When dealing with plants in a lab, in a garden, or in an agricultural field, we often forget that there was originally a whole biome that those plants were involved in in the wild. We often try to simulate those conditions to the best of our capabilities, but there is still so much knowledge we are lacking on how certain interactions work that we are usually only obtaining a fraction of the possible results we could otherwise include. That’s one of the primary difficulties in plant research, how to better understand the symbiotic functions of the plant microbiome.
But, if we can determine what specific microbes or particular traits result in good or bad outcomes for plants, we can use that to synthetically manipulate how plants grow, all without touching the plants themselves. Add in biotechnological applications and we could even design particular microbes to cause the desired effect and type of growth for certain situations, soils, and growth parameters. To do so, however, requires taking that actual step of finding out what microbes cause what effects in the plant microbiome and that is no simple task. Measuring the impact of a single microbe or even species of microbe is largely impossible with current technology, but determining families of microbial life and their relative amounts in conjunction with each other is more feasible.
Establishing A System
Researchers at the University of North Carolina chose a single effect to study and how microbes interact with that plant growth system. They decided on the response of plants to phosphate starvation and how microbes can exacerbate the results or improve the plant’s growth despite the nutritional limitation. They chose to use bioinformatics involving phylogenies and a constructed neural network to identify microbes of interest. The first step was to separate the available bacteria into smaller communal chunks to experiment with. This was done by setting up association assays to grow the bacteria on and splitting them into groups or blocks of synthetically derived communities.
These forced together communal groups were then tested against multiple plants with different traits and transcriptomics was used on the plants to determine their transcribed genetic response to the microbes. The results of these were then used to create a neural network capable of predicting what other groups of microorganisms that were not tested would do if used on plants, essentially predicting further results through statistics and focused around a single plant phenotype, that of phosphate starvation. The effectiveness of this neural network was then challenged by making the actual communal groups it suggested and seeing if they did indeed have the impact on the plants that the network predicted.
The reason why phosphate starvation was chosen for this predictive analysis is because of it being a dual essential macronutrient for plants and microbes. It is a limiter on plant growth and only is found in certain quantities in the soil. Since microbes also take up the phosphate and have very efficient transport mechanisms for doing so as compared to plants, there is severe competition between the two for the resource. Therefore, it is a perfect mechanism for measuring plant transcriptomic and metabolic response, along with symbiotic (or parasitic) interactions with the soil microbiome.
A Back And Forth Microbiome
The first thing the researchers found was that, in situations of phosphate deficiency, plants both extend their root architecture and begin producing certain metabolites to exude in order to better explore for more phosphate rich soils. Due to this, the kinds of bacterial microbiomes they have to deal with may change over time. These effects were tested using the model organism Arabidopsis thaliana.
Once this plant response was measured, they tested whether the community of 440 bacterial strains found around the roots altered their growth in response to the roots growing as well. The bacteria were grouped into 10 clusters of strains that each had a different growth response to the plant root alteration. For some, the exuded metabolite substances from the roots boosted their growth and, for others, it inhibited it. Thus, it was shown that the exudates from the roots can modify the bacterial microbiome’s growth purposefully depending on the plant’s personal conditions.
A total of 183 of the strains were selected for more in-depth, direct testing on how they change the ability of plants to grow. The scientists made sure these strains spanned the 10 different response groups already found and also the ones that seemed to have the strongest response to phosphate levels and root exudates. The phosphate content in the plant shoots, a prime response to phosphate starvation, was used as the measurement of effect and was compared against control groups without microbes involved. The plants were also split into experimental groups with four different phosphate levels available to the seedlings while growing.
A Focused Effect
In general, the bacteria appeared to have a slight negative impact on the plants ability to gather phosphate, which isn’t entirely unexpected. This imbalance was made more extreme under stronger phosphate limitations, making the bacteria take up a larger share of the phosphate that was available. The most restricted experimental group saw a number of strains with high phosphate intake that had a severe negative effect on the plants.
Conversely, under the group with the most available phosphate, there were far more strains that formed that helped the plant’s ability to store phosphate in its shoots. Therefore, the bacterial strains that take prominence in a microbiome depends on the nutritional status and availability of macromolecules for the plants. Additionally, the effect of the bacteria on the plant growth was tied only to the status of the phosphate and there was no variance based on their personal ability to colonize the plant root or of the bacterial concentration in other areas of the plant.
The next question this raised is if the phosphate starvation response from the plant itself modulates the interactions with the bacteria and whether delaying it would change the strains that rise to prominence in the microbiome. To test this, the researchers pretreated Arabidopsis with phosphite, an analog of phosphate that can’t be metabolized by the plant, but still delays the onset of the starvation response. What they found is that germinating the plants, even on low phosphate medium, in a manner that had a delayed starvation response caused the number of bacterial strains with a negative effect on phosphate shoot accumulation to decrease.
However, in a similar opposite fashion, none of the strains had the positive impacts that was seen originally with the high phosphate availability, for example. This means that the plant starvation response does alter the types of bacterial strains that grow in the microbiome, but this is done completely independently of how much phosphate has been stored in the plant shoots. The researchers believe this modulation of the bacteria is likely controlled by the plant immune system, as a comparable mechanism had previously been described between Arabidopsis and a fungus.
Creating A Neural Model
After all of this testing, they went onto the synthetic block communities method based on the neural algorithm as discussed previously. What this taught them is that the abundance of bacteria in any community has little effect on the plant’s growth, but it is the variability of bacteria and the interactions between the bacteria in the microbiome that seem to alter the growth patterns. When added into the model, every case showed that bacteria abundance actually resulted in a decrease in the variability of the plant, meaning that the true impact abundance has is in decreasing phosphate concentrations in the plant shoot.
Running a transcriptomic analysis of the plants showed that phosphate starvation responses and their activation relied heavily on the composition of the synthetic community blocks. This was especially true based on the inclusion of so-called “negative blocks” of strains that had a negative impact on the plant’s ability to uptake phosphate. This was more likely to cause the starvation response. The plant genes that were up or down regulated were obviously those connected to such a response and also the plant immune system. The most common gene activation was for salicylic acid biosynthesis, a hormone important in many plant factors, including transport of ions like phosphates.
Three disparate bioinformatic models were used, two linear based ones and the neural network, to predict what sort of synthetic communal blocks would result in positive plant phosphate uptake. The models were largely brought into agreement on the blocks, though the neural network has higher sensitivity and could also determine the results in different phosphate availability amounts.
Using the network, 25 blocks were predicted to bring the most positive shoot phosphate intake based on the bacterial composition. 23 of the blocks succeeded in their goal, 1 had a neutral effect, and only 1 was found to be negative toward the goal. This prediction rate was within the desired parameters and allows for novel synthetic communities of bacteria to be generated for predicting the impact on a plant’s trait expression. Not just phosphate starvation traits either, but any other traits that interact with the microbial microbiome.
To A Better Understanding
With this method using synthetically combined bacterial communities and a predictive neural network, the researchers believe it can be applied to any number of studies looking into the soil microbiome and its influence on plant growth. Designed microbes with derived genes and expressions using biotechnology can also be included in the model for finding the optimal growth patterns and making plants that are able to resist stresses from their environment no matter the situation.
Photo CCs: A thaliana seedlings (cropped) from Wikimedia Commons
.jpg){kind=link}