On the nature of mycobacteriophage diversity and host preference

Abstract

The complete genome sequences of over 220 mycobacteriophages reveal them to be highly diverse, with numerous types sharing little or no nucleotide sequence identity with each other. We have determined the preferences of these phages for M. tuberculosis and for other strains of M. smegmatis, and find there is a correlation between genome type (cluster, subcluster, singleton) and host range. For many of the phages, expansion of host range occurs at relatively high frequencies, and we describe several examples in which host constraints occur at early stages of infection (adsorption or DNA injection), and phages have the ability to expand their host range through mutations in tail genes. We present a model in which phage diversity is a function of both the ability of phages to rapidly adapt to new hosts and the richness of the diversity of the bacterial population from which those phages are isolated.

Introduction

The bacteriophage population is vast (Hendrix, 2002, 2003; Suttle, 2007), dynamic (Wilhelm et al., 2002), and likely extremely old (Abrescia et al., 2012). The genome sequences of approximately 1,000 phages, isolated from a broad range of bacterial hosts, have been deposited in public databanks, and it is clear that we have only begun to appreciate the enormous sequence diversity of the phage population as a whole. For some bacterial genera, a substantial number of phage genome sequences have been determined, including Escherichia, Lactococcus, Mycobacterium, Pseudomonas, Staphylococcus, and Streptococcus, and these also typically span a broad range of diversity (Adriaenssens et al., 2012; Brussow, 2001; Hatfull, 2012b; Kwan et al., 2005, 2006; Smith et al., 2012). However, for many such collections, the phages were not isolated on the same species or strain, the host ranges are not well-defined, and there are many examples where phages have preferences for different strains of the same bacterial species (Lobocka et al., 2012).

The constraints on infection are not completely understood, but involve many different mechanisms, including restriction-modification (Hyman and Abedon, 2010), receptor availability (Hyman and Abedon, 2010), abortive infection, which can be mediated by a variety of processes (Fineran et al., 2009; Hoskisson and Smith, 2007), Clustered Regularly Interspaced Short Palindromic Repeats (CRISPRs) (Deveau et al., 2010), and modification of the translational apparatus (Amitsur et al., 1987). In turn, phages have evolved ways to overcome these constraints, such as through spontaneous mutation (Desplats and Krisch, 2003), diversity generating systems (Medhekar and Miller, 2007), site-specific recombination (Plasterk et al., 1984; Sandmeier, 1994), or molecular mimicry (Roberts et al., 2012). The totality of these processes can be viewed in the context of a grand microbial battle, in which the hosts are constantly under attack from the viruses, bacteria evolve effective ways to resist infection, and the viruses co-evolve to either overcome the resistance, or acquire the ability to infect a new host. After perhaps more than three billion years, an equilibrium has been established where both bacterial and phage populations are maintained (Stern and Sorek, 2011).

Mycobacteriophages are viruses that infect mycobacterial hosts and are of interest for two primary reasons. First, the large set of complete mycobacteriophage genome sequences shows them to be highly varied, thus offering insights into phage genome diversity and the evolutionary mechanisms acting to generate such diversity (Hatfull, 2010). Secondly, mycobacteriophages provide a rich set of tools for developing genetic systems for their hosts –including Mycobacterium tuberculosis, the causative agent of human tuberculosis (TB) – as well as fueling novel strategies for diagnostic, preventative, and therapeutic approaches for TB (Hatfull, 2012b). In addition, the discovery of new mycobacteriophages and their genome characterization represents a powerful platform for introducing novice scientists to authentic research experiences (Hanauer et al., 2006; Hatfull et al., 2006).

Currently, there are 221 completely sequenced mycobacteriophage genomes (Hatfull, 2012a; Henry et al., 2010; Pope et al., 2011a; Pope et al., 2011b). All of these [with the sole exception of DS6A (Redmond and Carter, 1960)] were either isolated on M. smegmatis mc²155, or are known to infect this strain. Comparative genomic analysis at both the nucleotide and gene content levels shows that even though these phages enjoy a common host and are thus capable of being in genetic contact, they span a broad range of diversity, albeit in a heterogeneous manner (Pope et al., 2011b). Specifically, there are groups of phages that share substantial levels and extents of DNA similarity, as revealed by either Dotplot analyses or BLAST comparisons. These can be grouped into clusters, where genomes within a cluster have recognizable sequence similarity over a span of more than 50% of their genome lengths (Hatfull et al., 2006). Some phages, however, have no close relatives, and these are designated as singletons. The 221 sequenced genomes fall into 15 clusters, and eight singletons, and there is little or no DNA sequence similarity between these groups (Hatfull, 2012a). The host ranges of only a few of these phages have been reported, suggesting at least a partial correlation between host preference and genetic relationships (Rybniker et al., 2006).

Both the size and the diversity of the different clusters vary enormously. The largest is Cluster A, containing 60 genomes, and the smallest – aside from the eight singletons – are Clusters M, N and O, each with two. Within some clusters, the genomes are extremely similar, and the five Cluster G phages are a good example of this. These all have a similar (although non identical) gene content, and differ mostly by a small number of nucleotide substitutions (for example, Angel and BPs differ by the insertion of a small mobile element MPME1 in BPs, and 138 nucleotide substitutions) (Sampson et al., 2009). Other clusters are more diverse and can be further divided into subclusters, such that genomes within a cluster have similar gene contents and genome organizations, but different subclusters have different degrees of nucleotide similarity. For example, Cluster A can be divided into at least nine subclusters. The total number of non-divided clusters, subclusters and singletons is 44 (Hatfull, 2012a). Mycobacteriophage genomes are also pervasively mosaic such that each genome can be considered as a specific assemblage of individual segments – commonly single genes – joined together in a particular combination (Hatfull, 2010; Pedulla et al., 2003). However, these mosaic relationships are most evident through comparisons of amino acid sequences, and represent distant evolutionary relationships generated by illegitimate recombination over an extended period of evolutionary time (Hatfull, 2010; Hendrix et al., 1999).

We describe here the ability of a large set of mycobacteriophages to infect M. tuberculosis and two strains of M. smegmatis distinct from their known common host (M. smegmatis mc²155). We show that relatively few of these phages infect M. tuberculosis, but that these generally fall within particular clusters/subclusters. In contrast, many of these phages infect other M. smegmatis strains, but there are groups of clusters or subclusters that either don’t infect or only form plaques at low frequencies. There is thus a correlation between cluster/subcluster type and host preference. We also describe two examples, both involving tail fiber mutations, by which mycobacteriophages overcome host barriers, which strongly implicate host recognition or DNA injection as the basis for constrained host range. Finally, we present a model to account for the diversity of mycobacteriophage genome sequences based on host diversity, and their ability to rapidly switch or expand host preferences.

Results

Mycobacteriophage diversity

The sequenced mycobacteriophage genomes can be grouped into clusters and subclusters based on sequence identity. However, while this classification is pragmatically useful, it is also inherently untidy, in that the grouping is essentially arbitrary, and the degrees of diversity within and between groups are highly varied. This is illustrated by representation of the phylogenetic relationships based on gene content (i.e. the distribution of homologous genes – phamilies – among different phages) (Fig. 1). For example, some singleton phages – such as Giles and Marvin – appear to be rooted such that they are either unrelated to any other phage, or they are no more closely related to any one group of phages than to the collection as a whole. Both phages actually contain a significant proportion of their genes (21.5% and 23.3%, respectively) that are shared with other mycobacteriophages, but the phages with homologues are broadly distributed across the phylogenetic landscape. Similarly, the Cluster C phages, which have a different virion morphology (i.e. myoviral) than all other mycobacteriophages (siphoviral), share about 10% of the genes with them. Thus these phages presumably have access to the larger pool of mycobacteriophage genes, but with different levels of access and participation in genetic exchange.

Figure 1.

Gene content relationships among mycobacteriophages. The relationships among 220 mycobacteriophages are displayed using the NeighborNet function in Splitstree4 (Huson, 1998). Data were generated from the database Mycobacteriophages_220 in the program Phamerator (Cresawn et al., 2011). Phages within a cluster or subcluster are circled and labeled accordingly. Singletons are shown in grey circles.

We have noted previously that there is little variation in GC% content between phages within clusters, but that the GC% content varies considerably from cluster to cluster (Hatfull, 2012b). There are also GC% content differences, albeit to a lesser degree, between subclusters within a given cluster such as for the subclusters of Cluster A (Table 1; Fig. 2). In this case, several of the subclusters have a narrow range of GC% content that does not overlap with that of other subclusters (e.g. A5, A6, and A1 (Fig. 2). The overall span of GC% content is impressive, ranging from 50.3% in Patience, to 69.1% in Cooper. This contrasts to the 67.4% GC% content of M. smegmatis, fairly typical for the actinomycetes. M. avium has the highest GC% among the mycobacteria (69.0%), and M. leprae has the lowest (57.8%); M. tuberculosis is 65.6% GC. There are several plausible factors that may play a role in the wide range of mycobacteriophage GC% contents. First, amelioration of phage DNA GC% content towards its host may be relatively slow compared to the frequencies with which phages can change or expand their host ranges. Second, M. smegmatis may represent a plausible host, but it may not be the host that was most recently used by the phages within the environment from which they were isolated. As such, the variety of ‘preferred’ hosts, or those most recently infected within their natural history, could be potentially large and varied. Furthermore, although M. smegmatis mc²155 is a common host, the range of host preferences could vary considerably among the phages. We note that the microbial population in soil and compost – from where many of the phages were isolated – is highly diverse (Maron et al., 2011; Schloss and Handelsman, 2008). Furthermore, in this scenario, we would predict that phages isolated from environments of limited bacterial diversity would be much less diverse, and this is what is observed with phages of Propionibacterium acnes isolated from sebaceous follicles of skin, even though about 25% of the genes are homologues to mycobacteriophage genes with weak but detectable amino acid sequence similarity (Marinelli et al., 2012).

Table 1.

Efficiencies of phage plating on mycobacterial strains

Phage	Cluster	GC%	Jucho1	MKD81	Mtb1
Bethlehem	A1	63.2	2.2×10−2	1.0×10−3	<10−7
BillKnuckles	A1	63.4	3.3×10−3	<10−8	<10−8
BpBieps31	A1	63.4	9.0×10−2	<10−8	<10−8
Bruns	A1	63.6	1.4×10−1	<10−4	<10−4
Bxb1	A1	63.6	8.0×10−1	1.0×10−4	1.0×10−2
DD5	A1	63.4	6.6×10−3	<10−6	<10−6
Doom	A1	63.8	4.0×10−2	<10−8	<10−8
Dreamboat	A1	63.9	7.6×10−2	<10−8	<10−8
Euphoria	A1	63.7	2.5	<10−8	<10−8
Jasper	A1	63.7	1.7×10−1	<10−7	<10−7
JC27	A1	63.6	1.5×10−1	<10−7	<10−7
KBG	A1	63.6	2.0×10−1	<10−6	<10−6
KSSJEB	A1	63.6	3.0×10−1	<10−9	<10−9
Kugel	A1	63.8	1.3×10−1	<10−10	1.0×10−6
Lesedi	A1	63.8	1.3	<10−9	<10−9
Lockley	A1	63.4	4.3×10−2	<10−6	<10−6
McGordo	A1	63.8	2.2	<10−8	<10−8
Museum	A1	63.6	3.6×10−1	<10−4	<10−4
Perseus	A1	63.7	6.7	<10−8	<10−8
RidgeCB	A1	64.0	3.3	<10−6	<10−6
SkiPole	A1	63.6	1.0×101	<10−7	<10−7
Solon	A1	63.8	1.6×10−1	<10−4	<10−4
U2	A1	63.7	4.5×10−1	4.5×10−6	4.5×10−3
Violet	A1	63.8	1.5×10−1	<10−7	<10−7
Che12	A2	62.9	1.7×10−1	<10−9	5.8×10−4
D29	A2	63.5	5.9×10−1	5.9×10−3	5.9×10−1
L5	A2	62.3	5.0×10−1	<10−7	7.5
Pukovnik	A2	63.3	1.0	7.0×10−5	<10−10
RedRock	A2	64.5	1.2	<10−9	<10−9
Trixie	A2	64.5	3.3	3.3×10−1	<10−6
Turbido	A2	63.3	7.7×10−2	7.7×10−1	7.7×10−1
Bxz2	A3	64.2	1.0	<10−6	1.5
HelDan	A3	64.0	7.7×10−2	<10−7	<10−7
JHC117	A3	64.0	3.3×10−6	<10−9	<10−9
Microwolf	A3	64.0	1.0	<10−4	2.2
Rockstar	A3	64.3	1.0	1.0	5.0
Vix	A3	64.0	1.0	<10−6	4.5×10−1
Backyardigan	A4	63.7	1.0×10−1	1.0×10−1	<10−7
Eagle	A4	63.9	1.0	3.3×10−1	<10−6
LHTSCC	A4	63.9	1.0	<10−8	<10−8
MeeZee	A4	63.9	2.0	1.0×10−6	<10−9
Peaches	A4	63.9	3.7×10−5	1.1×10−5	<10−9
Shaka	A4	63.9	3.3	3.3×10−1	<10−6
TiroTheta9	A4	63.9	1.0×10−1	<10−9	<10−9
Wile	A4	63.7	6.6×10−1	<10−8	<10−8
Airmid	A5	60.0	7.6×10−1	3.9×10−1	<10−9
Benedict	A5	59.8	1.0	2.3×10−1	<10−9
Cuco	A5	60.9	1.0	<10−8	<10−8
Blue7	A6	61.4	1.0	5.5×10−2	<10−8
DaVinci	A6	61.5	3.0×10−1	3.0×10−2	<10−9
Gladiator	A6	61.4	1.0	7.7×10−2	<10−8
Hammer	A6	61.3	1.0	1.4×10−2	<10−6
Jeffabunny	A6	61.6	<10−7	6.6×10−2	<10−7
TimShel	A7	63.1	<10−6	<10−6	<10−6
Saintus	A8	61.2	1.5×10−1	<10−5	<10−5
Alma	A9	62.5	1.0×10−1	<10−7	<10−7
PackMan	A9	62.6	1.0×10−2	<10−6	4.3×10−4
ABU	B1	66.5	<10−9	8.1×10−6	<10−8
Chah	B1	66.5	<10−10	3.0×10−6	<10−10
Colbert	B1	66.5	<10−7	<10−7	<10−7
Fang	B1	66.5	<10−9	6.5×10−7	<10−9
Harvey	B1	66.5	<10−7	<10−7	<10−7
Hertubise	B1	66.4	<10−9	4.3×10−7	<10−9
IsaacEli	B1	66.5	<10−9	<10−9	<10−9
JacAttac	B1	66.5	<10−7	<10−7	<10−7
Kikipoo	B1	66.5	<10−9	4.8×10−6	<10−9
Morgushi	B1	66.4	<10−8	<10−8	<10−8
Murdoc	B1	66.4	<10−8	<10−8	<10−8
Oline	B1	66.4	<10−10	3.3×10−6	1.3×10−7
Oosterbaan	B1	66.5	<10−7	<10−7	<10−7
Orion	B1	66.5	<10−8	3.3×10−6	<10−8
PG1	B1	66.5	<10−7	<10−7	<10−7
Phipps	B1	66.5	<10−4	1.0×10−1	<10−4
Puhltonio	B1	66.4	<10−8	<10−8	<10−8
Scoot17c	B1	66.5	1.8×10−1	7.6×10−3	5.9×10−3
Serendipity	B1	66.5	<10−8	<10−8	<10−8
TallGrassMM	B1	66.5	<10−9	<10−9	<10−9
Thora	B1	66.5	<10−9	5.4×10−7	<10−9
ThreeOh3D2	B1	66.5	<10−9	<10−9	<10−9
UncleHowie	B1	66.5	<10−8	<10−8	<10−8
Vista	B1	66.5	1.0×10−4	3.0×10−4	<10−7
Vortex	B1	66.5	<10−10	<10−10	<10−10
Yoshand	B1	66.5	6.7×10−2	4.3×10−7	2.9×10−6
Arbiter	B2	68.9	8.5×10−2	5.0×10−2	5.0×10−6
Ares	B2	69.0	<10−8	1.0×10−2	3.0×10−6
Hedgerow	B2	69.0	1.0	2.3×10−3	1.0×10−6
Qyrzyla	B2	68.9	6.7×10−21	<10−3	<10−3
Rosebush	B2	68.9	<10−10	1.0×10−6	<10−10
Athena	B3	67.5	2.0×10−1	<10−5	<10−5
Akoma	B3	67.5	2.3×10−2	2.3×10−2	<10−9
Daisy	B3	67.6	1.8×10−1	1.5×10−5	<10−9
Gadjet	B3	67.5	6.6×10−3	<10−5	<10−5
Kamiyu	B3	67.5	5.9×10−2	<10−10	1.8×10−6
Phaedrus	B3	67.6	3.0	5×10−6	<10−9
Phlyer	B3	67.5	7.5×10−1	<10−8	<10−8
Pipefish	B3	67.3	1.1×10−1	5.0×10−6	<10−9
ChrisnMich	B4	69.1	<10−6	<10−6	<10−6
Cooper	B4	69.1	5.0×10−1	<10−6	<10−6
Nigel	B4	68.3	<10−9	1.0×10−4	<10−9
Stinger	B4	68.6	3.0×10−1	1.0×10−2	<10−5
Zemanar	B4	68.9	<10−9	<10−9	<10−9
Acadian	B5	68.4	1.0×10−3	<10−9	<10−9
Bxz1	C1	64.8	2.6×10−2	1.0	<10−8
Cali	C1	64.7	1.0×10−2	1.0	<10−8
Catera	C1	64.7	1.0×103	1.0×102	<10−6
Dandelion	C1	64.7	3.0×10−3	1.0×10−1	<10−9
Drazdys	C1	64.7	1.0×10−2	3.3	<10−6
ET08	C1	64.6	3.3×10−3	3.3	<10−6
LinStu	C1	64.8	4.4×10−6	4.4×10−6	<10−9
LRRHood	C1	64.5	1.1×10−5	4.8×10−2	<10−9
MoMoMixon	C1	64.8	<10−6	1.0×10−3	<10−6
Nappy	C1	64.7	1.3×10−5	1.1×10−1	<10−9
Pio	C1	64.8	3.0	3.0×10−5	<10−9
Pleione	C1	64.7	<10−5	5.0×10−2	<10−5
Rizal	C1	64.7	<10−8	2.3×10−1	<10−8
ScottMcG	C1	64.8	<10−7	3.0×10−1	<10−7
Sebata	C1	64.8	6.0×10−2	6.0×10−2	<10−5
Spud	C1	64.8	<10−6	1.0×10−2	<10−6
Wally	C1	64.7	1.0	4.5	<10−10
Adjutor	D	59.9	<10−8	<10−8	<10−8
Gumball	D	59.6	1.0×10−2	<10−8	<10−8
Nova	D	59.7	5.4×10−1	<10−10	<10−10
PBI1	D	59.8	1.1×10−5	<10−8	<10−8
P-Lot	D	59.8	3.4×10−4	<10−8	<10−8
SirHarley	D	59.6	7.5×10−2	7.5×10−6	<10−9
Troll4	D	59.6	<10−8	<10−8	<10−8
Bask21	E	62.9	3.3	<10−5	<10−5
CJW1	E	63.7	5.6×10−1	<10−5	<10−5
Elph10	E	63.0	3.3	<10−6	<10−6
Eureka	E	62.9	5.0×10−2	<10−7	<10−7
Henry	E	63.0	1.7	<10−5	<10−5
Kostya	E	63.5	5.0×10−1	<10−6	<10−6
Lilac	E	63.0	2.5×10−1	3.5×10−2	<10−8
Porky	E	63.5	6.6×10−1	<10−8	<10−8
Pumpkin	E	63.0	1.3×10−2	1.0×10−6	<10−10
Rakim	E	62.9	1.0	<10−6	<10−6
Toto	E	63.0	2.0	<10−7	<10−7
Boomer	F1	61.1	9.5×10−1	4.2×10−7	<10−9
Che8	F1	61.3	9.0×10−4	2.7×10−4	<10−10
DeadP	F1	61.6	4.2×10−4	<10−7	<10−7
DLane	F1	61.9	2.3×10−3	<10−7	<10−7
DotProduct	F1	61.8	1.3×101	<10−8	<10−8
Drago	F1	61.2	8.0×10−1	<10−10	<10−10
Fruitloop	F1	61.8	3.3×10−2	<10−9	<10−9
GUmbie	F1	61.4	4.5×10−1	<10−6	<10−6
Ibhubesi	F1	61.2	4.4×10−1	<10−8	<10−8
Llij	F1	61.5	1.3	<10−9	<10−9
Mozy	F1	61.1	3.0×10−3	<10−9	<10−9
Mutaform13	F1	61.3	2.0×10−1	<10−9	<10−9
Pacc40	F1	61.3	4.0×10−1	<10−10	<10−10
PMC	F1	61.4	7.6×10−1	1.0×10−6	<10−9
Ramsey	F1	61.2	6.5×10−1	<10−8	<10−8
RockyHorror	F1	61.1	6.6	<10−8	<10−8
SG4	F1	61.9	<10−7	<10−7	<10−7
Shilan	F1	61.4	6.6×10−1	5.6×10−7	<10−9
Spartacus	F1	61.7	8.7×10−1	1.3×10−6	<10−10
Tweety	F1	61.7	4.3×10−1	<10−9	<10−9
Wee	F1	61.8	2.3×10−4	5.1×10−7	<10−9
Che9d	F2	60.9	3.7×10−4	<10−9	<10−9
Yoshi	F2	61.0	1.2×10−1	<10−10	<10−10
Angel	G	66.7	5.4	1.1×10−2	9.0×10−1
Avrafan	G	66.6	5.0×10−1	1.0×10−3	1.0×10−5
BPs	G	66.6	7.6×10−1	1.0×10−3	1.5×10−5
Halo	G	66.7	1.7	6.7×10−3	6.0×10−4
Hope	G	66.6	7.5×10−1	7.5×10−2	<10−7
Liefie	G	66.8	1.7×10−1	5.8×10−3	2.3×10−2
Konstantine	H1	57.4	7.5×10−5	7.5×10−6	<10−9
Predator	H1	56.4	<10−6	<10−6	<10−6
Barnyard	H2	57.5	<10−4	<10−4	<10−4
Babsiella	I1	67.1	3.0×10−3	<10−9	<10−9
Island3	I1	66.8	<10−8	4.4×10−5	<10−8
Che9c	I2	65.4	1.5×10−3	1.5×10−3	<10−9
Baka	J	60.7	2.9×10−3	<10−8	<10−8
Courthouse	J	60.9	1.1×10−4	<10−10	<10−10
LittleE	J	61.3	<10−7	<10−7	<10−7
Omega	J	61.4	<10−5	<10−5	<10−5
Optimus	J	60.8	3.3×10−2	<10−6	<10−6
Thibault	J	60.8	7.0×10−2	<10−9	<10−9
Adephagia	K1	66.6	1.0×10−2	<10−8	3.3
CrimD	K1	66.9	3.0	<10−6	1.0
Angelica	K1	66.4	<10−9	<10−9	<10−9
Jaws	K1	66.6	4.3×10−1	1.2×10−2	1.3
TM4	K2	68.1	1.5	2.3×10−5	7.7×10−1
Pixie	K3	67.3	3.3×10−1	2.0×10−2	4.6×10−1
Fionnbharth	K4	68.0	1.3×10−1	<10−8	3.0×101
Larva	K5	65.3	1.0	<10−8	<10−8
JoeDirt	L1	58.8	3.3×10−5	2.3×10−1	<10−8
LeBron	L1	58.8	<10−9	1.0×10−1	<10−9
UPIE	L1	58.8	3.0×10−5	4.0×10−2	<10−9
Faith1	L2	58.9	1.0	1×10−9	3.3×10−6
Rumpelstiltskin	L2	58.9	<10−9	<10−9	3.3×10−6
Bongo	M	61.6	5.2×10−1	5.4×10−3	7.7×10−3
Rey	M	60.9	2.2	<10−8	3.3×10−5
Charlie	N	66.3	1.3×10−1	1.0×10−4	<10−8
Redi	N	66.1	1.3×10−1	<10−9	<10−9
Corndog	O	65.4	<10−8	4.4×10−5	<10−8
Firecracker	O	65.5	<10−8	4.3×10−1	<10−8
BigNuz	Sin	66.7	2.9×10−1	1.4×10−5	<10−8
Dori	Sin	66.0	2.3	<10−9	4.0×10−2
Giles	Sin	67.3	2.9×10−3	2.5×10−5	<10−8
Marvin	Sin	64.7	<10−10	2.5×10−4	<10−10
Patience	Sin	50.3	<10−6	4.4×10−3	<10−6
Send513	Sin	56.0	<10−8	2.3×10−1	3.3×10−6
Wildcat	Sin	57.2	<10−10	1.7×10−3	1.0×10−4

¹Efficiencies of plating are relative to infection of M. smegmatis mc²155

Figure 2.

Gene content relationships among the subclusters of Cluster A mycobacteriophages. An expanded representation of the Cluster A phages from Figure 1 is shown with each of the subclusters circled and the GC% range for each subcluster indicated.

Mycobacteriophage infection of M. tuberculosis

Although a subset of the sequenced mycobacteriophages previously have been tested for infection of M. tuberculosis, we have extended this to the larger collection of phages (Table 1; Fig. S1). Interestingly, there is a strong correlation between cluster or subcluster designation and the ability to efficiently infect M. tuberculosis mc²7000 (using a cutoff value for infection at an e.o.p. of 0.1) in a chi squared test (x² = 97.85, df =26, P = 2.9×10⁻¹⁰). The infectivity across the clusters of phages is restricted primarily to phages in Cluster K, and two subclusters (A2 and A3) of Cluster A, although some plaques were also observed with Bxb1 and U2 from Subcluster A1 (Table 1, Fig. S1). The phages in other subclusters of cluster A do not form plaques on M. tuberculosis, even those most closely related phylogenetically, such as in subclusters A6/A9 and A4/A7, respectively (Fig. 2). However, Cluster A phages span considerable diversity (Fig. 2), and genome alignments do not provide obvious clues as to the genetic basis of this discrimination for M. tuberculosis. As reported previously for BPs and Halo (Sampson et al., 2009), all of the Cluster G phages are able to form plaques on M. tuberculosis but at a reduced efficiency of plating (Table 1, Fig. S1). We also observed plaques when plating large numbers of particles on M. tuberculosis for several other phages including some Cluster B phages.

Mycobacteriophage infection of M. smegmatis strains

M. smegmatis and M. tuberculosis are distantly related mycobacterial species with vastly different growth rates (doubling times of 3 hours and 24 hours, respectively) and genome sizes (7.0 Mbp and 4.4 Mbp, respectively). It is therefore perhaps not surprising that relatively few mycobacteriophages efficiently infect both strains. To further explore mycobacteriophage host preferences on more closely related bacteria, we tested these phages for their ability to infect two other strains of M. smegmatis, Jucho, and MKD8. M. smegmatis Jucho (Mizuguchi and Tokunaga, 1971) and MKD8 (Parsons et al., 1998) – a derivative of M. smegmatis PM5 (Tokunaga et al., 1973) – have been described previously, and have distinct patterns of conjugation with M. smegmatis mc²155, a derivative of M. smegmatis ATCC607 (Parsons et al., 1998; Snapper et al., 1990; Wang et al., 2003). We reasoned that since these strains have independent origins, some of the phages would discriminate between them, and exhibit patterns of infection distinct from those observed on M. tuberculosis. We also tested these phages on Mycobacterium aichiense – a fast grower and distant relative of M. smegmatis – but none of the tested phages are able to infect this bacterium (data not shown).

The plating properties of over 200 phages on the various M. smegmatis strains are shown in Table 1 and Fig. S1, and there are three main patterns that emerge. First, it is clear that the phages exhibit distinct patterns of infection on various M. smegmatis strains. Second, there are correlations between the abilities of phages to infect individual strains and their cluster/subcluster designation, although not as strongly as for M. tuberculosis (Jucho: x² = 40.00, df = 26, p = .039; MKD8: x² = 67.03, df = 26, p = 1.8×10⁻⁵; cutoff e.o.p. of 0.1) (Table 1, Fig. S1). A third observation is that although there are many examples where no plaques are observed at all on a particular strain, there are numerous examples where the e.o.p. is merely reduced as compared to mc²155, presumably either through escape of a restriction mechanism, or through phage mutations. Overall, these data are consistent with the idea that there is a correlation between cluster/subcluster designation and host range.

There are many potential mechanisms for resistance and host range determination, but their contributions to mycobacteriophage host preferences are poorly understood. One potential mechanism is superinfection immunity conferred by the presence of a prophage to which escape is typically very infrequent, and in the phage lambda prototype, multiple mutations (at least three) are required to confer a virulent phenotype (Hopkins and Ptashne, 1971). This could account for example for the inability of subcluster A1 phages to infect MKD8, although prophages in this strain have yet to be reported. However, for M. tuberculosis mc²7000 – an avirulent derivative of strain H37Rv – the genome sequence is known and there are no full-length prophages [there are two small prophage-like elements, but these do not encode any recognizable repressor proteins (Cole et al., 1998; Hendrix et al., 1999)]. Thus the inability for phages to infect M. tuberculosis is unlikely to result from phage immunity.

Expanding the host range of Cluster G phages

The Cluster G phages do not efficiently infect M. tuberculosis but can form plaques on this bacterium at a reduced frequency_, relative to M. smegmatis mc²155 (Sampson et al., 2009)(Fig. 3A). This pattern of infection might arise due to the escape from restriction or from the generation of mutations with expanded or altered host range. To explore these possibilities, 12 independently isolated plaques of either BPs or Halo were recovered from M. tuberculosis lawns; all were re-plated and found to infect M. smegmatis and M. tuberculosis at equivalent e.o.p’s (Fig. 3B). These were then plaque purified again on M. smegmatis, and the resulting phages were also shown to infect M. tuberculosis with an e.o.p. of one, relative to M. smegmatis (Fig. 3B). These properties strongly suggest that the enhanced efficiency of plating of the Cluster G phage isolates on M. tuberculosis results from mutational and heritable changes, rather than phenotypic survival, as would be expected if these phages were escaping from host restriction/modification systems. In this latter case, we predict that plating on M. smegmatis would have resulted in loss of the enhanced e.o.p. phenotype.

Figure 3.

Host Range Expansion of Cluster G mycobacteriophages. (A) Mycobacteriophages BPs and Halo form plaques on M. tuberculosis mc²7000 at an efficiency of 10⁻⁴ – 10⁻⁵, relative to M. smegmatis mc²155. A control phage, D29, infects both bacteria with equal efficiency. (B) A single Halo plaque picked from an M. tuberculosis and re-plated on M. tuberculosis and M. smegmatis infects both bacteria with equal efficiency (top part). A single plaque picked from the M. smegmatis plate in the top part of the figure retains the ability to efficiently infect M. tuberculosis (bottom part), showing it has acquired a heritable mutation conferring host range expansion. (C) The left arms of the BPs and Halo genomes contain the virion structure and assembly genes (1 – 26). The positions of a conserved domain and a weakly predicted coiled domain are shown. (D) Two of the mutations conferring host range expansion – T303A and A599D – were engineered back into mycobacteriophage Halo using BRED mutagenesis. Both mutant phage derivatives efficiently infect M. tuberculosis. (E) Halo and BPs host range mutants [A604E (blue squares); R331Q (purples crosses); T303A (brown squares); A306V (yellow triangles); A321E (green crosses); A599D (red circles) display increased adsorption rates on M. smegmatis relative to wild-type phage (light blue diamonds); adsorption to M. tuberculosis is unaffected.

To map the mutations responsible for the ability to efficiently infect M. tuberculosis, we first determined the complete genome sequence of one of the phage isolates, Halo enhanced host range 1 (ehr1). A single point mutation was identified at position 22,253 within gene 22, which changes an alanine residue at position 604 to a glutamic acid (A604E) (Fig. 3C). Halo gene 22 is located within a group of putative minor tail protein genes (Fig. 3C) and encodes a product of 797 residues; BPs encodes a nearly identical product that differs in only a single residue at position 12. The structure of BPs/Halo gp22 is not known, but it is rich in glycine (12.5%) and alanine (11.8%), characteristic of phage tail fiber proteins. There is a weakly predicted coiled domain between residues 270 and 330, but no significant matches were found using HHPred (Fig. 3C).

To map the locations of the mutations in the other host range mutants, segments containing gene 22 were amplified by PCR and sequenced. A single base substitution was found in each, corresponding to a total of seven different amino acid changes (Fig. 3C). Seven of the 12 mutants map to just two residues, at positions 306 and 604, and in general, the mutations cluster to two regions, one spanning residues 279 – 331 near the weakly predicted coiled domain, and the other around position 600 (Fig. 3C). The roles of these substitutions in gene 22 in host range was confirmed by using BRED mutagenesis (Marinelli et al., 2008) to separately engineer two of the substitutions (T303A and A599D) into a wild-type Halo genome. These engineered mutants recapitulated the phenotypes of the mutants isolated directly on M. tuberculosis (Fig. 3D). The mapping of host range mutations to tail fibers is not unexpected, and suggests strongly that the inability of Halo and BPs to efficiently infect M. tuberculosis results from surface interactions rather than defects in expression, replication, assembly, or lysis. In support of this interpretation, when DNA from either wild-type Halo or from a host range mutant is electroporated into M. tuberculosis and the transformed cells are plated on a lawn of permissive M. smegmatis, approximately equivalent numbers of plaques (~130 per 250 ng DNA) are observed.

To determine if these expanded host range mutants have altered surface interactions, we measured adsorption rates for M. smegmatis and M. tuberculosis (Fig. 3E). Although M. tuberculosis adsorption assays are technically challenging because of cell clumping, we observed no differences in the binding of the two mutants tested as compared to wild-type phage. Surprisingly, all of the mutants we tested have increased adsorption rates to M. smegmatis with A604E showing the strongest phenotype (Fig. 3E). This is a surprise because the mutants were isolated solely for their ability to infect M. tuberculosis, and although we might have expected to see phenotypic changes that compromise the ability of the mutants to infect M. smegmatis, dramatic improvements in adsorption were not expected. One plausible explanation for this phenomenon is that these mutants are not altered in receptor recognition at the initial stage of infection, but differ in a subsequent event that renders binding irreversible, such as DNA injection across the cell membrane, and which presumably requires dynamic events within the receptor-tail fiber complex. It should be noted, however, that these experiments were conducted with planktonically growing cells under laboratory conditions, and it is possible that these mutants could be impaired in infection of M. smegmatis growing environmentally [for example as biofilms (Ojha et al., 2005)], as noted for high adsorption rate mutants of phage λ (Gallet et al., 2009). This might explain why phages such as Halo have evolved to have rather poor adsorption rates.

Expanded host range of Subcluster B2 phages

To further explore the basis of host range determination in the mycobacteriophages, we examined the ability of the subcluster B2 phage Rosebush against the host strain M. smegmatis Jucho. As noted above, Rosebush does not efficiently infect Jucho, although plaques with a normal appearance are observed at a frequency of ~10⁻⁵ (Fig. 4A). We isolated and purified five independent mutants, all of which were shown to infect mc²155 and Jucho at equivalent efficiencies of plating (Fig. 4B). We determined the sequence of one of these mutants and identified a single base substitution at position 30,977 within Rosebush gene 32 that introduces a L297R substitution in gp32 (Fig. 4C). Gene 32 is located within a group of putative minor tail proteins, and – like BPs/Halo gp22 – Rosebush gp32 has a relatively high glycine (9.0%) and alanine (11.4%) content, consistent with its role as a phage tail fiber. Furthermore, like BPs gp22, Rosebush gp32 has a weakly predicted coiled domain (Fig. 4C). Comparative analysis suggests a modular structure to the protein. The N-terminal 400 residues are similar to a large group of putative tail proteins of phages from other clusters such as H, I, M, N, and other subclusters of B, and the C-terminal 500 residues have similarity to putative tail proteins of Cluster L phages, with much weaker similarity in the N-terminal domain. Interestingly, HHPred reveals a highly probable match (98.93%) of the segment containing residues 125-284 to the 147 aa carbohydrate-binding module of Thermoanaerobacterium polysaccharolyticum ManA (Bae et al., 2008), suggesting that Rosebush binds to a receptor with a carbohydrate moiety. However, the L297R substitution lies outside of this region.

Figure 4.

Isolation and characterization of Rosebush expanded host range mutants. (A) Plating of serial dilutions Rosebush on M. smegmatis mc²155 and M. smegmatis Jucho shows the reduced efficiency of plating; the control phage Qyrzula (which, like Rosebush, is in Subcluster B2). (B) A Rosebush plaque picked from the Jucho plate (mut1) plates on mc²155 and Jucho with equivalent efficiencies of plating (upper). A mut1 plaque recovered from M. smegmatis mc²155 plates with equivalent efficiencies on both strains, in contrast to wild-type Rosebush (wt; lower). (C) Locations of amino acid substitutions in gp32 and gp42 of Rosebush in the expanded host range mutants. Positions of the ManA domain and a weakly predicted coiled domain are shown, as well as a region at the C-terminus of gp32 that is homologous with gp42. (D) Adsorption assays (n = 2) of wild-type Rosebush (red squares) and expanded host range mutant 1 (blue diamonds) on M. smegmatis mc²155 and M. smegmatis Jucho as indicated.

The remaining four mutants were mapped by PCR amplification and sequencing of gene 32. Three of the mutants each have a single base substitution in 32, conferring the substitutions W121G, G288D and W293R. The identification of three substitutions within a span of fewer than 10 residues suggests that this is a particularly important region for host range determination; all of the substitutions lie outside of the ManA carbohydrate binding domain. The fourth mutant has a wild-type gene 32, and the complete genome sequence was determined to identify the mutation, a single base change at position 39,077 conferring a G140D substitution in gp42 (Fig. 4C). This suggests that gp42 is also important for host range determination. Gene 42 lies at the extreme end of the structural gene operon (the genes to its immediate right are transcribed leftwards) and encodes a 231 aa product. BLASTP comparison reveals that gp42 has sequence similarity albeit weakly (24% identity) to the C-terminal portion of Rosebush gp32 (Fig. 4C). HHPred analysis shows that Rosebush gp42 (residues 35 – 209) has highly probable matches (85.8 – 96.6%) to nine glycosyl hydrolases for which structures have been determined, with the closest being to a mouse Galactocerebrosidase (Deane et al., 2011). Curiously, HHPred also shows similarity (residues 35-188) to the baseplate protein gp10 of phage T4 (Leiman et al., 2006), although with a much weaker probability (37.1%). These data are consistent with the interpretation that gp42 is a tail component that assists in infection through hydrolytic degradation of sugar-containing molecules on the cell surface.

Collectively, these observations suggest that the inability of wild-type Rosebush to infect M. smegmatis Jucho is because of poor receptor recognition, or – as proposed for BPs – downstream events such as DNA injection. To confirm that there was no defect in replication, expression, assembly, or lysis, we performed a transfection experiment in which Rosebush genomic DNA was electroporated into competent Jucho cells, and following a period of recovery, plated with permissive M. smegmatis mc²155 cells (Fig. 5A). An equivalent number of plaques were recovered as when using M. smegmatis mc²155 competent cells, or when using Giles genomic DNA (as a control for cell competency) with both strains.

Figure 5.

Transfection and replication in non-permissive hosts.

(A) Transfection of electrocompetent M. smegmatis Jucho (top row) and mc²155 (bottom row) with 150 ng of either Rosebush genomic DNA (left column) or Giles genomic DNA (right column) followed by recovery and plating with M. smegmatis mc²155 cells. (B) Transfection of electrocompetent M. tuberculosis mc²7000 with 150 ng of either Giles genomic DNA (left plate) or D29 genomic DNA (right plate) followed by recovery and plating with M. smegmatis mc²155 cells. (C) Transfection of electrocompetent M. smegmatis mc²155 with either 200 ng (left plate) or 400 ng (right plate) of Streptomyces phage Zemlya followed by recovery and plating with S. lividans. Individual plaques were picked and tested for infection of S. lividans and M. smegmatis; all infected S. lividans but failed to infect M. smegmatis, showing that these are not mutants or contaminants.

Adsorption assays performed with the gp32 L297R mutant show no apparent difference in binding to Jucho cells as compared to wild-type Rosebush, reminiscent of the behavior of the BPs/Halo mutants in adsorption to M. tuberculosis (Fig. 4D). In addition, we observed that – like the BPs/Halo mutants – the Rosebush mutant adsorbs better to M. smegmatis mc²155 than wild-type Rosebush (Fig. 4D). As described above for the BPs/Halo host range expansion, the basis for this unexpected phenotype is not clear.

Cross-species and cross-genus transfection

The transfection assay described above for Rosebush is a powerful approach for determining whether a host range constraint derives from a surface or DNA injection blockage, or from subsequent events post-DNA entry into the cell. The observation that wild-type Rosebush can propagate in Jucho following DNA transfection is perhaps not surprising given that the bacterial hosts are strains within the same species. However, we have explored the possibility that this approach could be applied more broadly including both hosts of different species, and host from different genera. First, we tested the ability of M. tuberculosis to support the propagation of Giles DNA (Giles is normally unable to infect M. tuberculosis), by transfecting Giles genomic DNA into electrocompetent M. tuberculosis mc²7000 cells, allowing recovery, and plating with M. smegmatis cells (Fig. 5B). Equivalent numbers of plaques were recovered as when using electrocompetent M. smegmatis cells, indicating that surface recognition and DNA injection represent the barrier to Giles infection of M. tuberculosis. As described above, we performed similar transfection assays with wild-type Halo and one of the host range mutants and found that both produce similar numbers of plaques when equivalent amounts are electroporated into M. tuberculosis mc²7000 and plated on a lawn of M. smegmatis (data not shown).

To our knowledge, phages infecting either mycobacterial hosts or Streptomyces hosts do not cross the genus barrier between Mycobacterium and Streptomyces. In order to determine if surface interactions are responsible (at least in part) for this barrier, we performed a similar experiment to that described above, using DNA of Streptomyces phage Zemlya, which was electroporated into M. smegmatis, and plated with S. lividans cells (Fig. 5C). Although recovery was not efficient, using 200 ng and 400 ng of DNA, we recovered approximately 18 and 30 plaques, respectively, indicating that M. smegmatis can support the post-DNA injection events of Zemlya’s growth. In order to rule out host range mutants or contamination, 20 random plaques were picked and tested for their ability to infect both M. smegmatis and S. lividans. All 20 were able to infect S. lividans and none were able to infect M. smegmatis, indicating that these phages were not contaminants and were not host range mutants.

Discussion

Analysis of the host preferences for a large collection of mycobacteriophages indicates correlations between host range and genome type, with varying strengths of correlation depending on the strain. Although only a few hosts were tested, we strongly predict that these patterns will extend to other potential mycobacterial hosts. However, in general, these data provide good support for the idea that mycobacteriophages within different cluster or subcluster designations have distinct host preferences, and that these host preferences confer barriers to genetic exchange, even though under laboratory conditions they can all infect the same host. However, because the designation of phages into clusters and subclusters is predicated upon essentially arbitrary parameters, and genetic mosaicism from horizontal genetic exchanges contributes to a considerable variability within these groups, the correlation between cluster/subcluster type is a relatively crude one. Presumably the molecular basis of host preference depends on the presence and behavior of specific genes, raising the possibility of searching for a correlation between gene phamilies and host preferences. However, this is further complicated by the finding that host preference is also allele specific, as demonstrated here for phage BPs and Rosebush.

Although there are no similarly large collections of phages with sequenced genomes that are known to infect a common host that have been examined for host preferences, the patterns observed for the mycobacteriophages are not atypical. For example, phages rarely infect hosts from other genera, and have preferences for particular species and sometimes strains or serovars within a species, such as the Listeria phages (Loessner and Busse, 1990; Loessner and Calendar, 2006). However, there are abundant examples of phages that can expand their host range within a species and to a different species as observed here, either by adapting to a new receptor or overcoming restriction, immunity, CRISPRs or Toxin-antitoxin systems (Deveau et al., 2010; Fineran et al., 2009; Hopkins and Ptashne, 1971). There are also examples of phages that can ‘jump’ to new genera (Bielke et al., 2007; Jensen et al., 1998). To our knowledge, no phages have been described that traverse the barrier between gram-negative and gram-positive bacteria, or from which mutants can be isolated that permit jumping of the gram-barrier.

There are a multitude of mechanisms that can determine host range, including receptor availability, restriction/modification, CRISPRs, toxin-antitoxin systems and immunity; infection is also influenced by conditions such as the growth state of the cells and metal ion availability (Fullner and Hatfull, 1997). The number and variety of such systems is no surprise given the highly dynamic nature of host-phage interactions that have dominated the microbial world for perhaps more than three billion years. Interestingly, we note that not only is infection of M. tuberculosis restricted to specific genomic types – Cluster K, and subclusters A2, A3 – but of the others, only the Cluster G phages appear to be able to generate mutants that can overcome the constraint at a moderate frequency (10⁻⁴ – 10⁻⁵). This is different than the behaviors on other strains of M. smegmatis, where there are relatively few examples of clusters or subclusters failing to produce plaques at all, and it is more common to see reduced efficiencies of plating. Nonetheless, the mechanisms of mutant escape that we’ve described here both involve single residue substitutions in tail proteins.

The experiments reported here provide further insights into the nature of mycobacteriophage tails and their host interactions. The gp42 protein of Rosebush is a strong candidate for providing an enzymatic activity that allows for the phage tail tip to ‘bore’ its way through the complex outer surface of the mycobacterial cell for it to reach the membrane and complete the infection process, similar to the endorhamnosidase activity of the P22 tail spike (Goldenberg and King, 1982) or the lysozyme function of T4 gp5 (Arisaka et al., 2003). It is unclear how the mutation in Rosebush 42 generates an expanded host range phenotype, but it is plausible that the enzyme acquires a change in substrate specificity that permits processing of cell wall components in M. smegmatis Jucho that otherwise prevent infection. Such enzymes are of some interest as possible agents for anti-mycobacterial therapies.

M. tuberculosis contains two or three CRISPR loci and a total of 30 – 40 spacers, depending on the strain. It is unclear whether these are functional or not, but it is notable that there are no identifiable protospacer sequences corresponding to any of the 83 spacers in the entire phage collection (He et al., 2012). The CRISPRs are thus unlikely to contribute to the failure of any of the mycobacteriophages to infect M. tuberculosis. M. smegmatis mc²155 does not contain any CRISPRs; however, it is not known if any are found in either Jucho or MKD8 and therefore, whether they could contribute to the observed resistance patterns in these strains. We also note that M. tuberculosis encodes a large number of toxin-antitoxin systems, and although these are also not known to play any role in phage resistance, this has been demonstrated for a type III TA in Erwinia carotovora (Fineran et al., 2009) and is worthy of further exploration. Finally, although there is no evidence for restriction-modification systems in either M. smegmatis mc²155 or M. tuberculosis these could play a role in Jucho or MKD8 infection, and restriction systems have been reported in other strains such as M. chelonei and M. gordonae (Jones and Greenberg, 1977; Shankar and Tyagi, 1993a, b).

Based on these data, we present a model to account for the diversity of the mycobacteriophage population in which the diversity of the bacterial population in the environments where these microbes exist plays a central role (Fig. 6). In essence, the model recognizes that bacteriophages have the capacity to switch or expand their host ranges at frequencies that are vastly greater than the overall rates of genetic exchange. The changes in host preference are more likely to occur between hosts that are more similar – such as different strains of the same species (e.g. M. smegmatis mc²155 and M. smegmatis Jucho) – but can also occur with different species within the same genus (e.g. between M. smegmatis mc²155 and M. tuberculosis). Presumably, this switch can also happen between hosts of different genera, but this is likely to be a much less frequent event. For example, we have tested more than 60 mycobacteriophages for infection of S. lividans, and none formed plaques (data not shown), and no Streptomyces phages are known to infect M. smegmatis. Nonetheless, Streptomyces phage Zemlya can clearly replicate within and lyse M. smegmatis. Because the soil microbial diversity is large – especially given the substantial proportion of non-cultivatable bacteria (Rondon et al., 2000; Rondon et al., 1999) – and the genus barrier is not an insurmountable one, it is possible that a series of host range switches and expansions between closely related hosts collectively enable a random-walk across large swaths of the bacterial landscape.

Figure 6.

A model accounting for mycobacteriophage diversity. The large number of different types of mycobacteriophages isolated on M. smegmatis mc²155 can be explained by a model in which a) phages can readily infect a new bacterial host – either by a switch or an expansion of host range – and b) a highly diverse bacterial population, including many closely related strains, in the environments from which the phages are isolated. As such, phages with distinctly different genome sequences and GC% contents infecting distantly related bacterial hosts – such as those to the left (red) or right (blue) extremes of a spectrum of hosts – can migrate across a microbial landscape through multiple steps. Each host switch occurs at a relatively high frequency (~1 in 10⁵ particles, or an average of about one every 10³ bursts of lytic growth), and much faster than either amelioration of phage GC% to its new host, or genetic recombination. Two phages (such as those shown in red and blue) can thus ‘arrive’ at a common host (M. smegmatis mc²155) but be of distinctly different types (clusters, subclusters, and singletons). The variety of hosts is shown two dimensionally for simplicity, and the actual relationships among bacteria in environments such as soil and compost is likely to be considerably more complicated. Because host range switching or expansion is a common feature of bacteriophages, the model predicts that a high degree of phage diversity will be seen for any particular host if the microbial population from which the phages are isolated from is also highly diverse and rich in closely related strains. Because none of the phages isolated on M. smegmatis mc²155 also infect M. aichiense, we assume that this strain and its close relatives are absent from the soil and compost environments where most of these phages were isolated from [M. aichiense was isolated from soil in Japan (Ichiyama et al., 1988)].

Such diverse environments are known to exist in compost and soil, but other environments – such as in the human sebaceous follicles from which P. acnes phages can be isolated – are much less diverse, and this is reflected in the restricted diversity of P. acnes phages (Marinelli et al., 2012). Thus in a highly diverse bacterial context, phages can ‘arrive’ at a common host – such as M. smegmatis mc²155 – having ‘traveled’ from numerous phylogenetically distinct (from the host perspective) origins, with distinctive GC% contents, and particular genetic neighborhoods. The effects of this are expected to broadly impact the phage population, such that collections of phages isolated on any bacterial host from this environment will reflect the diversity of the underlying bacterial population. Furthermore, although it is often assumed that phage GC% will reflect that of its host (even though the ‘natural’ host may not be known), we predict that no such correlation will be seen for phages isolated from diverse microbial environments. We note that a broad range of GC% contents is also seen with both Pseudomonas and Staphylococcus phages (Ceyssens and Lavigne, 2010; Kwan et al., 2005, 2006; Marinelli et al., 2012). This model also underscores the potential for phages to readily act as vehicles of genetic exchange between different bacteria in a complex environment like the soil, as these phages readily move in and out of different hosts. Future studies aimed at characterizing phages of various hosts in both restricted and diversity rich environments will be important to determine the extent to which this model applies to a broad range of environmental niches, and ultimately, to better understand the complex ways in which bacteria and their phages interact in the context of the natural world.

Materials and Methods

Bacterial Strains and Media

M. smegmatis Jucho, M. smegmatis, MKD8, M. smegmatis mc²155 – a high efficiency transformation strain (Snapper et al., 1990) were grown in Middlebrook 7H10 supplemented with 0.5% glycerol, 1 mM CaCl₂ 10% and 10% Albumin Dextrose Complex (ADC), and the avirulent strain M. tuberculosis mc²7000 (Ojha et al., 2005; Sambandamurthy et al., 2006), was grow in Middlebrook 7H11 agar (Difco) supplemented with 0.5% glycerol, 1 mM CaCl₂ 10% and 10% Oleic Acid Albumin Dextrose Complex (OADC). Liquid cultures were grown in 7H9 liquid media (Difco), supplemented with 0.2% glycerol, 0.05% Tween 80, and 10% ADC for M. smegmatis and 10% OADC for M. tuberculosis; Tween 80 was omitted, and CaCl₂ was added at a final concentration of 1 mM to cultures used for phage infections. M. smegmatis was grown shaking at 250 rpm, and M. tuberculosis cultures were grown either on a roller or stationary; both were incubated at 37°C. Unless otherwise specified, carbenicillin (CB) (50 μg/ml) and cyclohexamide (CHX) (10 μg/ml) were routinely added to mycobacterial media. S. lividans (ATCC 69441) was cultured at 30°C in YEME (yeast extract, malt extract), on MS agar and on Nutrient agar containing 8mM Ca(NO₃)₂, 10mM MgSO₄, and 0.5% glucose (Kieser et al., 2000).

Propagation of Bacteriophages

All of the mycobacteriophages used here have been described previously (Hatfull, 2012a; Pope et al., 2011b). Streptomyces phage Zemlya is newly isolated, has a siphoviral morphology with a 51 kbp genome and will be described elsewhere. High-titer bacteriophage stocks and plate lysates were prepared as described previously (Sarkis and Hatfull, 1998). In some cases, phage were concentrated by ultracentrifugation in Ti45 rotor tubes at 20,000 μg for 1.5 hours and further purified by CsCl density gradient if necessary. Top agar overlays, seeded with ~300 μl of M. smegmatis or 1–2 ml of M. tuberculosis, were prepared in 0.35% mycobacterial top agar (MBTA) with 1 mM CaCl₂ and incubated at 37°C overnight or 1–2 weeks, respectively. Phage lysates were diluted in phage buffer (10 mM Tris-HCl, pH 7.5; 10 mM MgSO₄; 68.5 mM NaCl; 1 mM CaCl₂).

Host range tests

Lysates of phages were serially diluted in phage buffer and 3 μl of one hundred-fold dilutions were spotted onto fresh lawns of M.smegmatis mc²155, Jucho, MKD8, M. aichiense and M. tuberculosis mc² 7000. The plates were incubated at 37°C and plaque formation was assayed after 24–48 hours for M. smegmatis and M. aichiense strains and 6 days for M. tuberculosis.

Isolation of expanded host range mutants

Independent host range mutants of BPs and Halo were isolated by preparing multiple lysates on M. smegmatis mc²155 – each derived from a single plaque – and using these to infect M. tuberculosis mc²7000. Individual plaques were picked into phage buffer and re-plated on both M. smegmatis mc²155 and M. tuberculosis mc²7000. Individual plaques were picked from the M. smegmatis mc²155 plate, plaque-purified, and shown to infect both strains at equal efficiencies of plating. Halo ehr1 was identified by 454 sequencing, and the remaining mutants were identified by Sanger sequencing of BPs/Halo gene 22.

Rosebush host range mutants were isolated by plating independently prepared lysates onto M. smegmatis Jucho, and individual plaques were picked and purified. Following re-plating onto both strains, phage mutants recovered from M. smegmatis mc²155 were shown to infect both strains at equivalent efficiencies. A clone library of Rosebush mut1 DNA was prepared in pBluescript IIKS+ vector and sequenced to 92% coverage using an ABI 3730 sequencer. A mutation was identified at coordinate 30,977 in gene 32; no additional mutations were identified. Sequence coverage was subsequently extended to 99.9% using IonTorrent sequencing. Gene 32 of the remaining four mutants was amplified via PCR from phage lysates using primers RB_gp32_F and RB_gp32_R and sequenced using an ABI 3730 Sanger sequencer form the primers RB_gp32_F, RB_gp32_R, gp32seqF1-4, and gp32seqR1-4, identifying three additional mutations residing in 32. The final mutant (mut4) was sequenced using phenol-extracted genomic DNA via 454 to 99.8% coverage, identifying a point mutation in gene 42; no additional mutations were identified.

Transfection

Transfection was typically performed using phage genomic DNA (150–400 ng) and electrocompetent bacteria, and after electroporation allowed to recover for 1 hour in 7H9/ADC/CaCl₂ and then plated with addition of a permissive host. Plates were assayed for plaque formation after 48hr of growth at 37°C. For transfection of Zemlya into M. smegmatis, recovery was completed as above, but plated with soft nutrient agar and S. lividans pre-germinated spores (50°C for 10min in YEME) onto Nutrient agar plates containing Chloramphenicol (2ug/ml), which S. lividans is naturally resistant to.

BRED Mutagenesis

Point mutations in Halo gene 22 were introduced using BRED mutagenesis, as described previously (Marinelli et al., 2008). Briefly, for each mutation, induced electrocompetent cells of the recombineering strain mc²155:pJV53, prepared as described previously (van Kessel and Hatfull, 2007), were transfected with 100–300 ng phage DNA and 50 ng each of two complementary 71-mer recombineering oligonucleotides (oligos) containing the mutation to be incorporated centrally located. Reactions were recovered for approximately 1 hour in 7H9 with 10% ADC and 1 mM CaCl₂ and plated prior to lysis as top agar lawns with ~300 μl M. smegmatis mc²155 cells. Plaques were picked into 100 μl phage buffer, and 1 μl of these were screened by PCR using the highly sensitive mismatch amplification mutation assay (MAMA)-PCR (Swaminathan et al., 2001) and Platinum Taq High Fidelity DNA Polymerase (Invitrogen). Primary plaques containing a mixture of mutant and wild-type alleles were serially diluted, and re-plated on M. smegmatis. Individual secondary plaques were then screened for the mutation by MAMA-PCR, and mutants were confirmed by Sanger sequencing.

All oligonucleotides were purchased from IDT Inc; the 71-mer recombineering oligos were gel purified. PCR products were processed using QIAquick PCR-Purification (QIAGEN) or MinElute PCR Purification Kits (QIAGEN), eluting DNA in a minimal volume of sterile water.

Adsorption Assays

M. smegmatis mc²155, M. smegmatis Jucho, or M. tuberculosis mc²7000 cells in mid-log phase were diluted to an OD₆₀₀ of ~0.2 and infected with phage at an multiplicity of infection (M.O.I.) of 0.00005 – 0.01. At 10-minute intervals, aliquots were removed, the bacterial cells along with any adsorbed phage were pelleted by centrifugation, and the supernatant was serially diluted and tittered on a fresh lawn of M. smegmatis mc²155. The percentage of total infectious particles remaining in the supernatant was calculated and plotted.

Supplementary Material

01

Table S1. Oligonucleotides used in this study.

• Mycobacteriophage host preference correlates with genome type

• Mycobacteriophage expanded host range results from tail gene mutations

• A diverse host population and easy host range expansion promotes phage diversity