Spatio-temporal control of DNA replication by the pneumococcal cell cycle regulator CcrZ

Most bacteria replicate and segregate their DNA concomitantly while growing, before cell division takes place. How bacteria synchronize these different cell cycle events to ensure faithful chromosome inheritance is poorly understood. Here, we identified a conserved and essential protein in pneumococci and related Firmicutes named CcrZ (for Cell Cycle Regulator protein interacting with FtsZ) that couples cell division with DNA replication by controlling the activity of the master initiator of DNA replication, DnaA. The absence of CcrZ causes mis-timed and reduced initiation of DNA replication, which subsequently results in aberrant cell division. We show that CcrZ from Streptococcus pneumoniae directly interacts with the cytoskeleton protein FtsZ to place it in the middle of the newborn cell where the DnaA-bound origin is positioned. Together, this work uncovers a new mechanism for the control of the bacterial cell cycle in which CcrZ controls DnaA activity to ensure that the chromosome is replicated at the right time during the cell cycle.


Introduction
Most organisms have mechanisms ensuring that their genome is replicated and segregated prior to cell division. In many bacterial species, DNA replication and cell division occur concomitantly [1][2][3] .
Different models emerged from the mid-1900's to explain how bacterial cells handle DNA replication together with cell division in Escherichia coli or Bacillus subtilis [4][5][6][7] . The current cellsize control model suggests that cells initiate DNA replication independently from their original size, and grow to a constant size independently from their size at birth (adder model) [8][9][10][11][12] . How cells sense changes in cell size and convert it to trigger replication initiation is not known, but these models imply the existence of regulatory controls 3,[13][14][15] . However, no such cell cycle regulator has been reported yet for bacteria. Specific regulatory models have been proposed for E. coli [16][17][18] , but these are not applicable to most other organisms, and especially Gram-positive bacteria, that do not contain the proteins proposed to be the regulators.
Furthermore, most of the mechanisms known to regulate the initiation of replication and the activity of the replication initiator DnaA in E. coli do not exist in other bacteria [19][20][21][22][23] . This pinpoints a divergence between regulatory systems within bacteria. In line with this notion, changes in DNA replication initiation were shown FtsZ away from mid-cell 26,27 . Both species also have a nucleoid occlusion system (Noc) inhibiting Z-ring formation over the chromosome to prevent "cutting" of the chromosome during cell division 28 . Together, the Min and Noc systems ensure that cell division and septation occur when both sister chromatids have been fully replicated and segregated. These systems are however not conserved within all bacteria as the Gram-positive opportunistic human pathogen S. pneumoniae lacks homologs of the Min and Noc systems 29 . In contrast to E. coli and B. subtilis, the pneumococcal Z-ring forms readily over the nucleoid 29,30 .
Recently, a pneumococcal specific protein called RocS was identified that might fulfil a similar function as the Noc system by connecting chromosome segregation with capsule production 31 .
Another S. pneumoniae specific protein, called MapZ was shown to guide Z-ring formation, analogous to the Min system in other bacteria 32,33 . During cell growth, nascent MapZ rings are pushed apart by septal peptidoglycan synthesis, allowing for FtsZ polymers to continuously assemble at the newly formed septum 34 .
Importantly, the position of the origin of replication (oriC) was shown to be crucial for division site selection in S. pneumoniae and the origins mark the positions of future division sites 35 .
In S. pneumoniae, cell division and DNA replication are thus intimately connected. Critically however, it remains unknown how the cell senses when a new round of replication should be initiated.
We hypothesized that an unknown factor could be responsible for coordination of cell division and DNA replication in the pneumococcus. Using high throughput gene silencing with CRISPRi of all essential genes of S. pneumoniae 36 , we examined proteins leading to DNA content defects upon depletion. Here, we describe the identification of CcrZ, a conserved protein that activates DnaA to trigger initiation of DNA replication. Pneumococcal CcrZ localizes at the division site in a FtsZdependent manner and its inactivation leads to division defects.
Together, our findings show that CcrZ acts as a novel spatiotemporal link between cell division and DNA replication in S. pneumoniae.

CcrZ is a conserved bacterial cell cycle protein
We previously generated a knock-down library using CRISPRi (clustered regularly interspaced short palindromic repeats interference) targeting 348 conditionally essential genes of the serotype 2 strain S. pneumoniae D39V that were identified by Tn-Seq (transposon-insertion sequencing) 36 . Here, we investigated the function of spv_0476, encoding a protein of unknown function that is conserved in most Firmicutes (>30% identity) ( Supplementary Fig. 1a). Silencing of spv_0476 by CRISPRi led to a drastic reduction of the growth rate as well as appearance of anucleate cells as visualized by DAPI staining (Fig. 1a-b).
We renamed SPV_0476 to CcrZ (for Cell Cycle Regulator protein interacting with FtsZ) for reasons explained below. ccrZ is in an operon with trmB, which encodes a tRNA methyltransferase and this operon structure is conserved across Firmicutes ( Supplementary Fig. 1a). To exclude the possibility that the observed phenotypes of ccrZ silencing were caused by polar effects on trmB expression, we constructed a deletion of trmB.
This deletion did not lead to any growth defect ( Supplementary   Fig. 1b left panel). While Tn-seq indicated that ccrZ is essential 36 , we were able to generate a DccrZ deletion mutant, although cells grew very slow. We therefore constructed a depletion of CcrZ by ectopically expressing CcrZ under control of either an IPTG-or a ZnCl 2 -inducible promoter (P lac and P Zn respectively) and deleted ccrZ from its native location (ccrZ -/+ and P Zn -ccrZ -/+ respectively).
Depletion of CcrZ led to a significant growth delay at 37°C and 30°C, confirming the CRISPRi observations ( Supplementary   Fig. 1b). Immunoblotting using a specific antibody raised against purified CcrZ confirmed CcrZ depletion ( Supplementary Fig. 1c).
In line with the CRISPRi observations, DNA staining of cells depleted for CcrZ showed that 20% of cells lacked a nucleoid ( Fig. 1c, 442 Fig. 1d).
Interestingly, the S. pneumoniae ccrZ deletion could not be complemented by expression of ccrZ from either B. subtilis or S. aureus as only very small colonies were present on agar plate.
In contrast, depletion of S. aureus CcrZ was rescued by expression of CcrZ from B. subtilis ( Supplementary Fig. 1d).
In addition to an increase of the number of anucleate cells, CcrZ depletion in S. pneumoniae also led to slight morphological defects and modest changes in cell size when analyzed by phase contrast microscopy (Fig. 1d). Polysaccharide capsule production has previously been linked to the pneumococcal cell cycle 37 , but capsule production was not impacted as the amount of capsule was similar between a CcrZ mutant and wild type ( Supplementary  Fig. 1e). To visualize division sites in live cells, we constructed a translational fusion of mTurquoise2 to the tubulin-like protein FtsZ (as the only copy of FtsZ, expressed from its native genetic location), which assembles into distinct rings at new division sites where it recruits the machinery required to form septa 38 . Figure 1e, Z-rings were clearly mis-localized upon CcrZ depletion, with the presence of several aberrant Z-rings in a fraction of the cells (Fig. 1e). To obtain more insights into the morphological defects caused by CcrZ depletion and verify that the increased number of septa are not due to the fluorescent protein fused to FtsZ, we employed transmission electron microscopy (TEM) in untagged cells. While not evident by phase contrast microscopy, when ccrZ was depleted we observed frequent aberrant septum formation using TEM, in line with the FtsZ localization data, and many cells harbored two (18 %) to four (4 %) septa while typically only one septum is observed in wild type cells ( Fig. 1f-g).

Figure 1. Depletion of CcrZ leads to anucleate cells and division defects
a, growth curve of cells with ccrZ targeted by CRISPRi (ccrZ sgRNA + IPTG) indicates a clear growth defect when ccrZ is silenced. b, ccrZ silencing leads to appearance of anucleate cells, as visualized by DAPI staining. c, ccrZ depletion by ectopic expression via the IPTG-inducible P lac promoter also leads to cells lacking a nucleoid, as observed by DAPI staining. d, distribution of cell area of ccrZ-depleted cells, ccrZ depletion leads to a slight decrease in cell length and cell area (P value = 2.2 x 10 -16 , Wilcoxon rank sum test). e, when a deletion of ccrZ is complemented (left panel), FtsZ-mTurquoise2 shows a clear mid-cell localization, while it appears as a blurry signal in several cells upon ccrZ depletion (right panel). f, transmission electron microscopy (TEM) indicates that cells depleted for ccrZ form multiple, often incomplete, septa. g, distribution of number of septa per cell length as determined by TEM for 22 wild type cells, 28 CcrZ-depleted cells and 17 ccrZ-complemented cells (P value = 1 x 10 -6 for wild type vs CcrZ-depleted cells and P value = 0.0013 for ccrZ-complemented vs CcrZ-depleted cells, Wilcoxon rank sum test with Bonferroni adjustment).

S. pneumoniae CcrZ is part of the divisome
As CcrZ seems to be involved in both chromosome biology and cell division, we examined its subcellular localization. Strikingly, immunofluorescence on fixed cells using a CcrZ-specific antibody demonstrated a clear mid-cell localization ( Supplementary   Fig. 2a). To assess the localization of CcrZ in live cells, we created several functional fusions of a green fluorescent protein to the N-terminus of CcrZ (gfp-ccrZ) or a red fluorescent protein to the C-terminus (ccrZ-mKate2) and inserted either construct at the native ccrZ locus (Supplementary Fig. 1c). Visualization of fluorescently tagged CcrZ by epifluorescence microscopy in live bacteria showed that CcrZ localizes at mid-cell (Fig. 2a).
This localization was also conserved in both the TIGR4 and unencapsulated R6 strains ( Supplementary Fig. 2b). Interestingly, CcrZ Sa and CcrZ Bs did not localize as clear rings at mid-cell in S. aureus and B. subtilis ( Supplementary Fig. 2b), indicating that the activity and/or localization of CcrZ in these organisms is regulated differently. In order to get higher spatial resolution of S. pneumoniae CcrZ, 240 images (16 stacks) on live cells were acquired using 3D-structured illumination microscopy (3D-SIM) and reconstructed to generate a super resolution image and a 3D fluorescence image of GFP-CcrZ. As shown in  To test whether the mid-cell localization of S. pneumoniae CcrZ coincides with FtsZ, we constructed a CcrZ / FtsZ double-labelled strain (gfp-ccrZ ftsZ-mCherry). As shown in  Table 1). To determine which of the candidates might interact directly with CcrZ, we used the NanoBit complementation reporter assay 40,41 , which uses an enhanced luciferase separated into two different fragments (large bit (LgBit) and small bit (SmBit), respectively). Fusion of two different interacting proteins to each fragment can restore the activity of the luciferase and, in presence of a furimazine-based substrate, produce light. Accordingly, we fused the C-terminal extremity of CcrZ to LgBit (ccrZ-LgBit) and putative partners to  HlpA-SmBit as a positive control of interaction. After addition of the substrate, we could detect a strong and reproducible signal when FtsZ was fused to SmBit and CcrZ to LgBit, as well as a weaker signal for FtsA, EzrA and ZapA, and no detectable signal for any of the other proteins (Fig. 3a). This result indicates that FtsZ and CcrZ in S. pneumoniae are in very close proximity in space. Interestingly, using a strain expressing both CcrZ-LgBit and CcrZ-SmBit, a weak signal was observed indicating that CcrZ might also self-interact (Fig. 3a).
To confirm the observed interaction with FtsZ, we used a bacterial two-hybrid assay in E. coli 42 . Again, we observed a robust interaction between CcrZ and FtsZ, while T25-FtsZ did not interact with the empty vector alone, strongly suggesting that CcrZ directly binds to FtsZ (Fig. 3b). Co-immunoprecipitation of FtsZ-GFP from S. pneumoniae cells confirmed the in vivo interaction with CcrZ ( Fig. 3c). Affinity purification of CcrZ Sp -GFP when over-expressing FtsZ Sp in E. coli also confirmed this interaction as we were able to co-purify FtsZ in large amounts Time-lapse imaging indicated that cells with defective DNA content had either no DNA at all or chromosomes "guillotined" during septum closure suggesting reduced nucleoid occlusion control in DccrZ (Fig. 4a, Supplementary Video 6). We also co-localized FtsZ-CFP with HlpA-mKate2 while depleting CcrZ for a short period of time (2h). Interestingly, we observed many cells with a chromosome localized at only one half of the cell, at one side of the Z-ring (Fig. 4b). The absence of DNA in the other half of the cell could be explained by defective DNA segregation, by impaired replication or by DNA degradation.
When attempting to make clean ccrZ deletions, in addition to small colonies typical of slow growing mutants, there were also spontaneous large, wild type-sized colonies. Growth analysis of cells from three of these large colonies (ccrZ supp1-3 ) showed that cells behaved like wild type and DAPI staining revealed a restoration of wild type DNA content ( Fig. 4c-d). To verify whether these wild type-like phenotypes were caused by suppressor mutations, the genomes of these fast-growing strains were sequenced. All three strains still contained the ccrZ deletion and, in addition, contained a single nucleotide polymorphism To test this hypothesis, we quantified the copy number ratio between chromosomal origin and terminus regions (oriC/ter ratios) using real-time quantitative PCR. In a wild type situation, during exponential growth, the oriC/ter ratio varies between 1.4 -1.8, as most cells have started a round of DNA replication (note that in contrast to E. coli and B. subtilis, multifork replication does not occur in S. pneumoniae) 45 . Remarkably, depletion of CcrZ resulted in a significantly decreased DNA replication initiation rate with an oriC/ter ratio of 1.1 vs 1.8 for complemented cells (P value < 0.05) (Fig. 4f). Interestingly, the same observation was made for both B. subtilis and S. aureus, where deletion or depletion of CcrZ caused a clear reduction in oriC/ter ratios ( Fig. 4g). As the identified ccrZ-bypass mutations were found in DNA replication initiation regulators, we tested whether they would restore the oriC/ter ratio in a fresh ccrZ deletion background in S. pneumoniae. Indeed, the oriC/ter ratios for ΔccrZ dnaA-S292G, ΔccrZ dnaA-Q247H and for yabA-E93* (ccrZ supp3 ) were like wild type (Fig. 4h-i).
The point mutation found in yabA causes premature translation termination at the C-terminus of YabA. When yabA alone was replaced by an antibiotic resistance cassette, we observed an increase of replication initiation as well as a reduced growth rate; but when ccrZ was co-deleted, wild type like growth and a wild type oriC/ter ratio was restored ( Fig. 4i-j). DnaA suppressor mutations were located in the AAA+ ATPase domain of DnaA and Orange arrows indicate a cell with no nucleoid after cell division; white arrows indicate a cell with "guillotined" DNA. b, co-localization of FtsZ-CFP and HlpA-mKate2 when depleting ccrZ indicates that several cells have a nucleoid located only on one side of the Z-ring. c, three isolated ccrZ mutants (ccrZ supp1-3 ) restore wild type growth to ΔccrZ. d, DAPI staining of the three selected ccrZ suppressors mutants shows a restoration of DNA content. e, schematic representation of the localization of suppressor mutations in the domain III of DnaA and in the DnaA/DnaN binding motif (ANB) of YabA. TM: tetramerization domain. f, oriC/ter ratios as determined by RT qPCR for D39V wild type and CcrZ-depleted cells. Average values are indicated under the boxes. CcrZ depletion leads to a clear reduction in oriC/ter ratio. g, oriC/ter ratios for S. aureus upon ccrZ Sa depletion (left) and for B. subtilis with ccrZ Bs deletion (right). h, oriC/ter ratios of strains with dnaA mutations re-inserted into a ΔccrZ background show that these mutations restore replication initiation rates. i, yabA deletion leads to an increase in oriC/ter ratios, while suppressor mutation ccrZ supp3 (ΔccrZ, yabA-E93*) as well as co-deletion of yabA together with ccrZ (ΔyabA ΔccrZ) restored a wild type ratio. j, while yabA deletion alters the growth rate, a ΔyabA ΔccrZ double mutant grows like wild type. dnaA Q247H and dnaA S292G mutation also restore a wild type rate in a ΔccrZ mutant. k, dnaA mutation in a wildtype background increases the oriC/ter ratios. See Methods for statistical tests.
it was previously reported that specific mutations in this domain could increase the initiation rate in B. subtilis 46 . To determine if those mutations alone were able to induce over-initiation, we inserted each dnaA mutation into a wild type background strain.
Marker frequency analysis detected an increase in the oriC/ter ratio for both dnaA alleles (Fig. 4k). We conclude that mutations that increase the rate of initiation of DNA replication can rescue the ∆ccrZ phenotype.

CcrZ is a conserved regulator of DnaA
The results so far suggest that the division defects observed in the absence of CcrZ are due to perturbed Z-ring formation caused by under-replication of the chromosome. To examine whether disruption of DNA replication in general could lead to division defects similar to those of a ccrZ mutant, we took advantage of a thermosensitive dnaA mutant (dnaA TS ) in which DNA replication initiation is drastically reduced when cells are grown at the non-permissive temperature (40°C) 31 . As expected, when shifted to the non-permissive temperature, many cells were anucleate ( Supplementary Fig. 4a). Strikingly, localization of FtsZ-mTurquoise2 in the dnaA TS strain at 40°C phenocopied the ∆ccrZ mutant, and FtsZ was frequently mis-localized (Fig. 5a). Furthermore, examination by TEM at 40°C showed many cells with aberrant septa like CcrZ-depleted cells (Fig. 5b). As DnaA inactivation leads to strikingly similar phenotypes, these data are consistent with the idea that CcrZ exerts a control on DNA replication initiation.
To test whether CcrZ controls DNA replication via regulating DnaA activity, we made use of the fact that a B. subtilis ∆ccrZ Bs mutant also under-initiates (Fig. 4g) and a strain was constructed in which DNA replication was driven in a RepNdependent manner (from a plasmid origin of replication oriN) rather than from DnaA-dependent initiation (from oriC). This showed no significant ori-ter ratio differences when ccrZ was deleted (Fig. 5c), suggesting that CcrZ is an activator of DnaAdependent initiation of replication in B. subtilis. We therefore tested whether CcrZ interacts directly with DnaA to trigger DNA replication and employed bacterial two-hybrid assays and the Split-luc system using pneumococcal CcrZ and DnaA ( Fig. 3a and Supplementary Fig. 4b). However, none of these assays revealed a direct protein-protein interaction. It is still possible that CcrZ interacts directly with DnaA, but that we cannot detect it with these assays. Alternatively, another factor might be required for CcrZ's function or CcrZ indirectly affects the activity of DnaA in replication initiation.
CcrZ's conserved residues are essential for its function S. pneumoniae CcrZ is 264 amino acids long and is predicted to Nevertheless, as CcrZ is highly conserved in Firmicutes, we aligned CcrZ protein sequence with 1000 protein sequences from UniRef50 and identified three residues conserved in more than 95% of the proteins (D159, N164 and D177) and two other residues (H157 and D189) in more than 80% ( Supplementary   Fig. 4c). To determine the position of these residues, the S. pneumoniae CcrZ protein sequence was mapped onto the crystal structure of the best hit from the HMM alignment, the choline kinase LicA, in complex with adenosine monophosphate (AMP) (pdb 4R78). Interestingly, these five conserved residues appear to be in spatial proximity to AMP and thus to a putative nucleotide-binding pocket (Fig. 5d). Mutational analysis of these residues showed that at least H157, N164 and D177 are essential for CcrZ's function in S. pneumoniae (Fig. 5e), while mutating CcrZ-D159 or CcrZ-D189 did not lead to any growth defect. All three mutants were properly produced ( Supplementary   Fig. 1c) and CcrZ-H157A and CcrZ-D177A could still localize at the septum (Fig. 5f). Therefore, these three residues are crucial for the function of CcrZ. CcrZ-N164 and CcrZ-D177 residues correspond to LicA-N181 and LicA-D194, respectively, and both residues were shown to interact with the α-phosphate moiety of AMP 48 . It is therefore very likely that CcrZ-N164 and CcrZ-D177 also contribute to binding of a as of yet unknown nucleotide.

S. pneumoniae
We showed above that CcrZ is fundamental for DnaA-dependent  Fig. 5). In the absence of CcrZ, initiation of DNA replication is mis-timed and occurs too late relative to cellular growth and Z-ring formation, frequently leading to futile division events and anucleate cells (Fig. 6e).

Figure 5. CcrZ activates DnaA-dependent replication initiation
a, localization of FtsZ-mTurquoise2 in a thermo-sensitive DnaA strain (dnaA TS ) at permissive (30°C) and non-permissive (40°C) temperatures shows that dnaA inactivation leads to a similar phenotype as ccrZ inactivation. b, TEM of dnaA TS at non-permissive temperature (40°C) indicates the presence of multiple septa, similarly to a ΔccrZ mutant. c, when replication is driven in a RepN-dependent manner in B. subtilis (oriN), no decrease in ori/ter ratio can be observed in absence of ccrZ Bs (oriN, ΔccrZ Bs ). d, LicA choline kinase structure complexed with AMP and MES (2-(N-morpholino)ethanesulfonic acid). The 5 residues indicated in yellow are conserved between CcrZ and LicA (and highly conserved within Firmicutes). e, mutation of three of these five conserved residues in the putative ATP binding pocket leads to growth defects. f, localization of GFP-CcrZ H157A and GFP-D177A is not impaired.  Fig. 1a).
Besides the production of anucleate cells and cells with cleaved chromosomes, ccrZ mutants contain multiple aberrant division septa (Fig. 6e). Notably, this is phenocopied

Bacterial strains and culture conditions.
All strains, plasmids and primers used are listed in Supplementary   B. subtilis strains were derived from 1A700 or JH642 (pheA1 trpC2) 55 .

Construction of strains is described in the Supplementary
Methods.

Microtiter plate-based growth assay
For S. pneumoniae growth assays, cells were first grown in C+Y medium pH = 7.4 until mid-exponential growth phase (OD 595nm = 0.3) with no inducer at 37°C, after which they were diluted 100 times in fresh C+Y medium supplemented with IPTG or ZnCl 2 when appropriate. Cellular growth was then monitored every 10 min at either 37°C or 30°C in a microtiter plate reader (TECAN Infinite F200 Pro). Each growth assay was performed in triplicate.
The lowest OD 595nm of each growth curve was normalized to 0.004 Images were processed using ZEN (Zeiss). Signals was deconvolved, when appropriate, using Huygens (SVI) software.

Transmission Electron Microscopy (TEM)
Strains were grown in C+Y medium at either 37°C, or at 30°C for dnaA TS , until an OD 595nm = 0.3, with or without addition of ZnCl 2 (for ccrZ complementation or depletion, respectively) and diluted

3D Structured Illumination Microscopy (3D-SIM)
Samples for 3D-SIM were prepared as described previously

Image analysis and cells segmentation
All microscopy images were processed using Fiji (fiji.sc).
Cell segmentation based on phase contrast images was performed either on Oufti 57 or Morphometrics 58 and fluorescent signals where analyzed using Oufti (for CcrZ and FtsZ), MicrobeJ 59 (for CcrZ) or iSBatch 60 (for DnaN and oriC). Fluorescence heat-maps were generated using BactMAP 61 .

CcrZ-GFP
For affinity purification of CcrZ Sp -GFP while expressing FtsZ Sp ,

Large-scale purification of CcrZ-CPD for antibody production
In  GFP-proteins were eluted using SDS sample buffer at 95°C for 10 min and analyzed by immunoblotting.

Genome resequencing of ccrZ suppressors by NGS
Strains ccrZ supp1 , ccrZ supp2 and ccrZ supp3 were grown in C+Y medium at 37°C until OD 595nm = 0.3 and cells were harvested by centrifugation 1 min at 10,000 x g. Pellet was then re-suspended

oriC/ter ratios determination by RT-qPCR
Determination of S. pneumoniae oriC/ter ratios was performed as followed. Cells were pre-grown until OD 600nm = 0.4 in C+Y medium at 37°C, with or without inducer (ZnCl 2 or IPTG) for complementation and depletion conditions, respectively. Cells were then diluted 100 times in fresh C+Y medium supplemented when appropriate with inducer and harvested for genomic DNA isolation when they reached OD 600nm = 0.1 (exponential phase).
For normalization (oriC/ter ratio of 1), dnaA thermosensitive strain was grown for 2h at non-permissive temperature (40°C) in C+Y medium and harvested for chromosomal DNA isolation.

Quantifications and statistical analysis
Data analysis was performed using R and Prism (Graphpad).
When comparing wild type phenotypes with ccrZ depletion/ complementation, a Wilcoxon rank sum test with Bonferroni adjustment was used as we did not assume a normal distribution, since some mutant cells can behave like wild type because of the variable time of depletion or possible leakiness of P lac or P Zn .
Data shown are represented as mean of at least three replicates ± SEM if data came from one experiment with replicated measurement, and ± SD if data came from separate experiments.