Levels and Characteristics of mRNAs in Spores of Firmicute Species

ABSTRACT

Spores of firmicute species contain 100s of mRNAs, whose major function in Bacillus subtilis is to provide ribonucleotides for new RNA synthesis when spores germinate. To determine if this is a general phenomenon, RNA was isolated from spores of multiple firmicute species and relative mRNA levels determined by transcriptome sequencing (RNA-seq). Determination of RNA levels in single spores allowed calculation of RNA nucleotides/spore, and assuming mRNA is 3% of spore RNA indicated that only ∼6% of spore mRNAs were present at >1/spore. Bacillus subtilis, Bacillus atrophaeus, and Clostridioides difficile spores had 49, 42, and 51 mRNAs at >1/spore, and numbers of mRNAs at ≥1/spore were ∼10 to 50% higher in Geobacillus stearothermophilus and Bacillus thuringiensis Al Hakam spores and ∼4-fold higher in Bacillus megaterium spores. In all species, some to many abundant spore mRNAs (i) were transcribed by RNA polymerase with forespore-specific σ factors, (ii) encoded proteins that were homologs of those encoded by abundant B. subtilis spore mRNAs and are proteins in dormant spores, and (iii) were likely transcribed in the mother cell compartment of the sporulating cell. Analysis of the coverage of RNA-seq reads on mRNAs from all species suggested that abundant spore mRNAs were fragmented, as was confirmed by reverse transcriptase quantitative PCR (RT-qPCR) analysis of abundant B. subtilis and C. difficile spore mRNAs. These data add to evidence indicating that the function of at least the great majority of mRNAs in all firmicute spores is to be degraded to generate ribonucleotides for new RNA synthesis when spores germinate.

IMPORTANCE Only ∼6% of mRNAs in spores of six firmicute species are at ≥1 molecule/spore, many abundant spore mRNAs encode proteins similar to B. subtilis spore proteins, and some abundant B. subtilis and C. difficile spore mRNAs were fragmented. Most of the abundant B. subtilis and other Bacillales spore mRNAs are transcribed under the control of the forespore-specific RNA polymerase σ factors, F or G, and these results may stimulate transcription analyses in developing spores of species other than B. subtilis. These findings, plus the absence of key nucleotide biosynthetic enzymes in spores, suggest that firmicute spores’ abundant mRNAs are not translated when spores germinate but instead are degraded to generate ribonucleotides for new RNA synthesis by the germinated spore.

INTRODUCTION

Spores of Bacillales and Clostridiales species are formed in sporulation, are metabolically dormant, are resistant to a large variety of stresses, and can survive for long periods (1–4). Spores of some of these species also contribute to much food spoilage, as well as human disorders such as food poisoning, botulism, anthrax, and severe diarrhea (3). These properties make spores not only of general interest but also of applied interest in the health care, food, and decontamination industries.

One feature of sporulation is an orderly progression of changes in gene expression, with these changes achieved largely by the ordered appearance of a series of alternative σ factors for RNA polymerase (5), which play a major role in directing RNA polymerase to targeted genes, depending on their promoter sequences. The σ factors acting last in the developing Bacillus subtilis spore are σ^F and σ^G (σ^F/G), which direct transcription of genes encoding many important proteins found in spores’ inner regions, including the inner membrane and central core (5). The functions of many B. subtilis σ^F/G-dependent gene products are known, but there are also B. subtilis σ^F/G-dependent genes of unknown function. Notably, studies using microarray hybridization or transcriptome sequencing (RNA-seq) have also identified hundreds to thousands of different mRNAs in spores of many Bacillales and Clostridiales species (6–19). While these spore mRNAs include some that are transcribed by RNA polymerase with σ^F/G, there are also some spore mRNAs that are not sporulation specific, as well as a number of mRNAs thought to be expressed only in the mother cell in which the spore matures. Some clarity as to which spore mRNAs are most important was provided in recent RNA-seq analysis of mRNAs in B. subtilis spores (14). While this latter work found >300 spore mRNAs in spores, only 46 were present at >1 molecule/spore, with others present in only small fractions of spores in populations. Notably, most of the 46 most abundant B. subtilis spore mRNAs were encoded by σ^F/G-dependent genes and encode proteins present in spores (5, 14). This latter work, as well as a more recent study and older work, supported the conclusion that the role of B. subtilis spore mRNAs is to be degraded during spore germination and outgrowth to provide ribonucleotides for new RNA synthesis and the ultimate development of germinated spores into growing cells (19–21). However, it is not clear that the characteristics and function of spore mRNAs determined for the model organism B. subtilis are similar for mRNAs in spores of other firmicute species.

In the current work, RNA-seq analysis of spore mRNA and quantitation of total RNA in individual spores of B. subtilis and five other firmicute species allowed the following: (i) identification of mRNAs present in spores of four additional Bacillales species and one Clostridiales species, and ii) calculation of levels of different mRNAs/spore in these various species. These analyses indicated that, as with B. subtilis, only a small fraction of mRNAs in spores of Bacillus atrophaeus, Geobacillus stearothermophilus, Bacillus thuringiensis, Bacillus megaterium, and Clostridioides difficile are present at >1 molecule/spore and many of these abundant mRNAs are products of σ^F/G-dependent genes encoding proteins present in dormant spores, but at least some of these transcripts are significantly fragmented in spores. In addition, at least some of the less-abundant mRNAs in spores of these bacterial species are most likely transcribed in the mother cell compartment of the sporulating cell. However, at this time, how these mRNAs come to be in the dormant spore is not clear.

RESULTS

RNA levels in spores of various species.

The levels of total RNA in spores of the various species examined (Table 1) were determined in the following two ways: (i) spore rupture and quantitation of total RNA in spores by fluorometric analysis and by counting numbers of spores analyzed and (ii) from historic knowledge of the numbers of RNA nucleotides/spore determined for B. megaterium and B. subtilis spores many years ago (14, 22). Notably, the older work also measured DNA nucleotides/spore and was carried out before the B. subtilis and B. megaterium genomes were sequenced. However, the values for the latter genomes’ lengths are very close to those predicted from the pregenomic values when combined with data acquired later, showing that B. subtilis spores are monogenomic and B. megaterium spores are digenomic (14, 23). These values then allowed calculation of total RNA nucleotides in spores of all species analyzed (Table 1). As expected, B. atrophaeus, a close relative of B. subtilis, had only slightly fewer total RNA nucleotides/spore than B. subtilis spores, as did G. stearothermophilus spores, while B. thuringiensis and C. difficile spores had slightly more. The larger B. megaterium spores had ∼2.5 times more RNA nucleotides/spore. Using the assumed amount of total B. subtilis spore RNA that is mRNA as 3% (14, 24, 25) and applying this value to spores of all of the species examined allowed the calculation of the amount of mRNA nucleotides/spore (Table 1), with spores of B. megaterium having the most and G. stearothermophilus spores having the least.

TABLE 1.

Levels of RNA nucleotides in spores of different species^a

Species	Spore RNA
	Total RNA nt/sporeb	mRNA nt/sporeb
B. subtilis	3.3 × 107	106
B. atrophaeus	2.9 × 107	8.7 × 105
G. stearothermophilus	2.7 × 107	8.1 × 105
B. megaterium	8.3 × 107	2.5 × 106
B. thuringiensis	4.3 × 107	1.3 × 106
C. difficile	4.7 × 107	1.3 × 106

^aDuplicate samples of spores were mechanically disrupted and centrifuged, RNA in aliquots of the supernatant fluid was determined by Qubit analysis, and the numbers of spores/milliliter were determined by direct counting in a Petroff-Hauser chamber, all as described in Materials and Methods. Levels of total mRNA nucleotides per spore were calculated assuming that spore mRNA is 3% of total mRNA as was done previously for B. subtilis spores (14, 19). Values for total RNA nucleotides per spore for B. subtilis and B. megaterium spores are within 10% of the values determined many years ago by analyzing ³²P-labeled spores (22).

^bThese values are ≤ ±5%.

RNA-seq analysis of mRNA in spores of various species and levels of individual mRNAs.

RNA-seq analysis was carried out on Illumina TruSeq libraries generated in duplicate from rRNA-depleted samples and with ≥4 × 10⁶ reads for spore mRNAs from each species done in duplicate. Quantitation of B. subtilis spore mRNAs based on reads per kilobase of transcript per million mapped reads (RPKM) values revealed 49 spore mRNAs present at ≥1 copy/spore, similar to the value reported previously for rRNA-depleted RNA from B. subtilis spores (14), while spores of the close B. subtilis relative, B. atrophaeus, had 42 mRNAs at ≥1 copy/spore. Note, however, that the mRNAs in these spores at >1 copy/spore were only a minority of the total number of mRNAs detected in the spores of these two species (Tables 2 and 3). Spores of B. thuringiensis, C. difficile, and G. stearothermophilus had somewhat more mRNAs than B. subtilis spores at ≥1 copy/spore, but these abundant mRNAs were still only a small percentage of the total number of mRNAs in these spores. Even in B. megaterium spores, the ∼200 mRNAs at >1 copy/spore only represented 7% of individual mapped reads detected in spores of this species (Tables 2 and 3).

TABLE 2.

Levels of different abundance mRNAs in spores of various species^a

Group and avg RPKM	No. (avg)	Length (avg)b	RPKM (avg)c	Total ntd	Total nt/sporee	Transcripts/sporef
B. subtilis
>105	13	272	358,233	12,266,712	809,429	229
104–105	13	290	48,754	183,802	140,007	37
103–104	23	378	3,095	26,908	17,194	2
0–103	530	1,538	107	87,220	55,733	0.07
B. atrophaeus
>105	8	237	698,268	1,323,916	886,361	411
104–105	12	309	47,136	174,780	117,015	28
103–104	14	292	4,768	19,492	13,050	2.8
500–103	8	450	764	2,750	1,842	0.9
10–500	744	764	26.5	15,063	10,085	0.02
G. stearothermophilus
>105	7	252	399,851	96,805	406,420	230
104–105	10	336	41,375	135,458	106,935	32
5,000–104	16	732	6,678	1,218,048	130,964	12
3,000–5,000	12	767	4,010	598,260	61,261	6.4
1,000–3,000	12	328	1,868	51,168	5,501	1.3
450–1,000	16	495	664	55,440	5,956	0.6
1–450	534	869	54	464,046	49,894	0.1
B. megaterium
>105	5	261	229,770	298,701	812,466	498
104–105	36	270	31,843	309,514	841,878	70
5,000–104	22	321	6,974	49,250	133,933	16
2 × 103 to 5 × 103	47	369	3,015	52,289	142,226	6.4
103 to 2 × 103	93	450	1,376	57,586	156,634	3
5–1,000	2,741	936	131	336,090	914,166	0.3
B. thuringiensis
>5 × 104	7	339	115,342	313,236	95,658	40
104 to 5 × 104	9	328	14,732	48,118	14,690	4.7
5,000–104	15	358	6,596	42,960	13,115	2.6
3,000–5,000	19	549	3,827	74,060	22,610	2.2
2,000–3,000	27	472	2,418	48,427	14,722	1.3
1,000–2,000	99	589	1,389	164,643	50,266	0.2
1–1,000	2,268	1,502	212	3,406,536	1,040,016	0.15
C. difficile
>104	5	451	281,406	634,571	761,485	338
103–104	21	828	2,650	46,078	55,293	3
800–103	25	1,044	898	23,437	28,124	1
600–800	61	1,332	675	54,815	65,814	0.8
400–600	93	776	490	35,362	42,434	0.6
300–400	126	1,023	343	44,212	53,054	0.4
100–300	1,184	1,049	200	248,403	298,084	0.2

^aRNA-seq data were obtained as described in Materials and Methods and grouped by RPKM values as shown. All values shown are averages from two separate RNA-seq runs for each species.

^bLengths of all transcripts were the coding sequence plus an extra 50 nt at each end to account for 5′ and 3′ extensions (14).

^cIndividual RPKM values shown are averages of values obtained in 2 RNA-seq analyses, and these values differed by less than 15%.

^dTotal nucleotides were calculated as the average RPKM values times the average kilobases of each transcript times numbers of mRNAs in each group.

^eValues were corrected to give the sum of total mRNA nucleotides in each group as determined from the calculated mRNA nucleotides/spore in Table 1.

^fValues were calculated as total mRNA nucleotides in each group/average mRNA length times number of mRNAs in the group.

TABLE 3.

Levels of mRNAs present at ≥1/spore in spores of various species^a

Species	No. of different mRNAs	mRNAs ≥1 (no./spore)b	mRNAs ≥1 (% of total mRNAs)
B. atrophaeus	607	42	7
B. megaterium	2,944	203	7
B. subtilis	579	49	8
B. thuringiensis	2,446	67	3
C. difficile	1,515	51	3
G. stearothermophilus	607	57	9

^aValues were obtained from data in Table 2.

^bWe estimate that these values are ±∼10%.

Characteristics of the 60 most abundant spore mRNAs in different species.

Previous work (14) showed that most abundant B. subtilis spore mRNAs encoded proteins present in the dormant spore. This was also the case for the 60 most abundant B. subtilis spore mRNAs identified in the current work and for the abundant B. atrophaeus spore mRNAs as well (Table 4; see also Table S1 in the supplemental material). The percentages of the 60 most abundant mRNAs in spores of the other three Bacillales species that encoded spore proteins were lower than those in B. subtilis spores, although this is at least partly because of the following. (i) The complete proteomes of B. megaterium, B. thuringiensis, and G. stearothermophilus spores have not been determined. (ii) The proteins encoded by many of the most abundant mRNAs in these species have no obvious homologs in spores of Bacillus anthracis, Bacillus cereus, and B. subtilis, for which spore proteomes have been determined (26, 27). Thus, additional abundant mRNAs in spores of B. megaterium, B. thuringiensis, and G. stearothermophilus may well also encode spore proteins. The C. difficile spore proteome has, however, been determined (27), and using these data to analyze the 60 most abundant C. difficile spore mRNAs showed that at least 33% of these mRNAs encode proteins detected in dormant spores (Tables 4; see also Table S1).

TABLE 4.

Characteristics of 60 most abundant mRNAs in spores of different species^a

Species	% encoding spore proteins	% transcribed by RNA polymerase with:
		σF/G	σE/K	σA/B/D/H	Unknownb
B. subtilis	93	75	13	3	10
B. atrophaeus	93	75	6	15	5
B. megaterium	55	57 (30)c	8	17	18
B. thuringiensis	85	53 (20)c	10	33	3
G. stearothermophilus	90	77 (17)c	6	12	7
C. difficile	35	12	5	7	75

^aThe mRNA σ-factor dependence and whether they encode proteins in spores were determined as described in Materials and Methods, and this is all shown in Table S1 in the supplemental material.

^bThese mRNAs encode a protein not identifiable in spores due to the lack of spore proteomic data or data on the σ-factor dependence of the transcription of the coding genes.

^cValues in parentheses are percentages of σ^F/G-dependent genes identified by promoter mining as described in Materials and Methods.

A second characteristic of the abundant mRNAs in B. subtilis spores is that 75% are under σ^F/G control (Table 4; see also Table S1) and transcribed only in the developing forespore, as determined by transcriptomic analyses during sporulation (5, 14, 28). Not surprisingly, given the close phylogenetic relatedness between B. atrophaeus and B. subtilis, 45 of the 60 most abundant mRNAs in B. atrophaeus spores could also be assigned as σ^F/G dependent based on the homology of the coding genes with σ^F/G-dependent genes of B. subtilis (Table 4; see also Table S1). Notably, B. atrophaeus and B. subtilis spores also had a few abundant mRNAs likely transcribed in the mother cell compartment of the sporulating cell and under the control of the mother cell-specific σ factors, σ^E and σ^K (Table 4; see also Table S1). B. atrophaeus spores also had some abundant mRNAs that were transcribed by RNA polymerase with either the housekeeping σ factor σ^A or the stress response σ factor, σ^B (Table 4; see also Table S1).

Comparison of B. subtilis σ^F/G-dependent genes with the genes encoding the 60 most abundant mRNAs in B. megaterium, B. thuringiensis, and G. stearothermophilus spores identified 27, 33, and 60%, respectively, that were also likely under σ^F/G control (Table 4). However, the σ^F/G dependence of significant numbers of abundant mRNAs in spores of the other Bacillales species could not be determined by this method. To identify more σ^F/G-dependent genes in the other Bacillales species, we took advantage of the fact that B. subtilis σ^G promoter sequences are similar in at least some B. megaterium and G. stearothermophilus genes as well as in genes of B. cereus, a very close relative of B. thuringiensis (29). Additionally, the similarity between the σ^F and σ^G promoter sequences (28) facilitated our search for genes that were under σ^F control. Consequently, consensus B. subtilis σ^F/G promoter sequences (28) were used to search upstream of the translation start sites of genes encoding the 60 most abundant mRNAs in B. megaterium, B. thuringiensis, and G. stearothermophilus. This analysis found almost all genes initially identified as σ^F/G dependent by homology with known σ^F/G-dependent B. subtilis genes (data not shown), as well as a large number of additional σ^F/G-dependent genes (Table 4; see also Table S1). The additional σ^F/G-dependent genes identified raised the percentages of σ^F/G-dependent mRNAs in these other Bacillales species to 53, 57, and 77% (Table 4; see also Table S1). Note that eight of the most abundant B. subtilis spore σ^F/G-dependent mRNAs are from four bicistronic operons (14), and only the first gene would have a σ^F/G-dependent promoter. There were also a number of adjacent genes encoding abundant spore mRNAs in the other species examined, but it has not been definitively shown that these genes are cotranscribed. Thus, there may be more σ^F/G-dependent genes than we have identified. It was also notable that, again, there were significant numbers of abundant σ^E/K- and σ^A/B-dependent mRNAs in the spores of these other Bacillales species (Table 4; see also Table S1). The σ-factor dependence of most of the abundant C. difficile spore mRNAs has not been assigned, but of those that could be assigned, ∼45% are known to be σ^F/G dependent, and σ^A and σ^E/K-dependent mRNAs were also identified (11) (Table 4; see also Table S1). However, the B. subtilis σ^F/G consensus promoter sequences were not used to search for promoters in C. difficile since there are no σ^F/G-dependent promoter sequences known in this organism with any certainty.

In all of the Bacillales species examined, many of the abundant spore mRNAs encode proteins with no known function—“y” genes or hypothetical proteins (Table S1). However, one group of genes in these species is the ssp genes encoding small, acid-soluble spore proteins (SASP). Some of these proteins, the abundant α/β-type small, acid-soluble proteins, usually products of sspA or sspB genes and most likely sspF as well, bind to and saturate spore DNA, resulting in protection of the DNA from a variety of damaging agents, including radiation, heat, and genotoxic chemicals (1–3). Specific roles of other ssp gene products, with SspE being by far the most abundant, have not been identified, but almost all Ssp proteins are rapidly degraded as spore germination is completed, yielding amino acids that can be essential for protein synthesis in this period of development, as the dormant spores lack many amino acid biosynthetic enzymes (16, 20, 25, 26, 30, 31). Consistent with these results, we found that the most abundant C. difficile spore mRNAs also encode α/β-type SASP (Table S1), although SspE is not present in Clostridiales (32). It was also reported a number of years ago that three of the five most abundant mRNAs in Clostridium novyi spores also encode SASP, including at least two α/β-type SASP (9).

Coverage of RNA-seq reads for high and low abundance spore mRNAs.

Both the current as well as previous work has found large numbers of mRNAs in spores of all species that have been examined (14, 15, 17). However, it is clear in B. subtilis that at least 99% of the spore mRNAs, including both abundant and much less abundant ones, are not needed to direct synthesis of some essential protein(s) when spores germinate, since at least 99% of these mRNAs can be fragmented with no notable effects on spore germination or outgrowth (19). This information, as well as that RNA-seq measures only RNA fragments of ∼75 nucleotides, raises the question of whether spore mRNAs are intact and even capable of directing translation of full-length proteins. To determine the extent to which the most abundant of these spore mRNAs are fragmented, we examined overall read coverage for 10 of the 60 most abundant spore mRNAs from all species studied, including five of the most abundant of these mRNAs and five at lower abundance (Fig. 1; see also Fig. S1 and Table S1 in the supplemental material). The most abundant mRNAs were all under σ^F/G control and thus were almost certainly synthesized in the developing forespore, while at least some of the lower-level mRNAs were under σ^E/K control and were most likely expressed in the mother cell compartment of the sporulating cell (see Discussion). This analysis found, as seen previously (14, 19), that read coverage was generally relatively uniform for the most abundant spore mRNAs (Fig. S1). However, there were numerous examples of small gaps in the coverage for these more abundant mRNAs. Specific examples included sspE, ypzG, yizC, and sspJ mRNAs of B. atrophaeus spores; ykzP, sspF, and ytzL mRNAs of B. subtilis spores; yizC and yhcN mRNAs of B. thuringiensis spores; sspI, sspK, and yhdB mRNAs of B. megaterium spores; ypzG, yrrD, and yqfX mRNAs of G. stearothermophilus spores; and even larger gaps in the abundant C. difficile spore trxB3 mRNA and the σ^G-dependent mRNA encoding a hypothetical protein. Notably, a number of the gaps in the coverage of these various mRNAs were similar to the gaps in B. subtilis rRNAs from spores that had been incubated on plates for long periods at 37°C and greatly fragmented (19). While the coverage determined in the current work was generally less uniform for the less abundant mRNAs, the levels of reads for these mRNAs were low.

FIG 1.

Coverage of reads in RNA-seq for one high abundance (left) and one lower abundance mRNA (right) from among the 60 most abundant spore mRNAs in all six species studied as shown in Table S1 in the supplemental material. Coverage was determined as described in Materials and Methods, and the encoded protein and its location in the genome are shown above each panel. The left side of all panels gives the numbers of reads along the gene.

Analysis of spore mRNAs by RT-qPCR.

Previous work (19) indicated that spore mRNAs became fragmented when spores were incubated at 37°C such that while no mononucleotides were generated, they were no longer detected by RNA-seq, which requires reads of ∼75 nt. This finding as well as the observation that spore mRNAs do not seem to be translated to give protein early in outgrowth (13, 16) suggest that at least some of the mRNA isolated soon after spores are formed is fragmented. This was also suggested by the gaps in the reads from some spore mRNAs as noted above. To directly determine whether spore mRNAs were or were not largely intact, reverse transcriptase quantitative PCR (RT-qPCR) analyses were performed using cDNA generated from total RNA from either B. subtilis or C. difficile spores and normalized using full-length gene-specific amplicons for the corresponding RNA. These analyses used a variety of primer pairs located at the 5′ and 3′ ends of predicted mRNAs as well as in an internal region (Fig. 2A; see also Table S2 in the supplemental material). The results of this analysis for B. subtilis suggest that for two genes, only a miniscule fraction of the mRNAs includes the intact transcript, while the level of the third intact mRNA from the yqfX gene was ∼10-fold lower than the corresponding gene amplicon (Fig. 2B). In contrast, the levels of the full-length mRNAs for two of the genes examined from C. difficile (CDIF630erm_02337 and sspB) were decreased either minimally for the CDIF630erm_02337 gene or ∼30% for the sspB gene compared to those of internal regions (Fig. 2C). However, the relative abundance of the intact transcript for the C. difficile sspA2 gene was decreased >10-fold compared to that of internal regions (Fig. 2C).

FIG 2.

RT-qPCR analysis on cDNA libraries from B. subtilis and C. difficile spore mRNAs using primer pairs at both the 5′ and 3′ mRNA termini and in the mRNAs’ more internal regions. (A) Schematic for primer binding sites relative to the ribosomal binding site (RBS) and rho-independent terminator of the transcripts. Each condition was completed in biological and technical triplicate. However, biological replicate 2 with the ytzL internal forward and external reverse primer pair was excluded based on the negative control. The results represent mean values of normalized starting quantities from combined biological and technical triplicates. The error bars represent standard deviation (SD). Relative abundance of amplicons in B. subtilis (B) and C. difficile (C) produced using the four different primer sets. The relative abundance of each RT-qPCR amplicon was determined as described in Materials and Methods. P values were determined by a paired two-tailed t test comparing normalized starting quantities to the normalized full-length transcript; ns, not significant at P > 0.05; *, significant at P ≤ 0.05; **, significant at P ≤ 0.01; ***, significant at P ≤ 0.001; ****, significant at P ≤ 0.0001.

DISCUSSION

Analysis of the most abundant mRNAs in spores of the six firmicute species examined in the current work has led to a number of important conclusions. The first is that only a minority of mRNAs in the spores of these species are present at more than 1 molecule/spore, averaging only ∼6% of all individual mRNAs identified in these spores. This finding indicates that at least ∼94% of the individual mRNAs in spores are extremely unlikely to be important for generation of proteins that will affect the whole spore population. This conclusion is consistent with the idea that the primary function of spore mRNAs is to serve as a source of ribonucleotides for RNA synthesis soon after spore germination is triggered (14, 19), since dormant spores of at least B. subtilis, B. anthracis, and B. cereus, the latter two close relatives of B. thuringiensis Al Hakam, as well as spores of B. megaterium and C. difficile, lack several nucleotide biosynthetic enzymes, which are synthesized only at defined times in spore outgrowth (16, 20, 25, 26). It is also clear that, as was the case for almost all B. subtilis mRNAs present at >1 molecule/spore, most to many of the 60 most abundant mRNAs in all five Bacillales species examined are transcribed in the developing spore under σ^F/G control and encode proteins found only in dormant spores. The latter number would likely have been higher if the proteome of B. thuringiensis, B. megaterium, and G. stearothermophilus spores had been determined, although many spore proteins in these species were identified by homology to known B. cereus, B. anthracis, or B. subtilis spore proteins.

Perhaps not surprisingly, because of the long time that has elapsed since the Bacillales and Clostridiales orders split from a common ancestor, the C. difficile spore mRNAs were the outlier in this analysis, as only one-third of the 60 most abundant spore mRNAs encode known spore proteins. One factor contributing to this latter low abundance may be that many abundant C. difficile spore transcripts encode hypothetical proteins. In addition, while only 12% of the 60 most abundant C. difficile mRNAs have been shown to be under σ^F/G control (17), σ^F/G-dependent genes comprised ∼44% of those mRNAs whose σ factor dependence could be assigned. It is also notable that the study of the C. difficile spore proteome identified only 1,000 proteins, while similar studies with spores of other species identified up to 1,800 proteins (16, 27). Perhaps many low abundance C. difficile spore proteins were below the limit of detection. Hopefully, continued proteomic and transcriptomic studies of spores and sporulating bacteria of different species will allow assignment of the σ factor dependence of more of these genes as well as whether these genes encode spore proteins.

In contrast to the characteristics shared between mRNAs from spores of B. subtilis and the other five species examined, there are some obvious differences. This was true even for B. atrophaeus, a close relative of B. subtilis (32, 33). While the percentages of the 60 most abundant spore mRNAs under σ^F/G control were almost identical in these two species, the percentage of genes under σ^A control was ∼6 times higher in B. atrophaeus. Indeed, with the exception of B. subtilis, the percentages of the 60 most abundant spore mRNAs from all species exclusively under σ^A control ranged between 7 and 30%; many of these σ^A-dependent mRNAs encoded gene products involved in protein synthesis, including multiple ribosomal proteins (see Table S1 in the supplemental material). Notably, proteomic analysis of spores of B. cereus, B. subtilis, and C. difficile found σ^A peptides in all three replicates of the analysis with B. cereus and B. subtilis spores and in one of two replicates from C. difficile spores (16). Thus, it is not surprising that there are some σ^A-dependent transcripts in the collection of spore mRNAs. Indeed, proteins encoded by many σ^A-dependent mRNAs will be needed to assemble a spore, and some are needed early in forespore development to ensure efficient spore germination and spore outgrowth (34). This raises the question of what role mRNA stability might play in the eventual catalog of mRNAs that are packaged into the spore. For most mRNAs in these organisms, there has been minimal work on their stability. However, it might be informative to examine the stability of the σ^A-dependent mRNAs found at high levels in spores relative to those at low abundance.

One other group of spore mRNAs is notable, and this is mRNAs from genes in the σ^E/K regulon that are transcribed in the mother cell compartment of the sporulating cell, primarily genes encoding proteins in spores’ outer layers of the coat, crust, or exosporium (5); note that spores generally have only one of these latter two outermost layers (35, 36). These σ^E/K-dependent transcripts comprised an average of ∼8% of the 60 most abundant spore mRNAs in all species examined. However, in contrast to the findings with σ^A noted above, no σ^E peptides were found in spores of B. cereus, B. subtilis, or C. difficile (16); these latter analyses also found no σ^K peptides in B. cereus or C. difficile spores, although a possible σ^K peptide was found in one of three replicates of analyses of B. subtilis spores. Thus, while it seems likely that there could be σ^A-dependent transcription in the developing forespore, σ^E/K-dependent transcription seems less likely. In a study on σ factors governing sporulation pathways in C. difficile, Fimlaid and colleagues (11) showed that the activation of σ^G, which regulates the production of key forespore proteins (e.g., SspA, SspB, DacF, SpoVT, and SpoVAD), is only dependent on σ^F and is independent of σ^E (14). They also demonstrated that, contrary to findings in B. subtilis, σ^G is dispensable for σ^K activation in C. difficile. The differences in sporulation sigma factor regulation between C. difficile and B. subtilis can possibly explain why far fewer σ^E/K-dependent transcripts and no σ^E/K-dependent peptides were detected in C. difficile spores compared with σ^G-dependent mRNAs and peptides.

It is certainly possible that the σ^E/K-dependent transcripts in spores arise from weak σ^F/G-dependent transcription in the forespore of some of these genes. However, another possible explanation for the presence of σ^E/K peptides and putative σ^E/K-dependent transcripts in B. subtilis spores—and perhaps even σ^A-dependent transcripts in spores of all species examined—was suggested a number of years ago (17) because there is a connection between the mother cell and forespore cytoplasm termed a feeding tube (37–41). This feeding tube is likely how the mother cell transfers small molecules, such as ATP and amino acids, to the developing spore. Perhaps mRNAs or mRNA fragments also move from the mother cell into the forespore via this feeding tube. The precise time in sporulation when the feeding tube closes is not clear but must be rather late in forespore development since the developing spore cannot make either ribonucleotides or many amino acids and also likely cannot make its own ATP, as at least several TCA cycle enzymes are absent (16, 26, 42). Consequently, this feeding tube could be how mother cell mRNAs come to be in the forespore.

One major unanswered question is whether mother cell mRNAs in the spore, including even the most abundant ones, are intact or only fragments, and one analysis in which a number of B. subtilis spore mRNAs were isolated by capture of their 5′ ends indicated that at least some of these mRNAs were significantly fragmented (43). The occurrence of fragmentation of the mRNAs was also suggested by the analysis of the coverage of mapped RNA-seq reads in the current work, although these results were far from definitive. However, these latter results were supported by RT-qPCR analyses, which indicated that at least three abundant mRNAs in B. subtilis spores and one in C. difficile spores were significantly fragmented, such that either only 10% to less than 1% of these mRNAs (B. subtilis) or only ∼5% (C. difficile) represent intact coding mRNAs. This fragmentation could have taken place late in sporulation, or even during the incubation of sporulation plates after the phase bright spore has formed, in order to allow mother cell lysis to facilitate spore purification. Indeed, continued incubation of B. subtilis sporulation plates at 37°C resulted in fragmentation not only of spore mRNA but even rRNA (19). How this fragmentation occurs in dormant spores is not clear, including whether it is enzyme catalyzed, although it is most likely endonucleolytic, since even when almost all spore rRNA and mRNA are fragmented, there is no increase in spore ribonucleotide pools (19), and these RNAs’ degradation in B. subtilis cells is thought to be initially endonucleolytic (44). However, the fragmented RNAs and any intact spore mRNAs are almost certainly degraded completely soon after spore germination is initiated, providing ribonucleotides for new transcription (13, 19, 21).

The minimal levels of two full-length B. subtilis spore mRNAs and one C. difficile spore mRNA detected by RT-qPCR strongly suggest that the great majority of these spores’ mRNAs are incapable of directing synthesis of protein upon spore germination and outgrowth. However, in contrast to the RT-qPCR results noted just above, RT-qPCR analyses of two abundant C. difficile spore mRNAs, including sspB, indicate the presence of high levels of full-length mRNA. Thus, these latter mRNAs could be translated in the outgrowing spore. However, in B. subtilis and Escherichia coli, the presence of α/β-type SASP, such as SspA or SspB, results in the death of outgrowing spores or growing cells, respectively (45, 46), so it seems very unlikely that the sspB mRNA would be translated in outgrowing spores. Why the abundant C. difficile spore mRNAs were more stable than those in B. subtilis spores is not clear, but mRNAs in Clostridium novyi spores were also reported to be quite stable (9). Perhaps detailed studies of RNA degradation in dormant spores of Bacillales and Clostridiales species may elucidate the mechanisms of this degradation and why it may be slower in spores of Clostridiales.

MATERIALS AND METHODS

Bacterial species/strains used and spore preparation and purification.

The spores used in this work were from the following: (i) B. subtilis PS533 (47), a 168 derivative that carries plasmid pUB110 encoding resistance to kanamycin (10 μg/ml); (ii) B. atrophaeus ATCC 9372; (iii) B. thuringiensis Al Hakam (48); (iv) B. megaterium QMB1551, originally obtained from H. S. Levinson; (v) G. stearothermophilus ATCC 12980; and (vi) C. difficile 630Δerm obtained from A. Shen. Spores of B. atrophaeus, B. subtilis, and B. thuringiensis were prepared as described previously on double strength Schaeffer’s glucose agar plates (49, 50), which were incubated for ∼48 h at 37°C. B. megaterium spores were prepared on supplemented nutrient broth agar plates (49), which were incubated for 48 h at 30°C, and G. stearothermophilus spores were prepared as described previously (51) on agar plates that were incubated for ∼48 h at 65°C in lightly sealed bags with an empty plate containing water to keep the humidity high. C. difficile spores were prepared on 70:30 sporulation media agar plates in an anaerobic chamber as described previously (52). Spores of all species were scraped from plates into 4°C water and highly purified over ∼7 d as described previously (49, 53, 54), including a final purification by centrifugation through a very high-density (Histodenz) medium to remove spore-associated debris as much as possible. Final spore preparations were >98% free from growing or sporulating cells and germinated spores as determined by phase-contrast microscopy, and spores pelleted by centrifugation had no obvious debris on the pellet surface. Spores were stored in water at 4°C protected from light.

Extraction and analysis of spore RNA.

One milliliter of each strain of dormant spores at an optical density at 600 nm (OD₆₀₀) of 25 was pelleted by centrifugation. Prior to pelleting, to get a more accurate quantitation of the number of spores present, spore samples were diluted 1 to 25 in water, to give an OD₆₀₀ of ∼1.0, and the numbers of spores were counted in a Petroff-Hausser chamber to determine spores per milliliter. Spores were disrupted by shaking with glass beads in a MiniBeadbeater (Biospec Products, Bartlesville, OK) for 1 min, followed by 1 min on ice with a total of two disruption periods and giving >98% spore breakage (14, 19, 51). RNA was extracted using the RiboPure-bacteria kit (Thermo Fisher Scientific, Waltham, MA) as previously described (54). Total RNA was quantitated using the Qubit RNA HS (high sensitivity) assay kit (Thermo Fisher Scientific) (55) with the Qubit fluorometer, using appropriate dilutions to get measurements that fell within the standard curve (500 ng/ml maximum concentration). Generally, a dilution range of 1:500 to 1:1,000 of the purified initial total RNA gave quantifiable values. From this obtained concentration and the previously counted spore value, values of total RNA per spore were determined.

Total RNAs extracted and purified from spores of various species were submitted to University of Connecticut’s Center for Genome Innovation for rRNA depletion, Illumina TrueSeq RNA-seq library preparation, and next-generation sequencing. Briefly, total RNA was ribo-depleted to remove most rRNA using a Zymo-Seq RiboFree total RNA library kit (Zymo Research, Irvine, CA), the quality of the ribo-depleted RNA was examined using the Agilent TapeStation 4200 D1000 high-sensitivity assay, and RNA-seq on RNA from all species was carried out on duplicate samples using an Illumina Next-Seq 500/550, all as described previously (14, 19). Sequencing read depth was targeted at >4 × 10⁶ reads per sample. RNA-seq data were processed and analyzed as described previously (14), and levels of mRNA nucleotides in an individual spore in populations were calculated using values for total RNA nucleotides per individual spore determined, as described above, and assuming that spore mRNA comprised 3% of total RNA (14).

For all Bacillales species, the mRNAs that were present at ≥1 molecule/spore were further analyzed by comparison with transcripts whose expression are dependent on sporulation association σ factors as determined in B. subtilis (5, 14, 28, 56). This analysis was also carried out for the mRNAs present at >1 molecule/spore in C. difficile spores (11). Additionally, for spores of all of these species, the products of the genes encoding the mRNAs present at ≥1 molecule/spore were compared with the published spore proteomes of B. subtilis, a very close relative of B. atrophaeus; the spore proteomes of B. cereus and B. anthracis, close relatives of B. thuringiensis; and the C. difficile spore proteome (16, 26, 27). Unfortunately, the complete B. megaterium and G. stearothermophilus spore proteomes have not been determined, but comparisons with proteomic data for spores of the other Bacillales did allow identification of orthologs of many B. subtilis genes in these two species.

Identification of σ^F/G-dependent promoters in spores of other Bacillales species.

The genes encoding the 60 most abundant spore mRNAs from B. subtilis, B. megaterium, B. atrophaeus, B. thuringiensis, and G. stearothermophilus were considered in the promoter analysis. Briefly, 250 nt of sequence upstream of the translation start site and the first 48 nt of the coding sequence were included in the output files for the analysis. After upstream sequences were acquired, files were scanned for motifs of σ^F and σ^G consensus sequences as previously defined (28) through the Find Individual Motif Occurrences (FIMO) algorithm provided through MEME suite (57, 58). Outputs were individually assessed for confirmation of potential consensus sequences, >90% of which were <100 nt upstream of the translation start site, as seen in B. subtilis (28, 29).

Extraction and amplification of chromosomal DNA.

DNA was isolated from B. subtilis and C. difficile growing cells by standard procedures. Genomic DNA served as a template for PCRs (Phusion polymerase or Taq polymerase) (New England BioLabs, Ipswich, MA) to generate amplicons. Primers (see Table S2A in the supplemental material) were designed to bind an additional 50 nt (25 nt upstream and downstream) flanking the coding sequence to generate an amplicon. Amplicons were agarose gel purified and extracted and quantitated using the Qubit DNA HS assay kit (Thermo Fischer Scientific) with the Qubit fluorometer using a 1:100 dilution to get measurements within the standard curve. Purified amplicons were serial diluted to 10⁷ to 10² copies/μl to generate a standard curve.

RT-qPCR of spore RNA.

RNA was extracted from spores as described above and was quantified using a Take3 microvolume microplate along with a BioTek Synergy H1 multimode plate reader (BioTek Instruments, Inc., Winooski, VT); RNA samples were stored at −80°C until cDNA synthesis. One microgram of RNA was used to generate the cDNA library. First-strand synthesis was completed with random hexamer priming using the SuperScript III first-strand synthesis system (Thermo Fischer Scientific). A negative control was generated by omitting the SuperScript III reverse transcriptase. The samples were diluted 2-fold with molecular grade water, and cDNA libraries were stored at 20°C until quantitative PCR (qPCR). Primers utilized in qPCRs were selected to bind within the ribosomal binding site (RBS) (external forward primer) and before the rho-independent terminator (external reverse primer); internal primers were also used to amplify a 75-nt region within the mRNA’s coding sequence (Fig. 2A; see also Table S2B). The reactions were optimized for annealing temperature using SsoAdvanced Universal SYBR green supermix (Bio-Rad Laboratories, Hercules, CA) with or without MgCl₂ (50 μM). RT-qPCR was standardized using standard curves from purified DNA amplicons as described previously. A total of 2.5 μl of the cDNA library was used as a template for most PCRs with specific primer pairs (Table S2B). For each gene, primers included the following: (i) an external primer pair amplifying the region starting at the RBS to just prior to the likely transcription stop site; (ii) the external forward primer with an internal reverse primer, amplifying from the RBS to within the coding region; (iii) an internal forward primer with the external reverse primer; and (iv) an internal forward primer and an internal reverse primer amplifying an internal segment of the coding region (Table S2B). For C. difficile primer pairs and the B. subtilis yhcV internal primer pair, the cDNA library was diluted 1:10 before the template’s addition into the reaction mixture. A negative control containing water and a negative control of the cDNA library, where RNA served as the template while omitting SuperScript III reverse transcriptase, were run concomitantly. qPCR plates were run on a CFX96 real-time thermal cycler (Bio-Rad Laboratories). Starting quantities (SQs) were obtained by CFX Maestro software. SQs for each gene were normalized based on the most abundant gene segment to generate a normalized percentage. Data was visualized using Prism version 8 (GraphPad Software, San Diego, CA).

Statistical methods.

qPCR SQs of gene segments generated from combinations of primer pairs were compared by the paired two-tailed t test. P values of less than 0.05 were considered statistically significant. Statistical tests were performed with Prism version 8 (GraphPad Software, San Diego, CA).

Data availability.

All RNA-seq data discussed in this publication have been deposited in NCBI’s Gene Expression Omnibus (GEO) (59) and are accessible through GEO series accession numbers GSE173854, GSM5281858, GSM5281859, GSM5281860, GSM5281861, GSM5281862, GSM5281863, GSM5281864, GSM5281865, GSM5281866, GSM5281867, GSM5281868, and GSM5281869.