Knocking out GCN5L1 impaired surfactant production in MLE-12 cells

To investigate the potential function of GCN5L1 in surfactant biology, we examined its expression in the developmental lung using the Lung Gene Expression Analysis (LGEA) Web Portal database. We found that GCN5L1 is expressed in the mouse lung at all stages before and after birth (Additional file 1: Fig. S1A). Its expression was detected in different kinds of lung cells, including alveolar type II epithelial cells (Additional file 1: Fig. S1B). This finding was consistent with those of previous studies in which GCN5L1/BLOS1 and other components within the BLOC-1 complex were demonstrated to be widely expressed but with cell type-specific functions [14]. Next, we generated GCN5L1-mutant MLE-12 cells with the CRISPR/Cas9 system. Three targets for Cas9 were designed, and one with high efficiency, as evidenced by T7EI assay followed by sanger sequencing, was located near the start codon (T3, reverse strand) and used for subsequent experiments (Additional file 2: Fig. S2A-C). Multiple mutant clones were established (Additional file 2: Fig. S2D, Additional file 3: Fig. S3). M2 had a 43-bp deletion in one allele and a 146-bp deletion in the other; the M29 clone had 1-bp addition in one allele and a 4-bp deletion in the other (Additional file 2: Fig. S2D, Additional file 3: Fig. S3). We chose M2 and M29 for subsequent experiments, the mRNA levels were significantly decreased (Additional file 4: Fig. S4A), possibly through the nonsense-mediated decay mechanism. The successful KO of the GCN5L1 protein in these two mutant lines was also confirmed through western blotting (Additional file 4: Fig. S4B). Additional clones (M2, M8, M16, M20, M22, and M27) that contained nonframeshift mutations or WT sequences were discarded (Additional file 2: Fig. S2D, Additional file 3: Fig. S3).

To assess the influence of GCN5L1 deletion on MLE-12 cells, we first tested cell viability using an MTT assay. The results showed no obvious effect on cell viability after GCN5L1 disruption (Additional file 4: Fig. S4C). Then, we analyzed the influence of GCN5L1 deficiency on surfactant production. The coupled enzymatic reactions method was used to quantify phospholipids, revealing significantly reduced phospholipid release from GCN5L1-mutant versus WT cells (Fig. 1A). Consistently, a comparative lipidomic analysis of the culture supernatant demonstrated a significant reduction in secreted phospholipid species from GCN5L1-mutant cells (Additional file 5: Fig. S5). The ELISA analysis also showed a significant reduction in released SP-B and SP-C in the culture supernatant from GCN5L1-mutant cells (Fig. 1B). Moreover, restoring GCN5L1 expression in M2 cells restored the level of released surfactant proteins and lipids (Fig. 1C, D). These results suggest that GCN5L1, similar to its function in regulating swimbladder surfactant in zebrafish, is involved in regulating lung surfactant production in mammalian cells.

Fig. 1

GCN5L1 gene KO impaired surfactant production in MLE-12 cells. A Phospholipid secretion in WT and GCN5L1-KO MLE-12 cells. B ELISAs of secreted SPs in the culture supernatant of WT and GCN5L1-KO MLE-12 cells. C Reconstruction of GCN5L1 expression repaired phospholipid secretion in GCN5L1 mutant cells. D Reconstruction of GCN5L1 expression repaired SP-B and SP-C secretion in GCN5L1 mutant cells. The results are expressed as the mean ± SD of three independent experiments; **P < 0.01; t-test

Disruption of GCN5L1 influenced the expression of genes associated with surfactant biology

As demonstrated in our previous work, GCN5L1/BLOS1 mutation resulted in the profound disruption of gene expression in the zebrafish swimbladder and hampered normal postembryonic development of the epithelium [13]. Hence, we performed RNA-seq to examine changes in the transcription profile after GCN5L1/BLOS1 disruption in MLE-12 cells. DEGs between control (WT) and M2 mutant cells were extracted (Additional file 6: Fig. S6A), and GO and KEGG enrichment analyses were conducted. Among the downregulated genes, the top enriched GO terms included “multicellular organism development,” “cell adhesion,” “cell differentiation,” and “growth factor binding” (Additional file 6: Fig. S6B). Regarding the upregulated genes, enriched GO terms included “toll-like receptor 4 signaling pathway,” and “positive regulation of transforming growth factor beta production” (Additional file 6: Fig. S6C). In the KEGG analysis, the DEGs were mostly enriched in “focal adhesion,” “ECM-receptor interaction,” and “TGF-beta signaling pathway” (Additional file 6: Fig. S6D-E). These enriched terms could function in maintaining epithelial characteristics and/or AT II cell specific functions.

To further investigate the surfactant defect in the mutant cells, we mined DEGs related to surfactant biology. The expression levels of surfactant protein genes Sftpb and Sftpc and the lipid transporter Abca3 were reduced in M2 cells (Additional file 12: Table S1). However, no Sftpa1 or Sftpd expression was detected in WT or mutant cells (Additional file 12: Table S1). Since SP-B, SP-C, and Abca3 are LB components, we further examined the expression of other LB proteins by merging the LB proteome with our DEGs [15, 16], revealing more downregulated LB genes, including Alpl, Aldoc, Aldh3b1, C3, Ptgfrn, Vnn1, Susd2, Xpnpep, and Gsn (Additional file 12: Table S1). Q-PCR confirmed the dysregulation of Sftpb, Sftpc and Abca3, and some other LB genes in M2 and M29 cells (Fig. 2A). Additionally, we examined the mRNA expression of some transcription factors involved in surfactant production [17,18,19,20,21]. No significant changes in the expression levels of Nkx2-1, Foxa2, Gata6, Nfatc3, Stat3, Srebf1, and Srebf2 were observed after GCN5L1 deletion (Additional file 13: Table S2). However, the expression level of Cebpα was significantly decreased, as demonstrated by RNA-seq (Additional file 13: Table S2) and confirmed by q-PCR and western blotting (Fig. 2B, C). This decline in Cebpα expression was restored in lentivirus-mediated GCN5L1-reconstructed M2 cells (Fig. 2D). Moreover, lentivirus mediated overexpression of both Cebpα and GCN5L1 significantly restored the downregulated mRNA levels of Sftpb, Sftpc and other LB-associated genes in GCN5L1 mutant cells (Fig. 2D, E). In summary, these results suggest that GCN5L1 deletion dysregulates genes involved in surfactant biology, and Cebpα is a crucial associated regulator.

Fig. 2

Disruption of GCN5L1 altered the expression of Cebpα and surfactant-related genes. A Q-PCR verification of the surfactant-related DEGs in GCN5L1 mutant clones. B Q-PCR detection of Cebpα mRNA expression in WT and GCN5L1-KO MLE-12 cells. C Representative immunoblot image of Cebpα protein expression levels in WT and GCN5L1-KO MLE-12 cells. D Representative immunoblot images of Cebpα protein levels in WT cells, M2 cells, M2 cells infected with lentivirus expressing exogenous Cebpα (M2 + Cebpα), and M2 cells infected with lentivirus expressing exogenous GCN5L1 (M2 + GCN5L1). E Q-PCR assays of the RNA levels of surfactant-related genes in WT cells, M2 cells, and M2 cells infected with lentivirus expressing exogenous Cebpα (M2 + Cebpα), and M2 cells infected with lentivirus expressing exogenous GCN5L1 (M2 + GCN5L1). The results are expressed as the mean ± SD of three independent experiments; ns, not significant;*P < 0.05, **P < 0.01; t-test

Disruption of GCN5L1 altered the activity of the ROS-ERK-Foxo1-Cebpα axis in MLE-12 cells

To explore the upstream mechanism underlying Cebpα dysregulation, we first examined the protein levels of Nkx2-1 and Foxa2, both of which are known Cebpα regulators [21]. Consistent with the mRNA findings (Additional file 13: Table S2), no significant changes in the expression levels of these two proteins were detected in GCN5L1 mutant cells (Additional file 7: Fig. S7A). We next used the STRING database to search for potential Cebpα interactors. Ten primary interactors were found (Additional file 7: Fig. S7B), among which Foxo1 attracted our attention since a previous study revealed that the ROS-ERK-Foxo1 axis is dysregulated in GCN5L1-deficient hepatocytes [22]. Moreover, Foxo1 could interact with Cebpα and influence its expression [23, 24]. Thus, we examined the activity of this axis. While Foxo1 mRNA expression levels were similar (Fig. 3A), the protein levels in mutant cells were significantly reduced (Fig. 3B). ERK activity was also analyzed; consistently, ERK was found to be activated, as indicated by increased p-ERK levels (Fig. 3C). ROS level elevation after GCN5L1 deletion has been reported in different circumstances [7, 25,26,27]. As expected, the ROS levels were also increased in mutant MLE-12 clones (Fig. 3D). Moreover, the activity of the ROS–ERK–Foxo1 axis recovered in GCN5L1-reconstructed M2 cells (Fig. 3B–D). Together, these data indicate that alteration of ROS–ERK–Foxo1–Cebpα axis activity occurs following GCN5L1 depletion in MLE-12 cells.

Fig. 3

Disruption of GCN5L1 altered the activity of the ROS–ERK–FOXO1 axis in MLE-12 cells. A Q-PCR detection of Foxo1 mRNA expression in WT, GCN5L1-KO MLE-12 cells. B Representative immunoblot image of Foxo1 protein expression levels in WT, GCN5L1-KO MLE-12 cells and M2 cells infected with lentivirus expressing exogenous GCN5L1 (M2 + GCN5L1). C Representative immunoblot images of ERK and p-ERK protein expression levels in WT, GCN5L1-KO MLE-12 cells and M2 cells infected with lentivirus expressing exogenous GCN5L1 (M2 + GCN5L1). D Cellular ROS levels of WT, GCN5L1-KO MLE-12 cells, and M2 cells infected with lentivirus expressing exogenous GCN5L1 (M2 + GCN5L1). Cells were incubated with CM-H2DCFDA and measured by FACS; mean fluorescence intensity was normalized to WT group. E Representative immunoblot images of ERK, p-ERK, Foxo1, and Cebpα protein expression levels in WT cells, M2 cells treated with DPI (M2 + DPI) or SCH772984 (M2 + SCH772984), and M2 cells infected with lentivirus expressing exogenous Foxo1 (M2 + Foxo1). F Q-PCR assays of the RNA levels of surfactant-related genes in WT cells, M2 cells treated with DPI (M2 + DPI) or SCH772984 (M2 + SCH772984), and M2 cells infected with lentivirus expressing exogenous Foxo1 (M2 + Foxo1). The results are expressed as the mean ± SD of three independent experiments; ns, not significant; *P < 0.05,**P < 0.01; t-test

To assess whether alteration of the ROS–ERK–Foxo1–Cebpα axis was responsible for the surfactant defect in GCN5L1 mutant cells, we assessed the effect of modifying different components within the axis. As shown in Fig. 3E, reducing ROS with diphenylene iodonium (DPI) suppressed ERK activation and increased Foxo1 expression in M2 cells. Inhibiting ERK activity with SCH772984 also restored Foxo1 levels. These two modifications together with lentivirus mediated overexpression of Foxo1 all significantly promoted Cebpα protein expression and partially restored the expression of most dysregulated surfactant related genes (Fig. 3E, F). However, unexpectedly, the effect of these rescue experiments and Cebpα overexpression did not significantly affect surfactant production (Additional file 8: Fig. S8A-D). Collectively, these results imply that the ROS–ERK–Foxo1–Cebpα axis is involved in the misregulation of some surfactant related genes in GCN5L1-mutant cells.

Given that GCN5L1 is repeatedly reported to be a mitochondrial enriched protein [7,8,9,10,11,12], we tested its localization in MLE-12 cells. Consistent with previous reports, we observed strong colocalization of GCN5L1–FLAG with the mitochondrial marker TOM20 in MLE-12 cells (Fig. 4A, B). Thus, we tried to restore the GCN5L1-KO defect through the direct reconstruction of GCN5L1 expression in mitochondria. Unexpectedly, unlike GCN5L1 reconstruction (Fig. 1C, D), mito-GCN5L1 reconstruction failed to significantly restore surfactant release (Additional file 8: Fig. S8E–F); however, it rescued the expression level of dysregulated genes in mutant cells (Fig. 4C), indicating a specific role of mitochondria-localized GCN5L1 in the regulation of these genes.

Fig. 4

Mitochondria-localized GCN5L1 is responsible for the regulation of surfactant-related genes. A Colocalization of GCN5L1–FLAG with TOM20 in MLE-12 cells. Cells were cotransfected with TOM20- and GCN5L1–FLAG-expressing plasmids before staining with anti-TOM20 and anti-FLAG antibodies. B The colocalization scatter plot of anti-TOM20 and anti-GCN5L1–FLAG signals. Pearson correlation coefficient is 0.78, Manders’ overlap coefficient is 0.82. C Q-PCR assays of the RNA levels of surfactant related genes in WT cells, M2 cells, and M2 cells infected with lentivirus expressing mito-GCN5L1 (M2 + mito-GCN5L1).The results are expressed as the mean ± SD of three independent experiments; *P < 0.05, **P < 0.01; t-test

Disruption of GCN5L1 impaired lamellar body biogenesis and trafficking in MLE-12 cells

The above findings do not explain the decreased release of surfactant from mutant cells. Surfactant lipids and proteins are transported and stored in LBs before secretion. Moreover, a previous study demonstrated that GCN5L1/BLOS1 is a subunit of the BORC protein complex that plays a role in lysosomal organelle positioning/trafficking [5, 28,29,30]. Since the LB is an LRO and shares some common features with lysosomes [31], we explored potential LB positioning/trafficking defects after GCN5L1 mutation. Lysotracker Red was used to stain LB-like organelles in MLE-12 cells [32,33,34]. Strikingly, while these lysotracker-positive organelles were scattered in WT MLE-12 cells, they were perinuclearly clustered in GCN5L1 mutant M2 and M29 cells (Fig. 5A, Additional file 9: Fig. S9A). This phenotype was rescued in M2 cells when the expression of GCN5L1 was reconstructed (Fig. 5A, Additional file 9: Fig. S9A). The colocalization of these signals within ABCA3–EGFP vesicles indicated they were likely LBs (Fig. 5D, Additional file 9: Fig. S9D). To further confirm the identity of these organelles, we performed metabolic labeling with BODIPY phosphatidylcholine (a probe that undergoes native-like transport and metabolism in cells). As shown in Fig. 5B, these organelles were also lipid-enriched, similar to LBs in primary AT2 cells. BODIPY-labeled vesicles were also ABCA3–mCherry positive and showed accumulation in mutant cells (Fig. 5B, E, Additional file 9: Fig. S9B, E). Endogenous Lamp1, a frequently reported marker for LBs, was found to colocalize well with ABCA3–EGFP (Fig. 5C, F). Similarly, these Lamp1-positive structures also accumulated in GCN5L1 mutant cells (Fig. 5C, F, Additional file 9: Fig. S9C, F). Because autophagosomes are also stained by Lysotracker [35], and previous studies have revealed abnormal autophagosome accumulation in GCN5L1 deleted cells [36, 37], we investigated whether the Lysotracker Red stained organelles were or contained autophagosomes. Transient transfection of a GFP–LC3-expressing plasmid failed to label these organelles in GCN5L1 mutant cells (Additional file 10: Fig. S10A), suggesting they were not autophagosomes. Immunoblotting with LC3 antibody also showed no obvious alteration in the LC3 II/I ratio after GCN5L1 knockout (Additional file 10: Fig. S10B). Collectively, these results indicate that LB-like vesicles accumulate in MLE-12 cells after GCN5L1 disruption.

Fig. 5

Impairment of LB positioning/trafficking in GCN5L1-KO cells. A Lysotracker Red staining of WT, GCN5L1-KO clones (M2 and M29), and GCN5L1 reconstructed mutant cells (M2 + GCN5L1). B WT and M2 cells were metabolically labeled with BODIPY phosphatidylcholine. C WT and M2 cells were immunostained with Lamp1 antibody. D–F WT and M2 cells were transfected with h-ABCA3–EGFP or ABCA3–mCherry before staining with Lysotracker Red (D), BODIPY phosphatidylcholine (E), and Lamp1 antibody (F)

The size and number of these organelles differed between WT and mutant cells. Careful quantification revealed found that while the number of these LBs increased in mutant cells, their size significantly decreased (Fig. 6A–C). This phenotype indicated an LB biogenesis defect and is consistent with the downregulation of multiple LB-related genes in mutant cells (Fig. 2A). Interestingly, we found that despite the lack of increased surfactant release (Additional file 8: Fig. S8E–F), the mito-GCN5L1 expressing M2 cells had normal-sized LBs, indicating a direct role of mito-GCN5L1 in LB biogenesis (Fig. 6A, C). These restored LBs in mito-GCN5L1-expressing cells failed to efficiently transport at the same level as those in WT or GCN5L1 reconstructed M2 cells, resulting in accumulation within cells (Fig. 6A, B). Considering that no significant increase in surfactant release was observed, the relatively smaller number of LBs in mito-GCN5L1-expressing cells than that in M2 cells could have resulted from incomplete counting since accumulated large LBs partially overlapped in the images. Next, we performed TEM analysis on WT and M2 cells to better examine these organelles. As shown in Fig. 6D and E, MLE-12 M2 cells contained more but smaller LB-like organelles containing numerous membranous structures potentially constituting cell surfactant. However due to the relatively low activity of the surfactant biogenesis machinery in these cells, these membranous structures did not fill the LBs. Collectively, these results indicate that the disruption of GCN5L1 impaired LB biogenesis and positioning in MLE-12 cells. Additionally, the GCN5L1 signal clearly colocalized with the endogenous LB marker Lamp1 (Fig. 7, Additional file 11: Fig. S11), indicating a possible direct role of GCN5L1 in regulating the trafficking of these organelles.

Fig. 6

Abnormal size and number of LB-like organelles in GCN5L1-KO cells. A Representative images showing Lysotracker-stained organelles in WT, M2, M2 + GCN5L1, and M2 + mito-GCN5L1 cells. B, C The numbers and size (area) in WT, M2, M2 + GCN5L1, and M2 + mito-GCN5L1 cells; **P < 0.01 (Kolmogorov–Smirnov test). D Transmission electron micrographs of WT and M2 cells. Many LB-like organelles are present within the cytoplasm; arrows point to lamellar-like materials in the organelles. E, F The diameters and numbers of LBs in the TEM images were measured and counted with Image J. The results are expressed as the mean ± SD of four independently measured data; *P < 0.05; t-test

Fig. 7

Localization of GCN5L1 in LBs. Cells were transfected with GCN5L1–FLAG-expressing plasmid and stained with anti-Lamp1 and anti-FLAG antibodies

Disruption of GCN5L1 resulted in the accumulation of surfactant proteins within MLE-12 cells

To investigate the consequences of LB defects on surfactant production, especially surfactant protein secretion, we performed western blotting to examine the endogenous protein levels of SP-B and SP-C. Although mature SP-C was not detected in either control or mutant cells (data not shown), we detected a striking ~ 18KD band, corresponding to the SP-B dimer [38,39,40], with two different SP-B antibodies (Fig. 8A, B). While the expression of this band was weak in WT and GCN5L1-reconstructed M2 cells, it was strong in GCN5L1-mutant cells, implying accumulation of this protein within mutant cells. We further visualized the accumulated proteins within cells directly through immunofluorescence. Neither endogenous SP-B nor SP-C could be detected with our antibodies. However, when the cells were pretransfected with SP-B- or SP-C-expressing plasmids, we detected more perinuclear signals located within ABCA3-positive organelles in mutant versus control cells (Fig. 8C, D). Collectively, these results indicate that defects resulting from GCN5L1 deletion caused the accumulation of surfactant proteins within LBs in MLE-12 cells.

Fig. 8

Accumulation of SP proteins in GCN5L1-KO cells. A Immunoblot image of WT, M2, and GCN5L1 reconstructed M2 + GCN5L1 cells with an SP-B antibody (Santa Cruz). B Immunoblot image of WT, M2, and GCN5L1 reconstructed M2 + GCN5L1 cells with another SP-B antibody (Abcam). C WT and M2 cells were cotransfected with h-ABCA3–mCherry and m-Sftpb expressing plasmids and immunostained with an SP-B antibody (Abcam). D WT and M2 cells were cotransfected with h-ABCA3–EGFP- and m-Sftpc-expressing plasmids and immunostained with an SP-C antibody (Seven Hills Bioreagents). **P < 0.01; t-test