Changes of proteome components of Helicobacter pylori biofilms induced by serum starvation

  • Authors:
    • Chunhong Shao
    • Yundong Sun
    • Na Wang
    • Han Yu
    • Yabin Zhou
    • Chunyan Chen
    • Jihui Jia
  • View Affiliations

  • Published online on: October 4, 2013
  • Pages: 1761-1766
Metrics: Total Views: 0 (Spandidos Publications: | PMC Statistics: )
Total PDF Downloads: 0 (Spandidos Publications: | PMC Statistics: )


Biofilm is the adaptive living mechanism of Helicobacter pylori (H. pylori) during survival and propagation. Nutrient starvation is an environmental pressure for H. pylori in vivo and in vitro. Serum starvation effectively mimics the microenvironment in which H. pylori colonizes healthy humans who carry H. pylori and patients with chronic atrophic gastritis. In addition, it also mimics the in vitro environmental pressures of H. pylori. An H. pylori biofilm was successfully induced with serum starvation. To identify novel proteins associated with biofilm formation at the early stage in H. pylori, high-resolution 2-dementional gel electrophoresis was performed to obtain the proteome profiles of spiral H. pylori and early biofilm. Differential protein spots were identified using tandem matrix assisted laser desorption ionzation time of flight mass spectrometry, which revealed 35 proteins. These proteins are associated with various biological functions, including flagellar movement, bacterial virulence, signal transduction and regulation. To verify the results, the expression of cagA at the mRNA and protein levels was examined by fluorescence quantitative PCR and western blot analysis, respectively. This study indicates that H. pylori form biofilms by initiating multiple mechanisms involving a number of signaling pathways.


Helicobacter pylori (H. pylori) infection is a predominant cause of gastristis and peptic ulcer disease, as well as an early risk factor for gastric cancer. It is a globally epidemic pathogenic bacteria and more than half of the world’s population has been infected with this bacterium. H. pylori infection is common; however, its precise mode of transmission has not yet been identified. The main reason for this is that the etiology of H. pylori has not been solved effectively. Therefore restrictng the formulation of prevention measures against H. pylori.

A biofilm is a multicellular layer of bacteria, which attaches to surfaces, interfaces or other cells and is embedded within a matrix of extracellular polymeric substances under stress conditions (1). Growth in a biofilm provides a number of advantages for bacteria, including enhanced power of resistance to environmental stresses as well as to host defenses, which leads to serious clinical problems, particularly chronic and refractory infection. The biofilm of H. pylori has been observed in the water system, including drinking water, groundwater and sewage (2,3) and also on the human gastric mucosal epithelium (4,5). A number of studies have been performed on H. pylori biofilms (6,7). Thus, biofilms are hypothesized to be another adaptive survival mode of H. pylori. In conventional experiments, H. pylori is cultured in a rich medium containing fetal bovine serum (FBS). However, this may not effectively imitate the in vitro and in vivo survival environment of H. pylori. Nutrient starvation is a common environmental pressure that H. pylori is faced with. Serum starvation effectively mimics the microenvironment H. pylori within healthy human carriers of H. pylori and patients with chronic atrophic gastritis (8) and also the in vitro pressure environment pressure that H. pylori is subjected to. The current study identified that H. pylori adhered to each other and formed a population structure similar to that of a bacterial biofilm by removing serum from medium. Williams et al(9) observed that serum significantly affects the movement, morphology and adhesion of H. pylori and may result in agglutination in 1% serum concentration. Campylobacter jejuni also agglutinate in nutrient-limiting conditions. Reeser et al(10) observed that the autoagglutinated structure is a biofilm by analyzing its extracellular polymer matrix.

The formation of a biofilm is a complex and dynamic process. Under adverse conditions, the bacteria induce a unique set of genes, adhere to each other and form structural groups or subgroups. During biofilm formation, each link may be regarded as a key point to treat infection and drug design target, particularly at the early stage. In the present study, proteomic technologies were used to analyze the proteomic expression profiles of H. pylori biofilms at an early stage, and aimed to reveal the proteins and signaling molecules involved in H. pylori biofilm formation and identify novel pathogenic factors and antibacterial targets.

Materials and methods

Bacterial strain and culture conditions

The H. pylori strain 26695 was provided by Dr Zhang Jianzhong from the Chinese Disease Control and Prevention Center. It was cultured in Brucella broth with 10% FBS with 120 rpm agitation under microaerobic conditions (5% O2, 10% CO2, 85% N2, v/v) at 37°C. The same quantity of exponentially growing H. pylori was resuspended in Brucella broth with and without 10% FBS. Bacterial morphological changes were then observed using Gram staining at intervals of 2 h until the autoagglutinated structure formed. The initial time of biofilm formation was recorded. Subsequently, the planktonic H. pylori and the biofilm were fixed with 2.5% glutaraldehyde solution for >2 h. The critical-point dried sample was observed with a Hitachi S-520 scanning electron microscope (Hitachi Limited Company, Tokyo, Japan).

2-dimentional gel electrophoresis (2-DE) and image analysis

The planktonic H. pylori and its early biofilm cultured with and without 10% FBS for 4 h were harvested by centrifugation at 5000 × g for 10 min at 4°C and washed three times with ice-cold PBS (pH 7.2). The pellets were solubilized in lysis buffer containing 8 M urea, 2 M thiourea, 4% CHAPS, 1% DTT, 1% pharmalyte (pH range, 3–10), 1% protease inhibitor and 1% nuclease mix (Amersham Pharmacia Biotech, Piscatawway, NJ, USA). Following sonication, the solution was centrifugated at 20,000 × g for 60 min at 4°C to remove cell debris. Protein concentrations were determined using the Bradford method and proteins were stored at −80°C. The 2-DE analysis was performed as described previously (11). The SDS gels were silver-stained and the protein patterns were scanned using an ImageScannerII (Amersham Pharmacia Biotech). The gels were analyzed by ImageMaster 2D Elite 5.0 (Amersham Pharmacia Biotech) to determine differentially expressed protein spots whose vol% ratio increased >2-fold (P<0.05).

In-gel digestion and matrix assisted laser desorption ionzation time of flight mass spectrometry (MALDI-TOF-TOF MS)

Differentially expressed protein spots were excised, tryptic digested and identified with a 4700 MALDI-TOF-TOF proteomic system (Applied Biosystems) as described previously (11). The MS and MS/MS spectra were analyzed with a 50 ppm mass tolerance by GPS Explorer V.2.0.1 and Mascot V1.9 based on the NCBI and SWISSPROT databases.

Quantitative PCR

Total RNA was isolated from three independent cultures of planktonic H. pylori and biofilm at early stages using TRIzol reagent (Invitrogen Life Technologies, Carlsbad, CA, USA) according to the manufacturer’s instructions. The quantity was measured by absorbance at 260 nm. Subsequently, 4 μg total template RNA was reverse transcribed into cDNA using M-MLV reverse transcriptase and random hexamer primer (MBI). The primers for PCR are listed in Table I. The 16S rRNA gene served as the endogenous control. PCR was performed as described previously (12). The operating conditions were as follows: One cycle at 50°C for 2 min, one cycle at 95°C for 10 min, 40 cycles at 95°C for 15 sec and 60°C for 1 min. Gene expression was quantified by the comparative threshold cycle (Ct) method and normalized to the quantity of 16S rRNA cDNA in each sample. The relative quantity of the target was calculated by the ΔΔCT-method.

Table I

Primers used in the study

Table I

Primers used in the study

PrimersSequence (5′-3′)
Western blot analysis

Protein samples were separated by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (8% SDS-acrylamide gels) and transferred to a nitrocellulose membrane. The membranes were incubated in blocking buffer [TBS containing 0.1% Tween-20 (TBST) and 5% non-fat powdered milk] for 1.5 h and immunoblotted for 1.5 h with antibodies against β-actin or C-CagA (Santa Cruz Biotechnology, Inc. Santa Cruz, CA, USA). Membranes were washed three times with the TBST solution and incubated with the corresponding antibodies conjugated with horseradish peroxidase for 50 min. The washed membranes were developed by the Chemilucent ECL Detection system (Millipore, Billerica, MA, USA).

Results and Discussion

In recent years, increasing evidence has indicated that bacterial biofilms are important in infectious diseases. Biofilms effectively protect bacteria from external environmental damage and are an effective strategy to adapt to environmental pressures in vivo and in vitro. As one of the predominant pathogens affecting humans, H. pylori was subjected to various types of adverse conditions, including nutrient starvation. To produce the survival environment H. pylori generated when under starvation stress, exponentially growing H. pylori 26695(Fig. 1A) was inoculated into medium without FBS. Following 4 h, the bacteria began to adhere to each other and formed a subgroup structure (Fig. 1B). Typical community structures were observed following 12 h starvation and a number of H. pylori changed to a dormant coccoid form (Fig. 1C). Thus, a biofilm of H. pylori was successfully induced with serum starvation.

To identify novel proteins associated with H. pylori biofilm formation, high-resolution 2-DE was performed to obtain the proteome profiles of spiral H. pylori and its early biofilm, which was induced by 4 h of starvation (Fig. 2). 2-DE analysis was repeated in triplicate using independently grown cultures. The comparison of the protein patterns of these two forms of H. pylori revealed that 36 spots exhibited different levels of expression (>2-fold, P<0.05). These spots were subjected to MALDI-TOF-TOF MS analysis to determine their putative genes and infer their functions. A total of 35 proteins with different functions were identified (Table II); of which 29 were upregulated and six were downregulated in the early biofilm of H. pylori. These proteins belong to diverse functional classes, including movement, virulence, energy metabolism, regulator and chaperone proteins. This suggests that H. pylori invokes a multi-mechanism response to adapt to morphological changes under starvation stress.

Table II

Summary of protein spots showing altered expression between planktonic H. pylori and its early biofilm.

Table II

Summary of protein spots showing altered expression between planktonic H. pylori and its early biofilm.

Spot numberaProtein (gene)TIGR ORF numberbGiTop score%vol ratioc
1Cag pathogenicity island protein (cag26)Hp054723136641392.89
2Flagellar capping protein(fliD)Hp07522313869852.21
3Aconitase B (acnB)Hp077923139081052.36
4Aconitase B (acnB)Hp077923139081332.13
6Type I restriction enzyme R protein (hsdR)Hp08462313977462.05
6Urease β-subunit (urea amidohydrolase) (ureB)Hp00722313153452.05
6Arginase (rocF)Hp13992314565502.05
7H. pylori predicted coding region HP0013Hp00132313087642.02
9Transaldolase (tal)Hp149523146741752.04
10Holliday junction DNA helicase (ruvB)Hp105923142031612.04
10GTP-binding protein, fusA-homolog (yihK)Hp04802313589752.04
11 UDP-N-acetylenolpyruvoylglucosamine reductase (murB)Hp14182314592692.15
12Chaperone and heat shock protein (groEL)Hp001023130841067.45
13Poly E-rich proteinHp03222313421642.14
14DNA-directed RNA polymerase, β-subunit (rpoB)Hp11982314357642.26
15GTP-binding protein (obg)Hp030323134011283.24
15 UDP-N-acetylenolpyruvoylglucosamine reductase (murB)Hp14182314592613.24
16 Delta-aminolevulinic acid dehydratase (hemB)Hp016323132501112.13
17 Aspartate-semialdehyde dehydrogenase (asd)Hp118923143502572.31
19Aspartyl-tRNA synthetase (aspS)Hp06172313739653.16
20H. pylori predicted coding region HP0958Hp09582314102904.76
21Conserved hypothetical proteinHp15882314773730.32
213-oxoadipate coA-transferase subunit A (yxjD)Hp06912313815680.32
22Response regulator (ompR)Hp016623132521052.07
23H. pylori predicted coding region HP0406Hp040623135131312.01
24Alkyl hydroperoxide reductase (AhpC)Hp156323147471460.31
26Cag pathogenicity island protein (cag24)Hp05042313660770.44
26Translation elongation factor EF-Tu (tufB)Hp12052314366450.44
27Conserved hypothetical proteinHp10462314193762.57
27Hemolysin secretion protein precursor (hylB)Hp05992313716492.57
28Riboflavin synthase beta chain (ribE)Hp000223130791212.66
29Nonheme iron-containing ferritin (pfr)Hp06532313771820.03
30Hydrogenase expression/formation protein (hypA)Hp086923139961044.37
31H. pylori predicted coding region HP1377Hp137723145551112.55
32Co-chaperone (groES)Hp00112313085825.08
34H. pylori predicted coding region HP1029Hp102923141851422.41

a Spot numbers refer to the proteins labeled in Fig. 2.

b Nomenclature of the H. pylori strain 26695.

c Percentage volume ratio for each protein derived from floating unicellular H. pylori with respect to the protein derived from the early biofilm.

{ label (or @symbol) needed for fn[@id='tfn4-mmr-08-06-1761'] } H. pylori, Helicobacter pylori.

The adhesion of bacteria is the early stages of biofilm formation and the flagellar movement are important in this process. The present study identified that the flagellar filament capping protein (HAP2) was increased in the early biofilm of H. pylori. This protein is encoded by the fliD gene, constitutes the flagellar hook together with HAP1 and HAP3, and is involved in the flagellar assembly of H. pylori. Martin et al observed that proteins associated with flagella exhibited higher expression in the biofilm of Campylobacter and that the deletion of flagellar genes may seriously affect the biofilm formation of bacteria (13,10). In Escherichia coli (E. coli), flagellar motility may significantly affect the three-dimensional structure of the biofilm (14). In addition, aconitase was observed to be upregulated in H. pylori early biofilms. The structural analysis identified that this protein contained a domain similar to the HEAT domain, which mediates the interaction between proteins. In E. coli and Bacillus subtilis, the HEAT protein has been confirmed to have a dual function and regulate flagellar movement as a post-transcriptional regulator under iron and oxygen conditions (15,16). Under serum starvation conditions, the overexpression of aconitase in H. pylori is hypothesized to be involved in biofilm formation by the promotion of flagellar movement. However, this requires further study to be confirmed.

In present study, the cagA protein encoded by cag pathogenicity islands was identified to be induced in H. pylori biofilms. As an important virulence factor of H. pylori, this protein is injected into gastric mucosa epithelial cells through the type IV secretion system when the cagA positive H. pylori infect the individual. CagA is then localized in the inner surface of the plasma membrane and results in a series of changes, including excessive proliferation and phenotypic change, such as extended ‘hummingbird’ morphology (17). To confirm the results, the expression of cagA was examined at the mRNA and protein levels by quantitative PCR and western blot analysis, respectively. The results showed that the expression of cagA was not significantly changed following 4 h of starvation. This may be due to higher resolution of the 2-DE compared with 1-D SDS. However, after 8 h cagA expression was upregulated significantly (Figs. 3 and 4). Thus the upregulation of this protein was hypothesized to aid in the improvement of the viability of H. pylori under adverse environments.

UspA is a conservative and integral promoter of a number of stress-associated proteins in a variety of bacteria. The β-subunit of RNA polymerase positively regulates the expression of UspA under a number of environmental pressures, including starvation. Thus, a variety of stress-related proteins show higher expression levels and bacterial biofilms have strong resistance to the outside pressure (18). In the current study, the β-subunit of RNA polymerase and two chaperone protein GroES and GroEL of H. pylori were observed to be overexpressed under serum starvation. In addition, two oxidative stress-related proteins, alkyl hydroperoxide reductase and thioredoxin, were identified to have a high level of expression in the biofilm of H. pylori. This is consistent with the biofilms of E. coli and Campylobacter(13,19).

The two-component system is a common signal transduction mechanism in bacterial responses to various stresses. This system is composed of two types of proteins, a histidine kinase and a response regulator. These two components exchange signals by phosphate transfer. Two-component systems are involved in the biofilm formation of E. coli, Pseudomonas aeruginosa and other bacteria (20,21). For example, the EnvZ/OmpR two-component system in E. coli promotes biofilm formation by regulating the expression of proteins associated with bacterial fimbriae to enhance its adhesion (22). In the present study, the response regulator Hp0166 was observed to be induced in H. pylori biofilms. A previous study also demonstrated that the Ars two-component system is involved in H. pylori responses to acid stress by regulating the expression of a variety of genes (23).

In addition, the current study also identified that the expression of other proteins changed in H. pylori biofilms. The expression of urease and arginase was upregulated. These two proteins generate ammonia in the process of amino acid metabolism, which is essential to biofilm formation to maintain the pH balance. The accessory protein of urease, ureE, is highly expressed in Staphylococcus aureus biofilms (24). Peptidoglycan is a predominant component of the cell walls of bacteria, UDP-N acetylene alcohol acid glucosamine reductase is an enzyme that is required in peptidoglycan biosynthesis. It catalyzes the reduction of UDPN GlcNAc to UDPN acetyl muramic acid (25). This protein was overexpressed in H. pylori biofilms and it is consistent with a previous study of Staphylococcus aureus(14). The function of peptidoglycan remains unclear, but it was observed that fracture of the peptidoglycan may promote the formation of bacterial biofilms (26). In addition, this study also showed that the expression of a number of other unknown proteins, including Hp1377 and Hp1029, changed in H. pylori biofilm formation. However, this also requires further study.

In the present study, H. pylori biofilm formation was induced successfully by serum starvation and its proteome was analyzed using proteomic methods. This study showed that ≥35 proteins are involved in biofilm formation. These proteins are associated with a number of types of biological functions, including flagellar movement, bacterial virulence, signal transduction and regulation. Such results showed that H. pylori biofilms are formed through multiple mechanisms involving numerous signal pathways. These findings may provide valuable information in understanding the survival mechanism of this bacterium in animals and humans.


This study was supported by grants from the National Natural Science Foundation of China (grant nos. 81171536 and 30800037) and the Science Foundation of Shandong Province, PR China (grant nos. ZR2009CZ001 and ZR2009CM002).



Donlan RM and Costerton JW: Biofilms: survival mechanisms of clinically relevant microorganisms. Clin Microbiol Rev. 15:167–193. 2002. View Article : Google Scholar : PubMed/NCBI


Moreno Y and Ferrús MA: Specific detection of cultivable Helicobacter pylori cells from wastewater treatment plants. Helicobacter. 17:327–332. 2012.


Braganra SM, Azevedo NF, Simões LC, Keevil CW and Vieira MJ: Use of fluorescent in situ hybridisation for the visualisation of Helicobacter pylori in real drinking water biofilms. Water Sci Technol. 55:387–393. 2007. View Article : Google Scholar : PubMed/NCBI


Cammarota G, Sanguinetti M, Gallo A and Posteraro B: Review article: biofilm formation by Helicobacter pylori as a target for eradication of resistant infection. Aliment Pharmacol Ther. 36:222–230. 2012.PubMed/NCBI


Carron MA, Tran VR, Sugawa C and Coticchia JM: Identification of Helicobacter pylori biofilms in human gastric mucosa. J Gastrointest Surg. 10:712–717. 2006.


Yang FL, Hassanbhai AM, Chen HY, Huang ZY, Lin TL, Wu SH and Ho B: Proteomannans in biofilm of Helicobacter pylori ATCC 43504. Helicobacter. 16:89–98. 2011.


Grande R, Di Giulio M, Bessa LJ, Di Campli E, Baffoni M, Guarnieri S and Cellini L: Extracellular DNA in Helicobacter pylori biofilm: a backstairs rumour. J Appl Microbiol. 110:490–498. 2011.


Testerman TL, Conn PB, Mobley HL and McGee DJ: Nutritional requirements and antibiotic resistance patterns of Helicobacter species in chemically defined media. J Clin Microbiol. 44:1650–1658. 2006. View Article : Google Scholar : PubMed/NCBI


Williams JC, McInnis KA and Testerman TL: Adherence of Helicobacter pylori to abiotic surfaces is influenced by serum. Appl Environ Microbiol. 74:1255–1258. 2008.


Reeser RJ, Medler RT, Billington SJ, Jost BH and Joens LA: Characterization of Campylobacter jejuni biofilms under defined growth conditions. Appl Environ Microbiol. 73:1908–1913. 2007.


Shao C, Zhang Q, Sun Y, et al: Helicobacter pylori protein response to human bile stress. J Med Microbiol. 57:151–158. 2008. View Article : Google Scholar


Shao C, Zhou Y, Sun Y, et al: Analysis of aztreonam-inducing proteome changes in nondividing filamentous Helicobacter pylori. Curr Microbiol 2012. 65:108–115. 2012. View Article : Google Scholar : PubMed/NCBI


Kalmokoff M, Lanthier P, Tremblay TL, et al: Proteomic analysis of Campylobacter jejuni 11168 biofilms reveals a role for the motility complex in biofilm formation. J Bacteriol. 188:4312–4320. 2006.


Wood TK, González Barrios AF, Herzberg M and Lee J: Motility influences biofilm architecture in Escherichia coli. Appl Microbiol Biotechnol. 72:361–367. 2006. View Article : Google Scholar : PubMed/NCBI


Tang Y, Guest JR, Artymiuk PJ and Green J: Switching aconitase B between catalytic and regulatory modes involves iron-dependent dimer formation. Mol Microbiol. 56:1149–1158. 2005. View Article : Google Scholar : PubMed/NCBI


Tang Y, Guest JR, Artymiuk PJ, Read RC and Green J: Post-transcriptional regulation of bacterial motility by aconitase proteins. Mol Microbiol. 51:1817–1826. 2004. View Article : Google Scholar : PubMed/NCBI


Backert S, Moese S, Selbach M, Brinkmann V and Meyer TF: Phosphorylation of tyrosine 972 of the Helicobacter pylori CagA protein is essential for induction of a scattering phenotype in gastric epithelial cells. Mol Microbiol. 42:631–644. 2001.PubMed/NCBI


Kvint K, Hosbond C, Farewell A, Nybroe O and Nyström T: Emergency derepression: stringency allows RNA polymerase to override negative control by an active repressor. Mol Microbiol. 35:435–443. 2000. View Article : Google Scholar : PubMed/NCBI


Schembri MA, Kjaergaard K and Klemm P: Global gene expression in Escherichia coli biofilms. Mol Microbiol. 48:253–267. 2003.


Li YH, Lau PC, Tang N, Svensäter G, Ellen RP and Cvitkovitch DG: Novel two-component regulatory system involved in biofilm formation and acid resistance in Streptococcus mutans. J Bacteriol. 184:6333–6342. 2002. View Article : Google Scholar : PubMed/NCBI


Otto K and Silhavy TJ: Surface sensing and adhesion of Escherichia coli controlled by the Cpx-signaling pathway. Proc Natl Acad Sci USA. 99:2287–2292. 2002.PubMed/NCBI


Prigent-Combaret C, Brombacher E, Vidal O, Ambert A, Lejeune P, Landini P and Dorel C: Complex regulatory network controls initial adhesion and biofilm formation in Escherichia coli via regulation of the csgD gene. J Bacteriol. 183:7213–7223. 2001. View Article : Google Scholar : PubMed/NCBI


Wen Y, Feng J, Scott DR, Marcus EA and Sachs G: Involvement of the HP0165-HP0166 two-component system in expression of some acidic-pH-upregulated genes of Helicobacter pylori. J Bacteriol. 188:1750–1761. 2006. View Article : Google Scholar : PubMed/NCBI


Resch A, Leicht S, Saric M, Pásztor L, Jakob A, Götz F and Nordheim A: Comparative proteome analysis of Staphylococcus aureus biofilm and planktonic cells and correlation with transcriptome profiling. Proteomics. 6:1867–1877. 2006.


Andres CJ, Bronson JJ, D’Andrea SV, et al: 4-Thiazolidinones: novel inhibitors of the bacterial enzyme MurB. Bioorg Med Chem Lett. 10:715–717. 2000. View Article : Google Scholar : PubMed/NCBI


Mercier C, Durrieu C, Briandet R, Domakova E, Tremblay J, Buist G and Kulakauskas S: Positive role of peptidoglycan breaks in lactococcal biofilm formation. Mol Microbiol. 46:235–243. 2002. View Article : Google Scholar : PubMed/NCBI

Related Articles

Journal Cover

December 2013
Volume 8 Issue 6

Print ISSN: 1791-2997
Online ISSN:1791-3004

Sign up for eToc alerts

Recommend to Library

Copy and paste a formatted citation
Shao, C., Sun, Y., Wang, N., Yu, H., Zhou, Y., Chen, C., & Jia, J. (2013). Changes of proteome components of Helicobacter pylori biofilms induced by serum starvation. Molecular Medicine Reports, 8, 1761-1766.
Shao, C., Sun, Y., Wang, N., Yu, H., Zhou, Y., Chen, C., Jia, J."Changes of proteome components of Helicobacter pylori biofilms induced by serum starvation". Molecular Medicine Reports 8.6 (2013): 1761-1766.
Shao, C., Sun, Y., Wang, N., Yu, H., Zhou, Y., Chen, C., Jia, J."Changes of proteome components of Helicobacter pylori biofilms induced by serum starvation". Molecular Medicine Reports 8, no. 6 (2013): 1761-1766.