Maize seed endophytes

Abstract Maize is a vital global crop, and each seed (kernel) hosts an ecosystem of microbes living inside it. However, we know very little about these endophytes and what their role is in plant production and physiology. In this Microreview, I summarize the major questions around maize seed endophytes, including what organisms are present, how they get there, whether and how they transmit across generations, and how they and the plant affect each other. Although several studies touch on each of these areas, ultimately there are far more questions than answers. Future priorities for research on maize seed endophytes should include understanding what adaptations allow microbes to be seed endophytes, how the host genetics and the environment affect these communities, and how maize seed endophytes ultimately contribute to the next generation of plants.

host, such as nutrient solubilization in cacti (Puente et al., 2009), stress tolerance in cool-season grasses (Clay & Schardl, 2002), and hormone and secondary metabolite production across many plants (reviewed in White et al., 2019).
The divisions between beneficial, commensal, and harmful microbes are fuzzy, and some microbes can shift between roles at different points (e.g., Degani et al., 2021;Kloepper et al., 2013;Rai & Agarkar, 2016;Zhou et al., 2018). For this Microreview, I use the broad definition that an endophyte is any organism that colonizes internal plant tissue for at least part of its lifetime (Hardoim et al., 2015). This thus includes beneficial and commensal organisms along with mild or latent pathogens; although this definition technically includes severe pathogens, this review will not focus on them because of the extensive literature already available (e.g., Degani, 2021;Goko et al., 2021;McGee, 1988). Organisms on the seed surface also fall outside this review simply because they are not endophytes.

| WHAT MI CROB E S ARE IN MAIZE S EEDS?
Many different microbes can grow in seed, and the maize seed community is distinct from other parts of the plant (Johnston-Monje & Raizada, 2011). Only a handful of studies have looked at the maize seed endophyte community with modern deep sequencing methods (Johnston-Monje et al., 2021;Liu et al., 2017Liu et al., , 2020Majumdar et al., 2021;Santos et al., 2021), so most of our knowledge comes from isolation studies. However, based on both sequencing ( Figure 1) and isolation (Tables 1 and 2), the bacteria of the maize seed are dominated by Proteobacteria, Actinobacteria, and Firmicutes, with the majority of fungi placed in the Ascomycota phylum. Archaea appear to be rare in seeds (e.g., only 0.04% of reads in Johnston-Monje et al., 2021), and there appear to be no reports finding protists or nonpathogenic viruses in them.
Although Proteobacteria make up the majority of the community by sequencing, Firmicutes, particularly in the genus Bacillus, seem especially common among isolates (Bodhankar et al., 2017;Bomfim et al., 2020;Gond et al., 2015;Mundt & Hinkle, 1976;Pal et al., 2021;Rijavec et al., 2007;Yang et al., 2020). The reasons for this difference are unknown, but may have to do with many Firmicutes' ability to form spores for long-term survival. Pantoea species are also common (Johnston-Monje et al., 2014;Liu et al., 2013Liu et al., , 2017Majumdar et al., 2021), and the only time course available of maize seed endophytes across development showed that both Pantoea and Burkholderia came to dominate the community as the seeds matured . This study also showed a decrease in seed endophyte diversity over time, which could be due to either actual loss of some endophytes or to a small number of them growing to dominate the community. Further experiments are needed to determine which of these is the case.
In addition to species, it would be useful to know the number of actual microbial cells living in seeds. Unfortunately, quantifying microbes in seeds is surprisingly hard. Isolation methods suffer from the same selectivity bias as they do elsewhere because only a subset of microbes will actually grow in the laboratory. Many isolation efforts get zero colonies from some seeds (Fisher et al., 1992;Marag & Suman, 2018;Mundt & Hinkle, 1976;Rijavec et al., 2007), although they often have better success after imbibing them for 24-48 h Foley, 1962;Rijavec et al., 2007). This implies that imbibition either brings the microbes out of a dormant state or results in a large increase in their numbers, or (most likely) some combination of the two. (Readers interested in isolation methods are referred to the extensive review of Chowdhury et al., 2019.) So, the exact number of microbes in a seed is still unknown, although we can probably assume it is small. F I G U R E 1 Distribution of seed microbes based on community sequencing. Community sequencing data was downloaded from the National Center for Biotechnology Information Sequence Read Archive accessions PRJNA510484 (Liu et al., 2020) and PRJNA731997 (Johnston-Monje et al., 2021), and taxonomically classified with Kraken2 (Wood et al., 2019) and Bracken (Lu et al., 2017). Each plot shows the relative fraction of the community based on bacterial 16S ribosomal RNA (a, b) or fungal internal transcribed spacer (c) amplicon sequencing; part (c) excludes bacterial reads (c.56%). Labels indicate taxa present at >1% of the total dataset, with phyla/divisions on the left and genera on the right

| WHERE ARE S EED ENDOPHY TE S?
Endophytes have been localized to all the interior regions of the seed, including the pedicel, abscission layer, endosperm, radicle, and embryo (Fisher et al., 1992;Mitter et al., 2017).
Because seed-inherited microbes are difficult to track, little is known about where they end up in the mature plant. Transmission across multiple generations (see below) implies they must at least be in the stalk and aboveground vasculature. Electron microscopy of sterile, germinating seeds shows microbes in the pericarp, endosperm, radicle, and germinating root surface (Santos et al., 2021), and the beneficial microbe B. mojavensis RCC101 was seen to cover the root epidermis after germination (Bacon et al., 2001). The presence of endophytes on the root surface implies that microbes can move

| HOW DO ENDOPHY TE S G E T INTO THE S EED?
Endophytes have several potential routes to get into seeds ( Figure 2) (reviewed in Rodríguez et al., 2018;Shade et al., 2017 Frank et al., 2017) indicates that microbes can also associate with pollen and thus potentially come from the male parent as well. Whether this actually occurs in maize has not been shown, although there are some hints that it may (e.g., Wu et al., 2022).
Horizontal transmission is probably most common through the silks, as these represent a direct route from the environment to the developing seed. It has been suggested that the silk microbiome could be selected to control this route (Khalaf et al., 2021) because it is a favoured entry for fungal pathogens. Silks have even been used to artificially introduce endophytes into seeds (Mitter et al., 2017), and the ease and high success rate (90% in wheat, not reported in maize) implies it is probably common in nature. Horizontal transmission could also occur via wounds (such as from insects), although the author could find no studies on this for nonpathogenic endophytes.

| HOW WELL DO S EED ENDOPHY TE S TR AN S MIT ACROSS G ENER ATI ON S?
Traditionally, it was thought that seeds represented a minor source of microbes for the new plant, with most microbes (endophytes and otherwise) coming from the soil (McInroy & Kloepper, 1995). These results were based on isolating live microbes from seeds and plants.
More recent work with high-throughput sequencing challenges this view, however, with evidence that a large fraction of a plant's microbiota can be inherited via seed (Johnston-Monje et al., 2016, 2021. Because seed endophytes are arguably the best way maize could pass a microbiome on to progeny, the natural question is, . These results match work in rice, where only 45% of microbes from the parent generation were also found in the progeny (Hardoim et al., 2012). In some cases, this reduction could be due to loss of endophytes during storage, as endophytes are known to lose viability faster than the seed itself (Bacon & Hinton, 2019;Mitter et al., 2017;Mundt & Hinkle, 1976).
Despite this, some endophytes do seem capable of faithful transmission. For example, GUS-labelled Fusarium showed faithful transmission through three generations of maize (Bacon & Hinton, 2019).
Similarly, when GUS-labelled Paraburkholderia phytofirmans PsJN was introduced via silks, labelled bacteria were found in both infected seeds and the subsequent plant (Mitter et al., 2017). Ideally, we would also want to see transmission from that plant into its own seeds, but this was not tested.
Another reason for low seed transmission is that getting into seeds could be a stochastic process, and a given endophyte may only make it into a subset of seeds. Unfortunately, no analysis of the general seed microbiome and how it varies within a maize ear is available. A recent study on common bean (Phaseolus vulgaris) found relatively consistent transmission within a plant, with plant-to-plant variability being much higher than seed-to-seed within the same plant (Bintarti et al., 2022). This implies that maize seeds may be fairly consistent within a plant, although actual studies are needed to confirm this.
Different studies have come to conflicting conclusions as to how important the maize seed microbiome is for the next generation. Early studies found that surface-sterilized seeds grown on water agar had two to four orders of magnitude fewer microbes than the same seeds grown in soil, implying that soil was the major source of plant microbes (McInroy & Kloepper, 1995). More recent studies with sequencingbased approaches found that seeds had a larger effect on plant mi- It found that most microbes were shared between these conditions, implying they came via seeds. Because these seeds were not surfacesterilized, it is unknown how many of the microbes were endophytes versus living on the seed surface. Combining these results with the low transmission rate of many endophytes (see above) creates a paradox: we cannot predict which microbes make it into a seed, but those that do appear to have a significant impact on the seedling microbiome.
Resolving this apparent contradiction will require specific research on both how endophytes get into the seed and how they affect the resulting plant.

| HOW WELL C AN WE MANIPUL ATE MAIZE S EED ENDOPHY TE S?
A key goal for using endophytes in agriculture is that we want some way to manipulate them. Generally, the seed endophyte community is manipulated by removing existing endophytes, adding new ones, and/or modifying the endophytes themselves.
Removing existing endophytes is deceptively tricky. Surfacesterilization-usually via bleach and ethanol-can remove microbes on the exterior, but disinfecting the seed interior while keeping it alive is hard. The main methods are hot-water baths (Bacon et al., 1994;Daniels, 1983) or antibiotic treatments (Pal et al., 2021).
To our knowledge, no one has shown that either of these procedures fully removes the endophyte population, only that they significantly reduce it. Although various groups describe work with axenic (germfree) maize (e.g., Groleau-Renaud et al., 1998;Hussain et al., 2013;Niu et al., 2017;Shaharoona et al., 2006), this usually refers to surface-sterilization and sterile growing conditions; the presence of seed-transmitted endophytes is rarely, if ever, checked for.
Putting endophytes into seeds, on the other hand, is rarely attempted. Although there are many examples in the literature of people inoculating maize seeds with microbes (e.g., Bano et al., 2013;Casanovas et al., 2002;Oliveira et al., 2017;Viruel et al., 2014), the vast majority of these are not actually trying to get microbes inside the seed itself. Instead, the goal is to put inoculum ( for example, Mousa et al. (2015) only succeeded in putting GFPuv in one of four endophytes they worked on. When transformation fails, isolating natural rifampicin mutants (e.g., Bodhankar et al., 2017) may work instead, at least for bacteria, because they can be identified by plating onto selective media. In all of these cases, the goal of manipulation is to track the microbes in planta; the author is not aware of any cases where seed endophytes have been altered to test gene function or plant/seed associations. Such experiments will eventually be necessary to understand how seed endophytes work with and within the plant.

| HOW DO S EED ENDOPHY TE S AFFEC T THE MAIZE PL ANT ?
Endophytes in general are often studied for their growth-promoting capacities, and seed endophytes are no exception. Seed-transmitted endophytes can affect the plant either directly or indirectly through their interactions with other organisms.
Direct interactions include endophyte functions that are traditionally associated with plant-growth-promoting activity, such as phosphate solubilization, nitrogen fixation, siderophore production, and plant hormone synthesis (Gold et al., 2014;Gond et al., 2015;Pal et al., 2021;Ravichandran et al., 2021;Sandhya et al., 2017;Siddique et al., 2022). The presence of these genes or their activity in vitro does not, unfortunately, guarantee they actually increase plant growth in vivo. For example, one survey of seed endophytes found that most of the isolates with supposed growth-promoting activities actually decreased the growth of plants (Johnston-Monje & Raizada, 2011). A similar result was found by Bomfim et al. (2020), where 11 out of 51 seed isolates actually reduced germination and root growth. Sometimes the relationship is more complex, such as how a symptomless Fusarium infection was shown to reduce seedling growth after 7 days, but by 28 days the infected plants had recovered or even surpassed uninfected ones (Yates et al., 1997).
In contrast to these direct interactions between seed endophytes and the plant, many of their impacts occur indirectly through their interactions with other microbes. Much of the focus in this area is directed toward controlling pathogens (Chulze et al., 2015;Degani et al., 2021;Liu et al., 2016;Rijavec et al., 2007;Wicklow et al., 2005). Many of these interactions are tested in vitro, however, so it is usually not known if they actually impact the disease in realistic conditions. An exception found that endophytes antagonistic to Fusarium and other fungal pathogens did result in a significant yield increase when disease pressure was high, and that they also reduced mycotoxins during storage through unknown mechanisms (Mousa et al., 2015). Interestingly, the most consistent of these endophytes all came from wild teosintes, potentially supporting the hypothesis that domestication and breeding reduced maize's dependence on beneficial microbes (Berg & Raaijmakers, 2018;Pérez-Jaramillo et al., 2016;Soldan et al., 2021). Another study isolated 11 endophytes for their ability to antagonize late wilt (Magnaporthiopsis maydis) in vitro, two of which also improved performance in greenhouse trials .
The mechanisms of how seed endophytes interact with other microbes varies. Fusarium, for example, produces fusaric acid, which among other things interferes with quorum sensing in bacterial competitors (Bacon & Hinton, 2019). In the other direction, several Bacillus species inhibit Fusarium by producing lipopeptides (Gond et al., 2015;Yang et al., 2020) The authors also found that seeds treated with bacteria-specific antibiotics showed fungal growth during germination, while untreated ones did not, implying at least some bacteria in the seeds helped keep fungal pathogens under control (Pal et al., 2021). This antagonism may also explain why bacteria and fungi have different distributions in grown plants, so that there are more bacteria closer to the soil and more fungi further from it (Fisher et al., 1992;Hallmann et al., 1997), although to our knowledge this remains speculative.

| HOW DOE S MAIZE G ENE TI C S AFFEC T S EED ENDOPHY TE S?
The degree to which maize genetics affects seed endophytes-or  Liu et al., 2017Liu et al., , 2020. Despite these shortcomings, we can draw some conclusions.
First, there is a small amount of evidence that the pollen parent can affect microbes in the seed (Liu et al., 2017). This could occur through typical xenia effects (where the genotype of the seed changes its phenotype) or from microbes hitchhiking with the pollen, although in either case confirmation is needed from additional studies.
In addition, several studies have found differences in seed endophytes between domestic maize and its ancestor teosinte, as shown generally by Johnston-Monje and Raizada (2011). Desjardins et al. (2000) found a much lower rate of infection of symptomless Fusarium verticillioides infection in Zea parviglumis (4%) relative to modern maize (100%), although at least some of that could be due to the physical differences between them (e.g., presence of a protective fruitcase) (Bacon & Hinton, 2019). There is some speculation that beneficial endophytes could have been lost over the course of domestication (Berg & Raaijmakers, 2018;Pérez-Jaramillo et al., 2016;Soldan et al., 2021). Although the author is unaware of any systematic tests of this in maize, Mousa et al. (2015) did find that their three best biocontrol endophytes against Fusarium graminearum all came from teosintes. Taken together, these results imply that domestication altered at least some aspects of how maize interacts with seed endophytes, although the mechanisms and consequences are still mostly unknown.

| CON CLUS I ON S AND FUTURE DIREC TIONS
From the above information, we can conclude that maize endophytes are probably ubiquitous, and that we still know very little about them and their roles in the plant. There are far more questions than answers in this area, with major knowledge gaps in every area. Question 3: What is the effect of seed endophytes on the next generation of plants? This is the most practical question, as it lets us manipulate plant performance via seed endophytes. Growth promotion, stress tolerance, and disease resistance are the goals, but we also need to be testing for complex interactions with other microbes, nutrient utilization, seed viability, and the like. Ultimately, this is the question of how seed endophytes affect maize, and how we can use that to our advantage.
Answering these questions will take a significant amount of work, and will probably spawn yet more questions about these communities.
Yet given the importance of maize to global agriculture and the sheer amount of seeds (and thus seed endophytes) that we produce every year, answering these questions could open new avenues to more reliable and sustainable maize production.

ACK N OWLED G EM ENTS
This work was supported by the University of Georgia and the Foundation for Food and Agriculture Research New Innovator Award.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data sharing is not applicable to this article as no new data were created or analysed.