the transcription of hundreds of interferon stimulated genes (ISGs) (Platanias, 2005; Hall & Rosen, 2010). In spite of their common signaling pathway, IFNα and IFNβ display specific activity profiles.

Main biological activities resulting from IFN signaling are primarily antiviral but also antiproliferative, anti-angiogenic, and antigen-presenting effects as well as regulation of proand anti-apoptotic genes and proteins (Platanias, 2005). The current state of the cells and complex balances of feedback mechanisms determine the overall outcome of the IFN response, *e.g.* survival vs. apoptosis. Recombinant human IFNs (mainly IFNα and IFNβ) are used as therapeutic agents against a variety of cancers, multiple sclerosis, and viral diseases, such as hepatitis B and C.

Type II IFN is produced only by activated T or NK cells and its primary role is to modulate the adaptive immune response, *e.g.* by contributing to the activation of macrophages and T cell development (Hall & Rosen, 2010). Expression of the receptors of type III IFN is primarily limited to epithelial and DCs, thus restricting their scope of action. We will only focus on type I IFNs and their possible role in endometriosis.

High levels of type I IFNs are secreted rapidly by most cell types upon stimuli (Hall & Rosen, 2010). Type 1 IFNs have immunostimulatory effects on NK cells, macrophages and DCs, which all are essential effector cells in the innate immune system. Among the downstream targets of type I IFN signaling are the IFNs themselves as well as associated receptors, signal transducers and transcription factors enabling a strong feed-forward mechanism. It is strongly believed that type I IFNs are directly involved in the pathogenesis of autoimmune disease by enhancing the self-amplification of systemic autoimmunity (Hall & Rosen, 2010). The aberrant antigen-presentation of debris from apoptotic cells and the resulting production of autoantibodies in endometriosis could be speculated to give rise to a type I IFN-amplified autoimmune pathogenic mechanism. In autoimmunity, self antigens elicit an immune response that includes substantial production of type I IFNs. Upon DC presentation of endometrial autoantigens, type I IFN may then promote monocyte differentiation, DC survival and cytotoxic T cell activity, which may enhance killing of endometrial cells presenting autoantigens. Debris from dying cells is then taken up by DCs and presented for recognition by T cells in a self-amplifying loop. Type I IFNs also induce the differentiation of B cells, which would promote autoantibody production. Immune complexes of endometrial self antigens and autoantibodies would further amplify the IFN production, constituting yet another positive feedback loop.

The approved drug Intron A (human recombinant IFNα-2b) has been proposed as a possible immunomodulatory therapeutic against reoccurrence of endometriosis cysts after surgery. The hypothesis was that this could enhance the cytotoxic activity of macrophages and

indicates that these cytokines could be involved. A key constituent of the immune system that has been only slightly investigated in relation to endometriosis is the family of interferons (IFNs). IFNs are cytokines secreted from cells in response to viral challenge and various other stimuli. The human interferons are by structural homology classified into type I, which consists of IFNα (counting 13 subtypes), IFNβ, IFNε, IFNκ, and IFNω, type II, which is only IFNγ, and type III, the more recently discovered IFNλ (comprising 3 subtypes) (Platanias, 2005; Hall & Rosen, 2010). Secreted IFNs bind to specific plasma membrane receptors, which initiate intracellular pathways leading to the transcription of hundreds of interferon stimulated genes (ISGs) (Platanias, 2005; Hall & Rosen, 2010). In spite of their

Main biological activities resulting from IFN signaling are primarily antiviral but also antiproliferative, anti-angiogenic, and antigen-presenting effects as well as regulation of proand anti-apoptotic genes and proteins (Platanias, 2005). The current state of the cells and complex balances of feedback mechanisms determine the overall outcome of the IFN response, *e.g.* survival vs. apoptosis. Recombinant human IFNs (mainly IFNα and IFNβ) are used as therapeutic agents against a variety of cancers, multiple sclerosis, and viral diseases,

Type II IFN is produced only by activated T or NK cells and its primary role is to modulate the adaptive immune response, *e.g.* by contributing to the activation of macrophages and T cell development (Hall & Rosen, 2010). Expression of the receptors of type III IFN is primarily limited to epithelial and DCs, thus restricting their scope of action. We will only

High levels of type I IFNs are secreted rapidly by most cell types upon stimuli (Hall & Rosen, 2010). Type 1 IFNs have immunostimulatory effects on NK cells, macrophages and DCs, which all are essential effector cells in the innate immune system. Among the downstream targets of type I IFN signaling are the IFNs themselves as well as associated receptors, signal transducers and transcription factors enabling a strong feed-forward mechanism. It is strongly believed that type I IFNs are directly involved in the pathogenesis of autoimmune disease by enhancing the self-amplification of systemic autoimmunity (Hall & Rosen, 2010). The aberrant antigen-presentation of debris from apoptotic cells and the resulting production of autoantibodies in endometriosis could be speculated to give rise to a type I IFN-amplified autoimmune pathogenic mechanism. In autoimmunity, self antigens elicit an immune response that includes substantial production of type I IFNs. Upon DC presentation of endometrial autoantigens, type I IFN may then promote monocyte differentiation, DC survival and cytotoxic T cell activity, which may enhance killing of endometrial cells presenting autoantigens. Debris from dying cells is then taken up by DCs and presented for recognition by T cells in a self-amplifying loop. Type I IFNs also induce the differentiation of B cells, which would promote autoantibody production. Immune complexes of endometrial self antigens and autoantibodies would further amplify the IFN

The approved drug Intron A (human recombinant IFNα-2b) has been proposed as a possible immunomodulatory therapeutic against reoccurrence of endometriosis cysts after surgery. The hypothesis was that this could enhance the cytotoxic activity of macrophages and

common signaling pathway, IFNα and IFNβ display specific activity profiles.

focus on type I IFNs and their possible role in endometriosis.

production, constituting yet another positive feedback loop.

such as hepatitis B and C.

natural killer cells and thus reduce the growth of endometriosis tissue (Acién et al., 2002). Rats with surgically induced endometriosis showed persistent significant reductions in the implant sizes upon intraperitoneal or subcutaneous administration of human IFNα-2b. However, a clinical study showed that intraperitoneal administration of IFNα-2b following surgical treatment for endometriosis increased the cyst reoccurrence rate significantly after 21 months. Hence, the normal anti-proliferative effects of type I IFNs appear to be challenged in endometriosis.

The endometrial level of IFNα mRNA is increased in the mid-secretory phase of the human menstrual cycle (Li et al., 2001), which is the putative window of implantation (figure 2). It has been found that the transcription of IFNα and the interferon α receptor 2 (IFNAR2) is highly up-regulated in the endometrium of women suffering from endometriosis compared with healthy women (Kao et al., 2003). Moreover, *JAK1*, which has an important function in type I IFN signaling, is up-regulated in endometriosis stromal cells compared with the endometrial cells from endometriosis patients (Matsuzaki et al., 2006). The sum of these findings suggests a role of type I IFN in the human endometrium and a possible dysregulation in endometriosis.

The possible involvement of type I IFNs in endometriosis has recently been investigated directly (Vestergaard et al., 2011). A type I interferon-specific PCR Array indicated significantly down-regulated transcription of the genes *HOXB2* and *ISG20* in endometriosis lesions compared with endometrium from endometriosis patients and healthy controls, but no difference in the expression of any other interferon stimulated genes were observed. These results were independent of the menstrual phase. As only two out of 84 genes of the type I IFN response was significantly dysregulated in endometriosis, the type I IFNs do not appear to be generally involved in the pathogenesis of endometriosis. Yet, specific gene regulation involving type I IFNs could still play distinct roles in endometriosis. The putative involvement of *ISG20, HOXB2* and *HOXA10* in the pathogenesis will now be discussed.

#### **3.1 The ISG20 protein**

The transcriptional expression of *ISG20* (interferon-stimulated gene product of 20 kDa, also called *HEM45*) was shown by validated qRT-PCR to be highly down-regulated in endometriosis lesions (Vestergaard et al., 2011). *ISG20* transcription is induced synergistically by type I and II IFNs, induced by estrogen, and regulated in a progesteronedependent manner in mice, especially in the mouse endometrium. This gene encodes a 3′ to 5′ exonuclease with specificity for single-stranded RNA (and to a lesser extent for DNA), which is localized in the nucleus and is associated with promyelocytic leukemia (PML) protein nuclear bodies. The exonuclease has been proposed to down-regulate the estrogendependent transcriptional response by degrading estrogen-induced mRNAs within PML oncogenic domains (PODs). The subcellular localization of ISG20 in the nucleus argues for its involvement in the maturation rather than in the degradation of mRNAs.

ISG20 has also been implicated in the anti-angiogenic properties of interferon (Taylor et al., 2008). In an *in vitro* angiogenic assay system, ISG20 was found to be up-regulated in endothelial cells treated with interferon, however, overexpression of ISG20 did not lead to

Virus Infection and Type I Interferon in Endometriosis 257

(Langendonckt et al., 2010). The endometrial down-regulation of HOXA10 protein in women with endometriosis seems to be due to increased methylation of the *HOXA10* genomic enhancer region in the endometrium leading to epigenetic silencing of this gene (reviewed in Cakmak & Taylor, 2010). In conclusion, low levels of HOXA10 may result in

*HOXB2* is part of the *HOX* gene family involved mainly in embryonic development. A very solid down-regulation of *HOXB2* in endometriotic lesions compared with endometrium from both endometriosis and healthy women has been observed (Vestergaard et al., 2011). Little is known about *HOXB2* expression in the endometrium, but several studies have demonstrated that *HOXB2* expression is altered in tumours. In a xenograft breast tumour mouse model, HOXB2 acts as a negative tumour growth regulator, since *HOXB2* expression decreases proliferation of tumour cells (Boimel et al., 2011). Other results have shown that overexpression of HOXB2 in pancreatic, lung and cervical cancer was associated with malignancy. However, a more in-depth analysis correlated lower HOXB2 expression with higher grades of tumours. Finally, it has been reported that the HOXB2 protein binds the interferon-induced protein p205, involved in the growth inhibitory activities of interferon (see Vestergaard et al., 2011). Whether HOXB2 interaction with p205 modifies the growth inhibitory activities of p205 has not yet been investigated. Further studies are needed to determine the mechanism and implications of the abolished *HOXB2* expression in

The pathogenesis and the similarity to cancer invasiveness suggested that a viral background could be part of the pathogenesis of endometriosis, but so far no investigations have demonstrated this connection in the aetiology. The prevalence of pathogenic dsDNA viruses in the human endometrium was found to be generally low (0-10%), and nothing points towards any evidence that endometriosis is caused by currently known DNA viruses (Oppelt et al., 2010; Vestergaard et al., 2010). It can be speculated that the endometrium and endometriotic tissue is difficult to access or simply an unfavorable environment for virus progression, leading to a generally low prevalence in these deeper tissues. It is possible that viruses can infect the endometrium transiently but subsequently be either shed with the endometrial tissue during menstruation or be rapidly cleared by an efficient immune response. Thus stable infections of the endometrium would not be frequent. However, a pathogenic virus could theoretically initiate a malignant cell process during a shorter infectious period and then flee the scene. This "hit-and-run" strategy has been previously shown *e.g.* for CMV *in vitro* and indicated in clinical studies of both polyomaviruses and papillomaviruses (see Vestergaard et al., 2010). This could explain why no virus DNA so far

To address the viral "hit-and-run" strategy hypothesis, one could analyse for an elevated level of serum antibodies against the viruses, which would then show previous viral infections. Even though a broad selection of the most common pathogenic DNA viruses have been tested for, other more rare or even undiscovered viruses or bacteria might still be involved. Conclusively, the prevalence of pathogenic DNA viruses in the endometrium

resistance to progesterone action in the endometriotic tissue.

has been found associated with endometriosis lesions.

endometriosis lesions.

**4. Conclusion** 

reduced angiogenesis *per se*. However, overexpression of the enzymatically inactive, dominant-negative ISG20 mutant inhibited angiogenesis in this system and potentiated the anti-angiogenic properties of interferon. How ISG20 might be involved in angiogenesis is currently not clear.

The ISG20 protein mediates antiviral effects of interferons by inhibiting the replication of several RNA viruses, like vesicular stomatitis virus, influenza virus, encephalomyocarditis virus, West Nile virus, Dengue virus, hepatitis A and C viruses, yellow fever virus, and bovine viral diarrhea virus (Zhou et al., 2011). The antiviral activity of ISG20 is only observed with enzymatically active ISG20 expression, since expression of an ISG20 mutant without enzymatic activity, did not possess the same antiviral activities.

Importantly, ISG20 mRNA was found to be up-regulated in the uterine epithelium during the implantation window in mouse (Pan et al., 2006). Whether abolished ISG20 enzyme activity is implicated in the pathogenesis of endometriosis or a marker of altered hormonal expression, or both, needs to be further investigated.
