Peptidylarginine deiminase enzymes and citrullinated proteins in female reproductive physiology and associated diseases

Abstract

Citrullination, the post-translational modification of arginine residues, is catalyzed by the four catalytically active peptidylarginine deiminase (PAD or PADI) isozymes and alters charge to affect target protein structure and function. PADs were initially characterized in rodent uteri and, since then, have been described in other female tissues including ovaries, breast, and the lactotrope and gonadotrope cells of the anterior pituitary gland. In these tissues and cells, estrogen robustly stimulates PAD expression resulting in changes in levels over the course of the female reproductive cycle. The best-characterized targets for PADs are arginine residues in histone tails, which, when citrullinated, alter chromatin structure and gene expression. Methodological advances have allowed for the identification of tissue-specific citrullinomes, which reveal that PADs citrullinate a wide range of enzymes and structural proteins to alter cell function. In contrast to their important physiological roles, PADs and citrullinated proteins are also involved in several female-specific diseases including autoimmune disorders and reproductive cancers. Herein, we review current knowledge regarding PAD expression and function and highlight the role of protein citrullination in both normal female reproductive tissues and associated diseases.

Peptidylarginine deiminase enzymes are expressed in female reproductive tissues and post-translationally convert arginine residues in target proteins to citrulline.

Graphical Abstract

Graphical Abstract.

Introduction

Citrullination is the post-translational modification of arginine to the non-coded residue citrulline. Citrullinated proteins were first discovered in hair follicles in 1962, which subsequently led to the identification of a family of calcium-dependent enzymes named peptidylarginine deiminases (PADs or PADIs) [1–3]. Since their initial characterization, five PAD isozymes have been identified: PAD1, 2, 3, 4, and 6. The PAD isozymes are well conserved at the nucleotide and amino acid level, and PAD2 is thought to be the ancestral homolog of the gene family with the remaining members arising through a series of gene duplications [4, 5]. PAD enzymes are widely expressed across vertebrates, but absent in yeast, worms, and flies. Phylogenetic analysis suggests that PADs were introduced from cyanobacteria into animals by horizontal gene transfer indicating an unusual evolution of the gene responsible for this protein modification [6]. The PAD genes are clustered on the human chromosome 1p36.1 with PAD2 oriented in the opposite direction of PAD1, 3, 4, and 6 and this unique genomic arrangement is conserved across species [4, 5].

In this article, we first provide biochemical details regarding PAD enzyme structure and function followed by a discussion of what is currently known about PAD tissue distribution and targets for protein citrullination. Next reviewed are PAD expression and function in the anterior pituitary, uterus, ovary, and mammary gland and involvement in female reproduction. Lastly, we provide an overview of our current understanding of the function of PADs and citrullinated proteins in autoimmunity and female reproductive cancers.

Biochemistry of protein citrullination

PADs catalyze the hydrolytic conversion of positive peptidylarginine into neutral peptidylcitrulline in substrate proteins (Figure 1). Citrullination is also referred to as deimination because the imine group on peptidylarginine is exchanged for a carbonyl group, which leads to a 1 Da decrease with each modification [7, 8]. Because an arginine’s guanidinium group has one of the highest pKa values (pKa = 12.5), at physiological pH the residues are protonated and carry a positive charge. This positive charge allows for electrostatic interactions between an arginine and negatively charged moieties including the R-groups of acidic amino acids and the phosphate backbone of DNA. Citrullination elicits a net loss in positive charge, which alters the target’s inter- and/or intramolecular electrostatic interactions to affect protein structure and function (Figure 2) [4, 9, 10].

Figure 1.

PAD enzymes remove the positively charged imine group from arginine (red) resulting in the neutrally charged, non-coded, residue citrulline (blue).

Figure 2.

Conversion of a positive arginine (red) to neutral citrulline (blue) by PAD enzymes can alter intra- and inter-molecular interactions of citrullinated proteins, mediate proteolysis and immunogenicity, change the active site, and regulate chromatin structure.

The PADs require calcium (Ca²⁺) binding for full catalytic activity, and this is an essential property defining the enzyme family [10, 11]. Each PAD monomer can bind up to six calcium ions, which transforms a disordered cavity into a stabilized, catalytically competent active site [12, 13]. Calcium binding is thought to occur in a sequential fashion resulting in a conformational change that positions a critical nucleophilic cysteine residue into the active site [14]. As a result, PAD activity increases up to 10 000-fold compared to that of the calcium-free enzyme [12]. Optimal catalytic activity of PADs in vitro occurs at calcium concentrations in the 0.1–2 mM range, which is 10- to 10 000-fold higher than typical cellular physiologic concentrations that vary from 0.1 to 10 μM [13, 15, 16]. It has been proposed that PADs may temporarily localize near intracellular ion channels where calcium concentrations can reach millimolar levels allowing for maximal activation [13, 17, 18]. Work by Mechin et al. found that the calcium concentration required for optimal PAD activity is substrate dependent. Calcium requirements for PAD1, 2, and 3 activity varied for synthetic arginine substrates and filaggrin, suggesting that the structure of the target can modulate catalytic activity [19].

Interestingly, the binding of calcium also alters PADs in ways other than just mediating catalytic activity. Zheng et al. found that PAD2 interacts with cytoplasmic Annexin A5 in the absence of calcium, but that in the presence of calcium this interaction is weakened thereby allowing PAD2 to preferentially associate with ras-related nuclear proteins (RAN), which is well documented to facilitate nuclear transport [20]. As it does not contain a nuclear localization sequence, this unique calcium-mediated mechanism facilitates PAD2 nuclear translocation [21]. PAD calcium-binding sites are structurally conserved across isozymes, and this region is thought to act as a molecular switch that can be regulated by protein binding [13]. In support of this, antibodies that bind to PAD4 near this region decrease the requirement of calcium for catalytic activity [22].

Crystallographic studies demonstrate that PAD2, 3, and 4 can form homodimers in which two monomers bind in a head-to-tail fashion [23–25]. The catalytic activity and target specificity of PAD monomers as compared to homodimers remain unknown. This observation coupled with the high levels of homology between the isoforms advances another question for future research: Can PADs form heterodimers and, if so, how does this affect catalytic activity and substrate specificity?

PAD tissue distribution

PADs are expressed in a wide range of tissues and cells. PAD1 and 3 predominantly localize in the epidermis, hair follicles, and keratinocytes [4, 26–28]. PAD1 mRNA is present in all layers of the epidermis as well as the prostate, testis, placenta, spleen, and thymus [28, 29]. PAD2 is the most ubiquitously expressed and found in brain, skeletal muscle, skin, spleen, and leukocytes [4, 30]. PAD4 is expressed in the spleen, neutrophils, granulocytes, monocytes, and eosinophils [31–33]. PAD6 is the only catalytically inactive PAD isozyme, and it localizes in mouse oocyte cytoplasm and early embryos as well as in testis and peripheral blood leukocytes [5, 34, 35].

Some of the earliest work in the field characterized PAD expression patterns in female reproductive tissues [36–38]. Senshu et al. discovered that prepubescent female and male rat pituitaries have negligible PAD activity, whereas sexually mature females show markedly higher activity that correlates with age and stage of the estrous cycle [38]. Additionally, PAD mRNA transcript levels are 50-fold higher in female rat pituitaries as compared to the male [39, 40]. This sexual dimorphic pattern is most likely because PAD expression is strongly stimulated by 17β-estradiol (E2) but suppressed by androgens [36–38, 41]. In uteri, mammary glands, and the anterior pituitary gland, PAD expression fluctuates across the estrous cycle, with highest expression during estrus following elevated serum E2 in proestrus [9, 37, 40]. Many studies now support that PADs are expressed in female reproductive tissues across the lifecycle (Figure 3) [38, 42].

Figure 3.

PAD expression in female reproductive tissues.

Hormonal regulation of PAD expression

Estrogen regulation of PAD expression in female reproductive tissues is well documented. Takahara et al. was the first to show that PAD expression is low in mouse uteri following ovariectomy; however, expression is rescued in uterine luminal and glandular epithelial cells in a time and dose-dependent manner following treatment with exogenous estrogens [37]. Subsequent ovariectomy experiments show that treatment with exogenous E2 stimulates PAD expression in the uterus and mammary gland as compared to vehicle-treated controls [9]. Following exogenous E2 treatment, PAD expression is absent in uteri of estrogen receptor α (ESR1) knockout (KO) and ESR1 mutant mice, which contain an ESR1 that cannot bind estrogen response elements (EREs), but retains ligand binding, dimerization, and interactions with other transcription factors. These findings suggest that ESR1 mediates PAD expression [43]. Supporting this idea, chromatin conformation capture experiments investigating uterine development show that the Pad gene locus contains an ESR1 super-enhancer [44]. At the transcriptional level, the Pad 1, 3, and 4 gene promoters contain putative EREs [45]. Although its expression is highly E2 dependent in vivo, the Pad2 gene promoter lacks an ERE. The promoter does contain multiple binding sites for the transcription factor SP1, which may interact with ESR1 to promote PAD2 expression following E2 stimulation [46, 47].

While E2 stimulates PAD expression, other hormones such as progesterone can alter PAD activity. Ovariectomized mice treated for 30 h with exogenous progesterone do not show increases in PAD activity in uterine lysates compared to vehicle controls; however, progesterone does dampen PAD activity following E2 priming [37]. More recently, studies found that hormones can induce rapid calcium influx that promotes PAD2 nuclear translocation resulting in increased histone citrullination. In the gonadotrope-derived LβT2 cell line, treatment with buserelin, a gonadotrope releasing hormone (GnRH) agonist, induces a rapid calcium influx resulting in the translocation of PAD2 to the nucleus [40] with a similar finding following progesterone treatment of an ovine uterine luminal epithelial (OLE) cell line [48]. Further, progesterone stimulation of human neutrophils does not increase PAD4 expression compared to vehicle-treated cells, but does increase levels of citrullinated histone 3 [49]. Taken together, these studies suggest that E2 stimulates PAD expression but that progesterone treatment may increase intracellular calcium, allowing PADs to translocate to the nucleus and citrullinate histones.

Protein targets for citrullination

Each PAD enzyme preferentially citrullinates specific arginine residues in multiple proteins (i.e., the citrullinome), though target specificity can overlap between isozymes [10, 19, 23, 50–52]. Mass spectrometry (MS) has identified the citrullinome in a number of tissues [53–55], which shows that targets include metabolic enzymes, clotting factors, cytoskeletal proteins, transcription factors, and blood proteins [53, 54, 56–59]. While the functional consequence of citrullination on protein function is known for some targets such as histones and filaggrin, for most proteins the impacts of citrullination are still unclear. For example, citrullinated tubulin has been detected by MS, yet how this post-translational modification alters cytoskeletal dynamics and architecture is still not well defined. By far the best-characterized functional consequence of protein citrullination is the decreased affinity of citrullinated histones for DNA, which leads to changes in chromatin organization and gene expression [48, 60–65]. Although histone citrullination has clear physiologic importance, aberrant protein citrullination is increasingly associated with diseases. Hyper-citrullination of histones, as occurs during neutrophil extracellular trap (NET) formation, leads to large webs of disorganized chromatin that are extruded from the cell and function as part of the intrinsic immune response [66, 67]; however, NETs are also implicated in both autoimmune disorders and cancer progression [68–72]. Another characterized citrullinated protein is filaggrin, which helps aggregate keratin filaments in the epidermis. Citrullinated filaggrin is a component of the cornified envelope formed during terminal differentiation of keratinocytes and participates in epidermal barrier formation [27, 73, 74]. While numerous proteins are citrullinated, the effect of this post-translational modification at the protein and cellular level and consequences on reproductive function is still unresolved. The remainder of this article will comprehensively review the role of PADs and citrullinated proteins in female reproductive tissue.

PADs and citrullinated proteins in female reproductive physiology

PADs were originally characterized in the rodent uterus and pituitary gland in the late 1980s and early 1990s and have subsequently been discovered in reproductive tissues of several female mammals including humans, rodents, cats, dogs, and sheep [9, 61, 62, 75, 76]. Since the original characterization of expression patterns, work has shifted to focus on determining the function of PADs and citrullinated proteins in female reproductive tissues. In E2 regulated female tissues and cells, increased PAD expression results in elevated levels of protein citrullination, suggesting an important role for this post-translation modification in reproductive physiology. The sections to follow detail our current understanding of PADs and protein citrullination in female reproductive tissues.

Anterior pituitary gland

Seminal studies by Senshu et al. characterizing PAD expression and activity in the pituitary laid the groundwork for ongoing studies investigating the function of PADs and citrullinated proteins in this gland [38]. Of the five hormone secreting anterior pituitary gland cell types, PADs are best characterized in gonadotrope and lactotrope cells, which are both highly E2 regulated.

In response to hormonal inputs, gonadotrope cells synthesize and secrete luteinizing hormone (LH) and follicle stimulating hormone (FSH), which are required for follicle development and ovulation in all female mammals. PAD2 expression is highest in mouse gonadotrope cells during estrus, with lower but detectable levels of PAD4 [40]. PAD2 is also expressed in the gonadotrope-derived LβT2 cell line, and 30 min of GnRH agonist treatment results in the formation of distinct puncta of PAD2 in the nucleus. This nuclear translocation is dependent on GnRH-induced calcium influx through L-type voltage-gated calcium channels [21, 40]. Within 30 min of GnRH stimulation, PADs citrullinate histone H3 tail arginine residues at sites 2, 8, and 17, which increases transcription of the Lhb and Fshb genes [40]. Despite these results, many questions remain regarding PAD function in gonadotrope cells such as the full cohort of genes regulated by histone citrullination as well as the identity of other citrullinated proteins.

Lactotrope cells synthesize prolactin and undergo controlled hypertrophy and hyperplasia during pregnancy in response to increasing serum E2. This hormone-dependent transformation of the lactotrope population is vital to maximize prolactin synthesis, initiate lactation, and stimulate milk production by mammary secretory cells. Akiyama et al. first showed that PADs localize to lactotrope cells and that catalytic activity in rat pituitaries increases over 8-fold from pregnancy day 12 to 21, paralleling a 4-fold increase in serum E2 across the same period [77, 78]. PADs have also been studied in the lactotrope-derived MtT/S and GH3 cell lines, which were derived from lactotropes and somatolactotropes, respectively. In MtT/S cells, insulin increases PAD expression and activity prior to induction of prolactin synthesis [79]. In GH3 cells, PAD2 catalyzed histone citrullination suppresses the expression of miRNAs Let-7c, miR-23b, and miR-29c [61]. These miRNAs bind the 3′untranslated regions of important target mRNAs involved in cell proliferation and growth and decrease their expression. This histone citrullination/miRNA expression mechanism may regulate lactotrope hypertrophy and hyperplasia during pregnancy to support the increased production of prolactin needed to initiate lactation. In support of this idea, inhibition of PADs with the pan-PAD inhibitor biphenyl-benzimidazole-Cl-amidine (BB-ClA) results in a significant decrease in GH3 cell growth and proliferation. Further work is clearly warranted to elucidate the mechanism by which PAD catalyzed citrullination regulates lactotrope function during pregnancy and lactation.

Uterus

Initial genomic analysis revealed that PAD1, 2, and 4 mRNA levels are highest in mouse uterine tissue as compared to 50 other tissue types examined [80]. Single-cell RNA sequencing confirms this earlier work showing that PAD1, 2, and 4 are expressed in mouse uterine epithelia while PAD2 and 4 are the major isoforms in humans [81, 82]. PAD1, 2, and 4 are expressed in luminal and glandular uterine epithelial cells and levels change over the course of the estrous cycle [9, 30, 75, 83, 84]. Expression of both PAD2 and 4 peaks in the estrus phase in rodents while PAD1 is elevated in proestrus. Estrogen-dependent PAD expression was first observed in mouse uterine luminal and glandular epithelia and has been confirmed by additional studies [9, 37]. Work by Ledee et al. found that endometrial biopsies collected during the mid-luteal phase overexpress PAD1 in women with implantation failure following IVF/ICSI compared to fertile controls [85]. While this finding supports an important role for endometrial PAD1 in implantation, the mechanism underlying this defect is unknown.

Uterine PAD expression also fluctuates during pregnancy and parturition [86]. PAD2 and 4 are expressed in ewe uterine luminal and glandular epithelia at day 25 of gestation [62]. PADs most likely function to epigenetically regulate gene expression as ewe gestation day 25 uterine lysates and the OLE cell line both display high levels of citrullinated histone H3 tail residues at positions 2, 8, 17, and 26. Our previous work shows that progesterone stimulation of OLE cells induces citrullination of histones associated with ovine IGFBP1, which is an important component of the secreted histotroph fluid that is critical during early pregnancy. Mechanistically, progesterone stimulation triggers rapid calcium influx through L-type calcium channels, required for PAD2 nuclear translocation and, ultimately, histone citrullination [48].

In mice, PADs are expressed in decidual cells as they differentiate from endometrial stromal cells surrounding the implantation site at day 9, while at day 10 strong expression is detected in the decidua capsularis and along the cytotrophoblast-labyrinth structure [86]. PAD enzymatic activity in uterine lysates increases dramatically in mid-pregnancy (~days 9–11) before gradually decreasing near term [86]. Single-cell RNA sequencing studies found that during implantation, PAD1 is expressed in mouse glandular epithelial cells while PAD2 expression is detected in epithelial, plasmacytoid dendritic, and natural killer cells [87–89]. In humans, PAD1 and 2 are expressed in placental cytotrophoblast cells [90, 91]. To date, no work has examined PAD2 protein expression during placentation [28, 29]. To address this gap in knowledge, we examined PAD2 localization in different placental types across species (Figure 4). Immunofluorescence staining reveals that PAD2 localizes to fetal trophoblast cells in cotyledonary (sheep), diffuse (horse), zonary (dog), and hemochorial (Guinea pig) placentas, suggesting PAD2 plays a conserved role in placental cell differentiation. Clearly, additional work is necessary to determine the functional role of PAD2 in placentation.

Figure 4.

PAD2 (green) is expressed in placentas from different species. Placental tissue samples were collected under approval of the Colorado State University IACUC committee (protocol number 1303). Individual sheep, horse, dog, and guinea pig placentas were obtained at gestational days 75 (sheep) and at term (horse, dog, guinea pig) then fixed in 4% paraformaldehyde and paraffin embedded. Five micron sections were processed for H&E staining or routine immunofluorescence using an anti-PAD2 antibody (1:100 ProteinTech, 12110-1-AP) and imaged using Olympus BX63 fluorescence microscope. Blue nuclear staining is DAPI, small inserts are no primary antibody negative controls, and scale bar represents 20 μm. Light blue in zonary and hemochorial is autofluorescence from red blood cells.

Ovaries

In the ovary, PAD6 is predominantly found in maturing oocytes and early embryos [34, 92]. Although PAD6 is not catalytically active, female PAD6 KO mice are infertile, while males are fertile [93]. A similar phenotype occurs in women who have PAD6 mutations, which result in premature embryonic arrest and infertility [94, 95]. PAD6 expression is regulated by the oocyte-specific transcription factor NOBOX (newborn ovary homeobox) and downregulated in NOBOX KO mouse ovaries [96]. Mutations in PAD6 result in abnormal early embryo cleavage events [97]. PAD6 KO mouse oocytes display reduced content of acetylated microtubules, abnormal microtubule structure, diffuse distribution of endoplasmic reticulum and mitochondria, and abnormal distribution of organelles during maturation as compared to wild-type oocytes [98]. PAD6 is a component of the subcortical maternal complex in the oocyte cytoplasm, which is required for early embryo development. Studies also indicate that PAD6 is involved in genomic imprinting disorders such as Beckwith-Wiedemann syndrome [99].

Citrullinated histones are present in mouse oocytes and preimplantation embryos, suggesting that catalytically active members of the PAD family, beyond PAD6, function in oocyte development and embryogenesis [100]. When PAD activity is inhibited in pronuclear phase zygotes with a pan-PAD inhibitor, Cl-amidine, histone tail citrullination is reduced and early cleavage blocked. PAD inhibition and PAD1 depletion by morpholinos lead to a decrease in citrullination in 2 and 4 cell embryos, the loss of phosphorylation of RNA polymerase II at serine residue 2, and a decrease in overall transcription. The authors conclude that PAD1 catalyzed citrullination of histone tails facilitates early embryo gene transactivation [100]. PAD4 is also present in oocytes and preimplantation embryos, and PAD4 knockdown delays embryonic development, while PAD inhibition arrests embryos at the first cleavage [101, 102]. It should be noted that Cl-amidine irreversibly inhibits all PADs, so results obtained using this inhibitor cannot be attributed to a specific isozyme. Isozyme-specific PAD inhibitors are under development, which will undoubtedly be valuable in parsing out the relative roles of different PADs in all tissues including the ovary [103–107].

Mammary gland

The canine, feline, rodent, and human mammary glands all express PAD enzymes. During canine anestrus and estrus, PAD2 protein is sparse in mammary epithelial cells; however, during late diestrus PAD2 is widely expressed throughout the canine mammary epithelia [108]. Levels of citrullinated histones are also highest in the canine mammary gland during diestrus. In mice, PAD2 is expressed in mammary epithelial cells throughout the estrous cycle, peaking in estrus, while PAD4 expression is lower but detectible during this phase [9]. PAD2 is also expressed in human breast epithelial cells [76]. E2 strongly stimulates PAD2 and 4 expression in mammary epithelial cells of ovariectomized mice compared to placebo-treated controls [9]. PAD3 is present in the lactation day 9 mouse mammary gland [109], and studies show that prolactin stimulates PAD3 expression in the CID-9 cell line derived from mid-pregnant mouse mammary epithelial cells. In CID-9 cells, prolactin activates the JAK2/STAT5 signaling pathway to upregulate PAD3 mRNA and protein expression. Using mass spectrometry, we identified the lactation day 9 mouse mammary gland citrullinome, which contains 107 citrullinated proteins including histone H2A, α-tubulin, and β-casein. Interestingly, in vitro citrullination of purified tubulin appears to alter the overall polymerization rate of microtubules, suggesting that tubulin citrullination may play a role in vesicle trafficking or milk secretion by mammary secretory cells [53].

The most characterized target for PAD-catalyzed citrullination in mammary gland cells is histone arginine residues; however, the majority of this information stems from investigations using the breast cancer cell lines [60, 110, 111]. In MCF7 cells, E2 stimulates PAD2-catalyzed citrullination of histone H3 arginine residue 26 (H3R26). This histone modification subsequently colocalizes with ESR1 at decondensed chromatin loci surrounding EREs of target gene promoters [112]. Interestingly, the citrullination of H3R26, which upregulates gene expression, reduces the methylation of H3K27 that normally represses gene expression [63]. These findings indicate that there are both direct and indirect consequences of PAD activity on epigenetic gene expression. Li et al. detected citrullinated histone H2A in the lactating mouse mammary gland and investigated the functional significance of this observation using CID-9 cells [53]. Stimulation of the CID-9 cells with prolactin increases not only histone H2A citrullination but also the expression of the lactation-related β-casein (Csn2) and butyrophilin (Btn1a1) genes. The full cohort of genes regulated by the histone citrullination mechanism is unknown as are the functions of the additional citrullinated proteins (e.g., the major milk protein β-casein) detected in the lactating mouse mammary gland.

PADs and citrullinated proteins in disease

Citrullination and autoimmunity

PAD enzymes and citrullinated proteins are increasingly associated with the etiology of female-biased autoimmune diseases such as systemic lupus erythematosus, Sjogren syndrome, multiple sclerosis, and rheumatoid arthritis (RA); yet, citrullinated proteins are also implicated in non-sex-biased autoimmune disorders including type 1 diabetes and psoriasis [113–124]. A mechanistic role for citrullinated proteins in female-biased autoimmune disorders is potentially due to estrogen-mediated PAD expression combined with the fact that citrullinated proteins can act as antigens to stimulate autoantibody production.

Anti-citrullinated protein antibodies in female rheumatoid arthritis

Women have a 3-fold higher incidence of RA, develop the disease at a younger age, and have worse clinical outcomes as compared to men [125, 126]. RA can be clinically diagnosed by the presence of anti-citrullinated protein antibodies (ACPAs) in blood. ACPAs can be present 3–5 years before onset of clinical symptoms and are highly predictive of disease severity [127–131]. Strengthening this connection is the fact that genetic mutations that upregulate PAD expression are associated with increased RA disease risk [132–134]. A woman’s RA risk is correlated with the frequency and duration of exposure to female reproductive hormones. Moreover, RA flares are associated with specific phases of the menstrual cycle and the incidence of RA is higher in pre- versus post-menopausal women [125, 135, 136]. Additional RA risk factors include parity, the postpartum period, birth control type, and early age of menarche [137–139]. Given that PAD expression is stimulated by E2 in female reproductive tissues and suppressed by androgens, it follows that this would likely result in higher levels of citrullinated proteins and potentially increase exposure to autoantigens. Unfortunately, our understanding of the role of sex hormones in RA has been inconclusive and has not substantially improved our understanding of disease etiology. For instance, the synovial joint citrullinome from RA patients contains non-sex-specific citrullinated protein antigens such as fibrinogen, vimentin, and α-enolase [51, 54, 140, 141]. Therefore, identifying novel, sex-specific citrullinated protein antigens and determining their relative levels in key populations, such as women, is a key goal in the ongoing search for diagnostic tests and therapies to improve clinical outcomes for women with RA.

PADs in cancer

PADs are expressed in numerous cancers including those of the brain, colon, breast, liver, lung, esophagus, and kidney [142–144]. Alterations in PAD expression result in aberrant changes in the level of citrullinated proteins including histones and transcription factors that have a wide range of oncogenic effects including epigenetically regulating tumor gene expression, promoting epithelial-to-mesenchymal transition (EMT), and progression to metastatic disease [145–147]. Among others, extracellular matrix (ECM) proteins, cytoskeletal filaments, and enzymes such as α-enolase are citrullinated in tumor cell lines [148, 149]. In addition to α-enolase, the glycolytic enzyme pyruvate kinase M2 is citrullinated by PAD1 and PAD3 resulting in increased glycolysis during proliferation of multiple types of cancer cell lines [150, 151]. The following section will focus on our current understanding of PADs and citrullinated proteins in cancers of the female reproductive tissues.

Uterine and ovarian cancers

Just as the physiological role of PADs in uterine tissues is not well defined, neither is the role of citrullination in uterine cancer. Analysis of mRNA profiles from endometrial cancer samples show high levels of PAD2 as compared to other isoforms, while PAD4 is present in uterine cervicitis [144, 152]. To increase our knowledge of PAD expression in uterine cancers, we examined a US Biomax T094 human uterine tumor array by immunohistochemistry. While PAD2 and 4 are present in normal endometrial tissue, only 2 of 14 grade 2 malignant endometrial adenocarcinoma samples had detectable expression (Figure 5). To our knowledge, the only mechanistic study of PAD function in endometrial cancer shows that PAD2 citrullinates MEK1, which activates ERK1/2 to promote insulin-like growth factor 2 mRNA binding protein 1 (IGF2BP1) mediated SOX2 mRNA stability in the human endometrial cancer Ishikawa cell line [152].

Figure 5.

PAD2 and 4 are expressed in normal human endometrial tissue (n = 2/ out of 4 cores) and a subset of endometrial adenocarcinomas (n = 2/ out of 14 cores). Serial uterine tumor array slides were purchased from US Biomax (T094) and contain 12 cases with 24 cores and include associated pathology grade. Tumor array slides were examined by immunohistochemistry using an anti-PAD2 antibody (1:100 ProteinTech, 12110-1-AP), anti-PAD4 (1:100 Millipore Sigma, P4874) or an equal mass of non-specific IgG (data not shown). Images were taken with a Zeiss AxioVert.A1 with AxioCam MRc and scale bar represents 100 μm.

Slightly more is known about PAD expression and function in ovarian cancers. Ovarian tumors cultured in the presence of E2 express PAD4 in cystadenocarcinomas and dysgerminomas; however, PAD4 expression is low or non-detectable in other ovarian cancers such as granulosa cell tumors and malignant thecomas [153]. In PAD4 expressing ovarian cancer subtypes, the full role of PAD4 in tumorigenesis remains unclear, but it appears to function in the p53 pathway and contributes to tumor metastasis by regulating the expression of IGF1 and WAS/WASL-interacting protein family member 1 (WIPF1) genes [154]. As for other isozymes, PAD2 expression is associated with poor prognosis for epithelial ovarian cancer, and knockdown of endogenous PAD2 in A2780 and SKOV3 ovarian cancer cell lines suppresses colony formation, proliferation, migration, and invasion [155]. Given the link between hormonal regulation of these cancers and PAD expression, understanding the function of PADs in uterine and ovarian cancers remains an intriguing area for future research.

Breast cancer

Breast cancer is by far the most-characterized female reproductive cancer associated with PADs and citrullinated proteins. PAD4 is expressed in breast carcinomas and the original groundbreaking work that characterized PAD4 as an epigenetic enzyme utilized the MCF7 breast cancer cell line [60, 144]. In these cells, citrullination of histone H3 arginine residues is directly associated with E2-responsive gene expression [60]. Studies using these cells also demonstrate that PAD4 interacts with Elk1 to activate c-Fos gene expression and is predictive of actively transcribed genes [156, 157]. PAD4 can also downregulate the expression of the transcription factors NANOG and Oct4, thereby regulating breast cancer stem cells [158]. As for other citrullinated proteins, studies suggest that they have a functional role in cancer metastases. Shi et al. found that in breast cancer 4T1 cells high levels of PAD4 facilitate cancer extracellular chromatin networks, which promote lung metastasis in a xenograft model [110].

PAD2 is expressed in human, feline, and canine mammary cancers and is the most highly expressed isoform in primary breast tumors and luminal breast cancer cell lines [76, 159, 160]. In MCF7 cells, knockdown of endogenous PAD2 regulates the expression of multiple genes implicated in tumorigenesis including pleiotrophin (PTN) and melanoma antigen A12 (MAGEA12) [111]. PAD2 also promotes EMT and abnormal cell migration during tumorigenesis [161, 162]. Elegant work by Sharma et al. discovered that PAD2 citrullinates RNA polymerase II, which is required to overcome pausing and initiate transcription of genes important for cell cycle regulation and proliferation in the T47D breast cancer cell line [163]. The single-nucleotide polymorphism rs10788656 in PAD2 confers susceptibility to breast cancer, and PAD2 is hypothesized as a potential breast cancer biomarker and therapeutic target [161, 164]. Taken together, these studies indicate that citrullination is a component of the etiology and progression of breast cancer, but further work is needed to discover all the mechanisms involved to develop effective citrullination-specific strategies for detection and treatment of breast cancer.

Pituitary tumors

PAD2 and 4 and citrullinated histones are present in human prolactinomas and somatoprolactinomas [61]. Expanding upon this, we examined additional pituitary tumor subtypes for PAD2 expression by immunohistochemistry and found that PAD2 is present in non-functioning, corticotrope-, gonadotrope-, and somatotrope-derived tumors (Figure 6). Mechanistically, PAD function has only been investigated in the somatoprolactinoma-derived GH3 cell line. As discussed above, DeVore et al. discovered that PAD inhibition induces expression of miRNAs let-7c, miR-23b, and miR-29c, suggesting that PAD-catalyzed histone citrullination normally represses the expression of these pri-miRNA genes. Functionally, let-7c and miR-23b suppress the expression of the oncogene high mobility group AT-hook 1 (HMGA1), which is highly implicated in the pathogenesis and proliferation of human prolactinomas and somatoprolactinomas [61, 165]. Directly connecting PADs and miRNAs, HMGA1 mRNA and protein expression are suppressed following PAD inhibition but are restored by transfection of GH3 cells with let-7c and miR-23b antagomirs. Collectively, these studies suggest that histone citrullination suppresses let-7c and miR-23b expression, leading to increased HMGA1. Further work is clearly needed to elucidate the mechanism by which PAD-catalyzed citrullination regulates proliferation in pituitary adenomas.

Figure 6.

PAD2 expression is present in multiple subtypes of pituitary tumors. Pituitary tumor samples including subtype pathology were obtained from Institut D’Investigacions Biomediques August Pi i Sunyer Biobanc (Barcelona, Spain). Samples were de-identified and thus were exempt from University of Wyoming institutional review board (IRB) approval (protocol number 20140814BC00496). Tumors (n = 4 /tumor subtype) were examined by immunohistochemistry using an anti-PAD2 antibody (1:100 ProteinTech, 12110-1-AP) or an equal mass of non-specific IgG. Images were taken with a Zeiss AxioVert.A1 with AxioCam MRc and scale bar represents 50 or 100 μm.

PAD inhibitors as therapeutic modalities

Since PADs function in autoimmune diseases and the pathogenesis, invasion, and metastasis of tumors, PAD inhibitors have generated interest as potential therapeutic modalities [164, 166, 167]. To date, the most well-characterized pan-PAD inhibitor is BB-ClA, having been investigated in a number of different disease models [166, 168, 169]. In the human breast cancer MCF10DCIS cell line, BB-ClA treatment suppresses tumor spheroid growth by altering expression of p21, GADD45α, and Ki67, which are implicated in tumor progression [164]. In tamoxifen-resistant MCF7 cells, there is a dramatic upregulation of PAD2 expression; however, PAD2 depletion or BB-ClA treatment increase sensitivity to tamoxifen and efficacy of docetaxel [166, 170]. While the treatment of feline and canine mammary cancer cells with BB-ClA activated the endoplasmic reticulum stress pathway and reduced tumor viability in vitro, treatment of mouse xenografts with BB-ClA did not significantly reduce tumor size [166].

Since PAD inhibitors may represent adjuvant therapeutic options for treatment of breast and a range of other cancers, there is a strong push to develop and characterize new reversible and irreversible PAD inhibitors such as streptonigrin, NSC 95397, streptomycin, minocycline, and chlortetracycline [23, 164, 166]. Of these, streptonigrin, NSC 95397, and chlortetracycline already have documented anticancer/antitumor activity [171–175]. For example, streptonigrin is an antitumor antibiotic with function in lymphomas and carcinomas (reviewed in [176]) and was recently shown to be effective against renal carcinoma cells in an in vitro pre-clinical model [177].

PAD inhibition also disrupts NET formation, which is implicated in inflammatory events, tumorigenesis progression, and metastasis [69, 168, 178]. Tumors produce the chemokine receptor agonists CXCR1 and CXCR2, which stimulate NETosis and interfere with immune cytotoxic attack, while inhibition of NETosis sensitizes tumors to PD-1 and CTLA-4 dual checkpoint blockade [179]. NETs can interact with tumor cells to promote metastatic progression at secondary sites [180], and PAD4 KO mice, which are NET deficient, have decreased levels of spontaneous liver tumor metastasis compared to wild-type mice [181]. Collectively, these findings underscore the therapeutic potential of PAD4-specific inhibitors with acceptable treatment safety profiles for inhibition of NETosis.

It is highly likely that PAD inhibitors will also prove beneficial for the treatment of autoimmune disorders, given the well-documented relationship between citrullination, inflammation, and autoimmunity [4, 9, 182]. Specifically, PAD inhibitors can modulate the immune response and show promise in several animal models of autoimmunity. For instance, in lupus MRL/lpr mice, BB-ClA and Cl-amidine protect against lupus-associated kidney, skin, and vascular disease [168]. BB-ClA also prevents the development of diabetes in non-obese diabetic (NOD) mice, which are a model for autoimmune type I diabetes [183]. In a mouse collagen-induced arthritis (CIA) model, Cl-amidine treatment reduces the levels of synovial citrullinated proteins while decreasing clinical disease activity by approximately half [169]. Finally, Cl-amidine, given prophylactically and after disease onset, reduces disease symptoms in a mouse model of colitis [184]. Thus, there is a growing body of data to support the use of PAD inhibitors with acceptable safety profiles in clinical trials as adjuvant therapeutics for the treatment of cancer and autoimmune disorders.

Citrullinated protein cancer vaccines

A novel cancer therapy under development is vaccination with citrullinated proteins frequently found in tumors. The intermediate filament vimentin is citrullinated during EMT in many epidermal-derived tumors. When injected, citrullinated vimentin peptides are presented by major histocompatibility complex II (MHC-II) molecules and stimulate a CD4 T-cell-specific response. This therapy provides a survival benefit up to 14 days after tumor implantation in HLA-DR4 mice, which contain a MHC-II molecule with a humanized peptide-binding domain [185]. Injection with citrullinated enolase also generates an antitumor response against melanoma cells in C57Bl/6J and HLA-DR4 transgenic mice, with similar results using pancreatic and lung tumor cells. Importantly, non-citrullinated enolase does not induce antitumor response [149]. A potent CD4 T-cell-mediated antitumor response is stimulated in HLA transgenic mice after injection with combined citrullinated vimentin and citrullinated enolase [186]. This is an exciting area of research and work is under way to identify other citrullinated proteins that may provide therapeutic benefit for a variety of cancers [187, 188].

Conclusions and future studies

PADs are expressed in female reproductive tissues, yet the function of protein citrullination and the identity of the citrullinome in many tissues is still an open question. Of particular importance to reproductive biology is the role of E2 in stimulating PAD expression, which is followed by citrullination of numerous target proteins. With increased PADs, there is a subsequent increase in histone citrullination, which gives rise to epigenetic changes in gene expression in gonadotropes, lactotropes, mammary epithelial, and uterine cells [40, 48, 53, 61]. Clearly, further studies are required to identify other targets for PAD-catalyzed citrullination in these cell types. In non-reproductive models, the citrullination of cytoskeletal filaments regulate secretory mechanisms [189]. The uterus, mammary gland, and anterior pituitary gland are highly secretory and synthesize and secrete a wide range of molecules critical at all levels of reproduction. Given this, it is intriguing to speculate that citrullinated cytoskeletal proteins may contribute to secretory mechanisms and clearly warrant further investigation given the importance of this process to reproductive function. Additionally, there is a compelling need for the development and analysis of PAD KO mice. While independent global PAD2 and PAD4 KO lines exist, neither display any major reproductive phenotype, most likely due to compensation by other isoforms [190, 191]. Future studies would greatly benefit from the creation of tissue-specific single and PAD2/4 double KO mice lines to evaluate reproductive phenotypes.

In contrast to reproductive physiology, aberrant PAD expression and protein citrullination are associated with poorer outcomes in both autoimmune disorders and female reproductive cancers. Defining the mechanisms by which PADs and citrullinated proteins stimulate autoimmune initiation and progression and tumorigenesis remains to be elucidated. However, recent advances support the need for the continued development of selective PAD inhibitors for the treatment of autoimmune disorders and cancers, while simultaneously studying off target side effects and toxicities of PAD and citrullination directed therapies. In conclusion, determining the role of PADs and protein citrullination is critical not only to improve our understanding of reproductive biology, but also how their dysfunction leads to disease.

Conflict of Interest

Our manuscript contains original research, but it is not published nor is it under editorial consideration elsewhere. All research was carried out following ethical guidelines including strict guidelines governing the use of human and animal tissue. None of the authors has anything to disclose.

Author Contributions

The manuscript was written by AOC, HMR, and BDC. GL, CHY, BS, SAK, SBD, and AMN all helped edit and revise the manuscript based on their contributions to the field. SE, GJB, BDC, and HMR conducted the experiments in Figures 4–6. Financial support for the project was to BDC.