Background
PADIs are a family of posttranslational modification enzymes that convert positively charged arginine residues on substrate proteins to neutrally charged citrulline, and this activity is alternatively called citrullination or deimination. The PADI enzyme family is thought to have arisen by gene duplication and localizes within the genome to a highly organized cluster at 1p36.13 in humans. At the protein level, each of the five well-conserved PADI members shows a relatively distinct pattern of substrate specificity and tissue distribution [
1,
2]. Increasingly, the dysregulation of PADI activity is associated with a range of diseases, including rheumatoid arthritis (RA), multiple sclerosis, ulcerative colitis, neural degeneration, COPD, and cancer [
3‐
5]. While the presumptive function of PADI activity in most diseases is linked to inflammation, the role that PADIs play in cancer progression is not clear. We and others, however, have found that PADI4 appears to play a role in gene regulation in cancer cells via histone tail citrullination. For example, in MCF7 breast cancer cells estrogen stimulation enhances PADI4 binding and histone H4 citrullination at the canonical ER target gene,
TFF1, leading to transcriptional repression [
6]. On the other hand, stimulation of MCF7 cells with EGF facilitates activation of
c-fos via PADI4-mediated citrullination of the ELK1 oncogene [
7]. Additionally, others have shown that citrullination of the p53 tumor suppressor protein affects the expression of p53 target genes
p21,
OKL38,
CIP1 and
WAF1[
8‐
10]. Interestingly, treatment of several PADI4-expressing cancer cell lines with the PADI inhibitor, Cl-amidine, elicited strong cytotoxic effects while having no observable effect on non-cancerous lines [
11], suggesting that PADIs may represent targets for new cancer therapies.
Our current study suggests that PADI2 may also play a role in cancer progression, and this prediction is supported by several previous studies. For example, a mouse transcriptomics study investigating gene expression in MMTV-neu tumors found that
PADI2 expression was upregulated ~2-fold in hyperplastic, and ~4-fold in primary neu-tumors, when compared to matched normal mammary epithelium [
12]. In humans,
PADI2 is one of the most upregulated genes in luminal breast cancer cell lines compared to basal lines [
13,
14]. Additionally, gene expression profiling of 213 primary breast tumors with known HER2/ERBB2 status identified
PADI2 as one of 29 overexpressed genes in HER2/ERBB2+ tumors; thus, helping to define a HER2/ERBB2+ gene expression signature [
15]. Given these previous studies, our goal was to formally test the hypothesis that PADI2 plays a role in mammary tumor progression. For the study, we first documented PADI2 expression and activity during mammary tumor progression, and then investigated the effects of PADI inhibition in cell cultures, tumor spheroids, and preclinical
in vivo models of breast cancer.
Discussion
In this study, we show that PADI2 is specifically upregulated during mammary tumor progression and that the PADI inhibitor, Cl-amidine, is effective in inhibiting the growth of PADI2 overexpressing cell lines in both 2D and 3D cultures. In addition, we demonstrate here for the first time that Cl-amidine is successful in suppressing tumor growth in a xenograft mouse model of comedo-DCIS. Lastly, we document that PADI2 expression is highly correlated with HER2/ERBB2 overexpressing and luminal subtype breast cancers.
Given the previous correlations between PADI2 and the HER2/ERBB2 oncogene, the goal of this study was to carry out an initial test of the hypothesis that PADI2 plays a role in breast cancer progression. To accomplish this, we utilized the well-established MCF10AT model [
16,
17] and found that PADI2 expression was highly upregulated in MCF10DCIS cells, a cell line that forms comedo-DCIS lesions that spontaneously progress to invasive tumors [
30,
46]. Our finding that PADI2 expression is highest in comedo-DCIS lesions (defined by their necrotic centers) was perhaps not too surprising, given the close association of PADIs with inflammatory events. We are currently investigating the potential links between inflammatory signaling in these MCF10DCIS lesions and PADI2 activity.
Interestingly, PADI2 expression in the MCF10AT series coincided with HER2/ERBB2 upregulation which, again, was not entirely unexpected given previous reports correlating
PADI2 expression with
HER2/ERBB2[
15]. While we did find that HER2/ERBB2 and PADI2 protein expression correlated well across the MCF10AT cell lines, PADI2 protein levels are particularly high in the MCF10DCIS line, relative to HER2/ERBB2. We cannot currently explain this finding; however, it is possible that cell-line-specific factors are stabilizing the PADI2 transcript, thus allowing for increased protein expression [
51,
52].
While our data show a potential relationship between PADI2 and HER2/ERBB2 in the MCF10AT model, we wanted to examine this correlation at higher resolution. To accomplish this we queried our RNA-seq dataset of 57 breast cancer cell lines with known subtype and HER2/ERBB2 status and found that: (a)
PADI2 expression is highest in luminal cell lines and that (b)
PADI2 expression is highly correlated with HER2/ERBB2 overexpression across the basal-NM, claudin-low, and luminal lines. The observation that
PADI2 is upregulated in the luminal subtype confirms previous gene expression data where
PADI2 was identified as one of the top upregulated genes in luminal breast cancer lines compared to basal lines [
13,
14]. In order to test whether the observed correlation between
PADI2 and
HER2/ERBB2 would be retained at the protein level, we also tested a small sample of cell lines representing the four common breast cancer subtypes and found that PADI2 expression was only observed in the HER2/ERBB2+ BT-474 and SK-BR-3 lines. However, we did observe some discordance seen between
PADI2 transcript and protein levels, but we predict this difference may be due to post-transcriptional regulatory mechanisms. This prediction is based, in part, upon the observation that PADI2 possesses a long 3’UTR [
53] that contains several AU-rich elements [
54,
55] that have been shown to bind the stabilizing regulatory factor HuR [
56]. HuR binding has been shown to enhance the stability of mRNAs involved in proliferation [
57‐
59], while also playing a role in breast cancer, as cytoplasmic accumulation of HuR promotes tamoxifen resistance in BT-474 cells [
60] and the stability of
HER2/ERBB2 transcripts in SK-BR-3 cells [
52]. Interestingly, from these studies, the level of HuR was reported to be high in both BT-474 and SK-BR-3 cells, while it was relatively low in MCF7 cells. It is important to note that while we observed low levels of PADI2 protein expression in MCF7 (Additional file
1: Figure S1a), recent work from our lab has confirmed the expression of PADI2 in MCF7 cells [
49,
50].
We also examined two mouse models of mammary tumorigenesis, the luminal MMTV-neu and the basal MMTV-Wnt-1, and found that, as predicted, PADI2 levels are highest in the HER2/ERBB2 overexpressing MMTV-neu mice compared to normal mammary tissue and to hyperplastic and primary MMTV-Wnt-1 tumors. Taken together, these findings indicate that PADI2 is predominantly expressed in luminal epithelial cells, and that there appears to be a strong relationship between PADI2 and HER2/ERBB2 expression in breast cancer. Subsequent studies are now underway to test whether PADI2 plays a functional role in HER2/ERBB2 driven breast cancers, potentially by functioning as an inflammatory mediator.
Previous studies have shown that the inhibition of PADI enzymatic activity by Cl-amidine is effective in decreasing the growth of several cancer cell lines (i.e. HL-60, HT-29, U2OS, and MCF7 cells), and that administering the drug in combination with doxorubicin or the HDAC inhibitor SAHA can have synergistic cytotoxic effects on cells [
8,
9,
11,
45]. Cl-amidine is highly specific for all PADI enzymes, with dose-dependent cytotoxicity and little to no effect in non-cancerous cell lines (i.e. HL-60 granulocytes and NIH3T3 cells) [
11]. Our studies expand on these previous results by showing that Cl-amidine suppresses the growth of the transformed lines of the MCF10AT model, especially the MCF10DCIS cell line, in both 2D and 3D cultures. In addition, we show for the first time that Cl-amidine is successful in treating tumors
in vivo using a mouse model of comedo-DCIS from xenografted MCF10DCIS cells. Given that the loss of basement membrane integrity is an important event during the progression of DCIS to invasive disease, it is significant that Cl-amidine treated xenografts maintain their basement membrane integrity and show reduced leukocytic infiltration across the basement membrane compared to the control group. These observations suggest that Cl-amidine treatment might enhance the ability of tumor ductular myoepithelial cells to deposit continuous and organized basement membranes.
While we chose the subcutaneous model of MCF10DCIS tumorigenesis, future studies on the effect of Cl-amidine could examine alternate methods of transplantation, such as the previously described intraductal method [
48]. In addition, different models of DCIS could be examined, such as xenografted SUM-225 cells, which show high
HER2/ERBB2 and
PADI2 levels (see Figure
3 for relative levels). Of note, we found that while Cl-amidine suppressed tumor growth, the drug was well tolerated by mice in this study. Similarly, our previous work found that doses of Cl-amidine up to 75 mg/kg/day in a mouse model of Colitis [
3], and up to 100 mg/kg/day in a mouse model of RA [
5], were well-tolerated without side effects. Further work into studying the pharmacokinetics and biodistribution of Cl-amidine, or perhaps the development of an isozyme specific inhibitor of PADI2, will be an important step in helping to find a potent drug for the treatment of DCIS patients.
The actual mechanisms by which Cl-amidine reduces cellular proliferation have yet to be fully elucidated, though evidence here suggests that PADI2 may play a role (direct or indirect) in regulating the expression of both cell cycle and tumor promoting genes. Previous reports have shown that Cl-amidine effectively upregulates a number of p53-regulated genes, including
p21,
PUMA, and
GADD45[
8,
45]. Our qRT-PCR cell cycle array results confirm that two of these genes,
p21 and
GADD45α, are upregulated after treatment of MCF10DCIS cells with Cl-amidine by 17.68- and 13.53-fold, respectively. Furthermore, we have identified additional genes downregulated by Cl-amidine, including
MKI67,
MCM5, and
MCM2, each with known functions in cancer progression
. We have also quantitatively analyzed for apoptosis levels (Caspase-3) after Cl-amidine treatment via flow-cytometry, and see a dose-dependent decrease in proliferation and increase in apoptosis. Moreover, we also show that the cells arrest in S-phase after Cl-amidine treatment, thus leading to S-phase coupled apoptosis, which is a known response to DNA damage [
44]. Taken together, the observed inhibitory effects of Cl-amidine on tumor growth may be due to the suppression of genes involved in oncogenesis and the activation of genes involved in apoptosis, though additional work is needed to define the mechanisms behind these potential relationships.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
JLM, SM, OLG, HCB, BDC, AMP, and LJA all acquired primary data and helped in the analysis of the research found in this manuscript. JLM was involved in the design of the study and drafted the manuscript, in addition to performing molecular genetic studies on both in vitro and in vivo models. SM generated the 3D-spheroids and performed pathological analyses. OLG performed RNA-seq and ALEXA-seq, and statistical analysis on the data from the collection of breast cancer cell lines. HCB participated in generating MCF10DCIS xenografts and in the in vivo drug study. BDC performed IF experiments and helped with data analysis. AMP assayed PADI activity/citrulline levels. LJA performed flow-cytometry experiments and FACS analysis. VS and CPC designed and synthesized the Cl-amidine used for the experiments. LRH provided MMTV mouse models and helped with data analysis. SM, OLG, BDC, and PRT helped revise the manuscript. SAG helped in the design of the study and contributed to the manuscript revision. PRT, JWG, and SAG supervised the study. All authors read and approved the final manuscript.