Comparison of human PEX knockout cell lines suggests a dual role of PEX1 in peroxisome biogenesis

Generation of a PEX-KO cell line collection

For the generation of PEX-KO cell lines, we applied the CRISPR/Cas9 technique (Jinek et al. 2012). After designing specific RNAs that target the respective genes, we have used them to guide the Cas9 endonuclease, which then created a double strand break (DSB) at the desired location (Figure 1A). Because the repair of the DSB by non-homologous end-joining leads to different insertions or deletions (indels) on each allele, we amplified the respective gene area by PCR and after sequencing used the TIDE (tracking of indels by decomposition) software to determine mutations in mixed DNA samples (Brinkman et al. 2014). In parallel, we cloned and sequenced 20 individual DNA molecules to obtain the actual sequence of the shortened expression products (Figure 1A). Both methods allowed verification of the successful knockout.

Figure 1:

Procedure of creating a PEX-KO collection. (A) Chimeric single-guided RNAs (sgRNAs) were used to target the respective genes and guide the Cas9 endonuclease. Cas9 creates a double strand break (DSB), which is repaired by non-homologous end-joining. To identify the resulting insertions or deletions (indel) the respecting gene area was amplified by PCR and after sequencing a TIDE (tracking of indels by decomposition) analysis was performed, which determines indel mutations in mixed DNA-sequences. In parallel 20 single DNA-molecules were cloned and sequenced for knockout verification. (B) The chimeric RNA contains a 20-nucleotide sequence required for target DNA binding of Cas9 (guide RNA), while the other part of the chimeric RNA orientates the RNA in a proper secondary structure enhancing binding of Cas9 (scaffold RNA). The 20-nucleotide sequence is complementary to the area of the gene to be targeted, here exon1 of PEX1, and requires an adjacent GG dinucleotide-containing PAM sequence. (C) Indel spectrum of the PEX1-KO cell line determined by TIDE. The dissolution of the mixed DNA sequences showed that the repair of the DSB on every allele of exon1 of PEX1 led to the deletion of 7, 8 and 10 bp. The deletions occurred in similar frequencies (y-axis). (D) In case of PEX1, the repair of the DSB led to the deletion of 7, 8 and 10 base pairs (bp) on the respective alleles, which result in frame-shift mutations and significant shortening of the expression products. (E) Indel mutations of all generated PEX-KO cell lines. For some cell lines the repair of the DSB led to the indel of the same number of base pairs. PEX5-and PEX14-KO cell lines were generated by a different procedure leading to indels of whole exons (Galiani et al. 2022). Some chromosomes are present in three copies.

Figure 1B shows a section of exon1 of the PEX1 gene with the labelled guide sequence as an example. In case of the generation of the PEX1-KO, the TIDE-analysis revealed that the repair of the DSB led to the deletion of 7, 8 or 10 base pairs (Figure 1C). Since HEK293 cells, from which the HEK T-REx293 cell line derived, contain varying chromosome numbers (Binz et al. 2019), the analysis detected three alleles which have been repaired individually. Cloning and sequencing of 20 individual DNA molecules confirmed the result from the TIDE-analysis and revealed expression products of neglectable length (Figure 1D). Following this procedure, we generated nine selected PEX-KO cell lines, which include at least one member of every peroxin-group: PEX3, PEX16 and PEX19 (peroxisomal membrane formation), PEX5 and PEX7 (matrix protein import receptors), PEX13 and PEX14 (docking complex), PEX10 (RING-finger complex), PEX1 (export complex). Figure 1E shows the indel mutations of all generated KO cell lines, while the respective sequences of the remaining expression products are presented in Supplementary Table S1.

Expression of the respective peroxin leads to complementation of peroxisomal defects

Microscopic inspection of T-REx293 cells and the PEX-KO cells revealed a punctate pattern of the artificial peroxisomal proteins EGFP-PTS1 (Figure 2A), whereas all generated KO cell lines except PEX7-KO exhibited an import defect for matrix proteins as indicated by the cytosolic mislocalization of the peroxisomal marker EGFP-PTS1 (Figure 2B). The PTS2-EGFP was used to test the PEX7-KO as PEX7 codes for the PTS2 receptor and its deficiency does only affect import of PTS2 proteins (Figure 2D, E), while deficiency in all other peroxins affects both import of both PTS1 and PTS2 protein. The observed mislocalization is typical for cell lines defective in peroxins of the peroxisomal import machineries and underlines the successful knock-out of the respective PEX genes. Cells with defective members of the membrane protein import machinery even lack detectable peroxisomal membranes, as indicated by the absence of the peroxisomal membrane marker PMP70 as seen for PEX3-KO (Figure 2B) or mislocalization of the peroxisomal membrane protein PEX14 to mitochondria as seen for PEX3-KO, PEX16-KO and PEX19-KO (data not shown). It has been identified that loss of peroxisomes leads to re-routing of several peroxisomal proteins, including PEX14 to mitochondria, which have been described as places for emergency landing for abandoned peroxins (Nuebel et al. 2021). The fuzzy punctate pattern seen for PMP70 in PEX19-KO cells most likely represents mislocalization or aggregation of protein.

Figure 2:

Peroxisomal defects in T-REx293 PEX-KO cell lines are complemented by transfection of the deleted PEX genes. Normal T-REx293 cells and PEX-KO cells were cultivated on coverslips and transfected with vectors encoding EGFP-PTS1 (A–C) or PTS2-EGFP (D–F) to monitor peroxisomal matrix protein import. In parallel, PEX-KO cell lines were transfected with vectors additionally coding for the corresponding deficient PEX gene to test for complementation of the PEX phenotype (C), (F). T-REx293, PEX1-, PEX5-, PEX7-, PEX10-, PEX13- and PEX14-KO cells were fixed 48 h and PEX3-, PEX16-, and PEX19-KO cells 96 h after transfection with formaldehyde. Peroxisomal membranes were labelled using αPMP70-antibodies, whereas nuclei were stained with DAPI. While the observed punctate pattern indicates the peroxisomal localization of the peroxisomal marker proteins EGFP-PTS1/PTS2-EGFP in T-REx293 cells, there is a cytosolic mislocalization of the marker proteins in PEX-KO cells. PEX3-KO cells even lack any detectable peroxisomal membranes as indicated by the absence of the marker PMP70. A fuzzy pattern in PEX19-KO cells indicates mislocalization or aggregation of the membrane marker PMP70. While most of the PEX16-KO cells are devoid of peroxisomal membranes, some cells exhibit enlarged import-competent peroxisomes. The plasmid-encoded expression of the respective peroxins led to the complementation of both membrane and matrix protein import defects. Scale bars: 10 µm.

The PEX16-KO cells show an unusual heterogeneous phenotype. Most of the cells are defective in peroxisomal import of matrix and membrane proteins as indicated by the cytosolic mislocalization of the peroxisomal matrix marker and mislocalization or depletion of peroxisomal membrane proteins, as indicated by absence and aggregates of the membrane marker PMP70 (Figure 2B, Supplementary Figure S1) and mislocalization of PEX14 to mitochondria (data not shown). However, there is a subpopulation of PEX16-KO cells, which contain peroxisomal structures that are less in number and enlarged when compared with normal peroxisomes (Figure 2B, Supplementary Figure S1). Interestingly, there is a colocalization with EGFP-PTS1 (Supplementary Figure S1), indicating that these peroxisomal structures can import matrix proteins, as recently has been described for PEX16-KO HeLa cells (Yagita et al. 2022). This mosaic phenotype suggests that although PEX16 is important for peroxisomal membrane biogenesis, some mammalian cells remarkably can still form peroxisomes in its absence.

To show that the effect on peroxisome number and targeting is a result of the genetic manipulation, we examined whether the peroxisomal protein import defects can be complemented by the plasmid-encoded expression of the respective peroxin. For all KO cell lines the expression of the respective peroxins restored not only the import of matrix proteins, but also induced the formation of peroxisomal membranes in case they were absent (Figure 2C, F), as seen for PEX3-KO. After plasmid-encoded expression of PEX16 in PEX16-KO cells, the enlarged peroxisomal structures vanished and peroxisomes of normal size and number were formed.

PEX-KO cell lines differ in the steady-state level of peroxisomal proteins

The immunoblot analysis of whole cell lysates of the generated PEX-KO collection showed that the absence of some peroxins affect the steady-state levels of other peroxisomal proteins (Figure 3). The deletion of PEX1 for instance results in a reduced PEX5 stability, which has already been observed in PEX1-deficient patient cell lines (Dodt and Gould 1996). The knockout of PEX14 also leads to an increased steady-state level of PEX13, while the absence of PEX5 destabilizes PEX14 as shown beforehand (Natsuyama et al. 2013). As new finding we observe that the deficiency in PEX1 also results in an increased PEX13 stability (Figure 3). To prove the significance of PEX13 stabilization in the absence of PEX1 we repeated immunoblot analyses of three independent clones (Supplementary Figure S2). Stabilization of PEX13 has been observed in Hansenula polymorpha upon deletion of PEX14 and RING-finger peroxins that seem to play a role in PEX13-degradation (Chen et al. 2018).

Figure 3:

PEX-KO cell lines show differences in steady-state levels of peroxisomal proteins. Whole cell lysates of normal T-REx293 and PEX-KO cells were prepared and analyzed by immunoblotting using indicated antibodies. While a deletion of PEX1 leads to a decreased PEX5-and an increased PEX13-stability, an increased PEX13-stability is observed in PEX14-KO cells. Some KO cell lines, especially PEX10-KO show residual import of the PTS2-protein thiolase (ACAA1) as indicated by abundance of mature thiolase (m-ACAA1), which migrates faster than precursor thiolase (p-ACAA1) during electrophoresis due to peroxisomal cleavage of PTS2.

Moreover, some KO cell lines show partial import of the PTS2-protein thiolase (ACAA1), as indicated by the presence of maturated thiolase. As maturation of thiolase occurs after peroxisomal import and is caused by cleavage of the targeting signal, the presence of the mature forms suggests that these proteins have been transported into the peroxisomal lumen (Figure 3). This is especially the case for PEX10-KO cells, which show a nearly similar ratio of mature-and premature thiolase. The PEX7-KO cell line is the only cell line, which not only shows no maturated thiolase at all but an increased amount of accumulated premature thiolase that exceeds the total thiolase amount in control or other KO cell lines (Figure 3).

PEX1-KO cells exhibit an increased number of PMP70-containing peroxisomal vesicles

Immunofluorescence microscopy of PEX1-KO cells revealed that they contained increased numbers of PMP70-and PEX14-labelled peroxisomal structures that also appeared to be of reduced size when compared to peroxisomes in control cells (Figure 4A). The plasmid-encoded expression of PEX1 in PEX1-KO cells resulted in complementation of this phenotype, demonstrated by a reduction in peroxisome numbers and enlargement of peroxisomes, indistinguishable from those in control cells. We then took a closer look at the influence of the deletion of other peroxins on the number of PMP70-labelled structures. To this end, quantification of whole cell peroxisome numbers in ghost-containing KO-cells was performed (Figure 4B). Most of the deletion cell lines showed a similar number of peroxisomal ghosts as normal HEK T-REx293 cells, which contained about 250 peroxisomes per cell, while PEX1-KO cells in average exhibit nearly twice as many peroxisomal vesicles (Figure 4B). The only other KO cell line with significantly increased peroxisomal numbers is the PEX10-KO, albeit it did not reach numbers as high as observed for PEX1-KO cells (Figure 4B).

Figure 4:

A deletion of PEX1 leads to an increased number of PMP70-labelled peroxisomal ghosts. (A) Representative fluorescence microscopy images of T-REx293 and PEX1-KO cells show that a deletion of PEX1 leads to a significantly increased number of PMP70-and PEX14-labelled peroxisomal ghosts, which appeared to be of decreased size. The expression of PEX1 restores peroxisomal size and number in PEX1-KO cells. Scale bar: 10 µm. (B) Peroxisomal number in T-REx293 and indicated KO cells was quantified from whole cell images after visualization of peroxisomal vesicles with αPMP70-antibodies using the spots function of IMARIS (BITPLANE Scientific Software). Data represent mean ± SD Peroxisomal number in PEX1-and PEX10-KO cells is significantly increased compared to T-Rex293 cells as analyzed by one-way ANOVA (n ≥ 2, ****p = 0.0001, **p = 0.0037). Per replicate at least three cells were quantified for each cell line. Total number of cells analyzed: T-REx293, 17; PEX1-KO, 16; PEX5-KO, 10; PEX7-KO, 10; PEX10-KO, 25; PEX13-KO, 17; PEX14-KO, 24.

A splicing variant of PEX1 cannot rescue the import defect of PEX1-KO cells but restores peroxisome abundance

For several peroxins isoforms exist that have not yet been characterized in detail. Hence, we used the PEX1-KO cell line to investigate known isoforms of PEX1. Beside the full-length PEX1, there is a second alternatively spliced, short isoform, which lacks amino acids 92–413 (UniProtKB Identifier O43933-2) (Figure 5A). The deletion is expected to affect the N-terminal domain (NTD), which exhibits phospholipid-binding activity (Shiozawa et al. 2006). However, the deletion does not concern the two AAA⁺-ATPase domains, which both are supposed to play critical roles for PEX1/PEX6 complex formation and extraction of the PTS1 receptor from the membrane (Ciniawsky et al. 2015; Saffert et al. 2017). To study the cellular function of the short isoform, we expressed PEX1 as well as PEX1(Δ92-413) in PEX1-KO cells (Figure 5B) and compared their complementing activities. While reintroduction of PEX1 fully restores the peroxisomal import defect of matrix proteins (Figures 2C and 5C), cells expressing PEX1(Δ92-413) still show cytosolic mislocalization of EGFP-PTS1 (Figure 5C). This implies that not only the AAA⁺-ATPase domains of PEX1 are essential for its role in receptor recycling, but also the N-terminal region.

Figure 5:

Functional analysis of an alternatively spliced PEX1 isoform. (A) Domain structure of the two known PEX1 isoforms. While both isoforms contain two AAA⁺-ATPase domains (D1 and D2) with two Walker motifs (WA/WB) each, the short isoform of PEX1 lacks a stretch of 321 amino acids within the N-terminal regions (corresponding to amino acids 92–413 in the long isoform). (B) Adherent T-REx293, non-transfected PEX1-KO cells and PEX1-KO cells transfected with bicistronic plasmids encoding either PEX1 or PEX1(Δ92-413) were lysed in SDS-PAGE buffer and analyzed by immunoblotting using antibodies against PEX1, GFP and GAPDH. (C) To test the ability of both PEX1 isoforms to complement the import defect of matrix proteins, both were expressed in PEX1-KO cells in addition to the artificial matrix protein EGFP-PTS1 and analyzed by immunofluorescence microscopy. While the expression of PEX1 restored the import defect of peroxisomal matrix proteins, leading to the peroxisomal localization of EGFP-PTS1, the expression of PEX1(Δ92-413) could not complement the cytosolic mislocalization of EGFP-PTS1. Scale bar: 10 µm. (D) Influence of both PEX1 isoforms on peroxisomal numbers. Quantification was performed from whole cell images using the spots function of IMARIS (BITPLANE Scientific Software). Data represent mean ± SD. While the expression of PEX1 in PEX1-KO cells led to a complete reduction of peroxisomal numbers reaching T-REx293 levels, PEX1(Δ92-413) also significantly decreased the number of peroxisomal ghosts as analyzed by one-way ANOVA (n ≥ 2, ****p < 0.0001, ns = not significant). Per replicate at least three cells were quantified for each cell line. Total number of cells analyzed: PEX1-KO + PEX1, 42; PEX1-KO + PEX1(Δ92-413), 25.

The other prominent effect of a PEX1 deletion we have observed was the increased number of peroxisomal vesicles (Figure 4B). After expression of PEX1 in PEX1-KO cells the peroxisomal number was highly reduced and even reached T-REx293 levels (Figure 5D). Similarly, expression of PEX1(Δ92-413) in PEX1-KO cells results in partial complementation, i.e., a significant decrease of number of pre-peroxisomal vesicles, however not completely reaching T-REx293 levels. This demonstrates that the structural requirements for PEX1 to perform protein import or abundance of peroxisomal vesicles is not identical: While the AAA⁺-ATPase domains are essential for both functions, the N-terminal region is only required for recovery of peroxisomal matrix protein import. Therefore, it is suggestive that the novel ‘moonlighting’ function of PEX1, which leads to a reduction of the number of vesicles is at least partially independent from extraction of the import receptor from the peroxisomal membrane.

Plasmids

The generation of a bicistronic plasmid (pIRES2) for the expression of solely EGFP-PTS1 (C-terminal PTS1 sequence -SKL) or in combination with PEX5 and PEX14 was described elsewhere (Neuhaus et al. 2014). For the creation of a PEX19-containing complementation vector, pGEX-4T-1-PEX19 was digested with BamHI/SalI and inserted in BglII/SalI-digested pIRES2-EGFP-PEX14. PEX1 was PCR-amplified with the primer pair 5′-gatcgctagc atgtggggcagcgatcgcctgg-3′ and 5′-gatcggatccttatgctaaagttactttctgtccag-3′ and inserted into BamHI/BmtI digested pIRES2-EGFP-PTS1. For construction of complementation vectors, pIRES2-EGFP-PTS1 was PCR-amplified (5′-gtcgacggtaccgcggg-3′, 5′-gtgaaccgtcagatccgctagc-3′) and PCR-amplified PEX-genes were inserted by overhangs (PEX3: pM-PEX3, 5′-accgtcagatccgctagcatgctgaggtctgtatggaattttctgaaacgc-3′, 5′-gtacccctcagcaactggagaaatgagtcgacggtaccgcgggc-3′; PEX10: pM-PEX10, 5′-accgtcagatccgctagcatggccccggccgccg-3′, 5′-catctaccttcggcactaccgctgagtcgacggtaccgcgggc-3′; PEX13: pM-PEX13, 5′-accgtcagatccgctagcatggcgtcccagccgcca-3′, 5′-gcccgcggtaccgtcgactcaaagatcttgcttttctccatctttccca-3’; PEX16: pM-PEX16, 5′-accgtcagatccgctagcatggagaagctgcggctcctgg-3′, 5′-cagaaaatctacttctacagttggggctgagtcgacggtaccgcgggc-3′). To test for PTS2-import, PTS2-EGFP (N-terminal PTS2 sequence RLQVVLGHL-) was PCR-amplified (pEGFPN1-PTS2-EGFP, 5′-atgataatatggccacaaccatgcatcggctgcaggtagtg-3′, 5′-ggcatggacgagctgtacaagtaacggccgcgactctagatcat-3′) and inserted into pIRES2-EGFP-PTS1, which was also amplified by PCR using the primer pair 5′-cggccgcgactctagatcataatc-3′ and 5′-gaaaaacacgatgataatatggccacaacc-3′. The PEX7-containing complementation vector was generated by ligation of PCR-amplified PEX7 (pM-PEX7, 5′-accgtcagatccgctagcatgagtgcggtgtgcggtgg-3′, 5′-ccctgcttgtcttactattcctgcttgagtcgacggtaccgcgggc-3′) and pIRES2-PTS2-EGFP (5′-gtcgacggtaccgcggg-3′, 5′-gctagcggatctgacggttcac-3′). Plasmids for generation of PEX5-and PEX14-CRISPR-KOs have been described elsewhere (Galiani et al. 2022), while for the remaining KO cell lines the following sense oligonucleotides were annealed with the corresponding antisense part and inserted into the pX458 vector (Addgene 48,138, Dr. Feng Zhang) to act as guide RNAs: PEX1 (5′-agcgatgcgctggcgggtgc-3′), PEX3 (5′-atgatattgtcgtcgtgctt-3′), PEX7 (5′-cgcgatgccgtagtgctgcg-3′), PEX10 (5′-cgcggatcacctccgggggg-3′), PEX13 (5′-ctgggagacccgccgaattc-3′), PEX16 (5′-ggtgacgagtcacgtactcc-3′), PEX19 (5′-ggggccccagaagagatcgcc-3′).

Cell culture

HEK T-REx293 cells were cultivated in Dulbecco’s Modified Eagle’s Medium high glucose (DMEM), supplemented with 10% fetal calf serum, 4 mM L-glutamine, 100,000 U/L penicillin and 100 mg/L streptomycin at 37 °C and 8.5% CO₂.

Transfections were done by using X-tremeGENE HP DNA Transfection Reagent (Roche, Germany) according to the manufacturer’s instructions.

Generation of knockout cell lines using the CRISPR/Cas9 system

For the generation of T-REx293 PEX1-, PEX3-, PEX7-, PEX10-, PEX13-, PEX16-and PEX19-KO cell lines, T-REx293 cells were transfected with specific CRISPR-vectors, containing the Cas9 endonuclease, the respective guide-sequence and EGFP. 48 h after transfection, EGFP-positive cells were sorted by fluorescence-activated cell sorting (FACS) in 96-well plates, containing DMEM, supplemented with additional 10% FCS. After expansion, several clones of each deletion cell line were seeded on coverslips and tested for the import of EGFP-PTS1/PTS2-EGFP or the presence of peroxisomal membranes by fluorescence microscopy. From clones showing an import defect or absence of peroxisomal membranes genomic DNA was isolated using the kit NucleoSpin Tissue (Macherey-Nagel), according to the manufacturer’s instructions. The exon with expected double strand break was then PCR-amplified using the Terra PCR Direct Polymerase Mix (Takara) and compared to the wild-type sequence by the TIDE algorithm (Brinkman et al. 2014). The amplified exons of clones that showed base pair insertions or deletions on every allele resulting in a frame shift were inserted in pGEM-T, amplified and sequenced to obtain the remaining expression products. The generation of the T-REx293 PEX5-and PEX14-KO cell lines has been described elsewhere (Galiani et al. 2022).

Preparation of whole cell lysates and immunoblot analysis

For the preparation of whole cell lysates, cells were seeded to >90% confluence on 6 cm plates. Cells were washed with Hanks’ Balanced Salt Solution (HBSS) without MgCl₂, trypsinized and resuspended in HBSS+MgCl₂. Cells were harvested (2000 rpm, 7 min, 4 °C) and resuspended in RIPA buffer (150 mM NaCl, 1% (v/v) IGEPAL, 0.5% (w/v) sodium deoxycholate, 0.1% (w/v) SDS, 50 mM Tris pH 8, cOmplete™ EDTA-free Protease Inhibitor Cocktail (Roche), 25 µg/mL DNase). After 30 min on ice cell debris was sedimented (13,000 rpm, 5 min, 4 °C) and SDS-sample buffer was added to the supernatant. Alternatively, adherent cells were directly lysed in SDS-PAGE sample buffer. Samples were analyzed by SDS-PAGE and immunoblotting using the following antibodies: rabbit αPEX1 (1:1,000, Proteintech), rabbit αPEX5 (1:5,000, (Fodor et al. 2015)), rabbit αPEX13 (1:1,000, Proteintech), rabbit αPEX14 (1:5,000, (Will et al. 1999)), rabbit αPEX19 (1:5,000, Thermo Fisher Scientific), rabbit αACAA1 (1:1,000, Sigma-Aldrich), mouse αGFP (1:2,000, Proteintech) and mouse αGAPDH (1:7,500, Proteintech). As secondary antibodies IRDye 800CW goat αrabbit and IRDye 800CW goat αmouse (1:15,000, LI-COR, Lincoln (USA)) were used.

Fluorescence microscopy

For fluorescence microscopy, cells were seeded on coverslips to an appropriate density, washed with D’PBS and fixed using 3% formaldehyde for 20 min. Afterwards, membranes were permeabilized in 1% Triton X-100 for 5 min and free binding sites were blocked with 2% BSA for 1 h. Next, cells were incubated for 30 min with the primary antibodies, rabbit αPMP70 (1:500, Invitrogen) and chicken αPEX14 (1:500, Ruhr-University Bochum), and afterwards with the secondary antibodies goat α-rabbit IgG (H+L) Alexa Fluor 594 (Invitrogen) and goat α-chicken IgG (H+L) Alexa Fluor 647 (Invitrogen). Coverslips were mounted on glass slides using Mowiol 4–88 (Calbiochem), supplemented with DAPI. Imaging was performed using the Zeiss ELYRA PS.1 Super-Resolution Structured Illumination Microscope (SR-SIM, Zeiss). All images were taken in one focal plane. Peroxisomal quantification was performed by applying the spots-function of IMARIS (BITPLANE Scientific Software) on microscopic whole-cell images.

Statistical analysis

Statistical analysis was performed by one-way ANOVA with Tukey’s multiple comparisons using GraphPad Prism (Graphpad Software, Inc, USA). Data were considered significant at a value of *p < 0.05.