Back­stroke Swim­ming

Note. This is a lon­gread (1,871 words, 10 min read time), not count­ing the time you spend watch­ing the movies.

by Christoph

Pic­ture your­self as a sym­bi­otic bac­terium of an in­sect that has just been sucked up along with other food. You are de­ter­mined to swim through your host's in­tes­tine – no prob­lem, you've got fancy po­lar fla­gella – to reach the sym­bi­otic or­gan and es­tab­lish your­self there as an en­dosym­biont. The sym­biotic or­gan is lo­cated at the end of a duct about 200 µm long that branches off from the mid­gut. It's a good two hun­dred times as long as you are (with­out fla­gella), so a long-dis­tance swim is on the agenda. Oddly enough, the di­am­e­ter of this fluid-filled pas­sage is about the same as your waist­line, but it has zero flow, so you can't just let your­self drift. Freestyle swim­ming the bac­te­r­ial way is not an op­tion, since the geo­metry of the pas­sage pre­vents your fla­gella from gen­er­at­ing enough drag to dis­place the slightly vis­cous liq­uid ahead. So what can you do to not get stuck?

Ca­ballero­nia in­sec­ti­cola has fig­ured it out. Once it has en­tered the nar­row duct lead­ing to the sym­bi­otic or­gan via the sort­ing or­gan (CR) in the midgut of its host, the bean bug Rip­tor­tus pedestris, its swim­ming mo­tion al­ter­nates be­tween " freestyle" (coun­ter­clock­wise ro­ta­tion of the fla­gella) and "back­stroke" (clock-wise ro­ta­tion of the fla­gella). In this way, it cov­ers the ~200 µm dis­tance to the sym­bi­otic or­gan in just a few min­utes (Movie 1).

Movie 1  Fla­gel­lar dy­nam­ics of C. in­sec­ti­cola cells in the midgut of R. pedestris . The sec­ond in­star nymphs were fed a sus­pen-sion of sym­bi­otic bac­te­ria cells with flu­o­res­cently la­beled body and fla­gel­lar fil­a­ments. The sym­bi­otic or­gans of the nymphs were dis­sected 2 h af­ter feed­ing. The M4 re­gion of the duct [end­ing in the sym­bi­otic or­gan] is lo­cated at the up­per right of the movie. Yel­low ar­rows in­di­cate fla­gel­lar wrap­ping cells ("back­stroke swim­ming"). Area 26.0 μm × 19.5 μm. Timer sec­onds [s]. Source

When C. in­sec­ti­cola switches the ro­ta­tional di­rec­tion of its fla­gella from CCW to CW, the fla­gella first form wider coils that then quickly wrap around the cell body. This in­duces a screw-like mo­tion dur­ing "back­stroke" swim­ming (see di­a­gram). The cell can thus move in op­po­site di­rec­tions with­out hav­ing to ro­tate its body 180° each time, which comes in handy when it has to move – more squeeze than swim – through the very nar­row pas­sage lead­ing to the sym­bi­otic or­gan.

In the fol­low­ing Movie 2, you can clearly see the tran­si­tion be­tween "freestyle" (CCW) and "back­stroke" (CW) swim­ming, with the cells sus­pended in liq­uid medium briefly coil­ing their fla­gella in be­tween. In this com­pos­ite, three close rel­a­tives of C. in­sec­ti­cola show that they can also "do the "back­stroke" – but not B. an­thina; more on that be­low.

Movie 2  Cell be­hav­ior and fla­gel­lar dy­nam­ics of 5 species of Burk­holde­ria sensu lato group. C. in­sec­ti­cola,C. megalo­chro­mo­somata,B. an­thina, P. norim­ber­gen­sis, and P. ox­ala­tivo­rans cells were la­beled by flu­o­res­cent dye, sus­pended in the liq­uid medium con­tain­ing 0.5% MC and cap­tured with flu­o­res­cent mi­croscopy at 5‑ms in­ter­vals. Area 35.1 μm × 31.2 μm. Timer sec­onds [s]. Source

The ca­pac­ity for "back­stroke" swim­ming is not a pe­cu­liar­ity of some bac­te­ria from the Burk­holde­ri­a­ceae fam­ily of the Be­tapro­teobac­te­ria, It has also been found in (phy­lo­ge­net­i­cally) dis­tantly re­lated bac­te­ria. A non-ex­haus­tive list in­cludes She­wanella pu­tre­fa­ciens CN-32, Pseudomonas flu­o­rescens SBW25, Campy­lobac­ter je­juni, He­li­cobac­ter suis, Pseu­do­monas aerug­i­nosa, and Ali­ivib­rio fis­cheri. In the lat­ter, the abil­ity to squeeze through very nar­row chan­nels by "back­stroke" swim­ming prob­a­bly comes fully into play dur­ing the col­o­niza­tion of the crypts of the light or­gan of its host, the bob­tail squid E. scolopes.

The fla­gel­lar hook

The nag­ging ques­tion of why the fla­gella of C. in­sec­ti­cola wrap around the cell body in "back­stroke mode" (CW), whereas those of B. an­thina can­not, prompted Yosh­ioka et al. (2026) to de­velop a mech­a­nis­tic model. The key pa­ra­me­ters that were "played through" in ex­ten­sive nu­mer­i­cal cal­cu­la­tions in­cluded 1. the torque gen­er­ated by the fla­gel­lar mo­tor, 2. the flex­i­bil­ity of the fla­gel­lar fil­a­ment (FliC) re­quired for coil­ing, and 3. the hook (FlgE) bend­ing stiff­ness. Their model sug­gested that fla­gel­lar wrap­ping can be ex­plained by a sin­gle fac­tor, the stiff­ness of the hook in po­lar fla­gel­lated bac­te­ria (Movie 3).

Movie 3  Com­par­i­son of nu­mer­i­cal cal­cu­la­tion and real im­ages. Nu­mer­i­cal re­sults for rigid and flex­i­ble hook were pre­sented with the dy­nam­ics of fla­gel­lar fil­a­ments in C. in­sec­ti­cola and B. an­thina. The cells were la­beled by flu­o­res­cent dye, sus­pended in the liq­uid medium con­tain­ing 0.5% MC, and fla­gel­lar fil­a­ments were cap­tured with flu­o­res­cent mi­croscopy at 5‑ms in­ter­val for 0.5 s. Area of each real movie 7.80 μm × 5.85 μm. Source

To ex­per­i­men­tally eval­u­ate the re­sults of their model, the au­thors swapped the hook (FlgE) be­tween C. in­secticola and B. an­thina and con­ducted swim­ming com­pe­ti­tions. The " freestyle " swim­ming speed of both swap­ping mu­tants was al­most the same as that of their wild­type (Fig. 1, left), while the σ (fla­gel­lar ori­en­ta­tion an­gle) was mea­sured to be 0.011 for C. in­sec­ti­cola with FlgE_Ba and 0.051 for B. an­thina with FlgE_Ci (Fig. 1, cen­ter). They found C. in­sec­ti­cola with FlgE_Ba de­creased cell dis­place­ment in Q‑1D ("back­stroke swim­ming" in a nar­row chan­nel, see be­low) to al­most zero, whereas B. an­thina with FlgE_Ci in­creased its cell dis­place­ment in Q‑1D but did not reach the level of wild­type C. in­sec­ti­cola (Fig. 1, right). Taken to­gether, this con­firms the pre­dicted im­por­tance of the bend­abil­ity of the fla­gel­lar hook.

Fig­ure 1. Mod­er­ately flex­i­ble hook for fla­gel­lar wrap­ping. Char­ac­ter­i­za­tion of flgE swap­ping mu­tants. Left: Swim­ming speed for 1.0 s in growth medium. Cen­ter: Vari­ance of the fla­gel­lar ori­en­ta­tion an­gle σ. Right: Cell dis­place­ment for 1min inQ-1D. Box plots present the min­i­mum, max­i­mum, sam­ple me­dian, and the first and third quar­tiles. Schematic of the mea­sure­ments is pre­sented at the top of each graph. Dif­fer­ent let­ters in­di­cate sta­tis­ti­cally sig­nif­i­cant dif­fer­ences (two-sided Wilcoxon rank-sum test with Bon­fer­roni cor­rec­tion, p < 0.05; ex­act p val­ues and test sta­tis­tic are pro­vided in Source data). Source

An aside. To the best of my knowl­edge, this has not been ex­plic­itly stud­ied, but the hooks of the periplas­mic fla­gella of Bor­re­lia that wrap tightly around the in­ner mem­brane of the cell body also ap­pear to be highly bend­able, as can be seen in Fig. 2.1. in Bor­re­lia and it's not-so-loose ends (2|2).

Back­stroke swim­ming in a quasi-one-di­men­sional de­vice

Since study­ing the swim­ming be­hav­ior of C. in­sec­ti­cola in the nar­row chan­nel lead­ing to the sym­bi­otic or­gan of R. pedestris is pos­si­ble – as they have shown, see Movie 1 – but overly te­dious, Yosh­ioka et al. (2026) de­vel­oped a multi-chan­nel quasi-one-di­men­sional de­vice (Q‑1D) that mim­ics the sort­ing or­gan and has a chan­nel cross-sec­tion of 1×1 µm. The fol­low­ing Movies 4–6 fea­tur­ing sin­gle cells are clips ex­tracted from films that recorded an en­tire Q‑1D and were used by the au­thors for the sta­tis­ti­cal analy­sis of the be­hav­ior of ~100 cells in a given ex­periment (you can see an ex­am­ple in this YouTube clip).

In the Q‑1D, Sal­mo­nella en­ter­ica al­ways swam with its per­itri­c­hous fla­gella in such a way that the fla­gel­lar bun­dle trailed be­hind the cell body, re­gard­less of the di­rec­tion of move­ment (Movie 4). The au­thors re­gard it as a cur­rent tech­ni­cal lim­i­ta­tion of de­vice fab­ri­ca­tion that nanoscale gaps be­tween the Q‑1D and the glass sur­face, and be­tween the cell and the chan­nel wall, al­lowed par­tial pro­tru­sion of the fla­gel­lar fil­a­ments.

Movie 4 Dy­nam­ics of fla­gel­lar fil­a­ments of S. en­ter­ica in Q‑1D. Cells were la­beled by flu­o­res­cent dye, sus­pended in the buffer con­tain­ing 0.4% MC and cap­tured with flu­o­res­cent mi­croscopy at 10-ms in­ter­val. The movie con­sists of two parts: the first half shows only the flu­o­res­cence sig­nal, high­light­ing the fla­gel­lar dy­nam­ics dur­ing cell swim­ming along the chan­nel; the sec­ond half presents the same se­quence with the ma­genta over­lay, in­di­cat­ing the chan­nel bound­aries. Area 33.3 μm × 8.8 μm. Timer sec­onds. Source

Now com­pare the swim­ming of S. en­ter­ica (Movie 4) with that of C. in­sec­ti­cola in the Q1‑D (Movie 5). C. in­secticola quickly al­ter­nates be­tween "freestyle" (CCW) and "back­stroke" (CW). The back­stroke ap­pears to be more ef­fi­cient in re­gard to dis­tance cov­ered when­ever the cell, with its fla­gella wrapped around it, briefly stays in touch with one of the chan­nel walls of the Q1‑D (see di­a­gram for flow-field cal­cu­la­tions). This dif­fer­ence is also no­tice­able when com­par­ing "back­stroke" swim­ming ef­fi­ciency in liq­uid medium (Movie 1) and in the Q1‑D (Movie 6, up­per chan­nel).

Movie 5  Dy­nam­ics of fla­gel­lar fil­a­ments of C. in­sec­ti­cola in Q‑1D. Cells were la­beled by flu­o­res­cent dye, sus­pended in the buffer con­tain­ing 0.4% MC and cap­tured with flu­o­res­cent mi­croscopy at 20-ms in­ter­val. The movie con­sists of two parts: the first half shows only the flu­o­res­cence sig­nal, high­light­ing the fla­gel­lar dy­nam­ics dur­ing cell swim­ming along the chan­nel; the sec­ond half presents the same se­quence with the ma­genta over­lay, in­di­cat­ing the chan­nel bound­aries. Area 18.6 μm × 3.9 μm. Timer sec­onds. Source

I'll wrap up this movie-heavy post with an ex­am­ple that high­lights the value of an­a­lyz­ing the swim­ming be­hav­ior of bac­te­ria in mi­crochan­nels. In their Q‑1D, Yosh­ioka et al. (2026) held a swim­ming com­pe­ti­tion be­tween wild-type C. in­sec­ti­cola in the "up­per lane" in Movie 6, and ΔcheA mu­tant cells in the "lower lane". Both are pro­fi­cient swim­mers, but ΔcheA cells lack the abil­ity to switch be­tween CCW and CW ro­ta­tion of their fla­gella. Their fla­gella are locked in CCW ro­ta­tion, as briefly out­lined in Hey fla­gel­lum, shift into re­verse gear! As I said above, the nar­row pas­sage pre­vents the CCW-ro­tat­ing fla­gella of the ΔcheA mu­tant from gen­er­at­ing enough drag to dis­place the slightly vis­cous liq­uid ahead. The cell is vir­tu­ally stuck in place of the en­tire ~3 sec­onds de­spite its fla­gella ro­tat­ing wild­ly. In marked con­trast, the wild-type cell cov­ers the en­tire dis­tance of ~30 µm by "back­stroke" swim­ming (CW) in ~3 sec­onds, maybe with a brief CCW "hickup" at ~2.7 sec.

Movie 6  Com­par­i­son of the dy­nam­ics of fla­gel­lar fil­a­ments in Q‑1D. The cells of C. in­sec­ti­cola WT and ΔcheA mu­tant cells were la­beled by flu­o­res­cent dye, sus­pended in the liq­uid medium con­tain­ing 0.4% MC and cap­tured with flu­o­res­cent mi­croscopy. Area of each real movie 32.9 μm × 5.2 μm. Timer sec­onds. Source

Fi­nally… the choice of terms

What I re­fer to above as "back­stroke swim­ming", Yosh­ioka et al. (2026) call the "wrapped fla­gella mode", which they ex­plain: "...the wrapped fla­gel­lum scrapes the fluid in the gap like a corkscrew, cre­at­ing a lam­i­nar flow struc­ture in the nar­row tube and con­tribut­ing to cell propul­sion". The term "back­stroke" does not catch the ob­served dif­fer­ence of swim­ming ef­fi­ciency in liq­uid and in con­fined space. The term "wrapped fla­gella mode" doesn't ei­ther, since this swim­ming mode also ex­ists in liq­uid, while its use by Yosh­ioka et al. (2026) em­pha­sizes the cell be­hav­ior in con­fined space. One could rightly ar­gue that "back­stroke" or "wrapped fla­gel­lar mode" swim­ming in con­fined space is in fact a spe­cial form of glid­ing motil­ity, in which, ac­cord­ing to the gen­eral un­der­stand­ing, no fla­gella are in­volved. But they are in the case of C. in­sec­ti­cola. A dead end here.

To me, the way C. in­sec­ti­cola swims in the nar­row tube looks just like a tiny, self-pro­pelling Archi­medes' screw. I can't think of a short, catchy, suc­cinct term for that – maybe you can?