**Persistent Organic Pollutants and Breast Cancer: A Systematic Review and Critical Appraisal of the Literature**

### **Kaoutar Ennour-Idrissi 1,2,3,4, Pierre Ayotte 3,5,6 and Caroline Diorio 1,2,3,4,\***


Received: 4 July 2019; Accepted: 24 July 2019; Published: 27 July 2019

**Abstract:** Persistent organic pollutants (POPs) bioaccumulate in the food chain and have been detected in human blood and adipose tissue. Experimental studies demonstrated that POPs can cause and promote growth of breast cancer. However, inconsistent results from epidemiological studies do not support a causal relationship between POPs and breast cancer in women. To identify individual POPs that are repeatedly found to be associated with both breast cancer incidence and progression, and to demystify the observed inconsistencies between epidemiological studies, we conducted a systematic review of 95 studies retrieved from three main electronic databases. While no clear pattern of associations between blood POPs and breast cancer incidence could be drawn, POPs measured in breast adipose tissue were more clearly associated with higher breast cancer incidence. POPs were more consistently associated with worse breast cancer prognosis whether measured in blood or breast adipose tissue. In contrast, POPs measured in adipose tissue other than breast were inversely associated with both breast cancer incidence and prognosis. Differences in biological tissues used for POPs measurement and methodological biases explain the discrepancies between studies results. Some individual compounds associated with both breast cancer incidence and progression, deserve further investigation.

**Keywords:** breast cancer; persistent organic pollutants; breast cancer risk; breast cancer prognostic; systematic review

#### **1. Introduction**

Persistent organic pollutants (POPs) are a group of chemical substances of synthetic origin used for industrial, agricultural or domestic purposes, that persist in the environment and bioaccumulate in the food chain due to their lipophilic properties [1,2]. POPs have been detected in human blood, adipose tissue and human milk and have been linked to the increase in the incidence of hormone-dependent breast cancers [1,3–8].

Given the abundance of adipose tissue in the human breast, mammary epithelial cells exposure to POPs sequestered in breast adipose tissue may promote carcinogenesis and progression of mammary cancers [9]. In fact, numerous in vitro studies have demonstrated that some POPs stimulate the

growth of estrogen receptor (ER)-positive breast cancer cells [10–12]. In animal studies, exposure to some POPs, particularly during the perinatal period, impairs breast tissue development and increases its susceptibility to carcinogens and the incidence of precancerous and cancerous breast lesions [13]. In addition to their endocrine disrupting effect either as agonists or as antagonists of endogenous hormones [14], POPs can interfere with estrogen synthesis by disrupting adipose tissue functioning [15,16], interact with transcription factors [17], induce genotoxic enzymes [17] and cytochrome 450 leading to increased levels of reactive oxygen species [18], and induce trans-generational phenotypic changes by altering the epigenome [19].

Although experimental studies demonstrate that POPs can cause and promote growth of breast cancer, several observational studies conducted in humans yielded inconsistent results regarding the implication of POPs in women breast cancers [20–26]. Observational studies are known to be prone to different biases that vary according to studies designs [27]. To draw meaningful conclusions about a causal relationship between POPs and breast cancer in women, a systematic comparison of the strengths and weaknesses of studies should be performed to triangulate their findings to provide assurance that the observed findings are actually real [27]. Thus, the objective of the present systematic review of the literature was to evaluate the observed associations between POPs and breast cancer risk and prognosis to identify individual POPs that are repeatedly found to be associated with both breast cancer incidence and progression, and to provide an explanation to the observed inconsistencies between studies.

#### **2. Materials and Methods**

A systematic review was conducted following a pre-established protocol and according to the general methodology of Cochrane reviews [28]. Considering the expected methodological diversity and heterogeneity between eligible studies, the great susceptibility of observational designs to selection bias and the variability in methods used to control for confounding, no quantitative synthesis was planned [28].

#### *2.1. Search Methods for Identification of Studies*

An electronic search of the following databases was performed, from inception to December 2018: MEDLINE (via PubMed), EMBASE and CENTRAL (Cochrane Central Register of Controlled Trials). Search strategies were developed for each of these databases with text words and index terms referring to POPs, breast cancer risk and breast cancer prognosis, and excluding animal studies (Table S1). No language or publication date restrictions were applied. Reference lists of relevant reviews and of included studies were scanned for any additional relevant studies not otherwise identified.

#### *2.2. Criteria for Considering Studies for This Review*

#### 2.2.1. Types of Studies

Any observational or intervention study that evaluated the association between POPs and breast cancer risk, survival or a meaningful breast cancer prognostic factor, whatever the design was eligible for inclusion. No restrictions were applied regarding language or type (articles, short reports and abstracts) of publication.

#### 2.2.2. Types of Participants

Women included in the studies before or after breast cancer diagnosis, regardless of age, menopausal status, breast cancer type, disease stage and treatment regimen, were eligible. No participants were excluded based on ethnicity.

#### 2.2.3. Types of Exposures

Studies that measured exposure to any lipophilic POP, in a lipid rich biological human sample (peripheral blood and adipose tissue), whatever the method of measurement, were eligible.

#### 2.2.4. Types of Outcomes

Breast cancer risk, measured by breast cancer incidence, prevalence or breast mammographic density (a recognized breast cancer risk factor) and breast cancer survival, including overall survival (all-cause mortality), breast cancer-specific survival (breast cancer-specific mortality), and breast cancer-free survival (breast cancer recurrence), were the primary outcomes. Studies that assessed the association of POPs with meaningful breast cancer prognostic factors (age, stage, tumor size, lymph node involvement, histological type, grade and molecular subtype) were also eligible.

#### *2.3. Data Collection and Analysis*

#### 2.3.1. Selection of Studies

The references identified by the search strategy were reviewed by one author (K.E-I.) in a two-step process. First, the title and abstract of each study were screened to exclude obviously non-eligible studies and second, the full text of retained articles was examined and subjected to evaluation using the predefined eligibility criteria. Whenever required, a second review author (C.D.) was consulted. When required, further information was sought from the authors by email.

#### 2.3.2. Data Extraction

Data extraction was performed using an exhaustive standardized form designed for this review. Information about study design (inclusion criteria, sample size and methodology), participants and tumors characteristics at diagnosis (age, menopausal status, tumor invasiveness, tumor ER status), exposure assessment (timing, tissue sample, method of measurement, lipid-adjustment, list of all contaminants evaluated, treatment of non-detectable values), measured outcomes and reported results (any reported measure of association, adjustment variables, and statistical model selection procedure) were collected. For observational studies, special attention was paid to distinguishing between adjusted and unadjusted results, and to the variable selection method used in multivariate analyses. Studies definition of each characteristic or variable retained was recorded. In the case of multiple publications related to the same study, the publication reporting the outcomes of interest to the present review or the one with the longest follow-up of these outcomes was considered as the reference, and information was supplemented by secondary publications as required. Abstracts with insufficient information and data to permit inclusion were excluded from the qualitative synthesis (Table S2). Data were extracted twice over the course of several days to ensure their consistency.

#### 2.3.3. Assessment of Risk of Bias in Retained Studies

Based on the "STrengthening the Reporting of OBservational studies in Epidemiology." (STROBE) statements [28], and the rating approach of the "Risk Of Bias in Non-randomized Studies-of Interventions" (ROBINS-I) tool [27], the following domains were evaluated for risk of bias of included studies: selection of participants into the study, exposure measurement, outcome measurement, potential confounding accounted for, missing data, and selective reporting.

Assessment of the risk of bias was performed twice by a review author (K.E-I.), both for the risk of bias in each study and for the overall risk of bias across studies. When required, a second reviewer (C.D.) was consulted.

#### 2.3.4. Assessment of Heterogeneity

Differences between studies, including study design, participant characteristics (age and menopausal status), tumor characteristics (invasiveness, ER status, and treatment received), exposure measurement (timing, type of tissue sample) and different levels of risk of bias were considered for exploring possible sources of heterogeneity.

#### 2.3.5. Data Synthesis

Given that high heterogeneity between studies was expected, quantitative synthesis of data was not considered appropriate. A systematic qualitative synthesis of study characteristics and results was performed for risk, mortality, and prognostic factors associations with POPs exposure, and separately for each type of tissue sample. The results were considered adjusted only when all important confounders were considered into the models. For breast cancer risk, authors should have considered at minimum age, body mass index or any other estimation of body fat, and breastfeeding or parity as potential confounders. For breast cancer mortality, authors should have adjusted at minimum for age. In addition, studies of breast adipose POPs should have considered breastfeeding or parity as potential confounders. A positive association was defined as an observed higher risk or mortality with higher POPs exposure whereas a negative association was defined as an observed inverse association.

#### **3. Results**

#### *3.1. Results of the Search*

Of the 11,015 references retrieved by electronic search, 95 met eligibility criteria (Figure 1), of which 85 reported breast cancer incidence or prevalence outcomes [29–113], six reported mortality outcomes [41,45,114–117] and nine reported breast cancer prognostic factors [66,90,116,118–123]. The majority of studies of breast cancer risk were case-control studies (*n* = 81) whereas studies of breast cancer prognosis included five cohort studies for mortality and nine cross-sectional studies for breast cancer prognostic factors. Overall, POPs were measured in peripheral blood in 63 studies, in breast adipose tissue in 32 studies, in adipose tissue other than breast in five studies and in breast tumors in four studies (Figure 1).

#### *3.2. Description of Studies*

The 95 included studies were published between 1976 and 2018, and involved between five and 902 breast cancer patients (median = 113).

#### 3.2.1. Studies of Breast Cancer Risk

Characteristics of the 61 studies that examined associations between peripheral blood POPs and breast cancer risk are summarized in Table 1. These studies included breast cancer patients between 40 and 66 years of mean age with varying proportions of premenopausal and postmenopausal patients. Ten studies included at least 80% of postmenopausal patients, of which three studies included exclusively postmenopausal patients. The proportion of invasive breast cancers was not reported in 35 studies and varied in the remaining 26 studies between 62 and 100%. Twenty studies included at least 80% of invasive breast cancers of which 13 studies included exclusively invasive breast cancers. The proportion of estrogen receptor (ER) positive breast cancers was not reported in 40 studies and varied in the remaining 19 studies between 32% and 87%. Two studies included at least 80% of ER-positive breast cancers (Table 1 and Table S3).

**Figure 1.** Flow Diagram according to PRISMA (Preferred Reporting Items of Systematic Reviews and Meta-Analyses) [PRISMA], with modifications. \*One cohort study on breast cancer mortality also reported cross-sectional analyses of prognostic factors.

Characteristics of the 26 studies that examined associations between breast adipose tissue POPs and breast cancer risk are summarized in Table 2. All these studies were hospital-based case-control studies with hospital-controls and included breast cancer patients between 40 and 63 years of mean age with varying proportions of premenopausal and postmenopausal patients. Only one study included at least 80% of postmenopausal patients. The proportion of invasive breast cancers was not reported in 11 studies and varied in the remaining 15 studies between 76% and 100%, with eight studies including exclusively invasive breast cancers. The proportion of ER-positive breast cancers was not reported in 15 studies and varied in the remaining 11 studies between 45% and 90%. No study included at least 80% of ER-positive breast cancers (Table 2 and Supplementary Table S4).


**Table 1.** Summary characteristics of studies of peripheral blood POPs and breast cancer risk (*n* = 61).

*n*: number of studies; NR: not reported; POPs: persistent organic pollutants; MS: mass spectrometry; GC-ECD: gas chromatography with electron Capture Detector; HLPC-MS-MS: high-performance liquid chromatography-tandem mass spectrometry; HPLC/FD: high-performance liquid chromatography with fluorescence detection; LC-MS-MS: liquid chromatography–tandem mass spectrometry; GC-MS-MS: gas chromatography-tandem mass spectrometry; GC-IDMS: gas-chromatography isotope-dilution mass-spectrometry; GC-ID-HRMS: gas chromatography-isotope dilution high-resolution mass spectrometry; HR-GC-ECD: high-resolution gas chromatography with micro-electron capture detection; GC: gas chromatography; PCBs:. polychlorinated biphenyls; PFAS: perfluoroalkyl substances; BPA: bisphenol A; PBBs: polybrominated biphenyls.



*n*: number of studies; NR: not reported; POPs: persistent organic pollutants; GC-ECD: gas chromatography with electron Capture Detector; GC-MS: gas-chromatography mass-spectrometry; GC: gas chromatography; PCBs: polychlorinated biphenyls; PFAS: perfluoroalkyl substances; PBDE: polybrominated diphenyl ethers.

Four studies examined associations between POPs in breast tumors and breast cancer risk, of which three included breast tissue surrounding malignant tumors for cases. Two studies compared malignant tumors of cases to benign tumors of controls whereas the two other studies compared malignant tumors of cases to normal tissue of controls. All these four studies were hospital-bases case-control studies with hospital controls, in which cases were on average 50 to 60 years old in the two studies that reported age at diagnosis. One study included exclusively invasive cancers, whereas the three other studies did not report proportion of invasive cancers. No study reported proportion of menopausal patients or the proportion of ER-positive breast cancers (Table S4).

Two studies examined associations between POPs in buttock adipose tissue and breast cancer risk. The recent one was a cohort-nested case-control study with incidence density (risk-set) sampling and included 409 postmenopausal breast cancer patients aged 58 years old in average, of which 78% had ER-positive breast cancers. The proportion of invasive tumors was not reported. The other study was a hospital-based case-control study with community controls including 265 breast cancer patients aged 62 years old in average. The proportions of postmenopausal patients, invasive tumors and ER-positive tumors were not reported (Table S5).

#### 3.2.2. Studies of Breast Cancer Prognosis

Characteristics of the 14 studies that examined the association between POPs and breast cancer prognosis are summarized in Table 3.


**Table 3.** Summary characteristics of studies of POPs and breast cancer prognosis (*n* = 14).

*n*: number of studies; NR: not reported; POPs: persistent organic pollutants; GC-MS: gas-chromatography mass-spectrometry; GC-ECD: gas chromatography with electron Capture Detector; HLPC-MS-MS: high-performance liquid chromatography-tandem mass spectrometric; PCBs: polychlorinated biphenyls.

Six studies examined mortality outcomes, with three measuring POPs in peripheral blood, one measuring POPs in breast adipose tissue and two measuring POPs in adipose tissue other than breast. Patients were aged between 58 and 66 years old in average with only two studies reporting the proportion of postmenopausal women (66% and 100% respectively), and two studies reporting proportion of invasive cancers (71% and 86% respectively). The proportion of ER-positive breast cancers varied between 72% and 78% in the three studies that have reported patients ER status (Tables S5 and S6).

Ten studies examined breast cancer prognostic factors, with three measuring POPs in peripheral blood, six measuring POPs in breast adipose tissue and one measuring POPs in adipose tissue other than breast. Patients were aged between 52 and 65 years old in average with varying proportions of premenopausal and postmenopausal women and none with at least 80% of postmenopausal patients. The proportion of invasive cancers varied between 85% and 100% and the proportion of ER-positive breast cancers varied between 50% and 86% with only two studies including at least 80% of ER-positive breast cancers (Table S8).

#### *3.3. Risk of Bias in Retained Studies*

Overall, studies reporting associations between peripheral blood POPs and breast cancer risk ranged from moderate to critical risk of bias, whereas studies reporting associations between adipose tissue POPs and breast cancer risk were more likely to be at serious or critical risk of bias.

Overall, studies reporting associations between POPs, measured in peripheral blood or in adipose tissue, and mortality outcomes were at serious risk of bias, whereas studies reporting associations with prognostic factors ranged from serious to critical risk of bias.

#### *3.4. Systematic Data Synthesis*

#### 3.4.1. Studies of Breast Cancer Risk

Among the 61 studies that examined associations between peripheral blood POPs and breast cancer risk, 30 reported a positive association with at least one POP, six reported a negative association with at least one POP, and 20 reported no association. Cohort-nested case-control studies with cumulative density (exclusive) sampling, population-based case-control studies not nested in a defined cohort and hospital-based case-control studies with both community- and hospital-controls were more likely to observe an association. Studies that included at least 80% of postmenopausal patients were more likely to observe an association whereas studies that included less than 80% of postmenopausal patients were more likely to observe no association. Studies that included at least 80% of invasive cancers and those with non-reported proportions of invasive breast cancers were more likely to observe an association. Studies with non-reported proportions of ER-positive breast cancers were more likely to observe an association whereas studies including less than 80% ER-positive breast cancers were slightly more likely to observe no association (Table S3).

Among the 26 studies that examined associations between breast adipose tissue POPs and breast cancer risk, 10 reported a positive association with at least one POP, two reported a negative association with at least one POP, and seven reported no association. Studies reporting the proportion of invasive breast cancers were slightly more likely to observe an association. Studies with non-reported proportions of ER-positive breast cancers were more likely to observe a positive association. The four studies of POPs in breast tumors did not report minimally adjusted estimates of risk (Table S4).

The two studies that examined associations between POPs in buttock adipose tissue and breast cancer risk reported negative associations (Table S5).

#### 3.4.2. Studies of Breast Cancer Prognosis

All three studies that examined peripheral blood POPs reported positive associations with both breast cancer all-cause and specific mortality, of which one study also reported a negative association with all-cause mortality (Table S6). The only study of breast adipose tissue POPs and breast cancer mortality reported a positive association with breast cancer recurrence. The two studies of POPs in adipose tissue other than breast reported negative associations with all-cause and breast cancer specific mortality respectively, of which one study reported a positive association with breast cancer specific mortality (Table S7).

Among the three studies that examined peripheral blood POPs and breast cancer prognostic factors, one study reported a positive association with tumor size and lymph-node involvement. Among the six studies of breast adipose tissue POPs, three studies examined associations with meaningful breast cancer prognostic factors but none of them reported minimally adjusted estimates. The only study of buttock adipose tissue POPs and breast cancer prognostic factors did not report minimally adjusted estimates (Table S8).

#### 3.4.3. Individual POPs and Breast Cancer Risk and Prognosis

One to 71 individual POPs were measured in studies of breast cancer risk (median = 9). Organochlorines were measured in 69 studies, of which 43 in blood, 21 in breast adipose tissue, two in adipose tissue other than breast and three in breast tumor. Polychlorinated biphenyls (PCBs) were measured in 57 studies, of which 38 were in blood, 16 in breast adipose tissue, one in adipose tissue other than breast and two in breast tumors. Dioxins were measured in four studies, of which two in blood and two in breast adipose tissue. Perfluoroalkyl substances (PFAS) were measured in three studies in blood. Bisphenol A (BPA) was measured in two studies in blood, polybrominated flame retardants (PBBs and PBDEs) in one study in blood and one study in breast adipose tissue whereas mono-ethyl phthalate (MEP) and parabens were measured in blood in one study respectively (Table S9).

One to 35 individual POPs were measured in studies of breast cancer prognosis (median = 25). Organochlorines were measured in five studies of breast cancer mortality, of which three were in blood, one in breast adipose tissue and two in adipose tissue other than breast, whereas six studies measured organochlorines in relation to prognostic factors, of which three were in blood, four in breast adipose tissue and one in adipose tissue other than breast. PCBs were measured in four studies of breast cancer mortality, of which three were in blood, one in breast adipose tissue and one in adipose tissue other than breast, whereas four studies measured PCBs in relation to prognostic factors, of which two in blood, four in breast adipose tissue and one in adipose tissue other than breast (Tables S9 and S10). Parabens were measured in breast adipose tissue in one study in relation to prognostic factors.

The magnitude of the reported associations between POPs and breast cancer risk and mortality are summarized in Table 4.


**Table 4.** Main results summary of studies reporting positive \* associations between POPs and breast cancer risk and mortality.

POPs: persistent organic pollutants; CI: Confidence interval; NA: Not applicable; NR: not reported; \* positive association: an observed higher risk or mortality with higher POPs exposure; \*\* Adjusted for all important confounders; OR: odds ratio; MRR: mortality rate ratio; PCB: Polychlorinated biphenyls; HR: hazard ratio, β-HCH: β-Hexachlorocyclohexane.

When considering POPs positively associated with breast cancer risk in at least 10% of studies and at least two studies and no reported negative associations, eight individual POPs were consistently positively associated with breast cancer risk in blood: *p,p'*-Dichlorodiphenyldichloroethylene (*p,p'*-DDE), total or not specified DDE, β- Hexachlorocyclohexane (β-HCH), Dieldrin, PCB 118, PCB 138, PCB 170, PCB 180. Three individual POPs were consistently positively associated with breast cancer risk in breast adipose tissue: *p,p'*-DDE, total or not specified DDE, PCB 105. When considering POPs positively associated with breast cancer risk in at least one study and no reported negative associations, six individual POPs were positively associated with breast cancer risk in both blood and breast adipose tissue: *p,p'*-DDE, total or not specified DDE, Hexachlorobenzene (HCB), β-HCH, PCB 118 and PCB 180 (Tables S11 and S12).

When considering POPs positively associated with breast cancer mortality in at least 10% of studies and no reported negative associations, total PCBs and four individual POPs were positively associated with breast cancer mortality in blood: *p,p'*-Dichlorodiphenyltrichloroethane (*p,p'*-DDT), Dieldrin, PCB 174, PCB 177. Total PCBs and three individual POPs were positively associated with breast cancer mortality in breast adipose tissue: PCB 118, PCB 153, PCB 167 (Tables S13 and S14). Six individual POPs were positively associated with breast cancer prognostic factors in blood, in at least one study and with no reported negative associations: *p,p'*-DDE, Oxychlordane, *trans*-Nonachlor, β-HCH, PCB 138, PCB 153 (Table S16).

Three individual POPs were positively associated with both breast cancer risk and prognosis either in blood or in breast adipose tissue: *p,p'*-DDE, β-HCH and PCB 118 (Tables S11–S15).

#### **4. Discussion**

The present systematic review of POPs and breast cancer indicates that studies of blood POPs and breast cancer risk accounted for much of the observed inconsistencies of epidemiological studies results. POPs measured in breast adipose tissue were more clearly associated with higher breast cancer risk. POPs were more consistently associated with worse breast cancer prognosis, whether measured in blood or breast adipose tissue, whereas POPs measured in adipose tissue other than breast were inversely associated with both breast cancer risk and prognosis. Some individual POPs measured in blood and breast adipose tissue were consistently associated with higher breast cancer risk and worse prognosis. However, the overall strength of evidence is weak, since few studies contributed to estimations of associations and the overall risk of bias in these studies ranged from moderate to critical.

The inconsistencies between studies of blood POPs and breast cancer risk could be explained by methodological biases. In fact, more than half of these studies have measured POPs at the time of diagnosis which does not necessarily reflect the cumulative lifetime exposure to POPs and early-life exposures during critical windows of vulnerability [124]. Even though the majority of population-nested case-control studies and the only cohort study have measured POPs several months to many years before breast cancer occurrence, a point measurement of blood concentration of POPs is more likely to reflect recent dietary intakes and liver function [125,126] and can be affected by various events over time, such as weight loss or gain, pregnancies and breastfeeding [124–126]. The complex misclassification of POPs exposure resulting from blood measurements could have biased the observed associations toward the null, i.e., toward the observation of weaker associations or no associations at all.

In this regard, adipose tissue, as a storage compartment for lipophilic POPs [127], is a more appropriate medium for estimating lifetime exposure to POPs. The observation of consistently positive associations with breast cancer risk in studies of breast adipose tissue POPs but consistently negative associations with POPs measured in adipose tissue other than breast is in line with the existent evidence of a protective function of adipose tissue in the wildlife and points toward the metabolic and toxicokinetic differences between different types of adipose tissue [127]. By accumulating POPs, adipose tissue away from breast decreases their availability to other tissues, thereby limiting their toxicity to the breast, whereas accumulation of POPs in breast adipose tissue exposes breast epithelial cells to their chronic local release. In fact, ultrastructural methods revealed regional differences in morphology of human subcutaneous tissues [128]. Abdominal adipose tissue, classified as deposit white adipose tissue, having large adipose cells and a poor collagenic component whereas adipocytes of breast adipose tissue, classified as structural white adipose tissue, are covered by a relatively dense connective capsule [128]. These regional differences in morphology explain the known regional differences in the metabolism of subcutaneous fatty depots that are related to their various functions. Thus, our results suggest that differences between adipose tissue subtypes may also have a toxicocokinetic impact on POPs.

Moreover, more than half of studies of blood POPs and breast cancer risk included more premenopausal than postmenopausal breast cancer patients. Although environmental exposures may be involved in premenopausal breast cancer occurrence, these cancers are primarily driven by a strong genetic susceptibility and are more often ER-negative cancers [129]. Furthermore, the increase in breast cancer incidence over the last decades reflects the increase in the incidence of postmenopausal breast cancers, which are more often ER-positive breast cancers [130], and thus more susceptible to the hormone-disrupting effects of POPs [8]. The selection bias created by inclusion of large proportions of premenopausal breast cancers, which are less likely to be related to POPs exposure, could have biased the observed associations toward the null. In fact, we observed that studies that included less than 80% of postmenopausal patients and those including fewer than 80% ER-positive breast cancers were more likely to observe no association.

Another important issue was related to statistical methods used for selecting potential confounders. If the majority of studies of blood POPs and breast cancer risk have considered important confounders in their statistical models, methods used for selecting potential confounders were not always appropriate. In particular, the majority of case-control studies used the change in estimate method, which is not appropriate for accurate estimations of associations. Changes in estimates may be observed when adjusting for colliders (i.e., non-confounders that introduce a selection bias) and when non-collapsible effect measures such as odds ratios are used [131]. The bias introduced by this method can be difficult to predict when numerous variables are tested for confounding and can lead to discrepant studies results.

The strengths of the present systematic review include the use of the Cochrane Reviews rigorous methodology, the extensive and highly sensitive search strategy to retrieve as many relevant studies as possible, the use of a pre-established protocol, the assessment of the risk of bias, and the systematic analysis of results, in addition to considering sources of heterogeneity between studies results. Limitations include the lack of high-quality evidence inherent in observational study designs and the overall critical risk of bias in included studies.

Finally, although the present systematic review has identified some individual POPs associated with both breast cancer risk and prognosis that deserve further investigation, it should be emphasized that different POPs have different metabolic profiles and can have synergistic or antagonistic effects, and that proportions of different POPs may vary from one person to another. Thus, approaches considering the simultaneous exposure to different POPs may be more relevant than the isolated analysis of individual POPs.

#### **5. Conclusions**

Over the past three decades, numerous epidemiological studies have attempted to assess the association between exposure to POPs and breast cancer. Despite the apparent inconsistencies between studies, which were mainly due to methodological biases and to differences in the biological sample used for exposure measurement, when considering all studies (peripheral blood and adipose tissue) and all outcomes together (risk and prognosis), there was a trend toward a positive association between exposure to POPs and breast cancer that deserves further investigation. Future studies need to use rigorous methodology by including the relevant study population, using an appropriate biological sample for POPs measurement, controlling properly for confounding and assessing combined effects of POPs.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2072-6694/11/8/1063/s1, Table S1: Search strategy for Medline via PubMed, Table S2: Studies not included in the qualitative synthesis, Table S3: Studies of POPs measured in peripheral blood and breast cancer risk, Table S4: Studies of POPs measured in breast adipose tissue and breast cancer risk, Table S5: Studies of POPs measured in adipose tissue other than breast and breast cancer risk, Table S6: Studies of POPs measured in peripheral blood and mortality among breast cancer patients, Table S7: Studies of POPs measured in adipose tissue and mortality among breast cancer patients, Table S8: Studies of POPs and breast cancer prognostic factors, Table S9: Results of studies of POPs and breast cancer risk, Table S10: Results of studies of POPs and mortality in breast cancer patients, Table S11: Results of studies of POPs and prognostic factors in breast cancer patients, Table S12: Summary results of POPs associated positively with breast cancer risk, Table S13: Summary results of POPs associated negatively with breast cancer risk, Table S14: Summary results of POPs associated positively with mortality among breast cancer patient, Table S15: Summary results of POPs associated negatively with mortality among breast cancer patients, Table S16: Summary results of POPs associated positively with breast cancer prognostic factors.

**Author Contributions:** Conceptualization, K.E.-I. and C.D.; methodology, K.E.-I. and C.D.; formal analysis, K.E.-I. and C.D.; resources, C.D.; writing—original draft preparation, K.E.-I.; writing—review and editing, C.D. and P.A.; supervision, C.D.

**Funding:** This research received no external funding. K.E.-I. is a recipient of the Vanier Canada Graduate Scholarship. C.D. was a recipient of the Canadian Breast Cancer Foundation-Canadian Cancer Society Capacity Development award (award #703003) and holds a Senior Investigator Award from the Fonds de recherche du Québec-Santé.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Review*

### **Regulation of Cell Signaling Pathways by Berberine in Di**ff**erent Cancers: Searching for Missing Pieces of an Incomplete Jig-Saw Puzzle for an E**ff**ective Cancer Therapy**

**Ammad Ahmad Farooqi 1, Muhammad Zahid Qureshi 2, Sumbul Khalid 3, Rukset Attar 4, Chiara Martinelli 5, Uteuliyev Yerzhan Sabitaliyevich 6, Sadykov Bolat Nurmurzayevich 7, Simona Taverna 8, Palmiro Poltronieri <sup>9</sup> and Baojun Xu 10,\***


Received: 7 March 2019; Accepted: 25 March 2019; Published: 4 April 2019

**Abstract:** There has been a renewed interest in the identification of natural products having premium pharmacological properties and minimum off-target effects. In accordance with this approach, natural product research has experienced an exponential growth in the past two decades and has yielded a stream of preclinical and clinical insights which have deeply improved our knowledge related to the multifaceted nature of cancer and strategies to therapeutically target deregulated signaling pathways in different cancers. In this review, we have set the spotlight on the scientifically proven ability of berberine to effectively target a myriad of deregulated pathways.

**Keywords:** berberine; signaling pathways; oncogenic cascades; TRAIL; microRNAs; cancer therapy

### **1. Introduction**

Berberine, a natural alkaloid compound, is found in several medicinal plants. Typically, berberine is commercially produced from a Chinese medicinal plant *Coptis chinensis*. Berberine has captivated a substantial proportion of appreciation because of its remarkable pharmacological properties. Recent advancements in high-throughput techniques have helped us to demystify various hierarchically organized signaling complexes which play an instrumental role in cancer development and progression. Basic reviews related to the ability of berberine to improve worsening conditions in different diseases

have previously been published, so our aim is not to summarize pharmacological importance of berberine in different diseases, but we have restricted our discussion specifically to berberine mediated targeting of signaling cascades in different cancers. In this review, we present current views related to how berberine effectively targets different deregulated oncogenic cascades and highlight key practical and conceptual questions that will be helpful to shape the next dimension of investigation into the ability of berberine to efficiently target different signaling cascades. We will start our overview with one of the most widely investigated cancer killing molecules: TNF-related apoptosis-inducing ligand (TRAIL).

#### **2. Berberine Mediated Restoration of TRAIL-Mediated Apoptosis**

There has always been a quest to identify the molecules having significant cancer killing activity and minimal off-target effects. In accordance with this view, the discovery of TRAIL revolutionized the field of molecular oncology [1,2]. However, initial claims were partially challenged by contemporary researchers because TRAIL was ineffective against different cancers. In-depth studies revealed that TRAIL transduced the signals intracellularly through death receptors (DR4, DR5) [3,4]. However, loss of cell surface appearance of death receptors was a frequently noted mechanism in TRAIL-resistance cancers. Moreover, imbalance of pro- and anti-apoptotic proteins was also reported in TRAIL resistant cancers. The advent of high-throughput technologies has helped us to uncover the highly orchestrated nature of TRAIL mediated signaling, which is initialized through extrinsic and intrinsic pathways. In this section, we will summarize recent developments and discuss unresolved and outstanding research questions.

Berberine has been shown to potently induce AMP-activated protein kinase (AMPK) in cancer cells [5]. Expectedly, berberine mediated apoptosis inducing effects were severely impaired in AMPKα-dominant negative (DN) expressing or AMPKα knockdown cancer cells. Knockdown of DR5 significantly abrogated TRAIL-berberine-induced apoptosis [5]. TRAIL and berberine combinatorially enhanced p38-MAPK phosphorylation [6]. p38-MAPK inhibition enhanced apoptosis in EGFR (epidermal growth factor receptor)-overexpressing MDA-MB-468 TNBC cells [6].

TRAIL and berberine significantly activated caspase-3 and cleavage of PARP in TRAIL-resistant MDA-MB-468 BCa cells [7]. In a murine 4T1 BCa model, berberine potentiated the efficaciousness of the anti-DR5 antibody and effectively blocked tumor growth and lung metastases [7]. Mcl-1 and c-FLIP have been shown to negatively regulate TRAIL-mediated apoptosis. Berberine dose-dependently induced degradation of Mcl-1 and c-FLIP [8]. However, treatment with a proteasome inhibitor MG132 interfered with berberine-mediated downregulation of Mcl-1 and c-FLIP [8].

While great efforts over the past few years have advanced our understanding of the berberine mediated regulation of the TRAIL-mediated pathway, much of this work has focused on preliminary information about the ability of berberine to improve TRAIL-induced killing activity. There are still many questions which need detailed research, for example, how berberine mechanistically regulates expression of death receptors in different cancers. Does it inhibit receptor degradation, or does it interfere with epigenetic silencing to restore expression of death receptors? How does berberine improve formation of death inducing signaling complex (DISC) while simultaneously targeting negative regulators which inhibit DISC formation? In the following section, we will discuss how berberine modulates the WNT/β-catenin pathway to inhibit cancer.

#### **3. Regulation of WNT Pathway by Berberine**

Levels of cytoplasmic β-catenin are controlled by a multi-proteins destruction complex which induces β-catenin phosphorylation, which is required for β-catenin ubiquitination and its subsequent degradation by proteasomes [9]. Berberine efficiently inhibited nuclear accumulation of β-catenin. Co-immunoprecipitation studies revealed that berberine increased the interaction between APC (adenomatous polyposis coli) and β-catenin [9] (shown in Figure 1). These findings shed light on the

ability of berberine to stimulate the expression of APC and negatively regulate β-catenin by increasing physical interaction between these proteins.

**Figure 1.** (**A**) WNT/ β-catenin mediated intracellular signaling. β-catenin moved into the nucleus to stimulate expression of target genes. (**B**) Berberine promoted interaction of β-catenin and APC to enhance degradation of β-catenin. (**C**) Berberine also promoted interaction of β-catenin with c-Cbl in nucleus that also induced degradation of β-catenin. (**D**) Berberine inhibited phosphorylation of LRP5/6 and GSK-3β. Abbreviations: c-Cbl (CASITAS B-lineage lymphoma protooncogene), LRP5/6 (Low density lipoprotein receptor-related protein), CK (Casein Kinase), GSK-3β (Glycogen synthase kinase), RXR (Retinoid X-receptor), WNT (Wingless/Integrase), SAMP (Ser-Ala-Met-Pro motif), RING (Really interesting new gene).

There is evidence of mechanistic regulation of the WNT pathway by berberine in hepatocellular carcinoma cell line (SMMC-7721). However, these findings should be tested in other cancers. In a report, researchers demonstrated that berberine worked synergistically with HMQ1611, a taspine derivative and suppressed the phosphorylation of LRP5/6 and GSK3β [10]. Berberine and HMQ1611 combinatorially downregulated WNT5A, Frizzled8, CK1 (casein kinase 1) and APC. Overall this study uncovered distinct steps of β-catenin phosphorylation and degradation by berberine and HMQ1611 [10]. Importantly, berberine was unable to significantly inhibit tumor growth individually in mice xenografted with SMMC-7721 cells.

Berberine decreased levels of WNT5A and cytoplasmic β-catenin in both SGC7901 and AGS cells [11]. Treatment with berberine and galangin decreased β-catenin and WNT3A in esophageal carcinoma cells [12]. Berberine concentration-dependently downregulated mRNA expression of β-catenin in colon cancer cells [13]. RXRα (retinoid X receptor α) agonists have been shown to promote RXRα binding to β-catenin to induce ubiquitination and degradation of β-catenin [14]. Berberine dose-dependently enhanced physical association of RXRα with β-catenin. Berberine induced nuclear translocation of E3 ubiquitin ligase c-Cbl which modulated degradation of β-catenin [14] (shown in Figure 1). Overall, these findings clearly suggested that berberine enhanced degradation of β-catenin by promoting its interaction with c-Cbl.

It seems clear that berberine has potential to regulate the WNT pathway in different cancers, but it needs to be tested tactically in xenografted mice. Future studies must converge on identification of the modes opted by berberine to inhibit the WNT pathway in different cancers. Does it interfere with importin and exportin proteins to inhibit nuclear accumulation? Are there further previously unexplored ubiquitin ligases which can target β-catenin? Can berberine effectively target LRP5/6 and Frizzled receptors in other cancers to also efficiently inhibit cancer proliferation? In the upcoming section we will highlight how berberine regulates the Janus kinases-signal transducer and activator of transcription proteins (JAK-STAT) pathway in different cancers.

#### **4. Targeting of JAK-STAT Pathway**

Kinases of the Janus kinase (JAK) family and transcriptional factors of the STAT (signal transducer and activator of transcription) family form a highly dynamic and orchestrated membrane-to-nucleus signaling module that has been extensively investigated, and an overwhelmingly increasing list of scientific reports have provided evidence of natural products mediated targeting of the JAK-STAT pathway in different cancers. More importantly and excitingly, coupling of massively parallel DNA sequencing with chromatin-immunoprecipitation has enabled researchers to capture thousands of STAT-binding sites.

Berberine reduced protein levels of STAT3 and inhibited the phosphorylation at 705th tyrosine and 727th serine in cholangiocarcinoma cell lines [15] (shown in Figure 2). Berberine exerted inhibitory effects on constitutive and IL-6-triggered activation of STAT3 in NPC (nasopharyngeal carcinoma) cells [16]. TAFs (tumor-associated fibroblasts) secreted IL-6 and the conditioned media harvested from the fibroblasts induced STAT3 activation in NPC cells. Activation of STAT3 by conditioned media of TAFs was blocked by berberine [16].

**Figure 2.** JAK-STAT signaling. Berberine has been shown to inhibit phosphorylation of STAT3 and STAT5. Berberine markedly reduced mRNA levels of STAT3. Functionally active STAT3 moved into the nucleus and stimulated expression of miR-21. Abbreviations: STAT (signal transducer and activator of transcription), JAK (Janus kinase), PIAS (Protein inhibitors of activated STATs), PTP (Protein-tyrosine phosphatase), SHP-2 (Src homology phosphatase-2).

Berberine significantly decreased the phosphorylated levels of JAK2 and STAT3 in colorectal cancer cells [17]. Interestingly, p-JAK2 and p-STAT3 were found to be remarkably enhanced in COX2 (cyclooxygenase-2) overexpressing colorectal cancer cells. COX2 overexpression induced activation of JAK-STAT signaling further upregulated matrix metalloproteinases (MMP)-2 and MMP-9 in colorectal cancer cells. However, berberine effectively interrupted COX2/JAK/STAT signaling [17].

Colonization of *Fusobacterium nucleatum* in the intestine may contribute to colorectal cancer (18). Levels of p-STAT3 and p-STAT5 were found to be enhanced after inoculation of *F. nucleatum* in C57BL/6-APCMin/+ mice and wild-type C57BL/6 mice. Moreover, *F. nucleatum*-induced increases in quantities of p-STAT3 and p-STAT5 were found to be considerably reduced in mice treated with berberine [18] (shown in Figure 2).

Doxorubicin, a widely used chemotherapeutic drug has been shown to induce activation of STAT3 in lung cancer cells [19]. However, berberine markedly inhibited doxorubicin-triggered STAT3 activation. Besides, berberine promoted degradation of STAT3 by enhancing ubiquitination [19]. Berberine also enhanced killing effects of 5-fluorouracil by STAT3 inactivation and repressing the expression of survivin in gastric cancer cells [20].

#### **5. Targeting of the mTOR Pathway: Could Berberine Modify Extracellular Vesicle-Composition?**

The mammalian target of rapamycin (mTOR), a serine/threonine kinase of the PI3K (phosphoinositide 3-kinases)-related kinase family, is conserved on an evolutionary scale that coordinates different cellular processes. mTOR forms two structurally and functionally active complexes: mTOR complex 1 (mTORC1) and 2 (mTORC2). These two multi-component complexes are involved in physiological and pathological functions, such as macromolecules synthesis, homeostasis maintenance, cytoskeleton remodeling, angiogenesis, survival, response to stress and autophagy [21]. Considering the key role of mTOR in cell proliferation and differentiation, its deregulation contributes to cancer onset and progression [22].

The cellular metabolism mediated by mTOR is involved in the connections between cancer cells and tumor microenvironment during cancer advance and drug resistance acquisition, indicating the potential benefits of PI3K-Akt (protein kinase B)-mTOR pathway blockage. This inhibition contributes to reduce proliferation, migration, and survival of cancer cells, and increase tumor immunosurveillance through down-regulation of immunosuppression and anti-tumor immune stimulation [23]. mTORC1 is a downstream component in several pathways frequently altered in cancer, including the PI3K/Akt and MAPK (mitogen-activated protein kinases) pathways, that induces mTORC1 hyperactivation in many human cancers. Besides, mTORC2 signaling has a key role in tumors for its role in Akt activation, that induces tumor growth mechanisms such as glucose metabolism and apoptosis inhibition [24]. Recent reports show that mTOR is involved in lipid metabolism [25]; the critical step of this signaling cascade is the activation of proteins by phosphorylation at different sites. An important upstream target of mTOR is ERK (extracellular signal-regulated kinases), that regulates mTOR negatively and in turn enhances autophagy. ERK, a protein of MAPK cascade, is a central integrator of extracellular signals which are transduced by single cytokines or hormones or activated by cellular mechanical stresses that influence lipid metabolism [26,27].

The mTOR inhibitors, called "rapalogs," used as anti-cancer drugs belonged to a class of rapamycin derivatives. The first rapalog, approved for advanced renal carcinoma management was temsirolimus, followed by everolimus. These rapalogs did not show significant results in clinical practice as compared to the results obtained in pre-clinical studies. The rapalogs and catalytic mTOR inhibitors were useful in immunosuppression in a small number of cancers [28,29].

Several natural compounds such as berberine, resveratrol, curcumin, quercetin and others can modulate the mTOR pathway [30–32]. Recent studies have revealed that berberine has anti-tumor effects, through inhibition of the mTOR-signaling pathway. Berberine, anisoquinoline alkaloid isolated from *Berberis vulgaris* L, has anti-diarrheic, anti-inflammation, and anti-microbial activities [33]. Nowadays, several studies have shown that berberine is effective against glioma, colorectal, lung, prostate and ovarian cancer [26].

Berberine can modulate different pathways, such as cellular glucose metabolism and the HIF-1α (hypoxia-inducible factor 1α)-mTOR axis. In this context, Wang et al. [26] indicated that berberine modulated the metabolism of glioblastoma multiforme cells, induced autophagy and reduced glucose metabolism. These changes reduced tumor growth and invasiveness, induced apoptosis, by AMPK/mTOR/ULK1 (Unc-51 like autophagy activating kinase) pathway inhibition. Berberine reduced cancer progression in vivo, which clearly indicated the potential clinical benefits of alkaloids extract from plants in cancer therapy [26].

Mao et al. provided evidence that berberine played a central role in regulation of cellular glucose metabolism in colon cancer cells [34]. They studied the effects of berberine in colon cancer cell lines and findings revealed that berberine inhibited glucose uptake and reduced the transcription of genes, such as *GLUT1* (glucose transporter 1), *LDHA* (lactate dehydrogenase A) and *HK2* (hexokinase 2), involved in glucose metabolism of colon cancer cells. This mechanism is mediated by HIF-1α protein synthesis inhibition through mTOR pathway suppression. The molecular study indicated that HIF-1α protein expression, a well-known transcription factor critical for dysregulated cancer cell glucose metabolism, was considerably inhibited in berberine-treated colon cancer cell lines [34]. It was reported that berberine activated AMPK that in turn inhibited mTOR, in in vitro studies and in mouse models of colon carcinogenesis in early stages of tumorigenesis. Berberine also interfered with the NF-κB (nuclear factor kappa-light-chain-enhancer of activated B cells) pathway and effectively inhibited colon cancer progression [33].

Berberine may also induce autophagy in human liver carcinoma cell lines, through activation of Beclin-1 and inhibition of mTOR signaling by suppressing the activity of Akt and up-regulating P38 MAPK signaling [35,36]. The role of berberine in mTOR pathway modulation has been also demonstrated in hematological malignancies. Ma and collaborators showed the synergism of TPD7 and berberine in leukemia Jurkat cell growth inhibition through ephrin-B2 signaling modulation [37]. There are direct pieces of evidence which shed light on synergistic antitumor activities of rapamycin and berberine treatment in hepato-carcinoma cell lines. There was a marked decrease in phosphorylated p70S6 kinase 1 protein levels, a downstream effector of mTOR in cells combinatorially treated with rapamycin and berberine as compared to cells treated with rapamycin or berberine alone [38]. It was also demonstrated that berberine and cinnamaldehyde reduced the susceptibility of mice to lung carcinogenesis induced by urethane, and reversed the urethane-induced AMPK, mTOR, AQP-1 (aquaporin 1) and NF-κB expression patterns [39]. Overall these reports advocated the role of berberine as a new compound for cancer therapy.

Recent findings indicate that extracellular vesicles (EVs) play a key role in different steps of cancer progression, transporting oncogenic proteins and nucleic acids [40–42]. EVs are named in different ways based on their origin, diameter and mechanism of release. The two population of EVs better characterized are exosomes and micro-vesicles [43,44]. Hypoxia induces wide changes in the tumor microenvironment, and several reports show EVs central role in this mechanism [45]. Besides HIF-1, other pathways such as PI3K/Akt/mTOR are induced in tumor cells under hypoxia. It was demonstrated that hypoxia promoted prostate cancer progression and hypoxis-induced-exosomes remodeled the cancer microenvironment [46]. Moreover, exosomes released by mesenchymal stem cells (MSC) dose-dependently reduced VEGF (vascular endothelial growth factor) expression and secretion mainly through interfering with the mTOR/HIF-1α axis in breast cancer cells. MSC-derived exosomes, enriched in miR-100, were taken up by breast cancer cells [47]. microRNA-100 efficiently downregulated VEGF in breast cancer cells.

This evidence suggests the possible role of berberine in modifying EV-composition, as has already been demonstrated for other natural compounds such as curcumin. It is exciting to note that exosomes released by curcumin-treated CML cells contained considerably higher levels of miR-21. Consequently, these miR-21 loaded exosomes were taken up by HUVECs (human umbilical vein endothelial cells) and it was mechanistically shown that miR-21 directly targeted MARCKS (myristoylated alanine-rich C-kinase substrate) and inhibited angiogenic phenotypes [48]. Curcumin also induced selective packaging of miR-21 in exosomes and played a central role in reshaping post-transcriptional network in recipient cells [49]. Furthermore, in CML cells, curcumin modulates other molecular pathways thus altering the metabolism of glucose that in myeloproliferative disease is a consequence of non-hypoxic activation of HIF-1α [50]. It was demonstrated that curcumin promoted miR-22 mediated targeting of importin 7 that resulted in a significant reduction in nuclear accumulation of HIF-1α [51].

It will be interesting to see if berberine demonstrated potent activity to promote release of exosomes loaded with tumor suppressor microRNAs and proteins.

#### **6. Regulation of Epigenetic Modulators by Berberine**

Histone marks are motives enabling the recruitment of chromatin complexes that activate or repress transcription. Histone modifications such as acetylation and methylation at specific positions are signals recognized by these complexes. Berberine was shown to upregulate some histone deacetylases (HDAC) of class II, such as sirtuin SIRT1 (sirtuin 1), producing an antiatherogenic effect, and suppression of foam cell formation in THP-1-derived macrophages treated with oxidized low-density lipoprotein [52]. RNA silencing of SIRT1 or AMPK blocked the berberine action.

Rel proteins have emerged as complex modulators of carcinogenesis and we still have to explore their functionalities in malignancies and response to cancer therapeutics [53]. Set9 (lysine methyltransferase) induced methylation of the RelA/p65 subunit, which inhibited nuclear accumulation of NF-κB and repressed transcriptional upregulation of miR-21. Berberine dose-dependently induced generation of ROS, arrested cancer cells in G(2)/M phase and induced apoptosis in U266 cells [53]. Overall, the findings clearly suggested that berberine promoted Set9 mediated methylation of p65 to limit shuttling of NF-κB into the nucleus. Berberine mediated inhibition of translocation of NF-κB into the nucleus resulted in inhibition of miR-21 and B-cell lymphoma 2 (Bcl-2).

The protective effects of berberine against metabolic syndrome might rely on increasing mitochondrial SIRT3 activity and stimulating glycolysis, independent of AMPK activation [54,55].

The growth arrest and DNA damage-inducible protein GADD45α (growth arrest and DNA damage 45α) is a DNA demethylation regulator recruited by TCF21 antisense RNA inducing demethylation (TARID) lncRNA to enable transcription of the TCF21 gene coding for tumor-suppressor gene transcription factor 21 [56]. As detailed in the next paragraphs, administration of *Coptidis rhyzoma* aqueous extract, resulted in a higher expression of miR-23a, and in up-regulation of GADD45a, a chromatin relaxer, decreasing DNA methylation through recruitment of the 5-hydroxymethylcytosine (5hmC) hyperproducing enzyme TET, and thymine DNA glycosidase (TDG) [57–59].

Berberine induced a decrease in activity of two DNA methylases, DNMT1 (DNA (cytosine-5)-methyltransferase 1) and DNMT3, that, through DNA hypomethylation, induced an increase of p53 [60]; p53 activation was observed following the increase in miR-23a, through repression of Nek6, in HCC cells in response to Rhizoma Coptis aqueous extract [60].

Various long noncoding RNA (lncRNAs), through their tertiary structure, work as scaffolds to recruit protein complexes such as chromatin modifiers, polycomb repressive complex (PRC), and transcriptional regulators to euchromatin regions: many lncRNAs have been associated with pharmacological drugs as well as to cisplatin treatment [61]. In various studies, berberine has been able to induce or repress some lncRNA [62].

#### **7. Regulation of microRNAs by Berberine**

Many bioactive compounds have been proposed for their regulatory effects on non-coding RNAs, either long ncRNAs (lncRNA), or small RNAs such as microRNAs [63–65]. These miRNAs have added new layers of complexity in the context of post-transcriptional regulation, controlling the availability of mRNAs, and selection of the mRNAs that they recognize as targets by complementary seed sequences to initialize the degradation process of mRNAs. miRNAs exert a regulatory role in the post-translational process.

OncomiRs are oncogenic due to their ability to support cell proliferation, apoptosis inhibition, cell stemness, while tumor suppressor miRNAs are involved in nodes or networks leading to differentiation, cell cycle inhibition and growth arrest and apoptosis.

There are feedback loops and feedforward loops, sustaining the oncogenic activity of ncRNAs: for instance, a feedback loop has been described in lymphoma between miR-17-92, MYC, the protein kinase Chk2 and hu antigen R (HUR) [66]. Transcription factors can induce the expression of protein-coding genes as well as of miRNAs, that can target the mRNAs of the induced genes, in a feedforward loop. In lymphoma cells, miR-17-92 regulates MYC mRNA levels through the inhibition of Chk2, causing the depression of RNA-binding protein HUR, and its binding to MYC mRNA, preventing MYC translation. Thus, berberine suppresses the growth of multiple myelomas, either by down-regulation of three oncogenic miRNA clusters and other mRNAs, or by involvement of p53 and MAP kinases [67].

Several researchers pointed out to a correlation between berberine treatment and expression of non-coding RNAs, either lncRNAs or microRNAs. In cancer studies, treatment of multiple myeloma cells with berberine, significantly suppressed three oncogenic miRNA clusters, miR-17-92, miR-106-25, and miR-99a-125b. Berberine mediated downregulation of miR-99a-125b was found to be correlated with the regulation of p53, MAP Kinases and ErbB oncogene, leading to cell cycle arrest in the G2-phase and to apoptosis [63].

In colon cancers, berberine was effective in downregulating miR-429, with increase in its target, SOX-2; berberine up-regulated miR-296-5p and effectively interfered with the Pin1–β-catenin–cyclin D1 signal transduction cascade. Berberine inhibited the growth of HepG2 cells and induced the upregulation of miR-22-3p. miR-22-3p directly targeted SP1 and suppressed expression of its target genes, BCL2 (B-cell lymphoma 2) and CCND1 (cyclin D1) [68] (shown in Figure 3).

It was shown that berberine suppresses interleukin 6 (IL6), a factor required for cell growth in multiple myeloma cells (U266), through negative regulation of the signal transducer and activator of transcription 3 (STAT3), and this induces inhibition of miR-21 expression [69]. STAT3 regulates miR-21 expression through binding to STAT3 binding sites in the promoter (shown in Figure 2). Additionally, in ovarian cancer cells (SKOV3), berberine sensitized to cisplatin treatment through inhibition of miR-21, and subsequent activation of PDCD4, a tumor suppressor.

It is not straightforward to determine the role of miRNAs in tumor promotion or suppression, that depends on the context-specific, cell-type specific dual role of certain miRNAs [66]. This depends on the varieties of targets of miRNAs. There are miRNAs, such as miR-25 and miR-125b, that act as tumor suppressors in some tumor types and as oncogenes in others. In particular, in stem cells, miRNAs that in other cell types are regulated and respond to the bioactive supplements, may not decrease in levels, for their role in cell stemness. Berberine, in the form of a Rhizoma Coptis aqueous extract, resulted not effective in decreasing miR-21 levels, while it supported a higher expression of miR-23a, by up-regulating p21/GADD45a tumor suppressor gene (shown in Figure 3), causing HCC cells to arrest the growth in G2/M phase [19,70]. Furthermore, miR-23a was shown to repress Nek6 and to regulate p53 transcriptional activity.

Autophagy influences glucose and lipid metabolism in adipocytes. Berberine was shown to decrease miR-30a and miR-376b, preventing basal autophagy in 3T3-L1 adipocytes. MiR-30 interacts with the 3- -untranslated region of Beclin 1 (BECN1), thus the reduction in miR-30a levels increased BECN1 to form BECN1 complexes that induce autophagy (beclin homolog in yeast, Atg6, is known as autophagy promoting factor) [71]. This leads to reduced fat deposits and an increase in brown fat tissue.

Berberine was shown to exert an anti-apoptotic role in development of preimplantation embryos in vitro, by maintaining high levels of miR-21. Berberine up-regulated Bcl-2, in 2- and 4-cell embryos and blastocysts, and down-regulated caspase-3 and PTEN.

When the pre-miRNA is processed by the RISC complex, the guide strand is considered the mature, active miRNA, while the complementary passenger strand (termed miRNA\*) is thought to be devoid of function. However, some miRNA star has been related to a function, some have a feedback role to regulate the RISC processing, and some have been shown to have anti-oncogene activity. This seems the case for miR-21\*, or miR21-3p, affecting cancer cell growth [72]. In HepG2 hepatocellular carcinoma cells, berberine increased the levels of miR-21-3p, with a role in tumor growth inhibition and induction of apoptosis. In hepatic cancers, miR-21\* antitumor activity relies on inhibition of MAT2A and MAT2B methionine adenosyltransferases mRNAs, with consequent increased levels of S-adenosyl-methionine (SAM). Methionine adenosyltransferase (MAT) played a central role in growth of hepatoma cells. miR-21-3p had previously been shown to directly target MAT2A and MAT2B in HepG2 cells. Berberine induced apoptosis in HepG2 cells mainly through miR-21-3p mediated targeting of MAT2A and MAT2B [72].

**Figure 3.** Regulation of non-coding RNAs by berberine. (**A**) CASC2 interacted with AUF1 and prevented its binding to AU rich sequences present within mRNA of Bcl-2. (**B**,**C**) AP-1 transcriptionally upregulated miR-101. miR-101 directly targeted COX-2. (**D**–**G**) P53 induced transcriptional upregulation of miR-23a. Additionally, miR-23a worked synchronously and stimulated the expression of GADD45a and p21. miR-22-3p directly targeted SP1. SP1 mediated upregulation of CCND1 and BCL2. Abbreviations: AUF1 (AU-rich element RNA-binding protein-1), BCL2 (B-cell CLL/lymphoma-2), *GADD45*α (Growth arrest- and DNA damage-inducible gene), CCND1 (Cyclin D1), CASC2 (long non-coding RNA), SP1 (specificity protein-1), AP-1 (Activator protein-1).

Ovarian cancer cells resistant to cisplatin, such as A2780 and A2780/DDP lines, were incubated with berberine combined with cisplatin, showing a significantly lower survival rate [73]. This effect was found to be related to inhibition of miR-93 expression, that translated in re-expression of PTEN tumor suppressor and recovery of AKT signaling [73]. Bcl-w has been shown to be targeted by miRNAs. In gastric cancers [60], berberine upregulated miR-203 and restored cisplatin-sensitivity in gastric cancer cells [74].

In liver cancers, berberine inhibited cell proliferation and viability in HepG2, Hep3B, and SNU-182 lines. Berberine treatment increased the expression of tumor suppressor such as Kruppel-like factor 6 (KLF6), activating transcription factor 3 (ATF3) and p21, a cell cycle inhibitor, and down-regulated the oncogene E2F transcription factor 1 (E2F1). A possible mechanism of the upregulation of protein coding genes may be hypothesized through downregulation of the respective miRNAs.

Berberine supplementation led to the miR29-b suppression, increasing insulin-like growth factor-binding protein (IGFBP1) expression in the liver; miR29-b suppression caused an increase in AMPK activity and a reduction of lipid storage in diabetic and obese patients [60]. Activation of AKT positively affected glucose uptake, reducing the glucose levels in blood.

LncRNAs are deregulated after berberine treatment in hepatocellular cancer, similarly to the effects seen after curcumin treatment on lncRNAs and on epigenetic changes seen in hepatocellular cancer [75]. A lncRNA, ANRIL, expressed at increased levels in type 2 diabetic (T2D) patients, causing an increase of CREB (cAMP response element-binding protein) expression, may be affected by berberine. Berberine down-regulated miR-122 and this caused a decrease in SREBP-1 levels in palmitic acid-treated HepG2 cells. Berberine was able to reduce the levels of glucose in T2D patients and in nonalcoholic fatty liver disease, by down-regulating lncRNA052686 and miR-122 [61,62,71,76,77]. The hyperlipidemic effect of berberine was linked to the inhibition of C/EBPα and PPARγ2 expression through inhibition of phosphorylated CREB binding to C/EBPβ promoter. MRAK052686, a lncRNA downregulated in diabetes, was induced by berberine. MRAK052686 co-localize at the 3- UTR of Zbtb20, coding for a protein regulating glucose homeostasis. The co-expression of MRAK052686 and Zbtb20 increases the level of the protein, improving glucose homeostasis [62].

#### **8. Nanotechnological Strategies to Improve the Delivery of Berberine**

Due to its outstanding antitumoral properties, many efforts have been devoted in designing carriers for berberine delivery as an anti-cancer therapeutic agent. Both inorganic and organic nanomaterials have been exploited for this purpose (shown in Figure 4).

**Figure 4.** Berberine delivery strategies. On the left, inorganic nanocarriers are shown: Ag, Silver nanoparticles; ZnO, zinc oxide nanoparticles; IO, Iron oxide nanoparticles. On the right, organic nanocarriers are shown: SLN, solid lipid nanoparticles; NLC, nanostructured lipid carriers; Liposomes; Dendrimers; Lipopolymeric nanoparticles.

Silver nanoparticles proved successful in delivering berberine to human tongue squamous carcinoma SCC-25 cells, blocking cell cycle and increasing Bax/Bcl-2 ratio gene expression, thus indicating activation of pro-apoptotic pathways triggered by mitochondrial dysfunction [78]. Interestingly, silver nanoparticles carrying berberine displayed elevated cytotoxicity in different

breast cancer cell lines and inhibition of tumor growth in vivo [79]. The same research group fabricated citrate-capped silver nanoparticles loaded with berberine and conjugated to polyethylene glycol-functionalized folic acid and demonstrated that they were able to induce apoptosis and variations in gene expression in breast cancer cells. In MDA-MB-231 athymic nude mice models, a significative reduction of tumor progression was observed [80].

Zinc oxide nanoparticles carrying berberine were recently synthesized for lung cancer therapy. They displayed antiproliferative activity in A549 cells and no significant toxicity in vivo. Moreover, due to the intrinsic properties of ZnO, these materials have been exploited as photothermal therapy (PTT) agents, leading to thermal ablation of cancer cells [81].

An innovative approach has been recently adopted by creating iron oxide nanoparticles complexed with hypoxic cell sensitizer sanazole together with berberine, able to target hypoxic tumors in vivo. Hypoxia is known to induce expression of HIF-1-alpha and consequent activation of angiogenesis related genes. Interestingly, after administration of nanoparticles, transcriptional downregulation of these and many genes linked to cell proliferation and metastasis was observed and clearly correlated with a reduction of tumor volume in vivo [82].

Many studies have demonstrated the feasibility of designing organic nanoparticles for berberine delivery. Solid lipid nanoparticles have been synthesized with good stability, high berberine loading and huge entrapment efficiency, essential parameters for successful clinical evaluation. These nanomaterials inhibited cell proliferation of MCF-7, HepG2, and A549 cancer cells inhibited cell cycle progression and apoptosis in MCF-7 cells [83].

Nanostructured lipid carriers efficiently delivered berberine to H22 hepatocarcinoma cells with high antitumor efficacy [84]. Lipid nanoparticles covered with the self-tumor targeting polymers lactoferrin and hyaluronic acid were fabricated for berberine and rapamycin delivery to lung cancer cells and displayed improved internalization and selective cytotoxicity. Detectable reduction in the number of lung foci and vascular endothelial growth factor levels were further observed [85]. The same approach was adopted by administering inhalable nanoparticles to in vivo models of lung cancer, achieving a significant decrease in lung weight, reduction in lung adenomatous foci number and diameter and in angiogenic markers expression [86].

Amine terminated G4 PAMAM have been developed with conjugated berberine and demonstrated specific cytotoxicity in different human breast cancer cell lines [87]. Lipopolymeric micelles have been developed that greatly improved berberine water solubility up to 300% with low toxicity and induced apoptosis in treated monolayer and spheroid cultures of human prostate carcinomas [88]. Finally, berberine loaded folate acid modified chitosan nanoparticles demonstrated effective in inhibiting proliferation and migration and inducing apoptosis and necrosis in human nasopharyngeal carcinoma cells CNE-1 [89].

#### **9. Conclusions**

Berberine has emerged as an excellent natural product having significant biological activity [90]. It has demonstrated premium activity against different cancers. Increasingly sophisticated cutting-edge research has uncovered tremendous chemopreventive ability of berberine to modulate signaling pathways [91,92].

In the last few years, many studies have been focused on unraveling intrinsic properties of berberine and its ability to interfere with intracellular pathways. In particular, there is an increasing interest in exploiting this natural compound as an effective anticancer drug. Although some issues remain to be solved, such as its poor water solubility/stability and low bioavailability, many nanotechnological approaches have allowed the design of ad hoc delivery systems, making berberine application for cancer treatment feasible. Many different nanocarriers, based both on inorganic and organic materials, have been developed and proven to be effective in overcoming some of the above-mentioned issues and in delivering berberine in an extremely efficient manner in many different cancer experimental

models. In the near future, further research will provide crucial answers needed to pave the way to berberine clinical application.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Abbreviations**


### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

#### *Article*
