OUP user menu

Tail-associated structural protein gp61 of Staphylococcus aureus phage φMR11 has bifunctional lytic activity

Mohammad Rashel , Jumpei Uchiyama , Iyo Takemura , Hiroshi Hoshiba , Takako Ujihara , Hiroyoshi Takatsuji , Koichi Honke , Shigenobu Matsuzaki
DOI: http://dx.doi.org/10.1111/j.1574-6968.2008.01152.x 9-16 First published online: 1 July 2008


A tailed bacteriophage, φMR11 (siphovirus), was selected as a candidate therapeutic phage against Staphylococcus aureus infections. Gene 61, one of the 67 ORFs identified, is located in the morphogenic module. The gene product (gp61) has lytic domains homologous to CHAP (corresponding to an amidase function) at its N-terminus and lysozyme subfamily 2 (LYZ2) at its C-terminus. Each domain of gp61 was purified as a recombinant protein. Both the amidase [amino acids (aa) 1–150] and the lysozyme (aa 401–624) domains but not the linker domain (aa 151–400) caused efficient lysis of S. aureus. Immunoelectron microscopy localized gp61 to the tail tip of the φMR11 phage. These data strongly suggest that gp61 is a tail-associated lytic factor involved in local cell-wall degradation, allowing the subsequent injection of φMR11 DNA into the host cytoplasm. Staphylococcus aureus lysogenized with φMR11 was also lysed by both proteins. Staphylococcus aureus strains on which φMR11 phage can only produce spots but not plaques were also lysed by each protein, indicating that gp61 may be involved in ‘lysis from without’. This is the first report of the presence of a tail-associated virion protein that acts as a lysin, in an S. aureus phage.

  • Staphylococcus aurerus
  • bacteriophage
  • tail-associated lysin


To initiate successful infection, an incoming phage must cross the barrier presented by the host cell wall (Rydman & Bamford, 2002). Numerous studies have clarified the initiation of the infection mechanisms of phages in Gram-negative bacteria at the molecular level (Kao & McClain, 1980; Caldentey & Bamford, 1992; Sandmeier, 1994). Generally, three types of muralytic enzymes play essential roles in the lysis of host bacteria by phages and in bacterial cell division: the glycosidases (muramidases, glucosaminidases, and transglycosylases), amidases, and endopeptidases (Young, 1992). The crystal structure of the tail base plate of phage T4 confirmed that the middle lysozyme domain of gp5 digests the bacterial peptidoglycan layer (Kanamaru et al., 2002). The molecular basis of receptor binding and destroying activity of Salmonella phage P22 tailspike protein has been described (Steinbacher et al., 1997). The internal head protein, gp16, of phage T7 may have transglycosylase activity, and the activity is suggested to be required for an efficient infection (Molineux, 2001). The entry-associated lysins of PRD1 and φ6 are located in the internal membrane (Rydman & Bamford, 2000, 2002) and between the phage membrane and the nucleocapsid (Bamford et al., 1987), respectively. Phage K1F possesses endosialidase at the tailspike to degrade the polysialic acid-capsule (Stummeyer et al., 2005).

Surprisingly, studies of the lytic molecule responsible for the initiation of infection of Gram-positive bacteria by phages are extremely rare at present. Recently, one report of a phage of the Gram-positive bacterium Lactococcus lactis, Tuc2009, showed that tal2009 encodes a tail-associated protein that may be involved in localized cell-wall degradation (Kenny et al., 2004). Detailed proteomics analysis of a staphylococcal phage 812 has also been reported recently (Eyer et al., 2007).

Bacteriophage endolysin, bacterial autolysin, and a large family of proteins encoded by Archaea and eukaryotes of the family Trypanosomidae have been linked by the presence of the cysteine, histidine-dependent amidohydrolase/peptidase (CHAP) domain (Bateman & Rawlings, 2003; Rigden et al., 2003; Donovan et al., 2006). Most of these proteins are uncharacterized, but it has been proposed that they function mainly in peptidoglycan hydrolysis (amidase) (Pritchard et al., 2004; Yokoi et al., 2005) and they are commonly associated with several families of amidase functions (Bateman & Rawlings, 2003; Rigden et al., 2003).

Staphylococcus aureus is usually resistant to lysozyme, resulting from O-acetylation and a high degree of crosslinking of its peptidoglycan (Bera et al., 2005, 2006, 2007). Although a novel plant lysozyme with antifungal and antistaphylococcal activity has been described (Wang et al., 2005), lysozyme subfamily 2 (LYZ2), a eubacterial enzyme (SMART accession number: SM00047; http://smart.embl-heidelberg.de) distantly related to the eukaryotic lysozyme, has been predicted to be widely distributed in Staphylococcus phage, Staphylococcus bacteria, and other related bacteria. However, it has not yet been purified or functionally analyzed.

Previously, we reported a novel S. aureus phage φMR11 in the context of phage therapy. φMR11 belongs to morphotype B1 of the Siphoviridae family, and is likely related to phages of serogroup B(Rosenblum & Tyrone, 1964; Ackermann & DuBow, 1987). It has an icosahedral head and a noncontractile tail with a knob-like structure at the tip (Matsuzaki et al., 2003). In this study, we describe a structural component of phage φMR11, the product of gene 61 (gp61), which has a CHAP domain in the N-terminal region and a lysozyme subfamily 2 (LYZ2) domain in the C-terminal region. Both these domains exhibit clear lytic activity towards S. aureus. This protein is located towards the end of the phage tail and may be involved in lysis from without. To the best of our knowledge, this is the first report of a protein with bifunctional (amidase and lysozyme) lytic activity, which is a structural component of a phage.

Materials and methods

Culture media and bacterial strains

All reagents and constituents of culture media used in this study were purchased from Nacalai Tesque (Kyoto, Japan), unless otherwise stated. Tryptic soy broth (TSB) and brain–heart infusion broth (BHI) were obtained from Becton Dickinson (BD; Sparks, MD). TSBM is TSB medium supplemented with 20 mM MgCl2. For phage plaque formation and the spot test, TSBM-based solid medium containing 1.5% or 0.5% agar was used for the lower or the upper layer, respectively. Terrific broth (TB) was purchased from Sigma-Aldrich Co. (St. Louis, MO).

Bacteria and phage

The methicillin-sensitive S. aureus (MSSA) strains and methicillin-resistant S. aureus (MRSA) strains used in this study and phage φMR11 have been described previously (Matsuzaki et al., 2003).

Cloning gene 61 and its deletion mutants

φMR11 phage DNA was prepared as described previously (Matsuzaki et al., 2003). A PCR reaction was carried out using φMR11 phage DNA (0.5 μg) as the template with specific primer sets (Supplementary Table S1) for the complete gene 61 and different deletion constructs, as indicated in Fig. 1. The PCR products were digested with EcoRI/PstI (TaKaRa Bio Inc., Kyoto, Japan), purified using the JETSORB Gel Extraction Kit (Genomed, Bad Oeynhausen, Germany), and ligated to the EcoRI/PstI sites of pTrc99A (Amersham Biosciences, Piscataway, NJ) to clone the full-length gene or to the sites of pColdIII vector (Takara Bio Inc., Kyoto, Japan) to generate deletion constructs of the gene. Escherichia coli DH5α was transformed with the ligated products and the transformants were selected on Luria–Bertani plates containing 100 μg mL−1 ampicillin. The plasmid DNA was purified from the transformants, and the positive clones were screened by EcoRI/PstI digestion. Several positive clones were sequenced over the total insert region, together with flanking sequences, for each construct. Plasmids containing the correct insert in-frame were transferred into E. coli BL21. A plasmid containing the full-length gene encoding gp61 was designated pTrc61 and the plasmids containing deletion mutant genes were designated pCold1–150, pCold151–400, and pCold401–624, according to the amino acids they encoded. The sequence of gene 61 of φMR11 was deposited in the DDBJ database, accession no. AB360386.


Schematic representation of φMR11 gp61 and its derivatives. (a) Full-length gp61, indicating the predicted N-terminal CHAP (aa 29–146) and C-terminal LYZ2 (aa 473–610) domains, with a His tag at the C-terminus. The CHAP region (aa 1–150) including the His tag at the C-terminus, the spacer region (aa 151–400) with the His-tag at the C-terminus, and the LYZ2 region (aa 401–624) with the His tag at the C-terminus are also shown. (b) Purification of full-length and three individual regions of gp61. The proteins were expressed in Escherichia coli BL21, purified by affinity chromatography, and analyzed by SDS-PAGE. Size markers are shown.

Purification of φMR11 gp61 and its derivative proteins

Escherichia coli BL21 cells carrying pTrc61 were cultured overnight at 37 °C in TB medium containing 100 μg mL−1 ampicillin (TB–amp). The culture was then transferred into 50 volumes of fresh TB–amp and grown further at 37 °C until the OD600 nm reached 1.2. The OD was measured using a spectrophotometer (U-2000, Hitachi, Tokyo, Japan). gp61 expression was induced with 1 mM isopropyl-β-d-thiogalactoside (IPTG) (Nacalai Tesque, Kyoto, Japan) for 2 h at 37 °C. The cells were pelleted at 4500 g for 5 min, washed with phosphate-buffered saline (PBS), and stored at −80 °C until used for protein purification. Escherichia coli BL21 cells carrying pCold1–150, pCold151–400, or pCold401–624 were cultured overnight at 37 °C in TB–amp. The cultures were then incubated at 15 °C for 30 min, and protein expression was induced with 1 mM IPTG for 24 h at 15 °C. The cells were processed as described above. The cells were suspended in 300 mM NaCl, 50 mM NaPB (pH 7) and disrupted using an ultrasonic disintegrator. The proteins were purified with BD TALON Metal Affinity Resin (BD Biosciences Clontech, Palo Alto, CA) according to the manufacturer's instructions. The peak fractions were collected and dialyzed against PBS (pH 7.2). The purified proteins were stored at −145 °C until use.

Preparation of anti-gp61 antibody

Purified gp61 (500 μg) was mixed with an equal volume of a complete adjuvant and inoculated into the back of a rabbit. Similar inoculations were performed twice, at 10-day intervals, with an incomplete adjuvant. Blood was collected 40 days after the first inoculation and antiserum was prepared by centrifugation of the blood at 3000 g for 10 min.

Bacterial lysis and viability tests

Staphylococcus aureus cells were grown to the early stationary phase and diluted with medium to 200 Klett units (KU), which were measured using a Klett–Summerson photoelectric colorimeter (filter #54) (Klett Manufacturing Co., Inc., NY). An aliquot (5 mL) of the suspension was centrifuged at 9000 g for 5 min and the medium was discarded. The cells were washed with PBS and resuspended in 5 mL of PBS. An aliquot (100 μL) of the cell suspension was transferred to a 1.5 mL tube (in triplicate for each sample) and centrifuged at 9000 g for 5 min. The supernatant was discarded. The cells were treated with 200 μL of protein solution (0.1 μg μL−1) or PBS and incubated at room temperature for 30 min. The cell suspension was transferred to a microtiter plate and the OD595 nm was measured for the triplicate wells. An aliquot (5 μL) of the cell suspension was serially diluted 10-fold and the colonies were counted on duplicate BHI plates. An aliquot (195 μL) of the suspension was centrifuged and digital photographs of the pellet were taken.

Electron microscopy

Staphylococcus aureus cells were prepared and treated with the recombinant proteins, as described above. After incubation for 30 min at room temperature, the lytic reactions were stopped by the addition of 2% glutaraldehyde (TAAB Laboratories Equipment Ltd, Berkshire, UK). The cell pellet was coated with 1% agarose. After the sample had been rinsed in PBS with 5% sucrose, it was postfixed at 4 °C for 1 h in 1.5% osmium tetroxide in 0.1 M phosphate buffer with 5% sucrose, dehydrated in a graded series of ethanol, substituted with propylene oxide, and embedded in epoxy resin (TAAB 812). Ultrathin sections were prepared using an ultramicrotome, stained with uranyl acetate and lead citrate, and then examined using a transmission electron microscope (H7100, Hitachi, Japan).

Immunoelectron microscopy

Phage particles were purified using stepwise CsCl density-gradient ultracentrifugation, as described previously (Matsuzaki et al., 2003). Immunogold electron microscopy was performed as described previously, with simple modifications (Lubbers et al., 1995; Matsuzaki et al., 1998). Phage (20 μL) were spotted onto parafilm and an electron microscopical grid was applied to it for 1 min. The grid was then applied to TAM buffer [10 mM Tris-Cl (pH 7.0), 10 mM NaNO3, 1 mM MgCl2] for 1 min and incubated with anti-gp61 antibody for 30 min at 37 °C inside a box containing water to prevent evaporation. Water was then applied to the grid for 1 min. Immunogold-conjugated secondary antibody (10 nm particles; British BioCell International, London, UK) was diluted with dilution buffer [0.5 M NaCl, 0.1% bovine serum albumin, 0.05% Tween 20, 5% fetal bovine serum (Clontech, Palo Alto, CA)] three times and the grid was applied to it for 1 h, before it was rinsed with water. The samples were then stained with 2% uranyl acetate for 2 min and examined using a transmission electron microscope (H7100, Hitachi, Japan) at 75 kV.

Western blot analysis

Purified φMR11 phage particles were solubilized in sodium dodecyl sulfate (SDS) sample buffer and the total virion proteins were separated by SDS-polyacrylamide gel electrophoresis (PAGE; 12.5%) (Laemmli, 1970), transferred onto a Pall FluoroTrans membrane (Pall Corporation, New York, NY), and blocked overnight with 5% skim milk in Tris-buffered saline [TBS; 100 mM Tris-Cl (pH 7.5), 150 mM NaCl]. The membrane was then incubated with anti-gp61 antibody (1 : 1000) for 3 h, washed three times with TBS with 0.1% Tween 20 (TBS-T), and blocked with 5% skimmed milk in TBS for 30 min. It was then stained with horseradish-peroxidase-conjugated goat anti-rabbit IgG (1 : 2000; Sigma-Aldrich), washed three times with TBS-T, and visualized by enhanced chemiluminescent (ECL) detection (GE Healthcare, Buckinghamshire, UK).

Phylogenetic analysis

Homologues of gp61 protein were identified by searching the ncbi database with the resident Protein blast program (Altschul et al., 1997; http://www.ncbi.nlm.nih.gov/BLAST/). Alignments were prepared using clustalw (Thompson et al., 1994; http://clustalw.ddbj.nig.ac.jp/top-j.html). Phylogenetic trees were constructed using the treeview (Page, 1996) program from multiple protein sequence alignments generated by clustalw.


Identification and purification of φMR11 gp61 and its derivatives

φMR11 phage has a double-stranded DNA genome (43 011 bp), which contains 67 putative ORFs (Matsuzaki et al., 2003). Bioinformatic analysis indicated that ORF61 (gene 61) encodes a possible cell-wall hydrolase, which consists of 624 amino acid (aa) residues. gp61, the product of φMR11 gene 61, was deduced to have two typical catalytic domains in the cell-wall hydrolase, i.e., amidase and lysozyme domains in the N-terminal and C-terminal regions, respectively. To identify and analyze the functions of gp61, we cloned gene 61 and mutant genes derived from it that encode only N-terminal aa 1–150 (gp611–150), middle region aa 151–400 (gp61151–400), or C-terminal aa 401–624 (gp61401–624) of gp61. We purified gp61 and these truncated proteins by affinity chromatography and analyzed them by SDS-PAGE (Laemmli, 1970). The purified proteins corresponded to the calculated molecular masses i.e., gp61 is 71.7 kDa with a 6 × His tag at the C-terminus. gp611–150, gp61151–400, and gp61401–624 have a 6 × His tag at the C-terminus and are about 17.8, 29.9, and 26.1 kDa, respectively (Fig. 1).

Lytic activity of the putative amidase and lysozyme domains of φMR11 gp61

Because gp611–150 and gp61401–624 are presumed to encode amidase and lysozyme domains, respectively, we examined whether they actually have bacteriolytic activity. Measurement of OD595 nm of the bacterial suspensions indicated that the addition of gp611–150 or gp61401–624 caused a significant decrease in bacterial density within 30 min in all the strains examined (Fig. 2, Table 1). To confirm these lytic effects of gp611–150 and gp61401–624, we also checked the decrease in viable cell counts in the representative strain SA37. Consistent with the OD measurements, the addition of gp611–150 or gp61401–624 to S. aureus SA37 rapidly killed 70–80% of the bacteria within 30 min relative to the control. In contrast, gp61151–400 had no effect on the OD or the viability of S. aureus SA37. These results indicate that gp611–150 and gp61401–624 have lytic activity against S. aureus.


Bacteriolytic activity of different domains of φMR11 gp61. (a) Lytic activity of different deletion mutants of gp61. Representative data of three independent experiments are shown. (b) Killing ability of different deletion mutants of gp61. Representative results of three experiments are shown. The line on the bar shows SD in (a) and (b). (c) Digital photographs of pelleted bacterial debris after application of deletion mutants of gp61.

View this table:

Summary of ‘lysis from without’ analysis

  • * Spot and plaque tests of φMR11 on various Staphylococcus aureus strains, as described previously (Matsuzaki et al., 2003). + indicates spot- or plaque-positive, and – indicates spot- or plaque-negative.

  • Lytic activities of gp611–150, gp61151–400, and gp61401–624 were tested against the S. aureus strains (see ‘Materials and methods’). + indicates a decrease in the initial OD595 nm of at least 50%, and − indicates no significant change in the initial OD595 nm compared with that of buffer-treated bacteria after 30 min treatment with individual proteins.

  • SA37 strain lysogenized with φMR11.

Electron microscopic examination of cells treated with gp611–150 or gp61401–624

To investigate the mechanism of bacterial lysis morphologically, we analyzed the bacteria treated with gp611–150 or gp61401–624 by electron microscopy (Fig. 3). gp611–150 caused leakage of the cytoplasmic contents from all over the cells, leading ultimately to the death of the bacteria. In contrast, the activity of gp61401–624 was somewhat similar to that of endolysin (Rashel et al., 2007). The hydrolysis of peptidoglycan by gp61401–624 caused the protrusion of the cytoplasmic contents at several sites on the bacterial cell, after which the cell burst.


Morphological analysis of the bacterial strain SA37 after treatment with the different purified domains of gp61. Representative figures from TEM analysis are shown. (a) Buffer-treated bacteria. (b) CHAP domain (aa 1–150)-treated bacteria. (c) LYZ2 domain (aa 401–624)-treated bacteria. The arrow indicates the protrusion site.

gp61 is a structural protein located at the tail tip

To clarify the nature of gp61, we prepared anti-gp61 antibody and, using Western blotting, checked whether it interacts with a virion protein. The gp61-specific antibody interacted specifically with a virion protein (Fig. 4a). The size of the protein detected was around 72 kDa, which is the expected size of gp61. This analysis indicates that gp61 is a phage structural protein.


Detection of gp61. (a) Left. SDS-PAGE analysis of the total structural proteins of phage φMR11 visualized with silver staining. Right. Detection of gp61 in the total structural protein pool with a gp61-specific antibody and Western blotting. Size marker confirms the expected size of gp61 deduced from the sequence data. (b) gp61 is located at the tail tip. Immunoelectron microscopic analysis. Control. φMR11 phage treated without gp61-specific antibody. Anti-gp61. φMR11 phage treated with gp61-specific antibody. Anti-gp61 (enlarged). Enlarged view of φMR11 treated with gp61-specific antibody.

We used immunoelectron microscopy to determine the location of gp61 on the φMR11 virion. In phage treated with both anti-gp61 antibody and immunogold-conjugated secondary antibody, the gold colloid was observed attached to the tail tip, indicating that gp61 is located at the tail tip.

gp61 may be involved in ‘lysis from without’

Previously, we reported that φMR11 could form spots on all the S. aureus strains examined (Matsuzaki et al., 2003), but could form plaques only on some of them (30/75 strains). φMR11 is thought to lyse S. aureus by so-called ‘lysis from without’ in the spot-positive and plaque-negative strains. Because gp61 is a structural protein and has clear lytic activity when applied exogenously, we thought it might be involved in lysis from without. The lytic activity of gp611–150 or gp61401–624 against two φMR11 plaque-forming and 20 spot-forming (but not plaque-forming) S. aureus strains was examined (Table 1). Both gp611–150 and gp61401–624 showed lytic activity against 18 strains. Interestingly, two of these strains, SA38 and MR15, were sensitive to gp611–150 but not to gp61401–624. Conversely, another two strains, MR11 and MR14, were sensitive to gp61401–624 but not to gp611–150. Both gp611–150 and gp61401–624 could also lyse SA37/φMR11, which is a φMR11 lysogen derived from SA37. φMR11 could not form plaques on SA37/φMR11. These results suggest that gp61 is involved in lysis from without.

Phylogenic relationships of gp61

A blastp search with full-length gp61 revealed 25 similar proteins with expected (E) values of 2e–136. The first nine of these orthologues were the same size as gp61, i.e. 624 aa, with only a few substitutions at different positions over the entire length of the protein (Supplementary Fig. S1). Phylogenic analysis of these 26 genes revealed that φMR11 gp61 is located in a cluster of predicted tail proteins of the other nine closely related S. aureus phages (data not shown). A homology search using the N-terminal 150 aa of φMR11 gp61 (encoding the CHAP domain) as the query or the C-terminal 224 aa of φMR11 gp61 (encoding the LYZ2 domain) as the query revealed that both the CHAP and the LYZ2 domains are diversely distributed, not only among Staphylococcus phages and their hosts but also among other related bacterial species (data not shown).


In this study, we report for the first time the purification and functional analysis of individual domains of a tail-associated cell-wall hydrolase of an S. aureus phage and possibly of phages of various Gram-positive bacteria. φMR11 gp61 is located at the tail tip, and is inferred to act in punching a hole in the bacterial peptidoglycan layer, allowing the penetration of the tail tip to the surface of the cell membrane of S. aureus.

So far, many S. aureus phage genomes have been sequenced (Kaneko et al., 1998; Yamaguchi et al., 2000; Iandolo et al., 2002; O'Flaherty et al., 2004; Kwan et al., 2005), and their tail-associated cell-wall hydrolase genes have also been predicted based on bioinformatic analysis (Yamaguchi et al., 2000). Although some of them are very similar (96–99% identical) to φMR11 gp61 in their amino acid sequences (Supplementary Fig. S1), no experimental evidence of their cell-wall-lysis activities has been reported until now. It was concluded that gp61 has two putative functional domains: the CHAP domain at the N-terminal (gp611–150) and the lysozyme (LYZ2) domain at the C-terminal (gp61401–624). These two domains have independent lytic activities against many S. aureus strains. Interestingly, some S. aureus strains are sensitive to either gp611–150 or gp61401–624 (Table 1), indicating that these dual catalytic domains may participate in extending the host range (Bateman & Rawlings, 2003).

As described above, gp61 consists of three modular regions (gp611–150, gp61151–400, and gp61401–624), which were also inferred by hydropathy analysis (data not shown). Homology searches indicated that both terminal domains of gp61 are highly conserved among many putative tail-associated cell-wall hydolases in the phages of S. aureus and other species and in bacterial autolysins (data not shown). This indicates that genetic exchange (or replacement) might have occurred among the catalytic modules of the cell-wall hydrolases during the coevolution of bacteria and their phages (Sandmeier, 1994).

Many phages, including φMR11, can produce lysed areas rather than individual plaques on host strains (Matsuzaki et al., 2003). Although this phenomenon has been recognized from the early days of phage biology as ‘lysis from without’, the genetic basis of the phenomenon has not been defined in detail (Ralston et al., 1957; Ralston, 1962). Our results strongly suggest that gp61 is a candidate molecule responsible for this effect, because gp611–150 or gp61401–624 can lyse S. aureus strains independently of their plaque-forming ability (Table 1). Thus, tail-associated cell-wall hydolases may be involved in lysis from without by phages that infect Gram-positive bacteria.

Recently, attention has been paid to active phage or its endolysin as an alternative therapeutic agent against multidrug-resistant S. aureus infections (Slopek et al., 1987; O'Flaherty et al., 2005; Rashel et al., 2007). We have shown that both gp611–150 (amidase domain) and gp61401–624 (lysozyme domain) have clear lytic activity against many S. aureus strains. These domains may be possible therapeutic candidates directed against multidrug-resistant S. aureus.

Supporting Information

Fig. S1. Alignment of φMR11 gp61 orthologues.

Table S1. Primers used in this study.

This material is available as part of the online article from:

(This link will take you to the article abstract.)

Please note: Blackwell Publishing are not responsible for the content or functionality of any supplementary materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.


We thank Mrs Kazuyo Ito for her excellent help in this study. We also thank Mr Ken-ichi Yagyu, Science Research Center of Kochi University, for transmission electron microscopy work. This work was supported in part by research grant the Special Research Project of Green Science, Kochi University.


  • Editor: Wolfgang Schumann


View Abstract