Rising cyclin-CDK levels order cell cycle events.
Rising Cyclin-CDK Levels Order Cell Cycle Events Catherine Oikonomou, Frederick R. Cross* Laboratory of Cell Cycle Genetics, The Rockefeller University, New York, New York, United States of America Abstract Background: Diverse mitotic events can be triggered in the correct order and time by a single cyclin-CDK. A single regulator could confer order and timing on multiple events if later events require higher cyclin-CDK than earlier events, so that gradually rising cyclin-CDK levels can sequentially trigger responsive events: the ‘‘quantitative model’’ of ordering. Methodology/Principal Findings: This ‘quantitative model’ makes predictions for the effect of locking cyclin at fixed levels for a protracted period: at low cyclin levels, early events should occur rapidly, while late events should be slow, defective, or highly variable (depending on threshold mechanism). We titrated the budding yeast mitotic cyclin Clb2 within its endogenous expression range to a stable, fixed level and measured time to occurrence of three mitotic events: growth depolarization, spindle formation, and spindle elongation, as a function of fixed Clb2 level. These events require increasingly more Clb2 according to their normal order of occurrence. Events occur efficiently and with low variability at fixed Clb2 levels similar to those observed when the events normally occur. A second prediction of the model is that increasing the rate of cyclin accumulation should globally advance timing of all events. Moderate (,2-fold) overexpression of Clb2 accelerates all events of mitosis, resulting in consistently rapid sequential cell cycles. However, this moderate overexpression also causes a significant frequency of premature mitoses leading to inviability, suggesting that Clb2 expression level is optimized to balance the fitness costs of variability and catastrophe. Conclusions/Significance: We conclude that mitotic events are regulated by discrete cyclin-CDK thresholds. These thresholds are sequentially triggered as cyclin increases, yielding reliable order and timing. In many biological processes a graded input must be translated into discrete outputs. In such systems, expression of the central regulator is likely to be tuned to an optimum level, as we observe here for Clb2. Citation: Oikonomou C, Cross FR (2011) Rising Cyclin-CDK Levels Order Cell Cycle Events. PLoS ONE 6(6): e20788. doi:10.1371/journal.pone.0020788 Editor: Daniel Lew, Duke University Medical Center, United States of America Received April 15, 2011; Accepted May 9, 2011; Published June 10, 2011 Copyright: ß 2011 Oikonomou, Cross. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: C.O. and F.C. were supported by NIH grant GM47238. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. * E-mail: firstname.lastname@example.org Introduction In budding yeast, a single essential Cyclin Dependent Kinase (CDK) is alternately activated by nine cyclins to trigger the major events of the cell cycle. These cyclins are differentially expressed, inhibited, and degraded, and their temporal order of activity contributes to the ordering of cell cycle events [1,2]. Within a given portion of the cell cycle, however, multiple events can be regulated by a single cyclin-CDK complex. Clb2, in the absence of the other mitotic cyclins, can promote all essential mitotic events with near-wild-type (WT) efficiency [3,4] (Fig. S1). Moreover, this simplified system likely reflects the ancestral eukaryotic cell cycle control system, given strong evidence for gene and genome duplication events leading to the extant cyclin diversity [5,6]. Importantly, even when driven by a single cyclin-CDK, mitotic events are temporally separated and exhibit a stereotyped order. Specifically, growth is depolarized, the spindle forms, and then elongates. In the absence of unique cyclin-CDK activators, what determines the order and timing of such mitotic events? Checkpoints  that halt mitotic progression are non-essential in budding yeast and largely do not contribute to normal cell cycle timing [8,9]. Mechanistic coupling, in which a later event is structurally dependent upon completion of an earlier event , is limited to events that involve the same structure, such as spindle formation and elongation. Alternatively, later events may require higher CDK activity levels than earlier events, a possibility known as the ‘‘quantitative model’’ of ordering . Recent work using a monomeric engineered cyclin-CDK in fission yeast has provided evidence that the ‘‘quantitative model’’ promotes the ordering of S-phase and mitosis in that organism. The cyclin-CDK was engineered so that it could be inhibited by an ATP analog. The authors found concentrations of analog that permitted S-phase but not mitosis, suggesting a higher cyclin-CDK threshold for the later phase . Two recent studies of cyclinB1CDK1 activation dynamics in HeLa cells and extracts have also provided support for this quantitative cyclin-CDK activity level model. It was shown that in prophase, later events require more cyclinB1-CDK1 activity than earlier events . It was also shown, in vitro, that later-acting cyclinB1-CDK1 substrates require higher levels of cyclinB1 for phosphorylation than earlier-acting substrates . Mitosis in budding yeast provides an ideal in vivo, organismal system to expand these studies. Clb2 can control all mitotic events, and the Wee1 and Cdc25 positive feedback loops that control CDK activation in higher eukaryotes are not essential in budding yeast , greatly simplifying the experimental system. In addition, genetic tools allow us to titrate cyclin levels, in vivo, within the endogenous range. This titration, in combination with single-cell timelapse imaging, allows us to quantitatively measure PLoS ONE | www.plosone.org 1 June 2011 | Volume 6 | Issue 6 | e20788 Rising Cyclin-CDK Levels Order Cell Cycle Events and compare thresholds for cyclin-CDK control of individual mitotic events. To determine whether the timing of mitotic events is normally controlled by Clb2-CDK level, we quantitatively measured the Clb2 requirements for three mitotic events – growth depolarization, spindle formation, and spindle elongation. We address whether these events exhibit thresholds in response, compare Clb2 requirements to the levels of Clb2 in freely cycling cells when these events occur, and analyze the effects of premature Clb2 accumulation on the timing of these events. Results A System to Measure Mitotic Cyclin-CDK Requirements There are four, largely redundant, mitotic cyclins in budding yeast: Clb1-4. Clb2, in the absence of the other three, can successfully promote all essential mitotic events, with near-WT timing (Fig. S1) [3,4]. We wanted to develop a system to measure the Clb2-CDK requirements for individual mitotic events. To do this, we constructed a clb1,3,4D GALL:CLB2 strain, in which CLB2 is the sole source of mitotic cyclin, and is under the control of an attenuated galactose-inducible (GALL) promoter [16,17]. We added a construct encoding a fusion of GAL4 (the activator of the GALL promoter) with a mammalian mineralocorticoid receptor . CLB2 expression was now dependent upon the presence of an exogenous hormone, deoxycorticosterone (DOC), so experiments could be carried out in a single carbon source, and the expression level achieved was within the physiological range for CLB2 (Fig. 1A) . To measure the Clb2 concentration in single cells, we assayed the fluorescence intensity of YFP-tagged Clb2, which is fully functional (B. Drapkin, personal communication; data not shown). Clb2 is predominantly localized to the nucleus , and nuclear size is tightly correlated with overall cell size , so we used a histone H2B-mCherry fusion to mark the nucleus, and measured the mean YFP intensity within this mask (with unlabeled cell background subtracted) to estimate total Clb2YFP concentration per cell. To allow cells to be incubated for long periods with a titrated level of Clb2, we prevented its degradation by turning off the Anaphase Promoting Complex (APC) activator CDC20 using a methionine-repressible promoter (MET3). Clb2YFP pulses induced in the presence of methionine (and therefore in the absence of CDC20 expression) were stable for at least two hours (Fig. 1B). The overall experimental protocol was the following (shown in Fig. 1C): 1. proliferate cells in the presence of DOC; 2. deplete Clb2-YFP by washing out the hormone, arresting cells prior to mitosis; deplete Cdc20 by adding methionine to the media (‘‘block’’); 3. give a pulse of Clb2-YFP by adding DOC and subsequently washing it out (‘‘induce’’); 4. use timelapse microscopy to correlate cell fate with mean backgroundsubtracted nuclear Clb2-YFP intensity in single cells, normalized to peak Clb2 expression in cycling cells . Previous studies have validated this quantification and shown the ‘proportion of peak’ units to have physiological meaning [19,22]. Clb2 was present at low but detectable levels in blocked, uninduced cells. This may reflect inheritance of Clb2 from the cell cycle before the block, or basal expression from the GALL:CLB2 construct. In any event, this level of Clb2 is insufficient to drive entry into mitosis, as .90% of these cells stably failed to make a spindle for at least two hours in the absence of Clb2 induction. We are using Clb2 protein level as a surrogate for Clb2-CDK activity. However, the Clb2-CDK-inhibitory kinase Swe1 is present and potentially active in these cells. On the other hand, Swe1 activity might not be expected to be high, since these cells are budded, so the ‘morphogenetic checkpoint’ should not activate Swe1 . To directly address the potential regulation of Clb2CDK activity by Swe1, we measured Clb2-CDK kinase activity throughout the experimental protocol, using SWE1 and swe1D cells. We found that the presence of Swe1 approximately halves the kinase activity of Clb2-CDK for about the first 30 minutes following the expression pulse (Fig. S2A). To compare this to the situation in normally cycling cells, we used alpha factor to synchronize WT and swe1D strains in G1, and measured the Clb2CDK kinase activity following release. We observed that Swe1 lowers Clb2-CDK kinase activity by a similar factor, again for approximately the first 30 minutes of Clb2 expression, corresponding to early mitosis (Fig. S2B). This indicates that our expression system faithfully recreates physiological conditions of transient kinase inhibition. Since the kinase assay is a bulk population measurement, we do not know if the early 2-fold reduction in Clb2 kinase is uniform across the population. We return to this issue below. To keep the experimental system as close as possible to physiological conditions in the cell, we conducted most subsequent experiments in SWE1 strains, with Clb2 levels reasonably reflecting Clb2-associated kinase activity. Clb2-CDK can also be stoichiometrically inhibited by Sic1. However, Sic1 is likely absent in the Clb-depleted cells in our protocol, since they express a high level of the G1 cyclin Cln2 , which promotes efficient Sic1 degradation . At later times in the SWE1 timecourses and throughout the swe1D timecourses, the activity of Clb2-associated kinase closely paralleled Clb2 protein levels. Therefore, it is unlikely that Sic1 inhibition is significantly regulating Clb2-CDK activity in these experiments. Sic1 is completely degraded by mid-cell-cycle  and if it were inhibiting Clb2-Cdk earlier, we should observe a change in specific activity later when inhibition is lifted. Ordering of Mitotic Events by Cyclin-CDK Level Using alpha-factor synchronization of a clb1,3,4D strain, we determined that Clb2 protein level, and associated Clb2-CDK kinase activity, ramped up over a period of about 40 minutes (Fig. 1D). Thus, if different activity levels promoted different events, these events could be significantly separated in time. We chose to measure the Clb2 requirement for three temporally separated mitotic events: depolarization of growth, spindle formation, and spindle elongation. In CLB2WT cells synchronized in G1 with alpha factor and released, growth depolarization (see Materials and Methods for details of measurement) occurs, on average, 20 minutes before spindle formation (Fig. S3); spindle formation occurs approximately 10 minutes before spindle elongation (anaphase). Importantly, all of these events occur as Clb2 levels are steadily increasing. Therefore, if the different events are induced by sequentially higher Clb2 levels, the gradual rise of Clb2 could in principle provide a simple ordering mechanism. Growth depolarization: regulation of an autonomous oscillator In budding yeast, cellular growth is polarized early in the cell cycle. At the time of bud initiation, all growth is focused to the bud tip, resulting in initial formation of an elongated bud with actin polarized to the bud tip (‘polarized growth’). Later in the cell cycle, bud growth and actin filaments are depolarized in a Clb-CDK dependent manner, resulting in rounded bud growth (‘isotropic growth’) . We used two metrics to assay polarized growth: localization of the Spa2 ‘polarisome’ component to the bud tip , and the rate of increase of bud length (which is significantly higher during polarized growth). Interestingly, in the absence of mitotic Clb-CDK activity, we observed that growth was not continuously polarized but rather exhibited alternating cycles of PLoS ONE | www.plosone.org 2 June 2011 | Volume 6 | Issue 6 | e20788 Rising Cyclin-CDK Levels Order Cell Cycle Events PLoS ONE | www.plosone.org 3 June 2011 | Volume 6 | Issue 6 | e20788 Rising Cyclin-CDK Levels Order Cell Cycle Events Figure 1. Experimental system to measure Clb2-CDK requirements. A Clb2-YFP protein from populations of cells pulsed with 5mM DOC for 109 or 159 compared to Clb2-YFP protein levels from clb2D cells, unpulsed cells of the experimental strain, and the peak expression level of clb1,3,4D cells synchronized with alpha factor and released. Below, Clb2-YFP levels are normalized to Pgk1 (which is not cell-cycle regulated) and expressed as a fraction of the peak value. B Mean Clb2-YFP intensity values in single cells following a pulse of DOC. Each colored trace represents a different cell. Approximately the first 40 minutes correspond to the maturation time of the fluorescent protein. C Schematic of experimental protocol for assaying mitotic Clb-CDK requirements for mitotic events. clb1,3,4D cells expressing Clb2-YFP in response to DOC are grown asynchronously in the presence of DOC (1). DOC is washed out, arresting cells through Clb2 depletion with replicated DNA and unseparated SPBs (2). Methionine is added to turn off CDC20 expression (and thus Clb2-YFP degradation). Cells are given a pulse of DOC (3) and followed by timelapse microscopy (4). Execution of mitotic events is correlated with stable Clb2-YFP level in single cells. D Clb2 is normally active for at least 40 minutes per cell cycle. clb1,3,4D cells were synchronized with alpha factor and released. Clb2 protein and kinase activity levels were assayed as shown. Quantification of blots is shown, with values normalized to Pgk1 and expressed as a fraction of the peak value at 709. doi:10.1371/journal.pone.0020788.g001 polarized and depolarized growth, each lasting approximately 40 minutes (Fig. 2A,B; Video S1) (polarized, 40.1615.89; depolarized, 39.0613.09). Due to the tight correlation of our two metrics, we assayed only bud length increase in subsequent experiments. These findings are consistent with previous findings suggesting an intrinsic mitotic cyclin-independent oscillator controlling cell polarization and budding . We pulsed cells with Clb2-YFP and tracked polarized growth to determine the response of this polarized growth oscillator as a function of Clb2-YFP level. Although there is significant variability in the measured response to low Clb2 levels in this assay, we cannot reliably assign it to either experimental error or true variability in the biological response because polarized growth is periodic in the absence of any Clb2. Since we do not know where in the polarized growth cycle any individual cell was at the time of the Clb2-YFP pulse, considerable variability in the assay is expected, even if the real response to a given Clb2 level were completely deterministic. However, we clearly saw that Clb2-YFP exerted dose-dependent effects: the higher the Clb2-YFP level, the less time a cell exhibited polarized growth (Fig. 2C). Clb2-CDK appeared to decrease both the frequency of the cycle, and the duration of the polarized period within each cycle. Remarkably little Clb2-YFP was sufficient to depolarize growth; 50% of cells completely depolarized their growth with 0.12 peak Clb2-YFP. In synchronized WT cells, depolarization occurs when cells contain approximately 0.3 peak Clb2-YFP (Figs. 1D, S3); consistently, in Clb2-pulsed cells, the same level induces 75% of cells to completely depolarize their growth. Recently, we presented evidence that a free-running oscillator controlling nucleolar localization of the Cdc14 phosphatase was entrained to occur once per cell cycle by mitotic cyclin-CDKdependent modulation of its frequency through a phase-locking mechanism . The finding that Clb2-CDK activity could also modulate the frequency of the bud growth oscillator suggested that phase-locking might similarly constrain this oscillator to once per cell cycle. Indeed, a simple phase-locking model confirm the involvement of PERK/ATF4/DDIT3 ER stress pathway in JC011 mediated PSC-specific cytotoxicity. siRNA knockdown was carried out in NCCIT cells for maximum transfection efficiency and confirmation of transcript knockdown was determined by qRT-PCR. Briefly, NCCIT cells were allowed to recover for a period of 24 hrs following siRNA knockdown and treated with JC011. Cell viability analysis indicated that ATF4 knockdown successfully attenuated JC011 cytotoxicity in NCCIT cells (P,0.05) thereby confirming the role of PERK/ATF4/ DDIT3 ER stress in JC011 induced PSC cytotoxicity. DDIT3 knockdown NCCIT cells also showed a similar attenuated cytotoxic response when challenged with JC011 but these data were not statistically significant (Fig. 7). This result could perhaps be due to other components of PERK/ATF4/DDIT3 ER stress pathway which may potentially substitute for loss of DDIT3 activity. ROS levels determined by DCHF-DA FACS were used as surrogate readout to confirm knockdown of ATF4 and DDIT3. ATF4 knockdown resulted in a recovery of ROS levels comparable to untreated controls while DDIT3 knockdown resulted in no significant ROS recovery (Figure S9 in File S1) Discussion Considerable progress has been achieved in establishing optimum conditions to propagate and differentiate PSCs into a variety of lineages of functional specialized cells for human cell replacement therapies. These advancements are further underscored by recent U. S Food and Drug Administration approval of clinic trials to use PSC-derived cells to treat spinal cord injury and macular degeneration [24,25]. Nonetheless, the teratoma risk associated with contaminating PSCs in differentiated cell populations still remains. This problem is further aggravated by the intrinsic propensity for some PSC derivatives to undergo dedifferentiation in situ [4,11]. Even if ensuing data from clinical trials supports the efficacy and safety of the PSC-based therapies, the teratoma risk might have to be continuously monitored when PSC-based therapies are routinely applied in a clinical setting. Antibody-based solutions to remove undifferentiated PSCs in vitro are currently popular. The high specificity of an antibody towards a particular antigen also has its drawbacks as any antibody-based strategy could in theory be limited by the specificity of the expression of its associated surface antigen. It is well documented that PSC cultures comprise of a mixed population of cells with slightly different phenotypes . Consequently, the undifferentiated cells from one sample may not express similar groups of surface antigens; this would necessitate the use of multiple combinations of antibodies for any antibodybased enrichment strategy. In addition, Fluorescence Activated Cell Sorting (FACS) or Magnetic Activated Cell Sorting (MACS) based enrichment protocols require single-cell suspensions. Manipulation of single cells may not be appropriate in all tissue engineering scenarios especially when complex 2-D or 3-D multilayer cell constructs are the final product for transplantation. It is also difficult to envisage the use of an antibody as a preventive drug in an established clinical transplantation regime where patients may require administration of a drug for a period of time following the transplantation procedure to prevent teratoma formation. Small molecules which are specifically cytotoxic to PSCs and non-toxic to normal specialized cells could be a better solution in these scenarios. Small molecules are cheaper and easier to synthesize. They also show favourable pharmacokinetic profiles and elaborate FACS or MACS set-ups and protocols are also not required. While this manuscript was in preparation, a high throughput screen study of over 50,000 small molecules reported the identification of 15 pluripotent cell-specific inhibitors (PluriSIns) . Interestingly, Ben-David et al determined that their most selective PluriSIn induces ER stress apoptosis in PSCs ; we corroborate these findings in identifying the ER stress pathway as a new target for understanding the signalling cascades which govern apoptosis in PSCs. However, in contrast to the PluriSIns described in the study by Ben-David et al, our compounds (JC011, JC010, JC017 and JC040) may not be involved in lipid metabolism as our molecules are structurally different and do not share a phenyl-hydrazine moiety which is a common feature of 9 PluriSIns . Instead, the 4 JC compounds we describe have high ORAC and DPPH values indicative of strong anti-oxidant properties (Fig. 3). We also establish that JC molecule PSC cytotoxicity is mediated mainly through the PERK/ATF4/DDIT3 signalling arm of the ER stress pathway and concomitant with a small but rapid reduction of intracellular ROS levels. This latter observation is unexpected because cell death as a result of the ER stress response is more commonly associated with an increase rather than a decrease in endogenous intracellular ROS levels [22,23]. We present in this report a group of novel small organic molecules that are cytotoxic and effective against PSCs. These molecules can be easily and efficiently synthesized besides having several advantages over existing antibody-based enrichment strategies for the removal of PSCs from their differentiated progenies. These molecules should also be useful as an in vitro tool to help enrich certain differentiated PSC derivatives by eradicating contaminating undifferentiated PSCs. Although toxicity towards primary cardiomyocytes (,5–9%) was observed following treatment with JC011 (20–100 mM) and toxicity towards astrocytes (,5%) observed at 20 mM, this should be acceptable as JC011 concentrations of ,20 mM were found to be sufficient in reducing PLOS ONE | www.plosone.org 7 March 2014 | Volume 9 | Issue 3 | e85039 Pluripotent Stem Cell Cytotoxic Small Molecules Figure 5. Comparative microarray analysis reveals involvement of the PERK/ATF4/DDIT3 ER stress pathways. Clustering of key differentially upregulated ER stress genes in BGO1V following 6 hr and 12 hr treatments with 20 mM JC011 (A). Top 10 components of the PERK/ ATF4/DDIT3 ER stress pathway that were found to be rapidly upregulated in JC011 treated BGO1V cells (B). qRT-PCR confirmation of upregulated UPR/ER stress pathway genes following JC011 treatment (C). doi:10.1371/journal.pone.0085039.g005 PSC contamination in differentiated cell populations by up to 6fold (Fig. 4e). Additionally, prolonged exposure of cardiomyocytes to JC011 (20 mM) for up to 5 days resulted in no further increase in cell death. These statistics are an improvement over the 4.5-fold enrichment figures obtained by single antibody sorting using classical hESC antibodies . Primary neurons however were found to be significantly affected by JC011 at 20 mM; this finding PLOS ONE | www.plosone.org 8 March 2014 | Volume 9 | Issue 3 | e85039 Pluripotent Stem Cell Cytotoxic Small Molecules PLOS ONE | www.plosone.org 9 March 2014 | Volume 9 | Issue 3 | e85039 Pluripotent Stem Cell Cytotoxic Small Molecules Figure 6. JC011 and JC040 reduce ROS levels in PSCs. Cytotoxic JC011 and JC040 induce a small but rapid reduction of intracellular ROS levels in BGO1V cells as confirmed by DCHF-DA FACS analysis 3 hrs after treatment (A). JC007 (non-cytotoxic) does not alter endogenous ROS levels while JC005 (non-cytotoxic) increases ROS levels but without any corresponding cytotoxicity to BGO1V (B). FACS histograms are representative outcomes of 4 independent experiments. DCHF-DA stained JC011 treated BGO1V cells (C, D). DCHF-DA stained JC005 treated BGO1V cells (E, F). doi:10.1371/journal.pone.0085039.g006 could severely limit the use of the JC compounds in neuronal differentiation strategies. Furthermore, we corroborate the findings of Ben-David et al  in identifying the ER stress pathway as a new target for understanding the signalling cascades which govern apoptosis in PSCs. A deeper understanding of the PSC apoptosis machinery will provide an effective means to eliminate teratoma-forming cells from cell preparations with clinically desirable phenotypes, triggering selective apoptosis in PSCs could potentially be one solution to eliminate unwanted teratoma-forming cells. Although additional safety studies have to be performed, our findings indicate that the molecules we describe may perhaps have the potential to be used safely in vivo. Downstream, together with other enrichment methods such as antibody-based sorting, these molecules could perhaps function as part of a strategy to safeguard against teratoma formation in PSC-based cell replacement regimes. Materials and Methods Ethics Statement This study was carried out in strict accordance with the recommendations of the Responsible Care and Use of Laboratory Animals guide of the National University of Singapore and Nanyang Polytechnic. All animal work was conducted according to relevant Singapore AM, Goodman SH. The association between psychopathology in fathers versus mothers and children’s internalizing and externalizing behavior problems. Psychol Bull. 2002; 128: 746–773. PMID: 12206193 15. Grace SL, Evindar A, Stewart DE. The effect of postpartum depression on child cognitive development and behavior: a review and critical analysis of the literature. Arch Women Ment Health. 2003; 6: 263– 74. 16. Hammen C, Burge D, Burney E, Adrian C. Longitudinal study of diagnoses in children of women with unipolar and bipolar affective disorder. Arch Gen Psychiat. 1990; 47: 1112–1117. PMID: 2244796 17. Luoma I., Tamminen T., Kaukonen P., Laippala P., Puura K., Salmelin R. et al. Longitudinal study of maternal depressive symptoms and child well-being. J Am Acad Child Adolesc Psychiat. 2001; 40: 1367–1374. 18. Brown GW, Harris TO. Life Events and Illness. New York: Guilford; 1989. 19. Hammen C. Stress and depression. Annu Rev Clin Psychol. 2005; 1: 293–319. PMID: 17716090 20. Mazure CM. Life stressors as risk factors in depression. Clin Psychol-Sci Pr. 1998; 5: 291–313. PLOS ONE | DOI:10.1371/journal.pone.0144131 December 8, 2015 10 / 12 Relationship Life Events and Depression during the Perinatal Period 21. Paykel ES. Life events and affective disorders. Acta Psychiatr Scand. 2003; 108: 61–66. 22. Lancaster C, Gold K, Flynn H, Marcus S, Davis M. Risk factors for depressive symptoms during pregnancy: A systematic review. Am J Obstet Gynecol. 2010; 202: 5–14. doi: 10.1016/j.ajog.2009.09.007 PMID: 20096252 23. Murray D, Cox JL, Chapman G, Jones P. Childbirth: Life event or start of a long-term difficulty? Further data from the Stoke-on-Trent controlled study of postnatal depression. Brit J Psychiat. 1995; 166: 595– 600. PMID: 7620743 24. Brown GW, Harris TO, Hepworth C. Loss, humiliation and entrapment among women developing depression: a patient and non-patient comparison. Psychol Med. 1995; 25: 7–21. PMID: 7792364 25. Finlay-Jones R, Brown GW. Types of stressful life events and the onset of anxiety and depressive disorders. Psychol Med;1981; 11: 803–815. PMID: 7323236 26. Kendler KS, Hettema JM, Butera F, Gardner CO, Prescott CA. Life event dimensions of loss, humiliation, entrapment, and danger in the prediction of onsets of major depression and generalized anxiety. Arch Gen Psychiat. 2003; 60: 789–96. PMID: 12912762 27. Kendler KS, Thornton LM, Prescott CA. Gender differences in the rates of exposure to stressful life events and sensitivity to their depressogenic effects. Am J Psychiat. 2001; 158: 587–93. PMID: 11282693 28. Zimmermann-Tansella C, Donini S, Lattanzi M, Siciliani O, Turrina C, Wilkinson G. Life events, social problems and physical health status as predictors of emotional distress in men and women in a community setting. Psychol Med. 1991; 21: 505–513 PMID: 1876655 29. Brown GW, Harris T. Social origins of depression: A study of psychiatric disorder in women. London: Tavistock; 1978. 30. Brown GW, Bifulco A, Harris TO. Life events, vulnerability and onset of depression: some refinements. Brit J of Psychiat. 1987; 150: 30–42. 31. Monroe SM, Rohde P, Seeley JR, Lewinsohn P. Life events and depression in adolescence: Relationship loss as a prospective risk factor for first onset of major depressive disorder. J Abnorm Psychol. 1999; 108: 606–614. PMID: 10609425 32. Gotlib IH, Whiffen VE, Wallace P, Mount JH. A prospective investigation of postpartum depression: Factors involved in onset and recovery. J Abnorm Psychol. 1991; 100: 122–132. PMID: 2040762 33. Dayan J., Creveuil C., Dreyfus M., Herlicoviez M., Baleytel J-M, & O’Keane V. Developmental model of depression applied to prenatal depression: Role of present and past life events, past emotional disorders and pregnancy stress. PLoS One. 2010; 5(9): e12942. doi: 10.1371/journal.pone.0012942 PMID: 20877652 34. Pilkington PD, Milne LC, Cairns KE, Lewis J, Whelan TA. Modifiable partner factors associated with perinatal depression and anxiety: A systematic review and meta-analysis. J Affect Disord. 2015; 178: 165–180. doi: 10.1016/j.jad.2015.02.023 PMID: 25837550 35. Sharp H, Pickles A, Meaney M, Marshall K, Tibu F, Hill J. Frequency of infant stroking reported by mothers moderates the effect of prenatal depression on infant behavioural and physiological outcomes. PLoS ONE. 2012; 7: e45446. doi: 10.1371/journal.pone.0045446 PMID: 23091594 36. Moffitt TE, Caspi A, Margolin G, Krueger RF, Magdol L, Silva PA et al. Do partners agree about abuse in their relationship? A psychometric evaluation of interpartner agreement. Psychol Assessment. 1997; 9: 47–56. 37. Noble M, Wright G, Dibben C, Smith G, McLennan D, Anttila C. The English Indices of Deprivation 2004 (revised). Report to the Office of the Deputy Prime Minister. London: Neighbourhood Renewal Unit; 2004. 38. Spitzer R, Endicott J. The schedule for affective disorders and schizophrenia, lifetime version, 3rd ed. New York: New York State Psychiatric Institute; 1977. 39. Caspi A, Moffitt TE, Thornton A, Freedman D, Amell JW, Harrington HL et al. The life history calendar: A research and clinical assessment method for collecting retrospective event-history data. Int J Method Psych. 1996; 6: 101–114. 40. Freedman D, Thornton A, Camburn D, Alwin D, Young-DeMarco L. The life history calendar: A technique for collecting retrospective data. In Clogg CC, editor, (Ed.) Sociological Methodology Vol.18. Washington, D.C.: American Sociological Association; 1988. p.37–68. 41. Belli RF, Shay WL, Stafford FP. Event history calendars and question list surveys: A direct comparison of interviewing methods. Public Opin Quart. 2001; 65: 45–74. 42. Brugha T, Bebbington P, Tennant C, Hurry J. The List of Threatening Experiences: a subset of 12 life event categories with considerable long-term contextual threat. Psychol Med. 1985; 15: 189–194. PMID: 3991833 PLOS ONE | DOI:10.1371/journal.pone.0144131 December 8, 2015 11 / 12 Relationship Life Events and Depression during the Perinatal Period 43. Caspi A, Karen S, Moffitt T, Taylor A, Craig IW, Harrington H, McClay L et al. Influence of life stress on depression: moderation by a polymorphism in the 5-HTT gene. Science. 2003; 301(5631): 386–389. PMID: 12869766 44. Kendler KS, Karowski LM, Prescott CA. Causal relationship between stressful life events and the onset of major depression. Am J Psychiat. 1999; 156: 837–841. PMID: 10360120 45. American Psychiatric Association. Diagnostic and statistical manual of mental disorders. 4th Ed. Washington, DC: American Psychiatric Association; 1994. 46. Harrington R, Hill J, Rutter M, John K, Fudge H, Zoccolillo M, Weissman M. The assessment of lifetime psychopathology: a comparison of two interviewing styles. Psychol Med. 1988; 18: 487–93. PMID: 3399595 47. Surtees PG, Rennie D. Adversity and the onset of psychiatric disorder in women. Soc Psychiat. 1983; 18: 37–44. 48. Cox DR. Regression models and life tables. J Roy Stat Soc B. 1972; 34: 187–220. 49. StataCorp. Stata Statistical Software: Release 12. College Station, TX: StataCorp LP; 2011. 50. Holmes TH, Rahe RH. The Social Readjustment Rating Scale. J Psychosom Res. 1967; 11:213–8. PMID: 6059863 51. Lewisohn PM, Mermelstein RM, Alexander C, MacPhillamy DJ. The Unpleasant Events Schedule: a scale for measurement of aversive events. J C Psychol. 1985; 41: 483–498. 52. Stuart S, O’Hara MW. Interpersonal psychotherapy for postpartum depression: a treatment program. J Psychother Pract Res. 1995; 4(1): 18–29. PMID: 22700210 53. O’Hara MW, Stuart S, Gorman LL, Wenzel A. Efficacy of interpersonal psychotherapy for postpartum depression. Arch Gen Psychiat. 2000; 57(11): 1039–1045. PMID: 11074869 54. Spinelli MG, Endicott J. Controlled clinical trial of interpersonal psychotherapy versus parenting education program for depressed pregnant women. Am J Psychiat. 2003; 160(3): 555–562. PMID: 12611838 55. Spinelli MG, Endicott J, Leon AC, Goetz RR, Kalish RB, Brustman LE et al. A controlled clinical treatment trial of interpersonal psychotherapy for depressed pregnant women at 3 New York City sites. J Clin Psychiat. 2013; 74(4): 393–399. 56. Swendsen JD, Mazure JD. Life stress as a risk factor for postpartum depression: Current research and methodological issues. Clin Psychol-Sci Pr. 2000; 7: 17–31. 57. Liu RT, Alloy LB. Stress generation in depression: A systematic review of the empirical literature and recommendations for future study. Clin Psychol Rev. 2010; 30: 582–93. doi: 10.1016/j.cpr.2010.04.010 PMID: 20478648 58. Beck CT. Predictors of postpartum depression: An update. Nurs Res. 2001; 50: 275–285. PMID: 11570712 59. Austin MP, Marcé Society Position Statement Advisory Committee. Marcé International Society position statement on psychosocial assessment and depression screening in perinatal women. Best Pract Res Cl Ob. 2014; 28(1): 197–187. PLOS ONE | DOI:10.1371/journal.pone.0144131 December 8, 2015 12 / 12