Caffeine, a naturally occurring purine analog, is a frequently consumed methylxanthine derivative found in both beverages and medications. It exhibits a diverse array of biological effects, targeting multiple cellular pathways in organisms ranging from fungi and plants to humans. In higher animals, caffeine elicits neuronal stimulation and confers antidepressant and diuretic properties (López-Cruz et al., 2018). It functions as a relatively non-selective pharmacological agent, impacting various cellular processes such as cell growth, proliferation, and energy metabolism. Mechanistically, caffeine inhibits the target of rapamycin (TOR) signaling and interferes with the DNA damage repair system (Zhou et al., 2010; Ruta and Farcasanu, 2020).

In yeast species, several investigations have demonstrated that caffeine induces cytotoxic effects by disrupting DNA damage repair pathways and cell cycle checkpoint signaling (Moser et al., 2000; Kuranda et al., 2006; Chung, 2021). Specifically, caffeine leads to failure in DNA double-strand break (DSB) repair by inhibiting the recruitment of nucleases Sae2 and Dna2, essential for break end resection, a crucial early step in homologous recombination (HR), and by interfering with Rad51 involved in repair homology detection as well (Zelensky et al., 2013; Tsabar et al., 2015). Inversely, caffeine has potential antioxidant activity and is involved in reduction of reactive oxygen species (ROS)-mediated DNA damage and mutagenesis (Czachor et al., 2020; Choi et al., 2022).

A genome-wide screening in the fission yeast Schizosaccharomyces pombe has unveiled that several mutants deficient in HR and oxidative stress response transcription factors, as well as drug efflux pumps, exhibit heightened sensitivity to caffeine (Calvo et al., 2009). In addition, the overexpression of Pap1, an AP1-like transcription factor, or up-regulation of Caf5, an ABC transporter, renders cells resistant to caffeine (Benko et al., 2004). In the budding yeast Saccharomyces cerevisiae, however, mutant cells lacking Yap1 or Skn7, two specialized transcriptional regulators for oxidative stress response, are not sensitive to caffeine, and neither Pdr5 nor Snq2, two major caffeine efflux pumps, is significantly required for caffeine resistance (Tsujimoto et al., 2015; Choi et al., 2022).

In this study, we identified ALF1 gene as a novel factor potentially involved in caffeine tolerance through a yeast deletion mutant library screening. Alf1 is an alpha-tubulin folding protein, similar to mammalian cofactor B, that is functionally and physically associated with α-tubulin (Tian et al., 1997; Feierbach et al., 1999; Fig. 1A) Microtubules are essential cytoskeletal elements of eukaryotic cells and are composed of heterodimers of α- and β-tubulin. Alf1 plays an important role in recycling tubulin by sequestering α-tubulin from interaction with β-tubulin, preserving the reservoir of ready-to-assemble tubulin in the cell (Li and Moore, 2020). Rbl2 is involved in the folding of β-tubulin, similar to mammalian cofactor A, and serves as a counterpart to Alf1, which is responsible for α-tubulin folding. Pac2 and Cin1 are microtubule effectors required for tubulin heterodimer formation, analogous to mammalian tubulin folding factors E and D, respectively (Fig. 1A).

Fig. 1. The loss of Alf1 results in abnormal cell shape and increased caffeine sensitivity. (A) A schematic diagram illustrating the α- and β-tubulin folding pathway. Tubulin heterodimer formation and microtubule assembly are mediated by a series of tubulin folding proteins and microtubule effectors. Several proteins containing the CAP-Gly domain (hatched boxes), implicated in various important cellular processes, are also shown. The numbers above the black line indicate the positions of amino acids at N-terminus and C-terminus of each protein. (B) The alf1 mutant shows unusual cell morphology. (C) 10-fold serial dilutions of yeast mutant strains were spotted on YEPD media in the presence of various concentrations of caffeine and incubated for 3 days at 30°C.

To dissect the unforeseen function of Alf1 in counteracting caffeine toxicity, we assessed the impact of Alf1 deficiency in the absence of Rad51, a pivotal protein in HR, given the significance of caffeine sensitivity in DSB repair efficiency. We investigated a specific genetic interaction between a tubulin folding cofactor and a DNA damage repair protein concerning drug sensitivity, formation of repair foci, and regulation of the cell cycle. Overall, our findings propose an innovative genetic interplay between cytoskeleton regulation and genome integrity, offering a novel synergistic mechanism for cellular survival.

Strains and media

All of the strains utilized in this study are isogenic descendants of S. cerevisiae BY4741 (MATa his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0) obtained from Yeast Knockout (YKO) collection (YSC1053 glycerol stock, Thermo Scientific). All strains used in this study and their genotypes are listed in Table 1. The rad51 alf1 double mutant was created by transformation of rad51 mutant with alf1::KAN gene deletion cassettes and confirmed by PCR analysis. The strains with C-terminal GFP fusion protein of Rad52 were constructed by oligonucleotide-directed in-frame tagging method as previously described (Huh et al., 2003). Yeast cell cultures and treatments with caffeine, hydroxyurea (HU), phleomycin (PHL), methyl methanesulfonate (MMS), camptothecin (CPT), or 4-nitroquinoline 1-oxide (4NQO) were performed in YEPD (1% yeast extract, 2% peptone, and 2% dextrose) liquid media or on solid agar plates.

Table 1

Strains used in this study

Strain name	Parental strain	Genotype	Reference
BY4741	-	MATa his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0
BY4742	-	MATalpha his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0
yWH537	BY4741	WT	Yi et al. (2016)
yWH1402	BY4742	rad51::LEU2	Yi et al. (2016)
yWH1639	BY4741	alf1::KANMX	This study
yWH1913	BY4741	rad51::LEU2 alf1::KANMX	This study
yWH1910	BY4741	nip100::KANMX	This study
yWH1911	BY4741	bik1::KANMX	This study
yWH1912	BY4741	pac2::KANMX	This study
yWH2077	BY4741	rbl2::KANMX	This study
yWH2078	BY4741	cin1::KANMX	This study
yWH1463	BY4741	WT RAD52-GFP-HIS3	Choi et al. (2018)
yWH1579	BY4741	rad51::LEU2 RAD52-GFP-HIS3	Choi et al. (2018)
yWH2503	BY4741	alf1::KANMX RAD52-GFP-HIS3	This study
yWH2507	BY4741	rad51::LEU2 alf1::KANMX RAD52-GFP-HIS3	This study

High-throughput caffeine sensitivity screening

More than 4,800 non-essential genes are covered by S. cerevisiae genome-wide haploid deletion mutant library. The entire library collection was first cultivated in fresh liquid YEPD medium and then spotted in each 3 μl volume with a manual replicator onto solid YEPD plates with or without caffeine in concentration of 10, 12, and 15 mM. The plates were incubated at 30°C for 3 days, and the sensitivity was analyzed. The caffeine sensitive mutant candidates were collected and spotted again to confirm their caffeine sensitivity. Total 192 deletion mutants were isolated showing repetitive growth impairment compared to WT strain against a series of caffeine treatments.

Measurement of drug sensitivity

For drug sensitive spotting analysis, cells were grown overnight in YEPD liquid media at 30°C, and the culture was diluted 1,000 fold and regrown in 5 ml fresh media to reach mid-log phase (approximately 3–4 × 10⁷ cells/ml). Cells were then diluted 10-fold serially and spotted at 3 μl volume in rows on YEPD plates containing various concentrations of selected chemical reagents. The plates were incubated for 3 days at 30°C and then photographed.

Fluorescence microscopy

Fluorescent images were acquired with Leica DMi8 automated microsystems. LAS X software was used for the image analysis. To measure the amount of DSB lesions and genomic instability of WT and mutant strains, the percentage of cells with subnuclear Rad52-GFP foci formation in the population was determined after 1 h PHL treatment. At least 300 individual cells for each genotype were counted manually.

Analysis of cell cycle progression

To examine the cell cycle arrest of WT and mutant strains, exponentially growing yeast cells were analyzed by flow cytometry using the NovoCyte 2000 (Acea Biosciences). Yeast cells were cultured overnight in 5 ml of YEPD at 30°C. On the following day, the cells were diluted in 5 ml of fresh media to an OD₆₀₀ of 0.4 and incubated with shaking for 3 h. Cells were harvested and fixed with 70% ethanol for 1 h at 25°C, followed by sequential treatment with RNase A for 1 h, proteinase K for 30 min, and propidium iodide (PI) for 30 min. Before analysis, the mixture was sonicated for 5 sec at low power (Haase and Lew, 1997). For the control of G1- and G2/M-arrested conditions, cells were treated with 100 μM α-factor and 15 μg/ml nocodazole for 4 h, respectively. At least 1 × 10⁵ cells were analyzed for each sample.

Deletion of ALF1 causes abnormal cytoskeletal morphology and caffeine sensitivity

To obtain a novel genetic factor involved in caffeine tolerance, we performed a genome-wide screening of S. cerevisiae non-essential gene deletion library for mutants displaying impaired growth on caffeine-containing plates. We isolated 192 candidate mutants and grouped them by several functional categories such as drug transport, cell cycle regulation, DNA damage repair, and autophagy, and found that some of them (snq2, mec1, rad53 mutants, etc.) are consistent with the previous reports in their caffeine sensitivity, verifying our experimental conditions were properly operated (Tsabar et al., 2015; Choi et al., 2022). Among them, we selected a mutant lacking Alf1, responsible for microtubule formation, for further investigation, due to implausible relatedness of cytoskeleton maintenance to caffeine resistance.

The ALF1 encodes α-tubulin binding protein involved in post-chaperonin tubulin folding, and alf1 null mutant is known to sensitive to anti-microtubule drug benomyl (Tian et al., 1997). Alf1 has a cytoskeleton-association protein-glycine-rich (CAP-Gly) domain, a highly conserved protein-interaction module implicated in maintenance of cell polarity and microtubule networks (Miller et al., 2006; Steinmetz and Akhmanova, 2008; Fig. 1A). We analyzed phenotypes of alf1 mutant together with nip100 and bik1 mutants since Nip100 and Bik1 have CAP-Gly domains as well (Andrieux et al., 2019). We also tested them together with pac2, rbl2, and cin1 mutants that are defective in tubulin heterodimer formation required for normal microtubule function (Feierbach et al., 1999).

The alf1 mutant exhibits an aberrant morphology, characterized by a lemon-like apiculate form (Fig. 1B). This unusual shape seems not to reflect cytoskeletal abnormality since all the other tubulin-related mutants show normal cell shape as WT strain. Given that Alf1 and Pac2 are known to act in the same pathway leading to functional α-tubulin, Alf1 might have separate role in cell morphology (Feierbach et al., 1999). The severe sensitivity to caffeine was observed only in alf1 mutant, but not in the other mutants, suggesting that Alf1 has an exclusive novel function involved in caffeine resistance other than microtubule formation (Fig. 1C).

Alf1 genetically interacts with DSB repair pathway

Since alf1 mutant shows unusual caffeine sensitivity and caffeine is known to impair several steps in DSB repair pathway, we evaluated whether ALF1 is involved in HR-mediated DNA damage repair by spot plating analysis with rad51 mutant cells (Fig. 2). As rad51 mutant is slightly sensitivity to caffeine, rad51 alf1 double mutant shows more sensitivity to caffeine compared to alf1 mutant. Surprisingly, however, rad51 alf1 double mutant exhibits highly synergistic sensitivity to several DSB-inducing reagents such as hydroxyurea, phleomycin, camptothecin, MMS, and 4NQO, compared to rad51 single mutant, whereas alf1 mutant itself shows almost no harmful growth defect to any of those drugs (Fig. 2). This observation implies that genetic interaction exists between tubulin folding cofactor and DNA damage repair pathway.

Fig. 2. Alf1 genetically interacts with Rad51 in response to caffeine and various DSB-inducing reagents. 10-fold serial dilutions of yeast WT and mutant strains were spotted on YEPD plates in the presence of indicated drugs and incubated for 3 days at 30°C. For the exposure to DSB-generating reagents, hydroxyurea (HU), phleomycin (PHL), methyl methanesulfonate (MMS), camptothecine (CPT), or 4-nitroquinoline-Noxide (4-NQO) was added in the plates before spotting.

Alf1 deficiency impedes DSB repair in the absence of Rad51

To clarify whether the synergistic effects of caffeine sensitivity and impaired growth of rad51 alf1 double mutant with DNA damage-inducing chemicals are due to failure in genome maintenance, we visualized nuclear foci of GFP-fused Rad52 protein, involved in DSB repair stimulating Rad51-mediated strand exchange, by fluorescence microscopy and assessed the proportion of foci-containing cells (Fig. 3A).

Fig. 3. Alf1 deficiency aggravates DNA damage in the absence of Rad51 and stimulates puntuate foci formation. DSB-induced Rad52-GFP foci were photographed in WT, rad51, alf1, and rad51 alf1 mutant cells treated with or without 10 μg/ml PHL for 1 h. White scale bar corresponds to a length of 10 μm. BF, bright field. (B) The proportion of cells containing Rad52-GFP foci was measured and graphed. At least 300 individual cells for each genotype were counted manually. Averages from three independent experiments are presented. (C) The proportion of cells showing multiple distinct foci in the nucleus with 10 μg/ml phleomycin treatment was measured and graphed.

In WT cells, only approximately 5% exhibited spontaneous Rad52 foci, which increased to ~35% following treatment with 10 μg/ml phleomycin for 1 h (Fig. 3B). The alf1 mutant cells showed almost no difference with WT cells. In contrast, rad51 mutant cells exhibited about threefold increase (~15%) in spontaneous foci compared to WT, whereas reached ~70% when treated with phleomycin. The rad51 alf1 double mutant cells showed even higher elevation in spontaneous Rad52 foci (~80%) than WT under DSB-inducing conditions.

Notably, cells lacking Alf1 tend to form punctate foci, characterized by multiple distinct dots in the nucleus, implying that these cells experience more severe DNA breaks at several different loci remaining unrepaired (Fig. 3C). The proportions of punctate foci were just less than 1% in WT, and about 2% in rad51 mutant, but increased to ~5% in alf1 mutant with 10 μg/ml phleomycin treatments. Surprisingly, the number of cells with punctate foci synergistically increases up to ~12% when both Rad51 and Alf1 are absent. This strongly suggests that Alf1 is involved in the prevention or repair of sporadic and/or induced DNA damages, thereby contributing to genomic stability in the nucleus, and that it might play an even more significant role in the absence of Rad51 in chromosome maintenance.

Alf1 deficiency induces DNA damage checkpoint activation

Next, we examined cell cycle progression of WT and mutant strains in exponential growth using flow cytometry. Given that Alf1 is involved in the prevention or proper repair of damaged DNA, cell cycle progression might be stalled before cell division due to G2/M DNA damage checkpoint activation (Cuddihy and O’Connell, 2003).

In contrast to WT, which shows an occupancy ratio of approximately 30% in G2/M phase, the rad51 mutant exhibits a significantly increased percentage of G2/M population, approximately 60% (Fig. 4A and B). Interestingly, both the alf1 mutant and the rad51 alf1 double mutant show an even higher ratio of G2/M population, up to 70%, which is close to that of nocodazole-induced arrest in G2/M, particularly at the expense of G1 cell population. However, the occupancy ratio of each cell cycle phase in cells treated with 50 mM caffeine is not significantly different from that of untreated cells. These results suggest that Alf1 plays a crucial and unexpected role in preventing the accumulation of unrepaired DSB lesions and the activation of damage checkpoint signaling to maintain genome stability.

Fig. 4. Alf1 deficiency induces G2/M phase cell cycle arrest. (A) FACS analysis demonstrates that cell cycle progressions of rad51, alf1, and rad51 alf1 mutant strains are mostly arrested at G2/M phase with or without 50 mM caffeine treatment. WT cells were treated with 100 μM α-factor or 15 μg/ml nocodazole as controls for G1- or G2-arrest of the yeast cell population, respectively. (B) The graph shows the percentage of cells in each phase of the cell cycle: G1, S, and G2/M measured in (A).

The mechanisms by which caffeine inhibits DNA damage repair, leading to mutagenesis and carcinogenesis, are of great interest and have long been investigated. Research on the inhibitory effects of caffeine on various DNA damage repair pathways has revealed several pivotal proteins involved in genome stability, such as DNA photolyase, UvrA, Sae2, Dna2, and Rad51 (Selby and Sancar, 1990; Tsabar et al., 2015; Chung, 2021). Here, we present novel findings that Alf1, the tubulin-assembling element responsible for microtubule formation, is also highly involved in DNA damage repair process in response to caffeine treatment.

Although Alf1 and Pac2 together are involved in the pathway for functional α-tubulin formation, and Rbl2 and Cin1 are responsible for β-tubulin, only Alf1 exhibits distinct roles in cell morphology, caffeine tolerance, and DNA damage repair. Additionally, among various cytoskeleton-associated proteins with the CAP-Gly domain, only Alf1 demonstrates unique phenotypes related to genomic stability (Fig. 1).

More interestingly, spotting analysis reveals that the unusual sensitivity of the alf1 mutant to caffeine is not reproduced at all with various DSB-inducing agents, such as PHL, MMS, CPT, and 4NQO (Fig. 2). This implies that the repair defect of DNA damage caused by caffeine is completely different from that caused by other drugs. Additionally, from the observation that the alf1 mutant alone is not sensitive at all to various DSB-inducing agents, while the rad51 alf1 double mutant is even more sensitive than the rad51 single mutant, we might infer that Alf1 plays a redundant role in genome stability, especially in the absence of Rad51.

Alf1 is primarily known to localize to cytoplasmic microtubules, contributing to tubulin biogenesis and regulation in yeast cells. Feierbach et al. (1999) reported that no specific fluorescent signal was observed when Alf1-GFP was expressed at endogenous levels, while overexpression of Alf1-GFP revealed its presence throughout the cytoplasm. However, based on our observation that Alf1 has a genetic interaction with Rad51, a crucial protein required for DSB repair in the nucleus, we propose that Alf1 might also reside in the nucleus, playing an additional role in the maintenance of genome stability, particularly in the absence of Rad51 (Fig. 3).

Taken together, the work presented in this paper raises an important question about the possible crosstalk between two distinct cellular pathways: DNA damage repair and cytoskeletal regulation, in maintaining cellular homeostasis in response to caffeine toxicity. In addition, why only Alf1, unlike other proteins involved in tubulin dynamics, plays a crucial role in caffeine resistance and DNA damage repair needs to be further elucidated.

카페인은 널리 소비되는 메틸잔틴 유도체로, 균류, 식물 및 인간을 포함한 여러 생물체에 걸쳐 다양한 생물학적 영향을 미친다. 효모에서 카페인은 DNA 손상 복구 경로와 세포 주기 체크포인트 신호를 방해하여 DNA 이중 가닥 절단 수리에 실패함으로써 세포 독성을 유도한다. 이 연구는 효모 결손 돌연변이체 라이브러리 스크리닝을 통해 카페인 내성과 관련된 새로운 요인으로 알파 튜불린 접힘 단백질을 암호화하는 ALF1 유전자를 확인하였다. alf1 돌연변이체는 비정상적인 세포 골격 형태와 카페인에 대한 높은 민감성을 나타내어, Alf1이 미세소관 형성 이외에 카페인 저항성에 독점적인 기능을 가지고 있음을 시사한다. 유전적 상호작용 연구는 ALF1의 결핍이 rad51 돌연변이체의 DSB 유도제에 대한 민감성을 악화시킴을 밝혀, 유전체의 안정성을 유지하는데 시너지 역할을 한다는 것을 보여준다. alf1 및 rad51 alf1 이중 돌연변이체에서 Rad52-GFP 초점의 형성은 Alf1이 DNA 손상 복구에 관여하고 있음을 더욱 뒷받침한다. 유세포 분석 결과, Alf1의 결핍이 DNA 손상 체크포인트를 활성화하여 G2/M 단계에서 세포 주기 진행을 멈추게 한다는 것이 밝혀졌다. 이러한 발견은 세포 골격 조절과 유전체 안정성 사이에 새로운 유전적 상호작용이 존재함을 보여주며, 세포의 생존을 위한 협력 메커니즘에 대한 통찰력을 제공하고, 특히 Rad51이 없는 경우 Alf1이 유전체 보존에 있어 특별히 중요한 역할을 한다는 것을 암시한다.

This work was supported by Priority Research Centers Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (NRF-2016R1A6A1A03007648 and RS-2023-00275966).

The authors have no conflict of interest to report.