PCNA interacts with hHus1/hRad9 in response to DNA damage and replication inhibition

Abstract

The hHus1 and several hRad proteins are involved in the control of DNA integrity checkpoints, although the mechanisms underlying these processes are unknown. Using a yeast two-hybrid system to detect protein-protein interactions, we found that human proliferating cell nuclear antigen (PCNA), a protein known to function in both DNA replication and repair, interacts with the human checkpoint-related protein Hus1 (hHus1). In human skin fibroblast cells, exposure to ionizing radiation of hydroxyurea triggers translocation of hHus1 from the cytosol to the nucleus, where it associates with PCNA as well as another checkpoint protein, hRad9. This nuclear translocation and the complex formation or hHus1 with PCNA and hRad9 correlate closely with changes in cell cycle distribution in response to radiation exposure. These results suggest that this multi-protein complex may be important for coordinating cell-cycle progression, DNA replication and repair of damaged DNA.

Main

Cell-cycle checkpoints play a critical role in the maintenance of genomic integrity by inhibiting progression through the cell cycle in the presence of damaged DNA or incomplete DNA replication (Hartwell and Kastan, 1994; Russell, 1998). The loss of genomic integrity can cause cancer and other genetic diseases (Hartwell and Kastan, 1994; Nojima, 1997). In fission yeast, a group of six checkpoint proteins, including Hus1, Rad1, Rad3, Rad9, Rad17, and Rad26, are required to block entry into mitosis when DNA replication is inhibited or DNA is damaged (al-Khodairy and Carr, 1992; al-Khodairy et al., 1994; Enoch et al., 1992; Rowley et al., 1992). Human homologs of all of these key checkpoint regulators, with the exception of Rad26, have been identified (i.e., hHus1, hRad1, ATR, hRad9, and hRad17; Cimprich et al., 1996; Hang and Lieberman, 2000; Kostrub et al., 1998; Lieberman et al., 1996; Parker et al., 1998; Udell et al., 1998), suggesting that the DNA integrity checkpoint pathway controlled by these proteins may be conserved from yeast to human. Many of these proteins can interact with each other through a complex network of homo- and heterodimers (Hang and Lieberman, 2000; Kondo et al., 1999; Kostrub et al., 1997; St. Onge et al., 1999; Volkmer and Karnitz 1999).

The S. pombe hus1 gene was originally isolated from a screen for fission yeast mutants capable of entering mitosis when DNA replication is blocked by hydroxyurea (HU) (Enoch et al., 1992). However, the hus1 mutants are not only abnormally sensitive to HU but also sensitive to radiation exposure, suggesting that many of these genes required for cell-cycle arrest in response to blocked DNA replication are also required for DNA damage checkpoint control. The Hus1 protein interacts with Rad1 and Rad9 (Caspari et al., 2000; Kondo et al., 1999; Kostrub et al., 1997), forming a large protein complex 450 kDa that is much bigger than an expected trimeric size of 150 kDa, implying a possibility that this multi-protein complex might contain other unidentified members. Similar interactions have been reported for hHus1, hRad1 and hRad9, strongly suggesting that these human homologs are functionally equivalent to their S. pombe counterparts. Interestingly, the reported findings that Hus1, Rad1 and Rad9 as well as their human homologs share regions of sequence similarity with the proliferating cell nuclear antigen (PCNA) (Aravind et al., 1999; Caspari et al., 2000; Thelen et al., 1999; Venclovas and Thelen, 2000) suggest that these checkpoint proteins may directly impinge on the DNA replication machinery. The biochemical mechanism by which these proteins regulate the DNA integrity checkpoint pathways, however, remains essentially unknown.

To identify cDNAs encoding proteins that are capable of interacting with human Hus1 (hHus1), we used a lexA-based yeast two-hybrid system (Golemis et al., 1994) for Matchmaker cDNA library screening. Strain EGY48 S. cerevisiae cells were transformed with plasmid pEG202 containing a full-length hHus1 subcloned in frame with the LexA open reading frame (ORF) and human cDNA libraries (fetal brain and Hela mixture) cloned into pJG4-5 (Clontech Lab, Inc). Candidate clones were isolated from yeast colonies formed on leucine-deficient agar plates with detectable β-galactosidase activity and retested by co-transformation with the bait expressing plasmids. In a screen of 1×107 independent transformants, we isolated a partial cDNA encoding the C-terminal portion (amino acids 83–261) of human PCNA. PCNA is a ring-shaped protein that encircles DNA and functions as a DNA sliding clamp for the replicative DNA polymerases (Kong et al., 1992; Krishna et al., 1994). It is an essential component for eukaryotic chromosomal DNA replication machinery and plays a critical role in DNA recombination, repair, and several other cellular processes (Kelman and Hurwitz, 1998; Stillman, 1994; Tsurimoto, 1999). The human PCNA ORF was amplified by the polymerase chain reaction (PCR) from a HepG2 cDNA library, and introduced into yeast cells to characterize further the interaction of PCNA and hHus1. As shown in Figure 1a, PCNA, whether fused to a LexA DNA-binding domain (LexA-PCNA) or a B42 transactivation domain (AD-PCNA), interacted with hHus1, as determined by a β-galactosidase activity assay. In contrast, LexA-PCNA protein failed to form two-hybrid interactions with AD-hRad1 and AD-hRad9, suggesting that their yeast homologs do not mediate or bridge the interaction of hHus1 with PCNA. The association between PCNA and hHus1 was specific as judged by the failure of these proteins to form two-hybrid interactions with the parental ‘empty’ vectors or unrelated proteins (not shown).

Figure 1
figure1

Specific interaction of hHus1 with PCNA. (a) S. cerevisiae EGY48 cells were transformed with various combinations of the pEG202 expression plasmids producing LexA DNA-binding domain (BD) fusion proteins and pJG4-5 plasmids encoding B42 transactivation domain (AD) fusion proteins (Estojak et al., 1995; Komatsu et al., 2000). Two-hybrid interactions between these fusion proteins were determined by using lacZ reporter genes under the control of lexA operators. Three independent transformants were tested for their ability to produce blue color on X-Gal plates containing either galactose (Gal) or glucose (Glu), to either induce or repress, respectively, the gal-1 promoters in the two-hybrid plasmids. The interaction between LexA-BAD and AD-Bcl-xL served as a positive control in the yeast two-hybrid assays (Komatsu et al., 2000). (b, c) Exponentially growing 293T cells in 10 ml of DMEM containing 5% FCS were transiently co-transfected with 5 μg pFlag-CMV2 encoding human PCNA and 5 μg pcDNA3-HA-hHus1 expression plasmid. Cells were lysed after 2 days in 0.65 ml of ice-cold lysis buffer (50 mM HEPES, pH 7.5, 150 mM NaCl, 10 mM NaF, 30 mM Na4P2O7, 1 mM EDTA, 1% Triton X-100) containing protease and phosphatase inhibitors (Komatsu et al., 2000) and immunoprecipitations were performed by incubating 0.2 ml of lysate with 20 μl of protein G-Sepharose preadsorbed with 5 μg of the indicated monoclonal antibodies. The resulting immune complexes were washed three times in the same buffer, and subjected to 12% SDS–PAGE/immunoblot analysis using antibodies specific for PCNA (Santa Cruz Biotech, Inc.) or hHus1 and detected by an enhanced chemiluminescence (ECL) system (Amersham). The anti-hHus1 polyclonal rabbit antiserum was generated by using a synthetic peptide corresponding to amino acids 215 to 233 of the hHus1 protein (NH2-ser-glu-ser-thr-his-glu-asp-arg-asn-val-glu-his-met-ala-glu-val-his-ile-asp-COOH). (d) 293T cells were transiently transfected with 10 μg pcDNA3-HA-hHus1 and immunoprecipitation was performed using monoclonal antibodies specific for PCNA, Bcl-xL or HA, followed by immunoblot analysis of the resulting immune complexes using anti-PCNA polyclonal antibody as a probe. (e) 293T cells were transiently transfected with 10 μg pFlag-hHus1 or parental pFlag-CMV2(CNTL) plasmid DNA. Immunoprecipitation was performed 2 days later using anti-Flag M2 monoclonal antibody, followed by immunoblot analysis of the resulting immune complexes with anti-PCNA polyclonal antiserum

Next, we determined whether the interaction of PCNA and hHus1 could occur in mammalian cells. 293T cells were transiently co-transfected with expression plasmids encoding Flag-tagged PCNA and hemagglutinin (HA)-tagged hHus1, and immunoprecipitates were prepared using anti-Flag and anti-HA monoclonal antibodies, or with anti-Myc antibody as a negative control, and subjected to SDS–PAGE/immunoblot assays using anti-sera specific for PCNA or hHus1. Under these conditions, PCNA was co-immunoprecipitated with hHus1 (Figure 1b,c). In addition, endogenous PCNA could be co-immunoprecipitated with HA-tagged or Flag-tagged hHus1 from 293T cells expressing HA-hHus1 (Figure 1d) or Flag-hHus1 (Figure 1e), but not with the unrelated protein Bcl-xL (Figure 1d). The specificity of the association between hHus1 and PCNA was further shown by the failure of the anti-Flag (Figure 1e) or anti-HA (not shown) antibodies to coimmunoprecipitate with PCNA in 293T cells transfected with control empty vector.

To investigate the cellular localization of hHus1, we expressed hHus1 as a green fluorescent protein (GFP) fusion protein in 293 epithelial cells. Analysis of 293 cells by fluorescence microscopy after transfection with an expression plasmid encoding the GFP-hHus1 protein revealed a diffuse, mostly cytosolic staining (Figure 2b). In contrast, when the GFP protein was expressed without the appended hHus1 cDNA insert, a mostly diffuse pattern of fluorescence was found throughout the cell (Figure 2a). Similar results were obtained when using the Flag-tagged hHus1 protein (Figure 2d). The mostly cytosolic distribution of hHus1 in 293 cells was further shown by immunofluorescence staining of the endogenous hHus1 protein with anti-hHus1 specific antibody (Figure 2f). The staining of control vector-transfected cells (Figure 2c) or using normal rabbit serum (Figure 2e) confirmed the specificity of these results. In contrast, the murine Hus1 protein was detected mostly in the nucleus of Balb/c 3T3 cells using the same rabbit antiserum (not shown). Moreover, the S. pombe Hus1 protein has been reported to be a nuclear protein, but this nuclear localization of Hus1 requires the presence of Rad9 and Rad17 (Caspari et al., 2000). These results indicate that the cellular localization of hHus1 may differ from those found in murine and yeast cells, but we do not exclude the possibility that it is also mammalian cell type dependent.

Figure 2
figure2

The hHus1 protein localizes in the cytosol of 293 cells. 293 cells were transiently transfected with pEGFP-hHus1 (b), pFlag-hHus1 (d), parental pEGFP-C2 (a) and pFlag-CMV2 (c) plasmid DNA, or left untransfected (e,f). In (cf), cells were fixed in methanol at −20°C for 10 min and permeabilized in 0.5% Triton X-100 at room temperature for 10 min. After preblocking in 5% (v/v) skim milk and 3% (v/v) BSA for 1 h, cells were incubated with anti-Flag M2 mouse monoclonal antibody (c and d; 50 μg/ml, Sigma), anti-Hus1 rabbit polyclonal antiserum (f; 1 : 100) or normal rabbit serum (e), followed by rhodamine-conjugated goat anti-mouse IgG (c and d; 1 : 100, Chemicon) or fluorescein isothiocyanate (FITC)-conjugated goat anti-rabbit IgG (e and f), and analysed using fluorescence microscopy

Flow2000 cells are a precursor diploid strain of human skin fibroblasts that undergoes rapid cell cycle arrest in the absence of apoptosis following exposure to ionizing radiation (IR), thus providing a model commonly used to investigate the cellular response to radiation. We used Flow2000 cells to monitor changes in the intracellular locations of endogenous PCNA and hHus1 proteins in response to DNA damage and inhibition of DNA synthesis. Immunofluorescence analysis of methanol-fixed exponentially growing cells revealed a punctate nuclear immunostaining of PCNA protein (Figure 3a), consistent with a localization confined to the sites of DNA replication (Bravo and Macdonald-Bravo, 1987). Cells labeled with anti-hHus1 antibody showed that hHus1 was located throughout the cell (Figure 3b), and in general failed to co-localize with PCNA (Figure 3c). In contrast, exposure of Flow2000 cells to the DNA synthesis inhibitor hydroxyurea (HU) or IR resulted in re-localization of hHus1 to the nucleus and co-localization with PCNA (Figure 3e,f,h,i). These results were further confirmed by co-immunoprecipitation experiments, in which the protein complex of endogenous hHus1 and endogenous PCNA could be detected when DNA was damaged or the replication of DNA was blocked (Figure 3j,k). Nuclear translocation of hHus1 and association with PCNA were evident within 4 h and persisted for at least 12 h after IR exposure (Figure 3h,i,k). However, a constitutive interaction between hHus1 and PCNA was observed when both proteins were overexpressed (Figure 1), thus suggesting that enforced expression may bypass the in vivo control of endogenous hHus1 and PCNA shown in Figure 3.

Figure 3
figure3

DNA damage or replication inhibition induced hHus1 translocation into the nucleus and binding to PCNA. (ai) Logarithmically growing Flow2000 cells were cultured without (ac) or with 1 mM HU for 12 h (df), or exposed to 8 Gy IR followed by 5 h culture (g to i). Cells were fixed and stained with monoclonal mouse anti-PCNA (1 : 20; Santa Cruz Biotech, Incorporated) and polyclonal rabbit anti-hHus1 (1 : 100), followed by rhodamine-conjugated goat anti-mouse IgG and FITC-conjugated goat anti-rabbit IgG (1 : 100; Chemicon), and analysed using a confocal laser-scanning microscope (Carl Zeiss). Control immunoglobulin G1 was also used for confirming the specificity of immunostaining (not shown). (j) Flow2000 cells were cultured for 12 h with or without 1 mM HU, as indicated (+) or (−), respectively. Cell lysates were normalized for protein content and immunoprecipitations were performed using anti-hHus1 antiserum or normal rabbit antiserum (CNTL) as a negative control. The resulting immune complexes were subjected to SDS–PAGE and immunoblot analysis with specific antibodies for PCNA or hHus1. In addition to immune complexes, the total lysates (50 μg) were analysed directly, confirming no significant changes in PCNA and hHus1 expression after HU treatment. (k) Logarithmically growing Flow2000 cells were exposed to 8 Gy IR or left untreated (−). At various times after ionizing radiation exposure, cell lysates were made and anti-hHus1 immunoprecipitates were prepared, followed by immunoblot analysis using antibodies specific for PCNA or hHus1. The cell lysates (50 μg) were also applied directly to the gel without immunoprecipitation, confirming that ionizing radiation had no effect on protein expression levels of PCNA and hHus1

It has been shown that hHus1 can interact with hRad9 and hRad1, forming a DNA damage responsive protein complex (Hang and Lieberman, 2000; St. Onge et al., 1999; Volkmer and Karnitz, 1999). We explored the possibility that hRad9 might reside in the same complex formed by PCNA and hHus1 in the presence of incomplete DNA replication or DNA damage. Consistent with previous reports (Komatsu et al., 2000; St. Onge et al., 1999), hRad9 protein was detected mostly in the nucleus, and co-localized with PCNA in Flow2000 cells as measured by indirect immunofluorescence microscopy (Figure 4a,c). In contrast, the co-localization of hRad9 and hHus1 was barely detected in logarithmically growing Flow2000 cells (Figure 4f), although association was detected when cells were exposed to HU or IR (Figure 4i,l), implying that hHus1 forms a complex with PCNA and hRad9 when DNA replication is inhibited or DNA is damaged. We have previously reported that treating Hela cells with methyl methanesulphonate (MMS) resulted in perinuclear localization of Flag-hRad9, co-localization with GFP-Bcl-2, and apoptosis induction (Komatsu et al., 2000). In Flow2000 cells, IR failed to induce re-localization of endogenous hRad9 (Figure 4j) and subsequent apoptosis (not shown). It is possible that hRad9 controls several downstream pathways: when damaged DNA is potentially repairable, hRad9 forms a nuclear protein complex with hHus1, hRad1, and possibly PCNA to arrest the cell cycle, providing extra time for repair before cells re-enter the cell cycle and progress through critical phases; when cells are severely damaged, hRad9 could migrate out of the nucleus to trigger apoptosis through binding to Bcl-2.

Figure 4
figure4

hRad9 co-localizes with PCNA and hHus1. (ac) Logarithmically growing Flow2000 cells were fixed and double-stained with mouse anti-hRad9 monoclonal antibody (1 : 50; Imgenex) and rabbit anti-PCNA polyclonal antiserum (1 : 100; Santa Cruz Biotech, Inc.), followed by rhodamine-conjugated goat anti-mouse IgG and FITC-conjugated goat anti-rabbit IgG secondary antibodies. (dl) Flow2000 cells were cultured as described in Figure 3 and stained with antibodies specific for hRad9 (d, g and j) or hHus1 (e, h and k), followed by application of either rhodamine- or FITC-conjugated secondary antibodies, to detect hRad9 or hHus1, respectively

Activation of the DNA-damage checkpoint transiently delays cell cycle progression by slowing movement through S phase and arresting cells in the G1 and G2/M phases. We analysed propidium-iodide-stained logarithmically growing Flow2000 cells by flow cytometry to determine the alterations in the distribution of DNA content after ionizing radiation. As expected, DNA damage triggered cells to begin to accumulate in S phase of the cell cycle at 3 h post-irradiation, reaching a maximum at 6 h, and slowly passing through S phase, then arresting in G0/G1 and G2/M phase 12 h after ionizing radiation exposure (Table 1). This correlates closely with the nuclear translocation of hHus1 and its interaction with PCNA and hRad9 in response to DNA damage (Figures 3 and 4), suggesting that the PCNA/hHus1/hRad9 complex may contribute to coordination of cell cycle progression and DNA replication.

Table 1 Effects of ionizing radiation on the cell cycle distribution of Flow2000 cells

Rad17 has been reported to contain domains of sequence similarity with the replication factor C (RFC) (Griffiths et al., 1995), a replication fork associated protein that is required for the loading and unloading of PCNA onto and off the DNA (Waga and Stillman, 1998). In fission yeast, Rad17 forms a protein complex that is distinct from the Rad9/Hus1/Rad1 complex, but it is necessary to locate this protein complex in the nucleus (Caspari et al., 2000). Thus, it will be interesting to determine whether hRad17 functions as a molecular matchmaker to generate the linkage between the hRad9/hHus1/PCNA complex and DNA in replication forks or sites of DNA repair.

PCNA can interact with a large number of proteins, including DNA polymerases δ and ε, RFC, FEN1 nuclease, DNA ligase I, MLH1/MSH2, p21WAF1, p57, cyclin D, and the damage induced protein Gadd45 (Kelman and Hurwitz, 1998; Stillman, 1994; Tsurimoto, 1999), suggesting that it is an adapter molecule capable of linking cellular factors to their DNA substrates for DNA replication and repair, cell-cycle progression or cellular differentiation. It has been reported that mutations in PCNA increased the sensitivity of yeast cells to DNA damage (Ayyagari et al., 1995; Eissenberg et al., 1997), indicating involvement of PCNA in cellular DNA replication/repair. In addition, PCNA has been shown to participate in a replication checkpoint that controls entry into mitosis (Waseem et al., 1992), suggesting that the formation of the PCNA/hHus1/hRad9 complex might mediate this checkpoint. Although the precise roles of this complex have to be elucidated, it may contribute to coordination of cell cycle progression and DNA replication by decreasing the efficiency of replication fork movement, providing extra time for DNA repair or recruitment of replication factors.

References

  1. al-Khodairy F and Carr AM. . 1992 EMBO J. 11: 1343–1350.

  2. al-Khodairy F, Fotou E, Sheldrick KS, Griffiths DJ, Lehmann AR and Carr AM. . 1994 Mol. Biol. Cell. 5: 147–160.

  3. Aravind L, Walker DR and Koonin EV. . 1999 Nucleic Acids Res. 27: 1223–1242.

  4. Ayyagari R, Impellizzeri KJ, Yoder BL, Gary SL and Burgers PM. . 1995 Mol. Cell Biol. 15: 4420–4429.

  5. Bravo R and Macdonald-Bravo H. . 1987 J. Cell Biol. 105: 1549–1554.

  6. Caspari T, Dahlen M, Kanter-Smoler G, Lindsay HD, Hofmann K, Papadimitriou K, Sunnerhagen P and Carr AM. . 2000 Mol. Cell Biol., 20: 1254–1262.

  7. Cimprich KA, Shin TB, Keith CT and Schreiber SL. . 1996 Proc. Natl. Acad. Sci. USA 93: 2850–2855.

  8. Eissenberg JC, Ayyagari R, Gomes XV and Burgers PM. . 1997 Mol. Cell Biol. 17: 6367–6378.

  9. Enoch T, Carr AM and Nurse P. . 1992 Genes Dev. 6: 2035–2046.

  10. Estojak J, Brent R and Golemis EA. . 1995 Mol. Cell. Biol. 15: 5820–5829.

  11. Golemis EA, Gyuris J and Brent R. . 1994 In: Current Protocols in Molecular Biology. J. Wiley & Sons, Inc., pp13.14.1–13.14.17.

  12. Griffiths DJ, Barbet NC, McCready S, Lehmann AR and Carr AM. . 1995 EMBO J. 14: 5812–5823.

  13. Hang H and Lieberman HB. . 2000 Genomics 65: 24–33.

  14. Hartwell LH and Kastan MB. . 1994 Science 266: 1821–1828.

  15. Kelman Z and Hurwitz J. . 1998 Trends Biochem. Sci. 23: 236–238.

  16. Komatsu K, Miyashita T, Hang H, Hopkins KM, Zheng W, Cuddeback S, Yamada M, Lieberman HB and Wang HG. . 2000 Nat. Cell. Biol. 2: 1–6.

  17. Kondo T, Matsumoto K and Sugimoto K. . 1999 Mol. Cell Biol. 19: 1136–1143.

  18. Kong XP, Onrust R, O'Donnell M and Kuriyan J. . 1992 Cell 69: 425–437.

  19. Kostrub CF, al-Khodairy F, Ghazizadeh H, Carr AM and Enoch T. . 1997 Mol. Gen. Genet. 254: 389–399.

  20. Kostrub CF, Knudsen K, Subramani S and Enoch T. . 1998 EMBO J. 17: 2055–2066.

  21. Krishna TS, Kong XP, Gary S, Burgers PM and Kuriyan J. . 1994 Cell 79: 1233–1243.

  22. Lieberman HB, Hopkins KM, Nass M, Demetrick D and Davey S. . 1996 Proc. Natl. Acad. Sci. USA 93: 13890–13895.

  23. Nojima H. . 1997 Hum. Cell 10: 221–230.

  24. Parker AE, Van de Weyer I, Laus MC, Oostveen I, Yon J, Verhasselt P and Luyten WH. . 1998 J. Biol. Chem. 273: 18332–18339.

  25. Rowley R, Subramani S and Young PG. . 1992 EMBO J. 11: 1335–1342.

  26. Russell, P. . 1998 Trends Biochem. Sci., 23: 399–402.

  27. St. Onge RP, Udell CM, Casselman R and Davey S. . 1999 Mol. Biol. Cell 10: 1985–1995.

  28. Stillman B. . 1994 Cell 78: 725–728.

  29. Thelen MP, Venclovas C and Fidelis K. . 1999 Cell 96: 769–770.

  30. Tsurimoto T. . 1999 Front Biosci. 4: D849–D858.

  31. Udell CM, Lee SK and Davey S. . 1998 Nucleic Acids Res. 26: 3971–3976.

  32. Venclovas C and Thelen MP. . 2000 Nucleic Acids Res. 28: 2481–2493.

  33. Volkmer E and Karnitz LM. . 1999 J. Biol. Chem. 274: 567–570.

  34. Waga S and Stillman B. . 1998 Annu. Rev. Biochem. 67: 721–751.

  35. Waseem NH, Labib K, Nurse P and Lane DP. . 1992 EMBO J. 11: 5111–5120.

Download references

Acknowledgements

We thank Nikola Valkov for assistance with fluorescence confocal microscopy, the Molecular Biology, Flow Cytometry and Molecular Imaging core facilities at the Moffitt Cancer Center and Research Institute, and the NIH (CA82197, CA72694, GM52493, and CA68446) for support.

Author information

Affiliations

Authors

Corresponding author

Correspondence to Hong-Gang Wang.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Komatsu, K., Wharton, W., Hang, H. et al. PCNA interacts with hHus1/hRad9 in response to DNA damage and replication inhibition. Oncogene 19, 5291–5297 (2000). https://doi.org/10.1038/sj.onc.1203901

Download citation

Keywords

  • hRad9
  • hHus1
  • PCNA
  • DNA damage
  • cell-cycle checkpoints

Further reading