Multi-generational Analysis and Manipulation of Chromosomes in a Polyploid Cyanobacterium

Faithful inheritance of genetic material from one generation to the next is an essential process for all life on earth. Much of what is known about microbial DNA replication and inheritance has been learned from a small number of bacterial species that share many common traits. Whether these pathways are conserved across the great diversity of the microbiome remains unclear. To address this question, we studied chromosome dynamics in a polyploid photosynthetic bacteria using single cell, time-lapse microscopy over multi-generation lineages in conjunction with inducible CRISPR-interference and fluorescent chromosome labeling. With this method we demonstrated the long-term consequences of manipulating parameters such as cell growth, cell division, and DNA replication and segregation on chromosome regulation in a polyploid bacterial species. We find that these bacteria are surprisingly resilient to chromosome disruption resulting in continued cell growth when DNA replication is inhibited and even in the complete absence of chromosomes.

To determine whether growth phase has an effect on chromosome number, as observed for other 102 polyploid bacterial strains (Soppa, 2017), we grew chromosome labeled cells to either mid-exponential or 103 late-linear phase and imaged cells. Cells in exponential phase were both larger in size and had a higher 104 average chromosome number than cells in linear phase (Fig. 1 F-G). We also grew Wild Type (WT) cells 105 to either exponential or stationary phase and measured DNA content using Hoechst staining (Fig. 1H). 106 We observed a shift toward increased staining in exponential cells compared to stationary phase cells 107 ( Fig. 1 H-I), indicating that DNA content decreases in slow growing WT cells, similar to what we 108 observe for chromosome labeled cells. We confirmed these observations using a previously described 109 quantitative-PCR based method (Table S1) (

125
Chromosome Dynamics during PCC 7002 Growth

126
We captured chromosome dynamics in growing cells by performing time-lapse microscopy of 127 chromosome labeled strains. Unless otherwise noted cells were grown on 1% (w/v) agarose pads made 128 with A+ media under 150 µmol photons m -2 s -1 of red light (640nm) as previously described (Moore et   129 al., 2018). Using this method, we were able to image chromosome labeled cells for ~24 hr with a 30 130 minute (min) frame rate, which allowed us to track lineages for 3-4 generations ( Fig. 2A-B, Movie S1).

131
Using our image analysis software, we estimated both average and per frame chromosome number over 132 time ( Fig. 2B-C). Because of the dynamic nature of chromosome replication, which includes binding and 133 release of DNA binding proteins such as TetR, as well as limitations of our frame rate, the number of 134 puncta in a cell may not represent the true number of chromosomes in that cell. To address this challenge, 135 we created an algorithm to predict chromosome number based on previous and future frames for each 136 cell. (Fig. S2 and methods section for details). These corrected values were used for all data analysis.

137
Unless otherwise noted, only cells with traces that both started and ended during the imaging time (i.e. not 138 first or last generation cells) were analyzed.

155
Using this platform, we were able to analyze chromosome dynamics in a polyploid bacterial strain over 156 multi-generational lineages for the first time. One of the most striking observations from our movies was 157 the lack of regularity in both chromosome number and replication. Although we do not have the temporal 158 resolution to determine precise chromosome numbers at all times, we are able to visualize distinctly 159 different patterns of chromosome replication throughout our time-lapse images. Replication appears to be 160 continuous throughout the cell cycle and does not pause before cell division (Fig 2C, Movie S1). The 161 average chromosome number over a cell lifespan varied between 2.6-7.2 with a median value of 4.7, and 162 did not appear to depend on microcolony position (Fig. 2B). Consistent with previously observed data 163 (Chen et al., 2012; Jain et al., 2012), we note that chromosomes segregate relatively evenly during 164 division, but that there is some variation. Furthermore, the absolute number of chromosomes at division is 165 not constant, and not always an even number, indicating that daughter cells within a population, even 166 those from the same mother cell, may not inherit the same amount of genetic material (Fig 2D). We did 167 not observe specific positioning of chromosomes prior to division as was observed for in PCC 7942, but 168 did see relatively even spacing of chromosomes throughout the cell (Chen et al., 2012).

170
We calculated metrics such as growth rate for each cell (Fig. 2E). Cells with labeled chromosomes grew 171 similarly to WT cells as noted by the characteristic microcolony formation of PCC 7002 (Moore et al., 172 2018). Measured growth rates were also similar in chromosome labeled and WT cells, with median length 173 doubling times of 266 and 245 min, respectively. Chromosome number to cell area ratio remained 174 relatively constant for cells in different stages of microcolony formation, with a slight, but statistically 175 significant, increase in the ratio at the 8-cell stage compared to the 2-and 4-cell stages (Fig. 2F). We also 176 observed that the rate of chromosome replication, measured by dividing the difference between starting 177 and ending chromosome number in each cell by the amount of time between cell birth and division, was 7 similar in different stages of microcolony growth with a small decrease in the 2-cell stage compared to the 179 4-and 8-cell stages (Fig. 2G). As we and others have observed in non-time lapse imaged cells (Zheng & 180 O'Shea, 2017), mean protein expression was similar during growth for cells with different numbers of 181 chromosomes under these conditions (Fig. 2H). Interestingly, we did not detect differences in either the 182 total amount or rate of mOrange2 accumulation for cells with differential chromosome replication 183 patterns (Fig. 2I)    DNA replication for any conditions tested (Fig. 5A-B). ∆dnaA cells with labeled chromosomes grow and 295 have similar chromosome to area ratios as cells that are WT at the dnaA locus under standard growth 296 conditions (Fig 5C-D). dnaA deletion was confirmed by PCR (Fig. S3A). There are no additional dnaA 297 homologs within PCC 7002, and canonical clustering of DnaA binding sites is also absent from the 298 genome, further indicating that the role of DnaA is not conserved across prokaryotic species.

310
To inhibit DNA replication, we used CRISPRi to target dnaX, an essential component of the DNA 311 polymerase holoenzyme (Blinkova et al., 1993). This method allowed us to visualize dilution of 312 chromosomes over time (Fig 5E, Movie S4). Initial inhibition of DNA replication did not affect cell 313 growth (Fig. 5F -compare initial length measurements between 1 st and 2 nd generation cells) even though 314 chromosome to area ratio had already dropped significantly lower than -sgRNA cells by the 2 nd 315 generation (compare Fig. 5G and Fig. 2F). As chromosome depletion continued, cell division was 316 inhibited, resulting in cells that either did not divide or were delayed in division, leading to significantly 317 longer cells. These trends became more pronounced in later generations (Fig. 5G-H). We observed an 318 inverse correlation between cell length at division and chromosome per area ratio (Fig. 5I) (Synpcc7002_A0432). parA deletion was confirmed by PCR (Fig. S3B). ∆parA cells did not appear to 332 have any growth defects, nor chromosome segregation errors (Fig 6A-B). Additionally, we were not able larger and smaller cells than average in -sgRNA strains (Fig. 6C). As evident from both visual and 353 quantitative analysis, it is clear that chromosomes were not split evenly into large and small cells, but 354 rather that the number of chromosomes segregated into each daughter cell was proportional to cell size 355 (Fig. 6D, Movie S6). MinD depleted cells provided us with an additional tool to study differences between large and small cells with very different absolute chromosome numbers. When we grouped cells 357 by cell length at birth, we did not observe distinct differences in cell physiology (Movie S6). The number 358 of chromosomes per area were also almost identical for the smallest and largest cells (Fig. 6E) (Fig. 6F, Movie S8). Surprisingly, cells that did 376 not receive a chromosome after division continued to grow (Fig. 6G, dark purple and dark green traces), 377 and ~40% of these cells were able to divide at least once, indicating that the absence of a chromosome has 378 no effect on initial cell growth, and that neither the presence of a chromosome nor its replication is 379 absolutely required for cell division in PCC 7002 (Fig. 6H). After ~1-2 cell length doubling periods,  (Fig 6I -Left panel). The 386 consistency of the timing between chromosome loss and spikes in endogenous fluorescence was relatively 387 regular, indicating that cells have the capacity to grow, divide, and perform photosynthesis for a 388 surprisingly long period of time without genetic inputs (Fig 6G).  Table S5, noted 676 plasmids, or amplicons containing homologous recombination arms and inserts, were transformed into 677 WT or mutant backgrounds, and transformants were selected on the specified antibiotic(s). To transform, 678 we mixed ~1 ug of plasmid or amplicon with day old cells and allowed them to incubate for 4-14 hr 679 before plating on 1% Bacto-agar plates with antibiotics. Plates also included 0.5 µg/mL aTC for all 240x 680 tetO-array: TetR-sfGFP strains. Once individual colonies were detectable, they were patched to new 681 plates with single or combined antibiotics and aTC. After ~48 hr of growth patches were checked for 682 segregation and intact tetO arrays. Strains were considered segregated when no WT products could be 683 detected from PCR using primers flanking the insert and/or gene specific primers (Table S2). All 684 CRISPRi strains were freshly transformed for each experiment due to genetic instability of the constructs 685 following repeated passaging. We transformed strains in the following order: 1) 240x tetO array 686 containing construct, 2) TetR-sfGFP containing construct, and 3) mOrange2, sgRNA, or deletion 687 construct. In our system, the strand specificity of the TetR-sfGFP gene had an effect on tetO array 688 stability, with negative strand constructs being more stable than positive stand constructs.
PCR amplified inserts and backbones from base plasmids described in the STAR Methods Resource 692

704
To determine average chromosome copy number from bulk culture, we followed the procedure 705 described by Pecoraro et al., (2011). Briefly, we counted either WT or scJC0147 cells at the noted growth 706 phase and then extracted DNA using phenol:chloroform extraction (Green, Sambrook, & Sambrook, 707 2012) after initial treatment with 5 mg/mL lysozyme for 1hr shaking prior to SDS/proteinase K lysis. To 708 create samples for a standard curve, we purified and quantified ~1000 bp PCR products amplified from 709 the PCC7002 genome. These fragments were serially diluted to create standards of known concentrations.