The Rab11 effectors Fip5 and Fip1 regulate zebrafish intestinal development

The Rab11 apical recycling endosome pathway is a well-established regulator of polarity and lumen formation; however, Rab11-vesicular trafficking also directs a diverse array of other cellular processes, raising the question of how Rab11 vesicles achieve specificity in space, time, and content of cargo delivery. In part, this specificity is achieved through effector proteins, yet the role of Rab11 effector proteins in vivo remains vague. Here, we use CRISPR/Cas9 gene editing to study the role of the Rab11 effector Fip5 during zebrafish intestinal development. Zebrafish contain two paralogous genes, fip5a and fip5b, that are orthologs of human FIP5. We find that fip5a and fip5b mutant fish show phenotypes characteristic of microvillus inclusion disease, including microvilli defects, inclusion bodies, and lysosomal accumulation. Single and double mutant analysis suggest that fip5a and fip5b function in parallel and regulate apical trafficking pathways required for assembly of keratin at the terminal web. Remarkably, in some genetic backgrounds, the absence of Fip5 triggers protein upregulation of a closely related family member, Fip1. This compensation mechanism occurs both during zebrafish intestinal development and in tissue culture models of lumenogenesis. In conclusion, our data implicate the Rab11 effectors Fip5 and Fip1 in a trafficking pathway required for apical microvilli formation.


29
Development of many organs, such as the gastrointestinal system, kidneys, and respiratory tract 30 requires morphogenetic remodeling of cells to form a hollow tube, or lumen (Jewett and Prekeris, 2018).

31
Whereas the mechanisms cells use to form a lumen vary by organ, a common feature is that cells adopt 32 a highly polarized conformation including establishment of apical structures such as primary cilia, motile 33 cilia, or microvilli (Apodaca and Gallo, 2013). Intestinal epithelia are one of the few vertebrate cell types 34 to lack primary cilia, but their apical cell surface is covered with a brush border composed of actin-rich 35 membrane protrusions called microvilli to aid in nutrient absorption (Apodaca and Gallo, 2013). The 36 molecular basis of cell polarization is well defined, but much less is understood about how trafficking 37 pathways govern formation of these apical structures, especially in vivo.

38
The Rab11 apical recycling endosome pathway is a well-established regulator of polarity and

53
Polarization is critical for cell function such that polarity disruption results in a number of diseases.

54
Microvillus Inclusion Disease (MVID) is one such example, arising from the inability to form and maintain microvilli at the apical cell surface (Al-Daraji et al., 2010). Patients with MVID suffer from intractable 56 diarrhea and malabsorption due to absent or very sparse microvilli and typically do not live past 57 childhood. At the cellular level, patients display characteristic trafficking defects of lysosome 58 accumulation and inclusion bodies containing microvilli (Phillips et al., 1985, Phillips andSchmitz, 1992, 59 Ruemmele et al., 2006). Mutations in MYO5B are found in patients with MVID and mutations in the

83
Luminal organs such as the intestine, spinal cord, and notochord expressed fip5a and fip5b ( Figure S2A, 84 B). In measuring mRNA levels of fip5a and fip5b, we found that both transcripts showed increased levels 85 around 3 dpf, and high levels of fip5b mRNA persisted throughout 8 dpf ( Figure S2C). We thus focused 86 our efforts first on fip5b.

88
Endosome maturation and terminal web keratin organization require Fip5b function

89
To study the function of Fip5b, we used CRISPR/Cas9 gene editing. We selected two different 90 fip5b alleles that introduced a premature stop codon right after the C2 domain at the N-terminus ( Figure   91 1A, Figure S1B), thereby eliminating the Rab-binding domain (RBD) at the C-terminus essential for Fip5 92 function. We maintained these fip5b mutant stocks in a heterozygous state and performed intercrosses 93 to generate zygotic mutants for analysis. Stage matched wild-type siblings were used as controls. We 94 performed qRT-PCR to measure fip5b expression in fip5b CO40 homozygous mutant larvae and observed 95 an almost complete loss of fip5b mRNA levels ( Figure 1B), suggestive of nonsense-mediated decay.

107
Instead, intestinal cells of homozygous mutant larvae showed an accumulation of small (less than 108 500nm) apical vesicles ( Figure 1G, H) and large (greater than 500nm) organelles that resided medially in the cells (Figure 1G arrows, I) compared to wild-type cells which did not show an accumulation of 110 intracellular vesicles. Moreover, microvilli were shorter in both the anterior intestinal bulb and posterior 111 midgut of 6 dpf homozygous mutant fish compared to wild-type siblings ( Figure 1G, J). Finally, the 112 terminal web, an apical cytoskeletal network anchoring microvilli into the cell, was disrupted in mutant 113 fish. Wild-type larvae had a defined electron dense line at the base of the microvilli and an organelle-free 114 zone just below the apical cell surface which was absent in mutants ( Figure 1G, brackets). These data 115 revealed trafficking and microvilli defects in fip5b CO40 mutant larvae.

116
To investigate the identity of the large organelles observed in fip5b CO40 mutant larvae intestinal 117 cells, we performed immunohistochemistry to detect proteins that serve as common endosome markers.

119
Rab11 vesicles localized just beneath the apical cell surface, as revealed by actin staining ( Figure 1K). In  Figure 1L). These Rab7 endosomes were consistent in size and localization with the structures revealed 127 by electron microscopy ( Figure 1M). Taken together, these data suggested that Fip5b is required for network resided just below the apical actin network; however, in fip5b CO40 mutant cells, keratin 137 mislocalized to lateral and cytoplasmic regions of the cell ( Figure 1N

140
Terminal web defects result in microvilli abnormalities, which can be exacerbated by physical 141 stress from intestinal activity. We therefore hypothesized that fed mutant larvae would show more severe 142 microvilli phenotypes than unfed 6 dpf larvae still living off the yolk. To test this, we began feeding the 143 larvae daily at 7 dpf and then analyzed larvae at 11 dpf. Mutant larvae showed moderate trafficking 144 defects at 11 dpf ( Figure 1P, arrows, Q, R); however, the terminal web defects recovered, and microvilli 145 were now significantly longer than wild-type siblings ( Figure 1P, bracket, S). This phenotypic recovery 146 was unexpected and perhaps explains in part why adult mutant fish were homozygous viable.

147
Importantly, these trafficking and microvilli phenotypes were recapitulated in another fip5b mutant allele  Whereas fip5b mutant phenotypes were prominent during early developmental stages, these 154 mutant fish recovered from these defects and were viable as homozygous adults. One possible 155 explanation is a compensatory mechanism, perhaps through upregulation of another trafficking pathway, 156 and an obvious candidate for compensation is the zebrafish fip5b paralog, fip5a. To test Fip5a's role in 157 intestinal development, we again used CRISPR to create fip5a mutant alleles (Figure 2A, Figure S1C).

158
fip5a mutant stocks were maintained in a heterozygous state and intercrossed to generate zygotic 159 mutants for analysis. Stage matched wild-type siblings were used as controls. Similar to fip5b mutants, 160 fip5a CO38 homozygous mutant larvae were morphologically normal and viable as homozygous adults. To 161 study the role of fip5a during intestinal development, we performed the same transmission electron 162 microscopy analysis on fixed sections through the mid-intestinal region. Notably, fip5a CO38 mutant fish recapitulated phenotypes seen in fip5b mutant fish. At 3 dpf, fip5a CO38 mutant larvae formed a lumen, but 164 exhibited subapical organelles resembling inclusion bodies ( Figure 1B, C). By 6 dpf, inclusion bodies 165 cleared, and fip5a CO38 mutant cells now accumulated small apical vesicles ( Figure 1D, E) and large 166 organelles ( Figure 1D arrows, F) not present in wild-type larvae. Additionally, midgut microvilli were 167 shorter ( Figure 1D, G) and the terminal web was also disrupted in mutants compared to wild-type larvae 168 ( Figure 1D, brackets). These large organelles were Rab7-positive in fip5a CO38 mutant fish and terminal 169 web defects appeared to be the result of mislocalized keratin from the apical cell surface ( Figure 2H-K).

219
We thus used a MDCK tissue culture model of lumenogenesis to ask if FIP1 could compensate for FIP5.

220
When MDCK cells were grown in an extracellular matrix, the majority of wild-type cells formed a single 221 continuous lumen inside the cyst of cells; however, most FIP5 and FIP1 double KO cells showed a 222 multilumenal phenotype and a small percentage showed an inverted polarity phenotype ( Figure 4E, F).

223
These luminal phenotypes were significantly more severe than FIP5 KO alone ( Figure 4F).

239
Rab11 specificity for a particular cellular pathway is achieved through interacting with effector 240 proteins, and our work revealed a role for the Rab11 effector paralogs Fip5a and Fip5b in apical cargo 241 delivery and microvilli formation during zebrafish intestinal development. In particular, we observed 242 enlarged Rab7-positive, Rab11-negative organelles in mutants. Normally, there is a homeostasis 243 established between Rab11 recycling from endosomes and maturation from early endosomes to lysosomes (Stenmark, 2009). We propose that without Rab11-Fip5 mediated removal and apical 245 recycling of essential apical cargo, this homeostasis is disrupted such that cargo to be recycled builds up 246 and the maturation process is delayed resulting in engorged Rab7-positive organelles.

247
One characteristic of MVID is loss of microvilli at the apical cell surface, yet the mechanism 248 behind microvilli phenotypes is still being revealed. Work from intestinal tissue culture and MYO5B

415
Blocks were mounted in OCT and 20um sections cut using a Leica CM 1950 cryostat microtome.
Sections were placed on FisherBrand charged slides (Cat # 12-550-15) and rehydrated in PBS for 30 417 minutes. Excess liquid was dried, and then a wax pen was used to draw around the edge of the slide.

418
Slides were then blocked with 2% BSA and 5% donkey serum (ThermoFisher Cat # NC9624464) in PBS 419 for 1 hour, then incubated in primary antibody (see antibodies in Table 1) diluted in block at room 420 temperature for 2-3 hours. Slides were then washed 4x 15 minutes each with PBS and incubated in 421 secondary antibody (see antibodies in Table 1) diluted in block for 1-2 hours at room temperature. Slides

451
All primer sequences are listed in Table 2. Guide RNA (gRNA) oligos were designed using ZiFIT  Table 2).
The RNA probes were transcribed with the T7 polymerase and labeled using the DIG RNA labeling kit