OUP user menu


DOI: http://dx.doi.org/10.1093/alcalc/agl021 300-305 First published online: 31 March 2006


Aims: Serum protein profiles were examined in naïve, ethanol self-administering and ethanol abstinent cynomolgus monkeys (Macaca fasicularis) to search for differences in protein expression which could possibly serve as biomarkers of heavy ethanol consumption. Methods: Surface-enhanced laser desorption ionization time-of-flight (SELDI-ToF) mass spectrometry was used for proteomic profiling of serum. Results: Two proteins were identified by SELDI-ToF to be increased in ethanol self-administering compared with abstinent animals. These proteins were identified to be apolipoprotein AI (Apo-AI) and apolipoprotein AII (Apo-AII) by peptide mass fingerprinting and comparison with spectra of purified human Apo-AI and AII proteins. Immunoblot analysis of Apo-AI and Apo-AII was performed on a separate group of animals (within-animal ethanol-naïve and self-administering) and confirmed a statistically significant increase in Apo-AII, while Apo-AI was unchanged. Conclusions: An open proteomic screen of serum and confirmation in a separate set of animals found Apo-AII to be increased in the serum of ethanol self-administering monkeys. These results are consistent with previous clinical studies of human ethanol consumption and serum apolipoprotein expression. Moreover, these results validate the use of non-human primates as a model organism for proteomic analysis of ethanol self-administration biomarkers.

(Received 30 September 2005; first review notified 14 December 2005; accepted in revised form 1 March 2006)


Biomarker tests for monitoring ethanol consumption have many potential applications in health care, public safety, and alcoholism treatment. Currently, much of the data on drinking relies on self-reporting of alcohol intake. While useful (Del Boca and Darkes, 2003), self-reporting is known to be inconsistent (Phillips, 1984; Toneatto et al., 1992; Lapham et al., 2004) and in many cases there can be a motivation for patients to deny drinking. This has led to the search for biochemical markers of alcohol abuse (Rosman and Lieber, 1994; Sillanaukee, 1996; Lakshman et al., 2001). Several biochemical markers of alcoholism have been proposed (Haber et al., 2002; Helander, 2003; Javors and Johnson, 2003), but there are limitations to the specificity and sensitivity of these assays that have restricted their use clinically (Anton et al., 2002; Conigrave et al., 2002; Alte et al., 2003; Montalto and Bean, 2003). One way to avoid the limitations of existing tests has been to combine multiple tests to screen and confirm heavy drinking (Harasymiw and Bean, 2001; Javors and Johnson, 2003). However, serum biochemical markers with better specificity and sensitivity would be useful clinical tools in alcoholism treatment.

Developments in proteomic technology have made the screening of the plasma proteome for biomarker(s) of ethanol consumption possible (Kasinathan et al., 2004) and could potentially serve as a method to find patterns of serum protein expression that are indicative of heavy drinking suitable for use as a clinical diagnostic. Traditionally, proteomic screening has been accomplished through two-dimensional polyacrylamide gel electrophoresis (2-D PAGE) resolution and visualization of protein species (Robinson et al., 1995). New mass spectrometry-based approaches such as surface-enhanced laser desorption ionization time-of-flight mass spectrometry (SELDI-ToF MS) coupled to selective protein capture surfaces has proved useful as a biomarker discovery approach (Fung and Enderwick, 2002). This methodology has been used to identify a number of biomarkers for prostate cancer (Adam et al., 2001), ovarian cancer (Zhang et al., 2004), Alzheimer's disease (Davies et al., 1999), and HIV infection (Zhang et al., 2002). Previously, SELDI-ToF has been used to examine the serum profile of hospitalized alcoholics (Nomura et al., 2004).

Biomarker studies in humans have notable limitations such as comorbid substance abuse, variability in diet, and limited drinking history information. Non-human primates are an alternative population from which to develop biomarkers, in a much more controlled setting (Grant and Bennett, 2003). In this study, we used SELDI-ToF MS to profile serum proteins in non-human primates that had voluntarily self-administered large quantities of pure ethanol in an effort to find biomarkers of ethanol consumption.



Adult cynomolgus monkeys (Macaca fasicularis) were housed individually within 76 × 60 × 70 cm3 stainless steel cages in a vivarium maintained at 21 ± 1°C, with 30–50% humidity and a 12:12 h light:dark cycle. Monkeys were maintained in a positive caloric and fluid balance throughout the experiments. This was demonstrated by consistent increases in body weight and food (which was periodically increased) as well as fluid remaining at the end of the majority of sessions. These studies were conducted in accordance with the Wake Forest University Animal Care and Use Committee and the guidelines for the care and use of laboratory animal resources.

Monkeys were trained to self-administer ethanol in a 3 month behavioural induction as previously described (Vivian et al., 2001; Floyd et al., 2004). For the initial SELDI-ToF screening of drinking and abstinent animals, serum was collected from male and female monkeys after 6 months of ethanol self-administration (drinking) and after 6 months abstinence from ethanol preceded by 6 months of drinking (abstinent). There was no significant difference in cumulative intake between the two groups. These subjects were used in the initial testing because a useful diagnostic should be able to discriminate between current drinking and a past history of drinking followed by extended abstinence. To confirm the SELDI-ToF findings a second set of animals was used for immunoblot analysis. This set of animals was composed of within-subject samples from 10 males [ethanol naïve (naïve) and after 5 months of self-administration (drinking)]. These were collected to confirm changes within animals over time and limited to males to avoid sex-based differences. Individual animal data for both groups is given in Table 1.

View this table:
Table 1.

Animal characteristics

Preparation of samples

Serum was obtained from cynomolgus monkeys subjected to the ethanol self-administering protocol described above that included naïve, drinking, and abstinent periods. Whole blood was collected through leg venipuncture (unanaesthetized voluntary presentation of the leg) and then centrifuged to collect serum. Ten μL of serum was diluted 1:10 in 0.1% Triton X-100 in phosphate-buffered saline (PBS) (Sigma, St Louis, MO) and assayed for protein concentration using the BCA Protein Assay Kit (Pierce Endogen, Rockford, IL).

Serum profiling by SELDI-ToF MS of drinking and abstinent animals

SELDI-ToF MS analysis was performed using a reverse phase chip (H4 ProteinChip, Ciphergen, Fremont, CA). The H4 chip contains a chemical surface that binds proteins through hydrophobic interaction. The chip surface (8 spots per chip) was pretreated with 3 μl of 100% acetonitrile (Sigma). Ten micrograms of serum protein from the four drinking animals and four abstinent animals was applied to separate spots on the chip. The chip was placed in a humidity chamber for 30 min followed by washing with 0.1% TX-100 in PBS (5 min) and 1× PBS (5 min). Energy-absorbing matrix [1 μl, saturated solution of 3, 5-dimethoxy-4-hydroxycinnamic acid (2 mg/ml; sinapinic acid, Sigma)] was applied to each spot of the chip and allowed to air dry and crystallize. The chips were then subjected to analysis on the Protein Biology System II (PBSII; Ciphergen) SELDI mass spectrometer to obtain a mass spectral profile of the proteins in the samples. Based on experience with our Ciphergen PBSII system, we began with a laser intensity of 220 and a detector setting of 7. The range for the laser is 0–300 and for the detector the range is 1–10. Single exposures of the laser are taken from each spot on the chip to determine the best settings that will give optimal spectra. Once these settings are determined, 130 laser illuminations are made from sites on each spot to produce the composite spectra for that spot. There are 100 positions on each spot to which the laser can be directed. Typically, the data collection starts at position 20, 10 laser illuminations are collected, and then the laser is indexed up by 5 transients until the final collection is performed at position 80.

Spectra are then divided into HE (high excitation energy) and LE (low excitation energy) and calibrated with the appropriate set of internal control protein mass markers (low mass = hirudin, cytochrome C [bovine], myoglobin, and carbonic anhydrase; high mass = carbonic anhydrase, enolase, and albumin). The spectra were analysed using the Biomarker Wizard analysis software (Ciphergen) to identify differentially expressed serum markers.

Peptide mass fingerprinting

To determine the protein identity of the biomarkers found by SELDI-ToF profiling, pools of serum protein from drinking and abstinent animals were resolved on a Novex 4–20% gradient Tris–glycine (Invitrogen, Carlsbad, CA) 1-D PAGE gel. Gels were stained with Colloidal Blue (Invitrogen) to visualize the proteins and prominent bands at the approximate molecular weights (MWs) of 8.7 and 25 kDa were excised from the gel. Proteins were then in-gel digested with trypsin according to the manufacturer's protocol (Ciphergen) and 2 μl of the digested solution was spotted onto an H4 chip and allowed to air dry. The energy-absorbing molecule, α-cyano-hydroxy-cinnamic acid (1.0 mg/ml; Ciphergen) in 50% acetonitrile (v/v; Sigma), 0.5% trifluoracetic acid (v/v; Sigma) was applied, allowed to dry, and the ProteinChip was analysed in the PBSII. Peptide standards were used for external calibration of MW. For internal calibration, the 2163.3 Da trypsin auto-digestion peak as well as a matrix peak at 840 Da were used. This internal calibration allows for more accurate resolution of the peaks in the mass spectra. Mass spectral peaks from a trypsin only negative control digestion were then removed from the analysis, leaving only peptide peaks from the proteins of interest. This peptide mass fingerprint (Pappin, 2003) was then submitted to the ProFound search engine (Zhang and Chait, 2000). The search parameters in ProFound used for the 27 kDa protein were set to a mass range 27–32 kDa; pI range 4–6; single protein only; trypsin digest with acrylamide modification; two maximum missed cuts; charge state of M; and a mass tolerance set at 200. For the 8.9 kDa protein, the search parameters were the same as above except for the mass range 8–9 kDa and a mass tolerance of 300.

Immunoblot analysis of naïve and drinking animals

To confirm both the quantitation and identification of the serum biomarkers, serum samples from 10 independent animals (not used in the initial SELDI-ToF screening) were used for a within-subject immunoblot analysis. Samples were collected from the animals before any ethanol consumption (naïve, non-drinking) and after induction of ethanol self-administration and 5 months of consumption (drinking). Serum samples were diluted in 0.1% Triton X-100 in PBS and protein concentration was determined as previously described using the BCA kit (Pierce Endogen). Ten micrograms of protein was resolved by denaturing PAGE (SDS–PAGE) on 12% NuPAGE pre-cast polyacrylamide gels (Invitrogen). The proteins were transferred to a nitrocellulose membrane (Immobilon-P; Millipore Corporation, Bedford, MA) and blocked overnight in 5% non-fat milk, washed in PBST and probed with apolipoprotein AI (Apo-AI) and apolipoprotein AII (Apo-AII) specific polyclonal antibodies (goat anti-apolipoprotein AI, 1.0 mg/ml; rabbit anti-apolipoprotein AII, 1.0 mg/ml; Biodesign, Int., Saco, ME). Membrane-bound antibodies were detected with donkey anti-goat IgG horseradish peroxidase (HRP) (Santa Cruz, Santa Cruz, CA) and mouse anti-rabbit IgG-HRP (Amersham Pharmacia Biotech, Piscataway, NJ), respectively, and visualized using enhanced chemiluminescence (Perkin-Elmer; Fremont, CA).


Immunoblot signal intensities of naïve and drinking samples were analysed using paired two-tailed Student's t-test, α, P < 0.05.


SELDI-ToF analysis of serum protein from drinking and abstinent animals revealed ∼60 peaks (using a signal to noise ratio of 5) on the H4 hydrophobic chip, most of which showed no differential expression. In addition, screens of serum samples on two additional surface chemistries, the Strong Anion Exchange (SAX2) and the Weak Cation Exchange (WCX2), detected 21 and 55 protein peaks, respectively; none of which were notably altered by ethanol consumption (data not shown). Two peaks, with MWs of 8.7 and 27.9 kDa, from the H4 chip demonstrated differential expression between the drinking and abstinent animals (Fig. 1). To aid in identification of these proteins, 1-D PAGE was performed, and bands at the approximate MW of 8.7 and 25 kDa were excised and subjected to peptide mass fingerprinting. SELDI-ToF was used to obtain mass spectra for the peptides derived from each band and peptide mass spectral maps were submitted through ProFound to identify the proteins. The 25 kDa band, was identified as Apo-AI, (Z-score of 2.43). The peptide map for the 8.7 kDa peak was identified as Apo-AII, (Z-score of 2.39). Both of these identifications were made with a greater than 99% confidence (Zhang and Chait, 2000). The difference seen between the SELDI-ToF MS (27.9 kDa) and 1-D PAGE (25 kDa) apparent MWs is due to the lower resolution of the PAGE approach, as well as anomalous migration of selected proteins (particularly hydrophobic species).

Fig. 1.

SELDI-ToF MS of four ethanol-naive and four ethanol-consuming animals. Spectra showing increases in a 27.9 and 8.7 kDa proteins between animals that had self-administered ethanol for 6 months and those which had abstained from drinking for 6 months after self-administering for 6 months. x-Axis is the mass to charge ration (m/z) and y-axis is peak intensity.

To ensure that this difference in MWs was not the result of a different protein species, purified human Apo-AI and AII were also subjected to SELDI-ToF analysis. The purified human Apo-AI yielded a peak at a mass of 28.1 kDa, which, when compared with the experimental monkey sample, is 199 Da larger. This corresponds to a two-amino acid difference between the Apo-AI in these two species. The purified human Apo-AII yielded a peak at a mass of 8.79 kDa, which, when compared with the experimental monkey sample is almost identical. The combination of the peptide mass fingerprinting results and the purified protein mass spectra provided for an unambiguous identification.

Next, immunoblot analysis was used to confirm the SELDI-ToF findings. A different group of animals was used for this portion of the study (no samples remained from the animals used in SELDI-ToF experiment). The animals used for confirmation were 10 males from which samples were collected prior to any ethanol exposure and after 5 months of daily drinking. These samples were chosen to eliminate sex-based differences and to confirm that the induction of Apo-AI and AII occurs with drinking and not just during abstinence from drinking. Apo-AII protein levels increased by an average of 50% within subjects after chronic drinking (paired t-test, P < 0.001) (Fig. 2). Apo-AI levels were not different between naïve and drinking animals. The effect of cumulative dose on Apo-AI and Apo-AII expression was checked using Pearson correlation testing and no significant correlation between dose and protein expression was found.

Fig. 2.

Immunoblot analysis of Apo-AI and AII. Levels of Apo-AI and Apo-AII were compared, within subject (n = 10), in serum samples before exposure to ethanol (naïve) and after 5 months of ethanol self-administration (drinking). Apo-AII expression was increased by 50% in a within-subject comparison, while Apo-AI was not significantly altered. *P < 0.001, paired t-test.


In this study, we set out to identify biomarkers associated with the consumption of ethanol. Such proteins could be of clinical importance in the treatment of ethanol abuse and to aid in determining a patient's drinking profile. In using cynomolgus monkeys in this study, we were able to control the diet and the amount of ethanol consumed, dramatically reducing the confounding factors present in clinical populations. As well, these experiments were conducted with unadulterated 4% ethanol. Therefore, other chemical constituents that vary with the source of ethanol are not a factor in this study. Moreover, non-human primates serve as an important bridge for translating animal models of alcoholism into the clinic (Grant and Bennett, 2003).

In this study, two differentially expressed proteins were detected on SELDI-ToF MS that appeared to be up-regulated in a comparison between ethanol drinking animals and those which were abstinent for 6 months after 6 months of daily drinking. The two proteins were identified as Apo-AI and AII by SELDI-ToF MS. Ethanol's effects on Apo-AI and AII have been previously documented in humans but have not been examined in non-human primates (Fraser et al., 1983; Duhamel et al., 1984; Camargo et al., 1985; Masarei et al., 1986; Branchi et al., 1997; De Oliveira E Silva et al., 2000; Nomura et al., 2004). It should be noted that the SELDI-ToF screening approach is unbiased and that this screening resulted in two known ethanol-responsive serum proteins. Generally, increases in both Apo-AI and AII have been observed in humans with ethanol consumption. There are exceptions, such as a previous human study using SELDI-ToF in which an Apo-AII fragment was decreased during abstinence after chronic drinking (Nomura et al., 2004).

Furthermore, there has been disagreement as to whether changes in serum apolipoprotein levels are the result of ethanol itself of other substances found in alcoholic beverages. Human studies across a range of alcoholic beverages such as red wine, beer, and grain liquor (Camargo et al., 1985; Masarei et al., 1986; Gottrand et al., 1999; Senault et al., 2000) have all found increases in apolipoproteins. In this study, we have extended that finding to a non-human primate model with strict experimenter-controlled diet and consumption of pure ethanol diluted to 4%. These results suggest that changes in apolipoproteins are directly caused by ethanol consumption.

The exact mechanism by which ethanol affects levels of Apo-AI and AII is still unknown. One report has hypothesized that the increase is due to the fractional catabolic rate (FCR) being decreased, leading to an increase in apolipoprotein levels (Gottrand et al., 1999). A more recent study demonstrated no change in FCR, but an increase in the transport rate of apolipoproteins out of the liver (De Oliveira E Silva et al., 2000). This increase would lead to more apolipoproteins being brought into the bloodstream, thereby aiding the efflux of cholesterol from the blood to the liver by way of reverse cholesterol transport (Segrest et al., 2000). Nonetheless, the increase in high-density lipoprotein concentrations and the reduction on low density lipoprotein fractions may account for the cardio-protective effects of alcohol (Taskinen et al., 1987).

This study confirms the increases of Apo-AII reported in earlier human studies in a non-human primate model with well-controlled diet and complete knowledge of ethanol history. Due to the high variability of Apo-AI, we were unable to document statistically significant increases in Apo-AI. This variability of Apo-AI expression has been observed in humans (De Oliveira E Silva et al., 2000). Other possibilities of why the Apo-AI SELDI-ToF finding was not confirmed include the sex-based effects and that the initial finding was based on a small number of animals (n = 4 vs n = 10 for confirmation). The finding that pure ethanol, with no brewing or fermenting additives and under low fat dietary conditions, increases the serum concentrations of Apo-AII coincides with human studies (Senault et al., 2000). This study also demonstrates that ethanol has similar effects on serum profiles of the macaca compared with humans, further supporting the importance and utility of this animal model in studying the physiological effects of chronic ethanol consumption. The potential use of Apo-AII as a diagnostic biomarker for alcohol intake is limited by the effects of anti-hyperlipidemia drugs, diet, genetics, and exercise on serum levels. As a validation of the experimental proteomics approaches, it is noteworthy that an anonymous screen of ethanol-regulated proteins identified these two important apolipoprotein species. The initial goal of the study was to find biomarkers in an unbiased screen of hundreds of proteins. Further studies are needed which examine the proteome in greater depth and breadth, making use of depletion (Echan et al., 2005; Freeman et al., 2006) and fractionation (Tang et al., 2005) strategies.


Supported by NIH Grants P50-AA11997, T32-AA07565, and R01DA013770, as well as a NARSAD Young Investigator Award. The authors wish to thank Robert Diggs and Kruti Patel for helpful comments on the manuscript.


  • The first two authors contributed equally to the work.


View Abstract