It is widely accepted that most bacteria encounter low nutrient conditions during their life cycles and that adaptation strategies must be in place to survive those adverse conditions. Starvation-induced activities include differentiation into resistant forms that maintain viability in absence of nutrients
. Some of the resistant forms that bacteria can differentiate into include spores, ultramicrobacteria and viable but not culturable (VBNC) cells
. A common denominator in bacteria subjected to starvation is the ‘rounding up’ phenomenon by which cells become rounder, adopting a coccus shape morphology
. In addition, starved cells tend to show a reduction in size and therefore an increase in their surface-to-volume ratio, which may facilitate the uptake of substrates from a nutrient-poor environment. Our study showed that F. columnare develops a very unique cell configuration when subjected to starvation characterized by ring or coiled forms that, overtime, developed an envelope layer. Cells maintained their length but their overall shape changed from long and thin bacilli to round forms by curving over themselves. The strategy adopted by F. columnare did not increase the surface-to-volume ratio of the cell but reduced the surface exposed to the elements. The secretion of amorphous extracellular polysaccharides have been described in other Gram negative bacteria and data suggest they conferred protection against osmotic and oxidative stresses during starvation
. If the matrix that was observed around the F. columnare starved cells in the later stages was indeed secreted to provide protection against starvation or unfavorable environments then, the phenomenon of ‘coiling’ could be considered a starvation-induced activity since it would allow the cells to save energy by producing less of the protective envelope to cover themselves.
The presence of coiled or curved cells in old F. columnare cultures was first reported by Garnjobst
 in 1945 who described those cells as degenerative since the author failed to recover colonies after passing them onto fresh medium. Since then, the presence of spheroplasts or degenerative forms have been reported in several Flavobacterium species
 described how those cells, in their latter stages, were covered by a ‘veil of secreted slime’ that make the ‘coiled’ or ‘ring’ cells appeared as coccus-shaped cells. Her descriptions matched our observation precisely, both based on light-microscopy (see Additional file
1: Figure S2) and SEM (Figure
2) but our results showed that the ‘coiled’ forms are not degenerative but viable and culturable after at least one month of starvation. This was proven by comparing the growth curves between fresh and 1-month starved cultures in where no differences were observed. If starved cells were degenerative forms and observed growth was due to the few remaining bacilli observed then, a significant lag phase should be observed in cultures with a predominant population of coiled forms. The main difference between her study and ours is that, Garnjobst
 aged F. columnare cultures in high nutrient solid medium while we maintained our cultures in liquid and in absence of any organic nutrient. Excess of toxic metabolites and oxygen radicals in agar plates could explain the differences observed in culturability of aged F. columnare cells.
When starved cells were exposed to a different range of nutrients, their morphology transitioned from coiled forms to short bacilli. We failed to observe the cells ‘uncoiling’ but they morphed into noticeable smaller cells rather quickly. Cells exposed to nutrients produced numerous membrane vesicles that seem to be secreted into the medium thus reducing the overall volume of the cells. After this transition phase in where the cells reduce their volume and recovered their rod morphology, cells started to actively divide as confirmed by a parallel increase in cells numbers (SEM) and cell density values. Nutrients clearly reversed the structural changes induced during starvation. From our experiments, we conclude that F. columnare ‘coiled’ forms are viable but do not reproduce unless they revert back to the rod morphology.
Survival under long-term starvation conditions in freshwater has also been demonstrated in the close species F. psychrophilum[14, 25]. However, the morphological changes observed in F. psychrophilum during starvation were less dramatic than those observed in F. columnare. Few cells adopted a ‘ring-type’ structure but the main distinctive characteristic of starved F. psychrophilum cells was the formation of enlarged areas along the length of the cells or at one of the ends. SEM images of F. psychrophilum starved cells did not show the matrix layer covering the cells that we observed in F. columnare. Nevertheless, ultrastructural similarities were found between these two species. Surface blebbing and membrane vesicle formation was observed in fresh cultures of F. columnare and during the revival process of starved cells similar to those reported in F. psychrophilum. Although the role of bleb formation and release of membrane vesicles is not clear, it has been postulated they may play a role in host-pathogen interaction due to the high content of antigenic proteins present in F. psychrophilum membrane vesicles. Further studies on the role that these ultrastructures may play in F. columnare pathogenesis are needed. The typical capsule described for F. columnare and F. psychrophilum was missing from our TEM images probably due to different sample preparation methods. It is likely that during sample preparation for TEM, the capsule or mucus layer observed by SEM was removed since we did not use a capsule stabilization protocol.
Differences in cell culturability were observed between strains although those could not be correlated with their genetic group. The strains used in this study were choosen based on their genotype and source of isolation
. Strains ARS-1, ALG-00-530 and AL-02-36 behaved similarly throughout the experiment and the numbers of coiled forms at 14 days were statistically identical. The initial length of the cells seemed not to influence the coiling process since both the shortest (ARS-1) and the longest (ALG-02-36) strains behaved similarly. In the type strain ATCC 23643, coiled cells were covered by a matrix layer that made difficult to observe the cell structure in detail. SEM observations of starved ATCC 23643 cells resembled those described in starved Vibrio cholerae cells by Chaiyanan et al.
 in where V. cholerae cells had remained viable for a 2-year period. The appearance of coiled cells covered by a matrix was also observed in strain ALG-00-530 after 5 months in ultrapure water. Cells were connected by what appeared to be fimbriae, a characteristic structure that has also been reported in other long-term starvation studies
[13, 27, 28]. Our results showed that strains of F. columnare followed a similar strategy to survive under lack on nutrients by adopting a coiled conformation and secreting a matrix layer, although this process occurred faster in some strains.
Under starvation conditions and in absence of organic nutrients, F. columnare can survive for at least 5 months at ambient temperature in sterile water. In a previous study
, the authors proposed that F. columnare survived the nutrient-deprived conditions by utilizing nutrients released from dead cells that allowed cultures to maintain constant growth over time. Our results contradict this assumption because in all our microscopic observations we failed to detect any cells undergoing cell division although we did note some lysed cells in our cell preparations that likely released nutrients into the medium. Based on our data, and at 5 months under starvation, more than 99% of the F. columnare cells underwent a dramatic change in morphology and cell structure into what can only be considered dormant or resistant forms. This behavior is typical of copiotrophic bacteria that can survive under oligotrophic conditions but without active reproduction
. Moreover, 3-month old F. columnare cells were not able to outcompete with young cells when provided with nutrients which indicates F. columnare lose fitness overtime when subjected to starvation conditions.
The new observations presented in this study demonstrate a unique state in the F. columnare life cycle induced by starvation. This state (coiled form) should not be regarded as degenerative but an active adaptation to lack of nutrients allowing F. columnare to remain viable in water, in absence of organic matter, and even without salts for an extended period of time. This bacterium is likely to encounter starvation conditions after nutrients provided by the host are exhausted and bacterial cells are released back into the water column. This stage in the life cycle of F. columnare indicates that water can act as reservoir and served as dispersant mechanism for this pathogen. However, F. columnare should not be considered a facultative oligotroph since no cell replication was observed under very limited nutrient content (originated from lysed cells) suggesting that water is a transient environment for this bacterium. Furthermore, starved cells failed to infect channel catfish thus low organic waters should not be considered the primary reservoir for this pathogen. The notion that F. columnare may have a restrictive ecological niche is supported by the recently published complete genome of F. columnare that predicts a lifestyle in close association with its host
. However, further studies on the biology of F. columnare are required to fully understand its life cycle.