There is a great interest in using systems biology approaches and system theory to understand all interactive processes in a cell and in an organism. This systems approach depends upon generating enormous amounts of data that describe all gene sequences and their transcript and protein levels, as well as their regulatory controls and dynamic interactions. Mathematical models based on this information should then be able to explain the various networks and to make predictions concerning any perturbation of the system. However, models are only as good as the data used to generate them, and good models also depend upon their testability in different genetic and environmental conditions. This will be an intimidating job for most complex organisms as we can infer by the various attempts to describe the genetic and developmental networks of a simple phage like λ (1, 3, 41, 69). Although λ may be the most completely understood organism, we know that there is a lot more to learn (16, 45). This review describes several new findings about λ regulation, which add to previously unknown levels of regulation and question certain dogma and which will be essential for meaningful advances in systems biology.

The bacterial RNA polymerase (Pol) binds and transcribes from the early promoters pL and pR (Fig. 1). RNA Pol transcribes as far as the transcription terminators tL1 and tR1 beyond pL and pR, respectively (54, 56, 57). The product of the N gene, located on the tL1-terminated transcript, modifies subsequent RNA Pol molecules initiating at pL and pR so that they transcribe through not only tL1 and tR1 but also other terminators downstream (7, 8; for see reviews, see references 11, 22, 23, and 53). Thus, through a mechanism of transcription antitermination, N allows other genes distal to the tL and tR terminators to be transcribed. These genes encode various functions of the phage required for either lytic or lysogenic development. The genes in the pL operon include those involved in general and site-specific recombination. The genes in the pR operon include replication genes O, P, and ren, as well as another positive regulatory gene Q. Like N, Q protein is an antitermination factor (55). Q modifies RNA Pol as it initiates at the strong constitutive promoter pR′ that is the promoter for genes that encode lysis and morphogenetic proteins of the phage. Transcripts from pR′ normally terminate at tR′ some 200 bases away (Fig. 1). Q action on RNA Pol prevents tR′ termination, allowing expression of the rest of the lytic genes beyond Q. Once Q activates these genes, the cell is committed to lysis. Thus, the lytic pathway, activated by N and Q, is a temporal cascade of transcription from the early promoter pR to the late promoter pR′, providing lytic gene expression through sets of termination and antitermination events.

FIG. 1. Gene and transcription map of λ. Genes are shown in the shaded rectangle. The early transcripts for pL and pR promoters are shown as red arrows. The late transcript from pR′ is indicated with black arrows. The CII-activated pI, pRE, and (more ...)

Two phage proteins, Int and CI, are required to form stable lysogens. Int allows the integration of the phage genome into the bacterial chromosome, and CI represses the two early phage promoters to prevent any lytic phage gene expression. When λ first infects, Int and CI are not initially made, and the phage initiates gene expression along a set of events that are common to both the lytic and lysogenic pathways. If conditions are favorable during this initial phase, Int and CI synthesis can be switched on to enable lysogenic development. This activation depends primarily upon the phage CII function.

The cII gene is located between the tR1 terminator and the replication genes and, thus, is transcribed with the early lytic genes. However, CII protein is required only for lysogenic development of infecting phages. Another gene required for lysogenic development is cIII, located beyond tL1 in the pL operon. Mutations in these genes as well as in the cI gene encoding the repressor function cause λ plaques to be clear, unlike the normal turbid plaques where the turbidity indicates growth of lysogenic cells. Whereas the CI repressor is required to maintain the repressed lysogenic state, the CII and CIII proteins are only required to initially activate CI synthesis (34). Once CI has been made, the CII and CIII functions are no longer required because CI can maintain its own synthesis.

The CII protein is the primary switch element for establishing repression after infection. CII activates and thereby coordinates transcription from three promoters, which are silent until sufficient CII accumulates. Promoter pI transcribes the int gene (59), promoter pRE transcribes the cI gene (60), and promoter pAQ transcribes an antisense RNA within the Q gene (31) (Fig. 1). Whereas Int and CI proteins are required to form stable lysogens, the Q antisense RNA inhibits Q function and in so doing preempts lethal lytic expression as cells become lysogenic (36).

Coordination in turning on all three of these promoters is required to allow a stable λ lysogenic response with CII acting as the “switchman.” Without activation of this switch by CII, λ will follow the lytic pathway, a default course. The tripartite action of CII requires a high threshold level of the protein. The levels of CII are reduced by the bacterial HflB (FtsH) protease, which binds to the C-terminal part of CII, causing its rapid decay (13, 35). Like CII, the CIII protein is also important to establish lysogeny, but unlike CII, its role is indirect. CIII inhibits the bacterial protease HflB (4, 32). Thus, without CIII, HflB degrades CII and not enough CII is available for the switch. Inhibition of HflB by CIII allows accumulation of CII to the threshold level critical for its coordinated action on its three promoters. It is known that a cell infected by one phage predominantly follows the lytic default pathway (about 99% of the time), but infection by more than one phage greatly enhances the lysogenic switch (about 50% of the time). Since calculations indicate that a cell infected by two phage is sufficient to tip the balance in favor of lysogeny (9, 36, 37), it is reasonable to expect that more CIII and CII would be made from twice the number of infecting phage, thus enabling the CII-mediated switch. CIII concentration is important for this higher lysogenic response because of the limiting levels of HflB in cells (65). We will discuss in more detail the regulation of CII expression and its activation of the three silent promoters.

Once CI repressor is made from pRE, the repressor shuts off the early promoters pL and pR by binding at operators OL and OR (Fig. 1 and 2). This shuts off all λ functions in the pL and pR operons including CII and CIII, thereby precluding continued CI expression from the CII-activated pRE promoter. CI repressor continues to be made, however, by the enhancement of a weak promoter, pRM, located downstream of pRE. CI itself activates the pRM promoter. Thus, CI acts both as a repressor and an activator.

FIG. 2. Looping of the λ operators OL and OR during CI binding. The promoters pL and pR control the transcription of the early genes N and cIII and the genes cro and cII, respectively (Fig. 1). The promoter pRM transcribes CI, rexA, and rexB in a lysogen. (more ...)

The OR operator is composed of three repressor binding sites, OR1, OR2, and OR3. Three similar sites are present at OL (Fig. 2). CI binds tightly and cooperatively to OR1 and OR2 as a tetramer to repress pR. A similar tetramer bound at OL also represses pL (48). These two sets of tetramers interact to form an octamer looping the DNA between OL and OR (Fig. 2) and generating a more stable repression complex of the early promoters (15, 17, 52, 64). In this octamer complex, the CI bound at OR2 represses pR and at the same time activates the adjacent pRM promoter to stimulate cI transcription. As CI expression increases, the higher CI concentration enables a CI dimer to bind to each of the weaker OR3 and OL3 operators in the looped complex (Fig. 2). These dimers interact with each other and repress CI synthesis by occluding pRM (17, 64). This positive and negative autoregulation at pRM by CI ensures that a narrow range of CI repressor level is maintained, which is optimum for stable lysogeny but is adjusted low enough for efficient induction of the lysogen. The stable lysogen makes sufficient repressor to block not only prophage lytic development but also that of any extraneous infecting λ phage, thus imparting immunity to the lysogen against lytic superinfection.

The cro gene, the first gene expressed from the pR operon, is required for lytic development. Cro is a weak repressor that binds the same OL and OR sites as CI but with different relative affinities. Cro binds best to OR3, by which it blocks CI synthesis from pRM (33). These two proteins, CI and Cro, which block each other's transcription, have been described as the elements responsible for the classical bistable λ genetic switch (47). CI is responsible for the immune state by turning off pL, pR, and Cro. Cro, once made, is able to maintain an anti-immune state by turning off pRM and CI expression while leaving pL and pR and its own expression at least partially on (19, 25, 64). This anti-immune condition, however, can only be established with a genetically modified, defective prophage (19).

Cro has been believed to play an active role in switching lysogenic cells to the lytic state following induction. However, recent findings have raised questions about this role. Originally, a specific OR3 mutant was shown to prevent Cro repression of pRM and to inhibit prophage induction. From this result, it was proposed that the role of Cro was to aid the transition from lysogeny to lytic development by reducing repressor levels as induction ensued. This is the basis for the classical Cro/CI bistable, genetic switch (47). Yet recent experiments demonstrated that the OR3 mutation used in the early studies had a second effect. The mutation also prevented CI binding and negative autoregulation, thereby causing higher than normal steady-state levels of CI repressor in the lysogen. This higher level of CI was a cause for the reduced induction (15, 17) reported by Ptashne and his colleagues.

The initial stages of lysogenic induction might be expected to reduce CI functional levels and initiate Cro expression. A recent study in which CI was partially inactivated did not reveal any difference in induction levels in the presence or absence of cro, supporting the idea that the CI/Cro bistable switch is not essential for the induction process (64). Atsumi and Little constructed a λ hybrid phage in which the cro gene is replaced with the lacI repressor gene; in this hybrid, LacI at the lac operator controls pR transcription during lytic growth (2, 49). In this arrangement, the λ OR3 operator and pRM promoter are unchanged and controlled only by CI repressor; Cro is not present. The hybrid phage can be induced and completes the lytic cycle in the absence of Cro binding to OR3, supporting the idea that Cro is not absolutely required for induction. However, Atsumi and Little did notice that, compared to wild-type λ, induction required about a 40% greater level of UV irradiation in the absence of Cro, suggesting that the presence of Cro may aid the induction process at suboptimal UV doses.

It is known that Cro turns down transcription from pL and pR by four- and twofold, respectively (reference 64 and references therein), thereby reducing CII levels indirectly by its effect on CIII as well as directly by reducing cII gene expression (51). By this cumulative reduction of CII levels, Cro helps the default lytic course. Consistent with this, a mutation in cro causes an infecting phage to abort the lytic pathway, which can be restored by a second mutation in cII (20, 36). This suggests that too much CII protein accumulates in a cro mutant.

There is an unexpected delay period in lytic development between the time of activation of λ genes by N antitermination and the time of activation of the late λ genes from pR′ by Q antitermination. Although the molecular mechanism is unknown, the delay is believed to be caused by a high threshold requirement for Q in its role of antiterminator, as indicated by direct measurements of Q protein accumulation and Q antitermination activity (36). The delay in committing to the lytic state is likely very important for the ability of λ to switch over to the lysogenic state. This allows time for CII function to be made and for the accumulation of CI and the Q antisense RNA, with the latter blocking Q function. Premature expression of the Q gene prevents lysogeny, as evidenced by the effect of a mutation, byp, which results in a new promoter just upstream of Q. This promoter causes Q to be expressed immediately after infection, resulting in clear plaque formation (8, 63). Mutation of the pAQ promoter prevents the antisense RNA synthesis and also causes clear plaque formation (30, 31) because Q function and lytic expression are not inhibited. How the antisense transcript inhibits Q function remains to be determined. It is not known whether pAQ RNA inhibits Q synthesis by an antisense mechanism or Q activity by binding to the protein.

N is a positive regulator for the lytic pathway. N protein, once made, binds to specific nascent RNA sites NUTL and NUTR. The NUTL site is between the N gene and the pL promoter, and NUTR is downstream of the first gene, cro, in the pR operon (Fig. 1). A set of bacterial proteins called Nus, which takes part in cellular transcriptional and translational processes, interacts with N and RNA Pol during N-mediated transcription antitermination (10, 12, 40, 61). BoxA and BoxB are elements of the RNA Nut sites (24). BoxB is a stem-loop structure in the RNA and binds N (6). BoxB-bound N associates with NusA and RNA Pol. BoxA RNA binds the NusB and NusE complex (43, 46). Thus, N, Nus factors, and Nut RNA interact with RNA Pol while tethered by the nascent RNA transcript (44, 66-68) (Fig. 3). N is the driving force for antitermination, while Nut and the Nus factors provide stability and full activity to the system (14, 50). Although N overrides both Rho-dependent and -independent terminators, the exact way in which N binds and changes the elongating RNA Pol is still being debated (21).

FIG. 3. RNase III control of N-mediated translation repression. RNA Pol is modified by N and Nus factors bound to the NutL RNA structure to become transcription termination resistant. This transcription antitermination complex represses N gene translation (top). (more ...)

The importance of N′s role as an early regulatory protein of phage λ is reflected by the variety of its own regulation. First, N transcription is down-regulated from OLpL by CI- and Cro-mediated repression. Second, N is degraded by protease(s), causing a relatively short half-life. Third, and most recently discovered, is that N represses its own translation in conjunction with the bacterial endoribonuclease RNase III, which regulates this autorepression. This rare ability of N to modulate both transcription and translation is described below.

N is the first gene in the pL operon (Fig. 1). The 220-nucleotide RNA segment upstream of N contains the NutL site to which N binds, a large stem-and-loop structure following NutL, which is cleaved by RNase III (39, 62), and finally the ribosome binding site (Shine-Dalgarno) of N (Fig. 3). N and RNase III act at their respective binding sites to repress and enhance N translation, respectively (67, 68). When RNA Pol first transcribes the pL operon after phage infection, it transcribes through the N gene and terminates at tL1 (Fig. 1). From these terminated transcripts, the N protein is made. As N accumulates, it binds subsequently initiated pL transcripts at NutL and modifies RNA Pol to the N-antitermination form with the help of Nus factors. The N-modified RNA Pol transcribes through N and tL1 as well as distal genes and terminators. N translation from the pL-antiterminated transcripts is dependent upon the level of RNase III in cells. If it is high, processing of the RNase III site (Fig. 3) between NutL and the N message is rapid, allowing continuous translation of N from the message. If the RNase III level is low, processing of RIII occurs too slowly, resulting in autorepression of N translation. Since the cellular RNase III level increases and decreases with growth rate (5), N translation also increases and decreases with growth rate (67).

N translation repression is abolished by mutating components of the N antitermination complex such as N, NutL, and Nus factors, or by replacing E. coli RNA Pol with T7 RNA Pol for the transcription of N (68). Thus, N translation repression is dependent on forming the E. coli RNA Pol transcription antitermination complex. The N-antitermination complex, while extending RNA beyond tL1, represses N gene translation from the same RNA molecule. In contrast, if the N-antitermination complex does not form, the N message is terminated at tL1, and the N gene is translated. RNase III interferes with translation repression by the N-antitermination complex by splitting the NutL site where the complex is bound from the N message. Ultimately, the level of the global regulator, RNase III, controls N gene expression; growth in rich medium leads to high RNase III and N levels, favoring lytic development, whereas growth in poor medium or under starvation conditions leads to low RNase III and N levels, favoring lysogenic development (37, 47).

We previously mentioned the lytic/lysogenic decision in terms of the importance of CII and the multiplicity of infection in causing a switch from lytic to lysogenic growth. High multiplicities of infection ensure that CII accumulates to a threshold allowing the lysogenic switch. This CII multiplicity-dependent switch is epistatic to the N effect in rich or minimal medium (18). On the other hand, infection of a cell by a single phage leads predominantly to lytic development under all growth conditions. However, the number of rare lysogens that occur is dictated to some extent by the growth condition and its effect on N expression. As discussed above, during an infection, λ senses cell growth conditions through RNase III levels, and it is RNase III that affects N levels and lysogeny. When a single phage is infecting a cell in minimal medium where N levels are low, lysogeny increases compared with rich medium infections where N levels are high (37). Is the occurrence of the extremely rare lysogens found during single infections in rich medium stochastic or determined by cellular variables? Since the population of cells being infected is in various stages of the cell cycle, it is difficult to address the issue without infecting synchronous cells.

CII is required for Int expression following λ infection. As previously described, int gene transcription is activated from the pI promoter by CII, and this ensures coordinate synthesis of Int and CI proteins to form a stable lysogen. The finding that CII is required for Int synthesis raised the issue as to why Int is not expressed from the pL promoter by N antitermination, even though the N-antitermination complex transcribes the int gene. It was demonstrated by genetic studies that a site in the b (sib) segment of λ beyond int and attP (Fig. 1) inhibits Int expression from pL but not from pI (27). The sequence of the sib site revealed the presence of a large stem-loop structure in the RNA, which is sensitive to processing by RNase III (Fig. 4). When λ infects host mutants defective in RNase III, Int is expressed from pL transcripts. Thus, processing of the pL transcript at Sib by RNase III prevents Int expression. This cleavage event initiates degradation of the int transcript from the 3′ processed end by polynucleotide phosphorylase (PNPase). Thus, RNase III and PNPase are both required to retroregulate Int expression from a downstream site (26).

FIG. 4. Regulation of int gene expression. (A) Int transcription from pL is antiterminated by N and reads through a terminator tI (see panel D) generating an extended stem structure (sib). (B) This structure is processed by RNase III to generate a new RNA 3′ (more ...)

If RNase III processes the pL transcript at Sib to prevent Int expression, how does the pI transcript avoid being processed? This is because the RNA Pol initiating from pI terminates at tI without transcribing the entire Sib site, and the transcript is thus resistant to RNase III and PNPase. On the other hand, N-modified transcription from pL is resistant to termination at tI and extends beyond it, generating the fully sensitive RNase III site, Sib (Fig. 4).

Although Int is needed for phage DNA integration after infection, Int along with Xis is required for prophage DNA excision after induction (42). After integration of the phage DNA into the bacterial chromosome via attP, the sib site is physically separated from int and is located on the opposite side of the prophage (Fig. 5). Because the integration process splits the sib site from int, the prophage int gene transcript from pL no longer carries Sib, and the RNA is stable. This stable pL-int transcript expresses Int and the upstream Xis protein at high levels, allowing the excision to occur immediately following induction (28, 58).

FIG. 5. l DNA integration and excision. The l DNA molecule circularizes following infection (top), and with Int protein of the phage and IHF protein of E. coli, recombines at specific sites attP-P′ in the phage and attB-B′ in the bacterial chromosome (more ...)

λ was discovered by Ester M. Lederberg in 1951 as a prophage in E. coli K12 (38). Our knowledge about λ biology has continued to grow and evolve over the intervening 55 years. During this entire time, λ has served as a paradigm for studying gene regulation and development. Models to explain λ's regulatory and developmental switches have also been evolving as new discoveries continue to be made, right up to the present. The study of λ continues to be important for many reasons, not the least of which is as a paradigm for systems biology studies.

Although λ may be the most completely understood organism, from a systems biology viewpoint there is a lot more to learn (3, 41, 69). In this review, we discuss recent findings that reveal new aspects of λ genetic circuitry that (i) deemphasize the role of the classical Cro/CI bistable switch in lysogenic induction, (ii) show the existence of a DNA operator-repressor loop, and (iii) explain the extreme stability of λ lysogenic immunity. Furthermore, we describe how stability of critical regulatory proteins, like N, CII, and Q, determines their threshold requirements in the temporal context of λ's development and provides a paradigm for all genetic and developmental regulation. It is clear that any systems biology approach to understand the big picture is doomed to fail unless the details of the system are known. Complete and accurate experimental results are crucial for mathematical modeling. Earlier mathematical models of the systems of λ require reevaluation in the light of the new revelations, as discussed here, and need to evolve as more knowledge accumulates.

We dedicate this article to three friends and lambdologists, Hatch Echols, Nat Sternberg, and Ira Herskowitz, who made seminal contributions to science and were devoted to the biology of lambda.

This research was supported in part by the Intramural Research Program of the NIH, National Cancer Institute, Center for Cancer, and by a Trans NIH/FDA Intramural Biodefense Program Grant from NIAID to D.L.C. The research carried out at Hebrew University was supported in part by The Israel Science Foundation (grants 489/01-1 and 340/04).

Amos B. Oppenheim passed away after submission of the manuscript.