CRISPRi-Library-Guided Target Identification for Engineering Carotenoid Production by Corynebacterium glutamicum

Corynebacterium glutamicum is a prominent production host for various value-added compounds in white biotechnology. Gene repression by dCas9/clustered regularly interspaced short palindromic repeats (CRISPR) interference (CRISPRi) allows for the identification of target genes for metabolic engineering. In this study, a CRISPRi-based library for the repression of 74 genes of C. glutamicum was constructed. The chosen genes included genes encoding enzymes of glycolysis, the pentose phosphate pathway, and the tricarboxylic acid cycle, regulatory genes, as well as genes of the methylerythritol phosphate and carotenoid biosynthesis pathways. As expected, CRISPRi-mediated repression of the carotenogenesis repressor gene crtR resulted in increased pigmentation and cellular content of the native carotenoid pigment decaprenoxanthin. CRISPRi screening identified 14 genes that affected decaprenoxanthin biosynthesis when repressed. Carotenoid biosynthesis was significantly decreased upon CRISPRi-mediated repression of 11 of these genes, while repression of 3 genes was beneficial for decaprenoxanthin production. Largely, but not in all cases, deletion of selected genes identified in the CRISPRi screen confirmed the pigmentation phenotypes obtained by CRISPRi. Notably, deletion of pgi as well as of gapA improved decaprenoxanthin levels 43-fold and 9-fold, respectively. The scope of the designed library to identify metabolic engineering targets, transfer of gene repression to stable gene deletion, and limitations of the approach were discussed.


Introduction
Metabolic engineering offers the possibility to construct production strains that overproduce a valuable compound of interest. Besides the overproduction of native products, also non-native products can be accessed by introducing heterologous pathways. The ability of rational strain engineering holds tremendous value for molecular biology and biotechnology and requires methods to precisely and predictably target genes for expression or repression [1,2].
The clustered regularly interspaced short palindromic repeats (CRISPR) system provides a method for targeted gene editing and gene regulation [3,4]. The CRISPR system from Streptococcus pyogenes is applied most often. One approach, called CRISPR interference (CRISPRi) [5], allows for controlled gene repression. It relies on modified protein dCas9 that is catalytically dead due to two amino acid changes in its RuvC and HNH endonuclease domains (D10A and H841A) [5]. The dCas9-sgRNA complex binds to the 20 bp complementary DNA target sequence of the sgRNA, which results in the inhibition of transcription by either blocking transcription elongation or inhibiting transcription initiation [6]. In contrast to a stable gene knockout (deletion), a gene knockdown temporarily suppresses the RNA level of the target gene, and it is a powerful technique to repress genes where a knockout would be lethal [7]. CRISPRi enables fast and robust, but reversible For construction of the CRISPRi library plasmid pS_dCas9, the vector pRG_dCas9 [46] was restricted with PstI and SalI (NEB, Frankfurt, Germany) and dephosphorylated (Antarctic phosphatase, New England Biolabs, Frankfurt, Germany). The sequences of the dCas9 handle and terminator from S. pyogenes were amplified by high-fidelity PCR (All-in HiFi, Kraichtal, Germany) from the plasmid piCas [66] with the oligonucleotides vgag and vgam (Table S1), and the PCR amplicon was purified with a PCR and gel extraction kit (Macherey-Nagel, Düren, Germany). The PCR product was cloned into the PstI-and SalI-restricted, dephosphorylated vector pRG_dCas9 by Gibson Assembly [67], resulting in plasmid pS_dCas9. Standard genetic procedures were performed as described previously [68].

Construction of the CRISPRi Library
The sgRNAs were designed to contain a 20 bp region homologous to the non-template strand of the chosen DNA targets. The genome sequence of C. glutamicum ATCC 13,032 [39] was used as a basis for the selection of the 20 bp targeting sequences using the CRISPy-web tool [69]. CRISPRi library plasmids were constructed in E. coli DH5α through the annealing oligo method, with single-stranded oligonucleotides covering sgRNA (20 bp) and 20 bp overlaps with plasmid pS_dCas9. Equal volumes (5 µL) of equimolar oligonucleotides (100 µM) were mixed with the annealing buffer (990 µL), and the mixture was incubated at 95 • C for 5 min and then cooled down to room temperature. The CRISPRi library plasmid pS_dCas9 was restricted with PstI (NEB, Frankfurt, Germany) and dephosphorylated (Antarctic phosphatase, New England Biolabs, Frankfurt, Germany) before the double-stranded oligonucleotides were annealed by the Gibson Assembly method [67]. The concentration of DNA was measured with an ND-1000 spectrophotometer (Thermo Fisher Scientific, Schwerte, Germany). The oligonucleotides (Table S1) used in this study were obtained from Metabion (Planegg/Steinkirchen, Germany). E. coli DH5α cells were transformed by heat shock after preparation of CaCl 2 -competent cells [70]. Transformants were screened by colony PCR, and plasmids were isolated by a plasmid miniprep kit (GeneJET, Thermo Fisher Scientific, Schwerte, Germany). Library vectors were confirmed by sequencing with oligonucleotides vgai and vgaj (Table S1). C. glutamicum cells were transformed by electroporation [71].

Construction of C. glutamicum Deletion Mutants
For deletion of the sdhCAB operon, the suicide vector pK19mobsacB was used [65]. The genomic flanking regions of sdhCAB were amplified from the genomic DNA of C. glutamicum WT using the oligonucleotide pairs del-sdhCAB1/del-sdhCAB2 and del-sdhCAB3/del-sdhCAB4 (Table S1). The PCR amplicons were purified, linked by crossover PCR, and subsequently cloned into a SmaI-restricted pK19mobsacB. The resulting deletion vector pK19mobsacB-sdhCAB was introduced into C. glutamicum via trans conjugation with E. coli S17-1 [65]. Deletion of sdhCAB was achieved by two-step homologous recombination using the respective deletion vector, as previously described [70]. Integration of the vector into one of the gene-flanking regions represents the first recombination event and was selected via kanamycin resistance. Integration of the vector into the genome results in sucrose sensitivity due to levansucrase, encoded by sacB. Selection for the second recombination event, loss of the vector, was carried out via sucrose resistance. Deletion of sdhCAB was verified via sequencing with primers del-sdhCAB-5 and del-sdhCAB-6 (Table S1).

Quantification of the mRNA Levels of Targeted Cells by CRISPRi
RNA levels were determined by quantitative reverse transcription PCR (qRT-PCR). Total RNA was isolated from C. glutamicum strains growing exponentially in CGXII medium. Biological triplicates were analyzed. Aliquots of 500 µL were centrifuged at 14,000 rpm for 15 s (Eppendorf centrifuge 5810 R), and the pellets were immediately frozen in liquid nitrogen and stored at −80 • C until further use. For RNA isolation, the samples were homogenized by resuspending the cells in 100 µL of TE buffer (10 mM Tris-HCl, 1 mM EDTA; pH 8) containing 5 mg mL −1 of lysozyme. After incubation at 37 • C for 30 min, total RNA was extracted using a NucleoSpin ® RNA kit (Macherey-Nagel, Düren, Germany) according to the manufacturer's instructions. After extraction, RNA samples were treated with DNase restriction using RNase-free DNase Set and RNeasy MinElute kits (Qiagen, Hilden, Germany) to eliminate possible genomic DNA contamination. The total RNA concentration was measured using a spectrophotometer (NanoDrop ® , ND-1000; ThermoFisher Scientific, Schwerte, Germany). Quality control was performed to determine the purity and integrity of isolated RNA.
Equal amounts of 50 ng of each sample were used to perform cDNA synthesis. qRT-PCR was performed using the SensiFAST TM SYBR ® No-ROX One-Step Kit (Bioline, London, UK) and the CFX96 cycler system (Bio-Rad, Hercules, CA, USA). The temperature profile was (1) Table S1. The ∆Cq method was used in calculations [72,73].

Design and Initial Testing of a CRISPRi Library for Gene Repression in C. glutamicum
This work aimed to design, construct, and test a CRISPRi library suitable for gene repression screening in C. glutamicum. In principle, the approach is generalizable, but here, we focused on scoring the effect of repressing genes of the central carbon metabolism and carotenogenesis as well as regulatory genes on the biosynthesis of decaprenoxanthin, the natural pigment of C. glutamicum.

Construction of a Vector CRISPRi Library for C. glutamicum
The CRISPRi library was based on pRG_dCas9 [56]. To ease library preparation, the dCas9 handle followed by the terminator from S. pyogenes was amplified from plasmid piCas [66] and inserted between the PstI and SalI restriction sites, resulting in vector pS_dCas9 ( Figure 1). The various gene-specific 20 bp sgRNA sequences of the target library  (Table S2) were generated from oligonucleotides by the annealing oligo method before being inserted into the PstI cloning site.
natural pigment of C. glutamicum.
3.1.1. Construction of a Vector CRISPRi Library for C. glutamicum The CRISPRi library was based on pRG_dCas9 [56]. To ease library preparation, the dCas9 handle followed by the terminator from S. pyogenes was amplified from plasmid piCas [66] and inserted between the PstI and SalI restriction sites, resulting in vector pS_dCas9 ( Figure 1). The various gene-specific 20 bp sgRNA sequences of the target library (Table S2) were generated from oligonucleotides by the annealing oligo method before being inserted into the PstI cloning site. Figure 1. Construction of the dual-inducible CRISPRi expression plasmid pS_dCas9. Adapted pRG_dCas9 plasmid carrying the dCas9 handle followed by the terminator from S. pyogenes amplified from plasmid piCas between the PstI and SalI restriction sites. A 20 bp sgRNA sequence can be inserted in the PstI cloning site. For multiplexing of more than one specific sgRNA, the restriction site SalI after the first sgRNA-cs can be used. It has chloramphenicol resistance. P1: tetR/tetA promotor; P2: tac promotor; T1: rrnB T1 terminator; T2: terminator from S. pyogenes; T3: lambda terminator; oriEc: p15A; oriCg: pCG1; cat: chloramphenicol resistance.

Testing of the CRISPRi Library Vector for the Repression of crtR
To test the sensitivity and function of the CRISPRi library vector, the repression of crtR coding for the transcriptional repressor CrtR of the carotenogenic crt operon [63] was chosen. We have shown previously that deletion of crtR in C. glutamicum MB001 increased the accumulation of the native carotenoid pigment decaprenoxanthin about 30-fold [63]. Therefore, we assumed that targeting crtR by CRISPRi will increase decaprenoxanthin biosynthesis. The sgRNA sequence for crtR (and later for all tested genes) was chosen according to the following strategy: the sgRNA (i) was identified using CRISPy-web [69] targeting the non-template strand, (ii) is located in the coding sequence of the target gene for inhibition of transcription elongation, (iii) preferably is in the 5′ proximal region of the coding sequence, and (iv) is unique in the C. glutamicum genome. The plasmid pS_dCas9_crtR and the empty vector pS_dCas9 were used to transform C. glutamicum MB001, and the resulting recombinant strains were cultivated in glucose minimal medium Figure 1. Construction of the dual-inducible CRISPRi expression plasmid pS_dCas9. Adapted pRG_dCas9 plasmid carrying the dCas9 handle followed by the terminator from S. pyogenes amplified from plasmid piCas between the PstI and SalI restriction sites. A 20 bp sgRNA sequence can be inserted in the PstI cloning site. For multiplexing of more than one specific sgRNA, the restriction site SalI after the first sgRNA-cs can be used. It has chloramphenicol resistance. P1: tetR/tetA promotor; P2: tac promotor; T1: rrnB T1 terminator; T2: terminator from S. pyogenes; T3: lambda terminator; oriEc: p15A; oriCg: pCG1; cat: chloramphenicol resistance.

Testing of the CRISPRi Library Vector for the Repression of crtR
To test the sensitivity and function of the CRISPRi library vector, the repression of crtR coding for the transcriptional repressor CrtR of the carotenogenic crt operon [63] was chosen. We have shown previously that deletion of crtR in C. glutamicum MB001 increased the accumulation of the native carotenoid pigment decaprenoxanthin about 30-fold [63]. Therefore, we assumed that targeting crtR by CRISPRi will increase decaprenoxanthin biosynthesis. The sgRNA sequence for crtR (and later for all tested genes) was chosen according to the following strategy: the sgRNA (i) was identified using CRISPy-web [69] targeting the non-template strand, (ii) is located in the coding sequence of the target gene for inhibition of transcription elongation, (iii) preferably is in the 5 proximal region of the coding sequence, and (iv) is unique in the C. glutamicum genome. The plasmid pS_dCas9_crtR and the empty vector pS_dCas9 were used to transform C. glutamicum MB001, and the resulting recombinant strains were cultivated in glucose minimal medium with 1 mM IPTG for induction of the sgRNA and with 0.25 µg mL −1 of aTc for induction of the dCas9 gene. Exponentially growing cells were harvested for RNA extraction, while the decaprenoxanthin content was quantified after cultivation for 28 h. After qRT-PCR, the ∆Cq value was calculated using the vegetative RNA polymerase sigma factor gene sigA as a reference. The qRT-PCR analysis revealed a statistically significant and about fourfold lower crtR RNA level upon targeting crtR using pS_dCas9_crtR ( Figure 2a).
Phenotypically, C. glutamicum MB001(pS_dCas9_crtR) indeed showed a more intense yellow pigmentation than strain MB001(pS_dCas9) (Figure 2b). Accordingly, HPLC analysis revealed an about ninefold, significantly (p-value < 0.001) higher decaprenoxanthin content (0.44 ± 0.05 mg (g CDW) −1 ) for C. glutamicum MB001(pS_dCas9_crtR) as compared to the empty vector carrying the control strain MB001(pS_dCas9) (0.05 ± 0.00 mg (g CDW) −1 ; Figure 2c). Thus, based on visual observation of pigmentation, qRT-PCR analysis of crtR RNA levels, and decaprenoxanthin quantification, the CRISPRi system proved suitable to score the effect that repression of a gene of interest has on carotenogenesis. To test the duration of the inhibitory effect of the CRISPRi targeting crtR, serial transfers from a culture grown for 6 generations in the presence of inducers to a medium without inducers were performed. CRISPRi targeting of crtR visibly increased decaprenoxanthin levels to above those of the empty vector control strain in both serial cultures. However, the effect faded gradually from serial transfer one (5 generations without inducers) to serial transfer two (8 generations without inducers) ( Figure S1). with 1 mM IPTG for induction of the sgRNA and with 0.25 µg mL −1 of aTc for induction of the dCas9 gene. Exponentially growing cells were harvested for RNA extraction, while the decaprenoxanthin content was quantified after cultivation for 28 h. After qRT-PCR, the ∆Cq value was calculated using the vegetative RNA polymerase sigma factor gene sigA as a reference. The qRT-PCR analysis revealed a statistically significant and about fourfold lower crtR RNA level upon targeting crtR using pS_dCas9_crtR ( Figure 2a). Phenotypically, C. glutamicum MB001(pS_dCas9_crtR) indeed showed a more intense yellow pigmentation than strain MB001(pS_dCas9) (Figure 2b). Accordingly, HPLC analysis revealed an about ninefold, significantly (p-value < 0.001) higher decaprenoxanthin content (0.44 ± 0.05 mg (g CDW) −1 ) for C. glutamicum MB001(pS_dCas9_crtR) as compared to the empty vector carrying the control strain MB001(pS_dCas9) (0.05 ± 0.00 mg (g CDW) −1 ; Figure 2c). Thus, based on visual observation of pigmentation, qRT-PCR analysis of crtR RNA levels, and decaprenoxanthin quantification, the CRISPRi system proved suitable to score the effect that repression of a gene of interest has on carotenogenesis. To test the duration of the inhibitory effect of the CRISPRi targeting crtR, serial transfers from a culture grown for 6 generations in the presence of inducers to a medium without inducers were performed. CRISPRi targeting of crtR visibly increased decaprenoxanthin levels to above those of the empty vector control strain in both serial cultures. However, the effect faded gradually from serial transfer one (5 generations without inducers) to serial transfer two (8 generations without inducers) ( Figure S1).

Characterization of a CRISPRi Library to Interrogate 74 Target Genes with Potential Relevance for Carotenogenesis in C. glutamicum
In total, 74 target genes from C. glutamicum were repressed by CRISPRi; the growth parameters of the respective strains are listed in Table S3. The CRISPRi library comprising a subset of genes from central carbon metabolism as well as regulatory genes was selected to cover glycolysis (19 gene targets), the pentose phosphate pathway (8 gene targets), and Figure 2. Testing the CRISPRi system for identification of metabolic engineering targets relevant for carotenoid production. Strain C. glutamicum MB001(pS_dCas9_crtR) for CRISPRi-mediated repression of crtR was compared to the empty vector carrying control strain MB001(pS_dCas9) with respect to (a) crtR RNA levels, (b) color phenotypes, and (c) cellular decaprenoxanthin content. For qRT-PCR analysis, exponentially growing cells were harvested and sigA was used as a reference (a). The color phenotype (b) was judged by visual inspection after growth in the Biolector ® flowerplate microcultivation system. Cells were grown in 40 g L −1 of glucose CGXII minimal medium for 28 h and induced at 0 h with 1 mM IPTG and 0.25 µg mL −1 of aTc. The cellular decaprenoxanthin content (c) is given as ß-carotene equivalents, as determined by HPLC analysis. Mean values and standard deviations of three biological replicates are given. The p-value of <0.001 (***) was calculated using Student's t-test (two sided, unpaired).

Characterization of a CRISPRi Library to Interrogate 74 Target Genes with Potential Relevance for Carotenogenesis in C. glutamicum
In total, 74 target genes from C. glutamicum were repressed by CRISPRi; the growth parameters of the respective strains are listed in Table S3. The CRISPRi library comprising a subset of genes from central carbon metabolism as well as regulatory genes was selected to cover glycolysis (19 gene targets), the pentose phosphate pathway (8 gene targets), and the tricarboxylic acid (TCA) cycle (17 gene targets) for general use in metabolic engineering ( Figure 3). With the goal to improve isoprenoid and carotenoid production as the chosen application example in this study, genes were selected to cover the MEP pathway of isoprenoid pyrophosphate biosynthesis (6 genes targets) as well as terminal carotenogenesis (8 gene targets) ( Figure 3). The CRISPRi library approach is commensurate with multivariate modular metabolic engineering [74]. The workflow of the approach is illustrated in Figure 4.

CRISPRi-Based Repression of Genes of the MEP Pathway and of Carotenogenesis-Reduced Decaprenoxanthin Pigmentation
Repression of genes of the MEP pathway and of carotenogenesis was expected to reduce pigmentation. To test this hypothesis, the respective CRISPRi library transformants were analyzed for decaprenoxanthin production. Indeed, repression of ispG lowered the cellular decaprenoxanthin content significantly, whereas repression of the other MEP pathway genes did not affect decaprenoxanthin content in a statistically significant manner, albeit some reduction was observed (Figure 5a). CRISPRi targeting of the crt operon genes crtE, mmpL, crtB, crtI, and crtEb reduced decaprenoxanthin biosynthesis significantly (Figure 5b). This was observed neither upon targeting idsA nor upon targeting crtX. These genes are not part of the crt operon. While CRISPRi repression of crtX did not reduce the decaprenoxanthin level (Figure 5b), unglucosylated instead of diglucosylated decaprenoxanthin accumulated, which was not observed for all other strains. This finding was commensurate with CrtX functioning as decaprenoxanthin glucosyltransferase [33].
In the case of idsA, qRT-PCR analysis revealed that repression reduced the mRNA level by about sixfold (Figure 6), while decaprenoxanthin biosynthesis was hardly affected. Possibly, CrtE compensates for the loss of IdsA, since crtE and idsA both encode geranylgeranyl diphosphate (GGPP) synthases [75]. The reduced decaprenoxanthin level observed upon CRISPRi repression of crtE may either be due to IdsA not being able to compensate for reduced CrtE levels or be due to polar effects on downstream crt operon genes, since crtE is the first gene in the crt operon. To directly test for downstream effects, qRT-PCR analysis of the RNA level of crtEb, the last gene of the crtE-mmpL-crtB-crtI-crtY e -crtY f -crtEb operon, was performed in strains where CRISPRi repressed either the first or the second crt operon gene (crtE or mmpL). Indeed, the crtEb mRNA level was significantly reduced upon CRISPRi targeting of crtE as well as mmpL ( Figure 6); thus, the observed reduced decaprenoxanthin biosynthesis upon CRISPRi repression of crtE and mmpL is, at least in part, not due to lowered CrtE or MmpL levels but due to repression of the whole crt operon.

Figure 5.
Influence of CRISPRi-mediated repression of genes of the MEP pathway (a) or of carotenogenesis (b) on decaprenoxanthin production by C. glutamicum. In addition, thiE, deoC, and fixB were analyzed. Mean values of biological duplicates are given. Statistical analysis was calculated with ANOVA against all measured decaprenoxanthin production of C. glutamicum MB001 from all Biolector ® flowerplates and is marked by a star (*). As a reference, the decaprenoxanthin production of the empty vector strain C. glutamicum MB001 (pS_dCas9) in biological duplicates of the corresponding experiment is shown. For abbreviations, see Figure 3.
In the case of idsA, qRT-PCR analysis revealed that repression reduced the mRNA level by about sixfold (Figure 6), while decaprenoxanthin biosynthesis was hardly affected. Possibly, CrtE compensates for the loss of IdsA, since crtE and idsA both encode geranylgeranyl diphosphate (GGPP) synthases [75]. The reduced decaprenoxanthin level observed upon CRISPRi repression of crtE may either be due to IdsA not being able to compensate for reduced CrtE levels or be due to polar effects on downstream crt operon genes, since crtE is the first gene in the crt operon. To directly test for downstream effects, qRT-PCR analysis of the RNA level of crtEb, the last gene of the crtE-mmpL-crtB-crtI-crtYe-crtYf-crtEb operon, was performed in strains where CRISPRi repressed either the first or the second crt operon gene (crtE or mmpL). Indeed, the crtEb mRNA level was significantly reduced upon CRISPRi targeting of crtE as well as mmpL ( Figure 6); thus, the observed reduced decaprenoxanthin biosynthesis upon CRISPRi repression of crtE and mmpL is, at least in part, not due to lowered CrtE or MmpL levels but due to repression of the whole crt operon.
Three further genes were analyzed. Since the MEP pathway provides precursors also for thiamin biosynthesis, repression of the first gene of the thiamin biosynthesis operon thiEOSGF coding for thiamine-phosphate pyrophosphorylase ThiE was studied. The gene Figure 5. Influence of CRISPRi-mediated repression of genes of the MEP pathway (a) or of carotenogenesis (b) on decaprenoxanthin production by C. glutamicum. In addition, thiE, deoC, and fixB were analyzed. Mean values of biological duplicates are given. Statistical analysis was calculated with ANOVA against all measured decaprenoxanthin production of C. glutamicum MB001 from all Biolector ® flowerplates and is marked by a star (*). As a reference, the decaprenoxanthin production of the empty vector strain C. glutamicum MB001 (pS_dCas9) in biological duplicates of the corresponding experiment is shown. For abbreviations, see Figure 3.
Three further genes were analyzed. Since the MEP pathway provides precursors also for thiamin biosynthesis, repression of the first gene of the thiamin biosynthesis operon thiEOSGF coding for thiamine-phosphate pyrophosphorylase ThiE was studied. The gene for deoxyribose-phosphate aldolase (deoC) that catalyzes a condensation reaction similar to that of the MEP pathway enzyme Dxs was also chosen. Finally, fixB coding for a protein of the electron transfer flavoprotein (ETF) family with some resemblance to a novel β-cyclase enzyme CruA from Chlorobium tepidum [76] was repressed. Upon CRISPRi targeting of thiE, deoC, and fixB, decaprenoxanthin levels were unaffected, somewhat reduced, and significantly decreased, respectively ( Figure 5).
for deoxyribose-phosphate aldolase (deoC) that catalyzes a condensation reaction similar to that of the MEP pathway enzyme Dxs was also chosen. Finally, fixB coding for a protein of the electron transfer flavoprotein (ETF) family with some resemblance to a novel βcyclase enzyme CruA from Chlorobium tepidum [76] was repressed. Upon CRISPRi targeting of thiE, deoC, and fixB, decaprenoxanthin levels were unaffected, somewhat reduced, and significantly decreased, respectively ( Figure 5).

CRISPRi-Based Repression of Genes of the Central-Carbon-Metabolism-Identified Supply of GAP and Entry into the Pentose Phosphate Pathway as Potential Bottlenecks in Decaprenoxanthin Biosynthesis
To identify the influence of the central metabolism on the decaprenoxanthin production, selected genes were repressed by CRISPRi and decaprenoxanthin was quantified (Figure 7). While the repression of the glycolytic gene pgi improved decaprenoxanthin production significantly, CRISPRi targeting of the pentose phosphate pathway genes tkt, tal, zwf, and pgl negatively affected decaprenoxanthin pigmentation (Figure 7b). Repression of aceE encoding a subunit of the pyruvate dehydrogenase complex that is relevant for carbon entry into the TCA cycle improved decaprenoxanthin biosynthesis (Figure 7c). CRISPRi targeting of the TCA cycle genes sdhA, sdhB, and sdhCD lowered decaprenoxanthin accumulation (Figure 7c) in each case; however, these changes were not statistically significant.

CRISPRi-Based Repression of Genes of the Central-Carbon-Metabolism-Identified Supply of GAP and Entry into the Pentose Phosphate Pathway as Potential Bottlenecks in Decaprenoxanthin Biosynthesis
To identify the influence of the central metabolism on the decaprenoxanthin production, selected genes were repressed by CRISPRi and decaprenoxanthin was quantified (Figure 7). While the repression of the glycolytic gene pgi improved decaprenoxanthin production significantly, CRISPRi targeting of the pentose phosphate pathway genes tkt, tal, zwf, and pgl negatively affected decaprenoxanthin pigmentation (Figure 7b). Repression of aceE encoding a subunit of the pyruvate dehydrogenase complex that is relevant for carbon entry into the TCA cycle improved decaprenoxanthin biosynthesis (Figure 7c). CRISPRi targeting of the TCA cycle genes sdhA, sdhB, and sdhCD lowered decaprenoxanthin accumulation (Figure 7c) in each case; however, these changes were not statistically significant.

Interrogation of Regulatory Genes by CRISPRi with Respect to Carotenoid
Biosynthesis in C. glutamicum C. glutamicum possesses seven RNA polymerase sigma factors, which in part are regulated by their cognate anti-sigma factors [39]. The deletion of the sigma factor gene sigB is known to increase carotenoid production [77]. To identify the influence of regulatory genes on carotenoid production in C. glutamicum, RNA polymerase sigma factor and anti-sigma factor genes and the genes for the transcriptional regulators of carbon metabolism SugR, GlxR, and RamB were chosen. CRISPRi targeting of sigB and sugR increased and decreased, respectively, the cellular decaprenoxanthin content; however, the effects were not statistically relevant (Figure 8). Notably, CRISPRi-mediated repression of glxR significantly increased the cellular decaprenoxanthin by about twofold (Figure 8).

Interrogation of Regulatory Genes by CRISPRi with Respect to Carotenoid Biosynthesis in C. glutamicum
C. glutamicum possesses seven RNA polymerase sigma factors, which in part are regulated by their cognate anti-sigma factors [39]. The deletion of the sigma factor gene sigB is known to increase carotenoid production [77]. To identify the influence of regulatory genes on carotenoid production in C. glutamicum, RNA polymerase sigma factor and antisigma factor genes and the genes for the transcriptional regulators of carbon metabolism SugR, GlxR, and RamB were chosen. CRISPRi targeting of sigB and sugR increased and decreased, respectively, the cellular decaprenoxanthin content; however, the effects were Statistical analysis was calculated with ANOVA against all measured decaprenoxanthin production of C. glutamicum MB001 from all Biolector ® flowerplates and is marked by a star (*). As a reference, the decaprenoxanthin production of the empty vector strain C. glutamicum MB001 (pS_dCas9) in biological duplicates of the corresponding experiment is shown. For abbreviations, see Figure 3. not statistically relevant ( Figure 8). Notably, CRISPRi-mediated repression of glxR significantly increased the cellular decaprenoxanthin by about twofold (Figure 8).

Deletion of Selected Target Genes Identified by CRISPRi Repression
Gene deletion is a favored metabolic engineering strategy since the constructed strains are genetically stable (except when suppressor mutations occur in trans). Gene deletion results in the complete loss of function (knockout), while CRISPRi-mediated repression reduces gene function (knockdown). A gene can be repressed by CRISPRi, but not deleted, if its function is essential. The MEP pathway genes, the RNA polymerase sigma factor A gene sigA, and, according to some reports, the regulatory gene glxR cannot be deleted in C. glutamicum. Deletion of the crt operon is known to abolish decaprenoxanthin biosynthesis [33,40], thus supporting the evidence obtained by CRISPRi ( Figure 5). Deletion of crtX is known to result in accumulation of unglucosylated decaprenoxanthin [32]. In addition, we chose other genes to study the effect of gene deletion on decaprenoxanthin production, i.e., pgi, gapA, and aceE, as their deletion was expected to increase decaprenoxanthin levels, as well as sdhABCD and sugR, since their deletion was expected to reduce carotenoid biosynthesis ( Figure 9).
In line with the CRISPRi results, deletion of the sdhABCD operon significantly reduced decaprenoxanthin levels in comparison to C. glutamicum WT, while deletion of pgi and gapA improved the cellular decaprenoxanthin content significantly (Figure 9a). Since the gapA deletion mutant showed impaired growth, the decaprenoxanthin concentration in the culture was not increased (Figure 9b). Therefore, the pleiotropic effects of gapA deletion nullified the increased decaprenoxanthin content per cell biomass as the biomass concentration was reduced. SugR represses many target genes and binds their promoter DNA regions with different affinities in vitro [78]; thus, besides pleiotropic effects, gradual regulatory differences may be expected as well. In contrast to the CRISPRi results, deletion of sugR increased the decaprenoxanthin content in comparison to the control strain fivefold. Although not understood mechanistically, these results indicate that the complete absence of SugR upon gene deletion is beneficial, while SugR levels below those of the WT strain upon CRISPRi targeting may limit decaprenoxanthin biosynthesis.
Deletion of aceE is known to be possible, but the resulting mutant requires a source of acetyl-CoA, such as acetate, for growth [60]. Therefore, with glucose minimal medium, the positive effect observed of CRISPRi targeting aceE could not be tested by the deletion Figure 8. Influence of CRISPRi-mediated repression of RNA polymerase sigma factor and transcriptional regulator genes on decaprenoxanthin biosynthesis. Mean values of biological duplicates are given. Statistical analysis was calculated with ANOVA against all measured decaprenoxanthin production of C. glutamicum MB001 from all Biolector ® flowerplates and is marked by a star (*). As a reference, the decaprenoxanthin production of the empty vector strain C. glutamicum MB001 (pS_dCas9) in biological duplicates of the corresponding experiment is shown. For abbreviations, see Figure 3.

Deletion of Selected Target Genes Identified by CRISPRi Repression
Gene deletion is a favored metabolic engineering strategy since the constructed strains are genetically stable (except when suppressor mutations occur in trans). Gene deletion results in the complete loss of function (knockout), while CRISPRi-mediated repression reduces gene function (knockdown). A gene can be repressed by CRISPRi, but not deleted, if its function is essential. The MEP pathway genes, the RNA polymerase sigma factor A gene sigA, and, according to some reports, the regulatory gene glxR cannot be deleted in C. glutamicum. Deletion of the crt operon is known to abolish decaprenoxanthin biosynthesis [33,40], thus supporting the evidence obtained by CRISPRi ( Figure 5). Deletion of crtX is known to result in accumulation of unglucosylated decaprenoxanthin [32]. In addition, we chose other genes to study the effect of gene deletion on decaprenoxanthin production, i.e., pgi, gapA, and aceE, as their deletion was expected to increase decaprenoxanthin levels, as well as sdhABCD and sugR, since their deletion was expected to reduce carotenoid biosynthesis ( Figure 9).
In line with the CRISPRi results, deletion of the sdhABCD operon significantly reduced decaprenoxanthin levels in comparison to C. glutamicum WT, while deletion of pgi and gapA improved the cellular decaprenoxanthin content significantly (Figure 9a). Since the gapA deletion mutant showed impaired growth, the decaprenoxanthin concentration in the culture was not increased (Figure 9b). Therefore, the pleiotropic effects of gapA deletion nullified the increased decaprenoxanthin content per cell biomass as the biomass concentration was reduced. SugR represses many target genes and binds their promoter DNA regions with different affinities in vitro [78]; thus, besides pleiotropic effects, gradual regulatory differences may be expected as well. In contrast to the CRISPRi results, deletion of sugR increased the decaprenoxanthin content in comparison to the control strain fivefold. Although not understood mechanistically, these results indicate that the complete absence of SugR upon gene deletion is beneficial, while SugR levels below those of the WT strain upon CRISPRi targeting may limit decaprenoxanthin biosynthesis.
strain grown with glucose.
Taken together, these results highlight the advantage of CRISPRi screening over deletion analysis. On the one hand, it allows for fast target gene identification for metabolic engineering, and transfer to the construction of genetically stable strains by gene deletion often is straightforward (sdhCAB, pgi, gapA). On the other hand, CRISPRi is a well-suited method to analyze essential genes, conditionally lethal genes such as aceE, as well as pleiotropic genes or regulatory genes.

Discussion
In this study, the first C. glutamicum CRISPRi library for the repression of 74 genes (e.g., those for central metabolism, coding for global regulators and RNA polymerase sigma factors, besides the genes for carotenoid biosynthesis) was constructed and tested to identify targets affecting carotenogenesis. As expected, decaprenoxanthin levels were reduced when the MEP pathway or carotenogenesis genes were repressed. In addition, eight new targets to improve carotenoid production were identified. Using deletions instead of CRISPRi repression for five selected genes supported CRISPRi results for some Deletion of aceE is known to be possible, but the resulting mutant requires a source of acetyl-CoA, such as acetate, for growth [60]. Therefore, with glucose minimal medium, the positive effect observed of CRISPRi targeting aceE could not be tested by the deletion of aceE since aceE deletion is conditionally lethal under these conditions. To circumvent this problem, the aceE mutant was assayed for decaprenoxanthin biosynthesis in glucose medium supplemented with acetate. The decaprenoxanthin content per biomass of the aceE mutant grown with glucose and acetate was, however, not higher than that of the WT strain grown with glucose.
Taken together, these results highlight the advantage of CRISPRi screening over deletion analysis. On the one hand, it allows for fast target gene identification for metabolic engineering, and transfer to the construction of genetically stable strains by gene deletion often is straightforward (sdhCAB, pgi, gapA). On the other hand, CRISPRi is a well-suited method to analyze essential genes, conditionally lethal genes such as aceE, as well as pleiotropic genes or regulatory genes.

Discussion
In this study, the first C. glutamicum CRISPRi library for the repression of 74 genes (e.g., those for central metabolism, coding for global regulators and RNA polymerase sigma factors, besides the genes for carotenoid biosynthesis) was constructed and tested to identify targets affecting carotenogenesis. As expected, decaprenoxanthin levels were reduced when the MEP pathway or carotenogenesis genes were repressed. In addition, eight new targets to improve carotenoid production were identified. Using deletions instead of CRISPRi repression for five selected genes supported CRISPRi results for some genes (sdhABCD, pgi, gapA). This was not the case for the conditionally lethal aceE deletion and for the pleiotropic deletion of the global regulatory gene sugR.
CRISPRi screenings were useful to identify promising targets for increased carotenoid production in other organisms [79,80]. Clearly, CRISPRi repression is complementary to gene deletion, not only because CRISPRi allows functional analysis of essential genes by assigning phenotypes as a consequence of their repression, but also because regulatory genes can be titrated, resulting in phenotypes that are intermediate between the levels in the wild type and the null levels in a deletion mutant. In C. glutamicum systems biology, loss of function analysis by gene deletions is often coupled with transcriptome, proteome, metabolome, and fluxome studies to gain insight, for example, into regulons and modulons [81]. Similarly, CRISPRi libraries may contribute to a systems-level understanding of an organism when coupled with omics experiments, allowing one to assess responses to gradual rather than absence/presence perturbations imposed by gene deletions. Transcriptome analysis may unravel multiple levels of global gene expression patterns associated with down-regulated transcriptional regulator protein levels complementing insight from regulatory gene knockouts. The carbon metabolism regulator SugR binds to its target promoter DNA sequences with a wide range of affinities in vitro [25,[82][83][84][85]. The differences observed in vivo between CRISPRi repression and deletion of the gene sugR with regard to decaprenoxanthin accumulation (Figures 8 and 9) support this notion. This may call for an in-depth analysis of the effects of gradually reduced SugR protein levels between wildtype levels and zero in the deletion mutant using CRISPRi combined with genome-wide transcriptome analysis. These analogous vs. digital (CRISPRi vs. deletion) perturbation experiments may also provide insight into the gradual changes at the proteome, metabolome, or fluxome levels [86].
Systems metabolic engineering finely balances enzymes of the pathway of interest (including contributory pathways) using different promoters, ribosome-binding sites, or synthetic operon structures [2,[87][88][89] in order to achieve high productivities. Recently, CRISPRi was used to adjust different levels of the arginine biosynthesis repressor ArgR in E. coli, which accelerated growth twofold as compared to the deletion of argR, while specific arginine production remained similar [90]. In a similar study, CRISPRi was applied to adjust cell growth by repression of key growth-related genes. Upon proper choice of sgRNAs, addition time, and induction level, carbon flux was precisely redistributed between biomass formation and synthesis of N-acetyl-glucosamine as the target product, which was produced up to about 90 g L −1 [91]. The approach described here may in the future be used for multiplexing since it is likely that CRISPRi multiplex screening will identify synergistic effects if two or more genes are repressed simultaneously by CRISPRi.
Decaprenoxanthin biosynthesis was significantly improved when pgi was repressed by CRISPRi or was deleted (Figures 7, 9 and 10). NADPH in C. glutamicum mainly derives from the pentose phosphate pathway [92] and is known to limit L-lysine production [93], especially during growth on carbon sources that support low PPP fluxes, such as acetate [20] or fructose [94,95]. Deletion of pgi is known to be beneficial for L-lysine production [96]. While production of 1 molecule of L-lysine requires 4 molecules of NADPH, 35 molecules of reduction equivalents (NADPH, reduced ferredoxin) are required per molecule of decaprenoxanthin. For the biosynthesis of the 10 C5 isoprenoid pyrophosphate precursors of decaprenoxanthin (2 DMAPP, 6 IPP, and 2 HMBPP) from GAP and pyruvate, 38 reduction equivalents are needed: 1 NADPH and 3 reduced ferredoxins per IPP or DMAPP (in the reactions of Dxr, IspG, and IspH) as well as 1 NADPH and 2 reduced ferredoxins per HMBPP (in the reactions of Dxr and IspG). Conversion of the 10 C5 precursors to the C50 decaprenoxanthin yields 3 reduced ferredoxins (1 in the reaction catalyzed by lycopene synthase CrtI and 2 in the elongation reactions catalyzed by lycopene elongase CrtEb; Figure 3). Thus, biosynthesis of decaprenoxanthin imposes a high demand for reduction equivalents to the cell. Improved NADPH provision may also be reached by overexpression of PPP genes, e.g., the genes coding for glucose-6-phosphate dehydrogenase (Zwf) [45], transketolase (Tkt), and transaldolase (Tal) [44]. Engineering of the PPP and the TCA cycle increased β-carotene production in E. coli by 64% [44] and in S. cerevisiae by 81.4% [45]. The importance of the PPP for decaprenoxanthin production was confirmed here when targeting the genes tkt, tal, zwf, and pgl by CRISPRi significantly decreased cellular decaprenoxanthin content (Figure 7b).
Microorganisms 2021, 9, x FOR PEER REVIEW 17 of 22 of the PPP and the TCA cycle increased β-carotene production in E. coli by 64% [44] and in S. cerevisiae by 81.4% [45]. The importance of the PPP for decaprenoxanthin production was confirmed here when targeting the genes tkt, tal, zwf, and pgl by CRISPRi significantly decreased cellular decaprenoxanthin content (Figure 7b). Figure 10. Scheme of C. glutamicum metabolism with CRISPRi target genes that significantly improved (red arrows) or reduced (black arrows) decaprenoxanthin biosynthesis in C. glutamicum when repressed. Grey is used to depict all other reactions in glycolysis (orange shading), the pentose phosphate pathway (blue shading), the TCA cycle (green shading), as well as the MEP pathway and the carotenogenesis (yellow shading). Abbreviations are explained in Figure 3.
Supply of GAP and pyruvate as precursors for decaprenoxanthin biosynthesis was found to be crucial by the CRISPRi analysis ( Figure 10). Repression of aceE (encoding subunit E1 of the pyruvate dehydrogenase complex that oxidatively decarboxylates pyruvate to acetyl-CoA) increased decaprenoxanthin. Although not statistically significant, CRIS-PRi of the fba gene coding for fructose-1,6-bisphosphate aldolase decreased decaprenoxanthin formation, while CRISPRi of the genes gapA and tpi encoding GAP converting glycolytic enzymes positively affected decaprenoxanthin formation ( Figure 7). Notably, deletion of gapA increased the cellular decaprenoxanthin concentration significantly ( Figure 9). However, CRISPRi repression and/or deletion of genes central to glycolysis do not come without side effects. For example, deletion of gapA cells showed an increased decaprenoxanthin content, but led to lower biomass concentrations; thus, the decaprenoxanthin titer was not higher (compare Figure 9a with Figure 9b). Moreover, deletion of aceE has been shown to abolish growth with glucose alone [60], and a positive effect on decaprenoxanthin production was not observable after growth on a glucoseacetate mixture (Figure 9). The results with aceE and gapA clearly showed that rerouting a major pathway of the central metabolism toward the production of decaprenoxanthin may critically interfere with growth. For example, the central carbon metabolism provides ATP, and indirectly CTP, which are required for isoprenoid pyrophosphate and carotenoid biosynthesis [15]. Thus, targets identified by CRISPRi library screening may require fine balancing of gene expression/repression to optimize decaprenoxanthin production without impairing growth.
Inherent problems in the interpretation of CRISPRi results when targeting an upstream gene of an operon exist due to polar effects on the downstream co-transcribed genes of that operon [97]. The crtE-mmpL-crtB-crtI-crtYe-crtYf-crtEb operon ( Figure 6) consists of seven co-transcribed genes [30,33]. CRISPRi repression of the chosen genes of this Figure 10. Scheme of C. glutamicum metabolism with CRISPRi target genes that significantly improved (red arrows) or reduced (black arrows) decaprenoxanthin biosynthesis in C. glutamicum when repressed. Grey is used to depict all other reactions in glycolysis (orange shading), the pentose phosphate pathway (blue shading), the TCA cycle (green shading), as well as the MEP pathway and the carotenogenesis (yellow shading). Abbreviations are explained in Figure 3.
Supply of GAP and pyruvate as precursors for decaprenoxanthin biosynthesis was found to be crucial by the CRISPRi analysis ( Figure 10). Repression of aceE (encoding subunit E1 of the pyruvate dehydrogenase complex that oxidatively decarboxylates pyruvate to acetyl-CoA) increased decaprenoxanthin. Although not statistically significant, CRISPRi of the fba gene coding for fructose-1,6-bisphosphate aldolase decreased decaprenoxanthin formation, while CRISPRi of the genes gapA and tpi encoding GAP converting glycolytic enzymes positively affected decaprenoxanthin formation (Figure 7). Notably, deletion of gapA increased the cellular decaprenoxanthin concentration significantly (Figure 9). However, CRISPRi repression and/or deletion of genes central to glycolysis do not come without side effects. For example, deletion of gapA cells showed an increased decaprenoxanthin content, but led to lower biomass concentrations; thus, the decaprenoxanthin titer was not higher (compare Figure 9a with Figure 9b). Moreover, deletion of aceE has been shown to abolish growth with glucose alone [60], and a positive effect on decaprenoxanthin production was not observable after growth on a glucose-acetate mixture (Figure 9). The results with aceE and gapA clearly showed that rerouting a major pathway of the central metabolism toward the production of decaprenoxanthin may critically interfere with growth. For example, the central carbon metabolism provides ATP, and indirectly CTP, which are required for isoprenoid pyrophosphate and carotenoid biosynthesis [15]. Thus, targets identified by CRISPRi library screening may require fine balancing of gene expression/repression to optimize decaprenoxanthin production without impairing growth.
Inherent problems in the interpretation of CRISPRi results when targeting an upstream gene of an operon exist due to polar effects on the downstream co-transcribed genes of that operon [97]. The crtE-mmpL-crtB-crtI-crtY e -crtY f -crtEb operon ( Figure 6) consists of seven co-transcribed genes [30,33]. CRISPRi repression of the chosen genes of this operon reduced decaprenoxanthin biosynthesis ( Figure 5). Besides gene-specific effects, polar effects have to be considered. CRISPRi targeting of the first and second crt operon genes (crtE and mmpl) significantly reduced mRNA levels of crtEb, the last gene in the operon ( Figure 6). In particular, reduced decaprenoxanthin upon repression of crtE came as a surprise since deletion of crtE did not abolish carotenogenesis, while deletion of the crtE paralog idsA did [75]. Therefore, the finding that CRISPRi repression of idsA did not reduce decaprenoxanthin was not anticipated (Figure 5b). Since it has been shown that plasmid-borne overexpression of crtE compensates for the deletion of idsA [75], it is likely that partial repression of idsA by CRISPRi was compensated for by CrtE.

Conclusions
The C. glutamicum CRISPRi library designed in this study was shown to be suitable to score targets for improving decaprenoxanthin production, the application example chosen here. The inherent limitations of this screening approach with regard to the transfer of CRISPRi results to clean gene deletions became obvious as they reflect the change from a gradual analogous approach to a digital yes/no approach. With these limitations in mind, many applications of this CRISPRi library and its future extensions become evident, in particular when combined with systems approaches targeted at gaining a basic physiological understanding or at achieving superior biotechnological performance of C. glutamicum.
Supplementary Materials: The following are available online at https://www.mdpi.com/2076-2 607/9/4/670/s1: Figure S1: Duration of the inhibitory effect of the CRISPRi targeting crtR; Table  S1: Oligonucleotides used in this study; Table S2: Annealing oligo primers for construction complementary regions of sgRNA templates; Table S3: Data generated with the CRISPRi library for all targets. Funding: This research was funded in part by the European Regional Development Fund (ERDF) and the Ministry of Economic Affairs, Innovation, Digitalization and Energy of the State of North Rhine-Westphalia (grant number EFRE-0400184 (Bicomer) and grant number EFRE-0300095/1703FI04 (Cluster Industrial Biotechnology Kompetenzzentrum Biotechnologie CKB)). Support for the article processing charge by the Deutsche Forschungsgemeinschaft and the Open Access Publication Fund of Bielefeld University is acknowledged.

Informed Consent Statement: Not applicable.
Data Availability Statement: All data are present in the manuscript and its supplement.