Migration of aerobic bacteria from the duodenum to the pancreas with tumors: a mechanistic understanding

More aerobic bacteria are found in the pancreas with tumors than in the healthy pancreas. We 17 provide a mechanistic understanding of the migration of intestinal bacteria from the duodenum to 18 the pancreas with tumors. Mathematical models of migration of aerobic bacteria from the 19 duodenum to the pancreas with tumors in the hepatopancreatic duct were developed. In addition, 20 the behaviors of GFP E. coli under a pH gradient in a microfluidic device were analyzed. Moreover, 21 upstream migrations of Pseudomonas fluorescens against flow were measured in a polydimethylsiloxane (PDMS) T-shaped cylinder mimicking a pancreatic duct. The simulated 23 bacterial concentration of the pancreas with tumors was higher than that of the healthy pancreas 24 and agreed reasonably well with the literature. Migration of aerobic bacteria in the hepatopancreatic 25 duct is counteracted by bile and pancreatic juice flow but facilitated greatly by bacterial pH taxis 26 from lower pH in duodenum fluid toward slightly alkaline pH in pancreatic juice, favorable for them. 27 Migration of bacteria to the pancreas with tumors is made easier by solid tumors on the pancreatic 28 duct, which compresses the pancreatic duct and thus reduces the fluid flow rate. On the other hand, 29 GFP E. coli migrated under the pH gradient in a microfluidic device from acidic areas toward neutral 30 or slightly alkaline pH, validating pH taxis. Furthermore, Pseudomonas fluorescens migrated 31 upstream from hydrochloride solution but not from bicarbonate solution against bicarbonate flow at 32 >20 μm/s, with an advancing velocity of app roximately 60 μm/s, validating the models ( 244 words).


103
The migration of aerobic bacteria in the hepatopancreatic duct from the duodenum into the 104 pancreas was simulated (Figures 4, 7, 1S). Factors that influence bacterial transport are summarized in Table 1. The simulated bacterial concentration in the healthy pancreas (figure 4 106 blue) was lower than that in the literature (figure 4 orange)(12). However, the bacterial amount 107 estimated using the typical weight of the pancreas at 80 g at 3.2 CFU seems consistent with the 108 literature that 15% of healthy pancreas contained detectable bacteria (10). Bacteria did not migrate 109 into the pancreatic duct due to motility alone, even at the periphery of duct, where fluid flow velocity 110 is lower ( Figure S1b black dotted). However, bacterial pH-taxis under the pH-gradient at the T-         The measured pH-tactic velocity in a T-shaped cylinder is over 50 μm/s (figure 6b, S5b), much 184 faster than the typical chemotactic velocity at 10 μm/s. This may be due to the following reasons.

185
First, the gradient under flow is made greater since the flow inhibits diffusion (figure 3 green).

186
Second, an increase in pH leads to exponential decreases in hydrogen ion concentration. Thus,    (Table S1), which do not show motility or pH-taxis, or migrate into the pancreas even in the reduced flow ( Figure 7). Thus, the latter route is not neglected. Moreover, the intestinal barrier in patients 212 with obstructive jaundice is impaired, which is frequently accompanied by pancreatic cancer(36); 213 thus, bacterial translocation via the bloodstream is promoted (42). On the other hand, bacterial 214 colonization in the pancreas was not detected in a mouse model with defective intestinal 215 permeability with increased permeability by Campylobacter infection (5). However, Pseudomonas 216 putida, which is motile and highly aerobic, was the most common strain in pancreatic tumors (10) 217 (Table S1) Table 1. The details of the modeling follow. The parameter list is 266 provided in Table S2.
aerotaxis to oxygen aerotaxis away from carbon dioxide motility 1 1 1              Oxygen transport is not included in the healthy pancreas, assuming no oxygen concentration 444 difference between healthy pancreas and duodenum, but was included for transport in the

504
This PDMS mixture was cured at 75°C for two hours as a basis for the cylinder. Then, glass tubes 505 were placed in T-shaped tubes, and another PDMS mixture was poured there ( Figure S15b, c).

506
The tubes were then removed carefully by incising with a cutter, leaving a hollow T-shaped 507 cylinder (figure S15d). End tips of the hollowed cylinders were filled with remaining cured PDMS 508 so that the PDMS that would be poured later would not be filled in. Finally, the PDMS mixture was 509 poured into the whole device and cured (figure S15e).

516
The pH distribution was measured by bromocresol purple (FujifilmWako, Japan). Bacteria were 517 measured in the same manner as Sec. 5.1, but movies were taken using a CMOS image sensor 518 (IMX586, Sony, Japan). The obtained movies were analyzed using MATLAB 2021 (MathWorks,

519
Japan), as shown in Figure S16. Horizontal distance in millimeters was calculated from a ruler in 520 an image placed near the device.