Natural variations of copper and sulfur stable isotopes in blood of hepatocellular carcinoma patients

Significance

In cancer, the metabolism of copper and sulfur are dysregulated, leading to deleterious side effects. These issues are commonly addressed by studying the variations of concentrations of the elements, but here we have used, for the first time to our knowledge, copper and sulfur stable isotope compositions variations, using methods widespread in Earth sciences. We show that in hepatocellular carcinomas patients, blood copper and sulfur are enriched in light isotopes compared with control subjects. These isotopic signatures are not compatible with a dietary origin, but rather reflect the massive reallocation in the body of copper immobilized within cysteine-rich proteins such as metallothioneins. We also propose that sulfur isotope compositions could serve to track sulfur originating from tumor-derived sulfides.

Abstract

The widespread hypoxic conditions of the tumor microenvironment can impair the metabolism of bioessential elements such as copper and sulfur, notably by changing their redox state and, as a consequence, their ability to bind specific molecules. Because competing redox state is known to drive isotopic fractionation, we have used here the stable isotope compositions of copper (⁶⁵Cu/⁶³Cu) and sulfur (³⁴S/³²S) in the blood of patients with hepatocellular carcinoma (HCC) as a tool to explore the cancer-driven copper and sulfur imbalances. We report that copper is ⁶³Cu-enriched by ∼0.4‰ and sulfur is ³²S-enriched by ∼1.5‰ in the blood of patients compared with that of control subjects. As expected, HCC patients have more copper in red blood cells and serum compared with control subjects. However, the isotopic signature of this blood extra copper burden is not in favor of a dietary origin but rather suggests a reallocation in the body of copper bound to cysteine-rich proteins such as metallothioneins. The magnitude of the sulfur isotope effect is similar in red blood cells and serum of HCC patients, implying that sulfur fractionation is systemic. The ³²S-enrichment of sulfur in the blood of HCC patients is compatible with the notion that sulfur partly originates from tumor-derived sulfides. The measurement of natural variations of stable isotope compositions, using techniques developed in the field of Earth sciences, can provide new means to detect and quantify cancer metabolic changes and provide insights into underlying mechanisms.

Copper is an essential trace element (1), which has a pivotal role in the balance of oxidative stress: High levels are harmful because reduced copper promotes the generation of reactive oxygen species (2), and in the meantime, copper is involved in several enzymes (ceruloplasmin, hephaestin, Cu/Zn superoxide dismutase) that prevent the generation of reactive oxygen species (e.g., ref. 3). Three major types of ligand are associated with copper binding. Copper binds to nitrogen in histidine, and to sulfur in cysteine and methionine. The ratio of naturally occurring stable isotopes of copper, ⁶⁵Cu/⁶³Cu, varies according to the nature of the donor ligand. Light isotopes favor soft bonds relative to hard bonds. Bonds with Cu⁺ are softer than those with Cu²⁺, while bonds with sulfur are softer than those with nitrogen. These fairly general principles have been confirmed by ab initio calculations and density functional theory, and Cu isotope fractionations among various organic and inorganic compounds are now available (4). Heavy ⁶⁵Cu is preferentially oxidized compared with ⁶³Cu and prefers nitrogen donor ligands, such as histidine, or oxygen donor ligands, such as glutamate or aspartate (4). Sulfur possesses four stable isotopes, among which the ³⁴S/³²S ratio is the easiest to measure because ³²S and ³⁴S represent >99% of total sulfur (Method). The magnitude of variation of the ³⁴S/³²S ratio can reach one tenth permil during the incomplete reduction of sulfate and sulfide (5). Regardless of the element, differences in coordination and redox states are generally associated with isotopic effects known as isotopic fractionation. In animal models, the normalized ⁶⁵Cu/⁶³Cu ratio, expressed in delta units (Method), varies between organs from −1‰ in liver to +1‰ in kidney (6) (for comparison, this ratio ranges from −3‰ to +2.5‰ in terrestrial materials). In humans, the ⁶⁵Cu/⁶³Cu ratio in blood and bone differs between men and women (7, 8), partly because a sizeable proportion of the women's blood copper comes from the liver to balance menstrual losses (9, 10). The body ⁶⁵Cu/⁶³Cu ratio varies according to that of diet (11) and also seems to be age dependent (12). In life sciences, sulfur isotopic ratios are mainly used to trace animals’ diet in wildlife (13) and in stockbreeding (14) contexts. To this date, one study reports on bodily sulfur isotopic systematics, which shows no significant variations relative to diet in a murine model (15).

The natural variations of the isotopic ratios of an element offers a new means to study the imbalances linked to pathological conditions (16–18). Here, we used the natural variations of the ⁶⁵Cu/⁶³Cu and ³⁴S/³²S ratios, expressed as δ⁶⁵Cu and δ³⁴S values (Method), to track the copper and sulfur imbalance in hepatocellular carcinoma (HCC). Liver cancer was chosen because the liver is the main copper reservoir in the body and is known to be a pivotal organ for the metabolism of sulfur amino acids, and because the metabolism of copper and sulfur has been shown to be disrupted in cancer (e.g., refs. 19 and 20). To this end, we measured the δ⁶⁵Cu and δ³⁴S values in the serum and red blood cells of a series of Thai male HCC patients (n = 23, Material) and control subjects (n = 20, Material). We also analyzed biopsies of liver tumor and surrounding unaffected tissues from seven Caucasian patients (Material).

Results

Correlation of Copper and Sulfur Isotopes with Biological Parameters.

Several parameters were measured in the framework of the International Liver Cancer Study (ILCS) initiative in Thailand, including alkaline phosphatase (ALP), aspartate aminotransferase (AST), and alanine aminotransferase (ALT), which are classical tests for liver function and latent-transforming growth factor β binding-protein 2 (LTBP2), a new candidate biomarker of HCC (21). In general, tests for liver function fail to discriminate HCC patients from controls [analysis of variance (ANOVA); ALP, P = 0.456, AST, P = 0.172, ALT, P* = 0.010], while LTBP2 values do (ANOVA; P*** < 10⁻⁴). In serum, copper and sulfur isotope compositions are not correlated to any of the tests for liver function (copper; ALP, R = 0.350, P* = 0.027, AST, R = 0.231; P = 0.151, ALT, R = 0.144, P = 0.379; sulfur; ALP, R = 0.219, P = 0.271, AST, R = 0.245; P = 0.218, ALT, R = 0.186, P = 0.353). Serum copper isotope compositions are not correlated to LTBP2 values (R = −0.093, P = 0.569). However, serum sulfur isotope compositions are correlated to LTBP2 values (R = −0.509, P** = 0.006), but this mostly results from a group effect between HCC patients and controls.

In serum, copper and sulfur isotope compositions are not correlated to any anthropometrical parameter i.e., age, height, and body mass index (Tables S1 and S2). This holds for red blood cells (RBC) too (Tables S2 and S3), except that an inverse correlation (R = −0.490, P* = 0.028) is observed between the age of control subjects and the copper isotope composition of RBC (Fig. S1), as previously observed in a remote Yakut population (12).

Copper and Sulfur Isotopes in Blood Fractions of HCC Patients and Control Subjects.

Copper concentrations and isotope compositions were measured along with ceruloplasmin (Cp) concentrations and Cu/Zn superoxide dismutase (SOD1) activity in the blood fractions of HCC patients and control subjects. Results are given in Table S1 for serum and in Table S3 for RBC. As expected from a wealth of data (e.g., ref. 22), copper concentrations were higher in the serum and in RBC of HCC patients compared with controls (serum, P < 10⁻⁴, Fig. 1A; RBC P = 0.006, Fig. 1B). In addition, we observed a small increase of Cp (P = 0.012, Fig. 1C) and a decrease in SOD1 activity (P = 0.015, Fig. 1D), two effects previously reported in a variety of cancers (23, 24).

Fig. 1.

Chemical and biochemical compositions of serum and RBC of HCC patients and controls. (A) Copper concentrations in serum. (B) Copper concentrations in RBC. (C) Cp concentrations in serum. (D) SOD1 concentrations in serum. (E) Copper isotope composition of serum. (F) Copper isotope composition of RBC. (G) Sulfur isotope composition of serum. (H) Sulfur isotope composition of RBC. For all panels, *P = 0.01–0.05, **P = 0.001–0.01, and ***P < 0.001.

The copper isotope composition of blood fractions showed a tendency toward enrichment in ⁶³Cu in HCC patients compared with control subjects (Fig. 1 E and F). This effect is significant in RBC only (RBC, P < 10⁻⁴, serum, P = 0.134). However, there is a generally good correlation between serum and RBC copper isotope composition in HCC patients (P < 10⁻⁴, R = 0.579), suggesting that the copper isotopic imbalance is reflected in both fractions (Fig. 2A). We hypothesize that the lack of significant isotopic difference in the serum of HCC patients is due to the short half-life of serum copper, which is determined by that of Cp ($t_{1/2}^{Cp}$ ≈ 5 d, ref. 25), whereas in RBC, the half-life of copper is about one order of magnitude higher ($t_{1/2}^{RBC}$≈ 60 d).

Fig. 2.

Copper and sulfur isotope systematics in blood. (A) Copper isotope compositions in serum as a function of that of RBC. The black line stands for the least square correlation for all samples (y = 0.504(±0.115)x − 0.316(±0.088), R² = 0.335, P < 10⁻⁴). (B) Sulfur isotope compositions in serum as a function of that of RBC. The black line stands for the least square correlation for all samples (y = 1.051(±0.084)x − 0.161(±0.406), R² = 0.863, P*** < 10⁻⁴). (C) Sulfur isotope compositions as a function of that of copper in RBC. (D) Sulfur isotope compositions as a function of that of copper in serum.

Measurements of sulfur isotope compositions indicate that sulfur is significantly ³⁴S-depleted in both serum and RBC in HCC patients compared with control subjects (serum, P*** < 10⁻⁴, Fig. 1G; RBC, P** = 0.004, Fig. 1H), and that the sulfur isotope compositions in serum are correlated to those in RBC (P*** < 10⁻⁴, R = 0.929, Fig. 2B). HCC patients were characterized by lower δ³⁴S and δ⁶⁵Cu values compared with controls in RBC [multivariate analysis of variance (MANOVA); Wilks' λ = 0.47, F(2,40) = 22.52, P*** < 10⁻⁴, Fig. 2C] and in serum [MANOVA; Wilks' λ = 0.59, F(2,24) = 8.23, P** = 0.002, Fig. 2D].

Discussion

Otto Warburg and Adolf Krebs found in 1928 (26) that serum copper levels increased in various chronic diseases, including several types of cancers, resulting into a systemic and oncogenic copper accumulation (19, 27, 28). The present isotopic copper results suggest that this extra copper burden is unlikely of exogenous (dietary) origin. Enhanced exogenous copper uptake would be expected to attenuate the isotopic fractionation between dietary sources and blood, leading the δ⁶⁵Cu value of blood fractions to tend toward that of the diet. Such a mechanism has been proposed to explain the iron isotope composition of blood in patients with hemochromatosis (16). Assuming that the composition of a typical human diet may give typical δ⁶⁵Cu values of about +0.4‰ (6), enhanced dietary copper uptake would result in a ⁶⁵Cu enrichment in the blood of HCC patients, which is not compatible with our observations. If not exogenous, the extra copper burden may be caused by (i) reduced copper losses through bile and (ii) release of copper from endogenous stores.

Bile is the major normal pathway of copper excretion. The impairment of hepatic cells in HCC can lead to a reduced bile production, and consequently to an increase of systemic copper. Testing this hypothesis requires comparison of bile production and chemical composition for HCC patients and healthy controls. This approach can be easily achieved in further experiments by using murine models. Such experiments should focus on HCC models but also on other cancer models where there is, in principle, no relationship with liver function impairment.

The release of copper from endogenous stores could be directly demonstrated by analyzing the copper concentration and isotope composition for various organs in cancer mice and comparing the results with control mice. Here, we use an indirect approach, which consists to predict the isotope composition of copper released from normal cells. In cells, copper is laden mainly to three metalloproteins, i.e., SOD1 and metallothionein (MT) in the cytoplasm and cytochrom c oxidase (CcO) in mitochondria (29–31). As heavier isotopes bind preferentially with ligands with stronger electronegativity (oxygen > nitrogen > sulfur, ref. 4), and taking the binding characteristics of the main three copper-laden proteins into account, one can predict that δ⁶⁵Cu_SOD1 > δ⁶⁵Cu_CcO >> δ⁶⁵Cu_MT. The low δ⁶⁵Cu value of the extra copper burden may thus be explained by the release of intracellular copper from cysteine clusters, with MT being the most likely source. Therefore, the links between copper imbalance in cancer and MT should be the focus of further attention.

A third mechanism, which is not exclusive from the previous ones, can explain the low δ⁶⁵Cu value of the blood extra copper burden. It involves mass conservation of copper isotopes between ⁶⁵Cu-depleted blood and ⁶⁵Cu-enriched tumors. In a series of liver biopsies of HCC patients sampled along with adjacent normal liver tissues, we found that tumors are systematically ⁶⁵Cu-enriched relative to normal tissues (Table S4 and Fig. S2). The accumulation of ⁶⁵Cu in tumors can therefore enhance the existing ⁶⁵Cu depletion observed in blood originally triggered by the release of intracellular copper from MT. This hypothesis implies that the blood δ⁶⁵Cu value would decrease as a function of the severity of the cancer, which would be of interest for the estimation of tumor burden.

The observation of lower δ³⁴S values in the serum and the RBC of HCC patients is unexpected because the two blood fractions have different proteomes, thereby suggesting that a common mechanism might be at work. The hypothesis of enhanced formation of sulfides, in which the δ³⁴S value is 20‰ less than in coexisting sulfate (32), may be involved. Tumor-derived hydrogen sulfide (33) and allosteric formation disulfide bonds in cancer-related proteins (20) are possible candidates, but it is still unclear how prevalent is the formation of these compounds. Considering that the sulfur content of proteins is stoechiometric and therefore constant, the blood δ³⁴S value of HCC patients (δ³⁴S_HCC = 4.0‰) can be considered as a mixture of the normal sulfur isotope composition (δ³⁴S_N = 5.4‰) and of total sulfides (δ³⁴S_TS), with proportion x and 1 − x, respectively:

$${\mathrm{\delta }}^{34}{\mathrm{S}}_{\mathrm{HCC}}=x\ast {\mathrm{\delta }}^{34}{\mathrm{S}}_{\mathrm{N}}+\left(1-x\right)\ast {\mathrm{\delta }}^{34}{\mathrm{S}}_{\mathrm{TS}}.$$

Such a simple mass balance indicates that about 6% of total sulfur in blood of HCC patients is coming from tumor produced sulfide with a δ³⁴S_TS of −20‰. This number decreases to 4% with a δ³⁴S_TS of −30‰. These numbers are very high and require further investigations, notably the direct determination of the δ³⁴S value of tumor produced H₂S. However, they already suggest that blood sulfur isotopic variations of HCC patients is a sizeable process that may provide new biomarkers for cancer detection and monitoring. However, the sulfur isotope compositions need to be measured on other types of cancer to know whether the origin of the low blood δ³⁴S values in HCC patients are linked to disorders of hepatic origin or sulfides production.

Material

Study Participants.

Blood samples from patients and controls from Thailand were collected in the framework of a hospital-based case−control study conducted at the Cancer Control Unit of the National Cancer Institute of Thailand (Bangkok) from April 2008 to December 2009. All cases of primary liver cancer were recruited, and matched controls were obtained from outpatient clinics. Differential diagnosis of HCC versus cholangiocarcinoma was established by a combination of clinical examination, imaging using ultrasonography, computerized tomography or Magnetic Resonance Imaging, biochemistry (alpha-fetoprotein and liver function enzymes testing) and histological confirmation on a small subset of patients from whom needle biopsies were available. Individuals from the reference group presented no clinical evidence of liver disease. All study participants provided informed consent, and both Thailand and international institutional review boards approved the study protocol (21, 34).

Liver Biopsies Collection.

Liver tumor and nontumor samples were collected in spring 2010 at the Paul Brousse Hospital, Centre Hépato-Biliaire, Villejuif, France, on seven patients further diagnosed for hepatocarcinoma. Copper concentrations and isotope compositions of tumor and nontumor samples are given in Table S1.

Method

Sample Preparation for Copper Isotope Analysis.

RBC and serum samples were digested by a mixture of concentrated subboiled distilled HNO₃ and 30% H₂O₂. Further information concerning copper separation from the matrix is given in SI Text.

Copper Isotope Analysis.

Copper isotope compositions were measured at the Laboratoire de Géologie de Lyon (LGLTPE). The ion exchange chromatography and mass spectrometry techniques have been described extensively elsewhere (35, 36), and are described in SI Text. The delta value is given by δ⁶⁵Cu = [(⁶⁵Cu/⁶³Cu)_sample/(⁶⁵Cu/⁶³Cu)_standard − 1] × 10³. The NIST-SRM 976 solution was used as the copper isotopic standard. Copper isotope compositions are given in Tables S1 and S3.

Copper Concentration Analysis.

Copper concentrations were measured by quadrupole inductively coupled plasma mass spectrometry (Q-ICPMS) at LGLTPE using a 7500 CX quadrupole mass spectrometer (Agilent Technologies). The following operating parameters of Q-ICPMS were optimized: rf power (1,550 W), the plasma gas flow rate (15 L⋅min⁻¹), the auxiliary gas flow rate (2 L⋅min ⁻¹), the carrier gas flow rate (0.9 L⋅min⁻¹), and the makeup gas flow rate (0.15 L⋅min⁻¹). The performance of the Q-ICPMS instrument was optimized and checked daily by testing a standard mixture containing 1 ng⋅mL⁻¹ Li, Y, Co, Ce, and Tl in 3% HNO₃. Indium at 1 ng⋅mL⁻¹ was used as an internal standard to correct for any long-term instrumental drift. Copper concentrations are given in Tables S1 and S3.

Sulfur Isotope Analysis.

Sulfur isotope analysis was undertaken by Elemental Analysis Isotope Ratio Mass Spectrometry at IsoAnalytical and duplicated for some samples at LGLTPE. Further details on the analytical procedures are given in SI Text. In both laboratories, international standards NBS-127 (barium sulfate, δ³⁴S_CDT = +20.3‰) and IAEA-SO5 (barium sulfate, δ³⁴S_V-CDT = +0.50‰) were measured as quality control checks during batch analysis of the samples (Table S5 and Fig. S3). Sulfur isotope compositions are given in Tables S1 and S3.