Two More Ques­tions about CRISPRs

by Merry

Over the past eight years, step-by-step, re­searchers have es­tab­lished a ba­sic un­der­stand­ing of the CRISPR de­fenses against for­eign DNA so widely used by both bac­te­ria and ar­chaea. We re­lated the early story on STC in 2008 and com­mented on six ad­di­tional ques­tions in 2011. Still, ques­tions re­main. Here is an up­date on two of them. (For the ba­sic nuts and bolts of CRISPR-Cas sys­tems, please visit those two ear­lier posts.)

Fig­ure 1. Fol­low­ing in the foot­steps of the bac­te­ria, we bol­ster our adap­tive im­mune de­fenses by vac­ci­na­tion with in­ac­ti­vated viruses. Source

Ques­tion #1: How do you sur­vive lytic phage in­va­sion long enough to ac­quire a spacer?

A CRISPR-Cas de­fense (CRISPR for short) can quite ef­fi­ciently dis­patch a phage in­vader, but do­ing so re­quires hav­ing cr­RNAs on pa­trol that tar­get that phage. To have such armed guards on duty, not only must you have a spacer in your CRISPR li­brary ac­quired from that phage, but you must have tran­scribed the li­brary and processed the tran­script into ac­tive cr­RNAs. This takes time. A lytic phage can cause ir­re­versible dam­age more quickly. So how does a bac­terium or an ar­chaeon ac­quire a new spacer in time and live?

Two pur­ported sources of spacer se­quences are 'dead' phages, both those dead on ar­rival and those dis­patched by the host. In many nat­ural ecosys­tems, phage viri­ons have a half-life on the or­der of a few days due to dam­age from UV ir­ra­di­a­tion, en­zy­matic at­tack, and other agents. In the up­per 200 m of clear oceanic wa­ters, for ex­am­ple, "a large pro­por­tion" of viri­ons may not be in­fec­tious. Many of these, how­ever, may still be ca­pa­ble of de­liv­er­ing at least part of their ge­nomic DNA into the cell, thus pro­vid­ing the bac­terium with a sam­pling of in­nocu­ous DNA from lo­cal phages. As early as 2010, re­searchers had pos­tu­lated that cells could ac­quire spac­ers from such DNA frag­ments and thereby pre­pare to de­fend them­selves should they en­counter the same phage again. In ad­di­tion, bac­te­ria and ar­chaea that carry a re­stric­tion-mod­i­fi­ca­tion (R‑M) sys­tem can cre­ate harm­less frag­ments from the genomes of in­vad­ing phages. If the in­com­ing phage DNA con­tains un­methy­lated re­stric­tion recog­ni­tion sites, the cell's re­stric­tion en­donu­cle­ase will cleave the DNA at those sites, thus in­ac­ti­vat­ing the phage and pro­vid­ing ma­te­r­ial for po­ten­tial spacer ac­qui­si­tion.

An­other group of re­searchers has now put these ideas to the test us­ing Strep­to­coc­cus ther­mophilus SMQ-1279. This strain car­ries two ac­tive type II‑A CRISPR sys­tems (CR1 and CR3) and a type II R‑M sys­tem — two de­fenses that work in con­cert to pro­vide a more ef­fec­tive anti-phage de­fense. It is sen­si­tive to in­fec­tion by lytic phage 2972 but its R‑M sys­tem re­duces suc­cess­ful in­fec­tion by that phage by five or­ders of mag­ni­tude. Typ­i­cally re­stric­tion is a 'leaky' de­fense be­cause the cell con­tains ac­tive DNA methy­lases as well as re­stric­tion en­donu­cle­ases. On rare oc­ca­sion the phage DNA is methy­lated be­fore it is cleaved, the phage repli­cates, and host ly­sis re­leases a swarm of fully-methy­lated prog­eny that can in­fect that host strain with im­punity. Tak­ing ad­van­tage of this fea­ture these re­searchers were able to pre­pare stocks of both un­methy­lated phage 2972 and fully-methy­lated phage 2972M+ for use in their ex­per­i­ments.

Fig­ure 2. Bac­te­rio­phage-in­sen­si­tive mu­tant (BIM) count ex­pressed as the av­er­age ra­tio of the num­ber of BIMs from each treat­ment (L, M, H, B×2) over the num­ber of BIMs in the con­trol. Cells were ex­posed to phages at a base MOI (B) of 0.1 of 2972M+, plus ei­ther a Low (0.01), Medium (0.05) or High (0.10) MOI of un­methy­lated phage 2972, or, as a con­trol, anad­di­tional MOI of 0.1 of 2972M+ (B×2). Source

First ex­per­i­ment: In­fect this host with phage 2972M+. When at­tacked by fully methy­lated phages, the host's R‑M sys­tem is use­less. More­over this host strain does not have a spacer tar­get­ing this phage. Host sur­vival de­pends on the rapid de­vel­op­ment of CRISPR im­mu­nity by ac­qui­si­tion of a spacer tar­get­ing the at­tack­ing phage — a rare event that yields few sur­viv­ing colony-form­ing units (CFUs). These sur­vivors, re­ferred to as BIMs (Bac­te­rio­phage In­sen­si­tive 'Mu­tants'), al­most al­ways have a newly ac­quired spacer. Next step: re­peat this ex­per­i­ment but this time chal­lenge the cells with a mix­ture of both methy­lated and un­methy­lated phage. If their hy­poth­e­sis was cor­rect, re­stric­tion en­donu­cle­ase cleav­age of the un­methy­lated phage genomes would fa­cil­i­tate spacer ac­qui­si­tion and thereby in­crease the num­ber of new BIMs. Their re­sults sup­port this hy­poth­e­sis as the num­ber of BIMs in­creased as much as ten-fold, the great­est in­crease be­ing ob­served with the high­est MOI (mul­ti­plic­ity of in­fec­tion) of the un­methy­lated phages. When these BIMs were tested by PCR for spacer ac­qui­si­tion, at least 97% had new spac­ers.

Their sec­ond ex­per­i­ment mim­ic­ked the first, only here the un­methy­lated phages were re­placed by UV-in­ac­ti­vated phages. This pro­to­col in­creased the num­ber of BIMs three-to-four-fold. They at­trib­uted the lesser in­crease to the like­li­hood that UV-dam­age ren­dered many of the phage viri­ons in­ca­pable of DNA de­liv­ery into the hosts, thus ef­fec­tively de­creas­ing the MOI. To­gether these two ex­per­i­ments strongly sup­port the hy­poth­e­sis that the pres­ence of non-in­fec­tious phage DNA, whether due to dam­age prior to en­try or to cleav­age by the host af­ter en­try, can pro­vide a route to safe spacer ac­qui­si­tion.

Still one more ex­per­i­ment was needed to pro­vide de­fin­i­tive ev­i­dence that the spac­ers were ac­quired from the non-in­fec­tious DNA. For this they used two dif­fer­ent, but re­lated, phages: 2972 and 858. Their genomes are 90.9% iden­ti­cal at the nu­cleotide level and share some, but not all, po­ten­tial CRISPR tar­gets. (A po­ten­tial tar­get is sim­ply a DNA se­quence close to the pro­to­spacer ad­ja­cent mo­tif [PAM] that is rec­og­nized by the host's CRISPR-Cas sys­tem.) First they chal­lenged hosts with only 2972M+, then iso­lated the re­sul­tant BIMs and se­quenced their newly ac­quired spacers.A third of the newly ac­quired spac­ers also matched phage 858 as was ex­pected due to the par­tial iden­tity of their genome se­quences.

Next they pre­treated the hosts by in­cu­ba­tion with com­pletely UV-in­ac­ti­vated phage 858 for 4 h at an MOI of 5. They then chal­lenged the cells by in­fec­tion with fully methy­lated phage 2972M+, as be­fore. Now 87.2% of the new spac­ers matched phage 858, sug­gest­ing that spac­ers had been ac­quired from the in­ac­ti­vated phage. In ad­di­tion, one par­tic­u­lar spacer clinched their ar­gu­ment. Here I need to first men­tion that al­though some CRISPR-Cas sys­tems re­quire an ab­solute match be­tween the archived spacer and the in­vad­ing DNA, many are less de­mand­ing. This par­tic­u­lar CRISPR-Cas sys­tem tol­er­ates one mis­match. This is a good strat­egy for the bac­terium as it means a phage can­not evade with a sin­gle point mu­ta­tion in the pro­to­spacer. Re­turn­ing to their ex­per­i­ment, that par­tic­u­lar newly-ac­quired spacer per­fectly matched the 858 genome, but had one mis­match with the 2972M+ genome. Thus this spacer could only have been ac­quired dur­ing the pre­lim­i­nary chal­lenge with the UV-in­ac­ti­vated phage 858.

In other words, bac­te­ria use in­ac­ti­vated viruses as a 'vac­cine' to elicit adap­tive im­mu­nity against spe­cific pathogens. It seems we weren't the first.

Ques­tion #2: Must you choose be­tween host­ing a prophage and hav­ing a CRISPR de­fense?

Not all phage in­fec­tion is a bad thing. Tem­per­ate phages can launch a lytic in­fec­tion im­me­di­ately upon ar­rival, but they also have the op­tion to qui­etly in­te­grate into the host chro­mo­some as a prophage, i.e., to lyso­g­e­nize the host. A prophage can re­side peace­fully long-term, all the while con­fer­ring ben­e­fits on the host. Ben­e­fits? Typ­i­cally prophages block in­fec­tion by re­lated phages. Of­ten they en­code and ex­press vir­u­lence fac­tors or other meta­bolic genes that as­sist their host. How­ever, given dam­age to host DNA or other stresses, a prophage can ex­cise, repli­cate, and ul­ti­mately lyse the host. The truly canny host would want a CRISPR de­fense that would al­low a prophage to in­te­grate and per­sist but de­stroy it when it ex­cises.

The ac­cepted view has been that CRISPRs shoot on sight, tem­per­ate and lytic phages alike. How­ever, the known CRISPR-Cas sys­tems are a di­verse lot. Based on dif­fer­ences in their Cas genes, they are clas­si­fied into three types and at least 12 sub­types. Since the Cas pro­teins func­tion in cr­RNA pro­cess­ing and tar­get­ing, these mech­a­nisms vary from type to type. Type I and II CRISPR-Cas sys­tems had been shown to be in­com­pat­i­ble with prophages, but when re­searchers re­cently in­ves­ti­gated type III, they found a dif­fer­ent story.

For this study, re­searchers as­sem­bled a test sys­tem com­posed of parts from a va­ri­ety of sources. For the bac­te­r­ial host they em­ployed S. au­reus RN4220 armed with a plas­mid, pGG3, car­ry­ing the type III‑A CRISPR-Cas sys­tem from S. epi­der­midis RP62a and a chlo­ram­pheni­col-re­sis­tance gene for plas­mid main­te­nance. For the phage, they used four λ‑like tem­per­ate phages that had been found as prophages in a S. au­reus clin­i­cal iso­late (New­man), thus were cre­atively named φNM1 – 4. All four make them­selves use­ful to that host by car­ry­ing genes that en­hance host path­o­genic­ity. They added an­other key com­po­nent: a spacer known as 32T that had been found in a CRISPR lo­cus of an­other S. au­reus strain. Spacer 32T had near per­fect iden­tity to a tar­get se­quence that was con­served in φNM1, φNM2, and φNM4. They added this spacer to their host's ar­ma­ment by in­sert­ing it into the CRISPR ar­ray on the plas­mid. Lastly, they in­serted an ery­thromycin-re­sis­tance gene in the phage genome so they could later se­lect for cells with a sta­bly-in­te­grated prophage.

S. au­reus, equipped with the 32T spacer on the plas­mid, is highly im­mune to φNM1. Lytic in­fec­tion is re­duced by seven or­ders of mag­ni­tude rel­a­tive to con­trols that lack that spacer. In con­trast, the fre­quency of lyso­g­e­niza­tion by φNM1 was the same whether the host car­ried the spacer or not. A CRISPR spacer that in­ter­feres with lytic in­fec­tion but tol­er­ates lyso­g­e­niza­tion? This tol­er­ance was un­ex­pected. Af­ter all, the tar­get se­quence would be present in the phage genome both be­fore and af­ter in­ser­tion as a prophage.

What could be go­ing on here? The re­searchers car­ried out ap­pro­pri­ate tests to show that the spacer still matched the tar­get and that the CRISPR-Cas sys­tem had not been in­ac­ti­vated. Some­how this CRISPR de­fense sensed a dif­fer­ence be­tween a phage genome launch­ing a lytic in­fec­tion and the same genome when in­serted as a prophage, and re­sponded ap­pro­pri­ately to both. What is that dif­fer­ence? Most phage genes are si­lenced in the prophage whereas most are ac­tively tran­scribed dur­ing lytic repli­ca­tion. Thus a hy­poth­e­sis:  type III‑A CRISPR-Cas sys­tems tar­get only tran­scrip­tion­ally ac­tive tar­gets.

Fig­ure 3. Con­di­tional tol­er­ance is achieved via tran­scrip­tion-de­pen­dent CRISPR-Cas tar­get­ing. Tol­er­ant spac­ers 17T, 43T, and 61T‑1 tar­get lytic genes than are si­lenced dur­ing lysogeny, whereas in­tol­er­ant spac­ers 2B and 17B tar­get genes that are con­sti­tu­tively ex­pressed dur­ing lysogeny. C is a pGG3 non-tar­get­ing con­trol. Blue bars = CFUs un­der ery­thromycin and chlo­ram­pheni­col se­lec­tion (i.e., cells car­ry­ing a sta­bly in­te­grated prophage and a pGG3 plas­mid). Source

As their first test of this hy­poth­e­sis, these re­searchers cre­ated five spac­ers that tar­geted var­i­ous re­gions of the phage genome and in­serted each one in turn into the CRISPR ar­ray on the plas­mid. Three of these spac­ers tar­geted lytic genes that are si­lenced dur­ing lysogeny. All three were tol­er­ant, i.e., did not in­ter­fere with lyso­g­e­niza­tion. The other two spac­ers tar­geted genes ex­pressed dur­ing lysogeny and these were in­tol­er­ant, re­duc­ing the num­ber of CFUs with in­te­grated prophages. More ev­i­dence that tran­scrip­tion across the tar­get was re­quired for CRISPR im­mu­nity came from a par­tic­u­lar mu­tant that es­caped tar­get­ing al­though its tar­get se­quence was un­changed. Its es­cape was at­trib­uted to a point mu­ta­tion in the up­stream pro­moter re­gion that strongly re­duced tran­scrip­tion of the tar­get.

Dur­ing prophage in­duc­tion, pre­vi­ously si­lenced genes be­come ac­tively tran­scribed. Thus, their hy­poth­e­sis also pre­dicts that when a tol­er­ated prophage ex­cises and "goes lytic," even a tol­er­ant CRISPR spacer will strike. To test this they used mit­o­mycin C to in­duce prophage ex­ci­sion. In lyso­gens that lacked a spacer tar­get­ing the phage, the prophage ex­cised, repli­cated, and lysed its host. When a lyso­gen car­ry­ing a tol­er­ant spacer was in­duced, the CRISPR de­fense struck and de­stroyed the ex­cised phage, thereby pre­vent­ing host ly­sis.

The mol­e­c­u­lar mech­a­nism un­der­ly­ing this CRISPR tol­er­ance has yet to be un­rav­eled. This is one more ex­am­ple re­mind­ing us that al­though the ba­sics of CRISPR de­fenses are well es­tab­lished, many in­ter­est­ing ques­tions re­main. There are sub­tleties, adap­ta­tions, and nu­ances that we are just be­gin­ning to ex­plore.

 

Other Posts

  • The Dy­namo at Work

    Pic­tures Con­sid­ered #15 by Christoph — Why on earth would one stare – even for a few sec­onds – at a blurry, low-res, pix­e­lated movie, watch­ing a jit­tery rod tum­bling coun­ter­clock­wise around its pivot? Be­cause it's ATP syn­thase, a key en­zyme of the en­ergy me­tab­o­lism in every liv­ing cell, that's ro­tat­ing there. And when...

  • Three to Tango

    by Elio — How many sym­bioses that in­volve three part­ners can you think of? Surely, not many. But here is one, in­volv­ing a path­o­genic fun­gus, a bac­terium, and a fun­gal virus.

  • Why CRISPR Doesn't Work in E. coli

    by Elio — We re­ceived this query: "I en­joyed the ar­ti­cle on your blog 'Six Ques­tions About CRISPRs' by Merry Youle. I am an ex-lamb­dol­o­gist, hav­ing quit phage lambda in the early 70s and moved to GM-plants. There is one thing about CRISPR that I do not un­der­stand: Why did lamb­dol­o­gists not find CRISPR? They found phage..."

  • Tal­mu­dic Ques­tion #94

    by Mark O. Mar­tin — Bac­te­r­ial bi­o­lu­mi­nes­cence is ubiq­ui­tous in ma­rine en­vi­ron­ments, both among free liv­ing and mu­tu­al­is­tic mi­crobes. Why is this phe­no­type com­par­a­tively rare in ter­res­trial and fresh­wa­ter en­vi­ron­ments (the fas­ci­nat­ing Pho­torhab­dus be­ing a no­table ex­cep­tion)?