Does conjugation support gene selection?

Does conjugation support gene selection?

We are searching data for your request:

Forums and discussions:
Manuals and reference books:
Data from registers:
Wait the end of the search in all databases.
Upon completion, a link will appear to access the found materials.

I have already posted this on chat but haven't got any response. A recent question on group selection stimulated me to ask this here.

QUESTIONS: Why should bacteria conjugate? If we consider that a bacterium is another bacterium's rival (in terms of obtaining resources) why should a bacterium "share" it's antibiotic resistance gene, for example, with another bacterium? I can think of 2 reasons. One is group selection, and the other is "Gene Selection". Group selection has been highly criticized in recent years, and individual selection has dominated thinking, but here my support goes for gene selection rather than individual selection.

Gene selection seems to me the easier explanation over individual selection. Even if we consider the fact that a colony is made from divisions of single cells and the cells are extremely similar and also that conjugation is more probable between cells of a colony, we should also consider that conjugation can happen between very different bacterial species.

Interesting Question (glad the "group selection" post got you thinking!)

Me and two friends thought about your question during our Biology BSc at ICL (in fact we also thought about an even more puzzling phenomenon called "bacterial transformation" which is like conjugation, but they incorporate genes from dead bacteria!!) our first year course convener recognized our answers as "most likely candidates" so here they are:

1. Stress-induced Experimentation Hypothesis

The likelihood of conjugation/transformation occurring may change for a bacterial population as a whole depending on how stressed it is (or perhaps how unfamiliar the environment is?). Stress signals could cause the population to be more "risky" and take up the DNA from surrounding living/dead bacteria and incorporate it with its own genome: "the hell, we're out of options… lets transform into something that resembles the indigenous species of this strange ecosystem!". Indeed Bacteria have memory (see: chemotaxis) this means that perhaps they can be able to tell whether they have been receiving more stress now as opposed to before, meaning they can "trial and error" with various genes that have been imported within their plasmids! - "Let's try this gene, oh wait… I'm dying, hmm let's kill off this plasmid and take some more genes!".

In reply to this my course convener said:

"ah yes, a switch from asexual to sexual reproduction occurs in many organisms under stress. The genes that cause the sex/transformation to occur may gain a net benefit from recombination with genes taken from other organisms/the environment."

2. Diversity Plagarism Hypothesis

If we look at the entire population of conjugating bacteria as a single competitive unit itself (e.g. like a tissue is a coordinated group of specialized cells) that can compete with other populations of bacteria who cannot conjugate the conjugating population will "win". This is because a conjugating population will always have much higher phenotypical diversity as it can mix traits between the individuals. This means a conjugating population is much more efficient at undergoing natural selection for beneficial traits or trait combinations.

Think of natural selection (coupled with mutation) as an inventor… something that helps you invent much "fitter bacteria" is the ability to pull things apart from this bacterium and that bacterium and glue them together, this is essentially what conjugation/transformation does!! it allows you to pull bits from this bacterium and stick it to bits from that bacterium to get a "super-bacterium". It's like open source technology (free to share) as opposed to patented technology!! Conjugation is immediately a good thing because it means you can "steal" cool technology from living/dead bacteria and incorporate it into your population's gene pool!

Hope this was helpful!


Ancillary Genes Carried on Conjugative Plasmids

Conjugative plasmids have been categorized in several ways, depending on the aspect of interest:

Accessory genes. In most cases, conjugational plasmids carry numerous other accessory genes not related to the conjugation process. Importantly, for humans, antibiotic resistance genes are among those carried on plasmids, and such transmissible antibiotic resistance factors are called R factors.

Replication genes. Conjugative plasmids unable to stably replicate in the same cell are said to belong to the same incompatibility (Inc) group. Features of replication control are responsible for the incompatibility. At least 30 different Inc groups of plasmids have been identified among the Enterobacteriaceae alone.

Antibiotics as a selective driver for conjugation dynamics

It is generally assumed that antibiotics can promote horizontal gene transfer. However, because of a variety of confounding factors that complicate the interpretation of previous studies, the mechanisms by which antibiotics modulate horizontal gene transfer remain poorly understood. In particular, it is unclear whether antibiotics directly regulate the efficiency of horizontal gene transfer, serve as a selection force to modulate population dynamics after such gene transfer has occurred, or both. Here, we address this question by quantifying conjugation dynamics in the presence and absence of antibiotic-mediated selection. Surprisingly, we find that sublethal concentrations of antibiotics from the most widely used classes do not significantly increase the conjugation efficiency. Instead, our modelling and experimental results demonstrate that conjugation dynamics are dictated by antibiotic-mediated selection, which can both promote and suppress conjugation dynamics. Our findings suggest that the contribution of antibiotics to the promotion of horizontal gene transfer may have been overestimated. These findings have implications for designing effective antibiotic treatment protocols and for assessing the risks of antibiotic use.

Horizontal gene transfer (HGT) is a main contributor to the spread of antibiotic resistance genes 1–3 . Conversely, it has been generally assumed that antibiotics promote HGT 4,5 . One of the most common mechanisms for HGT, particularly for the transfer of plasmids such as those carrying antibiotic resistance, is conjugation 2,6,7 . There are two broad ways in which antibiotics can promote HGT via conjugation. First, when dosed at sublethal concentrations, antibiotics can increase the conjugation rate by either activating the excision of transferrable genes from the host chromosome or by inducing the expression of conjugation machinery (or both) 8–12 . Second, it has been speculated, but not proven, that antibiotics can cause global cellular responses, such as changes in cell wall composition or upregulation of key survival genes 13,14 , which can indirectly increase the conjugation rate 15–17 .


As an example, conjugation was performed using the donor strain S17-1 containing conjugative plasmids pARO181 or pARO190 and recipient strain JM109. After conjugation on LB agar containing nalidixic acid and a selectable marker, kanamycin or ampicillin, following antibiotic selection as mentioned earlier, the transconjugants were counted to calculate the efficiency of conjugation (Table II). Total DNA was isolated from some of the colonies on the conjugation plates by vortex mixing with 15 μL of TE buffer for 3 minutes and run on 0.8% agarose gel electrophoresis using the conjugative plasmid DNAs isolated from the donor strains as positive controls. As shown in Fig. 1, total DNA from transconjugants should give the same pattern as conjugative plasmids isolated from donor strains. In addition, the miniprep alkaline lysis method [ 12 ] can also be used for isolating the plasmids for donor, recipient, and transconjugants.

Agarose gel electrophoresis of transconjugants compared with the conjugative plasmids. Lane M: 1-kb DNA ladder. Lane 1: plasmids (a) pARO181 and (b) pARO190 isolated from the donors using the mini-prep. Lanes 2–3: total DNA involving (a) pARO181 and (b) pARO190 isolated from the donors. Lanes 4–6: total DNA isolated from the transconjugants containing (a) pARO181 and (b) pARO190. Lane 7: total DNA isolated from the recipient. Lane 8: total DNA isolated from donor E. coli S17-1. The total DNA extracts from transconjugants are similar to those from the donor S17-1 containing pARO181 or pARO190, but there is no plasmid in the recipient strain JM109. The thick expected bands in the lanes of donors (lanes 2–3) and transconjugants (lanes 4–6) are the supercoil plasmid DNA (lower band) and open circular plasmid DNA (higher band). The highest band in the lanes 2–8 is the chromosomal DNA.

Conjugative plasmid Number of donors in 1 mL Number of recipients in 1 mL Donor: recipient ratio Number of transconjugants in 1 mL Efficiency of conjugationa a The efficiency of conjugation is the number of transconjugants per donor cell.
pARO181 1.62 × 10 11 2.83 × 10 11 2:3 1.53 × 10 6 9.44 × 10 −6
pARO190 1.73 × 10 11 2.83 × 10 11 2:3 1.84 × 10 6 1.06 × 10 −5
  • All values were from two replicates in two independent experiments.
  • a The efficiency of conjugation is the number of transconjugants per donor cell.


Acquisition of RP4 impacts growth dynamics

To investigate how plasmid acquisition might affect transconjugant growth, we sought to compare the growth dynamics of de novo transconjugants (which have not undergone physiological adaptation) to those previously established (and therefore fully adapted). De novo transconjugants (T) can be readily generated using our previously established protocol for estimating conjugation efficiencies (Lopatkin et al, 2016a ): donors (D) and recipients (R) carrying unique resistance genes are mixed under conditions that minimize growth. Since T is uniquely resistant to both antibiotics, it can then be directly selected from the population, and its growth tracked over time in a microplate reader (Fig 1B). This procedure is ideal for our purposes since it minimizes growth and adaptation during the conjugation window, ensuring that subsequent dynamic characterization of de novo T captures emergent phenotypic changes. In contrast, adapted T can be isolated by streaking conjugation mixtures onto dual antibiotic agar plates individual clones can then be grown and stored for subsequent testing (Dahlberg & Chao, 2003 Rozwandowicz et al, 2019 ). These established transconjugants exhibit stably reproducible growth rates and are often used to quantify fitness costs and/or determine the timescale of compensatory mutations (Harrison et al, 2015 Hall et al, 2020 ).

Using this approach, we first focused on the well-established, large conjugal plasmid RP4 (

60 kb). We chose RP4 since initial characterization revealed it as imposing a fitness cost on the cell, which allows us to distinguish between immediate acquisition costs and compensatory mutations thereafter (Appendix Fig S1A). Briefly, D and R were established using Escherichia coli MG1655 strains (Appendix Table S1A) R expresses spectinomycin (Spec) resistance, and D, which carries the RP4 plasmid, encodes kanamycin (Kan) resistance. To measure the growth of de novo T, D and R were mixed for one hour at 25°C, diluted 1,000× into Spec-Kan liquid media, and tracked via OD600 in a microplate reader. In parallel, adapted T clones were incubated under identical conditions (e.g., one hour, 25°C) to control for any physiological effects of the protocol itself, and subsequently diluted into Spec-Kan liquid media at a comparable initial cell density, as verified with colony forming units (CFU) (Appendix Fig S1B).

Strikingly, de novo T appeared to grow overall slower than adapted T (Fig 1C). Indeed, curve-fitting using the established Baranyi equation (Baranyi & Roberts, 1994 ) revealed that de novo T’s growth rate was significantly lower, and the lag time significantly higher, than that of adapted T (Fig 1D, Appendix Fig S1C, P = 1.12e-08 and P = 5.71e-09, respectively). These results were independent of the conjugation duration: mixtures incubated for both 15 and 120 min exhibited similar trends (Fig 1E and Appendix Fig S1C). However, when these populations were diluted and re-grown after 24 h, both the growth rate and lag time were fully restored (Fig 1F). Importantly, in all cases, the recovered growth rates remained lower than that of the plasmid-free strain, indicating the plasmid retained its fitness cost (Fig 1D). Finally, these results were independent of the method used to quantify growth parameters since three additional quantification methods resulted in qualitatively consistent trends (Appendix Fig S2, Appendix Table S2).

Although these results initially suggest that RP4 acquisition is costly, we identified several protocol-related factors that could potentially account for these observations. First, the protocol requires de novo T to adjust to a new growth environment, possibly altering growth dynamics independently from conjugation-specific effects. However, adapted T was subjected to identical experimental conditions, which accounts for any effect of environmental adaptation on growth independent of conjugation. Second, competition with residual R/D cells could alter de novo T growth. To test this, we initiated growth of adapted T with R/D diluted 1,000× in the background media this approximates the parental densities present during the conjugation experiment. Doing so did not qualitatively affect the trajectory of adapted T, nor the growth discrepancy between adapted and de novo T (Appendix Fig S1D). Moreover, there was no appreciable background conjugation of R/D at this density (Appendix Fig S1E), indicating that neither population survived long enough to conjugate during this time window. Overall, we conclude that RP4 is indeed costly to acquire, and that plasmid acquisition can modulate both the growth rate and lag time.

Introducing a quantitative metric for the plasmid acquisition cost

That RP4 acquisition induced transient changes in both growth rate and lag time is intuitive: The reduced growth rate is consistent with fitness cost interpretations as a metabolic burden associated with replication/protein expression. Moreover, a cell’s immediate response to a metabolic perturbation is known to manifest in altered lag dynamics, such as following a nutrient shift (Madar et al, 2013 ). Therefore, to facilitate quantification, we sought to define a rigorous and accurate plasmid acquisition metric that would capture all effects of plasmid acquisition on growth dynamics. To this end, a minimal model of cell growth suggested that the time required to reach a predetermined threshold density is an inclusive proxy for changes in both growth rate and lag time (Fig 2A and Appendix Fig S3) this is consistent with a recent study that used an analogous “time to threshold” method to compare conjugation efficiencies in vitro (Bethke et al, 2020 ).

Figure 2. Acquisition cost quantification for RP4

  1. The time (t * , orange) at which a specified cell density threshold (T * , top purple) is reached uniquely depends on the initial cell density (T0, bottom purple), and the growth rate (µ, aqua) and lag times (λ, orange). Assuming background subtraction, the line can be represented by the equation that is shown.
  2. Representative standard curve generation is shown. Left: To generate the standard curve, adapted T are diluted in 10-fold increments and growth is measured over time (dark to light gray is T0). Circles indicate the t * (purple line) corresponding to T * (orange line). Aqua line represents the out-growth of T following a conjugation experiment (i.e., de novo T). Right: The initial cell density is plotting against t * black line indicates the standard curve.
  3. Left: RP4 adapted T growth initiated with decreasing true T0 (dark to light gray) blue markers show the time to reach OD600 of 0.275 (t * ). Right: Standard curve is shown in blue. Error bars are standard deviation of three biological replicates.
  4. True and predicted CFU of RP4 with the recipient E. coli strain MG1655. Scatter points represent three biological replicates, and bar height is the average.
  5. True and predicted CFU of RP4 with the recipient K. pneumoniae (KPN) strain AL2425. Scatter points represent four biological replicates, and bar height is the average.

Source data are available online for this figure.

Briefly, let T(t) describe the growth of transconjugants over time, and µ be the maximum specific growth rate. Consistent with previous literature (Métris et al, 2006 ), extrapolating the exponential growth region to the horizontal axis allows us to define the geometric lag time (λ), which corresponds to the observable onset of exponential growth (Fig 2A). During the exponential phase, T(t) can thus be described by the line: ln(T(t)) = ln(T0) + µ(t-λ), where T0 is the true initial population density. Under these conditions, the time (t * ) it takes the population to reach a specific detection level (T * ) is inversely correlated with T0. In other words, we can predict an unknown initial cell density (Tpred) from its observed time to threshold (t * ) value, assuming T * , µ, and λ are constant (Fig 2B). Conversely, a discrepancy between Tpred and true T0, which CFU can determine, specifically indicates that µ and/or λ have changed. For our purposes, consider a standard curve relating t * and T0 using adapted T, and a Tpred generated from a de novo T population as described above. In that case, any discrepancy between Tpred and true T0 (e.g., Tpred/T0 < 1) can be attributed to growth-specific consequences of acquiring a plasmid. Moreover, the use of T0 as a metric of acquisition cost, rather than t * , allows us to simultaneously compare conjugation rates across diverse experimental conditions and plasmids. Additionally, generating an a priori standard curve spanning a broad range of initial densities also avoids the variability inherent in manually diluting adapted T to a specific target number, as we had done initially (Fig 2B). As such, predicted compared to true CFU represents a more robust and trustworthy quantification method.

We verified the utility of this metric by confirming it could capture the observed discrepancies in RP4 growth. Specifically, we built a standard curve using adapted T (Fig 2C) and predicted the initial de novo T population densities (Tpred) based on times to threshold quantitated from observed growth curves. Having previously verified the true T0 with CFU counts, we found that Tpred was significantly less than true T0 (P = 0.0143, right-tailed t-test), indicating an acquisition cost for RP4 (Fig 2D), as expected. We note that a standard curve generated with R and D diluted 1,000× in the background media to approximate the parental densities present during the conjugation experiment does not significantly change the observed discrepancy between true T0 and Tpred (Appendix Figs S4A and B), consistent with earlier.

Generality of the plasmid acquisition cost

To investigate the generality of the acquisition cost, we first determined whether it was unique to our particular experimental conditions. Specifically, we compared true T0 and Tpred estimates for RP4 using different experimental parameters (e.g., dilution factor, conjugation time window, recipient strain). Results revealed that the RP4 acquisition cost was independent of the conjugation time window as shown previously (Appendix Fig S4C i), recipient strain (Appendix Fig S4C ii), and the dilution factor (Appendix Fig S4C iii-iv) indeed, dilution factors of 150×, 500×, 1,000×, and 5,000× predicted a separate but parallel standard curve (Appendix Fig S4D), indicating a systematic difference between the two estimates. Moreover, RP4 was costly to acquire for the clinically isolated recipient strain Klebsiella pneumoniae(KPN), indicating species-level generality (Fig 2E). The drastic difference in RP4 acquisition costs between E. coli and KPN recipients suggest that cost is not solely a function of particular plasmids strain/species-level attributes are likely key as well.

Next, we reasoned that the long replication time of the large RP4 plasmid (

60 kb), coupled with the significant amount of energy required to synthesize conjugation machinery upon acquisition, likely imposed an immense energetic burden that led to transient growth inhibition, and thus, a high acquisition cost (San Millan & MacLean, 2017 ). Conversely, we hypothesized that mobilizable plasmids, which are transferred by conjugation in trans but do not themselves encode conjugation machinery, would most likely result in a minimal acquisition cost. The absence of genes encoding for conjugation machinery reduces both the plasmid size and the burden of the machinery production. To test this, we used the FHR helper plasmid system described previously (Dimitriu et al, 2014 ), which is not self-transmissible but can mobilize any co-residing plasmid that encodes the recognition sequence oriT. We chose the mobilizable plasmid pR which carries chloramphenicol resistance. As with RP4, this plasmid exhibits an overall fitness cost (Appendix Fig S5). Consistent with our hypothesis, post-conjugation growth curves overlapped with those used to determine the standard curve (Fig 3A). Indeed, experiments revealed a statistically identical match between true T0 and Tpred (Fig 3B, P = 0.34 and 0.86 for two- and one-tailed t-tests, respectively). Thus, we conclude that pR does not induce a significant acquisition cost. More generally, these results confirm that observed acquisition costs, as determined by the difference in transconjugant estimates (T0 and Tpred), are not artifacts of the experimental method and can be detected using this quantification metric.

Figure 3. Generality of acquisition cost

  1. OD600 for de novo (aqua) and adapted transconjugants (gray) for the plasmid pR are shown over time. Black lines are best-fits. Individual curves are biological replicates.
  2. True and predicted CFU for the plasmid pR are statistically identical (P = 0.34 and 0.86 for one and two-tailed t-tests, respectively). Scatter points represent biological replicates, and bar height is the average.
  3. Growth rates are shown for adapted T carrying RP4 under variable glucose (glu) and casamino acid (caa) concentrations. Values represent % w/v. Scatter points represent biological replicates.
  4. Acquisition costs were quantified for the same glucose and casamino acid concentrations from (C). Scatter points represent the average, and error bars represent standard deviation, of three biological replicates. Aqua and red indicate glucose at 0.4% and 0.04% w/v, respectively. Y-axis is acquisition cost (Tpred/T0) normalized to the cost in the absence of casamino acids.
  5. Tpred compared to T0 for six well-characterized plasmids. Representatives are shown of two biological replicates (see Appendix Fig S10 for day-to-day variability). R1, R1drd, and pRK100 do not show a significant acquisition cost (P = 0.93, 0.79, and 0.28, respectively), whereas RIP113, R6K, and R6Kdrd do (P = 6.79e-05, 7.57e-05, and 0.037, respectively, one-tailed t-test, Appendix Table S3A).
  6. Tpred was significantly less than T0 for clinical plasmids p41, p168, p193, and p283 at 37°C (P = 7.10e-05, 2.10e-05, 0.021, 1.90e-05, respectively, n = 4, 4, 3, 2, respectively, one-tailed t-test, Appendix Table S3A). In all cases except p283, bars represent averages and scatter points individual measurements from at least three biological replicates p283 has two biological replicates.
  7. Tpred for two clinical plasmids, p193 and p168, at 30°C was significantly less than T0 (P = 2.80e-04, 2.72e-04, respectively, n = 3, 4, respectively, one-tailed t-test, Appendix Table S3A).
  8. All acquisition costs scattered against the fitness cost measured under the identical condition for each plasmid. Acquisition costs are measured as the ratio between Tpred/T0. Black line is the linear regression line of best fit, and R 2 = 0.01 (shown in the bottom left). Error bars represent standard deviation the type of replicates used for these error bars is listed in Appendix Table S3A.

Source data are available online for this figure.

Given that RP4 and pR-specific differences likely arise due to differences in energetic demand, we hypothesized that altering growth efficiency (e.g., the amount of substrate consumed that is converted to biomass) (Chudoba et al, 1992 ) would modulate acquisition costs. Intuitively, inefficiently growing cells generate excess available energy (Russell & Cook, 1995 Russell, 2007 ) that may be readily applied to plasmid-related metabolic demands, potentially resulting in a lower acquisition cost. In contrast, efficiently growing cells devote the bulk of available energy to biomass production (Low & Chase, 1999 ), and thus, reallocating that energy to plasmid demands may increase acquisition costs. To test this hypothesis, we focused on modulating growth conditions. It is well-established that excess glucose yields highly inefficient E. coli growth (Liu, 1998 Basan et al, 2015 ), but that efficiency is restored with exogenous amino acid supplementation (Akashi & Gojobori, 2002 Waschina et al, 2016 ). Adapting a recent approach that leveraged this trade-off (Lopatkin et al, 2019 ), we quantified plasmid acquisition costs under three casamino acid (CAA) concentrations (0, 0.01, and 0.1% w/v) and excess glucose (0.4% w/v). Since higher CAA increases both efficiency and growth rate, we included a fourth combination (0.04%/0.1% w/v glucose/CAA) where growth rate is comparable to 0.4%/0.01% w/v, but results in higher efficiency due to the lower glucose level (Fig 3C). In accordance with our intuition, increased CAA resulted in significant decreases in acquisition costs (Fig 3D P < 0.05, one-tailed t-test, see Appendix Table S3B for exact P-values). Moreover, this trend was not an effect of increased growth: acquisition cost in 0.04%/0.1% glucose/CAA was significantly less than in 0.4%/0.01% glucose/CAA conditions (P = 0.04, one-tailed t-test). Together, these results suggest that environmental conditions significantly modulate acquisition cost through changes in growth efficiency.

We next looked at acquisition costs across a broad panel of plasmids to further investigate its generality. Specifically, we first quantified acquisition costs for six well-characterized conjugal plasmids (R1, incF R1drd, incF R6K, incX R6Kdrd, incX pRK100, incF RIP113, incN Appendix Table S1B) covering four additional incompatibility groups and a range of fitness costs (Appendix Fig S5). Importantly, two pairs of these plasmids represent derepressed mutants and their native repressed counterparts (R1 and R6K). Although both plasmid types express conjugation machinery immediately following their transfer, repressed plasmids tightly regulate machinery expression shortly thereafter, minimizing their fitness costs (Lundquist & Levin, 1986 ) (Appendix Fig S5). Interestingly, we observed no qualitative differences in R1 or R6K acquisition costs, regardless of conjugation repression (Fig 3E): Both R6K/R6Kdrd, and neither R1/R1drd, were costly to acquire. Since all variants express machinery immediately upon entry, these results suggest that repression of conjugative machinery occurs on a timescale longer than that of acquisition cost for these four plasmids. Rather, we noted acquisition cost differences across incompatibility groups. Specifically, all four incF plasmids (R1, R1drd, and pRK100), along with pR from earlier, imposed no significant acquisition cost, whereas R6K and R6Kdrd (incX) and RIP113 (incN) induced a strong acquisition cost. Incompatibility groups are differentiated by their plasmid replication/partitioning mechanisms as well as specific copy number (Kittell & Helinski, 1993 ) for example, incF plasmids typically exist at low copy numbers (≤

2) (Burger et al, 1981 ), whereas incX plasmids can be present at 10–15 copies per cell (Rakowski & Filutowicz, 2013 ). Thus, these results suggest that acquisition costs may arise as a consequence of establishing plasmid-specific replication and maintenance mechanisms.

Finally, we quantified the acquisition cost for four clinical self-transmissible plasmids previously obtained from pathogenic isolates encoding extended-spectrum β-lactamase (ESBL) enzymes (Lopatkin et al, 2017 ). These enzymes confer resistance to a wide range of β-lactam drugs. As with the well-characterized plasmids, these encompass a range of fitness costs, including one that proved beneficial to the host (Appendix Fig S5). Despite this diversity, Tpred was consistently and significantly lower than true T0 for all ESBL plasmids (Fig 3F, P < 0.05, one-tailed t-test, see Appendix Table S3A for exact P-values). This finding held true at lower out-growth temperatures and for different donor/recipient pairs as well (Fig 3G). Moreover, we observed a significant acquisition cost even when the overall fitness cost was beneficial, as was the case with p193. Overall, we conclude that different conjugal plasmids can incur unique and varying acquisition costs, the magnitudes of which are likely dependent on a host of environmental and host-specific factors.

We summarized all acquisition costs as the ratio between Tpred and true T0 and compared them to fitness costs to determine whether there was any significant relationship between the two measurements. Combined, these cover 12 plasmids and five incompatibility groups (Appendix Table S1A). Doing so revealed a statistically insignificant relationship between the two variables (Fig 3H), leading us to conclude that although the two costs may have similar underlying constraints, they are ultimately imposed, at least in part, by independent factors.

Mathematical model of conjugation that accounts for de novo transconjugants

Thus far, we have shown that transconjugants may exhibit a reduced growth rate and overall prolonged lag time immediately following conjugation. These altered growth dynamics returned to expected levels within 24 h however, steady-state growth rates indicated that the plasmid retained its fitness cost, suggesting that compensatory mutations had likely not occurred. While these short-term experiments enabled us to rigorously quantify plasmid-dependent growth effects, they do not necessarily provide insights into longer-term population dynamics. Indeed, over extended periods, de novo transconjugants are continually generated, leading to potential competitive effects and non-homogeneous adaptation within a mixed population. To capture and elucidate this additional complexity and better describe how acquisition cost modulates overall population structure, we modified a previously published model of conjugation (Lopatkin et al, 2017 ) to further investigate the impact of transient physiological plasmid adaptation on both short- and long-term dynamics (Appendix Equations S1-S2 Appendix Fig S6A).

In the simplest case, consider a population S that either does (S 1 ) or does not (S 0 ) have the conjugal plasmid (Appendix Fig S6A). S 0 gains the plasmid from S 1 at a rate constant, the conjugation efficiency (η), thereby becoming S 1 . Likewise, S 1 can lose the plasmid at a rate constant associated with plasmid segregation error (κ), thus transitioning back to S 0 . Critically, the growth dynamics in this original model are governed primarily by the relative fitness cost (e.g., µ1 = µ, and µ0 = αµ), where µ1 and µ0 are the growth rates of S 1 and S 0 , respectively, and α is a scalar that represents the plasmid fitness impact: α > 1 implies the plasmid is costly to the host, whereas α < 1 denotes benefit.

Figure 4. Conjugation model incorporating acquisition cost

  1. Network diagram of conjugation model. The plasmid-free population (S 0 ) acquires the plasmid from the plasmid-adapted population (S A ), turning into a transient de novo transconjugant (S D ) at a rate η (the conjugation efficiency). Finally, S A can revert to S 0 according to the plasmid segregation error rate κ. The de novo population, in turn, transitions into the adapted population at the rate β. Growth rates for S 0 and S D are scaled relative to the plasmid-adapted population (µ) based on the scalars α and ρ, respectively. Not included in diagram: dilution of all populations out of the system at rate D.
  2. RP4 data were fit to the model to calculate β and ρ. Dotted line shows model fit. Data are from Fig 1C.
  3. Model predicts accurate growth rates and lag times based on fitted parameters.
  4. Parameter sensitivity to ρ. Bottom: Average observed growth rate is defined as (S D ρµ+ S A µ)/(S D + S A ), and is measured for increasing β from light to dark. Top: Corresponding population density of S 1 over time. β is fixed to 0.01 based on RP4 fitting.
  5. Parameter sensitivity to β. Bottom: Average observed growth rate measured for increasing β from light to dark where ρ = 0.3 (left) or ρ = 0 (right). Top: Corresponding population density of S 1 over time.
  6. RP4 data (Fig 4B) are re-fit with fixed ρ (top) and fixed β (bottom).

Here, conjugation between S 0 and S A results in the formation of S D , which subsequently transitions into the fully adapted transconjugant population S A at a plasmid-specific transition rate β. As in the original model, all populations grow logistically with a common carrying capacity and are diluted at a rate D. Moreover, S D grows at a rate relative to S A given by µD = ρµ, where ρ is a scalar between 0 and 1. Thus, as with our experimental data, the combined effect of ρ and β accounts for the plasmid acquisition cost. Finally, based on our data that acquisition costs were not qualitatively different between derepressed plasmids and their natively repressed counterparts, we assume that both conjugation and plasmid loss from S D is negligible compared to S A however, we note that this assumption does not qualitatively change the modeling results nor impact our main conclusions (Appendix Fig S6C).

Intuitively, the presence of S D should not drastically alter the qualitative behavior of the expanded model compared to the original. Indeed, when both S 0 and S A are initially present, S D accumulates over short time scales (e.g., 24 h), but is limited due to carrying capacity constraints (Appendix Fig S6B). Rather, S D introduces a temporal delay in overall transconjugant dynamics. Specifically, depending on the values of ρ and β, S D growth decreases the overall growth rate of all plasmid-carrying cells (e.g., S 1 = S D +S A ), thereby delaying peak growth and steady-state density. This interpretation of plasmid acquisition cost is fully consistent with our experimental methods, since it allows for changes in both growth rate and lag times. Moreover, our measurements did not distinguish between S D and S A .

To investigate whether this model could account for our observed experimental results, we first sought to determine whether the expanded model could recapitulate the RP4 dynamics. In particular, we assume that both dilution and plasmid segregation are negligible, consistent with this batch growth setup (Appendix Table S4). We then fit the remaining unknown parameters (ρ and β) to the entire transconjugant population (S 1 ) using a simulation initiated with 100% de novo transconjugants (S D ) (Fig 4B). Doing so resulted in quantitatively accurate predictions of both observed growth rates and lag times (Fig 4C).

We next sought to quantitatively examine the impact of ρ and β on the observed growth rate of S 1 to better understand their individual effect. Since S D transitions to S A , the observed growth rate of S 1 (μobs) can be defined as the weighted average of both transconjugant populations (Appendix equation S6), and therefore changes over time as transconjugants are newly generated and adapted. Simulations revealed that, for a fixed β, a faster de novo transconjugant growth rate allows the observed growth rate to remain higher throughout the entire duration (Fig 4D bottom), and S 1 is modulated accordingly (e.g., growing faster with increasing ρ, Fig 4D top). Likewise, for a fixed ρ, S 1 reached the maximum growth rate (e.g., that set by S A ) more rapidly as the transition rate (β) increased this corresponds to growth inhibition when transition is slow (Fig 4E left). Interestingly, though, even when de novo transconjugants cannot grow (ρ = 0), the observed growth rates largely followed the expected β trajectories, reaching the maximum level within

24 h (Fig 4E right). These simulations suggested that the transition rate (β) rather than the scaled growth rate (ρ), may be primarily responsible for fitting accuracy. Indeed, fitting either β or ρ individually revealed that β was sufficient to predict the RP4 growth dynamics, while ρ was not (Fig 4F). Based on these observations, we considered β the primary driver of the acquisition cost for subsequent investigation into its interplay with fitness costs.

Incorporating plasmid acquisition costs better predict temporal conjugation dynamics

As described above, it is well-established that fitness costs impact population structure and temporal dynamics in heterogeneous communities over time. To this end, we recently used the original model (Appendix Fig S6A, Appendix equations S1-S2) to accurately predict plasmid fate in mixed communities of plasmid-free and plasmid-carrying cells we found that even costly plasmids could persist with a sufficiently rapid conjugation rate (Lopatkin et al, 2017 ). This model, however, was not always able to quantitatively capture temporal behaviors for the range of plasmids we tested. Given that plasmid acquisition and subsequent adaptation is a continuous process, we reasoned that our expanded model may provide further insight and accuracy toward predicting long-term conjugation dynamics. Indeed, simulations initiated with identical conditions as in our previous setup (e.g., initial S D = 0) revealed that S D may still significantly contribute to the overall population structure when an acquisition cost is present (Fig 5A). Thus, we next sought to use our expanded model to determine the relative contribution of both fitness and acquisition costs: To what extent do both these processes modulate overall temporal population dynamics?

Figure 5. Fitness cost versus acquisition cost

  1. Long-term temporal dynamics of S A , S 0 , and S D are shown in red, blue, and gray, respectively, from the main model (Fig 4A). X-axis is time over 21 days, and y-axis is the fraction of each population.
  2. A population is initiated with a 1:1 ratio of S 0 and S A and the total plasmid-carrying population fraction (S 1 = S A +S D ) is tracked over time. Left: β is held constant (β = 0.01) and α is increased from no cost (α = 1) to high cost (α = 1.5). Right: α is held constant (α = 1.2) and β is increased from slow transition (β = 10 −4 ) to rapid transition (β = 1).
  3. Heat map shows where the observed growth rate of S 1 (μobs, calculated using Appendix equation S6) differed from the maximum growth rate under ideal conditions (e.g., μ, if there is no acquisition cost). A 98% threshold was used to numerically define the region where μobs differed significantly from μ (e.g., μobs/μ < 0.98). Any α and β combination meeting this criterion is colored blue and are red otherwise. Changing this threshold did not qualitatively change conclusions (Appendix Fig S9). Changing the conjugation efficiency (η) shifts the boundary (increasing from light to dark shades of blue).
  4. Validation of modeling predictions using four plasmids from left to right: RP4 (in this study), p193, p41, and p168 (from previous work). Data are reproduced with permission from Nature Communications (Lopatkin et al, 2017 ), under the Creative Commons Attribution 4.0 International License. Marker shapes and colors were modified for visualization purposes. Solid line shows original model fit (e.g., Appendix equation S1-S2). Dotted lines show updated model fit (e.g., Appendix equations S3-S5). Experiments were performed at least twice. Error bars represent the standard deviation of four to six measurements.

To address this question, we first examined temporal dynamics in the limiting case described above (Fig 4E), i.e., that de novo transconjugants accumulate and transition, but do not grow (ρ

0). We chose this scenario initially due to the challenge of experimentally measuring de novo growth rates in isolation (i.e., quantifying ρ). Therefore, the long-term fitness cost of carrying the plasmid remains (α) and the acquisition cost is approximated by the transition rate, β. Here, we note that this formulation preserves the observed reduction in overall transconjugant growth rates. Under these conditions, increasing the transition rate increases the adapted fraction (compared to de novo cells), with β = 1 approximating our original model (i.e., S D ≈ 0 and S A ≈ S 1 .

Sensitivity analysis revealed that both the fitness cost (α) and acquisition cost (β) modulate the time until the steady state of S 1 was reached (Fig 5B). To quantify the contribution from each cost, we calculated the difference between the potential maximum growth rate of S 1 (µ, as set by S A ), and the observed steady-state growth rate of S 1 (µobs, Appendix equation S6), which may be less than µ, since it consists of the combined populations S D and S A . In the case where the observed and maximum growth rates are equivalent (μ = μobs), β does not impact overall S 1 growth, and the effect of acquisition cost on the temporal dynamics is negligible compared to fitness cost. Conversely, any deviation (i.e., μobs/μ < 1) indicates a β-specific effect. Indeed, a boundary defined by α and β delineates the parameter regions where observed growth deviates from the maximum growth (Fig 5C). Specifically, when the acquisition cost is sufficiently low, growth of S 1 is largely unaffected by S D therefore, fitness cost alone will likely capture the temporal dynamics. However, when the acquisition cost is high, S D does impact the growth of S 1 , particularly for costly plasmids in this case, acquisition costs should be incorporated to improve temporal predictions. This conclusion was maintained when ρ > 0 (Appendix Fig S6D).

Consistent with the original model, the boundary dictating these scenarios is dependent on the conjugation efficiency (η): when conjugation is sufficiently slow, neither acquisition nor fitness costs alter μobs/μ. Intuitively, in this case, since there is minimal accumulation of S D , de novo transconjugants do not appreciably change the overall growth rate. Surprisingly, however, rapid conjugation is not necessarily beneficial to the plasmid-carrying population. If generation of S D via conjugation is significantly faster than the transition from S D to S A (i.e., η>>β), the observed growth rate decreases this counterintuitive finding suggests that faster conjugation may not favor plasmid retention unless paired with correspondingly rapid adaptation.

These results give rise to two general scenarios of plasmid persistence. First, if the observed and maximum growth rates are approximately equal (μobs ≈ μ), the acquisition cost has a minimal effect and α is sufficient to predict overall dynamics (Fig 5C, red region) this roughly corresponds to α < 1, but can include regions where α > 1 if the conjugation efficiency is sufficiently fast. Moreover, as predicted by our original model, a fast conjugation efficiency can overcome a plasmid burden (α > 1), thus accounting for the expanding red region as a function of η. Second, as μobs diverges from μ (Fig 5C, blue region), increasing fitness costs amplify the effect of β here, acquisition cost impacts growth rate and thus population dynamics.

To test these predictions, we re-examined previously generated data whereby transconjugants were co-cultured with the corresponding plasmid-free population over the course of 14–21 days. In each case, we independently fit β with short-term growth data (Appendix Fig S7) and simulated long-term dynamics (Fig 5D). We found that our experimental data were indeed fully consistent with the expanded model predictions. In particular, although p193 was predicted to have a high acquisition cost, incorporating it does not significantly affect predictive accuracy as the plasmid is beneficial. Conversely, for plasmids that did have a fitness cost (e.g., RP4, p41, and p168), the magnitude of β correlated with whether it was needed for predictive accuracy: plasmids with lower β diverged more strongly from the original model and were better predicted by the updated one that accounted for its effects. Here, we note that changing the value of ρ such that it remained low did not appreciably alter simulation results. Collectively, these results validate our model and demonstrate the potential impact acquisition costs may have on long-term population dynamics. For example, p41 and p168 exhibited similar fitness costs however, the high acquisition cost of the latter would likely favor the former in a mixed population.

Materials and Methods

Data on complete chromosomes and plasmids of prokaryotes were taken from Genbank Refseq (, last accessed November 2011). This included 1,207 chromosomes, 891 plasmids that were sequenced along with these chromosomes, and 1,391 plasmids that were sequenced independently. We used the annotations of the Genbank files, having removed all pseudogenes and proteins with inner stop codons. The information on T4SS was taken from Guglielmini et al. (2011).

Construction of Protein Profiles and Genome Searches

Unless mentioned explicitly, the protein profiles used are those described in Guglielmini et al. (2011). To study the presence/absence of the different components of the vir system, we made additional protein profiles, namely for VirB1, VirB2, VirB5, VirB7, VirB10, and VirB11. We first used PSI-Basic Local Alignment Search Tool (BLAST) (e value < 0.1) to search for distant homologs, using as query each of these genes from the VirB locus of the A. tumefaciens plasmid pTi SAKURA (Refseq entry NC_002147) and the aforementioned databank of completely sequenced replicons. Given the problems of convergence of PSI-BLAST when using complete genomes, and the extensive similarity of plasmid and chromosomal conjugative systems ( Guglielmini et al. 2011), we restricted homology searches to plasmid sequences when building protein profiles. We retrieved the proteins with hits for each protein family and built multiple alignments using MUSCLE ( Edgar 2004). We removed the few proteins with sizes very different from the average. We then rebuilt the multiple alignments with MUSCLE and trimmed them to remove the sites at the edges that were poorly aligned. We used HMMER 3.0 ( Eddy 2011) to produce hidden Markov model (HMM) profiles and to perform searches within genomes. In the analysis of the evolution of the MPFT system, we only considered the hits that colocalized with previously detected vir proteins (VirB3, VirB4, VirB6, VirB8, VirB9). FtsK proteins were retrieved directly by using the PFAM PF01580 profile. TraB proteins, being closely related to FtsK, were retrieved by BLASTP searches of TraB from Streptomyces plasmid pCQ3 (YP_003280879) on the Actinomycetales proteins from the Refseq database. We sampled the top results and then built a protein profile for this protein and searched for its occurrences as for the other profiles. We built a web server to allow running the protein profiles. This is available at

Phylogenetic Analysis

Unless explicitly stated, all phylogenetic analyses were performed with the following procedure. First, sequences were aligned using MUSCLE with default parameters as implemented in SeaView ( Gouy et al. 2010). Second, all columns in the multiple alignment matrix with more than 80% of gaps were removed. Third, 100 replicate trees were built with RAxML 7.2.7 ( Stamatakis 2006) using the model GTRGAMMA. We kept the one with the best likelihood. We calculated bootstraps with the standard implementation and used the autoMR stop criterion to obtain confidence values for each node. There were two exceptions to this method. We aligned the ATPases using MAFFT ( Katoh and Toh 2010) with the G-INSI algorithm and removed the sites containing more than 60% of gaps. We performed the phylogenetic inference as mentioned earlier and additionally with PhyML 3.0 ( Gascuel et al. 2010) under the LG model and with the bioNJ starting tree to get aLRT support values. The alignment of the set of VirB4 and VirD4 was built with MAFFT with the E-INSI algorithm, since these two proteins show different domain organization, and then manually edited. MAFFT was used instead of MUSCLE because it provided better alignments in these cases. The computation of 100 replicates plus hundreds of bootstrap trees was excessively time consuming, given the size of the data set in the VirB4/VirD4 analysis. Thus, we used PhyML 3.0 to build the phylogenetic tree, under the LG model and with the bioNJ starting tree. aLRT support values were also calculated for each node.

The support tests we conducted revealed in this last tree some weak support that conflict with the aLRT values. To further investigate this, we used a reduced data set composed of VirB4 proteins, excluding the distant homolog TraU. Using this data set, we performed the tests described later. All multiple alignments and phylogenetic reconstructions are freely available on DRYAD (

Tests to the Phylogenetic Analysis

To test the robustness of our conclusions based on phylogenetic analysis, we made a number of tests. These analyses aimed at testing the robustness of the conclusions to the multiple alignments, to the identification of informative sites in multiple alignments, and to the use of a protein model matrix. We therefore produced two automatic methods where we make the alignment of the protein using MAFFT and MUSCLE. Informative sites were extracted from the alignments using BMGE ( Criscuolo and Gribaldo 2010). We fine-tuned BMGE parameters for each alignment to obtain a good compromise between the quality and the number of informative sites. The best model to analyze the data was chosen with ProtTest ( Darriba et al. 2011). Note that ProtTest does not analyze the GTR model for proteins, so we cannot assess whether the model chosen by ProtTest is better than ours. Trees were built as before using RAxML, and we generated 100 bootstrap trees for each analysis. To compare the different analyses, we computed the quality of multiple alignment score using the Core component of T-Coffee ( Notredame et al. 2000) for the three methods (our expert analysis, the MAFFT and MUSCLE-based analyses). This score, ranging from 0 to 100, is computed by comparing the consistency of the alignment with a list of precomputed pairwise alignments called library. We used the default “Mproba_pair” library. The key results, e.g., monophyly or basal position of certain clades, were tested for the three methods and are displayed in table 1 and supplementary table S1, Supplementary Material online. Each of these tests has an identification number in the tables. This number is displayed in the respective node in the phylogenetic trees. For example, in figure 2, the node with ID no. 3 refers to the monophyly of TraB and is indicated in table 1 as having 99% bootstrap support in our expert analysis, 100% in the automatic analysis using MAFFT, and 96% in the automatic analysis using MUSCLE. In supplementary table S1, Supplementary Material online, it is indicated that for this analysis the best alignment, as given by T-Coffee, is the one of the expert alignment (score 88), followed by MAFFT (76) and then MUSCLE (67). The node no. 3 in figure 2 is thus indicated in a black circle (high bootstrap support).

Phylogenetic analysis of the AAA+ ATPases associated with conjugation. The position of the root was determined using the AAA+ ATPase VirB11 in a separate analysis. Names along the FtsK tips correspond to the taxonomic origins of each protein, reflecting the width of sampling. Bold vertical black lines represent nodes with a high support value (bootstrap >70% and aLRT >0.7). Bold gray lines represent nodes with high aLRT score (>0.7) but a weaker bootstrap (<70%). The homologs of TcpA are found only in Firmicutes. The homologs of TraB are found only in Actinobacteria. Numbers in circles refer to the analysis of robustness in table 1 (identified in the third column of table 1) black background stands for a high support (≥70% bootstrap in the best-scoring alignment) and gray background for a moderate support (≥50% bootstrap in the best-scoring alignment).

Phylogenetic analysis of the AAA+ ATPases associated with conjugation. The position of the root was determined using the AAA+ ATPase VirB11 in a separate analysis. Names along the FtsK tips correspond to the taxonomic origins of each protein, reflecting the width of sampling. Bold vertical black lines represent nodes with a high support value (bootstrap >70% and aLRT >0.7). Bold gray lines represent nodes with high aLRT score (>0.7) but a weaker bootstrap (<70%). The homologs of TcpA are found only in Firmicutes. The homologs of TraB are found only in Actinobacteria. Numbers in circles refer to the analysis of robustness in table 1 (identified in the third column of table 1) black background stands for a high support (≥70% bootstrap in the best-scoring alignment) and gray background for a moderate support (≥50% bootstrap in the best-scoring alignment).

Analysis of the Robustness of Key Phylogenetic Results.

a Proteins included in the data set.

b The different hypotheses for which we present the bootstrap supports.

c When the hypothesis correspond to what we observe in the reference phylogeny, and if the support value is greater than 50, it is displayed here and in the corresponding figure with a number.

d Bootstrap values for each hypothesis and for each alignment technique.

Analysis of the Robustness of Key Phylogenetic Results.

a Proteins included in the data set.

b The different hypotheses for which we present the bootstrap supports.

c When the hypothesis correspond to what we observe in the reference phylogeny, and if the support value is greater than 50, it is displayed here and in the corresponding figure with a number.

d Bootstrap values for each hypothesis and for each alignment technique.

Relative Decrease in Protein Similarity with Divergence

For each pair of T4SS loci, we made pairwise alignments of each of the orthologous pairs of genes. Alignments were done using an end-gap free version of the Needleman–Wunsch algorithm ( Mount 2004), with a BLOSUM60 matrix, open penalty of 1.2, and extension penalty of 0.8. We then plotted the percentage of similarity between VirB4 homologs and each of the other pairs of homologs. The points for each scatter plot were then fitted with a spline (λ = 1,500), and the curves were superimposed.


Transformation in bacteria was first demonstrated in 1928 by the British bacteriologist Frederick Griffith. [3] Griffith was interested in determining whether injections of heat-killed bacteria could be used to vaccinate mice against pneumonia. However, he discovered that a non-virulent strain of Streptococcus pneumoniae could be made virulent after being exposed to heat-killed virulent strains. Griffith hypothesized that some "transforming principle" from the heat-killed strain was responsible for making the harmless strain virulent. In 1944 this "transforming principle" was identified as being genetic by Oswald Avery, Colin MacLeod, and Maclyn McCarty. They isolated DNA from a virulent strain of S. pneumoniae and using just this DNA were able to make a harmless strain virulent. They called this uptake and incorporation of DNA by bacteria "transformation" (See Avery-MacLeod-McCarty experiment) [4] The results of Avery et al.'s experiments were at first skeptically received by the scientific community and it was not until the development of genetic markers and the discovery of other methods of genetic transfer (conjugation in 1947 and transduction in 1953) by Joshua Lederberg that Avery's experiments were accepted. [5]

It was originally thought that Escherichia coli, a commonly used laboratory organism, was refractory to transformation. However, in 1970, Morton Mandel and Akiko Higa showed that E. coli may be induced to take up DNA from bacteriophage λ without the use of helper phage after treatment with calcium chloride solution. [6] Two years later in 1972, Stanley Norman Cohen, Annie Chang and Leslie Hsu showed that CaCl
2 treatment is also effective for transformation of plasmid DNA. [7] The method of transformation by Mandel and Higa was later improved upon by Douglas Hanahan. [8] The discovery of artificially induced competence in E. coli created an efficient and convenient procedure for transforming bacteria which allows for simpler molecular cloning methods in biotechnology and research, and it is now a routinely used laboratory procedure.

Transformation using electroporation was developed in the late 1980s, increasing the efficiency of in-vitro transformation and increasing the number of bacterial strains that could be transformed. [9] Transformation of animal and plant cells was also investigated with the first transgenic mouse being created by injecting a gene for a rat growth hormone into a mouse embryo in 1982. [10] In 1897 a bacterium that caused plant tumors, Agrobacterium tumefaciens, was discovered and in the early 1970s the tumor-inducing agent was found to be a DNA plasmid called the Ti plasmid. [11] By removing the genes in the plasmid that caused the tumor and adding in novel genes, researchers were able to infect plants with A. tumefaciens and let the bacteria insert their chosen DNA into the genomes of the plants. [12] Not all plant cells are susceptible to infection by A. tumefaciens, so other methods were developed, including electroporation and micro-injection. [13] Particle bombardment was made possible with the invention of the Biolistic Particle Delivery System (gene gun) by John Sanford in the 1980s. [14] [15] [16]

Transformation is one of three forms of horizontal gene transfer that occur in nature among bacteria, in which DNA encoding for a trait passes from one bacterium to another and is integrated into the recipient genome by homologous recombination the other two are transduction, carried out by means of a bacteriophage, and conjugation, in which a gene is passed through direct contact between bacteria. [1] In transformation, the genetic material passes through the intervening medium, and uptake is completely dependent on the recipient bacterium. [1]

Competence refers to a temporary state of being able to take up exogenous DNA from the environment it may be induced in a laboratory. [1]

It appears to be an ancient process inherited from a common prokaryotic ancestor that is a beneficial adaptation for promoting recombinational repair of DNA damage, especially damage acquired under stressful conditions. Natural genetic transformation appears to be an adaptation for repair of DNA damage that also generates genetic diversity. [1] [17]

Transformation has been studied in medically important Gram-negative bacteria species such as Helicobacter pylori, Legionella pneumophila, Neisseria meningitidis, Neisseria gonorrhoeae, Haemophilus influenzae and Vibrio cholerae. [18] It has also been studied in Gram-negative species found in soil such as Pseudomonas stutzeri, Acinetobacter baylyi, and Gram-negative plant pathogens such as Ralstonia solanacearum and Xylella fastidiosa. [18] Transformation among Gram-positive bacteria has been studied in medically important species such as Streptococcus pneumoniae, Streptococcus mutans, Staphylococcus aureus and Streptococcus sanguinis and in Gram-positive soil bacterium Bacillus subtilis. [17] It has also been reported in at least 30 species of Proteobacteria distributed in the classes alpha, beta, gamma and epsilon. [19] The best studied Proteobacteria with respect to transformation are the medically important human pathogens Neisseria gonorrhoeae (class beta), Haemophilus influenzae (class gamma) and Helicobacter pylori (class epsilon) [17]

"Transformation" may also be used to describe the insertion of new genetic material into nonbacterial cells, including animal and plant cells however, because "transformation" has a special meaning in relation to animal cells, indicating progression to a cancerous state, the process is usually called "transfection". [2]

As of 2014 about 80 species of bacteria were known to be capable of transformation, about evenly divided between Gram-positive and Gram-negative bacteria the number might be an overestimate since several of the reports are supported by single papers. [1]

Naturally competent bacteria carry sets of genes that provide the protein machinery to bring DNA across the cell membrane(s). The transport of the exogenous DNA into the cells may require proteins that are involved in the assembly of type IV pili and type II secretion system, as well as DNA translocase complex at the cytoplasmic membrane. [20]

Due to the differences in structure of the cell envelope between Gram-positive and Gram-negative bacteria, there are some differences in the mechanisms of DNA uptake in these cells, however most of them share common features that involve related proteins. The DNA first binds to the surface of the competent cells on a DNA receptor, and passes through the cytoplasmic membrane via DNA translocase. [21] Only single-stranded DNA may pass through, the other strand being degraded by nucleases in the process. The translocated single-stranded DNA may then be integrated into the bacterial chromosomes by a RecA-dependent process. In Gram-negative cells, due to the presence of an extra membrane, the DNA requires the presence of a channel formed by secretins on the outer membrane. Pilin may be required for competence, but its role is uncertain. [22] The uptake of DNA is generally non-sequence specific, although in some species the presence of specific DNA uptake sequences may facilitate efficient DNA uptake. [23]

Natural transformation Edit

Natural transformation is a bacterial adaptation for DNA transfer that depends on the expression of numerous bacterial genes whose products appear to be responsible for this process. [20] [19] In general, transformation is a complex, energy-requiring developmental process. In order for a bacterium to bind, take up and recombine exogenous DNA into its chromosome, it must become competent, that is, enter a special physiological state. Competence development in Bacillus subtilis requires expression of about 40 genes. [24] The DNA integrated into the host chromosome is usually (but with rare exceptions) derived from another bacterium of the same species, and is thus homologous to the resident chromosome.

In B. subtilis the length of the transferred DNA is greater than 1271 kb (more than 1 million bases). [25] The length transferred is likely double stranded DNA and is often more than a third of the total chromosome length of 4215 kb. [26] It appears that about 7-9% of the recipient cells take up an entire chromosome. [27]

The capacity for natural transformation appears to occur in a number of prokaryotes, and thus far 67 prokaryotic species (in seven different phyla) are known to undergo this process. [19]

Competence for transformation is typically induced by high cell density and/or nutritional limitation, conditions associated with the stationary phase of bacterial growth. Transformation in Haemophilus influenzae occurs most efficiently at the end of exponential growth as bacterial growth approaches stationary phase. [28] Transformation in Streptococcus mutans, as well as in many other streptococci, occurs at high cell density and is associated with biofilm formation. [29] Competence in B. subtilis is induced toward the end of logarithmic growth, especially under conditions of amino acid limitation. [30] Similarly, in Micrococcus luteus (a representative of the less well studied Actinobacteria phylum), competence develops during the mid-late exponential growth phase and is also triggered by amino acids starvation. [31] [32]

By releasing intact host and plasmid DNA, certain bacteriophages are thought to contribute to transformation. [33]

Transformation, as an adaptation for DNA repair Edit

Competence is specifically induced by DNA damaging conditions. For instance, transformation is induced in Streptococcus pneumoniae by the DNA damaging agents mitomycin C (a DNA cross-linking agent) and fluoroquinolone (a topoisomerase inhibitor that causes double-strand breaks). [34] In B. subtilis, transformation is increased by UV light, a DNA damaging agent. [35] In Helicobacter pylori, ciprofloxacin, which interacts with DNA gyrase and introduces double-strand breaks, induces expression of competence genes, thus enhancing the frequency of transformation [36] Using Legionella pneumophila, Charpentier et al. [37] tested 64 toxic molecules to determine which of these induce competence. Of these, only six, all DNA damaging agents, caused strong induction. These DNA damaging agents were mitomycin C (which causes DNA inter-strand crosslinks), norfloxacin, ofloxacin and nalidixic acid (inhibitors of DNA gyrase that cause double-strand breaks [38] ), bicyclomycin (causes single- and double-strand breaks [39] ), and hydroxyurea (induces DNA base oxidation [40] ). UV light also induced competence in L. pneumophila. Charpentier et al. [37] suggested that competence for transformation probably evolved as a DNA damage response.

Logarithmically growing bacteria differ from stationary phase bacteria with respect to the number of genome copies present in the cell, and this has implications for the capability to carry out an important DNA repair process. During logarithmic growth, two or more copies of any particular region of the chromosome may be present in a bacterial cell, as cell division is not precisely matched with chromosome replication. The process of homologous recombinational repair (HRR) is a key DNA repair process that is especially effective for repairing double-strand damages, such as double-strand breaks. This process depends on a second homologous chromosome in addition to the damaged chromosome. During logarithmic growth, a DNA damage in one chromosome may be repaired by HRR using sequence information from the other homologous chromosome. Once cells approach stationary phase, however, they typically have just one copy of the chromosome, and HRR requires input of homologous template from outside the cell by transformation. [41]

To test whether the adaptive function of transformation is repair of DNA damages, a series of experiments were carried out using B. subtilis irradiated by UV light as the damaging agent (reviewed by Michod et al. [42] and Bernstein et al. [41] ) The results of these experiments indicated that transforming DNA acts to repair potentially lethal DNA damages introduced by UV light in the recipient DNA. The particular process responsible for repair was likely HRR. Transformation in bacteria can be viewed as a primitive sexual process, since it involves interaction of homologous DNA from two individuals to form recombinant DNA that is passed on to succeeding generations. Bacterial transformation in prokaryotes may have been the ancestral process that gave rise to meiotic sexual reproduction in eukaryotes (see Evolution of sexual reproduction Meiosis.)

Bacterial Edit

Artificial competence can be induced in laboratory procedures that involve making the cell passively permeable to DNA by exposing it to conditions that do not normally occur in nature. [43] Typically the cells are incubated in a solution containing divalent cations (often calcium chloride) under cold conditions, before being exposed to a heat pulse (heat shock). Calcium chloride partially disrupts the cell membrane, which allows the recombinant DNA to enter the host cell. Cells that are able to take up the DNA are called competent cells.

It has been found that growth of Gram-negative bacteria in 20 mM Mg reduces the number of protein-to-lipopolysaccharide bonds by increasing the ratio of ionic to covalent bonds, which increases membrane fluidity, facilitating transformation. [44] The role of lipopolysaccharides here are verified from the observation that shorter O-side chains are more effectively transformed – perhaps because of improved DNA accessibility.

The surface of bacteria such as E. coli is negatively charged due to phospholipids and lipopolysaccharides on its cell surface, and the DNA is also negatively charged. One function of the divalent cation therefore would be to shield the charges by coordinating the phosphate groups and other negative charges, thereby allowing a DNA molecule to adhere to the cell surface.

DNA entry into E. coli cells is through channels known as zones of adhesion or Bayer's junction, with a typical cell carrying as many as 400 such zones. Their role was established when cobalamine (which also uses these channels) was found to competitively inhibit DNA uptake. Another type of channel implicated in DNA uptake consists of poly (HB):poly P:Ca. In this poly (HB) is envisioned to wrap around DNA (itself a polyphosphate), and is carried in a shield formed by Ca ions. [44]

It is suggested that exposing the cells to divalent cations in cold condition may also change or weaken the cell surface structure, making it more permeable to DNA. The heat-pulse is thought to create a thermal imbalance across the cell membrane, which forces the DNA to enter the cells through either cell pores or the damaged cell wall.

Electroporation is another method of promoting competence. In this method the cells are briefly shocked with an electric field of 10-20 kV/cm, which is thought to create holes in the cell membrane through which the plasmid DNA may enter. After the electric shock, the holes are rapidly closed by the cell's membrane-repair mechanisms.

Yeast Edit

Most species of yeast, including Saccharomyces cerevisiae, may be transformed by exogenous DNA in the environment. Several methods have been developed to facilitate this transformation at high frequency in the lab. [45]

  • Yeast cells may be treated with enzymes to degrade their cell walls, yielding spheroplasts. These cells are very fragile but take up foreign DNA at a high rate. [46]
  • Exposing intact yeast cells to alkalications such as those of caesium or lithium allows the cells to take up plasmid DNA. [47] Later protocols adapted this transformation method, using lithium acetate, polyethylene glycol, and single-stranded DNA. [48] In these protocols, the single-stranded DNA preferentially binds to the yeast cell wall, preventing plasmid DNA from doing so and leaving it available for transformation. [49] : Formation of transient holes in the cell membranes using electric shock this allows DNA to enter as described above for bacteria. [50]
  • Enzymatic digestion [51] or agitation with glass beads [52] may also be used to transform yeast cells.

Efficiency – Different yeast genera and species take up foreign DNA with different efficiencies. [53] Also, most transformation protocols have been developed for baker's yeast, S. cerevisiae, and thus may not be optimal for other species. Even within one species, different strains have different transformation efficiencies, sometimes different by three orders of magnitude. For instance, when S. cerevisiae strains were transformed with 10 ug of plasmid YEp13, the strain DKD-5D-H yielded between 550 and 3115 colonies while strain OS1 yielded fewer than five colonies. [54]

Plants Edit

A number of methods are available to transfer DNA into plant cells. Some vector-mediated methods are:

  • Agrobacterium-mediated transformation is the easiest and most simple plant transformation. Plant tissue (often leaves) are cut into small pieces, e.g. 10x10mm, and soaked for ten minutes in a fluid containing suspended Agrobacterium. The bacteria will attach to many of the plant cells exposed by the cut. The plant cells secrete wound-related phenolic compounds which in turn act to upregulate the virulence operon of the Agrobacterium. The virulence operon includes many genes that encode for proteins that are part of a Type IV secretion system that exports from the bacterium proteins and DNA (delineated by specific recognition motifs called border sequences and excised as a single strand from the virulence plasmid) into the plant cell through a structure called a pilus. The transferred DNA (called T-DNA) is piloted to the plant cell nucleus by nuclear localization signals present in the Agrobacterium protein VirD2, which is covalently attached to the end of the T-DNA at the Right border (RB). Exactly how the T-DNA is integrated into the host plant genomic DNA is an active area of plant biology research. Assuming that a selection marker (such as an antibiotic resistance gene) was included in the T-DNA, the transformed plant tissue can be cultured on selective media to produce shoots. The shoots are then transferred to a different medium to promote root formation. Once roots begin to grow from the transgenic shoot, the plants can be transferred to soil to complete a normal life cycle (make seeds). The seeds from this first plant (called the T1, for first transgenic generation) can be planted on a selective (containing an antibiotic), or if an herbicide resistance gene was used, could alternatively be planted in soil, then later treated with herbicide to kill wildtype segregants. Some plants species, such as Arabidopsis thaliana can be transformed by dipping the flowers or whole plant, into a suspension of Agrobacterium tumefaciens, typically strain C58 (C=Cherry, 58=1958, the year in which this particular strain of A. tumefaciens was isolated from a cherry tree in an orchard at Cornell University in Ithaca, New York). Though many plants remain recalcitrant to transformation by this method, research is ongoing that continues to add to the list the species that have been successfully modified in this manner. (transduction): Package the desired genetic material into a suitable plant virus and allow this modified virus to infect the plant. If the genetic material is DNA, it can recombine with the chromosomes to produce transformant cells. However, genomes of most plant viruses consist of single stranded RNA which replicates in the cytoplasm of infected cell. For such genomes this method is a form of transfection and not a real transformation, since the inserted genes never reach the nucleus of the cell and do not integrate into the host genome. The progeny of the infected plants is virus-free and also free of the inserted gene.

Some vector-less methods include:

    : Also referred to as particle bombardment, microprojectile bombardment, or biolistics. Particles of gold or tungsten are coated with DNA and then shot into young plant cells or plant embryos. Some genetic material will stay in the cells and transform them. This method also allows transformation of plant plastids. The transformation efficiency is lower than in Agrobacterium-mediated transformation, but most plants can be transformed with this method. : Formation of transient holes in cell membranes using electric pulses of high field strength this allows DNA to enter as described above for bacteria. [55]

Fungi Edit

There are some methods to produce transgenic fungi most of them being analogous to those used for plants. However, fungi have to be treated differently due to some of their microscopic and biochemical traits:

  • A major issue is the dikaryotic state that parts of some fungi are in dikaryotic cells contain two haploid nuclei, one of each parent fungus. If only one of these gets transformed, which is the rule, the percentage of transformed nuclei decreases after each sporulation. [56]
  • Fungal cell walls are quite thick hindering DNA uptake so (partial) removal is often required [57] complete degradation, which is sometimes necessary, [56] yields protoplasts.
  • Mycelial fungi consist of filamentous hyphae, which are, if at all, separated by internal cell walls interrupted by pores big enough to enable nutrients and organelles, sometimes even nuclei, to travel through each hypha. As a result, individual cells usually cannot be separated. This is problematic as neighbouring transformed cells may render untransformed ones immune to selection treatments, e.g. by delivering nutrients or proteins for antibiotic resistance. [56]
  • Additionally, growth (and thereby mitosis) of these fungi exclusively occurs at the tip of their hyphae which can also deliver issues. [56]

As stated earlier, an array of methods used for plant transformation do also work in fungi:

  • Agrobacterium is not only capable of infecting plants but also fungi, however, unlike plants, fungi do not secrete the phenolic compounds necessary to triggger Agrobacterium so that they have to be added e.g. in the form of acetosyringone. [56]
  • Thanks to development of an expression system for small RNAs in fungi the introduction of a CRISPR/CAS9-system in fungal cells became possible. [56] In 2016 the USDA declared that it will not regulate a white button mushroom strain edited with CRISPR/CAS9 to prevent fruit body browning causing a broad discussion about placing CRISPR/CAS9-edited crops on the market. [58]
  • Physical methods like electroporation, biolistics (“gene gun”), sonoporation that uses cavitation of gas bubbles produced by ultrasound to penetrate the cell membrane, etc. are also applicable to fungi. [59]

Animals Edit

Introduction of DNA into animal cells is usually called transfection, and is discussed in the corresponding article.

The discovery of artificially induced competence in bacteria allow bacteria such as Escherichia coli to be used as a convenient host for the manipulation of DNA as well as expressing proteins. Typically plasmids are used for transformation in E. coli. In order to be stably maintained in the cell, a plasmid DNA molecule must contain an origin of replication, which allows it to be replicated in the cell independently of the replication of the cell's own chromosome.

The efficiency with which a competent culture can take up exogenous DNA and express its genes is known as transformation efficiency and is measured in colony forming unit (cfu) per μg DNA used. A transformation efficiency of 1×10 8 cfu/μg for a small plasmid like pUC19 is roughly equivalent to 1 in 2000 molecules of the plasmid used being transformed.

In calcium chloride transformation, the cells are prepared by chilling cells in the presence of Ca 2+
(in CaCl
2 solution), making the cell become permeable to plasmid DNA. The cells are incubated on ice with the DNA, and then briefly heat-shocked (e.g., at 42 °C for 30–120 seconds). This method works very well for circular plasmid DNA. Non-commercial preparations should normally give 10 6 to 10 7 transformants per microgram of plasmid a poor preparation will be about 10 4 /μg or less, but a good preparation of competent cells can give up to

10 8 colonies per microgram of plasmid. [60] Protocols, however, exist for making supercompetent cells that may yield a transformation efficiency of over 10 9 . [61] The chemical method, however, usually does not work well for linear DNA, such as fragments of chromosomal DNA, probably because the cell's native exonuclease enzymes rapidly degrade linear DNA. In contrast, cells that are naturally competent are usually transformed more efficiently with linear DNA than with plasmid DNA.

The transformation efficiency using the CaCl
2 method decreases with plasmid size, and electroporation therefore may be a more effective method for the uptake of large plasmid DNA. [62] Cells used in electroporation should be prepared first by washing in cold double-distilled water to remove charged particles that may create sparks during the electroporation process.

Selection and screening in plasmid transformation Edit

Because transformation usually produces a mixture of relatively few transformed cells and an abundance of non-transformed cells, a method is necessary to select for the cells that have acquired the plasmid. [63] The plasmid therefore requires a selectable marker such that those cells without the plasmid may be killed or have their growth arrested. Antibiotic resistance is the most commonly used marker for prokaryotes. The transforming plasmid contains a gene that confers resistance to an antibiotic that the bacteria are otherwise sensitive to. The mixture of treated cells is cultured on media that contain the antibiotic so that only transformed cells are able to grow. Another method of selection is the use of certain auxotrophic markers that can compensate for an inability to metabolise certain amino acids, nucleotides, or sugars. This method requires the use of suitably mutated strains that are deficient in the synthesis or utility of a particular biomolecule, and the transformed cells are cultured in a medium that allows only cells containing the plasmid to grow.

In a cloning experiment, a gene may be inserted into a plasmid used for transformation. However, in such experiment, not all the plasmids may contain a successfully inserted gene. Additional techniques may therefore be employed further to screen for transformed cells that contain plasmid with the insert. Reporter genes can be used as markers, such as the lacZ gene which codes for β-galactosidase used in blue-white screening. This method of screening relies on the principle of α-complementation, where a fragment of the lacZ gene (lacZα) in the plasmid can complement another mutant lacZ gene (lacZΔM15) in the cell. Both genes by themselves produce non-functional peptides, however, when expressed together, as when a plasmid containing lacZ-α is transformed into a lacZΔM15 cells, they form a functional β-galactosidase. The presence of an active β-galactosidase may be detected when cells are grown in plates containing X-gal, forming characteristic blue colonies. However, the multiple cloning site, where a gene of interest may be ligated into the plasmid vector, is located within the lacZα gene. Successful ligation therefore disrupts the lacZα gene, and no functional β-galactosidase can form, resulting in white colonies. Cells containing successfully ligated insert can then be easily identified by its white coloration from the unsuccessful blue ones.

Other commonly used reporter genes are green fluorescent protein (GFP), which produces cells that glow green under blue light, and the enzyme luciferase, which catalyzes a reaction with luciferin to emit light. The recombinant DNA may also be detected using other methods such as nucleic acid hybridization with radioactive RNA probe, while cells that expressed the desired protein from the plasmid may also be detected using immunological methods.

Natural selection affects multiple aspects of genetic variation at putatively neutral sites across the human genome

A major question in evolutionary biology is how natural selection has shaped patterns of genetic variation across the human genome. Previous work has documented a reduction in genetic diversity in regions of the genome with low recombination rates. However, it is unclear whether other summaries of genetic variation, like allele frequencies, are also correlated with recombination rate and whether these correlations can be explained solely by negative selection against deleterious mutations or whether positive selection acting on favorable alleles is also required. Here we attempt to address these questions by analyzing three different genome-wide resequencing datasets from European individuals. We document several significant correlations between different genomic features. In particular, we find that average minor allele frequency and diversity are reduced in regions of low recombination and that human diversity, human-chimp divergence, and average minor allele frequency are reduced near genes. Population genetic simulations show that either positive natural selection acting on favorable mutations or negative natural selection acting against deleterious mutations can explain these correlations. However, models with strong positive selection on nonsynonymous mutations and little negative selection predict a stronger negative correlation between neutral diversity and nonsynonymous divergence than observed in the actual data, supporting the importance of negative, rather than positive, selection throughout the genome. Further, we show that the widespread presence of weakly deleterious alleles, rather than a small number of strongly positively selected mutations, is responsible for the correlation between neutral genetic diversity and recombination rate. This work suggests that natural selection has affected multiple aspects of linked neutral variation throughout the human genome and that positive selection is not required to explain these observations.

Conflict of interest statement

The authors have declared that no competing interests exist.


Figure 1. Correlations between summaries of genetic…

Figure 1. Correlations between summaries of genetic variation and recombination rate in the low-coverage dataset…

Figure 2. Comparison of Spearman's for genic…

Figure 2. Comparison of Spearman's for genic regions with the expected values based on forward…

Figure 3. Negative selection is required to…

Figure 3. Negative selection is required to match multiple aspects of the low-coverage data.

Figure 4. Correlation between neutral human-chimp divergence…

Figure 4. Correlation between neutral human-chimp divergence ( d ) and recombination rate.


We are grateful to M. Shirakawa for technical assistance. We thank K. Hikosaka for assistance in the culture of P. falciparum, M. Komatsu (Juntendo University) for Atg7 +/+ and Atg7 -/- MEFs, S. Sato (Universiti Malaysia Sabah) for the pSSPF2 vector, M. Meissner (University of Glasgow) for the P5RT70loxPKillerRedloxPYFP-HX vector, S. Yamaoka (Tokyo Medical and Dental University) for the retroviral pMRXIP vector, T. Kitamura (The University of Tokyo) for the retroviral pMXs-IP vector, and T. Yasui (Osaka University) for the pCG-gag-pol and pCG-VSV-G plasmids. This work was supported by funding from the National Key Research and Development Program of China (No. 2017YFD0500400 to H.J.) The Tokyo Biochemical Research Foundation (to M.H.S.) Grants-in-Aid for Scientific Research on Innovative Areas (No. 25111005, to N.M., and No. 16H0101200, to Y.S.) from the Japan Society for the Promotion of Science and Exploratory Research for Advanced Technology (ERATO) (No. JPMJER1702, to N.M.) and Core Research for Evolutional Science and Technology (CREST) (No. JPMJCR13M7, to N.N.N.) from the Japan Science and Technology Agency (JST).

Carbonyl-reactive Crosslinker Chemistry

Aldehydes (RCHO) and ketones (RCOR') are reactive varieties of the more general functional group called carbonyls, which have a carbon-oxygen double-bond (C=O). The polarity of this bond (especially in the context of aldehydes) makes the carbon atom electrophilic and reactive to nucleophiles such as primary amines.

Although aldehydes do not naturally occur in proteins or other macromolecules of interest in typical biological samples, they can be created wherever oxidizable sugar groups (also called reducing sugars) exist. Such sugars are common monomer-constituents of the polysaccharides or carbohydrates in post-translational modifications (glycosylation) of many proteins (i.e., glycoproteins). In addition, the ribose of RNA is a reducing sugar.

Periodic acid (HIO4) from dissolved sodium periodates (NaIO4) is a well-known mild agent for effectively oxidizing vicinal diols in carbohydrate sugars to yield reactive aldehyde groups. The carbon-carbon bond is cleaved between adjacent hydroxyl groups. By altering the amount of periodate used, aldehydes can be produced on a smaller or larger selection of sugar types. For example, treatment of glycoproteins with 1 mM periodate usually affects only sialic acid residues, which frequently occur at the ends of polysaccharide chains. At concentrations of 6 to 10 mM periodate, other sugar groups in proteins will be affected.

Carbohydrate modification is particularly useful for creating target sites for conjugation on polyclonal antibodies because the polysaccharides are located in the Fc region, away from the antigen-binding site. This results in labeling or crosslinking sites located away from antigen binding sites, ensuring that antibody function will not be adversely affected by the conjugation procedure.

The reaction of sodium periodate with sugar residues yields aldehydes for conjugation reactions. R and R' represent connecting sugar monomers of the polysaccharide. Red asterisks indicate sites of diol cleavage. Sialic acid is also called N-acetyl-D-neuraminic acid.

Carbohydrate modification is particularly useful for creating target sites for conjugation on polyclonal antibodies because the polysaccharides are located in the Fc region. This results in labeling or crosslinking sites located away from antigen binding sites, ensuring that antibody function will not be adversely affected by the conjugation procedure.

Learn more

Select products

This 45-page guide is of value to the novice as well as those who have previous experience with crosslinking reagents. It begins with a basic discussion on crosslinking and the reagents that are used. The guide also contains a discussion on various applications where crosslinking has been applied, including the powerful label-transfer technique for identifying or confirming protein interactions. Crosslinking chemistry is addressed in an easy-to-follow format designed to convey the important information you need without getting lost in details. Each Pierce crosslinking reagent is shown along with its structure, molecular weight, spacer arm length and chemical reactivity. The handbook concludes with a list of excellent references on crosslinker use and a glossary of common crosslinking terms.

Aldehyde-reactive crosslinker reactive groups

As already mentioned above, nucleophilic varieties of primary amines (–NH2) are the main class of compounds that are reactive with aldehydes. Since primary amines are abundant in proteins, it is important to remember that aldehydes represent an amine-reactive crosslinker chemistry just as much as primary amines constitute an aldehyde-reactive crosslinker chemistry (this page).

However, the natural primary amines of proteins (N-terminus of polypeptides and the side chain of lysines), being in the form R–C–NH2, are not strongly or permanently reactive with aldehydes, except in certain conditions and when additional compounds are added to stabilize the bond. The reaction, called reductive amination, is discussed at the end of this article, because its main applications do not involve the use of discrete reactive groups that can be incorporated into ready-to-use labeling or crosslinking reagents.

Instead, hydrazides and alkoxyamines are the most important aldehyde-reactive functional groups for incorporation into synthetic labeling reagents and crosslinkers. The terminal amino groups in these compounds are more strongly nucleophilic than protein amines, and they spontaneously react with aldehydes to form stable bonds.

Hydrazide reaction chemistry

Aldehydes created by periodate-oxidation of sugars in biological samples react with hydrazides at pH 5 to 7 to form hydrazone bonds. Although this bond to a hydrazide group is a type of Schiff base, it is considerably more stable than a Schiff base formed with a simple amine. The hydrazone bond is sufficiently stable for most protein-labeling applications. If desired, however, the double bond can be reduced to a more stable secondary amine bond using sodium cyanoborohydride (see section on reductive amination at the end of this page).

Hydrazide reaction scheme for chemical conjugation to an aldehyde. R represents a labeling reagent or one end of a crosslinker having the hydrazide reactive group P represents a glycoprotein or other glycosylated molecule that contains the target functional group (e.g., an aldehyde formed by periodate oxidation of carbohydrate-sugar groups, such as sialic acid).

With low molecular weights (<1,000 Da) and bright fluorescent signals, hyrazides are used as polar tracers that cross gap junctions and follow neuronal projections. Therefore, hydrazides are often used to investigate neuronal communication in conjunction with labeled dextrans, which are retained within the cell. Hydrazides exist in a range of fluorescent colors and can be multiplexed with other probes. The following confocal image provides a representative example of a hydrazide used for neuronal tracing.

Neuron staining. Confocal image stack of a 10,000 MW calcium green dextran–labeled climbing fiber in a sagittal cerebellar slice, showing incoming axon and terminal arborization (in yellow). The Purkinje cell innervated by this fiber was labeled with Invitrogen Alexa Fluor 568 hydrazide.

Learn more

Select products

Bioconjugate Techniques, 3 rd Edition (2013) by Greg T. Hermanson is a major update to a book that is widely recognized as the definitive reference guide in the field of bioconjugation.

Bioconjugate Techniques is a complete textbook and protocols-manual for life scientists wishing to learn and master biomolecular crosslinking, labeling, and immobilization techniques that form the basis of many laboratory applications. The book is also an exhaustive and robust reference for researchers looking to develop novel conjugation strategies for entirely new applications. It also contains an extensive introduction to the field of bioconjugation that covers all of the major applications of the technology used in diverse scientific disciplines as well as containing tips for designing the optimal bioconjugate for any purpose.

Aniline catalysis of hydrazide conjugation

Recently, it was discovered that aniline acts as a catalyst for hydrazide-aldehyde reactions. The aromatic amine of aniline rapidly and efficiently forms a Schiff base with the aldehyde, effectively increasing the activation of the aldehyde. As a result, the aniline is easily replaced by the hydrazide. Thus, aniline (also called GlycoLink Coupling Catalyst) allows significantly greater total coupling yields with hydrazides and/or greater efficiency (i.e., equal yields with less hydrazide reagent). GlycoLink Coupling Catalyst decreases reaction times and increases aldehyde-hydrazide coupling efficiency, resulting in greater than 90% coupling of glycoproteins in 4 hours.

See the following references for additional information about aniline catalysis:

  1. Byeon, J.Y., et al. (2010). Efficient bioconjugation of protein capture agents to biosensor surfaces using aniline-catalyzed hydrazone ligation. Langmuir 26(19):15430-5.
  2. Dirksen, A., et al. (2006). Nucleophilic catalysis of hydrazone formation and transimination: implications for dynamic covalent chemistry. J. Am. Chem. Soc. 128(49):15602-3.
  3. Dirksen, A., Dawson, PE. (2008). Rapid oxime and hydrazone ligations with aromatic aldehydes for biomolecular labeling. Bioconjug. Chem. 19(12):2543-8.

To learn more about aniline, see the table below.

ProteinMolecular Weight# Glycosylation Sites% Occupancy of Sites% Coupled
Monoclonal Rat IgG1150,0002variable72.93
Monoclonal Mouse IgG1150,0002variable84.74
Rabbit Serum IgG150,000210089.78
Polyclonal Chicken IgY170,000410090.62
Human Serum IgG150,000210097.13
Human Serum IgM970,0001010097.18

Table 1. Glycosylation properties of polyclonal antibodies (and ovalbumin) and efficiency of glycoprotein coupling to Thermo Scientific GlycoLink Resin. For each experiment, 0.4 mg of protein was reacted with 0.1 mL of resin

Learn more

Select products

Applications for hydrazide crosslinking

1. Using bis-hydrazide compounds

Commercially available homobifunctional hydrazide compounds (i.e., those which have a hydrazide group at each end) include carbohydrazide (MW 90.1) and adipic acid dihydrazide (ADH, MW 174.2). These bis-hydrazide compounds could be used in a single reaction to crosslink and polymerize prepared (aldehyde-containing) sugars or polysaccharides. However, they are more frequently used to modify and conjugate molecules in two stages, either homobifunctionally or heterobifunctionally.

For example, by reacting ADH in huge excess with periodate-oxidized dextran (polysaccharide), only one side of each individual ADH molecule will conjugate, resulting in hydrazide-activation of the dextran. After desalting to remove the excess, unreacted ADH, a small aldehyde-containing ligand could be added to conjugate at multiple locations onto the much larger dextran.

Because each hydrazide is also an amine, a bis-hydrazide compound can be used heterobifunctionally in a variety of ways to perform specialized conjugates. For example, ADH could be reacted in large excess with an RNA oligonucleotide whose ribose groups had been oxidized to create aldehydes. After desalting to remove excess ADH, EDC (Carbodiimide Reaction Chemistry) could be used to conjugate a carboxyl-containing label or carrier molecule to the amine-derivatized nucleic acid.

2. Prepare specific glycoprotein conjugates

Heterobifunctional hydrazide crosslinkers whose opposing end contains a sulfhydryl-reactive maleimide group are useful for conjugating glycoproteins to other proteins or molecules. See the page on Sulfhydryl-Reactive Crosslinker Chemistry for information on the maleimide group. Aldehyde-to-sulfhydryl crosslinkers are mentioned here, because maleimides are one of the few groups that can be paired opposite hydrazides in a single reagent. This is because the hydrazide group contains a primary amine and cannot be paired in a single molecule with an amine-reactive group, such as an NHS ester.

However, this limited reagent selection does not mean that applications for conjugation to carbohydrates are limited to proteins that have native sulfhydryl groups. It simply means that whatever macromolecule one wishes to attached to a carbohydrate must first be modified to contain a sulfhydryl group. Analogous to the bis-hydrazide scenarios described above, reagents exist to modify molecules to contain sulfhydryl groups. For example, Traut's Reagent (2-Iminothiolane, 2-IT) and SATA will add sulfhydryl groups onto primary amine sites.

When maleimide-hydrazide crosslinking is done sequentially, reaction to sulfhydryls is usually performed first followed by reaction to the molecule containing prepared aldehydes. The opposite sequence can be done, but initial steps must be performed quickly to prevent hydrolysis of the maleimide group.

The best crosslinking approach and strategy of modification to prepare respective molecular targets for conjugation depends on several factors. Although conditions can be adapted to use maleimide-hydrazide crosslinkers, many similar conjugation goals can be accomplished using different strategies. For example, see the discussion of reductive amination at the end of this page with respect to preparing antibody-HRP conjugates.