Introduction

Copper alloys and their antimicrobial properties have been shown to be effective in killing pathogenic and antibiotic-resistant bacteria (Wilks et al. 2005; Noyce et al. 2006). Studies demonstrated that pathogens are easily spread throughout the hospital environment on the gloves of healthcare workers and that thorough disinfection of surfaces decreases the likelihood for transmission or acquisition of infection (Boyce 2007). Current hospital trials have shown that copper alloys may improve the ability to effectively disinfect surfaces (Casey et al. 2010; Mikolay et al. 2010). However, the exact mechanism through which copper is toxic to the cellular activity of bacteria remains under investigation, in particular the mechanism of killing on metallic copper surfaces. Copper in the ionic form, while toxic in high concentrations, is usually effectively managed in Escherichia coli cells by copper homeostasis systems, such as the P-type ATPase CopA, the CusCFBA system, and the multi-copper oxidase CueO (Rensing and Grass 2003; Baker-Austin et al. 2006). E. coli strains may also contain plasmid-encoded determinants for additional copper resistance that further enhance their ability to withstand toxic copper concentrations (Lee et al. 2002). Enterococcus faecium also maintains copper homeostasis through membrane transporter proteins of the P-type ATPase family, which are encoded by the chromosomal operon copYZAB which is similar to copYZAB described in Enterococcus hirae (Solioz and Vulpe 1996; Wunderli-Ye and Solioz 1999). More recently, transferable copper resistance genes (tcrB genes) have been identified as being part of the plasmid-borne tcrYAZB operon in E. faecium which is similar to the copYZAB operon and possibly mediates co-selection for resistance to macrolides and glycopeptides (Hasman and Aarestrup 2002; Hasman et al. 2006). Isolates from animal farms containing the plasmid with tcrB showed resistance to copper and various antibiotics as well as resistance to compounds used for disinfection (Hasman and Aarestrup 2002; Aarestrup and Hasman 2004). Some of these genes involved in copper resistance and homeostasis also play a role in survival on copper alloy surfaces under specific conditions. It has been shown that disruption of the regulator in the cin operon in Pseudomonas aeruginosa results in faster killing rates on copper alloys when compared to the wild type (Elguindi et al. 2009). E. hirae ∆copB and ∆copAB deletion strains suspended in phosphate buffer were killed more rapidly on copper surfaces than was the wild type, and observed differences in killing rates could be linked to copper ion concentrations in different media used for the incubation on copper alloys (Molteni et al. 2010). Recent studies on copper alloys eliminating moisture on the surfaces revealed very rapid killing of E. coli cells which was inhibited by reactive oxygen species (ROS) quenchers but enhanced by inactivated superoxide dismutase, suggesting an increase in ROS resulting from an increase in copper ion concentrations in the cell (Espirito Santo et al. 2008).

In this study, the survival rates of copper-resistant E. coli and E. faecium strains were tested on different copper alloys while altering moisture contents during inoculation following EPA-recommended testing guidelines. The differences observed under different testing conditions may be related to moisture levels affecting copper surface corrosion and enhanced copper tolerance mechanisms of the strains. The measurement of corrosion rates presents a quantitative analysis of the level of copper ions released from the surface over time and determines how the flux of copper ions relates to bacterial survival. Surface treatments of copper alloys can influence corrosion rates and, subsequently, survival times. The commonly used organic corrosion inhibitor benzotriazole (BTA) is effective in binding copper ions in a single layer on the metal surface, thereby markedly reducing corrosion (Antonijevic and Petrovic 2008). A thermal oxide layer on a copper-electroplated surface provides a barrier to free copper ions as well and alters corrosion rates. In minimal medium such as 0.8% NaCl, bacteria would respond directly to the concentration of bioavailable copper ions since no sequestration of copper occurs due to other organic molecules in the medium. Copper-resistant E. coli and E. faecium cells inoculated in 0.8% NaCl on copper-electroplated surfaces showed significant differences in survival rates correlating with corrosion rates measured. The results presented indicate that copper resistance in bacteria does not play a significant role in survival on copper surfaces under the experimental conditions employed in this study. Some conditions only show a significant reduction of viable counts and not a complete die-off after 24 h of exposure. This is significant regarding bacteria in the healthcare environment since they frequently acquire resistance to antibiotics and disinfectants, and total elimination of bacteria from fomites remains a difficult task in the food and healthcare industries.

Materials and methods

Bacterial strains and growth conditions

The E. coli isolates (77-30009-6, 77-30013-2, 77-30253-2) and E. faecium isolates (75-30704-5, 75-30733-4, 75-30518-6) were collected as part of the DANMAP surveillance program (DANMAP 2004, 2006) from healthy animals at or just prior to slaughter in 2003 (E. faecium) and 2005 (E. coli). These strains can be obtained through DTU, the National Food Institute, or by contacting Henrik Hasman (hhas@food.dtu.dk). All isolates had been tested previously for susceptibility to the following antimicrobial agents using a commercially prepared dehydrated panel (SensiTitre)—avilamycin, bacitracin, chloramphenicol, erythromycin, gentamicin, kanamycin, penicillin, quinupristin/dalfopristin, streptomycin, tetracycline, vancomycin, and virginiamycin—according to CLSI standard and evaluated using CLSI breakpoints (Clinical and Laboratory Standards Institute 2008).

Cultures of E. coli copper ion-resistant strains were prepared in Luria–Bertani (LB) broth at 37°C, while E. faecium copper-resistant strains were grown in Lee’s liquid medium (10 g tryptone, 10 g yeast extract, 5 g lactose, 5 g sucrose, 0.5 g dipotassium phosphate in 1−1 L). Following overnight incubation, 100 μl of the cultures was placed in 3 ml of the respective liquid media and grown to mid-log phase at OD600 0.5–0.7. A 25 μl inoculum of these cultures was pipetted on each coupon to be tested and spread with a sterile glass rod. The samples were spread over the entire surfaces of the copper alloys and kept moist in a closed container (Wilks et al. 2005). In order to simulate drop size contamination, the inoculum was placed as a droplet on the coupons and kept moist in a closed container. Additional experiments involved the methods described in EPA-recommended protocols obtained from the Copper Development Association requiring the initial drying of the inocula for 30 min on copper alloys in open air followed by a timed survival assay (Nada 2005a) and a 24-h repeat inoculation with intermittent sampling. For 24-h repeat inoculation experiments, a 10 μl inoculum was spread over copper alloy coupons every 3 h and left drying in open air between applications (Nada 2005b). In order to compare survival rates to measured corrosion rates, the bacteria were grown to mid-log phase in culture medium, pelleted at rcf 18,000×g for 2 min, and pellets resuspended in 0.8% NaCl. After defined incubation times at room temperature (25°C) specific for each of the experiments, the coupons were transferred into 50-ml centrifugation tubes which contained 10 ml sterilized PBS and 20–25 sterilized 2-mm diameter glass beads. After thorough mixing on a vortex mixer for 30 to 60 s, serial dilutions were plated out on LB agar or Lee’s agar plates and incubated at 37°C for 24 h. The experiments were performed in triplicates and the mean and standard deviation calculated. For experimental controls, E. coli and E. faecium copper ion-resistant strains were placed on stainless steel coupons.

MIC determination of copper sulfate by the agar dilution method

Copper ion resistance was evaluated on brain heart infusion agar plates containing 0, 1, 2, 4, 8, 12, 16, 20, 24, and 28 mM copper sulfate (CuSO4⋅5H2O) adjusted to pH 7. Isolates were inoculated with a single drop (approximately 5 × 105 cells) of cells picked from a blood agar plate and suspended in a 0.9% NaCl solution adjusted to McFarland = 0.5 according to CLSI standards. The plates were then incubated at 37°C for 48 h and the growth assessed after 20 and 48 h. The tcr-positive E. faecium A17sv1 (Hasman et al. 2006) and the pRJ1004-carrying and pco-positive E. coli ED8739 (courtesy of Dr. Jon Hobman, School of Biosciences, The University of Nottingham, Loughborough, UK) were used as positive controls. The plasmid-free and copper-susceptible E. faecium BM4105 (CuS04 MIC = 4 mM) and E. coli MT102 (CuSO4 MIC = 12 mM) were used as negative controls. In cases where the growth was difficult to evaluate, PCR was used to examine the presence of the tcr genes (E. faecium) and pco genes (E. coli) as described previously (Hasman et al. 2006). Copper-resistant E. faecium isolates had a CuSO4 MIC value of 24 mM, and copper-resistant E. coli isolates had a CuSO4 MIC of 20 mM.

Preparation of test surfaces

The different copper alloy coupons employed in this study were a gift from the International Copper Association, and their compositions are listed in Table 1. Prior to use, the 1-in.2 copper alloy and stainless steel coupons were immersed for 30 s in a 3% NaOH solution heated to 70°C, rinsed in distilled H2O, and followed by a 3- to 5-s immersion in 10% H2SO4 and rinsed with DH2O. To prevent reoxidation of the surfaces, the coupons were rinsed in 95% ethanol, air-dried, and kept in a sterile closed container until used (Espirito Santo et al. 2008). Copper films of thickness ~600 nm electroplated on silicon wafers were provided by Freescale Semiconductors Inc. CuOX films, 60 nm in thickness, were prepared by thermally oxidizing electroplated copper films at 300°C in a laboratory tube furnace in a high-purity air ambient for 5 min. X-ray photoelectron spectroscopy results showed that a thin layer of CuO was present at the surface, and as the layer was sputtered away, the film almost entirely consisted of Cu2O. Using an electrochemical reduction process, the oxide film was characterized to be roughly 95% Cu2O and 5% CuO. Copper–BTA film test surfaces were created by pretreating electroplated copper samples with an isopropanol and subsequent distilled H2O rinse, then immersed for 60 s in 0.1 M HCl and followed by a distilled H2O rinse before being air-dried. The samples were immersed 15 min in a BTA (99% pure, Sigma-Aldrich) and distilled H2O-based solution in concentrations of 0.016, 0.001, 0.0004, and 0.0001 M for experiments, rinsed with distilled H2O, and air-dried. Before testing, the copper-electroplated samples were rinsed in 95% ethanol and air-dried.

Table 1 Copper alloys and their constituent compositions in percent values
Full size table

Surface dissolution of copper and corrosion inhibition

Approximately 280 copper alloys containing more than 60% Cu have been registered with EPA as antimicrobial copper materials (EPA 2008). When exposed to aqueous solutions, depending on the pH and redox potential of the system, copper may dissolve as cupric ions or form oxides, Cu2O and/or CuO. A potential–pH diagram for Cu–water system at a dissolved copper concentration of 10−6 M was constructed using a commercially available software package, STABCAL (Huang, Montana Tech. 2009). This diagram, shown in Electronic supplementary materials (ESM) Fig. S1, shows that at acidic pH values and under oxidizing conditions, i.e., more positive potential values, the oxidation of Cu to Cu2+ ions is favored. In the near-neutral pH regions, the formation of cuprous and cupric oxide is thermodynamically favorable if the solution potential is above zero. As a reference, the solution potential of 1 M NaCl is roughly 0.25 V (with respect to the standard hydrogen electrode). Hence, in saline test solutions at near-neutral pH, the copper surface is likely to be oxidized to copper oxide. At high pH values, depending on copper ion activity, copper oxides may be dissolved to form HCuO −2 and CuO 22 (ESM Fig. S1).

BTA (C6H5N3) is a well-known corrosion inhibitor for copper in various applications in a wide range of environments (Cotton and Scholes 1967). The interaction of BTA with copper has been studied extensively. It is generally accepted that BTA (as well as BTA− ion) chemisorbs on the copper (as well as cuprous oxide) surface and forms an insoluble cuprous surface complex. Under certain conditions, the formation of a thick, multilayered coating has been confirmed. The mechanism for BTA’s unique film formed on copper is suggested to be a planar structure of stoichiometry 1:1 Cu/BTA (Cotton and Scholes 1967; ESM, Figs. 2.1 and 2.2). The potential–pH diagram for the Cu–BTA–water system, constructed using the STABCAL software, is shown in ESM Fig. S3 for a copper ion concentration of 10−6 M and a BTA concentration of 10−4 M (ESM Fig. S3) It may be noted from this figure that the formation of CuBTA (s) is thermodynamically favorable in the pH range of 3–10. Oxidation of CuBTA to Cu++ and CuO (s) is thermodynamically favorable under oxidizing (higher solution potential) conditions, but it is kinetically very slow.

Results

Moisture content influences survival rates of copper-resistant bacteria

With 25 μl of bacterial culture spread thinly over the surface, copper alloys were kept in a moist, closed compartment, and survival rates of E. coli isolates were assessed by countable colony forming units (CFUs) remaining. For E. coli 77-30009-6 on copper alloys containing >88% copper, no countable CFUs were seen after 15–30 min, while on 70% copper alloys countable CFUS dropped to zero in 60 min (Fig. 1a). This method was also most effective in killing E. faecium isolates. Countable CFUs for E. faecium 75-30704-5 dropped to zero after 30–60 min. The 70% copper and 30% zinc alloy was the least effective, with only a onefold log decrease in countable CFUs after 60 min (Fig. 1b). This method has been frequently used to evaluate the efficacy of metallic copper surfaces against a variety of bacteria, but would be difficult to attain in the environment since the inocula dry within 10–15 min without cover in dehumidified surroundings.

Fig. 1
figure 1

Survival rates of copper ion-resistant bacteria on copper alloys. Bacterial suspension of E. coli (a) and E. faecium (b) spread over copper alloys and kept in moist environment. E. coli (c) and E. faecium (d) applied as droplets on copper alloys and kept in moist environment. E. coli (e) and E. faecium (f) spread over copper alloys and dried 30 min before timing in dry environment. Shown are 99.9% copper (filled square), 94.8% copper (diamond), 88.6% copper (triangle), 70% copper (circle), and stainless steel control (empty square)

Full size image

Droplets of a bacterial suspension on copper alloys in a moist environment resulted in markedly prolonged survival rates of E. coli and E. faecium isolates. E. coli 77-30009-6 cells were killed within 120–360 min depending on the copper content of the alloy (Fig. 1c). E. faecium isolate 75-30704-5 showed a decrease of countable CFUs to zero after 360 min on 94.8% copper, but survival was still prolonged on the remaining alloys (Fig. 1d). However, E. faecium isolates 75-30733-4 and 75-30518 only showed a maximum of onefold decrease in countable CFUs after 480 min (data not shown). When bacteria were inoculated with increased moisture levels and a smaller area of exposure on the metal the kinetics of copper ions in rich medium were altered, significant differences in survival rates between individual strains were seen, signifying the strains’ distinctive abilities to withstand copper ion-dependent toxicity.

The method of spreading the inocula of E. coli 77-30009-6 over the copper surfaces and letting it air dry for 30 min prior to sampling times left zero countable CFUs on all copper alloys (Fig. 1e). Countable CFUs on stainless steel controls were decreased by 0.5–1 log fold after 60 min. With this method, a prolonged drying period was achieved, and recovery of all E. coli copper ion-resistant strains from all copper alloys tested showed no remaining countable CFUs, except on stainless steel (data not shown). E. faecium survival rates utilizing this testing method revealed that countable CFUs for 75-30704-4 were decreased by 3 log fold after 60 min. Differences between the alloys were not well defined under drying conditions. There was no decrease in CFU count on stainless steel, which indicated that the drying process itself did not have a negative effect on E. faecium (Fig. 1f).

A suspension of E. coli spread over copper surfaces and air-dried between eight successive inoculations revealed no countable CFUs on all alloys tested over a 1- to 24-h period. However, on stainless steel controls, countable CFUs increased over the 24-h period (Fig. 2a). E. faecium incubation on four different copper alloys resulted in an overall decrease of countable CFUs ranging from 2 to 3 log fold over a 24-h period, while stainless steel surfaces showed an increase in countable CFUs (Fig. 2b).

Fig. 2
figure 2

Survival of copper ion-resistant bacteria on copper alloys with multiple inoculations. E. coli (a) and E. faecium (b) suspensions were reapplied on copper alloys in 3-h intervals without removal for a 24-h period. Each 10-μl sample was spread over the surface and air-dried between applications. Shown are 99.9% copper (filled square), 94.8% copper (diamond), 88.6% copper (triangle), 70% copper (circle), and stainless steel control (empty square)

Full size image

Accelerated killing of E. faecium 75-30704-5 was observed when suspended in 0.8% NaCl and placed as a drop on 99.9% copper alloys. No countable CFUs remained after 30 min of exposure in 0.8% NaCl (Fig. 3b). Therefore, a quantitative measurement of surface corrosion of copper materials was conducted in order to provide an estimate of the rate by which dissolved copper is released from the surface and how bacterial survival is influenced by that rate.

Fig. 3
figure 3

a Survival of copper ion-resistant E. coli 77-30009-6 influenced by corrosion of copper surfaces. Solid metallic copper (filled square), electroplated copper (diamond), electroplated copper with thermal oxide coating (circle), electroplated copper with BTA 0.016 M (triangle), and stainless steel control (empty square). b Survival of copper ion-resistant E. faecium 75-30704-5 influenced by corrosion of copper surfaces. Solid metallic copper (filled square), electroplated copper (diamond), electroplated copper with thermal oxide coating (circle), electroplated copper with BTA 0.016 M (triangle), and stainless steel control (empty square)

Full size image

Corrosion rates influence survival of copper-resistant bacteria on copper

Electrochemical measurements and calculation of corrosion rates of electroplated copper samples revealed that the corrosion inhibitor BTA at a concentration of 16 mM decreased the corrosion rate of electroplated surfaces by ~65%. In contrast, the 60-nm thermal oxide layer on electroplated surfaces decreased the corrosion rate by only 30% (Table 2). Bacterial survival rates on these surfaces also showed marked differences. E. coli 77-30009-6 cells suspended in 0.8% NaCl and placed as a drop on electroplated copper surfaces were killed within 60 min of exposure, but showed no decrease in countable CFUs after 60 min of exposure on BTA-coated electroplated copper (Fig. 3a). Thermal oxide coating also prolonged survival and showed a 3 log fold decrease of countable CFUs after 60 min. Similarly, E. faecium cells in 0.8% NaCl were killed within 45 min on electroplated copper surfaces with and without thermal oxide coating, and countable CFUs were decreased 3 log fold after 60 min on BTA-coated surfaces (Fig. 3b). However, bulk 99.9% copper surfaces were most effective in killing both E. coli and E. faecium strains suspended in 0.8% NaCl (Fig. 3a, b). E. coli cells were inoculated on electroplated copper surfaces bearing BTA concentrations ranging from 0.1 to 16 mM. After 30 min of exposure, no die-off was seen at concentrations of ≥1 mM (Fig. 4). Exposure of copper surface to 0.016 M BTA solution resulted in a decreased level of copper release and rate of bacterial kill. Survival rates of E. coli were dependent on the BTA concentration in the solution in which copper surfaces were treated. (Table 2 and ESM Fig. S4). Copper ion concentrations on dry surfaces were not determined.

Table 2 Electrochemically determined copper ion release in 1 min from untreated and treated electroplated copper surfaces in 0.8% NaCl electrolyte
Full size table
Fig. 4
figure 4

Survival of E. coli 77-30009-6 at 30 min on BTA-coated electroplated copper surfaces

Full size image

Images of E. coli cells were obtained utilizing an environmental scanning electron microscope and revealed no grossly damaged cell membranes after placing the cells on dry electroplated copper surfaces from 30 to 60 min. The most notable difference was the cell shape which appeared rounded with live cells on stainless steel vs. the characteristic oblong shape of dead cells on copper (ESM Figs. S5.1 and S5.2).

Discussion

The findings of this study support the mounting evidence that the release and influx of bioavailable copper ions from copper surfaces into bacterial cells is the driving force in the killing mechanism on metallic copper. Corrosion inhibitors such as BTA altered the concentration of copper ions (Cu+/Cu2+) released from copper surfaces, which directly affected bacterial survival in 0.8% NaCl medium. However, the exact concentration of copper needed to kill a particular microorganism would depend on other factors, such as moisture content, copper content of alloy, medium present, copper homeostasis mechanisms, and the membrane structure of, e.g., Gram-negative or Gram-positive organisms. Altered moisture levels on copper surfaces resulted in marked differences in survival rates related to the copper content of the alloys. The observed faster killing of E. coli copper-resistant strains in dry environments suggests different interactions of Gram-negative bacteria with copper ions released from the metallic surfaces than those of Gram-positive bacteria. It has been reported that Gram-positives survive longer on dry copper surfaces (Espirito Santo et al. 2010), shown also with E. faecium in this study (Figs. 1f and 2b). However, in 0.8% NaCl medium, the cells died off quickly, correlating with different corrosion rates (Fig. 3b). The higher copper tolerance in a dry environment may be due to copper ions bound to components of rich medium, and exchange with the thick peptidoglycan layer could form a buffer to copper ion influx. Neutral copper complexes such as Tris2–Cu or Cl2–Cu may not exchange copper ions with other molecules and, as suggested by Molteni et al. (2010), may actually facilitate cross-membrane transport of copper ions. Gram-negative E. coli may have an advantage when exposed to slow corrosion rates due to their periplasmic copper export systems, which Gram-positives lack. At low corrosion rates through BTA coating, i.e., 4.9 μM in 1 min, E. coli survival was not reduced after 60 min of exposure, suggesting that existing copper homeostasis mechanisms were adequate at this rate. A 100% increase in corrosion rates, i.e., 10.2 μM in 1 min through thermal oxide coating, reduced survival rates 3 log fold after 60 min. However, these mechanisms offered no protection under dry conditions and most likely allowed a very rapid and massive influx of copper ions into the cells, resulting in an almost immediate cell death. Copper ion concentrations on a dry surface could not be determined with the method used in this study. It is presumed that dry copper also ionizes at the surface, and when making contact, cells can rapidly take up copper ions but cannot effectively transport them out of the cell possibly due to the lack of extracellular fluid. The rapid accumulation of copper in the cytoplasm was recently demonstrated with E. coli during a 1-min contact on dry copper surfaces, suggesting a compromised cell membrane as the underlying mechanism (Espirito Santo et al., personal communication). Bacterial contaminations in the food and healthcare environment can vary greatly in concentrations, exposure times, media, and moisture levels. The bacterial strains in this study were rapidly killed on moist copper surfaces regardless of their protective copper homeostasis mechanisms. Therefore, it is imperative for copper surfaces in the food and healthcare environments to maintain higher corrosion rates without the application of corrosion inhibitors in order to achieve maximum bactericidal effects. Moreover, oxidized copper surfaces maintain sufficient corrosion rates and are effective for contact killing, but may appear unclean when compared to stainless steel surfaces. It has also been shown that Tris–Cl medium may contribute to increased copper corrosion and, consequently, to more effective killing of E. hirae on copper surfaces (Molteni et al. 2010). Therefore, copper alloys which have more durability and anti-corrosion properties may not perform well as bactericidal surfaces, as suggested previously (Wilks et al. 2005). More studies will be needed to estimate corrosion rates in order to identify which copper alloys could provide the highest corrosion rates under most environmental conditions, ensuring the maximum effectiveness of the anti-microbial copper materials.