Histological evaluation of AMPK signalling in primary breast cancer

AMP-activated protein kinase (AMPK) is an intracellular energy-sensing kinase that is inactive unless it has been phosphorylated by upstream kinases at a specific threonine residue (Thr-172) within the kinase domain. Phosphorylation at Thr-172 and consequent activation occurs in response to metabolic stresses that deplete cellular energy levels and thus increase the AMP/ATP ratio [1, 2]. Activation of AMPK can be assessed using a phosphospecific antibody (anti-pAMPK) that recognizes either catalytic subunit isoform (α1 or α2), but only when phosphorylated at Thr-172. AMPK is proposed as a "fuel gauge" that monitors changes in the energy status of cells [2, 3]. It is activated by metabolic stresses that inhibit ATP production such as glucose deprivation [4], ischaemia [5], or hypoxia [6], or by stresses that accelerate ATP consumption such as contraction in skeletal muscle [7]. Once activated, AMPK down-regulates energy-consuming processes, including cell proliferation, thus ensuring that these processes only proceed if there are sufficient metabolic resources available [8]. AMPK switches on catabolic pathways that generate ATP, while switching off biosynthetic pathways and other processes that consume ATP, and hence has a key role in maintaining energy balance both at the single cell and the whole body levels [2, 9].

One direct downstream target of AMPK is acetyl-Coenzyme A carboxylase (ACC). In its active, de-phosphorylated form, ACC catalyzes the conversion of acetyl-CoA to malonyl-CoA in the de novo lipid synthesis pathway [3, 10, 11]. When it is inactivated by phosphorylation, a decrease in malonyl-CoA occurs, thereby increasing the mitochondrial import and oxidation of long chain fatty acids (LCFAs), resulting in the generation of ATP [12]. ACC exists as two isoforms, ACC1/a and ACC2/b, with ACC1 being thought to produce the pool of malonyl-CoA involved in fatty acid synthesis, while ACC2 is thought to produce the mitochondrial pool that regulates fatty acid oxidation [13, 14]. AMPK phosphorylates ACC1 at multiple residues, although phosphorylation at a single serine (Ser-79 in rat and Ser-80 in human ACC1) accounts for the resulting inactivation [15, 16]. AMPK also inactivates ACC2 [17] via phosphorylation at the site equivalent to Ser-79 on ACC1 (Ser-221 in human ACC2). A phosphospecific antibody (anti-pACC) recognizes both ACC1 and ACC2 phosphorylated at these sites, and is a widely used marker for AMPK activation. In isolated hepatocytes from AMPK-α1^-/- and -α2^-/- double knockout mice, the signal obtained using this antibody is completely lost, showing that AMPK is responsible for phosphorylation at these sites [18].

There is a large body of evidence to support the link between metabolic disorders, such as obesity and type 2 diabetes, and dysfunctional lipid and energy metabolism causing increases in circulatory and intracellular fatty acids [10, 12, 19]. High levels of fatty acids are toxic to cells and may cause deleterious metabolic abnormalities [12, 19]. These unwanted effects could be prevented by activation of AMPK and consequent inactivation of ACC2 in peripheral tissues, leading to an increase in fatty acid oxidation [12, 19]. In addition many cancer cells, including many breast tumours, exhibit a markedly increased rate of fatty acid synthesis [20]. Some breast tumours express high levels of fatty acid synthase (FAS), a key enzyme for fatty acid biosynthesis [21]. FAS is up-regulated in about 50% of breast cancers, is an indicator of poor prognosis, and is associated with the HER2 oncogene [22‐24]. The drug Orlistat (Xenical™), which was developed as an inhibitor of gastric and pancreatic lipases and has been approved for weight loss by the FDA, blocks breast cancer cell cycle progression, promotes apoptotic cell death, and transcriptional repression of the HER2 oncogene [25]. In addition to these potentially beneficial effects of AMPK activation on lipid metabolism, AMPK would also be expected to directly inhibit cell growth and proliferation. The major kinase that phosphorylates Thr-172 on AMPK has been identified to be the tumour suppressor kinase, LKB1 [26, 27]. Although LKB1 is also the upstream kinase for a family of twelve additional AMPK-related kinases [28, 29], many of the tumour suppressor effects of LKB1 are likely to be mediated by AMPK. For example, AMPK activation causes a G1-S phase cell cycle arrest [30, 31], while it also inhibits cell growth by switching off the target-of-rapamycin (TOR) pathway [32, 33]. Consistent with the idea that AMPK activation might inhibit growth of tumour cells were findings that the oral anti-diabetic drug, metformin, inhibits proliferation of epithelial cells derived from breast, prostate and ovarian cancers, an effect that requires both LKB1 and AMPK [34]. These findings may be clinically relevant, because epidemiological studies show an association between a lower incidence of cancer, including breast cancer, and metformin usage in diabetic patients [35, 36].

Despite growing evidence in this arena, little is known about the state of phosphorylation of AMPK, and its downstream target ACC, in human breast cancers. Since malignant cells are likely to have an increased energy demand and may also experience hypoxia and other metabolic stresses, one might speculate that AMPK signalling would be activated. The objective of the present study was to evaluate the status of phosphorylation of AMPK and ACC signalling in human primary breast cancer using immunohistochemistry, and to correlate this with the clinical and pathological characteristics of the tumours.

For the first cohort of the 117 patients, 104 (88.9%) were infiltrating ductal carcinoma, 3 (2.6%) infiltrating lobular and 10 (8.5%) mixed or other types. All cases were successfully scored for Ki67 and HER2; 3/117 (2.6%) for pACC, and 4/117 (3.4%) for pAMPK were not evaluable on the TMAs.

In the second cohort of 237 patients, 197 (83.1%) were infiltrating ductal carcinoma, 16 (6.7%) infiltrating lobular carcinoma, and 34 (14.3%) were mixed and other types. One case (1/237 (0.4%)) was excluded from scoring for pAMPK and 19/237 (8%) for pACC were not evaluable (Table 1).

Cytoplasmic staining was seen for pAMPK and, as expected, nuclear staining for Ki67 and ERα; combined cell membrane and cytoplasm staining for pACC; cell membrane strongly and cytoplasm lightly for HER2. Normal breast epithelium did not stain for Ki67, HER2 or pACC; however strong staining of normal breast epithelial cytoplasm was observed for pAMPK (Figure 1A).

Reduced AMPK phosphorylation signal (scores 0 - 2) was seen in the vast majority of the breast cancer specimens (101/113 [89.4%] and 217/236 [91.9%] of the first and second cohort respectively), as light staining of the cytoplasm (Figure 1B), compared to strong expression in normal breast epithelium (score 3). A significant positive association was seen between pAMPK and pACC signals (FET: p = 0.007 and 0.014 respectively). Reduced AMPK phosphorylation signal was associated with higher histological grade (χ ² for trend = 6.644, df = 1, p = 0.010; and χ ² for trend = 5.291, df = 1, p = 0.021). AMPK phosphorylation signal reduction was also associated with axillary node metastasis (FET: p = 0.061 and 0.039 respectively) (Table 2 and additional file 1)

Ki67 expression (Figure 1C) was negatively associated with ER status (FET: p = 0.002 and p < 0.001 respectively) (see additional file 1). It was also associated with higher grade tumours (FET: p < 0.001 in both cohorts), increased tumour recurrence (FET: p = 0.041 and 0.005, Relative Risk (RR) = 1.89 and 1.59, Odds Ratio (OR) = 2.39 and 2.47, respectively) and lower overall survival (Kaplan-Meier Log Rank Test: p = 0.023 and 0.015, RR = 1.62 and 2.13, OR = 2.29 and 23.39, respectively) (Figure 2).

HER2 amplification, found in 11.9% (14/117) and 13.9% (32/230) of patients respectively (Figure 1D), was significantly associated with Ki67 expression (FET: p = 0.006 and 0.021 respectively), higher histological grade (FET: p = 0.073 and < 0.001 respectively), and inversely with ER status (FET: p = 0.006 and 0.001 respectively). HER2 positive patients had increased recurrence (FET: p = 0.027 and 0.001, RR = 1.72 and 1.96, OR = 3.75 and 7.2, respectively); and lower overall survival (Kaplan-Meier Log Rank Test: p < 0.001 and 0.075, RR = 2.81 and 1.99, OR = 5.19 and 2.3, respectively) (Figure 3).

ER positive tumours, identified in 62.6% (72/115) and 73% (173/237) of patients respectively, were associated with HER2 negative (FET: p = 0.006 and 0.001 respectively), low Ki67 expression (FET: p = 0.002 and < 0.001 respectively), lower grade tumours (FET: p = 0.029 and < 0.001 respectively), and lower disease recurrence (FET: p = 0.033 and < 0.001, RR = 1.41 and 2.75, OR = 2.33 and 4.36, respectively). In the second cohort of patients, ER positive patients also lived longer (Kaplan-Meier Log Rank Test: p < 0.001, RR = 1.44, OR = 5.81) (Figure 4).

Malignant transformation of cells is characterized by insensitivity to growth-inhibitory signals, unlimited replicative potential, angiogenesis, tissue invasion and metastasis [38]. Despite the physiological stresses in cancer cells, and the associated hypoxia in both primary and metastatic cancer [39], malignant cells continue to proliferate and invade. These hallmark characteristics of cancer raise the question of whether dysfunction of AMPK could have a role in malignancy. The aim of this study was to evaluate phosphorylation of AMPK and ACC as markers of AMPK signalling in human breast cancer in relation to clinical and pathological data.

The two cohorts of patients were representative of previously reported breast cancer populations as exemplified by the associations between HER2 amplification, Ki67 expression, ER negative cancers, higher histological grade, Ki67 labelling index [40] and higher disease recurrence and lower overall survival. The significant differences in menopausal status, age, tumour size and node status between the two cohorts (Table 1) perhaps emphasizes the consistent trends and significant associations of AMPK signalling with node status and tumour grade respectively (Table 2). Further possible caveats include the use of IHC rather than western blotting to assess the signalling pathways, although the distribution of signalling in normal and cancerous tissues on IHC allowed localisation of the proteins.

Despite the high energy consuming processes of uncontrolled cellular proliferation in breast cancer, the histological evidence suggests that AMPK does not seem to respond in primary breast cancer. In contrast, normal breast epithelium demonstrated appropriate AMPK staining, but as expected for a quiescent epithelium no activation of ACC. Reduced signalling of AMPK phosphorylation in the majority of the breast cancer specimens (89.4% in the first and 91.9% in the second cohort) compared to strong expression in normal breast epithelial cells, and the inverse relationship between staining with histological grade and node status supports the current view that activating this energy-sensing enzyme may be a potential therapeutic avenue in breast cancer [8]. Lower tumour growth fraction, assessed by lower Ki67 proliferation index in patients with activated ACC, a downstream target of AMPK, is consistent with in vitro evidence that AMPK activation leads to reduced proliferation [41] and growth inhibition in epithelial cells [34], also associated with decreased mTOR signalling [42, 43]. AMPK activation is known to lead to inhibition of the mTOR pathway [44, 45] through phosphorylation of TSC2 [32] and Raptor [33]. Activation of AMPK also directly limits translational initiation and protein synthesis through inhibition of translation elongation factor 2 (EF2) [46]. In this study we have focused on histological evidence for AMPK phosphorylation and its direct downstream target, ACC, as markers of AMPK activation, and have correlated them with the functionally important biological markers of breast cancer: ER, Ki67 and HER2.

In conclusion, the role of AMPK in the mechanism of action of metformin [47‐49], evidence in support of a putative relationship between type 2 diabetes and neoplasia with particular reference to breast cancer [50], and the reduced risk of cancer in patients with type 2 diabetes taking metformin compared with those on other medications such as sulphonylureas [35, 51], are all indicative of the need to explore this relationship and its potential to discover a novel targeted therapy in breast cancer. AMPK activation is a possible therapeutic target for cancers, since AMPK inhibits mTOR signalling, which in turn inhibit tumour growth in vitro and metastasis in experimental animal models [20]. The data presented here, the first immunohistochemical evidence of AMPK dysfunction in primary breast cancer supports further laboratory and clinical research targeting AMPK by metformin or its analogues in breast cancer.