I was using RStudio and did this. Both tables printed had the NA in there, without errors.

Don''t know if it will work for you:

x <- "NA"

data_summary2 <- data.frame(
  Fonte = c("Motor", "Piso", "Resíduo"),
  GL = c(2, 3, 6),
  `Soma dos Quadrados` = c(11.765, 2.296, 0.362),
  `Quadrado Médio` = c(5.883, 0.765, 0.060),
  `Valor F` = c(97.59, 12.70, x),
  `Pr(>F)` = c(0.0000265, 0.00522, x)
)