Cellular responses to compound stress induced by atmospheric-pressure plasma in fission yeast

Cells are continuously exposed to a variety of stresses and need to respond appropriately to them to cope with such stress conditions. The fission yeast Schizosaccharomyces pombe is a popular model organism for exploring diverse aspects of mammalian biology, including responses to cellular stress (Hayles and Nurse, 2018). In S. pombe, the stress-activated Sty1 (also known as Spc1) mitogen-activated protein kinase (MAPK) pathway, which is homologous to the mammalian stress-activated p38 MAPK pathway, plays a central role in cell survival under diverse stress conditions, such as oxidative and osmotic stress, ultraviolet light and heat shock (Degols et al., 1996; Degols and Russell, 1997; Vivancos et al., 2006). Stress-response mechanisms centered on Sty1 have been intensively studied. However, most, if not all, of these studies have focused on responses to a single stressor, and it remains to be proven how cells respond when they suffer from multiple stresses simultaneously.

In recent years, non-thermal atmospheric-pressure plasma has been extensively studied for various applications in the medical and agricultural fields (Ito et al., 2018; Tanaka et al., 2018). Plasma is the fourth state of matter and is an electrically charged gas. Recent advances have made it possible to generate ‘cold’ plasma at atmospheric pressure, allowing direct plasma irradiation to biological or living materials. Plasma irradiation is known to have various effects on living organisms, such as selective death of cancer cells and promotion of plant growth (Koga et al., 2016; Alizadeh and Ptasińska, 2021). It is also used for wound healing, blood coagulation, surface sterilization and cancer therapy (Brany et al., 2020). However, the detailed molecular mechanisms underlying the effects of plasma irradiation on living organisms remain largely unknown.

Atmospheric-pressure plasma produces charged particles, ultraviolet radiation and electromagnetic fields. It also produces a variety of reactive oxygen and nitrogen species from the oxygen, nitrogen and water present in ambient air (Yoshimura et al., 2020). Thus, plasma irradiation can be used to induce multiple stresses. We set out to elucidate the cellular responses to compound stresses induced by plasma irradiation using S. pombe.

To irradiate cells with plasma directly without heat stress, it is necessary to maintain the temperature of the irradiation area within an appropriate range. We previously developed a plasma device with strict temperature control (Yoshimura et al., 2019). In this system, the supplied helium gas is cooled by a Peltier device before discharging. Plasma generated by this device produces a variety of cellular stresses, including hydrogen peroxide (Yagi-Utsumi et al., 2021). Using this device, we investigated the cellular responses to a novel compound stress, plasma, in S. pombe. Although the effects of hydrogen peroxide are substantial, we shifted our focus to the cellular responses that become evident only when utilizing the novel stressor, plasma.

To clarify the effects of plasma irradiation on S. pombe, we applied plasma irradiation to growing wild-type cells in the log phase under various conditions. Irradiation of plasma generated by discharging helium gas inhibited S. pombe growth depending on the irradiation duration, whereas helium gas had only a slight effect when applied for prolonged exposure times (Fig. 1A). The effect of helium can be attributed to desiccation. We examined the cell morphology after plasma irradiation. Elongated and bent cells were observed 4 h after 30-s or 2-min irradiation (Fig. 1B), as shown previously (Yoshimura et al., 2019), suggesting a defect in cell cycle progression. DNA staining with DAPI revealed no significant defects in chromosome distribution after plasma irradiation. Some cells displayed no staining, potentially indicating the presence of dead cells.

We then compared plasma sensitivity between cells in the log and stationary phases. Stationary phase cells were more resistant to plasma irradiation than log phase cells (Fig. S1A). There was no difference in plasma sensitivity between haploid and diploid status in log phase cells (Fig. S1B).

Plasma irradiation induces various stressors, such as oxidative stress, UV stress and electric field stress. To examine which stress response pathways are important in responses to plasma irradiation, we exposed several mutants to plasma. A deletion mutant of the sty1 gene, which encodes a stress-responsive MAPK that plays essential roles in oxidative stress responses (Millar et al., 1995; Shiozaki and Russell, 1995; Vivancos et al., 2006), was highly sensitive to plasma irradiation (Fig. 1C). The DNA-damage sensitive mutants rad1-1 and rad3-136 [rad1 and rad3 encode a subunit of the highly conserved Rad9–Rad1–Hus1 (9-1-1) checkpoint-clamp and ATR checkpoint kinase, respectively; Jimenez et al., 1992; Rowley et al., 1992; Navadgi-Patil and Burgers, 2009], exhibited similar sensitivity to plasma irradiation to the wild-type strain (Fig. S1C). Deletion mutants of chk1 and cds1, encoding checkpoint effector kinases (Walworth et al., 1993; Murakami and Okayama, 1995; Stracker et al., 2009) also had similar sensitivity to the wild-type strain. These results suggest that plasma irradiation does not induce severe DNA damage, at least under our experimental conditions. We also tested the plasma sensitivity of cell cycle mutants, namely cdc2 and cdc10, which encode the master cell cycle regulator Cdk1 and a transcription factor required for G1-S phase transition, respectively (Lowndes et al., 1992; Morgan, 1997; Moser and Russell, 2000). Both temperature-sensitive cell cycle mutants, cdc2-33 and cdc10-129 (Nurse et al., 1976; Nurse and Bissett, 1981), exhibited plasma sensitivity comparable to that of the wild-type strain, at least at a permissive temperature, where some loss of function is expected (Fig. S1C).

To identify factors that might be involved in the cellular response to plasma irradiation, we screened for mutations that confer resistance to plasma irradiation stress. We randomly introduced mutations into wild-type cells and selected mutants that could survive after 2 min of plasma irradiation, which allows little or no growth in wild-type cells (Fig. 2A, see Materials and Methods). We isolated a plasma-resistant mutant and designated it as prm1, standing for plasma resistant mutant (Fig. 2B). Microscopic observations revealed that the prm1 mutant exhibited a deficiency in cell separation. It showed a chained and branched cell phenotype (Fig. 2C).

Next, we determined the gene responsible for prm1 by whole-genome sequencing (Iida et al., 2014; see Materials and Methods). The causative gene for prm1 was sep1, which encodes a forkhead family transcription factor (Ribar et al., 1997). The mutation in prm1 was a single nucleotide deletion in codon 406, which resulted in the premature truncation of the C-terminal 258 amino acids with the addition of three amino acids before the stop codon (Fig. 3A). Sep1 is a cell cycle-dependent transcription factor, with most of its targets functioning during mitosis (Rustici et al., 2004; Bahler, 2005). sep1 mutant cells have been shown to have defects in cell separation, leading to a chained cell phenotype, as observed in the prm1 mutant strain (Sipiczki et al., 1993; Ribar et al., 1999). The growth rate of the sep1 deletion mutant was comparable to that of the wild-type strain under non-stressed conditions, whereas the cell concentration at saturation was slightly lower (Fig. S2A). The sep1 deletion mutant was resistant to plasma irradiation (Fig. 3B). Plasma irradiation did not affect the cell separation-deficient phenotype of sep1Δ cells (Fig. S2B).

We then examined the sensitivity of sep1Δ cells to various stressors. In contrast to what was seen for sty1Δ cells, which exhibit a temperature-sensitive growth defect and osmotic stress sensitivity, sep1Δ cells did not show any growth defects at lower and higher temperatures and under high osmotic conditions (Fig. S2C). Notably, sep1Δ cells showed little, if not no, resistance to oxidative stress (Fig. 3C).

Sep1 regulates the expression of genes involved in mitotic progression (Rustici et al., 2004). One of the Sep1 targets is ace2, which encodes a C2H2 zinc-finger family transcription factor (Martin-Cuadrado et al., 2003). Ace2 upregulates the expression of genes important for cell separation, such as glucanase genes (Martin-Cuadrado et al., 2003; Alonso-Nunez et al., 2005; Petit et al., 2005). Deletion of the ace2 genes caused a cell separation-deficient phenotype, as in sep1Δ cells (Martin-Cuadrado et al., 2003) (Fig. S2D). Furthermore, ace2Δ cells exhibited similar stress sensitivity to sep1Δ cells, including resistance to plasma irradiation (Fig. 3C,D; Fig. S2C).

We confirmed that downregulation of ace2 was responsible for plasma resistance in sep1Δ cells. Cell separation defects in sep1Δ cells were suppressed by expressing ace2 from an exogenous promoter as shown previously (Alonso-Nunez et al., 2005; Fig. S2E). Artificial expression of ace2 restored plasma sensitivity in sep1Δ cells (Fig. 3E), implying that plasma resistance in sep1Δ is a consequence of the downregulation of the ace2 expression.

We next investigated plasma resistance in other cell separation mutants. mid2, ang1 and eng1 are transcriptional targets of Ace2, and encode an anillin-like protein, α-glucanase and β-glucanase, respectively, all of which are involved in the process of cell separation (Martin-Cuadrado et al., 2003; Alonso-Nunez et al., 2005; Petit et al., 2005). mid2 mutant cells expressing C-terminal truncated Mid2 exhibited severe cell separation defects (Fig. S2F; Berlin et al., 2003). agn1Δ eng1Δ cells showed cell separation defects, although to a lesser extent. A mutant of myo3 (also known as myp2), which encodes a type-II myosin heavy chain, also exhibited impaired cell separation (Fig. S2F) (Bezanilla et al., 1997; Motegi et al., 1997). In contrast to ace2Δ cells, mid2, myo3 and agn1 eng1 mutant cells did not show plasma resistance (Fig. S2G). These observations suggest that not all cell separation defects confer plasma resistance and that the Sep1-Ace2 pathway plays a specific role in the induction of plasma resistance, although the downstream constituents of Ace2 remain elusive.

To identify the entire repertoire of plasma response genes, we conducted transcriptome analysis after plasma irradiation. Cells were harvested at 10, 15 and 30 min after 30 s of plasma irradiation and RNA was prepared for RNA-seq analysis. We compared the gene expression profiles after plasma irradiation with that of hydrogen peroxide-treated cells (Fig. 4A). After genome-wide transcriptome analysis, we found that 211, 288 and 341 genes were upregulated and 31, 74 and 102 genes were downregulated more than two-fold at 10, 15 and 30 min after plasma irradiation, respectively. Hydrogen peroxide treatment induced upregulation of 326 genes and downregulation of 60 genes. Although ∼70% of the genes upregulated by plasma were also induced by hydrogen peroxide treatment, ∼70% of the downregulated genes were specific to the plasma treatment (Fig. 4B; Fig. S3A).

Gene Ontology (GO) analysis of protein-coding genes that increased expression levels at 30 min post plasma irradiation showed strong enrichment for functions involved in detoxification, response to chemicals, membrane transport and several metabolic processes (Fig. 4C). Genes that were upregulated in a plasma-specific manner and not by hydrogen peroxide treatment displayed enrichment for functions in amino acid membrane transport and lactate metabolism processes (Fig. S3B).

GO analysis of downregulated genes at 30 min revealed an enrichment for a function in cell wall organization or biogenesis (Fig. 4C). GO analysis categorized by gene expression through AnGeLi (Bitton et al., 2015) revealed that Sep1 target genes, most of which are involved in mitotic progression, were specifically downregulated by plasma irradiation (Fig. 4D; Fig. S3C). We examined the expression of ace2 and 14 potential Sep1-target genes, whose expression is known to be Sep1 dependent and Ace2 independent (Rustici et al., 2004). Indeed, plasma irradiation downregulated the expression of 11 of the 15 known Sep1-dependent genes, whereas only five genes were downregulated by hydrogen peroxide treatment (Fig. 5A). The expression of ace2 was downregulated by both plasma irradiation and hydrogen peroxide treatment. Taken together with the result that deletion of sep1 and ace2 conferred plasma resistance, downregulation of these transcription factors or their targets is likely to be vital to cope with plasma irradiation stress, whereas downregulation might be insufficient in wild-type cells, resulting in plasma sensitivity. In consistent with this, among the known Ace2 targets (mid2, agn1, eng1, adg1, adg2, adg3 and cfh4), only eng1 exhibited downregulation in response to plasma in wild-type cells. The expression of sep1 did not decrease after plasma irradiation; instead, it showed a slight increase. This suggests that Sep1 is downregulated post-transcriptionally.

GO analysis categorized by gene expression also demonstrated that, in addition to stress-responsive genes, caffeine- and rapamycin-induced genes were enriched in the set of genes upregulated by plasma (Fig. S3C) (Rallis et al., 2013). Caffeine- and rapamycin-repressed genes were enriched in genes downregulated by plasma. Genes in these categories showed strong enrichment for genes that changed expression levels in a plasma-specific manner (Fig. 4D). Treatment with caffeine and rapamycin has been shown to inhibit the highly conserved protein kinase complex TORC1 (Takahara and Maeda, 2012). Among the 343 caffeine- and rapamycin-induced genes, the expression of 125 increased more than two-fold within 30 min after plasma irradiation (Fig. 5B). The decrease in expression was not as pronounced, but the expression of 20 of the 349 caffeine- and rapamycin-repressed genes was reduced by more than two-fold.

We then examined whether TORC1 activity was indeed reduced after plasma irradiation by observing localization of the GATA family transcription factor Gaf1, which is known to change its localization depending on TORC1 activity (Laor et al., 2015; Ma et al., 2015). Gaf1 localizes to the cytoplasm of vegetatively growing cells, where TORC1 is active. When TORC1 is inactivated by nitrogen starvation or mutations involving TORC1 components, Gaf1 translocates to the nucleus. Plasma irradiation caused nuclear accumulation of Gaf1 in 39% of the cells, indicating a reduction in TORC1 activity (Fig. 6A,B). In contrast, hydrogen peroxide treatment only weakly induced distinct nuclear localization of Gaf1 (6%). We then confirmed the effect of plasma on TORC1 activity by using western blot analysis to observe the phosphorylation status of S6 kinase Psk1, a target protein of TORC1 (Nakashima et al., 2012; Otsubo et al., 2017). Psk1 phosphorylation decreased from 10 min after plasma irradiation, with a significant reduction after 15 min (Fig. 6C). These results imply that TORC1 activity is suppressed by plasma irradiation.

We investigated whether plasma resistance caused by the deletion of sep1 and ace2 is related to TORC1 activity. In sep1Δ and ace2Δ cells, Gaf1 localized to the cytoplasm as in wild-type cells, indicating that TORC1 was not compromised in the deletion mutants (Fig. 7A,B). Upon plasma irradiation, nuclear accumulation of Gaf1 was observed in both sep1Δ and ace2Δ cells at a similar frequency to that in wild-type cells (36% in sep1Δ cells and 37% cells in ace2Δ cells, compared to 39% in wild-type cells; Fig. 6B). This indicates that downregulation of TORC1 by plasma irradiation is independent of the Sep1-Ace2 pathway.

In this study, we investigated cellular responses to a novel compound stress induced by plasma irradiation in S. pombe. Plasma sensitivity was distinct in growing log phase cells, and stationary phase cells were less sensitive. Plasma irradiation has been reported to induce selective cell death in a variety of cancer cells (Tanaka et al., 2018; Brany et al., 2020; Alizadeh and Ptasińska, 2021), implying that plasma preferentially affects actively proliferating cells. It has also been suggested that the plasma-responsive pathway(s) is conserved from yeast to humans.

In S. pombe, mutants defective in DNA damage responses did not exhibit plasma sensitivity. These results indicate that plasma irradiation using our device could induce only weak DNA damage, which might be negligible in S. pombe. In addition, cell cycle mutants did not display plasma sensitivity. However, we performed experiments at a restrictive temperature for the mutants. To clarify the cell cycle dependency, further experiments, such as irradiation after cell cycle arrest, are required. In order to perform more detailed observations, additional equipment development is also necessary, because the current plasma irradiation device is limited to irradiation in a spot-like manner at a single point. This prevents simultaneous processing of a large number of cells under identical conditions and cells positioned away from the irradiation points inevitably escape the effects of plasma.

The stress-responsive MAPK Sty1 plays a crucial role in various stress responses, such as osmotic, heat and oxidative stress (Vivancos et al., 2006). Plasma irradiation induces the generation of hydrogen peroxide, and thus it is not surprising that sty1Δ cells show sensitivity to plasma. We found that sep1Δ and ace2Δ cells were more resistant to plasma irradiation than wild-type cells. Both Sep1 and Ace2 are transcription factors, and the expression of the latter is directly regulated by the former (Ribar et al., 1997; Martin-Cuadrado et al., 2003; Rustici et al., 2004). Ace2 induces transcription of target genes, most of which are involved in cell separation (Martin-Cuadrado et al., 2003; Rustici et al., 2004; Alonso-Nunez et al., 2005; Petit et al., 2005). Deletion of ace2 and sep1 results in reduced expression of these genes, leading to defects in cell separation. Our transcriptome analysis revealed that plasma irradiation induced the downregulation of Sep1-target genes. This suggests that S. pombe cells might cope with plasma stress by inhibiting the Sep1-Ace2 transcriptional pathway. In wild-type cells, however, downregulation might be insufficient, resulting in the plasma-sensitive phenotype. This is consistent with our findings that plasma irradiation in wild-type cells did not induce the cell separation-deficient phenotype and had no significant impact on the expression of most Ace2 target genes, whereas a cell elongation phenotype was observed in some cells after plasma irradiation. Our expression analysis revealed that the transcripts of sep1 did not decrease after plasma irradiation, suggesting that Sep1 is downregulated post-transcriptionally.

The precise molecular mechanisms by which deletion of sep1 and ace2 causes plasma stress tolerance remain elusive at this stage. sep1Δ and ace2Δ cells did not exhibit distinct resistance to oxidative stress. This suggests that the plasma-resistant phenotype of these cells might arise from alterations in pathway(s) other than oxidative response pathways. Plasma irradiation induced gene expression patterns distinct from those of hydrogen peroxide treatment, particularly in downregulated genes, further supporting this. It has been reported that in S. cerevisiae, deletion of the ACE2 gene prevents mother-daughter cell separation, leading to multicellular ‘snowflake’ yeast aggregates, which has been used as a laboratory model for the origin of multicellularity (Oud et al., 2013; Ratcliff et al., 2015). Multicellularity has been proposed to contribute to stress tolerance; however, the underlying mechanisms remain unclear. Further studies will shed light on the relationship between the regulation of cell separation and plasma resistance.

Our comprehensive gene expression analysis suggests a contribution of the TORC1 pathway to plasma responses. TORC1 downregulation upon plasma irradiation was observed in sep1Δ and ace2Δ cells. Inactivation of TORC1 in TORC1 mutants results in cell cycle arrest at G1 and does not cause cell separation defects (Alvarez and Moreno, 2006; Uritani et al., 2006; Hayashi et al., 2007; Matsuo et al., 2007; Weisman et al., 2007). These observations indicate that two independent pathways, namely the TORC1 and Sep1-Ace2 cell separation pathways, operate in plasma responses in S. pombe. A relationship between TORC1 and plasma has been reported in human tumor cell lines. Plasma-treated medium, which has been shown to possess anti-tumor activity, suppresses TORC1 activity in glioblastoma brain tumor cells (Tanaka et al., 2012) and myeloid leukemia cells (Kim et al., 2020), suggesting that downregulation of TORC1 by plasma stress might be widely conserved between S. pombe and humans. Dysregulation of TORC1 is linked to the development of human diseases, such as diabetes, obesity, cancer and neurodegeneration (Dazert and Hall, 2011). TORC1 is also closely associated with aging (Cornu et al., 2013). It is anticipated that further studies to clarify the role of TORC1 in plasma responses will provide useful information in addressing these diseases and aging.

The S. pombe strains used in this study are listed in Table S1. The general genetic manipulation procedures for S. pombe were performed as previously described (Moreno et al., 1991). Complete medium YE was used (Moreno et al., 1991). Mutagenesis of S. pombe cells with N-methyl-N′-nitro-N-nitroso-guanidine was performed as previously described (Arata et al., 2014). Gene disruption was performed as previously described (Bahler et al., 1998; Sato et al., 2005). To observe Gaf1 localization, the gaf1 gene encoding the C-terminally truncated protein, which shows similar localization patterns to the full-length protein, was fused to the mScarlet-I gene and integrated to the Z locus on chromosome II (Sakuno et al., 2009). To overexpress ace2, the ace2 gene was integrated into 2 L locus (Sakai et al., 2021) with the Padh15 promoter (Sugiyama et al., 2023 preprint).

Plasma irradiation was performed using an in-house device with temperature-controlled helium as the feeding gas (Yoshimura et al., 2019). Cells in the log phase or stationary phase were spotted onto YE agar medium or onto a membrane filter (Durapore, 0.65 µm PVDF membrane) on YE agar medium, and then irradiated for the indicated times. Irradiation was performed under the following conditions: distance from the nozzle, 20 mm; temperature at irradiation point, 30–32°C; discharge voltage, 6.5 kV; gas flow rate, 3.0 SLM. For isolation of plasma-resistant mutants, 1000 cells were spotted onto YE agar medium, and then irradiated for 2 min. After plasma irradiation, a total of 10,000 cells (10 spots) were collected, plated onto fresh YE agar medium, and incubated for 3 days at 30°C. At total of 15 colonies that had grown were subjected to a second round of irradiation, and the prm1 mutant was isolated.

For differential interference contrast (DIC) imaging, S. pombe cells were observed under a microscope (Axioplan 2, Carl Zeiss, Oberkochen, Germany) equipped with a chilled CCD camera (CoolSNAP HQ2, Photometrics, Tucson, AZ, USA) and MetaMorph software (Molecular Devices, Sunnyvale, CA, USA). For DAPI staining, 10⁶ cells were spotted onto a membrane filter on YE medium. After plasma irradiation, cells were collected, transferred to fresh YE liquid medium and cultured for 4 h at 30°C. Cells were then fixed with ice-cold methanol containing 1 µg/ml DAPI for 10 min at room temperature, washed twice with PEMS (1.2 M Sorbitol, 100 mM PIPES, 1 mM EGTA, 1 mM MgSO₄), and suspended into PEMS. Cells were imaged with an IX83 inverted microscope (Olympus, Tokyo, Japan) equipped with a Prime sCMOS camera (Photometrix) and an oil-immersion lens (UPLXAPO 100×). For fluorescence imaging, cells were prepared as for RNA-seq analysis and observed using the PersonaDV microscope imaging system (Applied Precision, WA, USA). Images were acquired by collecting 10 optical sections along the z-axis at 0.4 µm intervals, and were deconvolved and merged into single image projections.

Cells in log phase were spotted onto a membrane filter (Durapore, 0.65 µm PVDF membrane). After plasma irradiation, cells were harvested and disrupted with glass beads in 20% trichloroacetic acid, as described previously (Otsubo et al., 2018). The samples were separated using 7.5% SDS-PAGE. Mouse monoclonal antibodies against phospho-p70 S6 kinase (Thr389; 1:1000; 1A5; Cell Signaling Technology, Danvers, MA, USA) and γ-tubulin (1:2000; GTU-88; Sigma-Aldrich, St Louis, MO, USA) were used.

Determination of the causative gene of prm1 was performed as previously (Iida et al., 2014). Briefly, the prm1 strain was backcrossed with the wild-type strain, followed by tetrad analysis. Six spores with the plasma-resistant phenotype and six spores with the plasma-sensitive phenotype were pooled, and genomic DNA was purified using Genomic tip 20/G (Qiagen). Library construction and sequencing were performed by Macrogen. Sequencing data were analyzed using the Mudi platform (Iida et al., 2014).

Cells in the log phase were spotted onto a membrane filter (Durapore, 0.65 µm PVDF membrane) on YE medium. After plasma irradiation, cells were harvested, and total RNA was extracted with hot phenol (Bahler and Wise, 2017). For hydrogen peroxide treatment, cells were spotted onto a membrane filter on YE medium containing 1 mM hydrogen peroxide for 15 min and harvested. Library construction and sequencing were performed using the TruSeq stranded mRNA Library and NovaSeq6000 (Illumina) by Macrogen. Three biological replicates were used to generate the libraries. The reads were quality trimmed using sickle (v.1.33), aligned to the S. pombe reference genome (R64-1-1) using HISAT2 (v.2.2.1), and processed using Samtools (v. 1.13) and featureCouts (v.2.0.1). Differential expression was analyzed using DESeq2 (v.1.32.0) in R. Gene ontology analysis was performed using GOTermFinder (https://go.princeton.edu/cgi-bin/GOTermFinder; Boyle et al., 2004) and AnGeLi (http://bahlerweb.cs.ucl.ac.uk/cgi-bin/GLA/GLA_input; Bitton et al., 2015).

We thank Dr Naoko Iida for technical advice, Atsuko Nakade for technical support, Data Integration and Analysis Facility, National Institute for Basic Biology for computational resources to analyze RNA-seq data, and Dr Kiyokazu Agata and Dr Masaru Hori for their support and helpful discussions.

The peer review history is available online at https://journals.biologists.com/jcs/lookup/doi/10.1242/jcs.261292.reviewer-comments.pdf